3484 1 2 3 4 5 6 7 8 9 NATIONAL FEDERAL MILK MARKETING ORDER 10 PRICING FORMULA HEARING 11 12 DOCKET NO.: 23-J-0067; AMS-DA-23-0031 13 14 Before the Honorable Channing D. Strother, Judge 15 16 ---o0o--- 17 18 Carmel, Indiana 19 September 13, 2023 20 21 ---o0o--- 22 23 24 25 26 Reported by: 27 MYRA A. PISH, RPR, C.S.R. Certificate No. 11613 28 3485 1 A P P E A R A N C E S: 2 FOR THE USDA ORDER FORMULATION AND ENFORCEMENT DIVISION, USDA-AMS DAIRY PROGRAM: 3 Erin Taylor 4 Todd Wilson Brian Hill 5 FOR THE AMERICAN FARM BUREAU FEDERATION: 6 Danny Munch 7 FOR THE INTERNATIONAL DAIRY FOODS ASSOCIATION: 8 Steve Rosenbaum 9 FOR THE MILK INNOVATION GROUP: 10 Ashley Vulin (Remotely) 11 Charles "Chip" English 12 FOR THE NATIONAL MILK PRODUCERS FEDERATION: 13 Nicole Hancock Brad Prowant 14 FOR SELECT MILK PRODUCERS, INC.: 15 Ryan Miltner 16 17 For Edge Dairy Cooperative: 18 Dr. Marin Bozic 19 20 ---o0o--- 21 22 (Please note: Appearances for all parties are subject to 23 change daily, and may not be reported or listed on 24 subsequent days' transcripts.) 25 26 ---o0o--- 27 28 3486 1 M A S T E R I N D E X 2 SESSIONS 3 WEDNESDAY, SEPTEMBER 13, 2023 PAGE 4 MORNING SESSION 3489 AFTERNOON SESSION 3614 5 6 7 ---o0o--- 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 3487 1 M A S T E R I N D E X 2 WITNESSES IN CHRONOLOGICAL ORDER 3 WITNESSES: PAGE 4 Dr. Mark Stephenson: 5 (Continued) Cross-Examination by Ms. Hancock 3489 6 Cross-Examination by Mr. Miltner 3526 Cross-Examination by Mr. English 3582 7 Cross-Examination by Ms. Taylor 3584 Cross-Examination by Mr. Wilson 3598 8 Cross-Examination by Ms. Taylor 3600 Redirect Examination by Mr. Rosenbaum 3604 9 Recross-Examination by Ms. Taylor 3610 Recross-Examination by Mr. Wilson 3611 10 Recross-Examination by Ms. Taylor 3612 11 Dr. William Schiek: 12 Direct Examination by Mr. Rosenbaum 3614 Cross-Examination by Ms. Hancock 3642 13 Cross-Examination by Mr. Miltner 3674 Cross-Examination by Mr. English 3688 14 Cross-Examination by Ms. Taylor 3689 15 James DeJong: 16 Direct Examination by Mr. Rosenbaum 3708 17 ---o0o--- 18 19 20 21 22 23 24 25 26 27 28 3488 1 M A S T E R I N D E X 2 INDEX OF EXHIBITS 3 IN CHRONOLOGICAL ORDER: 4 NO. DESCRIPTION I.D. EVD. 5 179 E-Mail from Chris Allen 3532 6 176 Testimony of 3613 Mark Stephenson 7 177 IDFA-29 3613 8 178 IDFA-1 3613 9 180 Testimony of 3614 3707 10 William Schiek 11 181 IDFA-7 3614 3707 12 182 IDFA-8 3615 3707 13 183 IDFA-9 3615 3707 14 184 IDFA-10 3615 3707 15 185 IDFA-11 3615 3707 16 186 IDFA-12 3615 3707 17 187 IDFA-13 3615 3707 18 188 IDFA-14 3616 3707 19 189 IDFA-15 3616 3707 20 190 IDFA-16 3616 3707 21 191 IDFA-17 3616 3707 22 192 IDFA-18 3616 3707 23 193 IDFA-19 3616 3707 24 194 IDFA-20 3617 3707 25 195 IDFA-40 3617 3707 26 196 Testimony of 3708 James DeJong 27 197 IDFA-41 3709 28 3489 1 WEDNESDAY, SEPTEMBER 13, 2023 - - MORNING SESSION 2 THE COURT: Let's come to order. 3 Okay. We resume with this witness. 4 CROSS-EXAMINATION (Cont'd) 5 BY MS. HANCOCK: 6 Q. Good morning, Dr. Stephenson. Welcome back. 7 A. Good morning. 8 Q. I think we were -- I just needed to cover a couple 9 more pages in your survey study, and then we can move off 10 that and wind things up. 11 I'm on, just so we're clear, Exhibit 178, which is 12 your 2023 report. And I'm on page 26. We were talking 13 about the ledger. 14 A. Okay. 15 Q. And I don't know for sure if I asked this. I 16 might have. If I'm repeating myself, just forgive me. 17 I'll get going again. 18 That depreciation number that's assigned there, I 19 think you -- that's a number that either the person 20 filling it in can populate, or if there's nothing there 21 and it was zero, you would assign a number. 22 Is that where we landed? 23 A. No, I wouldn't assign a number. If -- if it was 24 left blank, then it's not included. It would be in the 25 sum of these costs as a zero value. 26 Q. Okay. I thought -- is that where you said you 27 used an economic depreciation number? 28 A. I had asked them to do that, to use an economic 3490 1 depreciation. 2 Q. Okay. And so if they didn't put in a number, it 3 would just be zero -- 4 A. Which would grossly understate the consumption of 5 capital in a plant. 6 Q. Which in turn would show that they had much lower 7 costs for the operation of their plant based on the 8 numbers that they would be providing. 9 A. Yes. 10 MS. TAYLOR: Excuse me. I apologize for 11 interrupting. But since everyone gets text messages, I 12 think the webcast is down at the moment. We're aware and 13 fixing it, in case you get a message. 14 I do think we continue. The webcast is nice, but 15 not necessary to have the hearing. So I just wanted to 16 let everybody know. And I do apologize for interrupting. 17 BY MS. HANCOCK: 18 Q. Okay. And then let's turn to the next page. I'm 19 on page 27. This is the final page that someone will get 20 to let them know that they have completed the survey; is 21 that right? 22 A. That's correct. 23 Q. And it says -- you have a message there that says, 24 "I will scrutinize your data for completeness and 25 consistency, and then if I feel there is questions, I will 26 contact you for clarification." 27 Did you do this for both your 2021 and 2023 28 surveys? 3491 1 A. Yes. I have always done that. 2 Q. Is -- when you say you "scrutinize the data for 3 completeness and consistency," are you just walking 4 through each one of the tabs for each one of the plants 5 that had responded? 6 A. Yes. I mean, and I -- we talked yesterday, I had 7 a bit of conversation about some of those cross-checks 8 that are kind of built into the data collection here. If 9 they are out of bounds, then that's a red flag for me to 10 go in and take a closer look and see what might be missing 11 and to follow up and try to get that resolved. 12 Q. Okay. And how many of -- or what percentage of 13 the plants did you have to cull back and follow up on? 14 A. A relatively small number. Sometimes I would send 15 an e-mail, just a quick e-mail about, I need this to be 16 completed. Because, typically speaking, there would be a 17 few entries or boxes that didn't have completion. 18 Just to give you an example of that, the market 19 value of assets, the way that this was asked for, you 20 know, suggested that what would you expect you could sell 21 your plant for today? And for a few companies, that was 22 just a -- a question that they couldn't wrap their head 23 around, had never been contemplated, they didn't know how 24 to answer that, and they left it blank. 25 Q. And then you could -- you would update that data 26 based on the conversation or the e-mail response that you 27 received from them? 28 A. Yes. Or they could update it directly. I mean, 3492 1 they can go back in and -- and enter or fix data. That's 2 what I prefer they do. But if it was a verbal response, 3 then I could go back in and do it. 4 Q. And I think you answered this yesterday, but you 5 said you don't -- this doesn't generate a report for plant 6 for you. You are just getting what we're looking at here, 7 which is the tabbed information for each plant? 8 A. That is just data collection at this point. And 9 it's an organized method of asking for and collecting that 10 data. And it does provide some sub summaries. So, for 11 example, when you take a look at that little packaging 12 cost that's being calculated, that is a packaging cost 13 that's going to show up for the plant at the -- on 14 individual reports. 15 Q. And -- and did you provide the responders with a 16 report back based on the calculations that you had done on 17 their plants? 18 A. For this particular study, not all of them. A few 19 of them had requested that, and I did send that back. 20 Historically, it's nice to be able to give people, 21 you know, reports back. Part of what folks who -- 22 participants, you know, like about this is that it gives 23 them that external benchmark, how am I doing relative to 24 the other body of plants that participated. 25 Q. So what did you provide them back to be able to 26 help them put their numbers in context with the other 27 plants that had responded? 28 A. Well, for this particular study, there were only a 3493 1 few plants that had asked for that report back, and I did, 2 you know, provide them information about what their 3 summary costs were. So it was very much like that table 4 that provides a summary of all the plants, but it did 5 provide the same data for their operation. 6 Q. Okay. And then by the time you issued your data, 7 they were able to use as their cross comparison? 8 A. Yes. If they'd asked for it. And many plants 9 didn't. 10 Q. Okay. They -- you said many plants did not? 11 A. Did not. They supplied the data, and that was 12 done at that point. 13 Q. Okay. Let's turn to on page 178 -- or I'm 14 sorry -- on Exhibit 178, let's turn to page 4. 15 And this is in your Plant Selection area of your 16 2023 report. 17 A. I'm not quite there. 178 -- I see 177 -- oh, 178. 18 Sorry. 19 Q. Exhibit 178, page 4. 20 A. Page 4. Okay. 21 Q. And this is under the Plant Selection heading. 22 You see that? 23 A. Uh-huh. 24 Q. And it -- it says you maintain a proprietary list 25 of about 687 dairy plants in the U.S. 26 And yesterday I had asked you how many you 27 surveyed. Did you use your proprietary list to reach out 28 to those plants? 3494 1 A. I -- I have in the past, but there have also been 2 times when I have asked for other help and guidance in 3 selecting the plants that, for example, were producing the 4 NDPSR products. So I know whether a plant is producing 5 cheese or if they are producing fluid milk or yogurt or 6 whatever it may be. I may even have some breakdown of the 7 kinds of products within that broader selection out there. 8 It's difficult to maintain a database like this 9 because plants are changing all the time, and capacities 10 are increased, or plants go out of business, you know, 11 some -- some of those types of things. But that plant 12 location database is used for several different things, in 13 my past, at least, when I was working on materials. So, 14 for example, when we would do the U.S. dairy sector 15 simulator model, we also used that plant location database 16 to identify where plants are and what products they 17 produce. 18 Q. And then you go on to say that NASS shows that 19 there were 1,266 dairy plants in the U.S. in 2019, and 20 then consistent with what I believe you just said, which 21 is some of those are just very small and they wouldn't 22 have products that would be reported to NDPSR. 23 A. Correct. 24 Q. And can you tell me why you can't just use the 25 data that comes out of NDPSR, why you need to conduct an 26 additional survey on top of that to be able to get to the 27 actual cost data? 28 A. Well, NDPSR doesn't include any costs. It 3495 1 includes prices received for products that were sold. It 2 does give you some idea about the number and the pounds 3 sold and product that was reportable. But it doesn't give 4 you any idea about the costs or even which plants 5 participated in that. 6 Q. So you can use that as a backdrop for some pieces 7 of information, but then you have to dive in deeper for 8 the costs that we have talked about in the survey? 9 A. Yes. Absolutely. Yeah. 10 Q. And then you go on in the next paragraph to say 11 that "participation in this study is voluntary." 12 And that's referring to the 2023 study; is that 13 right? 14 A. Every study I have ever done. 15 Q. Okay. So that was going to be my next question. 16 So you have never had a study that you have done 17 that's been a mandatory reporting study? 18 A. No. 19 Q. And the one that you had done with USDA, the 2021, 20 you note in here "captured a good portion of the butter 21 and nonfat dry milk sales that were included in NDPSR, but 22 the proportion of cheddar cheese and dry whey was not as 23 complete." 24 Do you know what percentage of the cheddar cheese 25 and dry whey reported in NDPSR products that you did not 26 capture? 27 A. Well, I don't recall that off the top of my head. 28 I would have to go back and take a look at that again. 3496 1 But, I mean, there are pounds that are reported every week 2 in NDPSR for the four products. And I know how many 3 pounds are reported by the plants that I have here. I 4 don't know how many of the pounds that are reported by 5 these plants were actually reportable to NDPSR but -- 6 Q. How -- oh, sorry. 7 A. No, I'm done. 8 Q. How did you know that -- that you -- well, it 9 sounds like you were less than satisfied with the amount 10 of response for cheddar cheese and dry whey; is that fair, 11 for the 2021 study? 12 A. It was lighter than I expected. 13 Q. How did you make that determination that it was 14 lighter than what you expected? What did you compare? 15 A. Well, based on the past. We always have plants 16 that are invited to participate and some that don't choose 17 to do so. But, typically speaking, I get a pretty high 18 proportion of plants that are invited that actually do. 19 And I think that for many of them, it's a sense of 20 curiosity, you know, about what would be there. And for 21 some of them, it is also just maybe a sense of obligation 22 that this is something we should do for the betterment of 23 the industry, to have a report like this put out. 24 Q. It's fair to say that any -- all of the responders 25 could have different motivations for why they are 26 responding to the survey? 27 A. I think that's absolutely fair. Although, you 28 know, I can't say that I have gone back in and done a 3497 1 follow-up to say, why did you or why didn't you respond to 2 this. 3 Q. And I think you estimated that it took several 4 hours to complete the study? 5 A. Yes. 6 Q. So it's not a light undertaking for someone to do; 7 is that fair? 8 A. No, it's -- I think it's a fairly substantial ask. 9 It is going to require a person that's fairly well up in 10 an organization to spend at least a better part of a day 11 doing that. 12 Q. And so in the 2021 study where you say that you 13 captured a "good" portion of the butter and nonfat dry 14 milk sales, do you know what -- what amount is a good 15 portion? 16 A. No. But I would have said if I got 50% or better, 17 I would feel reasonably comfortable about that as a 18 sample. 19 Q. Okay. So is it fair to say that in 2021 you felt 20 like you were able to capture at least 50% of the butter 21 and nonfat dry milk sales reported in the NDPSR? 22 A. Yes, I think so. And, you know, I -- I felt like 23 there was good representation, at least, of the number of 24 possible operations there. But it wasn't quite as true 25 for the cheese and the whey plants. 26 Q. Okay. So is it fair to say, then, for the cheese 27 and the whey plant, you captured something less than 50%? 28 A. Of the volume. I believe so. Again, you're 3498 1 trying to lead me down a path where I have already told 2 you I can't quite recall that off the top of my head. I 3 could go back and recalculate that if that were important 4 for you. 5 Q. And I'm not trying to lead you anywhere. And I 6 know that you didn't know the number. I'm just trying to 7 get rough estimates so that I can at least bucket it in 8 one category or the other. 9 A. Yeah. I mean, let me just give you a big example. 10 If I got 10% of something, I would feel like that's a 11 pretty thin margin to use as a -- you know, as something 12 that was significant and representative of the sample. If 13 I got something that was 50% or greater, then I would feel 14 pretty good about it. 15 So if that gives you an idea about, you know, how 16 representative some of the samples were, then -- 17 Q. Can I take it from what you just described then 18 that for cheddar cheese and dry whey, it probably falls 19 somewhere between that 10 and 50%? 20 A. Probably. But, as I said, I don't know without 21 going back and taking a look at it. 22 Q. And that was for the 2021 survey. 23 In the 2023 survey, the one that IDFA and WCMA 24 commissioned, the next paragraph you say, "With the urging 25 of IDFA and WCMA to their members, participation of cheese 26 and dry whey plants was higher." 27 Is that fair? 28 A. Yeah. Those were my words. 3499 1 Q. So when you say "with urging," do you mean that 2 IDFA and WCMA encouraged their participants to respond? 3 A. They felt it was important to get a good update of 4 the cost of processing study. 5 Q. And then you go on to say in that same paragraph 6 on page 4, "It must be noted that a different sample of 7 plants makes it more difficult to compare results from 8 different studies." 9 Are you referring back to the '21 study being 10 compared to the '23 study? 11 A. Well, I would refer that to any kind of -- I mean, 12 that's kind of a generic statement. If I had all the same 13 plants in the 2007 study, the 2019 data, and the 2022 14 data, then it would be fair enough to say are these 15 representative of the census of plants that might have 16 been reporting. Perhaps, perhaps not. But at least 17 within these plants you can see what the trend has been. 18 Plants may have made investments in capacity or automation 19 or technology, but at least within the same plant you 20 would be able to see how their costs had changed over 21 time. 22 Q. Okay. It's fair to say that, because you had 23 only -- I think you said about 15 plants that overlapped 24 between 2021 and 2023, that meant that a large majority of 25 the plants that were studied in 2023 were different than 26 your 2021 study? 27 A. They were different, and I hope I conveyed that in 28 the body of the text in here. I mean, certainly some of 3500 1 them were the same, but we -- we got plants in both of 2 those studies that just were in one but not the other. 3 Q. Okay. And then you were just noting here that 4 from the role that -- that you were in, that you're just 5 making a qualifier here that says you have to take this 6 2023 study with a grain of salt because that different 7 sampling of plants can make it different to compare those 8 two, if it were instead an apples to apples comparison? 9 A. I don't think I used "grain of salt," but I think 10 that it's fair to understand that if they aren't the same 11 plants in there, that you can get different results. I -- 12 I have tried to make that point rather -- many times I 13 think in here that the sample matters. 14 Q. Yeah. And I think that you have. I just want to 15 make sure that I'm exploring that as well. 16 And you did, actually, get different results 17 between your 2021 and 2023 survey results; is that right? 18 A. Yes, I did. 19 Q. And -- okay. Do you know if when the IDFA and 20 WCMA were urging their members to participate in the study 21 that they knew what the study was going to be used for? 22 A. I wasn't privy to any of those phone calls or 23 e-mails, so I -- I don't know. I would assume, but I 24 don't know. 25 Q. Did you know what the study was going to be used 26 for for 2023? 27 A. Oh, absolutely. I mean, I would not have 28 responded with a degree of urgency to get it done and get 3501 1 it prepared, you know, had it, you know, not been for the 2 hearing that was upcoming. 3 Q. So you knew it was intended to be used for this 4 hearing, to determine the Make Allowances that USDA would 5 be considering at this hearing? 6 A. I did. 7 And, you know, I could go a step further to say 8 that because I was not as satisfied with the results of 9 the 2019 data, felt that, you know, they had question 10 marks as far as I was concerned from what I might have 11 expected, that it would have been hoped that you might get 12 a better sample and better results here. And of course, 13 if you have an organization like IDFA or WCMA urging 14 members to participate, then, you know, I felt that we 15 might get a better sample. 16 Q. Okay. Why were you not happy with the results 17 that came out of 2021 survey? 18 A. Well, some of the results looked to me like -- I 19 mean, I report the data. I don't cook the numbers or do 20 anything with it. It is the data as I receive it. And 21 yet there were still some questions that you had with the 22 results of products, like butter, as a good example, in 23 the 2019 data that I might have expected would have been a 24 bit higher than that. 25 Q. And I think for butter you had $0.1411? 26 A. I'd have to look, but that is -- it is in that 27 ballpark, yes. 28 Q. Okay. And you ended up almost three times higher 3502 1 at $0.3176 after IDFA and WCM -- 2 A. I probably would have said a little more than 3 twice as high. 4 Q. Okay. It's a little less than three times, a 5 little more than twice, something like that? Okay. 6 Other than the butter number coming in low from 7 the '21 survey, anything else that you felt like was 8 deficient or lacking? 9 A. No. That was the one that stood out to me in 10 particular but -- and at any rate, I felt like it was a 11 worthwhile effort to go in and redo a study to see if we 12 couldn't get better sample. 13 Q. And I think you -- you also talked about changing 14 some of your methodology between 2021 and 2023. 15 Can you talk about that a little bit more? 16 A. Yeah. The 2021 study was where I introduced the 17 idea of a weighting of unallocated costs by the degree of 18 product transformation. And, you know, I gave examples. 19 So if you had a plant that brought in raw milk and perhaps 20 made nonfat dry milk and sold most of the cream as cream 21 instead of churning it to butter, then for a rather 22 lightly processed product of cream, you would have been 23 overallocating costs with the methodology of using the 24 pounds of components in the cream in comparison. So that 25 would have overallocated costs to cream and underallocated 26 them to powder as an example. 27 By the same token, it could have been the other 28 way around. If you churn butter and sold a lot of skim 3503 1 milk or condensed skim or something like that, then you 2 might have underallocated to butter and overallocated to 3 powder. 4 Q. And that was on a scale of 1 to 10 that you assign 5 that transformation? 6 A. It was on a scale of 1 to 10. I worked with 7 people from the Center for Dairy Research at the 8 University of Wisconsin. They are a group of academics 9 and others who work in dairy processing area, and I 10 explained to them what I was trying to do. And I didn't 11 want to get highly technical if I could. I just said, on 12 a scale of 1 to 10, for all of these products, can you 13 give me a ranking number that represents the degree of 14 product transformation that has to go on for the plants. 15 And so those were the numbers that they came up with. 16 Q. And so on that scale of 1 to 10, cream, for 17 example, was rated as a 2? 18 A. I -- 19 Q. I'm on page 7. 20 A. If you -- if you would have asked me, I would have 21 said a 3, but I couldn't remember. Yep. Cream is a 2. 22 Q. And then butter, for example, is a 6? 23 A. Yes. 24 Q. But whey protein concentrate powder would be a 10? 25 A. Yes. 26 Q. Okay. Meaning that whey protein concentrate 27 powder would take up the most costs for its 28 transformational value. 3504 1 A. For its transformational value, yes. 2 Q. Okay. And you did this to try and capture or help 3 you better analyze the financials that were coming in from 4 the more complex plants; is that fair? 5 A. It is fair. We have had -- we have had plants in 6 the past where they had reported data, and then when they 7 looked at results, you know, would say, ooh, that -- you 8 know, you have underestimated costs for this product and 9 overestimated them for that product. And, you know, when 10 you look to see what's happening, and your methodology, 11 you realize that you were putting a lot of weight on 12 components that were not being very heavily processed. 13 Q. And then in that 2021 survey, when you did publish 14 it, you noted in your report, in 2023, that the industry 15 reacts with some criticisms. 16 A. They did. And I think that, you know, part of the 17 criticism was the actual numbers, you know, that -- that 18 they looked at. Butter, as an example, was a fairly low 19 number. And this weighting scheme would have a tendency 20 to push butter values lower, would have a tendency to do 21 that. 22 But, in my opinion, in looking at the data and the 23 plants who were participating, it was more of a sample 24 problem than it was this weighting problem. But this is 25 what was obvious to people, that this had changed, and so 26 I think that folks were uncomfortable with that process. 27 The other process had been well established and -- and the 28 industry seemed to feel comfortable with it, so I went 3505 1 back to that. 2 Q. Okay. And when you say "went back to it," meaning 3 you didn't have a -- 4 A. I didn't transform the data. Sorry for jumping 5 in. 6 Q. No. That's okay. Your words are better than 7 mine. 8 So it just meant that when you went back in 2023 9 to your original methodology, you weren't assigning a 10 weighted value of cost, you were just allocating them on 11 your own based on the product mix that was being made at 12 the plant? 13 A. And the components in those final products. 14 That's correct. 15 Q. Okay. And do you remember when you were getting 16 those criticisms, was it from USDA, for example? 17 A. No. I -- I heard, you know, from a number of 18 people who looked at this with surprise. Some of them, 19 participants; some of them, industry organizations; some 20 of them -- well, maybe, regulatory, I don't think said, 21 gee, what's going on here. Nobody had -- had asked about 22 that in particular. 23 Q. Okay. 24 A. But, you know, legitimate I think to question 25 those data. 26 Q. And I thought that you had told me that you felt 27 like you had already captured a good portion of the butter 28 that had been reported on NDPSR in that 2021 survey. 3506 1 A. But not as much as this time around, the 2023 2 survey. 3 Q. Okay. So even though you felt like you captured 4 at least 50% of the butter in that 2021 survey, you still 5 feel like the results were somewhat impacted because you 6 didn't have enough of the butter responses? 7 A. Yes. And there were different plants. So even 8 though, you know, there was a reasonable volume of butter 9 in the sample, there were a different set of plants that 10 reported in 2021 versus the 2023 study. Some overlap, but 11 quite a few different plants. 12 Q. Okay. And do you think that if you would have had 13 all of the butter plants from 2021 and all of the butter 14 plants from 2023 combined, it would be even more accurate 15 than what you have in your 2023 study? 16 A. More is always better. 17 Q. Okay. And now on page 6, I think you have a 18 statement here that you were not trying to determine the 19 profitability of the plants; is that accurate? 20 A. Yes. So to determine profitability, you would 21 have needed quite a bit more data, such as the sales price 22 and value of the products from the plant as well as the 23 costs that were paid for dairy ingredients coming into the 24 plants, and those were specifically not included. And 25 some of the marketing costs that are incurred are not 26 included in this cost of processing. 27 Q. And then you used Moody's index value in 2022 to 28 calculate a return on the value of the assets at 5.07%. 3507 1 Is that right? 2 A. I believe that that was the number. 3 Q. And I'm pulling that off of page 9. So I'm not 4 making you guess. 5 A. Okay. Thank you. I will accept that you have 6 looked at the right number. 7 Q. And I'm just -- so for you, what do you -- how do 8 you interpret that return on the value of assets number? 9 What do you interpret that to mean? 10 A. Well, if you think about any -- any firm or 11 business that has a lot of asset value tied up in the 12 operation of the -- of the firm, then they would always 13 have an option -- maybe seems extreme -- but an option to 14 sell that plant and its assets to someone else and put 15 that money in a safe investment, or to invest it in a 16 different kind of operation or plant. So this is a means 17 of just saying that there is an opportunity cost to the 18 investment that you have, and your investment should 19 return something over time. 20 Q. But the cost of tying up your asset, because if 21 you didn't have them tied up here, you could deploy them 22 somewhere else? 23 A. I would like to think that your assets are working 24 for you whether, you know, it's just through a paper 25 investment or in physical assets of a plant. 26 Q. Do you believe that a 5.07 return on asset value 27 is a conservative number? 28 A. Well, I don't need to believe that, but when I 3508 1 look at the possibilities for what number you might pick, 2 I have used the same methodology that CDFA had used in the 3 past, and that was the Moody's Baa bond index. It is not 4 a risky bond or a junk bond, you know, that would have a 5 high interest rate. It is not a savings account type of 6 interest rate that would be exceedingly low. But it's a 7 very conservative safe bond that tends to have a lower 8 rate of return. 9 Now, I will say that that return on assets number 10 can be influenced by two things: Either, one, by the 11 value of the assets that you perceive you have, the market 12 value, or by the interest rate at the time. And our 13 interest rates have gone up, as I think most of us would 14 know over the last couple of years, thanks to action on 15 the part of the Fed, and bond interest rates have followed 16 that up. So it's a higher interest rate than we had in 17 earlier studies. 18 Q. And so you use that -- you use this return on 19 asset percentage and assign it to that market value of 20 asset number that we looked at on the survey on page 26; 21 is that right? 22 A. That's correct. 23 Q. And so depending on what the survey responder 24 plugs in as their own market value of assets, it can have 25 a big swing on the numbers; is that fair? 26 A. It could. It could have a big swing on that 27 particular number. That number is broken out, I think, in 28 the table. It's still a relatively small proportion of 3509 1 total -- yeah, there's a return on investment there. 2 Q. Are you on page 12? 3 A. Oh, I just flipped open to page 14, but let's take 4 page 12, that's fine. 5 So you are looking at the all plants number here 6 for nonfat dry milk processing. The return on investment 7 was $0.035 per pound. 8 Q. Okay. 9 A. So that's $0.035 per pound out of the $0.275 that 10 was reported for total costs. 11 Q. Okay. So tell me what your table -- if we're on 12 page 12, tell me what your Table 3 -- this is titled Plant 13 Costs for the Nonfat Dry Milk Processing -- what is that 14 table designed to capture there? 15 A. It's designed to capture all of the costs 16 partitioned into different cost centers, I guess, or ways 17 of identifying costs by -- by usage in the plant. So we 18 have tried to use the same methodology, again, that CDFA 19 has used in the past. Although CDFA did change a couple 20 of their segments over time, so -- but the more recent 21 ones I believe had segmented this into processing labor, 22 so in other words, what is labor, and you can kind of see 23 in here that was about 19% of the total cost from that pie 24 chart down there. 25 What about utilities? That's another major cost 26 center in plants, about 15% in this case. 27 What about packaging? About 7%. 28 Non-labor or utilities processing, that is a title 3510 1 that CDFA uses that I think confuses people, but it's -- 2 it's a host of other things that don't include labor or 3 utilities or some of the general and administrative costs. 4 Now, you can't look at the general and administrative 5 costs of that last ledger, general ledger chart that's on 6 there and just say, it is all of those. No, it's some of 7 those, and a few other costs as well. 8 So, for example, superintendent labor in a plant 9 is part of the general and administrative costs that are 10 listed here, as are secretarial support and, you know, a 11 few other things. 12 Q. Attorneys? 13 A. Attorneys, yes, of course. 14 Q. I just note that you called that out in there, and 15 so I thought it would be appropriate here to note. 16 So if we take the column -- so starting at 17 processing labor, utilities, packaging, non-labor or 18 utilities processing, and then G&A, those are all actual 19 costs that you have collected from the survey responders; 20 is that right? 21 A. That's correct. 22 Q. And then as well as the product pounds that are 23 noted there? 24 A. Yes. 25 Q. And on page 12 what we're looking at is all of 26 these are specific to their production and processing of 27 nonfat dry milk? 28 A. Yes. 3511 1 Q. And then you have a return on investment column 2 there. That's based on what we were -- a calculation that 3 you have allocated based on that return of asset value. 4 A. Yes. And, again, if you looked at that last 5 ledger page that was shown on page 26 of this report, the 6 market value of assets can be given to me as unallocated, 7 you know, for the entire plant, or you can try to break it 8 out to your cheese product or your butter products or your 9 powder products. 10 Q. If they didn't break it out and they just gave it 11 to you as one cumulative number, how did you allocate it 12 when you went back to -- for -- on page 12, for example, 13 to nonfat dry milk processing? 14 A. The same way that all of the other unallocated 15 costs were done as I have explained before. It would be 16 based on the pounds of components in those products. 17 Q. Okay. And so for the market value, though, you 18 didn't use -- you just used it based on allocation of the 19 pounds? 20 A. No, the market value would have been based on the 21 allocation. If it -- if they reported one number, you 22 know, for the market value of the plant, I would have 23 allocated that to, let's say, nonfat dry milk based on the 24 pounds of solids in the product. And then it would have 25 been divided -- I mean, the dollar number from that would 26 have been divided by the total pounds of nonfat dry milk 27 reported processed by the plant. 28 Q. Okay. And so -- and then -- back on page 12 3512 1 again. And then you take all of the actual costs that are 2 reported, and then you have added in a return on 3 investment that was based on that 5.07% return on asset 4 value. 5 A. Yes. 6 Q. And then as it's been applied to the market value 7 number that was input by the survey responders. And then 8 you have allocated that based on the pounds of solids for 9 this particular product if it was reported generally. And 10 then you have come up with this assigned number that we're 11 looking at on page 12 in this example? 12 A. Correct. 13 Q. Okay. So, for example, if we are looking at, for 14 low cost plants, it's $0.0152 -- is that per pound? 15 A. That would be per pound of nonfat dry milk powder. 16 Q. And that would be a return on investment that was 17 assigned to that low cost plant, and then you have added 18 that -- all of those actual costs and the return on 19 investment to come up with the total cost there. 20 A. Correct. 21 Q. And then the same would be true for a high cost 22 plant, they would have -- for a high cost plant in this 23 example, it is $0.0569 per pound for return on investment; 24 is that right? 25 A. Correct. 26 Q. And -- and then -- so if we do the math on a high 27 cost plant example, and the return on investment as a 28 percentage of the total cost, I come up with 17.5%. Does 3513 1 that look right to you? 2 A. I think you are in the ballpark. And I should 3 trust your math, but I don't. 4 Q. You probably shouldn't. Never trust an attorney's 5 math. 6 A. Okay. You are looking at the high cost plants? 7 Q. Yes. 8 A. 17.5%. 9 Q. Okay. And so in this example, the return on 10 investment for the total costs of producing nonfat dry 11 milk processing at a high cost plant, they would have a 12 17.5% profit margin built into their cost here; is that 13 right? 14 A. That would be the return on investment here, yes. 15 Q. Okay. And so if -- we have heard some examples of 16 kind of a rough calculation on how profitability works in 17 the sale of cheese. So I'm going to give it a whirl, so 18 you just have to bear with me. 19 But you have a USDA cheese price that's set in 20 this example, we're talking cheddar cheese, the price that 21 would be set for calculating; is that right? 22 A. I don't think they set the price. They discover 23 the price. 24 Q. Okay. They discover the price. 25 And then if you subtract out an assigned 26 Make Allowance, then the net of that is the value to 27 determine that Class III price? 28 A. In -- 3514 1 Q. Very simple terms. 2 A. Mostly, yes. That neglects yield factors, but, 3 yes. 4 Q. Well, and that's a good question. 5 Did you do anything in your study to -- to 6 calculate or factor in yield? 7 A. I didn't, no. There are some data that are 8 collected that could be used to look at approximate yields 9 for something like butterfat, for example. But it would 10 be difficult with what I have collected to get a complete 11 set of yields, yield factors. 12 Q. But nothing that we have in your study as it 13 exists right now? 14 A. No. 15 Q. Okay. 16 A. I mean, that would be an addition of a few more 17 questions and reporting to do that. 18 Q. And if we had the ability to provide you with a 19 wish list, would that be something that you would want to 20 have included? 21 A. If I were director of Dairy Programs, I would 22 probably want to take a look at every parameter that is in 23 my product price formulas. So, you know, the two 24 parameters that we have in there now are yield factors and 25 Make Allowances, but there are some interaction in terms 26 of the protein, values, and those kinds of things. I 27 think it would be worth examining those from time to time, 28 and I do think that yield factors can change over time as 3515 1 well. So practices do differ, even though we're making 2 some of these products to standards. 3 Q. Okay. So is that yes, you would like to have that 4 data if you -- 5 A. Yes, I would. 6 Q. Okay. And so if we use that Class III price 7 there, it's fair to say that the -- this return on 8 investment that we just looked at is built into that 9 Make Allowance that we have just subtracted out of the 10 cheese price; is that right? 11 A. If this were used as the Make Allowance, yes. 12 Q. Okay. And so if they just did nothing -- if a 13 processor did nothing else other than just sell exactly as 14 that formula allowed at the Class III price, they would 15 already have in this example that 17.5% return on 16 investment as a percentage of the total cost of that 17 product being made as a profit. 18 A. I think that's a little sloppy thinking because 19 you have selected the high cost plants and are assuming 20 that that is what would be included as the Make Allowance. 21 And maybe you would want to follow that through with the 22 all plants or something else. 23 Q. Okay. 24 A. But, I mean, at least, qualify your statement, I 25 think, for the high cost plant that's -- that's a 26 legitimate conclusion to draw. 27 Q. And that's fair because I'm here representing 28 National Milk, so I want to use the number that's best for 3516 1 me. 2 But we can -- we can use all plants. So I 3 calculate that one to be 11.1%. 4 A. Okay. 5 Q. Does that look about right to you? 6 A. I haven't done that calculation, but your earlier 7 calculation was spot on. 8 Q. Okay. Well -- and just for our record, I'm on 9 page 12 of 30 of your report, and so that's 11.1% for all 10 plants under the nonfat dry milk processing. 11 And so if we just use that number, and your number 12 was used for the Make Allowance, it's -- and all things 13 just being static, if -- if a plant were to sell a product 14 at the Class III price, that return on investment would 15 already be built into the sale of that product? 16 A. For the average plant that's in there, yes. If -- 17 if they were the average plant, then, you know, that's 18 what would be included in the Make Allowance, if that was 19 the Make Allowance. 20 Q. And if we harken back to day one, Dr. Vitaliano 21 told us the way averages work is that some go above and 22 some go below; is that fair? 23 A. Sure. Sure. 24 Q. Some will be higher than that then, and some will 25 be lower than that. 26 A. Yes. 27 Q. Okay. And so if -- if costs -- or if the plant is 28 able to improve, in the same scenario, if a plant is able 3517 1 to improve on its costs and build in some efficiencies, 2 they can make the product for less than the 3 Make Allowance, that's another opportunity for them to add 4 more profit to their bottom line; is that right? 5 A. It is. But, you know, the -- that's still -- I 6 mean, that is -- I think I stated there in the 7 commodity-based product orientation that the biggest 8 opportunity that plants have for increased profitability 9 is reducing their costs of production. 10 Q. And we have heard from other folks who have 11 testified already that if they can beat the 12 Make Allowance, they know that that helps build in some 13 additional margins for their products. 14 Is that how you understand it would work? 15 A. Well, I would think that it's always a legitimate 16 goal to reduce your costs of production. 17 Q. And so you understand that for the processors, 18 there's an inherent built-in goal of trying to increase 19 their Make Allowance as high as possible so that they can 20 try and beat it and build in their profit margin? 21 A. I do understand that we would always want to be 22 pursuing a higher profit in a plant, whether that's by 23 reduced cost or increased price of product sold or 24 whatever the efficiency may possibly be. 25 Q. And then if I can take the same example again, 26 where this Class III price has been set with this ROI 27 built into it, if -- if they are able to sell their 28 product for a price greater than the Class III price, so 3518 1 some premium pricing, that's then a third way that they 2 can also capture additional profits; is that right? 3 A. That's a possibility. 4 Q. And probably a goal, right? 5 A. Well, you would always want to have a price that 6 is the highest that you can achieve for the sale of your 7 product. 8 Q. Okay. 9 A. Other things being equal. 10 Q. And then did you look at Dr. Schiek's -- sorry, I 11 have to make sure I say that right, he's earned it -- did 12 you look at Dr. Schiek's modeling? 13 A. I didn't. I mean, I took a quick glance at it, 14 but I have not looked at it in any detail at all. 15 Q. Do you understand what methodology he deployed? 16 A. I understand roughly what was being done there, 17 yes. 18 Q. How would you characterize it? 19 A. How would I characterize it? 20 Q. Yeah. How would you describe his methodology? 21 A. It would be a standard methodology that economists 22 might use as one of the means of modeling plant costs. 23 Q. Is it an indexing of the costs? 24 A. From what I have seen, it is the use of prices 25 over time and factors that would be consistent with those 26 prices to carry them forward in time. 27 Q. Is it a methodology that you would use to set 28 Make Allowances? 3519 1 A. First of all, I'm not in that seat. I do think 2 it's probably better to survey plants. And this is 3 casting no dispersions (sic) on that particular 4 methodology, but if you survey plants, then you not only 5 get some idea about what the costs were that were 6 incurred, but also the factors that were used in the 7 production of the plants. So in other words, some of the 8 yields that were also changing with plants. 9 Q. Meaning you have to just take in actual conditions 10 as they continue to evolve over time, and that modeling 11 methodology doesn't take those into account? 12 A. No, it doesn't. I mean, a good example was 13 yesterday we had several discussions about things even at 14 the farm level, such as feed costs, for example, and how 15 those had been inflating over time. 16 One of the things that's been changing at the farm 17 level, because of higher feed costs, is increasing use of 18 genetics to try to promote more feed efficiency and 19 conversion in dairy cattle, so, you know, I want to use 20 less feed to make a hundred pounds of milk. Plants do the 21 same kind of thing. I want to use less electricity to 22 produce a pound of butter. And, you know, they have 23 energy recapture opportunities or technologies or 24 automation that may supplant some of the labor in 25 operations. So, sure, factors of production should be 26 identified over time. 27 Q. And an indexing methodology wouldn't capture those 28 items? 3520 1 A. It would be difficult to do that. I mean, you -- 2 you might make some corrections -- as I said, I haven't 3 read or looked at Dr. Schiek's studies, and I don't know 4 if he made any attempt to account for changing factors of 5 production. 6 Q. And I will say you were equally as maybe humble or 7 candid about your own methodology and some of the 8 shortfalls that were included in the process that you used 9 as well. Is that fair? 10 A. Nothing is going to be perfect. 11 Q. Are there improvements that we can make on the 12 2021 and 2023 studies to make sure that they are more 13 accurate? 14 A. Absolutely. I would be a strong promoter, I 15 guess, of making sure that we had the opportunity to have 16 a good and representative sample, if not a census of 17 plants, and that the data were -- at least had the ability 18 to be audited if you ever felt that was a question or 19 need. 20 Q. So if I just want to maybe summarize some of the 21 areas that I pulled out of either your testimony or your 22 statements, one of them would be having the study, if you 23 had your -- if you had your dream of making the most ideal 24 study, to make sure that it was the most accurate, it 25 would be changing it from a voluntary study to a mandatory 26 study. 27 Would that be one? 28 A. I think that that would be a good thing to do. I 3521 1 think the plants would recognize that it's another burden 2 that they would have to comply with, but that they would 3 at least feel that they are getting good and accurate data 4 to be using. 5 Q. And then if it was something that was collected on 6 a routine basis, it would -- it would be something that 7 the plant could have processes in place to make that 8 easier or more efficient for them? 9 A. They could for sure. And it would provide them 10 internally with a benchmark of their operation relative to 11 others. 12 Q. So there's a value externally but then also 13 internally for the plant as well? 14 A. I hope they would view it that way. 15 Q. And would another element be to make sure that the 16 information was standardized so that when a plant was 17 responding and inputting a number, we knew that we were 18 comparing items apples to apples? 19 A. I have tried very hard to do that myself. 20 Q. Are there ways that we can improve on it? 21 A. Boy, I -- I don't know other -- perhaps you would 22 want to get into the plants and be able to follow their 23 method -- internal methodologies all the way through to 24 make sure that the costs that are being reported are the 25 ones you expected them to be. 26 Q. So, for example, you have some broad categories 27 like when you talk about headquarters expenses. Is that 28 an expense category that we could have some 3522 1 standardization around that would make sure that everyone 2 was including the same numbers in that category? 3 A. Sure. I think that headquarters expense is kind 4 of a -- an imprecise title, you know, that may go in 5 there, that may include a variety of items for some plants 6 and not for others. So, you know, that's one of those 7 things, if you were taking it line item by line item, you 8 might look at and question, why is my headquarters expense 9 so high? 10 Well, if headquarters are also purchasing for 11 several of their plants, then you may not have a packaging 12 cost, for example, at the individual plant level or one 13 that's commonly reported but rather that's included in 14 headquarters cost offset. 15 Q. And I just used that as an example, but there's 16 other examples in there where we could be more precise 17 through a standardization; is that fair? 18 A. There could be. And when you get down to some of 19 the line items that are asked for in here, I think it's 20 hard to be any more precise than what we already have. 21 Q. And -- and then another area might be making sure 22 that the survey is auditable so that we could verify the 23 data that's in there. Would that be another way to 24 improve on the study? 25 A. I think that that's good. I mean, it helps to 26 provide understanding for AMS or whoever is conducting the 27 study to assure themselves that they know what's going on 28 in a plant, and I think it is always good to be getting 3523 1 into plants so that you have an idea of process and how 2 things may have changed. I was not in plants for this 3 particular study. I have been in the past many times. 4 And sometimes just walking around and looking at, you 5 know, what's being done, you know, you might ask a 6 question about, you know, product in cooler or something 7 else that you observe. 8 Q. Okay. And -- and another area that in a perfect 9 world would be more of a standardized allocation of costs, 10 whether it be using a weighting method or some systematic 11 way to make sure that the cost allocations are accurately 12 reflected in each product? 13 A. I think that you ought to agree as an industry on 14 how that allocation method could or should occur. Costs 15 have to be allocated in all plants, and we at least want 16 to make sure that it's being done the same way. 17 Q. And I think in your report you also noted that the 18 respondents have a great amount of latitude to include 19 information based on their own interpretation of their 20 financials. 21 Is that one of the areas that you think we could 22 improve on as well? 23 A. I don't recall saying that. 24 Q. And that -- 25 A. Can you point that out to me? 26 Q. Well, it would pertain to both the depreciation 27 assignments and then also the valuation of their own 28 market value assets. 3524 1 A. As a specific example, yes, that would be one of 2 the things that you would want to have a discussion on. 3 As I mentioned, CDFA used to construct and maintain a 4 depreciation schedule on every major piece of equipment in 5 the plants, and that's not something I was willing to do. 6 Q. Okay. And that's because once they have the data, 7 they can track that over time so that it can't be 8 something that somebody can manipulate each year through 9 responding to their studies; is that -- 10 A. Well, this would be where an audit would be 11 something that would be almost necessary, too, to 12 identify, do you have that piece of equipment, where is 13 it, how old is it, what did you pay for that. 14 Q. And it's fair to say those two categories of 15 numbers, both depreciation and the market value of the 16 assets, can have a considerable and significant impact on 17 the actual costs that is determined for each plant's 18 production of products? 19 A. They have an impact, for sure. I mean, as 20 Mr. Bauer mentioned, you know, his plant was fully 21 depreciated, and so to him this is a low cost plant. And 22 it's because he's identifying the non-cash costs of 23 depreciation, you know, as being zero for him, or close to 24 it. 25 I think that we could argue whether that was 26 really true in the world or not. But for his experience, 27 it's a low cost plant because he's not depreciating 28 equipment. 3525 1 Q. How do we account for the fact that the costs that 2 were collected in your 2023 survey were collected at a -- 3 at the peak of a global pandemic when costs were 4 extraordinarily high and -- and maybe not representative 5 of a wider time period? 6 A. I'm old enough to remember when we had 7 double-digit inflation, and I think that many people don't 8 recall that kind of thing at all. You know, it's 9 inflation less than 3, 4%, would be the very norm for 10 them, and interest rates in those kind of ranges. But 11 things will change over time. We had a large step, and I 12 don't know how rapidly we retreat from these interest 13 rates, but these interest rates may be with us for a long 14 period of time. 15 Q. So I understand on interest rates, and I'm just -- 16 maybe my question is a little bit different, because I'm 17 not just -- interest, obviously, is a big cost. But I'm 18 also just talking about just the general input costs or 19 the supply costs that a plant had in 2022, which I think 20 are -- were -- were extraordinarily high and not in a way 21 that was normal inflationary growth. 22 A. Well, if -- if cost studies were ongoing, I mean, 23 a regular thing, then you would have the opportunity to 24 watch and track those. 25 Back in the hearing for the changes that were made 26 for the 2008 updates to Make Allowances, the big 27 discussion at that point in time were the increases in 28 utility costs, electricity and gas. They had risen very 3526 1 rapidly, and the industry was very concerned about it. 2 They also retreated from those high levels. So costs do 3 change over time. I -- I would acknowledge that. And 4 they don't always go up. 5 MS. HANCOCK: Thank you, Dr. Stephenson. I 6 appreciate your time. 7 THE WITNESS: You're welcome. 8 THE COURT: Additional cross before AMS? 9 Yes, Mr. Miltner. 10 CROSS-EXAMINATION 11 BY MR. MILTNER: 12 Q. Good morning, Dr. Stephenson. 13 A. Good morning. 14 Q. Ms. Hancock covered some of the questions that I 15 had here, so I'm going to try to not repeat them as best I 16 can. I'm sure there will be some overlap, though. 17 Ryan Miltner for Select Milk Producers. 18 I was looking back -- and this is now the third 19 time I have had the opportunity to cross-examine you on 20 cost surveys, for what that's worth. 21 A. Congratulations. 22 Q. Yeah, thank you. Thank you. I'll get a 23 certificate and frame it or something. 24 I can promise you only the first couple of 25 questions will refer to those prior hearings. 26 In the 2006 report you provided, you calculated a 27 confidence interval for the plants that were reporting, 28 and you did not calculate a confidence interval for the 3527 1 2021 or the 2023 report. 2 Is that because the sample wasn't random? 3 A. No, it isn't. And, boy, Ryan, you are pulling me 4 back to a memory that barely tickles neurons on here. I 5 hadn't looked at that. But I believe that that was one of 6 the studies where we had, in fact, pulled from a 7 stratified random draw of geography and plant sizes. So 8 we could do some calculations to look at what was it that 9 we actually got in the way of the sample there. So that 10 was a different goal, I guess, we were trying to achieve, 11 to have representation across all geographies and across 12 an observed spectrum of plant sizes. And the goal this 13 time for me was to assure that we had as many plants as we 14 could from NDPSR-reportable products. 15 Q. Okay. And for the record, I -- I didn't have a 16 specific recollection of that exchange until I read it a 17 few days ago, so don't feel bad. 18 A. Okay. 19 Q. So similarly, in 2006, you calculated an R-squared 20 that tied the volume of product produced to the total 21 processing cost. 22 Same -- same rationale for why that wasn't 23 prepared for these two studies? 24 A. Yes. Over the years, many years, I have performed 25 different kinds of cost estimates of processing. Some of 26 them have been synthetic, you know, where we used an 27 economic engineering approach. Some of them have been 28 statistical, as I think we have seen with Dr. Schiek's 3528 1 study this time around. And some of them have been survey 2 approach. 3 My summary at the time I had done some different 4 approaches was that the sample approach, the survey 5 approach was the best estimate that we had of plants 6 within known sizes. If we were trying to go on size of -- 7 outside of observable practices, then you need to do 8 something else. 9 Q. So outside of a statistical measure, could you 10 reasonably draw any conclusions about the size of plants 11 and their cost of operations for the 2023 report you have 12 provided? 13 A. No, I haven't done that. I -- I mean, I can have 14 observations where -- in fact, you can see that in the 15 tables in here where you will look and find the 16 identification of the low cost and high cost groupings of 17 plants, and you will generally notice that they tend to be 18 larger volume plants, you know, that are low cost 19 operations. It's not absolutely true and characteristic. 20 We do have small plants that are very competitive, but it 21 does tend to be true in the aggregate. 22 Q. Okay. So now -- now we'll focus more on the 2023 23 and 2021 reports. 24 So in the 2021 report, there were 57 plants that 25 reported, if I pulled that out correctly. Does that sound 26 right to you? 27 A. That was plant product observations. So, yes. 28 It's a combination of -- of individual plants and the 3529 1 specific products that were reported as being produced 2 here in the NDPSR. So, for example, a plant that made 3 nonfat dry milk and butter would be counted as two 4 observations. 5 Q. Okay. So I'm looking at page 4 of that report, 6 and you don't necessarily need to look at it unless you 7 really want to, but I think there were 61 plants broken 8 down by product, and 57 plants -- so as I -- as I tried to 9 figure this out, I got 57 plants and 61 plant product 10 observations. 11 A. I don't believe that there were that many total 12 plants. I would have to -- 13 Q. Okay. 14 A. -- go back and check that. That's -- that's a 15 large number. 16 Q. Okay. I will take your take your word for that. 17 I think you know better than I. 18 In the 2023 report, you had 45 plants, correct? 19 A. Well, Ryan, do you know what page you pulled that 20 from? 21 Q. That's what I'm looking for. 22 A. Somewhere I know I described the number of -- I 23 think it was firms, plants, and plant product 24 observations. 25 Q. The problem is that we have about four different 26 documents to look at. 27 A. Yeah, okay. Maybe. 28 Q. Here we go. At the bottom of page 4 of the 3530 1 report, not your testimony. 2 A. Okay. 3 Q. Okay. So the last paragraph, there were 15 4 participating firms with ownership of 45 different plants, 5 and then later in that paragraph, a total of 55 plant 6 product observations. 7 A. Okay. 8 Q. Okay. Now, there was also a -- and I have -- 9 really don't recall if this was something you stated or 10 whether it was written in a statement -- approximately 15 11 overlapping plants. 12 A. Yes. 13 Q. Okay. Are those plants or plant product 14 observations that overlapped? 15 A. Those would be plant product observations. 16 Q. Okay. And -- 17 A. So they would correspond with this 55. 18 Q. Did you state where that overlapping occurred in 19 terms of which products overlapped? 20 A. I didn't. It was all four products. I don't 21 recall if there was a predominance of one or more 22 products, but I think it was fairly uniform. It wasn't a 23 high number. I would have hoped for more overlap but -- 24 Q. So I want to ask about the process for soliciting 25 the plants to participate in the 2023 report. 26 And in your 2021 report, you stated that you 27 referred to your proprietary list of plants. Based on 28 your experience you invited 153 to participate? 3531 1 A. Uh-huh. 2 Q. And you explained that on page 3 and 4 of your 3 2021 report. 4 For the 2023 report, did you send any invitations 5 to plants to participate in the 2023 report? 6 A. I think that I did. Although, honestly, I can't 7 recall as clearly. I do know that IDFA had urged 8 membership, which at the time was, you know, most of the 9 plants that we would have had in the report, to 10 participate in this. 11 Q. And I know you said you don't recall, so this is 12 going to be an awkward question, I suppose. Would you 13 have sent an invitation to the 96 plants who didn't 14 respond to your 2021 invitation and asked them to consider 15 participating this time around? 16 A. I was trying to make sure that we got at least the 17 plants that we had last time and then pick up some 18 additional plants. So I was more focused on assuring that 19 we had the same plants, if we could, so we could look at 20 how costs may have changed over this intervening years, 21 and then pick up some additional plants as well. 22 MR. MILTNER: Your Honor, I have a document I'd 23 like to ask the witness some questions about. Could I 24 approach him with this? 25 THE COURT: Yes, you may. 26 MR. MILTNER: I do have copies for your Honor and 27 the rest of the room. 28 THE COURT: Is this something you want marked? 3532 1 MR. MILTNER: Yes, please. 2 THE COURT: Mr. Miltner's handing this document 3 out. Let's go ahead and mark it for identification as 4 Exhibit 179. 5 (Thereafter, Exhibit Number 179 was marked 6 for identification.) 7 THE COURT: I guess we should have some 8 description. It's a February 16th, 2023, e-mail from 9 Chris Allen, A-L-L-E-N. It doesn't -- the "to" line is 10 redacted. The subject line is "FMMO Update and New 11 Stephenson Cost Survey." 179 for identification. 12 MR. MILTNER: I trust you over me, your Honor. 13 And were those comments off the record or on? 14 THE COURT: That should be on the record, just 15 like we always do, an identified document. Yeah. This 16 witness has been shown the document, and Mr. Miltner's 17 going to ask him some questions about it, as I understand. 18 MR. MILTNER: I am. 19 BY MR. MILTNER: 20 Q. And I would also -- I appreciate your Honor 21 characterizing the document. I really would like to focus 22 on what appears after the forward, and that is that this 23 is an e-mail from Michael Dykes, sent on February 16th, 24 2023, to himself with CC's and -- and a number of blind 25 copies. 26 Dr. Stephenson, have you -- did you receive this 27 e-mail by chance? Were you a blind copy recipient? 28 A. No. This is the first time I have seen this. 3533 1 Q. In the very last paragraph, it -- the e-mail 2 reads: "If you have any questions about your company's 3 participation in the survey, we encourage you to contact 4 Dr. Stephenson directly." And I have redacted phone 5 numbers and e-mails. 6 Did any members of IDFA contact you directly about 7 the participation in the 2023 report? 8 A. Yes. I did have a few. Most of them would have 9 CC'ed -- in fact, I think all of them probably did CC a 10 member of IDFA's staff. But there were a few plants -- 11 and this is not unusual for me -- who are concerned about 12 the proprietary nature of the data that's collected, and 13 they wanted a non-disclosure agreement signed, so I was 14 always willing to do that. And that was the nature of the 15 direct contact from plants, you know, inquiring about the 16 process. 17 Q. And I think you stated how important it is to 18 maintain the confidentiality of firm information for your 19 reputation and the integrity of the study? 20 A. I think it's critical. 21 Q. In the third paragraph of this e-mail, it reads: 22 "In anticipation of a possible USDA hearing to consider 23 possible adjustments to Make Allowances, IDFA and WCMA 24 have commissioned Dr. Mark Stephenson to update his most 25 recent cost of processing study to capture manufacturing 26 data from 2021 to 2022." 27 This e-mail, having been sent on February 16th, I 28 assume that you were contacted and commissioned prior to 3534 1 February 16th by IDFA? 2 A. I had been contacted about updating this study to 3 see if I was willing or prepared to do that. And, yes, I 4 had been prior to this e-mail. 5 Q. And earlier you testified that you might have, but 6 you don't specifically recall, whether you invited anybody 7 to participate in the 2023 report, but you did testify 8 that IDFA and WCMA urged their members to participate. 9 Other than the urging, which you reference, and 10 perhaps your invitation, would there have been any other 11 way for industry participants to participate in the 2023 12 report? 13 A. Anybody that would have contacted me and asked 14 about it, or even if they had the link to the online 15 application that did collect the data for me, would have 16 been able to participate. They would have been included 17 in here as any other plant, only if they made products 18 that were NDPSR reportable and had been completed to my 19 satisfaction. 20 Q. So if -- if a firm received the link somehow, and 21 they clicked on the link, it would take them to the 22 reporting software that you have included as an appendix 23 and went through yesterday, correct? 24 A. Correct. 25 Q. Would they have -- would they have been required 26 to have any specific access key or something in order to 27 start inputting data? 28 A. No. Like some other protected sites, if you have 3535 1 not been a guest there before, the first thing you do is 2 to enter a user name and password. And when you have done 3 that, you then have access to begin entering individual 4 data. 5 Q. But they would have to have the link in order 6 to -- that site -- I guess, I'm -- let me rephrase that, 7 make it more of a question. 8 Would there have been a way for someone to locate 9 the survey software without having been provided that link 10 by somebody? 11 A. It's conceivable. The search engines crawl 12 through websites and identify all kinds of things. And I 13 did not put something on that page that says "don't index 14 this" or "don't search here." But it was not a widely 15 advertised link. Well, I didn't want to deal with a lot 16 of mischief either. 17 Q. I can understand. 18 When did you complete the work on the 2023 report? 19 A. Oh -- I don't recall exactly the date, but it 20 would have been in June. I mean, the final report was 21 written up very shortly after. 22 I had urged folks -- this e-mail is not incorrect. 23 I mean, it is quite correct. I had hoped that by 24 April 14th we would have all of the entries. But as I 25 mentioned the other day, not everybody gets things done 26 quite as quickly as I might want. And we had a number of 27 plants that said, would it be okay if we extend that 28 deadline a little bit, and -- and I moved the goal post a 3536 1 couple of times. 2 Q. And "June" is a perfectly fine answer for my 3 question. 4 When you did complete the report, did you submit 5 it to IDFA and Wisconsin Cheese Makers? 6 A. I did. 7 Q. Did you submit the report to anyone else? 8 A. I believe that it was circulated at that time. I 9 was not trying to be closed or careful about that in 10 particular. I suggested that they circulate the link that 11 had the report posted on it. 12 Q. When you say "it was circulated," it was 13 circulated by IDFA or Wisconsin Cheese Makers? 14 A. I posted the final report on a former website that 15 I maintained at the University of Wisconsin. I was 16 retired at the time but still able to do that much. And 17 it was, you know, as circulated as people wanted it to be. 18 So in other words, anybody that had known about that could 19 rather easily download the report. 20 Q. Thank you. 21 On your 2021 report on page 3, you describe it as 22 a working paper. And I'm going to break my almost 23 promise. In 2026, that was a working paper, too, and in 24 your testimony -- I'm sorry, 2006 -- in 2006 -- maybe 25 we'll be here again in 2026. 26 In 2006 -- 27 A. I won't, Ryan. 28 Q. In 2006, you testified, "In academia, we refer to 3537 1 a working paper as something that is not the final paper 2 on the entire project you are doing." 3 Is that a fair description of the 2021 report 4 also? 5 A. No. I -- by the time that these have come out, I 6 want to make sure that when they are released to anybody 7 that they are what I would consider to be a final 8 document. And the reason for that is that I have just 9 discovered in the past that if you've got another plant 10 that's submitted data, or two, or three, that might 11 change, you know, reporting tables just a bit, you know, 12 with that extra plant data in there, it just becomes 13 confusing, because people start to reference two or three 14 different working papers that have slightly different 15 values. 16 Q. Do you have the 2021 report handy with you? 17 Exhibit 158, I believe. 18 A. I think so. Although you referenced a page 3, and 19 mine here starts with page 8 of 33. I don't know if I 20 pulled something else off. But it's -- I have here 21 Hearing Exhibit 158. To my point -- 22 MR. MILTNER: Mr. Rosenbaum has handed me a copy 23 of his Exhibit 158. 24 And, your Honor, could I hand this to the witness? 25 THE COURT: Yes. 26 Mr. Rosenbaum, did you want to say something? 27 MR. ROSENBAUM: This is probably unnecessary, but 28 because there had already been a copy of this exhibit 3538 1 entered as 158, we had originally planned to make it an 2 attachment to another document, and that's why he has a 3 copy that has a different numbering. But we're going to 4 give him the official 158 so that Mr. Miltner's questions 5 and -- will be -- have the same pagination as -- 6 THE COURT: No problem at all. 7 MR. ROSENBAUM: -- what the witness has. 8 THE COURT: Smoothly handled. I don't think 9 anyone is going to object. I also have a copy. 10 MR. MILTNER: And I appreciate Mr. Rosenbaum 11 helping us out with that. 12 BY MR. MILTNER: 13 Q. All right. Dr. Stephenson, do you see the third 14 paragraph on page 3 there? 15 A. Yes. "This report is considered to be a working 16 paper." 17 Q. So the document that is Exhibit 158, is that 18 indeed a working paper? 19 A. No. And I can explain I think why that's in 20 there. 21 Q. Okay. 22 A. This was a few paragraphs of copied text from an 23 earlier report. So copy and paste, that was an editing 24 error on my part. 25 Q. Great. So the -- 26 A. The final report that was posted on the website 27 and made widely available through IDFA -- or not through 28 IDFA -- through USDA and others was, in fact, the final 3539 1 report. 2 Q. And then so I note the 2023 report does not refer 3 to it being a working paper. 4 Is the 2023 report a final report? 5 A. Yes, it is. 6 Q. Okay. So, now, with respect to the 2023 report -- 7 the next questions I have deal with -- primarily with 8 NDPSR. 9 So in the 2023 report, on page 4, you stated, "In 10 the 2021 study, plant selection was more targeted. It was 11 felt important to assure that plants producing 12 products" -- I'm sorry -- "producing product that may be 13 included in the National Dairy Products Sales Report, 14 NDPSR, which determines the product prices used in the 15 PPFs should be solicited." 16 I didn't see a similar statement in your 2023 17 report, but from your testimony, can I conclude that that 18 was still an aim of the 2023 report? 19 A. It was. In other words, I didn't want to go back 20 to earlier criteria that we used, such as geography and 21 plant size, that type of thing, or frontier of best 22 practice plants. This was meant to be NDPSR targeted 23 operations. 24 Q. So even though you targeted plants in the 2021 25 report that produced the NDPSR-reportable commodities -- 26 and I'm looking now at page 4 of Exhibit 158 -- I see that 27 you targeted plants that produced 640-pound block cheddar. 28 A. Yes. 3540 1 Q. Which is not an NDPSR product, is it? 2 A. No. 3 Q. Why would you want to survey those plants? 4 A. This had been a request from USDA to look at or 5 include those, or not exclude them from the 2021 study 6 that was done. Not that we would use that in reporting 7 here, but if it was decided that this is something that 8 you might want to include, could we have the beginnings at 9 least of a benchmark of those costs. 10 Q. Were plants that produced 640-pound blocks 11 actually responding -- were they respondents to the 2021 12 request? 13 A. Yes. But all of the plants that produce 640s also 14 produced 40s. 15 Q. Were the costs of those plants producing 640-pound 16 blocks segregated from the cost of producing 40-pound 17 blocks? 18 A. The plants weren't segregated. The costs up to 19 the point of packaging were not deemed to be different or 20 I didn't treat that differently. And likewise, on the 21 general ledger post processing, you know, for table 22 summary data, would have been included packaging costs, 23 and handling of the 640s were not. 24 Q. Did any of the plants that produced both 640-pound 25 blocks and 40-pound blocks attempt to self-allocate their 26 costs between those two products? 27 A. I don't recall. I would have to go back in to 28 take a look and see. You may recall yesterday as we were 3541 1 going through some of the screenshots, that if you 2 reported cheddar cheese processing as a product in your 3 plant, then the next page also allowed -- or began to ask 4 questions about what package sizes were produced. So 5 40-pound blocks could have been one of those. 640s could 6 have been one. 500-pound barrels could have been one. 7 Mammoth could have been an opportunity. But, you know, 8 we're not -- and we wanted to take all of those plant 9 products then and be able to ask specific packaging costs 10 for the different package sizes if we're interested in 11 them. 12 Q. So you have hit on a term I'm not familiar with. 13 What's a mammoth? 14 A. It's a large wheel of cheddar cheese. They can be 15 of various sizes from probably 50 pounds to 250 pounds or 16 greater. 17 Q. Similar to a Parmigiano Reggiano wheel? 18 A. Yes. They are usually made for specialty 19 purposes. 20 Q. Just curious, did anybody report those to you? 21 A. No. 22 Q. Okay. Okay. So similarly, Table 1 of 23 Exhibit 156, you targeted nonfat dry milk or skim milk 24 powder? 25 A. Table 1 of 156. 26 Q. Sorry, did I say -- it is the -- 158, my 27 apologies. 28 A. 158. Okay. 3542 1 So question again, please, Ryan? 2 Q. Yes. You targeted plants producing nonfat dry 3 milk or skim milk powder? 4 A. Yes. 5 Q. Is it your experience that the costs for skim milk 6 powder are very similar to those for nonfat dry milk? 7 A. Very similar. The make could be different, I 8 understand that. If you are reblending lactose back into 9 a finished nonfat product, that may be more expensive 10 than -- than pulling protein out in the liquid form ahead 11 of time. 12 Q. Did you similarly exclude the packaging costs of 13 plants that were making skim milk powder in the same 14 manner as you did for plants producing 640-pound blocks? 15 A. No, not necessarily. If -- if they were 16 25-kilogram bags or 50-kilogram bags or totes, then that 17 would have been included as well as being not 18 significantly different packaging. 19 Q. And then finally in the same table you list plants 20 producing dry whey or WPC. 21 Did you -- did you have any plants that responded 22 that produced WPC? 23 A. I did, but not enough to report. 24 Q. So there was no reporting of WPC -- what do you 25 mean by "not enough to report"? 26 A. I would not report data if I didn't have at least 27 three plants that had reported data, for confidentiality 28 reasons. 3543 1 Q. Is it true that WPC is a very different product 2 from dry whey? 3 A. Absolutely. 4 Q. That the difference between dry whey and WPC is 5 very distinct? 6 A. It is very distinct. I understand the process of 7 manufacture as well. This, again, was a request on the 8 part of USDA, could we possibly include that and take a 9 look at this, but it was not reported to USDA because 10 there were too few to report. 11 Q. Did any of the costs of manufacturing WPC get 12 included in the results of the 2021 study? 13 A. No. 14 Q. Okay. So the eight plants that were reported, 15 those were all exclusively dry whey producers? 16 A. Yes. 17 Q. With respect to nonfat dry milk, you had an 18 exchange yesterday with Mr. Rosenbaum about an AMPI plant 19 that produced high heat nonfat. 20 Do you recall that? 21 A. I do. 22 Q. And if I heard correctly, you concluded that the 23 AMPI's plants costs were at or near the median of the 24 total reported plant costs; was that correct? 25 A. That's correct. 26 Q. And so including or excluding that specific plant 27 had a negligible impact on the overall survey; is that 28 right? 3544 1 A. That's correct. 2 Q. You also stated that that AMPI plant producing 3 high heat nonfat dry milk was included initially because 4 its response to you indicated only nonfat dry milk and 5 didn't indicate whether it was high heat or low heat; is 6 that correct? 7 A. That's correct. 8 Q. Do you know if there were any similar ambiguous 9 product designations that occurred with other reporting 10 plants? 11 A. I'm not aware of them. It -- it is a possibility. 12 This is one of the places where audited reporting might be 13 useful to have for something like that. I was not aware 14 that that was a high heat only plant. And, you know, the 15 reporting on it should have been nonfat dry milk high 16 heat, but it was just nonfat dry milk. 17 Q. And the next question I had, which I'll read for 18 the record, but we'll skip is: Is this the kind of error 19 that one would expect to avoid with an audited survey? 20 A. Yes. 21 Q. Okay. With respect to 40-pound blocks, I believe 22 it was Agri-Mark's witness earlier testified that they 23 produced 40-pound block cheddar, and then depending on the 24 quality of the make, they will decide what they will sell 25 as a commodity and what might be aged to become Cabot 26 cheese. 27 How would a situation like that be handled in your 28 survey, where a 40-pound block manufacturer, clearly 3545 1 making a 40-pound product, which might or might not be 2 reportable, how do you handle that type of inclusion in 3 your reports? 4 A. I don't make a distinguish -- or I don't 5 distinguish that kind of difference in plants that are 6 producing product for long-term aging or something a 7 little bit different. You may recall a discussion we had 8 yesterday as well where I indicated that long-term storage 9 is specifically excluded from this. So those costs, I 10 don't want to try to capture. 11 But my understanding is that in these plants where 12 we're looking at identifying product that are candidates 13 for long-term storage, is that you plug or sample the 14 block, and you would have an expert make a determination 15 as to whether they think this will take on aging of as 16 many months as you are hoping to get on it. 17 Q. And in questions you answered that the volume of 18 cheese and whey reported in the 2023 report was 19 approximately 50% of the NDPSR surveyed volume. 20 Is that -- did I get that correct? 21 A. Yes. 22 Q. And -- 23 A. Excuse me, of the NASS volume. 24 Q. Of the NASS volume. Very good. 25 And 80 to 85% of the butter and powder, correct? 26 A. Yes. 27 MR. MILTNER: Your Honor, I have more questions, 28 based on how long we have gone, 20, 30 minutes. We have 3546 1 been going an hour and 45. I don't know if the court 2 reporter would like a break. 3 THE COURT: I was going to say, I think we have 4 been going an hour and 45 minutes. 5 Let's take a 15-minute break. Let's come back at 6 10:00. 7 (Whereupon, a break was taken.) 8 THE COURT: Back on the record. 9 Your witness, Mr. Miltner. 10 MR. MILTNER: Thank you, your Honor. 11 BY MR. MILTNER: 12 Q. You spent about an hour yesterday, thereabouts, 13 with Ms. Hancock going over Appendix A, which I 14 appreciated because I think it helped give us greater 15 insight about the methodology and what the plants were 16 looking at when they provided you information. 17 And I was wondering, as we went through that and 18 we walked through some of the questions and decision 19 points, did any survey participants reach out to you 20 through this process as they completed the survey about 21 any issues with the online model or the online system? 22 A. Absolutely. There were a few places where the 23 model couldn't proceed past pages with it, and part of the 24 reason was that, I did do this fairly quickly. I didn't 25 have the chance to harden the model and make sure that 26 people couldn't do things that they shouldn't have been 27 doing. And I think I gave the example yesterday of 28 peeking ahead, and when it did that, it did save pages 3547 1 without data in it and would cause a problem for that user 2 to not be able to enter that data until I purged the page. 3 So there were a couple of things. That wouldn't 4 happen normally if there had been enough time to 5 completely debug a model like that. 6 Q. About how frequently, once you've got a report 7 back from a plant, did you have to reach back out to get 8 additional clarification on an outlier or an allocation 9 question or anything else that caught your attention? 10 A. Maybe 25% of entries. Some folks I think managed 11 to enter data in a straightforward manner, and the glance 12 at the data, looking through the information didn't raise 13 any red flags for me. So, you know, it was data that I 14 would flag as being accepted. 15 Q. Along those lines, you have in your testimony the 16 line, "There are several key cross-checks in the data 17 collection." I don't know if you have talked about some 18 of those cross-checks. 19 Could you let us know what you are looking for or 20 what those cross-checks might be, please? 21 A. One of the cross-checks, and a primary one I 22 mentioned yesterday, and that was the use of doing this 23 mass balance calculation there. And the mass balance 24 calculation just simply says, have we accounted for all of 25 the components that we think came into the plant versus 26 all of those that were sold out of the plant. And I think 27 that that would be a fairly standard kind of accounting 28 process to take a look at that. 3548 1 And by the way, Ryan, this is one of the places 2 where you may be able to do some yield calculations, and 3 if you wanted to fine tune that, you could -- you could 4 take a look at more detail asking just a few other 5 questions that would let you get some yield parameter 6 data. 7 There are other places where there are 8 cross-checks. So, for example, you're identifying the 9 pounds of dairy product by a package size that you 10 produced over the course of a year, and then you are 11 entering data as to how many pounds of this cheddar 12 product you manufactured in the each of the 12 months. 13 And those numbers need to be the same. They aren't always 14 the same, and if they aren't, then I need to understand 15 why. You know, I mean, what -- what is there that's 16 different. And there are five or six of those kinds of 17 cross-checks throughout the data entry form. 18 Q. When you are doing the mass balance, do you get 19 down to a level that is as granular as in-plant losses 20 versus losses in shrink versus, you know, loss through 21 fines at a cheese plant, or things like that? 22 A. I don't ask for those values, but, you know, I -- 23 I would do much like you might in a federal audit where 24 you are looking at shrink or overage in a particular 25 plant. And to me, this needs to be within a pretty small 26 tolerance. And for me -- I know what your question may 27 be, what is a pretty small tolerance? 2% is something 28 that I look at. And if it's outside that range, then it 3549 1 seems to me that I have at least got to ask some 2 questions. 3 Q. If it were over that, you would be asking, did you 4 have a bunch of off-spec material or off-spec product that 5 you had to scrap, or something like that? 6 A. Something like that, yes. 7 Q. After you talk about the cross-checks, you say, 8 "Submitting intentionally deceptive costs would raise red 9 flags and prompt questions from me." 10 Did you have any instances where you thought 11 somebody was submitting intentionally deceptive costs? 12 A. I have never had that instance. I have had plenty 13 of times when I got data that were flagged from -- from my 14 screening processes, whether by these cross-checks or by 15 the fact that they might be a statistical outlier of the 16 body of samples that I have had by the time I'm done. And 17 most of the time, it would be an innocent clerical error 18 or omission, you know, of something that had not been 19 included that -- sometimes I think we have folks in the 20 plants that have very narrow job descriptions, you know, 21 that are assigned this kind of task and maybe haven't 22 fully understood what I'm trying to do with it. So it is 23 almost like they are trying to fill out a form that they 24 would fill out for a plant report or something to AMS 25 but -- and a clerical error is certainly something that 26 would be possible, and those are usually very obvious. 27 Q. So yesterday there was some cross-examination from 28 Dr. Bozic, and you noted -- I think in your statement as 3550 1 well -- an increased variance in the reported costs in the 2 2023 report compared to certainly the 2006 report, but 3 also I think you said compared to the 2021 report. 4 What do you ascribe that increased variance to? 5 A. Well, the biggest variance that I witnessed was 6 actually in the 2021 data. I mean it had been a while 7 since I had done a plant study like this, but I had done 8 many of them before. And in the 2021 data is when I first 9 saw that the range of plant observations was quite 10 different. I don't recall saying that 2023 looked a lot 11 different than 2021. If it was, it was maybe a little bit 12 different. But the variance is much more than it had been 13 in the past. 14 Q. Dr. Bozic also noted, and I think this was his 15 characterization, a bimodal distribution of the costs. 16 And I think you agreed that the distribution was 17 bimodal; is that correct? 18 A. Tended to see clusters at both ends of low cost 19 plants and high cost plants. And in the past, you 20 normally saw something that was more like a normal 21 distribution, where the body of respondents would be 22 somewhere in the middle, you know, and the really low cost 23 plants or really high cost plants tended to be few in 24 number. 25 Q. Not to dive too far back into the way back 26 machine, but that normal distribution is part of the 27 statistical analysis you did in 2006, right? 28 A. Yes. I used to, at least, talk about the 3551 1 qualities of the data that were there, some of those 2 statistical measures. 3 Q. So if we think about a normal distribution and a 4 bimodal distribution, and now USDA has to decide where do 5 we peg a Make Allowance, which has been at the weighted 6 average usually, historically. 7 . Given that the distribution is now bimodal, 8 should that inform USDA as to where it might want to draw 9 a line with respect to a changed Make Allowance? 10 A. A couple of things, Ryan. 11 First of all, when I have looked at what has been 12 recommended in a decision from AMS, it's never been 13 completely obvious to me how that decision was made 14 because it did not necessarily reflect the average cost 15 that I had reported or seen. You know, a lot of times I 16 would look at that and recognize that it's maybe typically 17 somewhere on the higher cost side of average but well 18 below the highest cost plants. 19 So conceptually, with what you are saying, if they 20 were to follow that, they might capture most or all of the 21 low cost plants as being -- having their operating costs 22 covered, well covered, and a few of the plants at the 23 higher end not covered, by just the Make Allowance alone. 24 And I think that part of the danger is that we 25 also assume that, well, then, many of these plants are not 26 covering their costs, they must be losing money, just 27 hemorrhaging money, you know. And I don't believe that's 28 the case. Why would we stay in business if that were 3552 1 happening? There are probably also plants that are 2 selling their product at the higher end of the NDPSR price 3 observations. 4 Q. Do you have any information as to the ages of the 5 facilities that participated in the 2023 report? 6 A. No. I haven't been collecting that data. I used 7 to collect that data years ago to at least get some idea 8 of when the last significant investment in the plant was 9 done. 10 I -- by the way, I haven't made that explicit in 11 the 2023 report, but I have usually done that in the past 12 to ask whether the plant had had any significant 13 interruptions in operation during the course of the year. 14 So if a plant had to be shut down for a week or a month or 15 something else, for whatever reasons, then this may not be 16 typical data that I'm receiving. 17 Q. That's an interesting point, which makes me 18 wonder, if a plant didn't operate consistently throughout 19 the year because it was a balancing facility, would that 20 also skew the data? 21 A. No, I try to capture -- well, I mean, it may 22 change that. I -- it's -- it's one of the conjectures 23 that we might have as to why a plant operates at higher 24 costs. They are carrying capacity that's not used all of 25 the time, and, you know, that is certainly an additional 26 cost. 27 I do try to ask enough questions to be able to 28 look at, at least, the monthly differences in product 3553 1 manufacturing out here, so that we can see whether the 2 plant was providing a good deal of seasonal balancing. I 3 don't look at inter-week balancing or collect data at that 4 level. 5 Q. If a plant is seasonally balancing, would their 6 fixed costs all be loaded on to the months in which they 7 are producing product, or would you exclude some of those 8 fixed costs? 9 A. No, their fixed costs would be allocated across 10 the pounds of product that they produced in the 12 months. 11 Q. In response to another question from Dr. Bozic, 12 you noted that you see plants being built where it doesn't 13 matter if the plant is pooled or not. 14 Do you recall that statement? 15 A. I think so. 16 Q. Can you elaborate on that observation, please? 17 A. Well, there are places in our current economic 18 environment and regulatory environment where there's 19 really not much money in the pool to be attractive by the 20 time you zoned out to the edge, perhaps, of the regulation 21 that you would have access to. And plants are making 22 investment decisions that really don't depend on this. 23 When I first started at Cornell University many 24 years ago, we might have a firm, a company, that would 25 realize, I have got customers for more product, I need to 26 make more product, I intend to build a plant, where should 27 I build it? And we would look at such things as, well, 28 where do we see growth in milk production happening in the 3554 1 country that makes sense to be a part of catching that 2 wave? 3 And then we didn't get those kind of questions for 4 a period of time. And this was when we found plants and 5 dairy farms sitting down to talk together and saying, 6 where do you want to make milk? You know, let's go put a 7 plant and, you know, produce dairy products where it 8 hadn't been before. 9 And so it is a different environment, and many of 10 those regions are not finding that there's enough money 11 available to be a pool plant, and farms appear to be okay 12 with that. 13 Q. Can you give examples of states or regions where 14 you see that occurring? 15 A. Well, in what was my own backyard, anyway, in the 16 Upper Midwest, the I-29 corridor, in places where we have 17 seen some significant plant investments, and not all those 18 plants are pooling all the milk. They are making 19 investments that aren't based on expected equalization 20 payments. 21 Q. There's been testimony about large cheese plants 22 being constructed in -- I think specifically was 23 references to Texas, maybe New Mexico, and other locations 24 in the west, maybe South Dakota as well. 25 Would you include those plants in that category? 26 A. Possibly. Some of that new capacity coming online 27 I think will -- I would expect would change the behavior 28 of milk movements across the South and Southwest where we 3555 1 have seen -- that being a milk supply for the Southeast, 2 it's quite deficit. If there's a nearby home, even if 3 it's at a lower price, perhaps below Federal Order 4 minimums in the area, could very well be the case that 5 farms decide not paying that hauling cost into the 6 Southeast would be an advantage. 7 Q. On page 7 of your statement I think you either 8 directly or tangentially address that, where you stated 9 there are "safety relief systems in Federal Orders that 10 are expected to be employed when the system isn't working 11 properly." And then you suggest that insufficient 12 Make Allowance might be a reason -- or might be a reason 13 plants depool to allow themselves of one of those safety 14 relief mechanisms. You identified that as one of the 15 safety relief mechanisms. 16 What are some others that you might refer to? 17 A. Well, depooling is one of the obvious ones, and 18 ones that we can -- we can see in there. 19 I had two or three in my mind, and they are 20 escaping me right now. Let me think about it for a 21 minute, Ryan. 22 But depooling is -- is I think one of the obvious 23 ones where a Federal Order system may not be operating as 24 hoped for, but the safety mechanism is there to use as 25 intended by, you know, the Federal Milk Marketing Order, 26 should it need to be. 27 Q. You provided an example of a plant that might be 28 located in the Southwest but supplying milk to the deficit 3556 1 area in the Southeast just a moment ago. 2 I want to give you another possible example and 3 ask your opinion on it. There was a producer here from 4 New Mexico last week who is located in Clovis, or 5 Portales, one of the two, but very close to a large cheese 6 plant. And if the most significant available Class I 7 market to that part of New Mexico would be Dallas, say, a 8 very substantial distance, several hundred miles, but 9 still within the same marketing order, if that producer or 10 that producer's cooperative determines that it is more 11 economically advantageous to sell milk to a Class III 12 plant locally rather than to haul within the order a great 13 distance at a lower return, would that speak to a need to 14 change Make Allowances or perhaps a need to look at 15 Class I differentials? 16 A. Could be a variety of reasons, and I think that 17 both of those would need to be looked at. If there is not 18 enough money in the pool to create an equalization payment 19 for milk that's moving in a certain direction, then that 20 may need to be looked at with Class I differentials. 21 And I think you have to be a little bit careful to 22 go back and ask yourself questions, and I'm not sure I 23 want to write policy, but what are the orders trying to 24 do? If they are there to help assure that we have at 25 least the opportunity for access to fluid milk, or fluid 26 milk plants, or convince milk to move in the direction 27 where it's most needed, I think our Class I differentials 28 have certainly attempted to do that. They may not be 3557 1 accurate for time and place, but -- but they are moving 2 milk in directions where it is most needed. 3 And by the way, I didn't say the differentials 4 were big enough to get it there to move in the direction 5 in which it's needed. 6 Q. Okay. We'll make sure that's noted in the record. 7 In your testimony, Exhibit 176, on page 6, in your 8 concluding comments, you state, "I would also suggest that 9 any parameters in the product price formulas, such as 10 Make Allowances and yield factors, have periodic 11 assessment to assure validity of price announcements." 12 A. I'm just about there. Page 6 you said? 13 Q. Page 6. Second to the last paragraph, last 14 sentence. 15 A. Okay. 16 Q. If we oversimplify the end product formula, it's 17 the product price minus the Make Allowance times the yield 18 gives you the milk value, right? 19 A. Yes. 20 Q. So the product prices -- 21 A. Oh -- 22 Q. Go ahead. 23 A. I just wanted to interrupt you for clarification. 24 You're jumping straight to milk prices as opposed to 25 component values. 26 Q. Fair. 27 A. Okay. 28 Q. Gives you the value of -- gives you the value of 3558 1 the component. That's a good clarification. 2 We're having discussions in this hearing about 3 what should or shouldn't be included in the product price, 4 but we know those numbers, once they are surveyed, they 5 are fresh, right? They are a week old, correct? 6 A. Yes. 7 Q. And we're addressing Make Allowances in this 8 hearing, so they will be hopefully less stale than 9 15 years. 10 If you have fresh price data and relatively 11 current allowance data, and your yield information is 12 still 10, 15, 20 years old in terms of its underlying 13 assumptions, does your component value that you end up 14 with reflect the real value of the milk? 15 A. I would stand by my statement here, you know, that 16 does indicate that we should look at all of the parameters 17 in those formulas. So there are yield factors, and in the 18 case of the somewhat more complicated protein value, we 19 have butterfat and protein interaction factors in there. 20 Q. I'd like to ask about your ROI assumptions in your 21 report. You mentioned you recall double-digit inflation 22 and the corresponding high interest rates that they 23 brought. And I vaguely remember as a young kid getting 24 about 8% on my savings account and not understanding why 25 that didn't continue on forever, until I figured out why. 26 A. Blew your whole retirement plan? 27 Q. At age four I already had it planned out, and here 28 I am working. 3559 1 Are you familiar with the five-year break-even 2 rate that the St. Louis Fed publishes? 3 A. I'm not sure that I'm explicitly familiar with it. 4 The five-year break-even rate? 5 Q. It is a -- 6 A. Is this a five-year bond treasury bill? 7 Q. No. It is the St. Louis Fed's projection of the 8 inflation rate over the next five years. 9 A. Oh. No, I'm not familiar with that. I guess that 10 my fumbling here should have been an indication. 11 Q. That's okay. We can jump over that. 12 But it's been 15 years since we have updated 13 Make Allowances, correct? 14 A. Yes. 15 Q. And hopefully we don't do that again, but if we 16 don't, and you -- your model, if it were to be adopted as 17 the basis for Make Allowances solely, wouldn't you be 18 creating a 15-year bond payment to milk processors with a 19 5.5% coupon? 20 A. No. I don't think that that's quite right. We 21 did talk with -- a little while ago with Nicole about, you 22 know, the imputed return on investment that we have in 23 plants. But it's more than just that when we're taking a 24 look at what the plants actually return out of this. 25 It -- individual plants don't receive the Make Allowance 26 or the portion of that that is there for the returns. 27 They buy products and milk at minimum prices and premiums 28 and discounts from time to time. They also sell products 3560 1 that are not just at that limit. So I'm not sure that you 2 can impute that that is the expected return for the plant 3 over a long period of time. 4 Q. If -- if the Moody's index, though, reverts back 5 to where it was in January of 2020, would we be 6 overstating the ROI factor, though? 7 A. For that long a period of time? Yes, you would. 8 I do think that these costs ought to be considered and 9 monitored. And costs do change, both up and down. So I 10 have been a strong proponent of that, let's capture that 11 on a more frequent basis. 12 Q. And likewise, if the Fed raises rates another 13 point and a half, then your model would -- 14 A. Would understate. 15 Q. -- would understate it, right? 16 A. Yes. 17 Q. Do you have any information from the participants, 18 or just from your industry knowledge, if cheese 19 manufacturers use ROI like this to gauge their 20 profitability rather than EBITA or margin on sales or 21 something like that? 22 A. I'm not aware that they do. 23 Q. If you look at Exhibit 177 on page 13. This is 24 the summary costs for butter processing. 25 A. Okay. 13 of 27. This is the processing program? 26 Exhibit 177. 27 Q. Is that IDFA Exhibit 1? 28 A. IDFA Exhibit 29. 3561 1 Q. I'm sorry. I was looking at the -- 178, my 2 apologies. It's only going to get worse as we add more 3 Exhibit Numbers. 4 A. Okay. Page what? 5 Q. 13. 6 A. Yep. 7 Q. I'm looking at -- well, first of all, the row of 8 low cost plants and the row of high cost plants, is that 9 the weighted average of that particular subset? 10 A. It is. 11 Q. Okay. So essentially you have set a 25th, 75th, 12 and 50th percentile, right, in those three rows? 13 A. No, not necessarily. 14 Q. No? 15 A. What's done is that I rank all of the plant 16 observations, and I'm looking for, you know, the median 17 break. In this case there were 13 plants. Okay? So the 18 break would have been seven on one side and six on the 19 other. And when I make that break, it's not that I throw 20 the extra plant into low cost or high cost, necessarily. 21 I look to see whether there's a natural break in the data 22 and whether it favors that plant going into low cost or 23 high cost. So that's my characteristic of it. If there 24 were 12 plants, it would be easy, six in, six out. And 25 then I do a weighted average calculation of each of these 26 cost centers across cost lines. 27 Q. Okay. If I look at the return on investment 28 column, the low cost plants show an ROI of $0.0269, the 3562 1 high cost plants show an ROI of $0.0618, and all plants 2 $0.0392. 3 Does it seem anomalous that the plants with the 4 highest costs get the highest return on investment? 5 A. Well, these are plants that would have reported a 6 higher market value for assets. And, you know, this is 7 one of the places -- I have tried to explain that a number 8 of times, that it is a bit of a decision on the part of 9 the plant what they think they could sell this plant for. 10 And I don't throw a plant out because they reported too 11 low a value or too high a value. I have never bought or 12 sold a butter plant in my life. I have some ideas about 13 what the cost of a new plant might be or perhaps even the 14 sale of existing plants. But if it's an outrageous number 15 that's returned to me, then I would at least ask about 16 that as to, is this justifiable? So this does reflect 17 self-reported value of assets. 18 Q. In your work at either Wisconsin or Cornell, did 19 you participate in any studies on farm profitability? 20 A. Yes. 21 Q. What would be a reasonable ROI for a dairy farm? 22 A. I have never heard a dairy farmer say that there 23 is one. It's always more. 24 Q. Another joke there that I'll tell you off the 25 record. 26 A. But we saw Dr. Wolf's report, and I think that the 27 body of data that he was reporting on that observed that 28 it was about 6.1%. 3563 1 Q. You think that is regularly achieved by dairy 2 farms? 3 A. I think that it is regularly exceeded by some 4 farms, and it's an aspiration for others. So there is 5 quite a difference, there's no question about it. I -- I 6 do know many farms that would simply say, if that was what 7 I expected long-term in the way of an investment return, 8 I'll look for a different industry or business to work in. 9 Q. I want to talk about degree of transformation, and 10 you have noted the 2023 report doesn't have the degree of 11 product transformation allocation method used. It is in 12 the 2021 report. 13 I believe that in presenting your statement you 14 testified that you believe the degree of transformation 15 analysis was valid; is that correct? 16 A. I do, as a concept. 17 Q. And I think there's a quote in your statement that 18 you favor the weighting of unallocated processing costs by 19 the degree of transformation of the products as well as 20 the pounds of milk solids processed. 21 A. Yes. 22 Q. Why do you favor that approach? 23 A. As I mentioned two or three times in testimony, I 24 have seen cases where we have had plants that maybe had 25 unusual sales opportunities, but they were selling quite a 26 bit of some of their components as very lightly 27 transformed products from the plant. That can skew the 28 products of interest, the ones that are highly 3564 1 transformed, like skim milk powder or others, to where you 2 undervalue those -- or those costs for producing that 3 product. 4 Q. Where you stated in your written testimony that 5 you favored the weighting of unallocated processing costs 6 by the degree of transformation of the products as well as 7 the pounds of milk solids processed, was there any change 8 between 2021 and 2023 about how you allocated costs across 9 the pounds of milk solids processed? 10 A. No. That's done exactly the same way. You look 11 at the total pounds of solids, the butterfat and the 12 nonfat solids in the products, and based on the percentage 13 in those products of total milk solids, then you would 14 allocate based on that. Not different between any of 15 these studies. 16 Q. You have answered a few questions about this topic 17 before, but I wanted to dive a little further. 18 You stated that industry participants had asked 19 for a return to the previous methodology without the 20 degree of transformation applied. And I think, not 21 written, you said there were groanings from the industry, 22 which I liked, and you gave some indication as to where 23 those objections came from. 24 Were any of those objections or requests to change 25 the methodology from members of IDFA? 26 A. Both. And when I say "both," I mean IDFA and 27 National Milk Producers Federation. 28 Q. When IDFA commissioned you and Wisconsin Cheese 3565 1 Makers commissioned you, did they request that the degree 2 of transformation analysis not be included? 3 A. Yes. 4 Q. And I think in response to a question from 5 Ms. Hancock, you said that those groanings might be 6 probably directed to the sample size and the data rather 7 than the methodology. 8 Did I get that correct? 9 A. Yes. Well, not the sample size, but the sample. 10 Q. Okay. 11 A. So in other words, different plants. 12 Q. So the e-mail we have from IDFA acknowledges that 13 they have commissioned you to perform this analysis in 14 order to set Make Allowances, and they asked that this 15 valid analysis not be done, correct? 16 A. They asked that I return to the analysis that I 17 had used in the past and which CDFA had used. 18 Q. Okay. So looking at the 2023 report, on page 10, 19 where you start making observations, and you begin with 20 nonfat dry milk, you state that reported costs per pound 21 declined by a little more than 6%, but comparing the 22 non-transformed weighted average in the 2023 study, 23 $0.275, with the non-transformed weighted average values 24 for the 2021 study, $0.2154, the nonfat dry milk 25 processing costs were increased by 12%. 26 So does that mean that if you maintained the 27 degree of transformation analysis, the reported costs 28 would have decreased from $0.293 to $0.233? 3566 1 A. Let me look. 2 Could you restate your question? 3 Q. Sure. I think I may have my math wrong. Let's 4 take the numbers out. 5 A. That makes math easier. 6 Q. It does. 7 If you maintained the degree of transformation 8 analysis, would the costs for producing nonfat dry milk 9 have shown a decrease between your 2021 report and the 10 2023 report? 11 A. Theoretically, the transformation number for 12 butter, as a good example, is smaller than the degree of 13 transformation for nonfat dry milk, and leaving that in 14 there would have increased the value for butter and 15 decreased the value for nonfat dry milk. 16 Q. So if we compared the transformed cost in 2021 to 17 the transformed cost in 2023 -- 18 A. Yeah. 19 Q. -- would that have shown a decrease? 20 A. It might have. And this is one of the reasons I 21 made the comment about the sample matters, because we had 22 different plants in the two studies doing different 23 things. So in one case if you had a few plants that had 24 lightly processed products in one of the studies, and you 25 didn't have those plants in the other study, then that 26 degree of transformation may look quite different. And 27 that's why I did try to indicate that I felt this was more 28 of a sample impact than it was just the degree of 3567 1 transformation itself. I think the math is fairly 2 straightforward in the transformation, but, again, it's 3 being obfuscated to some extent by the sample itself. 4 Q. The transformed cost in 2021 for cheddar was 5 $0.2476. Do you have a transformed value for cheddar 6 costs in 2023? 7 A. No, I don't. I didn't do those. As I mentioned, 8 I was asked not to and to return to the previous study, so 9 I didn't bother doing that. 10 And that wasn't because I was trying to hide 11 anything, Ryan. I was just not looking for any more work. 12 Q. I would not have expected that you would have done 13 so. 14 So the whey processing costs that you report are 15 $0.3361. Now, for the last few months the NDPSR survey 16 dry whey prices have averaged around $0.26, and the range 17 of costs you report are $0.2848 to $0.3952. 18 So if the value of whey in the market is $0.26 and 19 the make is $0.336, why in the world are these plants 20 making whey? 21 A. I would assume that disposal costs or the lack of 22 the equipment to make other whey products is not available 23 to the plants, or this is simply viewed as being a 24 short-term phenomenon. But we have seen that before where 25 the implied value is negative. I mean this isn't -- would 26 not be the first time. We have seen that happen before. 27 Q. Do you -- do you look at the weekly NDPSR reports? 28 A. Not anymore. 3568 1 Q. How long has it been since you looked at them? 2 A. It's been a little while. I mean, I'm roughly 3 aware of what's happening in the marketplace, but really 4 not engaged like I was a while ago. 5 Q. Are you roughly aware of any reports that whey 6 production is declining, dry whey production is declining? 7 A. Not recently. Dry whey production has declined a 8 good deal over the past many years. It's transferred to 9 higher protein whey products. 10 Q. Okay. So just to recap a couple things. Your 11 2023 report had a smaller set of observations than the 12 2021 report, correct? 13 A. Yes. 14 Q. And as you noted often, the sample matters in 15 terms of the usefulness of the data reported, correct? Or 16 the conclusions reached perhaps is more accurate? 17 A. Correct. 18 Q. Of those plants that did report, fully two-thirds 19 didn't participate in the 2021 report, correct? 20 A. About that, yes. 21 Q. Those new plants that did participate were 22 identified by an e-mail from IDFA's CEO to its membership 23 and maybe an e-mail from you, correct? 24 A. Yes. 25 Q. And the 2023 report does not include the degree of 26 transformation factor which you believe is valid, correct? 27 A. Correct. 28 Q. And you abandoned using that at the request of 3569 1 IDFA who commissioned you to perform the research, right? 2 A. It was the request, yes. 3 MR. MILTNER: Thank you. That's all I have. 4 Could we move the admission of the e-mail that we 5 marked as 179? 6 THE COURT: Any objections? 7 MR. HILL: I do. 8 THE COURT: You object? 9 MR. HILL: I am objecting just because there's no 10 one to authenticate this document. The doctor has said -- 11 Dr. Stephenson has said that he never saw this e-mail 12 before today. There's no witness who can verify that this 13 document is authentic. 14 MR. MILTNER: That's okay. Mr. Allen will be here 15 later in the hearing, and we can have him authenticate and 16 more for admission then. 17 THE COURT: Let's do that. Let's -- everyone help 18 me remember to do that. 19 MR. MILTNER: I won't forget. 20 MS. HANCOCK: Your Honor -- 21 THE COURT: I can imagine what you are going to 22 say, Ms. Hancock. 23 MS. HANCOCK: You can imagine? Well, maybe I 24 shouldn't say it, but I'm going to say it because I just 25 want to make sure that we're treating equitably all of the 26 documents that we're offering for admission into the 27 record. And we have other exhibits that attorneys just 28 made up and weren't even accurate, and we allowed them 3570 1 into the record as exhibits. And my concern here is that 2 now we're applying a higher standard to an e-mail that 3 this witness has testified about, and that seemed to be 4 the only standard we applied previously. 5 And for the integrity of our record, I feel like 6 we have to apply the same standard for all of our exhibits 7 that are admitted, whether, you know -- just so that one's 8 not weighted more than the other, because I have serious 9 concerns about the integrity of previously admitted 10 exhibits and the value that they have. 11 And under -- under your authority, we have to have 12 you actually state on the record that you are allowing our 13 arguments to be included in the transcript in order to 14 maintain these comments in the transcripts, because I 15 think the rule provides that only your ruling is in the 16 transcript, not our actual arguments. 17 THE COURT: No, I'm ruling that your arguments can 18 be in the transcript. Let's make that clear, yes. 19 MS. HANCOCK: I appreciate that because I think 20 that's important for our record. 21 But I have serious concerns about not applying the 22 same standard across all of our exhibits. 23 THE COURT: Yeah, I hear you, and that's why I -- 24 I think you were going to say. I mean, I think the 25 ultimate determination of what's valid and appropriate 26 evidence lies with AMS or, you know, the Secretary acting 27 under that delegated authority. I want -- I think I want 28 any of these things to stay in the record so that they can 3571 1 consider that. I've thought about making it an offer of 2 proof for instance. 3 I think, when -- I mean, these are documents used 4 in aid of cross-examination. You can show a witness 5 virtually anything to refresh their recollection. And as 6 I understand it, if someone presents something and goes 7 through it and says, this number came from here, that 8 number came from here, would you agree, and then you 9 have -- I think having it in the document does not really 10 change whether you asked it orally or not. And I think -- 11 you know, I think that that's basically okay. 12 Something like this -- I guess the mischief I see 13 is when something comes into the record, it is sort of 14 there for all purposes, and this is an e-mail with all 15 sorts of things in it, names and things like that. So I 16 think it really should be authenticated. 17 But at the end of the day, I mean, I'm happy to 18 make my ruling on it, and then it is really up to AMS. In 19 this case AMS objected to it coming in without being 20 authenticated. 21 So, yeah. I mean, if you want to revisit the 22 earlier one. I'm guessing you don't. 23 MS. HANCOCK: I'm happy to revisit the earlier 24 one. I don't know if you can unring a bell though, so -- 25 I will say -- I will note just for the record, 26 USDA and National Milk both objected to the -- I don't 27 remember the exhibit number, but it's the one that 28 Mr. Rosenbaum created himself and asked a witness -- asked 3572 1 Dr. Vitaliano -- or someone, I don't actually remember 2 who -- to talk about the document. And I think that we 3 both objected on similar grounds, so -- 4 THE COURT: Yeah, I do remember that. And I have 5 had a chance to, you know, think harder about that and do 6 some research. And frankly, I guess I would ask the -- 7 actually Mr. Rosenbaum is standing up. I do want to hear 8 some argument on this because I think it's a legit matter 9 of practice and procedure for the this hearing. 10 What -- we have a big history of these cases in 11 the past, and I want to stay consistent with what we have 12 done in the past here. 13 Has this come up in previous hearings that people 14 have been involved in? What were the rulings then? I 15 mean, I haven't had the ability to research that. 16 I'll let Mr. Rosenbaum -- should we let you speak 17 first, Mr. Rosenbaum? I want to hear further on this. 18 MR. ROSENBAUM: Well, partly I'm standing out of 19 the motivation that we have 175 exhibits in the record 20 now. I really don't want -- I do not think it behooves us 21 to revisit their admission. You have made your decisions, 22 and if people want to challenge the reliability, they can 23 do that in their briefs. 24 And I think obviously, yeah, we are going to be 25 putting on IDFA witnesses. I mean, Mr. Miltner will have 26 plenty of opportunities to ask questions regarding the 27 e-mail he's put in, and probably has his own witness who 28 can do that as well. This is not a material -- that 3573 1 particular document does not present a material challenge 2 to counsel, and he's not objected to it, and I just feel 3 we should continue on. 4 THE COURT: Yeah, move on? 5 MR. ROSENBAUM: Move on, yes. 6 THE COURT: Well, I think you are right about 7 that. I don't think there's any -- I don't need to make a 8 decision on whether this document is admissible right now 9 because Mr. Miltner's offered somebody to authenticate it, 10 and I think when he does, it's perfectly admissible. 11 MR. ROSENBAUM: I mean, no one had any questions 12 about the nature of the document that was used with 13 Dr. Vitaliano, and your Honor explained why he found it 14 helpful personally. I thought it was helpful. I think it 15 is helpful in the record. In any event, I think that's 16 water under the bridge at this point. And I think we 17 should allow Dr. Stephenson to complete his testimony and 18 move on. 19 THE COURT: Well, I can live with that. And, you 20 know, I do the best I can here. Wouldn't be the first 21 time I have been reversed. And, you know, I hate to say 22 this on the record, but they do say consistency is the 23 hobgoblin of small minds, right? I will try my very best 24 to be consistent. 25 And I do not -- we do not have the transcript yet. 26 I suspect it would behoove anyone using a document they 27 have created to -- in that case, the document you used to 28 cross was not a document that was being presented as, you 3574 1 know, this is a piece of evidence that somebody else 2 created, or here are some numbers, you know. And I think 3 you probably went through and authenticated what was in 4 there, you know, and you proved at least far enough 5 that -- Ms. Hancock's shaking her head, so I take it she 6 disagrees. 7 Mr. English. 8 MR. ENGLISH: So without being able to provide a 9 specific example over the last decades that I have done 10 these hearings, I can assure your Honor and others that 11 there have been demonstrative exhibits. 12 I think this is a demonstrative document, and I 13 thought your ruling did not say that it was being admitted 14 for the truth of its purposes. It was there, and since it 15 had been referenced, it was appropriate to be along with 16 the record for that purpose. 17 To the extent that there was examination that 18 suggested, maybe, I don't buy into the argument that it 19 was discredited, but to the extent that was there, there 20 was an examination, and I think your ruling was correct 21 then. 22 I think, frankly, we're now spending a lot of time 23 on this issue when the reality is that the document that's 24 been marked as Exhibit 179, you know, is going to come in 25 at some point, whether Mr. Allen is here Friday, or thanks 26 partly to this argument, next Monday, it's going to 27 happen. 28 And besides that, you know, it leaves Mr. Brown, 3575 1 who is CC'd on the e-mail from Mr. Dykes, will be 2 testifying sometime this week. 3 So I think, you know, we're spending a lot of time 4 on that document that we don't need to spend, and I am 5 perfectly happy to move on to further examination of 6 Dr. Stephenson. 7 THE COURT: Well -- 8 MR. ENGLISH: I won't cut off Ms. Hancock, 9 obviously if she wants to respond. But I think at some 10 point if -- and I also agree with Mr. Rosenbaum, if we're 11 going to start revisiting one exhibit, then we almost have 12 to revisit others, and I think we may be here a long time. 13 THE COURT: Well, I think everything else has been 14 admitted, and I don't remember -- 15 MR. ENGLISH: Well, but if that ruling had gone 16 differently, we may have had different reactions to the 17 exhibits that came before, your Honor. 18 THE COURT: Well, that's true. 19 MR. ENGLISH: So we -- you know, in essence, since 20 that exhibit, there have been other documents. I just -- 21 I think that it's admitted for -- I think the typical 22 answer is it's admitted for the weight the Secretary will 23 give it. 24 THE COURT: Well, yeah. And I do feel that way, 25 which of course that could be true of anything I admit. 26 But that one in particular, yes, because I'm not the 27 ultimate decider here. 28 The reason I wanted to have this discussion on the 3576 1 record is to provide guidance going forward to people, and 2 Ms. Hancock raised a legitimate concern about consistency, 3 although I think we are being consistent. We're pretty 4 much letting in everything, it looks like to me. And 5 then -- but -- and I don't -- I -- you know, I don't want 6 to -- I don't want to do that. I mean, I don't want to 7 make a ruling in the first instance to be of help, I 8 guess, to the Administrator. 9 MS. HANCOCK: Your Honor, it is really truly only 10 the consistency. I understand that USDA will give it its 11 weight. 12 Two things I want to note. One, the document that 13 I'm referring back to -- and I'm sorry, I don't have the 14 exhibit number -- but my -- you know, one of my largest 15 concerns is it was created by an attorney for a -- you 16 know, that's representing a party here. It was titled 17 that it is an agreement between IDFA and National Milk. 18 No agreement has ever been made. 19 Mr. English just said, well, it's not being 20 offered for the truth of the matter asserted. It doesn't 21 matter, that's an evidentiary ruling based on a hearsay 22 determination standard. Once it's admitted, it can be 23 used for anything. 24 And I am not just concerned for the integrity of 25 this record, but what we see is, is that parties in 26 subsequent hearings will pull documents and portions of a 27 transcript out of a record and use it as binding precedent 28 going forward. 3577 1 So let's fast forward 15 years to another hearing 2 in the future, and somebody now has a document in the 3 record that's been admitted as evidence. And once it's 4 admitted, it's not admitted for a limited purpose, it's 5 being admitted as -- as evidence of the record, and it's 6 titled that it's an agreement between two parties. 7 Now, certainly somebody can try and put it into 8 context, but you're somewhat beholden to the room having 9 enough memory of the event and the context within which it 10 was admitted to put it back into context. 11 So that's my concern with the earlier ruling. 12 THE COURT: Uh-huh. 13 MS. HANCOCK: My concern now, for purposes of 14 this, is that we're now saying, hey, you have to go 15 authenticate this, Mr. Miltner, before we're going to 16 admit it. 17 Well, there's no way that any of our witnesses can 18 authenticate what Mr. Rosenbaum created and put into the 19 record, so we had zero authentication for -- for that 20 document, but now we're saying, you have to go 21 authenticate it for what it is now. And, now, we -- you 22 can't apply the same standard unless you just let it in. 23 THE COURT: Well, I'm letting it in. I mean, 24 we'll authenticate it right now. 25 You produced that document, didn't you, Mr. 26 Rosenbaum? You created that document, the previous 27 document? 28 MS. HANCOCK: He's not a witness. 3578 1 THE COURT: Well, we can swear him in and make him 2 a witness. I don't think there's a problem with that. He 3 admits that he took numbers from National Milk and numbers 4 from his own client and was trying to explore, like, what 5 are the differences here. 6 MS. HANCOCK: I understand -- 7 THE COURT: He asked the witness about that. Is 8 this my number? Is this your number? If the witness 9 said, I don't know your number, he said, well assume for 10 purposes of questioning, now, if this were my number would 11 that be a big difference? Do you have a difference with 12 that? 13 I mean, and as you said, I mean, you would be 14 willing to -- to accept the narrative cross-examination of 15 his words, and I don't see a big difference between that 16 being in paper. I don't think anyone's really going to 17 assert that National Milk and Rosenbaum's client have an 18 agreement just because he -- a document that the lawyer 19 created says that up at the top. 20 MS. HANCOCK: I mean, I do, I think that you 21 can't -- my point is that you can't authenticate it 22 through an attorney who represents a party. You can 23 authenticate it through the witnesses that testify about a 24 document. It was created by an attorney and given to a 25 witness. 26 My only point here is to raise the consistency and 27 treatment of the exhibits for authentication purposes, and 28 if Mr. Rosenbaum can have a document that's admitted that 3579 1 he created, that's titled that it's an agreement, when 2 there is no agreement that anybody has ever testified to 3 that, then I think that that is a different standard than 4 what we are applying to the -- to the e-mail that 5 Mr. Miltner just put before the witness. 6 And as I recall, the standard that we applied 7 previously was if the witness testified about the 8 document, and it's in the record, and there's an exchange 9 about that document, that you wanted the document to be 10 admitted so that the record was complete in the context of 11 that examination. 12 I just want the treatment to be the same for 13 exhibits. 14 THE COURT: Well, all right, Ms. Hancock, I hear 15 you. But there is a massive difference between an e-mail 16 between other parties that goes out and whether that's 17 authenticated or not, and something that a lawyer says, "I 18 put together these numbers to have a reasonable reference 19 to consider things." 20 The -- an e-mail, there's nobody that -- there was 21 nobody here to say this actually went out, was actually 22 received, or whatever else like that. There's no question 23 about what the other document is. And I'm sure that's 24 authenticated in the right sense. I think it's 25 appropriate to have it in the record, and I think we'll 26 stay consistent. 27 If there's a third-party document, somebody else 28 signs something, that is something that -- that needs 3580 1 authentication. 2 And, you know, I would, I guess forewarn folks 3 cross-examining that they need to, you know, authenticate 4 what's in these documents. 5 And I think this -- I think Mr. English and others 6 have talked to this. I don't think there's going to be a 7 problem about an exhibit -- one, there's going to be a 8 problem, but we're going to have somebody go jump through 9 all the hoops on that. 10 Mr. Miltner -- Mr. Hill. 11 MR. HILL: I am sorry I brought us down this 12 rabbit hole. So I do want to say that Dr. Stephenson does 13 have a limitation on his time, so whatever the case may 14 be, I would like to move forward with him to get him out 15 of here as quickly as possible. 16 THE COURT: That's fair. 17 Real quick, Mr. Miltner. 18 MR. MILTNER: Very quickly. I want to say that I 19 agree with much of what Ms. Hancock is saying. The fact 20 that I have a witness coming that was the recipient of the 21 e-mail and he can deal with this authentication issue is 22 why I said we'll deal with it later. Had that not been 23 the case, I think there was enough here with the practice 24 of this hearing and the way we have admitted evidence, as 25 Mr. English noted, that we tend to admit exhibits unless 26 there's a very, very valid reason for excluding it, such 27 as confidentiality or real questions about its providence, 28 and then we allow the Department to ascribe the weight to 3581 1 which the Secretary believes it's entitled. 2 And I think that under that standard, we certainly 3 could admit the e-mail now, but I don't want to belabor 4 this. We will have someone to authenticate it. But I 5 also wanted to make sure that I did get on record that I 6 do believe it is admissible at this point. So, thank you. 7 THE COURT: Very good. And everyone's objections 8 are preserved. 9 Let's -- for Dr. Stephenson's purpose, let's go 10 ahead. I do think the discussion was useful as guidance 11 to the questioning that other people may do. 12 And moreover, I suspect that you were right to 13 begin with, Ms. Hancock, that Mr. Rosenbaum probably got 14 in everything through back-and-forth questions and answers 15 that was in that document, so I don't think there's harm 16 or foul either way on that. And I do think it -- 17 This is a bad question to ask on the record, but I 18 will right now. I mean, I'm not sure what happens to 19 documents that are excluded. Do they go with the record 20 to the -- 21 MR. ENGLISH: That's my understanding, your Honor. 22 They've always gone -- they're definitely the documents 23 excluded, they are just not admitted, but they go along 24 anyway. 25 THE COURT: It may be an automatic offer of proof 26 anyway. I forget what the rules provide. 27 MR. HILL: Yeah, they get posted just like all the 28 other exhibits. 3582 1 THE COURT: Anyway, thank you, everyone. It helps 2 me refine my thinking, and glad to have everyone's 3 thoughts. 4 MR. ENGLISH: Your Honor, may I? 5 THE COURT: Yes, sir, your witness. 6 MR. ENGLISH: Chip English for the Milk Innovation 7 Group. 8 CROSS-EXAMINATION 9 BY MR. ENGLISH: 10 Q. Dr. Stephenson, good morning. I actually was 11 going to end with a question about Exhibit 179, but maybe 12 to close the door, I'll start with it. 13 The exhibit in the last paragraph references a 14 submitting data for the survey deadline of, in bold, 15 April 14th, 2023. 16 A. Which document are you -- 17 Q. This is Exhibit 179. This is the document we just 18 spent 20-some minutes talking about, which is, you know, 19 the e-mail purportedly from Mr. Allen back to Dr. Dykes. 20 And there's a reference to the deadline for submitting 21 data for the survey being April 14th, 2023, right? So 22 orient ourselves. 23 Did that deadline end up being a hard deadline? 24 A. No. I moved the goal posts a few times because 25 people requested that, could we please enter data within 26 the next two weeks, and that happened a few different 27 times. 28 Q. So going back quickly to a question asked by 3583 1 Mr. Miltner. If you were asked about confidentiality for 2 the 2023 study, you supplied, entities that requested, a 3 non-disclosure agreement, correct? 4 A. Correct. 5 Q. Given the public nature of IDFA's request to you 6 to update the 2019 study, or 2021 published, do you have 7 any view as to whether people in the industry knew about 8 your study and any facility could have participated if 9 they wanted to? 10 A. I believe that most of the participants knew that 11 the study was being updated -- or most of the previous 12 participants knew that it was being updated, and would 13 have been welcome to participate. 14 Q. What about other people who did not participate in 15 the earlier study, would they likely have known based upon 16 the invitation and the dairy industry's predilection to 17 share information? 18 A. There certainly would have been opportunities for 19 them to have understood and heard about that and 20 participated. I don't know how broadly it was. I don't 21 recall whether this was something that was picked up by 22 popular press or not, as has happened certainly in the 23 past. 24 Q. So at the time IDFA sent out this invitation in 25 February, mid-February of this year, to your knowledge, 26 was DFA a member of IDFA at that time? 27 A. I don't recall. I do remember hearing that they 28 had pulled out from their support of the organization, but 3584 1 I don't remember what the exact date and timeline was. 2 Q. But do you know for a fact that DFA did know about 3 the study? Did you discuss it with them at all? 4 A. I had some discussion with people about the study. 5 But, again, I don't recall if this was before or after 6 this e-mail. 7 Q. So going to the survey and your report as to 8 cheese, the plants that participated had relatively higher 9 volumes. 10 What is the implication for smaller volume plants' 11 ability to influence the study's results? 12 A. Well, their data would be included like any other 13 plant's data in the study. I will say that as I'm 14 reporting weighted average values on here, a plant that 15 produces smaller volumes will have a smaller impact on the 16 results than a larger volume plant would. That's just the 17 math. 18 MR. ENGLISH: That's all I have. Thank you. 19 THE COURT: Further questions for Dr. Stephenson? 20 AMS. 21 CROSS-EXAMINATION 22 BY MS. TAYLOR 23 Q. Good morning. 24 A. Good morning, Ms. Taylor. 25 Q. Kind of hard to believe there's even still 26 questions left to ask. 27 A. Hopefully there are still answers available. 28 Q. I hope so, too. 3585 1 I want to start -- I'm going to try to do this in 2 some semblance of a logical order. So let's start with 3 Exhibit 158, which is your December 2021 study. 4 A. Okay. 5 Q. I'm going to ask some questions that might be a 6 little repetitive, but I think for clarity of the record 7 would be helpful. 8 On page 4 when you talk about your products 9 targeted, you list cheddar cheese in 40-pound blocks, 10 640-pound blocks, and 500-pound barrels. 11 And if I remember from your discussions with 12 Mr. Miltner, costs for all those products were included, 13 except for packaging costs for 640s? 14 A. That is correct. 15 Q. Okay. And you also -- 16 A. And I didn't report packaging costs for barrels in 17 here. I have those, but they are not reported in the 18 table. 19 Q. Okay. So -- okay. Thank you for that 20 clarification. 21 And then on the whey, you have costs in here for 22 dry whey and WPCs, which you did collect, but the results 23 only reflect dry whey plants. 24 A. Could you restate that? I didn't quite clearly 25 hear it. 26 Q. Sure. For the whey category, you have in here you 27 collected information -- or you targeted cost data on dry 28 whey products and WPC products. 3586 1 A. That's correct. 2 Q. But the results are only for dry whey showing -- 3 A. That's the only thing that's reported. 4 Q. Okay. And that same characterization of the 5 products and which costs are reported applies to both the 6 '21 study and the 2023 study? 7 A. That's correct. 8 Q. Okay. On the top, in page 7, and you discuss your 9 transformation factors, which I know long before this 10 hearing was ever thought about we had some discussions of 11 those with you when you were working on this survey. 12 But one question is, you -- well, first, you only 13 used this transformation value on plants that could not 14 directly allocate their costs; is that correct? 15 A. Yes. If we had plants that were able to directly 16 allocate their costs, then those are the first things that 17 I take. 18 If they can indirectly allocate their costs, so, 19 for example, they have allocated the cost to cheese 20 products, in large quotes here, that might have included 21 cheddar cheese and other non-reportable NDS- -- 22 Q. NDPSR? 23 A. That's it. 24 -- then I would have performed an allocation 25 between cheese products, but it would have been restricted 26 to cheese products even if other products were produced in 27 the plant. So transformation values only occur when I'm 28 trying to allocate costs. 3587 1 Q. Can you talk about, if you can recollect, how 2 often you actually had to employ this transformation 3 value? 4 A. With some frequency. I mean, some plants will 5 report occasionally, just kind of like a bottom line 6 unallocated number for things, and at that point in time, 7 the allocation has to take place across it. 8 Most plants will have some degree of allocation 9 that they have actually done, but not complete. And I 10 don't really expect that to happen. There are always 11 going to be some costs that I need to allocate. 12 Q. Okay. And so would it be fair to say, maybe, 13 like, general and administrative costs might -- you might 14 use it that way, which were less, maybe have a -- there's 15 less ability to allocate between products? 16 A. I do have -- if you remember in that last input 17 screen of the general ledger, that there are at least 18 product categories, such as cheese or powders or butter, 19 where you can provide some degree of disaggregation of 20 your costs. I would say the majority of plants don't. 21 They will simply provide me unallocated costs on that 22 ledger page, but there are more than a few plants, 23 probably 40% or something like that, a rough estimate on 24 my part, that make some degree of allocation across their 25 product categories. 26 Q. Thank you. 27 I want to turn to the results which start on 28 page 11, and this is for nonfat dry milk. And what I took 3588 1 from your conversation earlier to -- well, first let me 2 ask this. You discussed how the 2023 results were more 3 bimodal distribution and less of a normal distribution on 4 the bell curve; is that correct? 5 A. Correct. And I had some of that same evidence for 6 the 2021 study. 7 Q. Okay. 8 A. And that comment was primarily made going back to 9 the earlier 2006 and '07 studies. 10 Q. Okay. So you saw the similarities between '21 and 11 '23? 12 A. Similarity, but differences. 13 Q. Okay. And so for your low cost and your high cost 14 divisions, if I'm looking at an N of 27, am I going to 15 always assume that the division is around 50%, or could it 16 be that -- oh, I guess that's my first question. 17 A. Yeah. I mean, I wouldn't have had 20 and 7, for 18 example, in a division here. It would have been 13 and 19 14. But where that additional plant falls, you know, 20 between the low cost or high cost, I would look to see 21 whether their price value has a natural break that is more 22 closely aligned with low cost plants or high cost plants. 23 Q. So you didn't observe where, because of the way 24 it's distributed here, that you had an uneven break? 25 Like, say, 15 happened to be around one end, and 12 around 26 the other end, and so the natural break was between 15 and 27 12 instead of between 15 and 14, if that adds up right? 28 A. No. When I start to do something like that, there 3589 1 are some statistical measures that might have been used to 2 look for natural breaks. But I didn't do that because 3 that, to me, began to feel a little bit more like I was 4 trying to impose my idea of what should be reported as 5 data or difficult to explain to others. So I tried to 6 break always at near the 50% -- 7 Q. Okay. Thank you. 8 A. -- in terms of number of plants. 9 Q. Okay. And then so if we're looking at the low 10 cost product pounds, that is the weighted average of the 11 number of plants that fall in that low cost category. 12 A. Correct. 13 Q. And then the same thing for the high cost. And 14 then the all plant product pounds is the weighted average 15 of all the plants together, or is that the average of the 16 two numbers? 17 A. Yeah, it's actually not the weighted average, it's 18 the average. Because it's all of the other values that 19 are weighted by those product pounds that are weighted 20 average values. 21 Q. Okay. And I wanted to turn to page 14 of this 22 exhibit for dry whey. Under the high cost plants -- and I 23 don't know if this is a typo -- it has a general and 24 administrative cost of zero, but you still have an average 25 of .0015, so I'm just wondering what the missing number 26 is. 27 A. Yeah. This was a case where, you know, sample, 28 again, imposed itself, and there were not very many 3590 1 plants, there were four operations that would have been in 2 that high cost. And this was a case, I do recall, where 3 actually if I carried that out to five or six decimal 4 points, there would have been a value there. But we had 5 many plants for the general and administrative numbers 6 where those were not supplied or broken out. 7 Recall, those are not all of those general ledger 8 numbers, they are specific data values from that ledger 9 page and from the labor page. 10 Q. Okay. 11 A. An example for that would have been clerical 12 values, it would have been the plant superintendent 13 values, and we may have had plants here that did not 14 report separate superintendent or plant manager values. 15 Q. So there are instances in the data where they 16 could have reported labor, utilities, other -- some of the 17 cost breakouts, but not all? 18 A. Yes. 19 Q. And if you did that, if they did that, they left 20 some cost categories blank, you didn't remove them from 21 the survey altogether? 22 A. No. I didn't remove them altogether. And there 23 are a few plants that don't report some of those values on 24 there that where -- let me give you a good example of 25 that. 26 On the return to investment category that's here, 27 if we had one plant out of the four that might have fallen 28 into one of these categories, it simply said, I can't come 3591 1 up with a market value of my plant assets. And that did 2 happen. Then, when I'm calculating this as a weighted 3 return value, that's not a zero value for those assets, 4 it's a non-number value, so it's excluded. It would be 5 weighted by the three observations that we have there. 6 Q. Okay. So then same thing for general 7 administrative -- 8 A. Yes. 9 Q. -- it's just not in the -- it's not in the 10 equation at all? 11 A. Exactly. 12 Q. Yeah. Okay. 13 For the plants for both '21 and '23 studies, I 14 think in your description you talk about how they 15 represent kind of, like, geogra- -- their -- word this in 16 a way that makes sense -- they are representative of the 17 geography of the entire United States, so they are all 18 over? 19 A. We had plants that were all over. I wasn't 20 targeting that like I have done in previous studies, but 21 we had representation from all parts of the country. 22 Southeast is a place where we didn't have any 23 observations. 24 Q. Okay. Excluding the Southeast part of the 25 country, were there any other parts of the country that 26 might have been overrepresented or underrepresented in 27 the -- 28 A. No. Actually, not. We had plants from all 3592 1 portions of the country. And I'll -- I guess you could 2 define regions into smaller and smaller geographies, but 3 for what I would consider major regions like Northeast, 4 Middle Atlantic, Upper Midwest, far West, Southwest, we 5 had representative plants in all of those areas. 6 Q. Okay. I have questions on like four different 7 documents, so -- 8 A. Okay. 9 Q. -- I'm trying not to go all over the place. 10 So for the data that you collected, I think most 11 of it was for -- occurred in 2022; is that correct? They 12 didn't have to do calendar year, they could have done 13 fiscal year, but -- 14 A. They could have. I don't recall specifically. It 15 was only either one or maybe two plants that did something 16 different than calendar year 2022 data. 17 Q. Okay. 18 A. And those fiscal years tended to be pretty close 19 to calendar year '22. 20 Q. Okay. I just want to make sure that the record is 21 clear because -- let's turn to -- let me -- I'm on your 22 statement, which is Exhibit 176, on page 5. 23 A. Okay. 24 Q. And you have in the table the 2008 25 Make Allowances, which are current levels, then your 26 results of the three surveys that you have done, and the 27 percentage changes are of '19 or '22, changed from 2006. 28 But I want to make sure it's clear that none of 3593 1 your survey results included any type of marketing costs 2 factor; is that correct? 3 A. No. There's no questions that are asked that 4 would comprise what I might consider marketing or sales 5 costs to be, and I do actually try to explicitly make sure 6 that those aren't included. 7 Q. And do you know if there's a marketing cost factor 8 included in the 2008 Make Allowance numbers that are 9 currently adopted? 10 A. The 2008 numbers? I -- I don't know, as I have 11 testified before, what had actually gone into the 12 discussion about what -- what final numbers came out for 13 Make Allowances. 14 I can tell you that the data that I had used back 15 in that time period and reported on 2007, I guess, did not 16 include marketing costs. 17 Q. Okay. So I'll ask a question of my lovely 18 technical people next to me. I did one set of math and it 19 worked, and then the other set apparently doesn't work. 20 The way they are looking at the -- the percentage 21 change you have listed between '06 and 2019, that column, 22 I think the way they have calculated it, it looks like you 23 actually did the percentage change over the current 24 Make Allowance, which is the first column. 25 But you did intend for it to be a change from the 26 2006 if we wanted to redo that math? 27 A. That was my intention was to compare it to my 28 results, not the Make Allowance results. 3594 1 Q. Okay. Thank you. 2 A. So if I did the wrong thing, apologies. 3 Q. We're just here to make sure we are all clear on 4 what we wanted to do. 5 Let's see. On the next page, for observations, 6 where you kind of give a general summary of what you saw, 7 and I want to summarize and make sure I have got this 8 right. 9 So for nonfat dry milk between 2019 and 2022, you 10 had less plants reporting, but they were bigger plants 11 that did report? 12 A. Correct. 13 Q. And -- and would you say -- as I'm looking at the 14 numbers between your '21 study and your '23 study, they 15 actually declined a little bit. In '21 you reported 16 $0.2933 cents, and in 2023 you've reported $0.2750. 17 So would you attribute that just to the larger 18 volume? 19 A. As I have mentioned a few times, we oftentimes see 20 lower costs in larger plants. But that's not an absolute 21 categorization, that sometimes small plants are very cost 22 competitive. 23 Q. Okay. I wanted to talk about the butter results. 24 And I know you discussed this some, and I think I might -- 25 and I think I'm being duplicative, I apologize, because I 26 missed half your answer when you were discussing it 27 earlier. 28 But for the butter, you had generally the same 3595 1 number of plants reporting, and it averages out to a 2 similar volume, but you think the difference is just a -- 3 it was just basically a different set of plants that 4 reported, a significantly different set of plants? 5 A. Yes, they were -- I mean, not a perfectly 6 different set, there were some same plants. But they 7 were -- in that -- in the two samples, really very 8 different plants that were reporting this time around. So 9 the volume was similar, the number of plants was similar, 10 the results not similar. 11 Q. Okay. On your dry whey survey, between the two, I 12 think you had a similar number of plants, but the volumes 13 reported were increasing, suggesting larger plants. 14 But if you look at the numbers between the 15 surveys -- so that's kind of similar to -- well, the 16 volume piece is similar to the nonfat dry milk, but we saw 17 costs increase from '21 to '23. Just wondering if you had 18 an idea of why we saw the increases there. 19 A. I am not exactly sure. I can characterize some of 20 those as being what we would think of as just the 21 operating costs in a plant, that labor had certainly gone 22 up, and that would have been true across all these plants, 23 not just whey. But we also had different plant sets here, 24 so they weren't the same whey plants, much like I reported 25 in others, and why I made kind of a point of saying the 26 sample matters. 27 Q. Okay. I'll turn to your Exhibit 178, your 2023 28 study. I don't know if you had heard earlier witnesses -- 3596 1 I think it was this week, I can't remember -- they talked 2 a lot about how insurance costs increased, and I just want 3 to be clear on where those costs might show up in the 4 reported costs. 5 Would those be under general administrative costs? 6 A. Yes. 7 Q. On page 10 of this exhibit, 10 of 30, under 8 observations, and for each of these you go through, each 9 paragraph discusses a different commodity, and you talk 10 about the non-transformed weighted average values for the 11 '21 and -- for the 2021 study. 12 So I take it you went back to the 2021 study and 13 did the allocation the old way on a solids basis; is that 14 correct? 15 A. I did. And I was curious about taking a look at 16 just seeing while impactful that that may have been, in 17 most cases not as much as you might expect but -- or hope 18 for perhaps. In other words, it didn't explain the low 19 cost of butter processing. 20 Q. Yeah. So -- 21 A. It wasn't that butter was undervalued and nonfat 22 overvalued. It -- I mean, to some extent that would have 23 been true. But this was more of a sample problem than it 24 was a weighting problem. 25 Q. Okay. I appreciate that. I was going to ask 26 about it. I wrote down what the numbers were using the 27 first transformation number, and then what the numbers are 28 using the old way on the solids basis, and actually, all 3597 1 of those costs declined. 2 A. Yeah. Again, sample. 3 Q. Okay. Okay. If we could turn to page 13. 4 So we took a look at the pounds reported in your 5 study here, multiplied it by the N of the relevant 6 commodity, to come up with a total pounds surveyed, and 7 then compared that to our NDPSR volumes. 8 And for butter especially, that number was 9 considerably higher than what we actually capture in 10 NDPSR, like eightfold. I'm just wondering if you might 11 know why that is? Are you capturing unsalted butter, 12 perhaps? Or it's bulk butter that isn't reportable to 13 NDPSR because eventually it goes into retail packaging? 14 Just kind of curious why there's such a large difference 15 there. 16 A. Yes. There are some of both things that you just 17 mentioned. There's some unsalted butter in here, and that 18 would have been a cost that was not included in the 19 ingredients cost, pretty small cost in the overall scheme 20 of things. But there were certainly many of these plants 21 that also produced consumer packages. Those aren't 22 included in here. 23 But up through the churn, processing costs are 24 pretty similar for the products that are going into 25 consumer packages. It's just additional labor in the 26 packaging room and other things that would not have been 27 included here. 28 Q. So labor is not included either? 3598 1 A. No. If we can sort -- if we can sort that out, 2 you know, then there is -- in the screenshots you can see 3 that there are -- well, I don't know if I have butter on 4 here or not. I think so, though. Yeah. Well, there's 5 butter processing, and then there's butter packaging -- 6 and I may have to own that one, too, that there's a 7 possibility that there could have been some of the labor 8 that would have been involved in butter packaging for 9 consumer products, that that could have gotten in there. 10 The packaging costs themselves would not have gotten in 11 there, but bulk packaging and consumer packaging labor may 12 have been commingled. 13 Q. And that would be the same in the 2021 -- 14 A. It would be the same in both. 15 Q. Yes, you could say the same in 2021? 16 A. Yes, both. 17 MS. TAYLOR: I think Mr. Wilson has a few 18 questions. 19 CROSS-EXAMINATION 20 BY MR. WILSON: 21 Q. Todd Wilson, USDA. Hello, Dr. Stephenson. 22 The -- just so that I can kind of get my head how 23 you break the low versus high. So you're doing that on 24 total dollar cost as a midpoint median? 25 A. Yes. So when we look at the total cost for 26 individual plants, all the plant observations are ranked 27 from low to high. And, again, if there's an even number 28 of plants in there, then it would be the 50% low, 50% 3599 1 high. If it's an odd number of plants, then that one 2 plant is going to be assigned to one or the other. 3 Q. And it would be all costs that you have listed in 4 this -- in these tables? 5 A. The break is made based on total costs. I mean, 6 that's the calculation and ranking that I do. The rest of 7 these carry from those individual plants. 8 Q. So the breaking would incorporate costs that are 9 not necessarily in these weighted numbers because of 10 consumer type packaging labor or whatever? 11 A. If the laboring -- labor costs got commingled in 12 here, then it would be included for all of those plants. 13 But, no, I -- I don't include -- 14 Q. Okay. It's not the total -- excuse me -- it's not 15 the total plant cost, it's the costs that are associated 16 with -- 17 A. With the product of interest, the one that's 18 reported here. That's correct. 19 Q. Okay. Thank you. I just wanted to try to be 20 clear. 21 In reference to the 15 plants that are overlapped. 22 Okay? Some of those were in different -- in all products. 23 Okay? Were there -- were there times, did you notice, 24 were there times when plants might flip from a high to a 25 low between '21 and '23, or do you know? 26 A. I didn't look for that, so I -- I can't really 27 comment at all about it, I guess. I could look back in 28 data when I'm at home and see whether that was the case, 3600 1 but I can't really comment here. 2 Q. Okay. I'm going to switch over -- or switch back 3 to whey, page 15. 4 I was just comparing some values across all 5 products, and I thought this of interest, and I wanted 6 maybe your opinion of why it might be. 7 So when we look at general and administrative on 8 the '21 study, it was the lowest of all the products, 9 butter, powder, cheese, whey. But in '23, it was the 10 highest. Just interesting that G&A would be that 11 divergent between those two population groups. 12 A. We did have, as I said, different plants in the 13 two samples. And for the G&A in particular, there were a 14 few plants in the whey category where we had 15 non-reporting, like, of superintendent salary values and, 16 you know, a couple of things of that nature, which gave a 17 very low G&A in the earlier study. 18 In this one, different plants reported 19 differently. So, again, when I come down to saying that 20 having the authority to compel participation and to audit 21 data like that, that I think these would be numbers that 22 you would want to assure were included. 23 MR. WILSON: Thank you. 24 MS. TAYLOR: I think I'm more organized now. 25 CROSS-EXAMINATION 26 BY MS. TAYLOR: 27 Q. We have heard throughout this hearing about 28 different investments, capital investments that plants 3601 1 have made. And I'm wondering if you could elaborate or 2 inform us on how that might -- how that's accounted for in 3 these numbers. We have heard about investments in things 4 like implementing UF at the start of a production process, 5 and we have also heard about capital investments such as 6 wastewater treatment investments. 7 So are those accounted for in these numbers, or 8 are some accounted for and not others accounted for? 9 A. Well, to the extent that it impacts the production 10 process and are costs that would be asked for and, thus, 11 captured in here, they are accounted for. So just as a 12 good example, water and waste treatment is a line item in 13 that ledger category that's there. So the attempt is 14 certainly made to capture some of those kinds of things. 15 UF in a plant or RO in a plant in front of the 16 vats would be capital equipment that would be purchased, 17 and at some point in time depreciated, and otherwise noted 18 in there. Hopefully, it's providing greater throughput 19 for plants, and so, you know, we -- we may see -- I would 20 assume that the investment was made to lower total overall 21 costs in the plant, and that those would be reflected in 22 the total numbers. 23 Q. Okay. And then I want to use an example of a 24 plant that has a cheese and a whey side. And say on their 25 whey side they did 20% dry whey, which would be 26 reportable, and 80% some type of value-added whey. 27 Are you -- are we at all perhaps capturing some of 28 that cost to the value-added side in here, or you were 3602 1 able to desegregate those? 2 A. To the fullest extent possible, we try to 3 disaggregate them. And, again, that's going to be based 4 on the pounds of solids in those two finished products. 5 And we are not going to get a complete allocation, I 6 guess, for that, because some of this is going through a 7 very similar stream in here. But we would have less of 8 some of the pounds of product in the value-added process 9 than we would have in the dry whey product, and the 10 allocation would be a bit different. 11 Q. Okay. 12 A. But, again, up to the point that we are actually 13 capturing the whey from the cheese making process, then 14 that's the place where the differences begin to express 15 themselves. 16 Q. So another difference I wanted to clear up for the 17 record is in your '06 survey you had a category of 18 repairs -- repairs and depreciation in ingredients. You 19 don't have that same category heading in your 2021 or '23 20 surveys. 21 So would that all fall under what is now called 22 nonlabor and utilities processing if we wanted to compare? 23 A. Yes. I think that CDFA towards end of their time 24 period, and maybe you will have the opportunity to look at 25 those more closely, changed the way that they reported 26 things. So I had looked to see what they were including 27 in their different titles. I'm not sure why they changed 28 that, but they had done it. And as I mentioned earlier in 3603 1 testimony, I had tried in the past to report the same way 2 CDFA did so people could have a -- look at an audited set 3 of data for at least a subset of plants, and then what was 4 done here in a more national level. 5 Q. Okay. So you have answered some other questions 6 from other parties about costs and whether 2022 -- I'm 7 generalizing here -- was a good year to use for cost data 8 because of supply chain disruptions and inflation, 9 etcetera. And I know you answered that question. All 10 costs go up or down, and I understand that. 11 But as you were looking through the data, did -- 12 the monthly data that was given to you -- did you see that 13 those costs were moderating at all at the end of the year? 14 A. I didn't -- I don't recall noticing anything that 15 really jumped out at me. And there are only a few places 16 where I'm actually capturing monthly costs, that's true in 17 utilities, and I don't recall that, you know, I noticed 18 the utilities changed very much over the course of a year 19 for any particular plant or buyer. 20 Q. Okay. So I have one more question, and I'll see 21 if you take the opportunity to answer it or not. 22 A. All right. 23 Q. Since we have this opportunity to talk to you, and 24 you certainly are an expert in this cost data that you 25 have been collecting for years, and we have to go back at 26 USDA and take this lovely record and help the Secretary 27 determine what he finds is appropriate based on the 28 record. 3604 1 And I was wondering if you had any offerings, 2 based on your experience, how we could use the cost data 3 to make a decision on minimum-regulated prices, given that 4 we're kind of working off of averages of various surveys 5 that you have done. So I didn't know if you wanted to 6 take this opportunity to offer your thoughts on that or 7 not. 8 A. I have tried to provide data and information for 9 folks to use in making determinations of exactly what you 10 are talking about, and it seems to me that it edges into 11 policy decisions. 12 If you were to ask me what would you do, I'm not 13 sure that I feel like I should be making those kinds of 14 decisions for you. This is an industry where you need to 15 put on the blindfold and hold up the scales of justice and 16 come out with something that is going to be good for the 17 industry. 18 Q. So you are deferring on that advice is what I'm 19 gathering. 20 A. I could have shortened it, and said, yeah, no 21 comment. 22 Q. So noted. 23 MS. TAYLOR: I think that's it from AMS. Thank 24 you. 25 THE WITNESS: Okay. 26 THE COURT: Mr. Rosenbaum? 27 REDIRECT EXAMINATION 28 BY MR. ROSENBAUM: 3605 1 Q. Just a couple points of clarification. 2 In your testimony, which is Hearing Exhibit 176, 3 you have a Table 1 on page 5 of 7, which AMS asked some 4 questions about. And -- and I think AMS, in doing their 5 calculations, suggested that the column that says '06/'19 6 percentage change, actually is a comparison between the 7 2008 Make Allowances and the 2019 results. 8 Do you remember that colloquy you had? 9 A. I do. 10 Q. And I think your testimony was, if that's the 11 case, that was not your intent. You had intended the 12 calculation to be a comparison between the 2006 results 13 and the 2019 results, correct? 14 A. Yes. That was what my intention was. I should 15 probably check that and see if I had actually done it or 16 not. 17 Q. I'm going to -- let me say, I'm going to suggest 18 to you that the numbers in the June 2022 percentage change 19 is, in fact, what I believe you intended, namely a 20 comparison between the 2006 results and the 2022 results. 21 In other words, I think you did that column the way you 22 intended. 23 A. Okay. 24 Q. But maybe you could just double-check both those 25 things just so we're sure that the record is clean on 26 this. 27 We have doubled-checked the 2022 numbers and -- 28 (Court Reporter clarification.) 3606 1 BY MR. ROSENBAUM: 2 Q. We double-checked the June '22 numbers and the 3 percentages, as we calculated them last night, were, in 4 fact, a comparison of the 2006 results and the -- and the 5 2022 results. 6 A. I just grabbed the wrong column, mentally -- 7 Yeah. I -- the percent change on the '06 to '19 8 is actually a comparison with the Make Allowance column. 9 Q. Okay. And to come up with the correct number for 10 cheese, for example, that should be in place of the 24%, 11 the math is simply -- correct me if I'm wrong -- the 12 correct math would be $0.247 minus $0.158 divided by 13 $0.158? 14 A. Yes. 15 Q. And so -- and similarly for whey, it should be 16 $0.265 minus $0.197 divided by $0.197? 17 A. Yes. 18 Q. And just to complete it, butter should be $0.141 19 minus -- well, minus $0.18 divided by $0.18, right? 20 A. Yes. 21 Q. And nonfat dry milk, to close it off, would be 22 $0.293 minus $0.166 divided by $0.166, correct? 23 A. Yes. 24 Q. And the numbers in two thousand -- in the column 25 2019, those reflect your use of the transformational 26 adjustments; is that correct? 27 A. Yes. 28 Q. Okay. Which I think you said affect butter and 3607 1 nonfat dry milk in opposite directions. Am I right about 2 that? The employment of transformational techniques. 3 A. Yes. I also indicated that the bigger difference 4 in these numbers had more to do with sample than it did 5 the weighting. 6 Q. All right. And have you been able to confirm for 7 me that the numbers in the column -- the percentage 8 numbers in the column June 2022 do reflect what you 9 intended, namely the 2022 results minus the 2006 results 10 divided by the 2006 results? 11 A. Yes. They do. 12 Q. And one thing that just leaps out at me is it's 13 actually a rather consistent number in terms of what the 14 percentage increase was from 2006 to 2022, among the four 15 different components on a percentage basis, correct? 16 A. It is. And I was a little surprised, I guess, 17 that it was that close, but I -- that gives me some degree 18 of comfort that maybe those samples were, you know, 19 reasonable in those two years as well. 20 Q. I mean, presumably the labor costs, electrical 21 costs, general inflation, would be factors faced, more or 22 less, equally regardless of what kind of commodity you are 23 making. Is that reasonable? 24 A. Yes. But there are a lot of things that I know -- 25 when you start to look at the individual plant data, one 26 of them is just how different things like utility rates 27 can be from one plant to the next. And I don't mean by a 28 small amount. I mean, they are substantially different. 3608 1 So clearly folks are able to secure different costs just 2 based on that. And over time, it's not really fair 3 necessarily to say these should all track, too, because we 4 have had investments that provide some energy recapture in 5 the plants, and not all plants have invested in that to 6 the same degree, so I mean factor usage is a little 7 different. 8 Q. Let me just ask a question relating to AMS's 9 questions about poundage. And if we could turn to page 13 10 of 30 just as an example, which I think was the one that 11 AMS was asking about. 12 A. On which exhibit? 13 Q. It is on Exhibit 178. I'm sorry. I'm turning now 14 to your June 2023 report. 15 A. Page 13? 16 Q. 13, yes. 17 And so you show the -- there's a row that says N 18 equals, and in this case it's 13. That's how many plants 19 reported costs for butter, correct? 20 A. That's correct. 21 Q. And every one of the commodities has an N on the 22 sheet that relates to it that provides how many plants 23 participated, correct? 24 A. Correct. 25 Q. And so if you multiply 13 times, in this case, 26 butter, 126,906,009 product pounds, does multiplying 13 27 against that number tell you the total poundage that was 28 covered by the plants covered in the survey? 3609 1 A. Correct. 2 Q. And if you wanted to determine what percentage of 3 total butter production in the United States that 4 represents, would you divide whatever number results from 5 multiplying 13 times 126,906,009, would you take that 6 number, divide it by the NASS survey of total butter 7 production in the United States for that year? 8 A. That would give you a percentage of reported 9 butter to NASS, yes. 10 Q. Okay. And would -- and reported to NASS should be 11 butter production in the United States, correct? I mean, 12 that's what that is a survey of it? 13 A. I'm not aware of them not tracking some plants or 14 reporting some plants for reasons. But, yeah, I would 15 think the U.S. total should be pretty much the total. 16 Q. So the 13 times 126,906,009 divided by the NASS 17 number would tell you what percentage of butter you are 18 survey picks up, correct? 19 A. Yes. 20 Q. And if that number is -- do the math, we'll have 21 someone do the math, testify to it -- 85% or some other 22 number similarly, you really not -- your sample at that 23 point is -- 24 A. It is pretty good. 25 Q. -- it is like very good, isn't it? 26 A. Yeah. It's nearly a census. I mean, there would 27 be probably many plants that didn't participate in this 28 that are going to be very small operations or quite small 3610 1 operations in comparison to these large plants that 2 participated this time around. Yeah. So there may be a 3 fair number of those small plants, but they didn't 4 comprise much volume. 5 MR. ROSENBAUM: That's all I have. Thank you. 6 MS. TAYLOR: I'd like to ask another question if 7 we may. 8 THE COURT: No objection? 9 MS. TAYLOR: Sorry. 10 RECROSS-EXAMINATION 11 BY MS. TAYLOR: 12 Q. We only get this one opportunity to make sure we 13 understand all these numbers, Dr. Stephenson, so -- I'll 14 just keep to the same page because I assume it's the same. 15 And I just want to make sure -- I thought we were clear, 16 but then we confused ourselves. 17 So the low class product pounds is the weighted 18 average of all the plants that fall in that category? 19 A. It's the average of all the plants in that 20 category. 21 Q. Yes. Okay. 22 And then the factors break -- cost breakouts in 23 that category are the weighted average? 24 A. Yes. 25 Q. Okay. Same for the high cost plants? 26 A. Yes. 27 Q. And so for all plants, is that a simple average or 28 is that a weighted average of 126 million? 3611 1 A. It's a weighted av- -- well, that is a simple 2 average as well, I mean, for the total pounds of product. 3 Q. Okay. Of the 13 plants? 4 A. Yes. 5 Q. Yeah. Okay. 6 And then the cost factors on that line? 7 A. Those are also weighted average cost values. 8 Q. Of all 13 plants? 9 A. Yes. 10 Q. Okay. 11 A. Was there a question? 12 Q. No, we just looked at your face and thought maybe 13 you were thinking about your answer again. 14 A. No, no, no. I -- no. No, I -- I stand by my 15 answer. Yes. 16 RECROSS-EXAMINATION 17 BY MR. WILSON: 18 Q. Could I ask a question maybe to clarify it in my 19 head. 20 The all plants row -- 21 A. Yes. 22 Q. -- is the sum of all the costs divided by all the 23 pounds? Not the average of the pounds, but all the 24 pounds? 25 A. This is the weighted average total costs. So, for 26 example, if you added up those weighted averages across 27 each of the categories, you should get the total cost. If 28 you added up all of the total costs of individual plants 3612 1 and divided by the total pounds, you should get the same 2 number. So, I mean, it's two different ways of getting to 3 that bottom right-hand corner. 4 MR. WILSON: Okay. Thank you. 5 THE WITNESS: You're welcome. 6 RECROSS-EXAMINATION 7 BY MS. TAYLOR: 8 Q. But for the example, because we were talking about 9 it, the general and administrative costs, if the plant did 10 not report those numbers -- 11 A. Right. 12 Q. -- then that number in the all plants line is only 13 for the pounds that were reported -- pounds where costs 14 were reported? 15 A. For that category, that's correct. 16 Q. For that category? 17 A. Yes. 18 Q. Okay. 19 A. So in other words, it would provide an accurate 20 representation of maybe 12 out of the 13 plants or 21 something to that effect, but it wouldn't be distorted by 22 being a zero value. 23 Q. Okay. We think we understand now. 24 A. Thank you. 25 MS. TAYLOR: Thank you. 26 THE COURT: Mr. Rosenbaum? 27 MR. ROSENBAUM: Your Honor, at this point I simply 28 would like to move Exhibits 176, 177, and 178 into 3613 1 evidence. 2 THE COURT: That's redirect, huh? Okay. 3 Any objections? 4 Exhibits 176, 177, 178 are made a part of the 5 evidentiary record in this hearing. 6 (Thereafter, Exhibit Numbers 176, 177, and 7 178 were received into evidence.) 8 THE COURT: With that, should be break for lunch? 9 MR. ROSENBAUM: Yes, your Honor. 10 Let me just say, our next witness after lunch is 11 going to be Dr. Schiek, and we are actually introducing a 12 number of documents into the record through him. And I 13 will put at the back, I'm going to put boxes -- not for 14 USDA, they get their own set -- but for anyone else who 15 wants a copy, just please take one each. It's going to be 16 IDFA Exhibits 2, 7 through 21, and 40. So you should have 17 one of each of those. 18 THE COURT: Can you gather up copies for me? 19 MR. ROSENBAUM: Yes, of course, your Honor. 20 THE COURT: Is there any controversy about that? 21 Ms. Hancock, you rose. 22 MS. HANCOCK: No, I'm just going to lunch. 23 THE COURT: All right. Let's go to lunch. Come 24 back at 1:05. 25 (Whereupon, a luncheon break was taken.) 26 ---o0o--- 27 28 3614 1 WEDNESDAY, SEPTEMBER 13, 2023 - - AFTERNOON SESSION 2 THE COURT: Another witness, Mr. Rosenbaum. 3 MR. ROSENBAUM: Your Honor, we call as the next 4 witness, Dr. William Schiek. 5 DR. WILLIAM SCHIEK 6 Being first duly sworn, was examined and 7 testified as follows: 8 THE COURT: Your witness. 9 DIRECT EXAMINATION 10 BY MR. ROSENBAUM: 11 Q. Dr. Schiek, you have prepared written testimony 12 today which has been marked IDFA Exhibit 2; is that 13 correct? 14 A. Correct. 15 MR. ROSENBAUM: Your Honor, I would ask that this 16 be given the next Hearing Exhibit number. 17 THE COURT: Yes. 180. 18 (Thereafter, Exhibit Number 180 was marked 19 for identification.) 20 MR. ROSENBAUM: And then, your Honor, we have a 21 series of reports by the California Department of Food and 22 Agriculture, which we would like to have marked seriatim. 23 And so the next one is IDFA Exhibit 7, which we 24 would ask to be marked as Exhibit 181. 25 THE COURT: Yes. IDFA Exhibit 7 is marked for 26 identification as 181. 27 (Thereafter, Exhibit Number 181 was marked 28 for identification.) 3615 1 MR. ROSENBAUM: And then IDFA Exhibit 8 would be 2 182, Hearing Exhibit 182. 3 THE COURT: And that is an 8 on there. Yes. 182 4 is so marked. 5 (Thereafter, Exhibit Number 182 was marked 6 for identification.) 7 MR. ROSENBAUM: IDFA Exhibit 8 -- 9, excuse me, 8 would be Hearing Exhibit 183. 9 THE COURT: Yes, so marked. 10 (Thereafter, Exhibit Number 183 was marked 11 for identification.) 12 MR. ROSENBAUM: IDFA Exhibit 10 is Hearing 13 Exhibit 184. 14 THE COURT: Yes. So marked. 15 (Thereafter, Exhibit Number 184 was marked 16 for identification.) 17 MR. ROSENBAUM: IDFA Exhibit 11 is Exhibit 185. 18 THE COURT: So marked. 19 (Thereafter, Exhibit Number 185 was marked 20 for identification.) 21 MR. ROSENBAUM: IDFA-12 is Hearing Exhibit 186. 22 THE COURT: So marked. 23 (Thereafter, Exhibit Number 186 was marked 24 for identification.) 25 MR. ROSENBAUM: IDFA-13 is Hearing Exhibit 187. 26 THE COURT: So marked. 27 (Thereafter, Exhibit Number 187 was marked 28 for identification.) 3616 1 MR. ROSENBAUM: IDFA Exhibit 14 is Hearing 2 Exhibit 188. 3 THE COURT: So marked. 4 (Thereafter, Exhibit Number 188 was marked 5 for identification.) 6 MR. ROSENBAUM: Hearing Exhibit 15 is -- sorry, 7 IDFA Exhibit 15 would be Hearing Exhibit 189. 8 THE COURT: So marked. 9 (Thereafter, Exhibit Number 189 was marked 10 for identification.) 11 MR. ROSENBAUM: IDFA Exhibit 16 will be Hearing 12 Exhibit 190. 13 THE COURT: So marked. 14 (Thereafter, Exhibit Number 190 was marked 15 for identification.) 16 MR. ROSENBAUM: Exhibit 17 will be Hearing Exhibit 17 191. 18 THE COURT: So marked. 19 (Thereafter, Exhibit Number 191 was marked 20 for identification.) 21 MR. ROSENBAUM: Hearing Exhibit 18 will be hearing 22 Exhibit 192. 23 THE COURT: So marked. 24 (Thereafter, Exhibit Number 192 was marked 25 for identification.) 26 MR. ROSENBAUM: IDFA Exhibit 19 will be Hearing 27 Exhibit 193. 28 THE COURT: So marked. 3617 1 (Thereafter, Exhibit Number 193 was marked 2 for identification.) 3 MR. ROSENBAUM: And then IDFA Exhibit 20 would be 4 marked Exhibit 194. 5 THE COURT: Oops -- so marked. 6 (Thereafter, Exhibit Number 194 was marked 7 for identification.) 8 MR. ROSENBAUM: And although I distributed IDFA 9 Exhibit 21, it turns out that actually is already in 10 evidence as Hearing Exhibit 156, so I'm not asking that 11 that be marked as a new exhibit. It's Hearing 12 Exhibit 156. That actually was put in by National Milk, 13 so it actually is NMPF-18A, but it's already in evidence 14 so I'm not going to ask that be given a separate number. 15 THE COURT: Very good, sir. 16 MR. ROSENBAUM: And then, lastly, Dr. Schiek is 17 about to go through a PowerPoint presentation, which has 18 been marked as IDFA Exhibit 40, and so we would ask that 19 that be marked as Hearing Exhibit 195. 20 THE COURT: So marked. 21 (Thereafter, Exhibit Number 195 was marked 22 for identification.) 23 BY MR. ROSENBAUM: 24 Q. Dr. Schiek, why don't we go ahead and put up your 25 PowerPoint presentation that you have put together, and 26 start by telling us a little bit about yourself. You 27 don't need to read this, but just tell us what your 28 background is. 3618 1 A. Yeah. Well, I'm currently executive director of 2 the Dairy Institute of California. Dairy Institute is a 3 dairy processor trade association similar to IDFA. We 4 engage in regulatory and legislative advocacy on behalf of 5 our members. Our core members are folks who have plants 6 in California, dairy plants in California, and buy milk in 7 California. 8 Q. And how long have you worked for the Dairy 9 Institute of California and in what positions? 10 A. So I have -- I started with Dairy Institute in 11 1997, and was the economist. And back in those days we 12 had a state regulatory pricing program, and it was my 13 responsibility to help the members coalesce around policy 14 positions that they wanted advanced at those hearings, and 15 I was a principal witness testifying on behalf of the 16 Institute. 17 Q. Okay. And in those days, California was not part 18 of the Federal Milk Marketing Order system; is that 19 correct? 20 A. Correct. 21 Q. And then when did you become executive director of 22 the Dairy Institute? 23 A. I became executive director on January 2020 24 following the retirement of my predecessor, who had been 25 there since 1997, and in another capacity before that. 26 And so since that time, I have been responsible for 27 basically all the advocacy efforts of Dairy Institute. 28 This would include legislative advocacy and other 3619 1 regulatory arenas beside milk pricing and policy. 2 Q. And taking -- going back in time, what did you do 3 before you started at the Dairy Institute? 4 A. So prior to joining Dairy Institute, I was an 5 assistant professor in the Department of Agricultural 6 Economics up the road here at Purdue University in West 7 Lafayette, Indiana, and was there from August '91 until 8 May -- through May '97. 9 Q. And prior to that? 10 A. Prior to that I was employed -- well, I was in 11 graduate school, so that time at the University of 12 Florida, but I was employed by the New York/New Jersey 13 milk Market Administrator. I started there in June '82, 14 when I had graduated from Cornell with my Bachelor's 15 degree working as a cooperative relations specialist, 16 which was basically a role of administering the 17 cooperative payments provisions that were part of the 18 Federal Order 2, New York/New Jersey Order at that time. 19 Later I trained my successor in that job, who I 20 don't see in the room anymore, that was Ed Gallagher. And 21 Ed took over that job, and I moved on to be just an 22 economist in the office working on special projects and 23 research projects. 24 And then he talked about, I think in his 25 testimony, being one of the Wilson Fellows -- and we're 26 not talking about Todd Wilson, we're talking about a 27 different Wilson. 28 Although, Todd, if you want to give me money to go 3620 1 to school, that will be good. 2 You know, I went to graduate school and did 3 research in dairy marketing, and I was funded, in part, by 4 the New York/New Jersey Milk Market Administrator office. 5 Q. And tell us what degrees you have. 6 A. So I have a Bachelor's degree in applied economics 7 and business management, with a specialization in business 8 management and marketing from Cornell University. That 9 was in 1982. And then at the University of Florida, I 10 have a Master's of Science and a Ph.D. from the Department 11 of Food and Resource Economics, with a specialty in dairy 12 marketing and policy. 13 Q. Okay. 14 MR. ROSENBAUM: Your Honor, I would ask at this 15 point that Dr. Schiek be declared an expert in 16 agricultural economics and food and resource economics. 17 THE COURT: Yes. 18 MR. ROSENBAUM: As well as applied economics. 19 THE COURT: I find him so qualified. 20 BY MR. ROSENBAUM: 21 Q. Dr. Schiek, at the time that the State of 22 California had its own milk marketing regime, did the 23 state conduct surveys of manufacturing costs for dairy? 24 A. Yes, they did. 25 Q. And were -- was participation in those surveys 26 mandatory? 27 A. It was for the plants that they selected to be -- 28 participate in the survey. I assume those were plants 3621 1 that met the product definition that they were interested 2 in studying. 3 Q. And were the -- were those surveys subject to 4 audit? 5 A. They were. 6 Q. Okay. And were they, in fact, audited? 7 A. They were. 8 Q. Now, we have marked a series of exhibits which 9 have been numbered as exhibit -- Hearing Exhibits 181 10 through 194. And then, as I mentioned, they is already a 11 document that has been marked as Hearing Exhibit 156. 12 Are these official publications of the California 13 Department of Food and Agriculture? 14 A. They are, yes. 15 Q. And are these the reports going from the period 16 2002 through 2016, in which the California Department of 17 Food and Agriculture was reporting the results of the cost 18 surveys they had conducted? 19 A. They are. 20 Q. Okay. And -- and did California cease performing 21 these audits after 2016? 22 A. That's correct. That was the last year they did. 23 Once California joined the Federal Order system, the 24 manufacturing cost unit was disbanded at CDFA, and they 25 were not doing those studies anymore. 26 Q. Okay. If we could go back to the PowerPoint 27 presentation, and to the next page, please. Tell us -- 28 tell us what it is you have undertaken for purposes of 3622 1 this hearing. 2 A. Yeah. Basically, what we have tried to do here is 3 to estimate dairy manufacturing costs from existing CDFA 4 data and other economic data to project what costs would 5 be in the more recent time periods since CDFA ceased 6 publishing this information where they -- you know, we no 7 longer have those audited cost numbers available. 8 CDFA collected manufacturing cost data from 9 California plants for many years beginning in 1989. A lot 10 of that earlier data was collected in a temporally 11 inconsistent manner. I guess that's a way of saying they 12 didn't do structured annual calendar year audits. They 13 would often cover a period, sometimes as long as two 14 years. Sometimes they -- one report to the next, they 15 would be overlapping. 16 So in doing this analysis for econometrics, you 17 want sort of discreet data, and so we selected the data 18 for which there was annual reports covering a calendar 19 year, so starting -- that started in 2002. And so that 20 began doing audited reports on an annual basis, on a 21 calendar year basis, beginning in 2002. 22 Current Federal Order Make Allowances were 23 established, in part, using work from Dr. Stephenson that 24 he did when he was at Cornell on manufacturing costs, as 25 well as CDFA data from 2006. So I just note that there's 26 a precedent for USDA using both survey data from 27 Dr. Stephenson and CDFA cost data. 28 Q. And, in fact, the Make Allowances that we are 3623 1 currently living under are, themselves, based at least, in 2 part, on both CDFA data and Dr. Stephenson's then 3 contemporaneous survey, correct? 4 A. That's correct. 5 Q. We go to the next slide, please -- 6 A. Yep. 7 Q. -- the approach. 8 So I think -- just so we're clear about this, you 9 were trying to take that CDFA audited data that takes us 10 through 2016 and use it to project what you -- what costs 11 would be as of what point in time? 12 A. Right, as of 2022. Basically we -- we have 13 projections for each calendar year beyond the survey out 14 through 2022. 15 Q. And tell us what the basic methodology was. 16 A. Yes. So basically we employed regression analysis 17 using ordinary least squares. Ordinary least squares is a 18 very commonly used type of regression analysis to fit 19 essentially a linear relationship between some independent 20 variables, things that we think will be impacting the 21 dependent variable, which in this case is manufacturing 22 cost. And so we're looking at variables that might -- 23 economic theory would suggest would have an impact on 24 manufacturing costs, and those would be things like energy 25 costs, labor costs, material costs, and we might also look 26 at how productivity changes are also impacting those. 27 Q. Okay. And tell us how the use of this data 28 captures productivity. 3624 1 A. So we explicitly included in the -- in the 2 equation for labor cost and the equation for other costs, 3 measures of productivity changes. One was a general 4 productivity change in terms of labor, and the other was a 5 total factor productivity change that captures more than 6 just labor productivity gains for the food, tobacco, and 7 manufacturing -- food and tobacco and beverage 8 manufacturing industries. 9 Q. And tell us how you dealt specifically with whey 10 manufacturing costs. 11 A. Yeah. Whey manufacturing costs presented a 12 problem because CDFA did collect some data on dry whey, 13 the cost of dry whey manufacturing. But that data is 14 pretty limited, they collected it for a period of four 15 years. And even that data, when it was collected, was 16 collected from plants that were probably higher cost 17 because of how they were operating. They weren't 18 necessarily running full volumes all the time, and so they 19 had very high costs during that period. And that data 20 ceased when -- when the number of plants producing dry 21 whey dropped below three in the state. So there's not 22 enough of a time series to do a whey model on this with a 23 regression analysis. 24 So what we did, and this had been used in the 25 past, is you try to come up with a number that represents 26 the incremental drying cost of whey, which is a more 27 dilute solution than skim milk, and so there would be a 28 higher cost of drying whey than there would be, say, 3625 1 drying nonfat milk powder. 2 And we just noted here, we don't have -- I didn't 3 go out and estimate those costs of -- incremental costs, 4 but we have a relationship between the current 5 Make Allowance for nonfat dry milk and for dry whey, which 6 is roughly $0.03 a pound. It is a little more but -- 7 so -- so what we did was we looked at the projections of 8 nonfat dry milk, the estimates of nonfat dry milk cost, 9 and the projections going forward, and added $0.03 to that 10 as an estimate of the dry whey cost. 11 Q. Okay. And let's go to the next slide if we could. 12 And what -- what are we seeing here? 13 A. Yeah. This is just a table that lists the 14 weighted average -- CDFA survey weighted average 15 manufacturing cost for each of the commodities from 2002 16 through 2016. So cheddar cheese is in the -- next to the 17 year, cheddar cheese is in the next column. Then you see 18 the four years for which there was dry whey data. And 19 then weighted average butter manufacturing costs under the 20 CDFA survey in the next column. And then finally, the 21 weighted average cost manufacturing cost for nonfat dry 22 milk. 23 Q. And are the figures that appear here on page 5, 24 are those taken from the various CDFA reports that we have 25 marked as exhibits? 26 A. Yes. 27 Q. And do they provide additional detail as to how 28 many pounds were subject to the survey and things of that 3626 1 nature? 2 A. Yeah, there's a lot more detail there. 3 Q. Okay. Let's go to the next page, please. 4 And tell us more about the approach you took. 5 A. Yeah. One of the things, even though we had 6 annual cost data for 2002 through 2016, I wasn't able to 7 locate the report for 2002 that has the extensive 8 breakdown of cost data. Part of the reason is those are 9 no longer up on the web by CDFA, so you -- you have to go 10 back and find them via another source or an archive of the 11 Internet. And so we don't have the breakdown on those 12 costs for 2002. We have labor costs and we have total 13 costs, but we don't have -- we don't have the breakdown of 14 the way I have broken down the costs here in terms of 15 utility cost and -- and other costs. 16 So we could only get the broken down data from 17 2003 to 2016. So we estimated -- we looked at CDFA dairy 18 manufacturing costs for utilities and other costs for 19 butter, nonfat dry milk, and cheddar cheese for 2003 to 20 2016. 21 For labor costs and to estimate trend models, we 22 actually were able to use the full set, 2002 to 2016, 23 because we had that extra year of data for those -- for 24 those analyses. 25 One of the things that became clear as we started 26 looking at the various variables that we -- independent 27 variables that we think will be influencing costs, that 28 economic theory suggests will be influencing costs, was 3627 1 that some of them are highly correlated. And that creates 2 a difficulty in estimation where you have got a lot of 3 variables that -- a few variables that are moving in the 4 same direction and are highly correlated with each other, 5 and it makes it difficult to estimate the parameters 6 associated with those variables that -- how they would 7 impact cost. 8 So to deal with that we -- we estimated the model 9 separately so you didn't have a lot of the correlated 10 variables in the same model. So we estimated, as I said, 11 utility costs, we estimated labor costs, we estimated all 12 other costs. So that would be everything that -- not 13 total -- everything that makes up total costs that's not 14 either labor or utility costs. So three equations, and 15 you can sum each component -- labor, utility, and other -- 16 to equal total cost. 17 Q. Okay. And if we can turn to the next page, 18 page 7. 19 Does this provide some information about, you 20 know, what -- what data you were including? 21 A. Yes. So in the utility cost equations we were 22 looking primarily at energy -- industrial energy prices. 23 So the data we used was from the U.S. Department of 24 Energy, Energy Information Agency, and these were 25 industrial rates for natural gas and for electricity that 26 were used. And it depended on the commodity which ones we 27 were using. 28 Q. That was in the state of California? 3628 1 A. In the state of California. So these are 2 California industrial energy prices. 3 Q. And then what for labor? 4 A. For labor we looked at another published series on 5 California wage rates for non-supervisory manufacturing 6 workers. This was data that was collected by the U.S. 7 Department of Labor, Bureau of Labor Statistics. And that 8 is our sort of proxy representation of plant -- plant 9 labor costs. 10 And we also used Bureau of Labor Statistics data 11 for non-farm labor productivity, which was included to 12 account for the impact of increasing productivity, and to 13 basically separate that out. I mean, we could have 14 estimated that model without including it, and the labor 15 productivity change would have been captured in the data. 16 But to be able to kind of talk about it, if productivity 17 growth outside of the model range, like as we start doing 18 forecasts, if productivity growth was higher or lower, 19 having it isolated, we can look at the impact of that as 20 well. 21 Q. Okay. And then the next page, please, you mention 22 the use of dummy variables. Can you just briefly 23 summarize the concept there? 24 A. Yeah. I just wanted to finish the last point on 25 that page. 26 Q. Sorry. 27 A. That's okay. 28 So the other cost category is a very broad 3629 1 category of costs. It includes things like repairs and 2 maintenance costs, depreciation, property taxes, plant 3 supplies, packaging costs, ingredients. The return on 4 investment allowance is in that number as well, as well as 5 general and administrative costs. 6 So it would be difficult to kind of get indicators 7 of drivers of all those individual costs separately, so 8 what we did was we used the U.S. Producer Price Index for 9 intermediate goods. Your factories, dairy manufacturing 10 plants are buying inputs from all over the country, and 11 essentially that is a measure of kind of inflation of 12 costs, at that level, at the intermediate level, over 13 time. We also included a productivity index that is a 14 little different. It's not labor productivity. It is 15 total factor productivity. So it looks at the 16 productivity of not just labor but materials and energy 17 and other things too. 18 Q. Okay. And let's go to the next page. 19 A. Yeah. So dummy variables, I know if you are not 20 in the econometric world, this is a strange sounding name. 21 But some of the changes can -- 22 Q. Can you put -- 23 A. Sorry. 24 Some of the changes in the -- in the cost numbers 25 can't easily be explained by changes in those other 26 variables that we have identified, like energy or labor 27 cost or productivity changes. 28 And basically what you see in the data is you will 3630 1 see a shift in the cost that isn't explained by these 2 changes in the other variables, and when you examine the 3 data, you recognize, okay, we have got -- something else 4 is going on here. And in some cases those are things we 5 know, just from knowledge of the industry. For example, 6 if one the largest manufacturers in the state opens a new 7 plant, for example, there's startup costs, and for -- for 8 a period of time when that new plant opens, usually the 9 first year, there will be much higher costs associated 10 with that. So that would be a known event. 11 Something else that might be known is if -- if we 12 knew somebody that -- you know, there was a big union -- 13 you know, unions represent workers in multiple plants, and 14 there's a new labor contract that shifts labor costs up 15 higher, we can account for that with one of these dummy 16 variables. And the reason that we -- and there may be 17 some shifts, sorry, that are -- that are unknown, but you 18 can see it clearly in the data and you know it isn't 19 related to one of the explanatory variables you have in 20 your model. So you -- you can deal with that with a dummy 21 variable and that kind of allows for that shift. 22 The reason you include these dummy variables is 23 that it does two things: One, it better captures -- it 24 helps you improve your explanatory power of costs, so 25 if -- your independent variables and your model does a 26 better job of explaining changes in cost when you include 27 them. 28 The other reason is that it -- it leads to better 3631 1 estimates, more accurate estimates of the impact of these 2 other explanatory variables like energy, prices, labor 3 costs, productivity, than you would -- better than you 4 would have if you excluded the dummy variables and 5 estimated the model without them. 6 Q. Okay. Next page, please. What are you showing 7 here? 8 A. Okay. So this is just a look at my next three 9 slides, really. Look at the three cost component -- 10 components that I'm estimating. 11 So this is the actual CDFA data, and this first 12 slide is cheese manufacturing costs from 2002 to 2016. 13 And I don't have total costs listed there, but I have the 14 three components. So you've got utility costs, which is 15 the gray line at the bottom; labor cost, which is sort of 16 an orange-brown line in the middle; and then all other 17 costs, which is the blue line at the top. 18 And for example, if you look at -- let's look at 19 the utility cost, which is the line at the bottom. You 20 will see starting in 2005 for like three years there was a 21 jump upward in that utility cost number. And actually, 22 when you look at the reports, you can see a jump in sewer 23 cost, which is a component of utility cost. And we know 24 at that time one of the larger cheese plants in the state 25 was having some issues with whey disposal and -- and so 26 they were having to dispose of whey in a way that was more 27 costly, and that was reflected in those sewer costs that 28 were picked up. So that's a case where you put a dummy 3632 1 variable in, and it is capturing that short-term event, 2 and that -- you know, that would be the case in that 3 situation. 4 Here's a -- here's another look at costs for 5 butter manufacturing in California. 6 Q. Okay. We're now on page 10. Go ahead. 7 A. Right. 8 And -- and here, for example, you see -- to that 9 page -- on the top line, you see a couple of years where 10 costs bumped up dramatically in 2008 and 2009. So that -- 11 that's one that corresponds to when a couple of new plants 12 were opening in California that were sizeable enough that 13 had an impact on costs. 14 Q. All right. Shall we go to the next page, which is 15 page 11, the nonfat dry milk page. Anything -- or is that 16 what you were just talking about? 17 A. No, I was talking about butter, but nonfat dry 18 milk looks similar, in part, because they often go 19 together when you open a new plant. So the corresponding 20 new plant is influencing nonfat dry milk as well as 21 butter. 22 We also saw a shift in utility costs that occurred 23 around the same time. 24 And then we -- sometimes you will see an ongoing 25 change that would be like a change in the slope. And so, 26 for example, if you -- if you look at the labor cost 27 number, sort of 2002 through 2011 or so, you have a 28 gradual increase in labor cost and then a steeper one 3633 1 afterwards going forward, 2012 and later. It might be 2 kind of hard to see that on there. But looking at the 3 numbers, you could see that change, and so you -- you 4 know, here is a case of not really sure what was going on 5 there, but it didn't seem to be related to the change in 6 wage rates. So you use a dummy variable to represent 7 there's something else happening that's accelerated the 8 rate of change in that cost. 9 Q. If we go to the next page, page 12, is this an 10 indication of the actual dummy variables that you used? 11 A. Yeah. So these were some of the dummy 12 variables -- or these were the dummy variables used in the 13 different models, and so you can just see, it is just -- 14 they are binary variables. So it's a 1 when that 15 particular condition that you -- you know is happening is 16 present, and it's a zero otherwise. So it has an impact, 17 you know, when -- when the -- when there's a 1 in place, 18 it has an impact on the model prediction. It does not 19 have an impact when the 1 is -- when it's a zero. 20 Q. Okay. Next page, please. 21 Tell us what the model and trend results were. 22 A. Right. So we -- we estimated those three 23 equations of those three cost components for each of the 24 three commodities, butter, cheese, and nonfat dry milk. 25 And what we found is that the estimated equations 26 generally showed good fit and strong overall correlations 27 for the equation. 28 Q. Okay. Explain to us what that means. 3634 1 A. Yeah. So basically there's a measure of fit that 2 we look at. It's called the R-square. And we actually 3 look at something call the adjusted R-square which is a 4 goodness of fit measure. It's determined by looking at 5 the sum of squared errors, which is a statistical term. 6 That is a result of the -- it is a total that you see in 7 the actual cost data. We compare that to what's explained 8 by the regression. 9 And so -- so basically one way of interpreting it 10 is that if an R-squared, say, has a value of 90, .90, it's 11 explaining 90% of the variation in the dependent variable. 12 So your model that you have estimated is explaining 90% of 13 that variation. That would be how you interpret the fit. 14 And the overall correlation is just -- you know, 15 it's very much like it. This is measured using an 16 F-statistic, where you look at does the model do a better 17 job of explaining the changes in the variables than no 18 model at all. So it's a pretty easy -- easy bar to clear, 19 but it -- it's -- it's certainly something you want to 20 look at. 21 Q. And -- and ultimately what did you -- sorry, you 22 also have something called correlation coefficients? 23 A. Right. 24 Q. Can you explain what that is? 25 A. So in addition to the F-stat, we have the t-tests 26 of the individual parameter estimates, and that's 27 basically a test of, is that parameter that you have 28 estimated, which is the -- you can think of it as a slope 3635 1 associated with the independent variables, like wage rates 2 and energy costs -- is that, in fact, statistically 3 different from zero, is it -- you know, does it have some 4 explanatory value in the model or does it not. 5 And then the correlation coefficients, those are 6 included instead of an R-square for the total cost because 7 we didn't actually estimate a total cost model. We 8 estimated a utility cost model, a labor cost model, and an 9 other cost model. Then we are summing up those predicted 10 values from our model, each of those individual three cost 11 models, summing those up to come up with our total cost 12 estimate. And then we're comparing that estimate from 13 those three equations to the actual total cost, and that's 14 the correlations that I'm talking about in that bullet 15 point. So the correlation between the predicted cost for 16 cheese and the actual cost was .92, where 1 is a perfect 17 correlation, so .92 is kind of getting at the same idea, 18 you are explaining a lot of the variation in the cost with 19 the model. 20 Q. Overall what was your conclusion as to whether the 21 model was doing a good job of predicting actual? 22 A. Yeah. So the fact that we had significant 23 regression F-statistics, we had a lot of individual 24 parameter estimates that were significant -- statistically 25 significant different from zero, and the fact that the 26 correlations of the overall cost predictions and the 27 actual costs were quite high, suggests the model does a 28 good job predicting actual costs. 3636 1 Q. And then if we go on to page 14, is this the 2 actual models? 3 A. So these are the actual models that we estimated. 4 And so you can see this is the cheese manufacturing cost 5 model. There's an -- just going through that first 6 equation for labor cost. There's a -- what we call an 7 intercept or a constant parameter; that's the a11. 8 There's a slope associated with -- with the manufacturing 9 wage rate; that's b11. And then there's a slope factor 10 c -- c11 associated with the labor productivity. And then 11 there's an error term. So the error term is, you know, if 12 we're predicting labor costs, are we above or below it. 13 Is the predicted version above it or below it, and that's 14 what the error term is. 15 Q. All right. And you devised a similar formula for 16 utility costs and other costs as well, correct? 17 A. Correct. 18 Q. And if we just go forward, you have a similar set 19 of modeling of equations for butter on page 15? 20 A. Correct. 21 Q. And then nonfat dry milk on page 16, correct? 22 A. Correct. 23 Q. And then tell us what page 17 is -- is discussing. 24 A. Yes. Page 17 is -- this is Table 3 from the IDFA 25 Exhibit 2, which was the actual paper that's been up for a 26 while. And this is basically all of the estimated 27 parameters. So if I go back a slide, if you are looking 28 at this, you know, a31, c -- it is the parameters 3637 1 associated with all those variables. 2 And so for cheddar cheese, for example, we have 3 got a constant that was estimated associated with each 4 cost component: Labor, utility, and other. We have 5 manufacturing wage and labor productivity in the labor 6 column. We have natural gas and electric prices in the 7 utility column. And then we have some of the dummy 8 variables like excess whey down there. 9 And the asterisks that are located to the right of 10 each of those estimates are an indicator of whether the 11 parameter estimate was statistically different from zero 12 at the -- there's two asterisks at the 5% level; if 13 there's one, it's at the 10% level. 14 And what does that mean? What does that 15 significance of the 10% level mean? It means, that if -- 16 if it's significant at the 10% level that there is only a 17 10% chance that you have erred in rejecting what's called 18 the null hypothesis, and the null hypothesis is that that 19 parameter is actually zero. So it's really not telling 20 you a lot other than the fact that we -- we think that 21 number is not zero, and we have been able to show that 22 statistically. 23 Q. Okay. So if we go to the next page, tell us now 24 how you took all of this information and analysis based 25 upon what happened between 2003 and 2016 or 2002 and 26 2016 -- 27 A. Yes. 28 Q. -- in California, how did you then forecast what 3638 1 manufacturing costs would be in 2022? 2 A. Right. So those estimated parameters on -- on the 3 screen there, those -- those were used to -- in concert 4 with data that is explanatory variable data in years 5 subsequent to 2016, to make projections forward at times. 6 So basically you multiply the data series -- and let's use 7 one example. 8 If I have got wage rates from 2017 to 2022, I have 9 got a parameter estimate associated with labor -- if I go 10 back -- a manufacturing wage rate associated with cheese 11 labor of .0049, right? So I'm going to estimate -- I'm 12 going to multiply that .0049 times the manufacturing wage 13 rate series that I have got in my dataset for 2017, for 14 2018. And doing that for all the parameters in the model, 15 you can then come up with a predicted value in those years 16 beyond the CDFA dataset. 17 Q. Okay. And you -- you performed that calculation 18 for each of the each of the variables you were tracking; 19 is that right? 20 A. That's right. 21 Q. And what was the end product of this effort? 22 A. Yeah. Again, reiterating the fact that for dry 23 whey what we did was we took the nonfat dry milk model 24 estimates and predicted values and added $0.03 to those 25 because we didn't have an estimated whey model. 26 One of the things that we found is that looking at 27 the 2022 numbers, they are all roughly around $0.10 a 28 pound higher than the 2006 numbers. So just to kind of 3639 1 give an idea on a per pound of product basis, kind of cost 2 change, that's -- that's kind of what it looked like. 3 So I have a table -- this was an extract from 4 Table 5. So this is part of Table 5 that's in the actual 5 report. I have pulled out, you can see the 2006 model 6 predicted value. That's not the actual CDFA model. Those 7 are in the earlier table that we went through. But I also 8 have 2016 listed there, which was the last year that CDFA 9 had audited data. And then the bolded numbers are all 10 predicted values going out -- or forecast values, if you 11 will, going out 2017 through 2022. 12 So you can see that the model had a predicted 13 value in 2006 of 18.66 for cheese. The actual value I 14 believe was 19.88. And this was a case where even within 15 the dataset the predicted value was lower that year. But 16 the predicted value is going out. You can see by 2022 we 17 have a predicted value of just over $0.30 a pound from the 18 model for cheese, cheese manufacturing costs. 19 Q. All right. And take us to the next page, page 20, 20 and tell us what that is showing. 21 A. Okay. So in addition to the -- to the models 22 we -- with the cost equations that we estimated, we also 23 just took a look at overall manufacturing costs and did 24 a -- estimated a trend. So this is just a straight line 25 that fits the data within the sample, and it is basically 26 done by taking the manufacturing cost and regressing it 27 against the year. So there's a constant term, then 28 there's a slope equation with the year, and then that will 3640 1 give you a predicted trend value. 2 And so we fit that trend and then, again, 3 projected those forward to make forecasts from the trend. 4 And this just shows what those would have been. And I 5 think what you -- what you see when you look at those 6 numbers is that certainly out in 2022, the trend values 7 are below what the model estimates are for manufacturing 8 costs. So the CDFA trend, for example, for cheddar cheese 9 was $0.27, a little over $0.27, and the model estimate was 10 $0.30. So the trend is $0.03 lower. 11 You go back, say, 2019, and the trend was $0.2573 12 cents, whereas the model estimate was $0.2521. So in that 13 one, the trend is actually higher than the model estimate. 14 So I -- the only reason I'm kind of making this 15 comparison is that the model is picking up the changes in 16 price levels or price variables like the PPI index or like 17 the manufacturing wage rate or like the energy cost. And 18 to the extent that those have accelerated in the last 19 couple years of the timeframe here that we're looking at, 20 say, 2021 and 2022, it's not surprising that the model is 21 picking that up whereas the trend wouldn't pick that up. 22 The trend is just dumb. It increases at a -- you know, 23 the same rate every year going forward, and so there's 24 some years where the trend is above what the model 25 predicts and some years where it's below. 26 Q. In the real world, have actual costs gone up in a 27 way that the model reflects? 28 A. Well, we had some testimony from some witnesses -- 3641 1 kind of just went through this slide. We have some 2 testimony from some other witnesses. So these are the 3 percentage increases that the model predicts. With cheese 4 and -- cheese costs in 2022 versus 2006 of an increase of 5 51.2%, whey 50.4%, butter 72.2%, and nonfat dry milk 6 59.4%. So that's what the model predicts. 7 And we have had some testimony -- and this is 8 just -- I was listening to most of this testimony, and, 9 you know, what I took away from it is that the AMPI 10 witness noted a 47% increase in the cost of manufacturing 11 bulk cheese from 2008 to 2022. We're showing a little bit 12 more than that from 2006 to 2022. Land O'Lakes noted a 13 combined 70.26% increase in the cost of manufacturing 14 butter and nonfat dry milk from 2007 to 2022. We had a 15 70-plus -- 72% increase on butter and a -- I forget what 16 the -- 59.4 on nonfat dry milk. So, again, similar range. 17 And then the Northwest Dairy Association had noted an 80% 18 increase in the cost of manufacturing products over, I 19 guess, it was -- I didn't have the time period in here. 20 I'm assuming it was the same general time period. 21 Q. Okay. And what utility do you believe this -- the 22 analyses you have performed have for purposes of USDA's 23 determinations of Make Allowances? 24 A. Well, I heard the question asked of Dr. Stephenson 25 a little bit earlier today, like, you know, was it better 26 to have an econometric model or was it better to have 27 actual plant data from surveys. And I would agree with 28 him, you know, plant data is -- is -- I would find that 3642 1 inherently more reliable than an econometric model. But 2 this is another data point, another way to look at 3 estimating those costs. And it seems to -- when used 4 alongside the cost testimony of other witnesses and 5 Dr. Stephenson's work, I think it -- it helps sort of fill 6 out the picture and -- and provide more corroboration of 7 the costs that are being talked about. 8 Q. Okay. And to -- going back to IDFA Exhibit 2, 9 which is Hearing Exhibit 180, is -- is -- is this a -- you 10 know, this is the entire report reflecting the work that 11 you did on this project; is that right? 12 A. Correct. 13 Q. And the PowerPoint presentation we have gone 14 through, which is Exhibit now 195, is that -- have you hit 15 the, sort of, if you would -- 16 A. The highlights. 17 Q. -- the highlights of that study? 18 A. Yes. 19 Q. Okay. 20 MR. ROSENBAUM: At this point, your Honor, the 21 witness is available for cross-examination. 22 THE COURT: Okay. Video -- did we lose the video 23 for a second? 24 All right. Cross-examination for this witness, 25 other than AMS? 26 CROSS-EXAMINATION 27 BY MS. HANCOCK: 28 Q. Good afternoon, Dr. Schiek. 3643 1 A. Good afternoon. 2 Q. You just ended by saying that you do agree with 3 Dr. Stephenson that a cost study model is a better data 4 source than using an econometric model, I guess. And then 5 I didn't hear the last part of it, which is where you 6 landed. 7 A. Oh, I was saying this is another approach, and I 8 think when used in concert with the other sources of data 9 testimony from individual companies about their costs and 10 Dr. Stephenson's work, it's just another point of 11 corroboration, I guess, a point of reference to kind of 12 understand where those costs are. 13 Q. Okay. And did you assist IDFA in putting together 14 their proposal for their Make Allowance increase? 15 A. Not in putting together their proposal. IDFA 16 approached me and asked if it -- if I thought it would be 17 possible to estimate costs from CDFA data. And I thought 18 about it for a while, and I said, well, we have a data 19 series; we probably could do that. And then they asked if 20 I would do it. So -- so my understanding is that they 21 utilized this work as well as Dr. Stephenson's work. 22 Q. Do you know how they used it for the proposal that 23 they made? 24 A. Other than looking at the numbers, my 25 understanding is they averaged the two cost estimates. 26 Q. Okay. So do you understand that they took 27 Dr. Stephenson's cost study results and then the 28 econometric results that you have put together, added 3644 1 them, divided by two, and then used that as the first-year 2 numbers that they are proposing? 3 A. I think they -- that would be the -- my 4 understanding, I don't know if I have got this right, is 5 that they -- that would be the final year of their 6 implementation schedule. So they would look at 7 Dr. Stephenson's '22 -- 2022 numbers, they looked at these 8 2022 numbers, as you described, added them, and divided by 9 two to create a simple average, and then that would be the 10 end year. And then they implemented it in a -- over a 11 four-year period, I think, so with a -- with a chunk at 12 the beginning and then equal steps after that. 13 Q. Okay. Thank you for that correction and 14 clarification. 15 Do you know how it was that they -- that they took 16 the average of yours and Dr. Stephenson's study that was 17 then divided by two to get to their full amount of the 18 Make Allowance increase, do you know how it was that they 19 backed out and got to year one? 20 A. Just from looking at it, I mean, I don't know -- I 21 don't know the exact process they went through, but from 22 looking at it, it looks like year one is about half of the 23 total increase. 24 Q. Okay. Just based on another percentage -- 25 A. That's what it looks like. 26 Q. -- allocation? Okay. 27 That's at least what you understand it to be when 28 you look at it? 3645 1 A. Yeah. 2 Q. So let's -- let's look at -- let's just take the 3 most recent CDFA manufacturing cost annual. And is this a 4 study or a survey? 5 A. I would say it's a survey of plants in California 6 regarding their manufacturing costs. And that was 7 audited, yeah. So they talk about it as a study. I use 8 the term study because that's how they refer to it here. 9 But, yeah, it's a survey of plant costs. 10 Q. Okay. Is there a difference between a study and a 11 survey? In your world, I guess I should clarify. 12 A. I think you could try to make one, but I think, 13 people could refer to a study -- a survey as a study. 14 Q. Okay. And I thought that I heard you say that the 15 CDFA data was audited. Did you say that? 16 A. It is. It was, yeah. 17 Q. Do you know when it was audited? 18 A. So -- okay. So my understanding of how that 19 practice worked is CDFA would request cost data from the 20 plants each of these years, and the plants would submit 21 their data. And then they had a team of -- it was the 22 cost -- manufacturing cost unit, a team of folks who would 23 look at the data, that would go into the plant, meet with 24 the folks, and ask questions about the data. They would 25 ask for documentation. They would look at those. And 26 then occasionally they would say, well, we think you need 27 to -- you know, you didn't include everything, so you need 28 to, you know, pull some more data from this area to get -- 3646 1 to ensure that the categories -- the cost categories were 2 accurate and to ensure that the data was accurate. 3 Q. If you look at Exhibit 156, which is the 2015 data 4 study, could you point to anyplace in here where it says 5 that this is an audited study? 6 A. Exhibit 156. Are we looking at the same one? I 7 have got 194 on this -- oh, was this introduced earlier as 8 156? 9 Q. This is the one that was introduced earlier, so 10 that's why my number is different. It is Exhibit 156. 11 A. I don't know that it says that it was audited, but 12 it says, "The auditors worked with plant management to 13 gather data on all aspects of the operation, review plant 14 records on site, and allocate plant expenditures to each 15 product manufactured by the plant. Studies are conducted 16 and developed in conformity with generally accepted 17 accounting principles, cost accounting techniques, and 18 instructions contained in the dairy marketing branch's 19 audit and cost procedures manual." That looks like an 20 audit to me. 21 Q. Okay. Other -- 22 THE COURT: Do we have a page number for that? 23 MS. HANCOCK: This is on page 3, your Honor. 24 THE WITNESS: Page 3 of -- yeah. 25 BY MS. HANCOCK: 26 Q. So is this what you are referring to when you say 27 that the information is audited? 28 A. Yes. 3647 1 Q. Okay. Has anybody from CDFA ever told you that 2 the information was audited? 3 A. Yes. 4 Q. Who was it? 5 A. Well, that would be the branch chief referred to 6 it as an audited study, and the head of that unit, Ed 7 Hunter, called it -- referred to them as audited. 8 Q. Okay. So it's your belief, then, that it is 9 audited based on those conversations and that statement? 10 A. That's correct. 11 Q. Okay. And you understand that the information 12 in -- that you collected or that you evaluated is all 13 based on California production? 14 A. Yes. 15 Q. And do you believe that California is 16 representative of the rest of the country? 17 A. Representative to the extent that California is 18 subject to a lot of the same -- first of all, they are 19 making the same products and subject to a lot of the same 20 cost influences. There are certainly differences, and 21 even back in 2006 there were differences between, you 22 know, Dr. Stephenson's numbers, from that time period -- 23 2005 to 2007 when he was collecting his information -- and 24 the 2006 CDFA numbers. So there are certain costs that 25 are different in California, you know, for -- for a while. 26 I think wage rates have been higher in California than 27 they are in other parts of the country. And -- but 28 there's variations between plants depending on their labor 3648 1 contracts, whether they are union or non-union, and what 2 kind of contract they have with their workers, so that 3 that kind of variation would be nationwide. 4 So I think it's -- it's representative. It may 5 not be exactly --you know, I wouldn't necessarily expect 6 that it is the exactly the same as U.S. weighted average 7 for sure. 8 Q. So you mentioned wage rates. Do you know what the 9 minimum wage is in California? 10 A. I believe it's $15 an hour now. 11 Q. I think it's $15.50. Does that sound right? 12 A. I'll take your word for it. 13 Q. Okay. Do you know what the minimum wage is in 14 Wisconsin? 15 A. I assume it's less, but I don't know what it is. 16 Q. It's actually 50%, so it's the federal minimum 17 rate. 18 A. Uh-huh. 19 Q. Does that sound right to you? 20 A. I -- I don't have a reason to dispute you. 21 Q. Okay. Do you have a way to factor into your 22 analysis that the minimum wage in Wisconsin is 50% of what 23 it is in California? 24 A. No. This is really a prediction of California 25 costs. 26 Q. Okay. 27 A. Because we're using California data to do it. 28 Yeah. 3649 1 Q. Okay. And extrapolate from that in a way that 2 would be -- 3 A. Reflective of California conditions. 4 Q. Okay. And so you're not suggesting that it would 5 be reflective of the rest of the country's conditions? 6 A. Not any more than the fact that they are operating 7 in the same environment -- you know, the same general 8 business environment because they are making the same 9 products. But, yeah, it's not -- the costs can be 10 different in other parts of the country for sure. 11 Q. And you would agree with me that there are other 12 parts of the country that have very different cost 13 structures than what California has? 14 A. I believe that's the case, yes. 15 Q. And in particular, in large part because of the 16 different regulatory schemes that -- regimes or schemes, 17 that have -- that have developed in California as compared 18 to the rest of the country? 19 A. Your -- I'm not sure which regulatory schemes you 20 are referring to, but if you are saying we seem to like a 21 lot of regulation in California, then I agree with that. 22 We have high costs -- we have a lot of regulations and 23 high cost -- higher cost of compliance. 24 Q. And businesses, whether it's the dairy industry or 25 other -- any other business in California, tends to pay a 26 higher cost because of the regulatory laws in California 27 that govern business transactions in general. Would you 28 agree? 3650 1 A. Oh, yeah. Yeah. 2 Q. Do you know why CDFA was conducting these 3 manufacturing cost studies? 4 A. Well, the primary motivation for it is my 5 understanding -- and, again they started doing this long 6 before I came to work at Dairy Institute -- but my 7 understanding is it was to support the determination of 8 the pricing formulas that they were using for the 9 manufacturing classes of milk. 10 Q. When they were under a state order system? 11 A. Right. Under the state order system. 12 Q. And that's why it stopped in 2016? 13 A. Correct. 14 Q. Because then it became the Federal Order 51? 15 A. Correct. 16 Q. Do you know how they were using the studies in 17 order to support their pricing systems? 18 A. They were using it as the basis for establishing 19 Make Allowances under -- under their system, and the sort 20 of primary benchmark was the weighted average 21 manufacturing cost. I think in the later years, latter 22 part of the series, it was common for the weighted average 23 manufacturing cost to simply become the Make Allowance. 24 Earlier years there was a little more discussion about 25 that. 26 Sometimes they might set a Make Allowance that was 27 above or below that number, depending on -- they were 28 looking at other factors, too, like, how much the volume 3651 1 of the plant -- of the volume of product produced would 2 have been covered by the Make Allowance, how much 3 dispersion there was among costs of -- sometimes you would 4 have costs that were bunched tightly together where 5 they -- the weighted average cost would be highly 6 reflective of all plants. 7 There were other cases where you might have more 8 of a spread, and then, I believe -- and I can't cite 9 specific examples, I'm just remembering, you know, hearing 10 discussions and decisions, but I believe there were some 11 cases where CDFA chose to implement a Make Allowance that 12 was either above or below the weighted average because 13 of -- just because of that dispersion, you might have an 14 outlier plant that had really, really low costs, and that 15 would have -- setting it at the weighted average might 16 have left many, many plants without their costs covered, 17 so they would maybe set it higher in that case, and in 18 another case they may set it lower because there were some 19 high cost outliers. 20 Q. Okay. So if I'm understanding you correctly, your 21 understanding of CDFA's use of this cost study that they 22 did annually was that they would use all of the data that 23 they collected, and then they would take real life 24 information that they have about the plants that were 25 operating in California, and make sure that it was 26 reflected in the data, and if it wasn't, they would make 27 additional adjustments, either above or below, and use the 28 totality of that information to set Make Allowance? 3652 1 A. Yeah. That's probably fair. I would -- I would 2 probably just say they would -- they would take into 3 account more than just the weighted average manufacturing 4 cost. They would look at other factors and establish the 5 Make Allowance. 6 Q. Okay. So they were really taking a 360-degree 7 comprehensive review? 8 A. They were taking a broader view, yes. 9 Q. Okay. But at the very core of it, was this 10 audited mandatory cost study -- 11 A. Correct. 12 Q. -- that they conducted on an annual basis? 13 A. Yes. 14 Q. Okay. And they did not ever use the econometric 15 modeling that you are using in order to set 16 Make Allowance? 17 A. No. They -- they collected the costs directly 18 from the plants. 19 Q. And the data that they collected each year was 20 even more comprehensive than what Dr. Stephenson collected 21 under his cost study as well? 22 A. More -- yeah. It was -- it probably represented a 23 higher percentage of the plants and the volume than his 24 study. 25 Q. And then also taking into account that totality, 26 that comprehensive approach that you just described? 27 A. Yes. 28 Q. Do you know what percentage of the California 3653 1 cheddar cheese or butter or nonfat dry milk and dry whey 2 were represented in those manufacturing studies? 3 A. I think it -- it varies from year to year, but 4 it's -- it's usually if you look at -- if you want to know 5 the numbers, it's usually in each of the summary tables. 6 I want to look at the nonfat dry milk study from -- I have 7 it marked as Exhibit 194, but it's your -- 8 Q. 156? 9 A. -- 156. 10 Q. And just so our record is clear, I don't believe 11 we have an Exhibit 194 yet. 12 A. Oh, okay. 13 Q. Because I think that was -- 14 A. 156 then. 15 So this is on page 8 of Exhibit 156. We're 16 talking about volume. There's a table, and then there's 17 some bullet points below the table but before the numbers 18 start. And I think that here it says the volume 19 includes -- let me back up go one higher. It's the first 20 bullet point, towards the end, it says, "The eight plants 21 processed 555.02 million pounds of nonfat dry milk during 22 the 12-month study period January through December 2016 23 representing 97.44% of the nonfat dry milk processed in 24 California." 25 Q. Okay. And -- ooh. 26 A. I was just going to say, and I think there's a 27 similar volume number reported for each of the 28 commodities, and I think they do that every year. 3654 1 Q. Okay. So greater than a 90% sample size; is that 2 fair? 3 A. In terms of volume, yeah. 4 Q. Okay. And did you hear Dr. Stephenson say that in 5 his cost study, for the one that -- the two categories 6 that he believed he had a good representation on, he 7 believed it was probably around 50%? 8 A. Yeah. 9 Q. And then there were two categories that fell 10 somewhere between 10 and 50%? 11 A. Yes, I -- I had heard the testimony. 12 Q. Yeah. And -- and then did you also hear him say 13 that sample size matters? 14 A. Yes. I heard him say that. 15 Q. Do you agree with that, that the sample size 16 matters in the accuracy of the information that's 17 reported? 18 A. Yes. The more representative your sample is, the 19 more representative it is of actual cost. So I would 20 agree with that. 21 Q. And we saw that when he had his 2021 study, and 22 then his 2023 study compared to that, the numbers were 23 different enough that he had to look into whether what 24 those differences were. Did you hear that testimony? 25 A. I heard that testimony, yes. 26 Q. And did you hear him have the conclusion that what 27 he -- what he had been able to figure out from that is 28 that it was just that the sample size matters? 3655 1 A. I heard him say that, yes. 2 Q. And so you can see that the sample size, at least 3 in that example, can make wide swings in information 4 depending on what data is collected? 5 A. I think it can. Sometimes -- sometimes smaller 6 sample sizes can be representative, but obviously the 7 more -- the more volume you have covered, the more 8 confidence you have that it's representative. 9 Q. Do you know how much it cost CDFA to conduct that 10 study every year? 11 A. I do not know. 12 Q. Okay. So I want to -- the totality of the 13 information that was used to compile your econometric 14 modeling stopped at 2016; is that right? 15 A. In terms of estimating the regression model, yes. 16 Q. Okay. And in terms of having the actual data that 17 populated your model? 18 A. Correct. 19 Q. And then after 2016, I think you described in your 20 summary in -- I'm sorry, I didn't seem to write down the 21 exhibit number on your PowerPoint. What's the exhibit 22 number on the PowerPoint? 23 THE COURT: 195. 24 MS. HANCOCK: Thank you. 25 BY MS. HANCOCK: 26 Q. Okay. And I think in your Exhibit 195, you had 27 provided a summary of the numbers on page -- on page 19 28 that shows you had the actual numbers, and then you have a 3656 1 bolded line, and then everything after that is just 2 predicted values based on your modeling; is that right? 3 A. Yeah. And actually the -- for example, if I'm 4 looking at what's -- what I have called Table 5a on that 5 page, where -- the table with the manufacturing cost model 6 predicted values, the -- I think I -- if I didn't make 7 this clear, the 2006 and 2016 numbers there are actually 8 the model estimates within the sample period where we had 9 the actual data. So those two aren't the actual costs. 10 After I did the regression model, what did the model 11 predict for that particular year? And then, of course, 12 after 2016 all we have are the model predictions because 13 there is no more actual CDFA audited cost data, so those 14 are the model estimate -- or projections going forward. 15 Q. Okay. So I remember you saying that, but it 16 didn't sink in with me in a way that I understood it. 17 A. No, but the -- where I talk about the percentage 18 increases, those are based on the percentage increases of 19 the predicted value versus the actual value that was in 20 the earlier table in that PowerPoint. 21 Q. Okay. 22 A. Just to draw a distinction. 23 Q. That's okay. And I'm just going to dive in a 24 little bit deeper just to see if I can understand it a 25 little bit better. 26 When you say that the 2016 and 2006 numbers that 27 are here are based on the model estimates, aren't the 28 model estimates, don't they originate with the actual 3657 1 numbers? 2 A. They do. But with a linear regression, even 3 within the period where you have the actual numbers, the 4 model is trying to fit a -- it's trying to compute a 5 linear fit of a model throughout the whole period. So 6 sometimes you are not going to hit the actual value 7 exactly, you will be above or below it with your estimate, 8 because it's a linear estimate, so it -- it -- sometimes 9 the costs don't necessarily go up in line with the model. 10 And so you will have some periods where you have an error 11 on the positive side and sometimes there will be an error 12 or the negative side. But the model is constructed in 13 such a way to minimize those errors so you get the best 14 fit possible of the data with your regression model. 15 Q. Is that a way that you can verify whether your 16 regression model is working properly, is when you do -- 17 when you do the lookback period where you had actual 18 numbers, that you can see if you come close to where those 19 numbers are? 20 A. Yeah, that's one way of thinking at it -- thinking 21 about it. But the bottom line is, you know, you expect 22 you are not going to hit it with perfection. I mean, it 23 is just not -- you're not going to -- unless it has some 24 sort of straight line, you know, the cost line up in a 25 very straight line. But you are going to fit a model 26 using this technique that has the -- does the best job of 27 sort of threading that needle. 28 Q. Okay. And in this modeling that you are doing 3658 1 is -- you heard Dr. Stephenson talk about how this is not, 2 in his view, the best way to predict Make Allowance costs 3 because you don't have the opportunity to factor in things 4 like specific product variations? 5 A. Yeah. I -- I heard what he said, and I don't 6 disagree with it. 7 Q. Okay. And then in addition to not being able to 8 factor in product variations, you also don't have the 9 ability to factor in different productivity measurements 10 at the plant. For example, if a plant was newly 11 constructed and was much more efficient, that's not 12 something that this model could take into account. 13 A. No, this model wouldn't take into account 14 individual productivity gains at an individual plant. You 15 wouldn't -- you wouldn't see that, at least, in the 16 forecast. Now, it would capture some of that if it's 17 happening during the period where you are estimating the 18 data, so it would capture some of that, in the 2002 -- or 19 2003 to 2016 period. But it -- it wouldn't have the 20 ability to capture specific plant innovations or, you 21 know, even brand new technology that happened subsequent 22 to that period where that -- where -- where the sample 23 period is. 24 Q. Okay. So I'm just going to paraphrase to make 25 sure I understand what you are saying. You are saying 26 if -- for example, if a plant today was constructed and it 27 was able to capture a 20% more efficiency in productivity, 28 if in the window of time between 2006 and '16 a plant had 3659 1 been constructed that similarly captured a 20% improvement 2 in efficiency, so it was in your input window, it could be 3 forecasted out in your modeling in a way that would still 4 capture it in your later time period? 5 A. It could capture some of that, yeah. 6 Q. Okay. But we don't really know as we look back if 7 that is the case; is that fair? 8 A. That's fair. 9 Q. And there's not really a way to ensure that this 10 modeling is capturing new innovation that might not have 11 occurred during that same time period? 12 A. That's correct. 13 Q. So if we become extra efficient now with AI or 14 some other new technologies that could be implemented in a 15 plant, robotics, whatever it might be, if it hadn't 16 happened at the same efficiency level previously, there's 17 no way for it to be captured? 18 A. Yeah. I think you are right. That is -- I mean, 19 that is the limitation of this kind of analysis. 20 Q. Okay. And then if we go to the next page in 21 Exhibit 195 on -- you have a trend line. You use -- this 22 is just the two methods that you used in your evaluation; 23 is that fair? 24 A. Yes. 25 Q. Is a trend line taking the input period and just 26 drawing a straight line forward? 27 A. Exactly. That's all it is. 28 Q. Okay. And I need to look at your -- do you have 3660 1 your Exhibit 180 in front of you? 2 A. Exhibit 180 is the -- is the report? 3 Q. IDFA Exhibit 2. Yeah. 4 A. Yep, I have that. 5 Q. Can we turn to page 12? 6 Is table 5 here capturing the data in both your 7 predicted modelling and then that straight line trend 8 analysis? 9 A. Yes. 10 Q. So if I drew a line down the middle of the page it 11 would be right under the word linear, and that would 12 separate those two? 13 A. Correct. 14 Q. Okay. And if I look at 2021 for cheese, for 15 example, we have $0.2707 that your model has predicted; is 16 that right. 17 A. .2707, yeah. That's correct. Right. 18 Q. And then if we look at what the trend line is, 19 that's 26.78% -- I'm sorry, $0.2678. 20 A. Right. 21 Q. And then fairly close to what your model 22 predicted -- 23 A. Correct. 24 Q. -- is that fair? 25 A. Yep. 26 Q. And then if we compare the whey column, that's 27 similarly close; is that right? 28 A. Correct. 3661 1 Q. And then the butter column is pretty close as 2 well, $0.2201 and $0.2193? 3 A. Correct. 4 Q. And then nonfat dry milk, pretty close as well at 5 $0.2447 and $0.2395? 6 A. Correct. 7 Q. Is it a fair conclusion to say that your model 8 that you created is just about as accurate as a straight 9 line linear trend model as well? 10 A. Well, the numbers you looked at, yeah. I think -- 11 I think the trend and the model estimates are fairly 12 close, and I think that's because there is a fairly strong 13 trend in the cost line from 2003 to 2016. 14 I think where you are seeing the divergence is 15 when there is a big change in those explanatory variables, 16 so we -- we begin to see -- in 2021, we had inflation take 17 off, and we begin to see more elevation in the cost. And 18 I think if you look back at, as I pointed out earlier, to 19 2019, the model costs were lower than the trend. And I 20 think we were picking up the fact that those price numbers 21 were -- you know, cost of energy and cost of materials and 22 those sorts of things were not moving up, in fact, had 23 eased because the Fed was -- well, we -- we -- the Fed had 24 raised interest rates in late 2018, and that led to a 25 slowdown in the economy in 2019, and so we didn't see 26 those price variables accelerating as fast. The model is 27 capturing some of that, the trend is not. 28 So, you know, there's going to be times where the 3662 1 model will be close to trend, there will be times where 2 the model estimates will be below, and there will be times 3 where the model estimates are above. 4 Q. Okay. And neither one of which is necessarily 5 based on actual market conditions? 6 A. Well, the model does take into account the 7 economic conditions that we have included as explanatory 8 variables, like wage rates, the Producer Price Index or 9 the -- you know, what we are using as a proxy for material 10 costs, energy costs. The trend does not take any of that 11 into account. 12 Q. Okay. And while your model can take it into 13 account, it doesn't provide an actual adjustment for it, 14 does it? 15 A. It provides an adjustment for it that's related to 16 how actual costs adjusted during the sample period. 17 Q. For California? 18 A. For California, correct. 19 Q. Okay. Do you agree with me that -- and maybe you 20 have already, but I'm just going to make sure I have asked 21 this -- do you agree with me that it's important for 22 manufacturing costs that the USDA uses to set 23 Make Allowances are accurate and current? 24 A. Yeah. I think it -- it is -- that's the goal, 25 right, is to set manufacturing allowances based on current 26 and accurate costs. 27 Q. When you were -- when you were calculating -- when 28 you were calculating your nonfat dry milk, you -- you 3663 1 didn't have the data in CDFA's materials, at least not for 2 all the years; is that fair? 3 A. Are you talking about the dry whey? 4 Q. I'm sorry. When you were calculating the dry 5 whey, you did not have all the figures from the CDFA data 6 for all the years? 7 A. I had all the figures there were. They just 8 didn't do the study -- they only did the study for four 9 years on dry whey. 10 Q. So you took the difference between nonfat dry milk 11 and dry whey that's in the current Make Allowance and used 12 that to establish your dry whey number and the numbers 13 that you put together? 14 A. For dry whey, yeah. 15 Q. And it's $0.03? 16 A. Correct. 17 Q. And so it's nonfat dry milk plus $0.03, and then 18 that's how you calculated the dry whey number? 19 A. Uh-huh. 20 Q. Is that a "yes"? 21 A. Yes. That is a yes. 22 Q. But Dr. Stephenson, if you look at his spread in 23 2023 between dry whey and nonfat, it has $0.0611; is that 24 right? 25 A. I -- let's see. I don't think I have that in 26 front of me, but I'll take your word for it. 27 Q. Did you review his numbers? 28 A. I looked at his numbers. I just don't have them 3664 1 in front of me right now. 2 Q. Did you consider using the spread that he came up 3 with as the spread for your calculation? 4 A. No. I -- when I was doing my work, we didn't have 5 his most recent numbers. And because the CDFA data used a 6 different way of allocating unallocated costs than his 7 2021 numbers, so I didn't use those because that was a 8 different methodology. So, you know, I just went with an 9 estimate that -- $0.03 is a number I have heard in the 10 industry that is representative of incremental drying 11 costs, whey versus nonfat dry milk, so -- and it -- it was 12 the approximate difference in the Make Allowance. So 13 those -- that was the reason I used it. 14 And I don't know if that number is accurate or 15 not. I -- I have asked people who are knowledgeable about 16 whey manufacturing, if that number still makes sense or 17 has it gone up, has it gone down. And, you know, I kind 18 of get, well, that might be -- yeah, might be $0.03. You 19 know, it's kind of -- I don't -- I don't -- I can't put a 20 lot of faith in whether $0.03 is the right number for 21 incremental drying costs or not, but that's -- that's kind 22 of what's in the current Make Allowance, so that's what I 23 used. 24 Q. Okay. Is it fair to say for your modeling it is 25 essentially a placeholder in that column for -- I should 26 say -- strike that. Let me say that again. 27 Is it fair to say that in the numbers that you 28 have concluded, it is somewhat of a placeholder based on 3665 1 the modeling that you did? 2 A. I don't know that I would use the word 3 "placeholder." It is an estimate using information 4 outside the modeling work that I did. It's, you know, 5 sort of using an external number that we think is 6 representative of that incremental drying cost, but it's 7 not -- it wasn't an analysis, an estimated whey cost or 8 estimated the cost of dry whey. It's just what -- you 9 call it a bootstrapped method maybe. 10 Q. Okay. Bootstrapped instead of placeholder. 11 A. Placeholder, yeah. Isn't that an important 12 distinction, though? 13 Q. And you said that at the time that you did your 14 modeling, you didn't have Dr. Stephenson's 2023 survey? 15 A. Correct. 16 Q. You only had his 2021 survey? 17 A. Correct. 18 Q. And if you would have used the numbers from his 19 2021 survey, the difference would have been instead of 20 plus $0.0611, the number would have been a negative 21 $0.0283. Does that sound right? 22 A. Yeah. And I remember that was one the things that 23 gave me pause, because there's very little -- it's hard to 24 understand how it could cost less to dry a product that is 25 more dilute than a product that is less dilute. And the 26 fact that it had a lower drying cost could be possible if 27 the dry whey plants were larger and more efficient than 28 the nonfat dry milk plants, but that's not -- that doesn't 3666 1 jive with my understanding of how dry whey plants operate 2 versus nonfat dry milk plants. 3 So, yeah, that -- that was one of the numbers from 4 the way he had allocated costs in that study that gave 5 some folks in the industry pause, as to maybe that 6 allocation cost isn't working quite right because that 7 doesn't -- that doesn't make sense to those of us who -- 8 who have kind of watched and think about how much -- how 9 much it costs to dry whey versus nonfat dry milk. 10 Q. Okay. And so the fact that his 2021 numbers 11 showed that the cost to dry whey was $0.2650, but then the 12 cost of manufacturing nonfat dry milk was $0.2933, that 13 didn't -- it didn't sound right to you? 14 A. Didn't sound right. Exactly. 15 Q. Did you hear him earlier when he said that there 16 were concerns in the industry about his 2021 butter number 17 as well? 18 A. Uh-huh. 19 Q. Did you have those similar concerns? 20 A. When I looked at those numbers when they were 21 released, I -- you know, I saw the cheese and the whey, 22 and I was like -- or the cheese number and the whey number 23 were kind of like, yeah, I could see those costs, those 24 sound like in the ballpark. But the butter/powder seemed 25 out of whack. You know, it seemed like, okay, this is 26 very different because the butter cost went down, 27 substantially, and the nonfat dry milk went up 28 substantially, and that that just didn't look right. 3667 1 But then, again, you know, I'm -- I'm most 2 familiar with the CDFA data, and they allocate those costs 3 on a solids -- total solids basis, which I understand is 4 closer to industry practice than the transformation 5 approach that Dr. Stephenson was using. 6 Q. Okay. So do you believe that the transformation 7 valuation allocation method caused some of the issues that 8 you were seeing in the numbers? 9 A. That was -- you know, without knowing the data -- 10 I mean, he's looking at the data, and he knows what plants 11 are included and that kind of thing. Just looking at the 12 results, that would be -- that would have been my 13 hypothesis as to why those numbers looked odd. 14 Q. Did you hear him testify, though, that he looked 15 into that to see if that was the cause of what was driving 16 the differences in the numbers? 17 A. I heard that, yes. 18 Q. And did you hear him also say that he concluded 19 that it wasn't that, but it was the sample size? 20 A. I heard him say that, yes. 21 Q. Do you trust that what he said is accurate? 22 A. I would say he knows the work he did better than I 23 do. So, yeah. 24 Q. Okay. And -- and then you understand that he 25 redid that study on behalf of IDFA for 2023? 26 A. Yes. 27 Q. And the numbers that he got in 2023, did that seem 28 to fit better with what you would have expected? 3668 1 A. Yeah. I mean, in terms of the relationship 2 between butter and powder particularly, yeah. 3 Q. Because the delta between dry whey and nonfat dry 4 milk is now three times difference; is that right? 5 A. Well, yeah. But it's -- the dry whey costs are 6 higher, which is what I -- more what I would expect. 7 Higher than nonfat dry milk. 8 Q. And so almost a $0.09 difference from 2021 to 9 2023; is that right? 10 A. I don't have the numbers in front of me, so I'll 11 take your word for it. 12 THE COURT: How much more do you have left? We 13 have been going for about an hour and a quarter, 14 20 minutes, something like that. 15 MS. HANCOCK: I have a little bit more. Do you 16 want to take a break? 17 THE COURT: Yeah, let's take a break. 18 Let's come back at 2:51. 19 (Whereupon, a break was taken.) 20 THE COURT: Back on the record. 21 Your witness. 22 MS. HANCOCK: Thank you. 23 BY MS. HANCOCK: 24 Q. Dr. Schiek, when you look at the -- let me say it 25 this way: How -- what percentage of accuracy would you 26 place on your modeling? 27 A. That's a -- what percentage of accuracy? I'm not 28 quite sure how to answer that or how you are defining 3669 1 accuracy. 2 Q. Let's say it's a cheese plant in Wisconsin. What 3 percentage of accuracy would you assign to the numbers 4 that you have modeled in your materials to their 5 manufacturing costs of cheese, cheddar cheese, in 6 Wisconsin? 7 A. I don't know that I could put a number on that. I 8 think, you know, as I said, we're using California cost 9 data. We're using explanatory variables that we think 10 impact the cost -- or are representative of cost changes 11 in California, so these are essentially estimates of 12 California costs. So to the effect that cheese plants in 13 Wisconsin are influenced by the same cost pressures and 14 cost factors that would be influencing dairy manufacturing 15 plants in California, I would say cost increases could be 16 related, the actual cost levels might be different. 17 Q. Okay. So the overall trends or percentages of 18 growth, if you knew Wisconsin's starting point, you would 19 be able to maybe assign that same percentage of growth to 20 get to a more accurate number; is that a way to describe 21 it? 22 A. That -- that probably would be a way of looking at 23 it in terms of comparing Wisconsin and California. Yeah. 24 Q. That would get you closer to an accurate number 25 than just assigning the model number results from 26 California to Wisconsin; is that fair? 27 A. I think depending on the premise of your -- based 28 on the premise of your question, if -- if I know upfront 3670 1 that the costs are different, and I say that costs are 2 different, but I believe they are subject to the same kind 3 of inflationary pressures, then, yes, I would say 4 you're -- the way you have presented it to me is accurate. 5 But I would have to know that to begin with, and I don't 6 know that I know that. 7 Q. Okay. And you didn't know it when you did your 8 modeling either; is that right? 9 A. Correct. 10 Q. Okay. And if you -- as an economist, if you were 11 trying to make an accurate prediction, to what decimal 12 number would you like to go out to for accuracy purposes? 13 A. Yeah, I don't know who started this. I guess it 14 was CDFA started going out to four decimal points, and so 15 that's what we have data on. And so when we do the 16 forecasts, they are coming out at four decimal places, 17 too. But, in reality, you know, if you are within a half 18 cent, I would think that's probably pretty close. 19 Q. Okay. So just to the tenth of a cent? 20 A. Yeah. 21 Q. Are you confident in the -- enough in the results 22 of your modeling that you would use those numbers to set 23 Make Allowance? 24 A. You know, I think if you had -- in the absence of 25 an audited cost study, yes. 26 Q. Okay. Do you agree that if you had an opportunity 27 to have an audited cost study that was mandatory, that 28 would be a better way to set Make Allowances? 3671 1 A. I think in the long run that would be a better way 2 to set Make Allowances. 3 Q. Would it also be a better way to set it in the 4 short term as well? 5 A. If you have it, it is better. I mean, I think 6 what we're -- what we're running into here is that it's -- 7 it's been many years, I've forgotten the number now, 8 2006 -- 2008 was when he changed it based on 2006, 2007 9 data. And I -- I could say with some confidence costs 10 have changed, and I think we have heard that from 11 processors and co-ops who have testified. 12 So the question is, you know, what's -- what's the 13 risk of waiting to update what is an outdated number 14 versus, you know, the accuracy gained that you would get 15 by waiting for an audited number. I think that's really 16 the question that the Department's going to be dealing 17 with. 18 Q. Okay. So making an adjustment now, even if it's 19 not accurate, is at least better than no adjustment now; 20 is that right? 21 A. I -- I would say you really want to know what your 22 best estimate of costs, current costs are, that that 23 should be what you are aiming for. And then how you 24 implement that, that's a policy question. 25 Q. Okay. And it's fair to say if we're looking on, 26 that you agree that the best way to set them would be on 27 an accurate number, right? 28 A. Yes. 3672 1 Q. And -- and if we look at kind of prioritizing 2 which one is the most accurate, the most accurate would be 3 a mandatory audited cost survey. Would you agree? 4 A. I -- yes. I think that's -- that's accurate, for 5 sure. 6 Q. And then in the level of priority, the next one 7 would be an actual cost survey, such as ones that 8 Dr. Stephenson has conducted, if you don't have a 9 mandatory survey. Would that be fair? 10 A. Yeah. I think, in general, real plant data is 11 preferable to -- to this approach in terms of -- 12 Q. And when you say -- 13 A. -- econometric approach -- 14 (Court Reporter clarification.) 15 THE WITNESS: I think the question was, would a 16 cost survey be preferable to an econometric model like I 17 have estimated; is that accurate? 18 BY MS. HANCOCK: 19 Q. That is correct. 20 A. Okay. And I was saying, I think you would always 21 prefer to have actual plant data from a survey as long as 22 it is, I guess, broad enough to encompass and be 23 representative, and then the econometric model would be 24 something that would be employed in support of -- of the 25 plant data. 26 Q. Okay. And so then if we have the mandatory 27 audited cost survey as the number one way that we could 28 achieve accuracy, and then a cost survey below that -- 3673 1 A. Uh-huh. 2 Q. -- such as the one that Dr. Stephenson has 3 conducted, either in 2021 or 2023, and then your modeling 4 would be below that; is that right? 5 A. In order of preference for -- for policy purposes 6 in setting Make Allowances, yes. 7 Q. Okay. So would you agree with me that by adding 8 your numbers in the modeling to Dr. Stephenson's 2023 9 numbers, and then taking an average of those two, that it 10 is something less than or less valuable than just looking 11 at his alone? 12 A. I think it depends on how confident you are that 13 his numbers are representative of the current conditions 14 in the marketplace. And, you know, I think one of the 15 things I said I think in the study that, you know, his -- 16 his estimate for the manufacturing costs for cheddar 17 cheese is lower than the estimate for 2022 that we came up 18 with in my modeling. I think all the other ones, dry 19 whey, nonfat dry milk, butter, his estimates are all 20 higher than what I came up with. 21 And, you know, I don't know, I -- I just -- I kind 22 of track these costs through 2016. I have a -- talk to 23 people, I don't know that, you know -- I can't say that 24 definitively that his cost estimates in all of the 25 commodities are -- are a better representation of where a 26 cost level is today than -- than -- than my estimates. 27 Q. And that's because based on your experience in the 28 industry, you see some anomalies in his number that don't 3674 1 match up with what you would expect to see in his actual 2 numbers; is that fair? 3 A. I would say some of those numbers seem a little on 4 the high side to me, yeah. 5 Q. Okay. 6 MS. HANCOCK: That's all I have. Thank you so 7 much, Dr. Schiek. 8 CROSS-EXAMINATION 9 BY MR. MILTNER: 10 Q. Good afternoon, Dr. Schiek. 11 A. Good afternoon, Mr. Miltner -- or should I say 12 Squire Miltner? 13 Q. You can say what you like. 14 Shall we put some folks to sleep? 15 A. I think it's too late for that. 16 Q. I'm looking at pages 7 through 9 of your full 17 report, Exhibit 180. 18 So at the bottom you have kind of a legend, and 19 you have two asterisks means the estimated parameter or 20 regression statistic is significant of the 5% level, and 21 then one asterisk means the same at a 10% level. 22 A. Correct. 23 Q. And in more basic terms, what does that mean? 24 A. More basic is probably going to put people to 25 sleep even faster. 26 So when we talk about statistical significance, 27 the way these tests are set up, so for the regression 28 statistic, which is I'm talking about the F, the F-stat 3675 1 here that you see, so same with each equation, the -- 2 there's a -- there's a hypothesis that's assumed, we call 3 it the null hypothesis, and that's the model that you have 4 just estimated has absolutely no ability to explain 5 variations in the variable, the dependent variable that 6 you are trying to measure, which in this case would be, 7 let's say, labor costs. 8 So what significance at the 5% level means is that 9 there is less -- if I say I'm going to reject that null 10 hypothesis in favor of the alternative, which it does have 11 some power to explain the variations in the -- in the 12 dependent variable, there's less than a 5% chance I'm 13 wrong if -- if I reject that null hypothesis. So 14 that's -- that's what the 5% is. 15 And so for 10%, which is a lower -- a less 16 rigorous statistical threshold, it's saying there is less 17 than a 10% chance that I'm wrong in rejecting the null 18 hypothesis. 19 Q. If I then look at your formulas or your equations, 20 under the cheese manufacturing cost model, beginning with 21 the first equation, labor cost equals an A constant -- 22 A. Uh-huh. 23 Q. -- plus a B constant times the manufacturing wage, 24 plus the C constant times labor productivity factor, plus 25 an E constant, there are two asterisks by your B, 26 constant, but there are no asterisks by your other factors 27 there. 28 A. Correct. 3676 1 Q. Does that mean that you cannot attest to the 2 significance of those other constants? 3 A. So, yes. What that means is that only the 4 parameter estimate associated with the manufacturing wage 5 rate was statistically significant at the 5% level, and 6 the other two were not statistically significant at the 7 10% level. 8 They may have been -- you know, if you relaxed the 9 constraint, they may have been significant at some other 10 level, but they are not significant at that level. 11 Q. Okay. So further to the right of that equation, 12 you have your F-statistic and your adjusted R-square. 13 And you mentioned the F-statistic, but if you 14 could, recap again for us what that F-statistic 15 represents. 16 A. Right. So that's the regression statistic. It 17 basically says that this equation that we have estimated, 18 the null hypothesis associated with that equation is that 19 it does not explain the variation in the labor cost 20 variable. And what this statistic value of two asterisks 21 means is it's significant at 5% level. If I reject that 22 null hypothesis, I would -- I'm -- I have less than 5% 23 chance of making an error by rejecting that. 24 Q. If you turn forward in -- in this exhibit to 25 page 19. 26 And at the top half of the page, I guess -- you 27 state it right there -- this is an ANOVA analysis of the 28 cheese labor estimate equation, correct? 3677 1 A. Correct. 2 Q. Where you have kind of a, I don't know, a quarter 3 of the way down, F, 24.432, is that the F calculation, the 4 F-statistic for this particular formula? 5 A. Yes. 6 Q. And so back on page 7 you didn't -- you didn't put 7 the F-statistic in there, you just indicated whether it 8 was significant at 5%, correct? 9 A. Correct. 10 Q. Does the actual number of the F-statistic have any 11 significance? 12 A. Well, the way this F-statistic is constructed, the 13 higher it is, the actual, you know, smaller your chances 14 are of making a -- what we call type one error, rejecting 15 the null hypothesis when it, in fact, was true. 16 Q. So a higher F-statistic is better -- 17 A. Better -- 18 Q. -- as far as being a predictive equation? 19 A. Correct. 20 Q. So back on page 7 next to that, you have the 21 adjusted R-square? 22 A. Correct. 23 Q. And I could try to explain what I think that 24 means, and I think I would be right, but I wouldn't do it 25 cogently. So can you help us with the adjusted R-square? 26 A. Okay. So the R-square is a measure of fit of the 27 model in terms of how much of the variation in the 28 dependent variable -- in this case let's use the labor 3678 1 cost as the example -- how much is explained by your 2 regression model. And the R-square is around 80%. 3 The adjusted R-square is a way of comparing models 4 that have different numbers of variables in them -- there 5 are a different number of -- yeah, numbers of explanatory 6 variables in them, so that you can always improve your 7 R-square by adding more regressors, more explanatory 8 variables, but you start losing efficiency, you start -- 9 over time. 10 And so the adjusted R-square is a way of 11 essentially penalizing you for the extra variables you are 12 adding and whether they really contribute much to the 13 explanatory power. And that's why it's kind of a better 14 measure of fit to use when you are comparing different 15 models than the straight R-square. So that's -- it's 16 another measure of fit that adjusts for the number of 17 variables that you are using to explain the -- explain 18 variations in the dependent variable. 19 Q. I want to give you my understanding of what that 20 means for the cheese manufacturing labor cost equation and 21 let me know if you agree with it. 22 The adjusted R-square of .77 means that if you had 23 an actual observation on the labor cost at a cheese plant 24 and a predicted labor cost for that same plant, 77% of any 25 variance is explained by the model; is that right? 26 A. Correct. 27 Q. Okay. So further down where you kind of have -- 28 on page 7, what I have called a legend for the various 3679 1 abbreviations in the equations, I want to look at the 2 outside factors that get pulled in. 3 So if I start with MFG wage, you explain what that 4 is. That's a California specific wage figure, correct? 5 A. Correct. 6 Q. And but labor productivity, lab pro, that's a 7 national average, correct? 8 A. It is, yes. 9 Q. Similarly, natural gas, you have chosen a 10 California specific measure of natural gas prices, 11 correct? 12 A. Correct. 13 Q. Same for electricity, that's California specific? 14 A. Correct. 15 Q. Okay. Turning to the next page, US PPI is 16 obviously a national figure. 17 A. Uh-huh. 18 Q. Food TFP, that's also a national figure, correct? 19 A. It is, correct. 20 Q. I think, the rest are dummies. 21 So it really is the labor and the energy costs 22 that are California specific numbers in your equations, 23 correct? 24 A. Correct. 25 Q. Okay. Here's where the No Doz would be helpful. 26 I want to go through the labor equation for 27 cheese. So if I -- if I take that equation, and I have 28 the first constant is a11, which if I look at page 9, and 3680 1 Table 3, I think that that is 0.0116, correct? 2 A. Correct. 3 Q. And then if I move further in the equation, I now 4 have b11. And if I look at Table 3, I think I'm supposed 5 to pull 0.0049; is that correct? 6 A. Correct. 7 Q. And then that B constant will be multiplied times 8 the manufacturing wage which you reference, correct? 9 A. The .0049? 10 Q. Yes. 11 A. Yes, that's correct. 12 Q. Okay. The next term is c11. And I think that 13 that is a negative 0.0004. 14 A. Correct. 15 Q. And then you multiply that times lab pro, the 16 non-farm labor productivity index, correct? 17 A. Correct. 18 Q. Okay. The next thing I have is e11. Now, your 19 E-constants are regression error terms, and I have not 20 been able to locate the e11 constant. 21 A. So e11 is not a constant. It is not a parameter. 22 It is just the error term. So it's -- it's -- so the 23 predicted value is everything before the error term. 24 Q. Okay. 25 A. The error term only measures the distance between 26 the predicted value and the actual value. 27 Q. Okay. Very good. I promise I won't go through 28 all of these, but let's look at utilities for nonfat dry 3681 1 milk. 2 A. Okay. 3 Q. So the first constant -- I'm pulling these, again, 4 off of Table 3 -- a32 is 0.0408, correct? 5 A. Yes. 6 Q. And then the B constant there would be 0.0008, 7 correct? 8 A. Correct. 9 Q. And -- and the C constant is 0.0094, right? 10 A. Correct. 11 Q. Okay. Have you had a chance to review the 12 testimony that Mike Brown is going to present on this 13 topic? 14 A. I have not. I have seen a couple of pieces of it 15 but, no, not in its entirety. 16 Q. I believe that he's going to testify that IDFA 17 recommends blending Dr. Stephenson's report and your 18 report weighted equally to set Make Allowances. 19 Is that your understanding of what IDFA is 20 proposing? 21 A. That's my understanding, yeah. 22 Q. So Ms. Hancock asked you some questions about the 23 relative costs of labor and energy in California versus 24 the U.S. And I don't recall if she asked if you had 25 looked at the differences in labor for the term or -- the 26 labor costs that you specifically reference for California 27 and how that compares to the same labor costs nationwide. 28 Did you examine that? 3682 1 A. Earlier on in doing this work, I estimated the 2 model -- it's not quite the same thing -- but I estimated 3 the model using national cost numbers, just to see, does 4 it make a big change in the parameter estimates. 5 Q. What did you conclude when you looked at that? 6 A. I concluded that, while the projected costs were 7 lower using the national numbers, they were -- I'll use 8 the term in the same ballpark. They were anywhere from a 9 half a cent to maybe a cent to a cent and a half lower. 10 They weren't an order of magnitude lower. 11 So that would be -- the hypothesis there would be, 12 if you were looking at California costs that were 13 increasing or changing at a rate that the national average 14 costs were changing, that would be what the model would 15 predict. 16 Q. Where you describe the MFG wage, you state that 17 that is the annual average hourly earnings for California 18 non-supervisory manufacturing workers, and so BLS lists 19 both a median and a mean. You said average. So I assume 20 that means you took the mean? 21 A. That was the mean, I'm pretty sure, yeah. 22 Q. Okay. So if I told you that California's rate 23 is -- mean rate is 125% of the U.S. mean rate, does that 24 sound reasonable to you? 25 A. Yeah, it wouldn't surprise me. 26 Q. So if we go back to the labor formula for cheese 27 that we just walked through and I plug in California's 28 mean rate, and then I plug it in for Michigan where 3683 1 Midwest Cheese is located -- 2 A. Uh-huh. 3 Q. -- that labor difference there, as I run it 4 through, I come to about $0.016 a pound. Does that sound 5 reasonable? 6 A. I can't -- I can't tell just from your 7 description. So how much -- give me -- if you wouldn't 8 mind, could I have the Michigan labor difference again? 9 Q. Well, again, I took -- I took the national mean of 10 18.68 and the California mean of 23.40, so it was 125% 11 difference. I used the -- 12 A. Yeah. 13 Q. -- national average versus California's. 14 A. Okay. So you are saying it is a cent? 15 Q. I'm saying that particular -- that particular 16 piece of your equation, that just the B constant times the 17 manufacturing wage, has an impact of about $0.016 as I ran 18 it through. 19 A. Yeah. That -- that may be the case. I -- I don't 20 know. The -- the -- I think the real comparison would be 21 not to run that number on a -- you know, a model of the 22 California dataset. But if you -- if you could run it on 23 the Michigan dataset, you know, if you are using Michigan 24 costs, you may have came up with a number that's lower 25 or -- or if the Michigan plant costs were higher, you'd 26 come up with a -- could come up with a number that's 27 higher even if the wage rate was lower because they may 28 have other costs. 3684 1 But, yeah, if you are just isolating that cost and 2 you are using California data that was estimated with 3 California wage rates, and then you are plugging in a 4 Michigan wage rate that's lower, yeah, I mean you could 5 have an impact that -- that is significant. 6 Q. So I want to ask also then about the California 7 utility numbers. And if I look at -- first of all, were 8 the electric -- the electricity factors in your formulas, 9 are they per kilowatt hour price? 10 A. I believe so, yeah. 11 Q. And then for natural gas, would that be per 12 thousand cubic feet? 13 A. Right. That's correct. 14 Q. Would it seem reasonable to you if, on average, 15 U.S. industrial electric rates -- let me rephrase that -- 16 California industrial electric rates are 116% higher than 17 the U.S. average for all other states? 18 A. California electric rates have increased a lot. 19 So, yes, that wouldn't surprise me. 20 Q. And for natural gas, California is 84.4% higher 21 than the national average. Does that sound reasonable? 22 A. That wouldn't surprise me either. And those are 23 industrial rates. 24 Q. Those are industrial rates. 25 So if I take the EIA data for the current month -- 26 and I recognize you did annual averages -- but if I took 27 the current month, and it reported California's per 28 kilowatt rate at $0.1937 and the rate for Texas at 3685 1 $0.0674, does that sound reasonable to you? 2 A. That sounds why you are not seeing a lot of new 3 plant investment in California, yeah. I -- I don't know 4 if it's reasonable or not, but it -- it -- based on the 5 numbers you are talking to me about, it doesn't seem out 6 of the realm of possibility. 7 Q. If I look at Figure A-3 on page 17 of your report, 8 utilities -- well, except for as they trailed off, but 9 through CDFA's report, they were, for the most part, the 10 second largest or maybe the single largest category of 11 costs for a nonfat plant, correct? 12 A. Utilities are -- well, it depends on how you 13 are -- how you are categorizing the costs. I guess, are 14 you looking at the CDFA categories or are you looking 15 at -- for here, I'm estimating three equations that have 16 labor, utility, and other. And other, in all cases, is 17 the largest component. 18 Q. I assumed other contains a bucket of various 19 costs. 20 A. I think I listed them there. Yeah. 21 Q. Yeah. So if utilities are such a large component 22 of the nonfat cost of production, would a -- would a 23 difference of nearly 100% in the natural gas costs be 24 meaningful when you are trying to apply California costs 25 to a national program? 26 A. Yes. I mean, really what you are asking is, are 27 costs in plants located in other states likely to be lower 28 than costs in California; would that be fair? 3686 1 Q. I think that's one way to look at it. If you want 2 to answer that question, I'll rephrase mine. 3 A. I think -- I think the fact that a location with a 4 lower utility cost could have lower plants costs of 5 manufacture is certainly a possibility. 6 Q. If California's electricity costs are more than 7 100% higher than the rest of the states, and natural gas 8 prices are nearly 100% higher than the rest of the states, 9 is it appropriate to use California's costs to set a 10 formula that applies to the entire country? 11 A. I think given the amount of nonfat dry milk that's 12 manufactured in California, certainly California costs as 13 a part of that equation is valid. 14 Q. And Dr. Stephenson included California plants in 15 his study didn't he? 16 A. I believe he did. 17 Q. And according to the 2002 dairy product summary 18 from NASS, California produces about 17.5% of the cheese 19 in the country, correct? 20 A. I don't have that in front of me, so I -- 21 Q. Does the Dairy Institute track that information? 22 A. We used to. 23 Q. Okay. 24 A. But not as much anymore. I haven't looked at that 25 number in a while. 26 Q. Does around 20% sound right? 27 A. Yes. 28 Q. And cheddar, 6.95%? 3687 1 A. That sounds right, too. 2 Q. Butter, about a third, 33.3% actually? 3 A. Yeah. That sounds about right. 4 Q. And nonfat, a larger percentage, 34.7% sounds 5 reasonable? 6 A. It does. 7 Q. Dry whey, about 25%? 8 A. Dry whey, 25%? That sounds high. 9 Q. That sounds high? 10 A. Uh-huh. 11 Q. Okay. I'm sorry. You know why that's high? 12 Because they don't report California, they report the 13 west. So -- 14 A. Okay. I could buy it for the west. 15 Q. Okay. And California's about 18% of U.S. milk 16 production, correct? 17 A. That sounds about right. 18 Q. And so Dr. Stephenson has already weighted 19 California -- or he's included California in his study, 20 correct? 21 A. Correct. 22 Q. And IDFA wants to take those California numbers 23 and have it account for a full half of it, correct? 24 A. Correct. 25 Q. On top of what's already in Dr. Stephenson's 26 study, correct? 27 A. Correct. 28 MR. MILTNER: Thanks. That's all I have. 3688 1 MR. ENGLISH: Chip English for the Milk Innovation 2 Group, your Honor. 3 CROSS-EXAMINATION 4 BY MR. ENGLISH: 5 Q. Dr. Schiek, do you still have Exhibit 156 in front 6 of you referred to by Ms. Hancock, or National Milk? 7 A. 156. Make sure, it is the gray cover 2016 data? 8 Q. Yes. 9 A. Yeah. 10 Q. So when you were asked questions, I think you 11 looked at one paragraph on page 3, and I would like to 12 point you to another paragraph before I actually go look 13 at some decisions by USDA. 14 So on page 3, the last paragraph, could you read 15 the second line that begins after the word "California"? 16 A. Yes. So this is page 3. 17 Q. Of 17 of Exhibit 156. 18 A. 156. So this is referring to the cost studies. 19 Q. Yes. 20 A. "They are the only studies in the U.S. which 21 present the audited and detailed processing costs of 22 butter, nonfat, dry milk, and cheddar cheese over several 23 years." 24 Q. So does that refresh your recollection of whether 25 CDFA audited the costs? 26 A. Well, they certainly claim they do, and I have no 27 reason to really reject that. 28 Q. And part of why we're talking about California, 3689 1 after all, is that in a series of hearings back in 2006 2 and 2007, CDFA data was presented, and USDA ended up 3 relying on CDFA data, correct? 4 A. Correct. 5 Q. Okay. Are you aware that in the 2006 decision, 6 which is Federal Register number 71, starting at 67467, 7 published in November 22nd of 2006, I represent to you 8 that USDA stated the following: "The CDFA witnesses 9 testified that all cost survey data collected is from 10 audited plant cost records." 11 Would that also refresh your recollection as to 12 whether CDFA data was audited? 13 A. Yes. I would -- I don't know if it refreshes, but 14 I would take that as CDFA attesting to the fact that they 15 were auditing their data. 16 Q. And later in the decision, if USDA said, we're not 17 going to rely on certain data because it's not audited, 18 but the reason we are using CDFA data is because it is 19 audited, USDA had concluded back as early as 2006 that 20 CDFA data was audited, correct? 21 A. Correct. 22 MR. ENGLISH: Okay. I have no further questions. 23 THE COURT: Anyone else have cross, other than 24 AMS? 25 Your witness. 26 CROSS-EXAMINATION 27 BY MS. TAYLOR: 28 Q. Good afternoon. 3690 1 A. Good afternoon. It is nice to see you, 2 Ms. Taylor. 3 Q. It's nice to see you. I bet you are surprised, 4 too, that I might still have questions for you given the 5 amount of questions you have already had. 6 A. Nothing surprises me. 7 Q. That's true. I am going to try to stay a little 8 bit out of the weeds. 9 A. Okay. 10 Q. Okay? Attempt. And just have a few questions 11 that left over from what people didn't already ask. 12 Your -- the data you have used in your model goes 13 to 2016. 14 A. Correct. 15 Q. So have you -- does your model account for any 16 changes in the makeup of California plants since that 17 time, new plants or closed plants? 18 A. It does not. 19 Q. I know you cite for -- well, let me ask you one 20 more question. So I don't -- do you -- can you talk about 21 what -- how the makeup of California manufacturing plants 22 have changed since 2016? 23 A. Generally, so throughout that sample period, you 24 know, just rethinking your first question to me, I -- 25 specifically with the cheese industry, the number of 26 cheddar cheese plants in California declined throughout 27 that period. We started that period with quite a number 28 of them, and by the last study, I think there were four 3691 1 plants. 2 Q. You are saying prior to 2016? 3 A. Prior to twenty -- last study in 2016, so in like 4 2003 there may have been eight or nine cheddar plants. I 5 could find out if I look. But that -- so that number 6 declined throughout that period. So the model estimates 7 cost in a period of declining numbers of cheddar cheese 8 plants. I mean, that's part of the background. So to 9 some extent there is sort of a movement towards fewer 10 plants kind of baked into the model, just because that's 11 what was going on in -- with the data during that 12 timeframe. 13 But in terms of since 2016 -- 14 Q. Can I ask you a different question -- 15 A. Okay. 16 Q. -- before you answer the post 2016? Can you speak 17 to how plants -- other manufacturing plants would have 18 changed during that time, butter plants, powder plants? 19 A. Yeah. I don't think the trend is quite as clear 20 there because we had some -- during that sample period, we 21 had some new plants open. And I know, you know, we had -- 22 we had a couple of plants close and then one reopen. So, 23 you know, there was less of sort of a linear trend in 24 terms of consolidation in nonfat dry milk during that 25 period, and I think butter would be the same answer as far 26 as nonfat dry milk. And, of course, dry whey we never had 27 a lot to begin with, so -- 28 Q. Right. Okay. So, now, how do you think that's 3692 1 changed since 2016? 2 A. We have had a little bit of consolidation in 3 nonfat dry milk, in other words, I think a couple of plant 4 have closed since that period, but it hasn't been a 5 particularly rapid consolidation. 6 Butter, I probably -- you know, Rob Vandenheuvel 7 would be able to speak to this better, but I don't know 8 that we have seen the same rapid number of decline in 9 butter plants. 10 And cheddar cheese, I think we have -- we have one 11 plant that ceased operation last year sometime, and that 12 was a smaller cheddar plant. 13 Q. I'm just wondering, since you have vast experience 14 in California, specifically, and the California Federal 15 Order came into existence in the end of 2018, and so 16 that's when they went from the state pricing formulas with 17 their Make Allowances to ours as currently exist. 18 And I'm just wondering if you -- I don't know, did 19 that change in price formulas in Make Allowances perhaps 20 have any influence in -- on investment decisions in the 21 state? Or California Make Allowances were higher than our 22 current Make Allowances? 23 A. It would be hard to tease that out. I mean, your 24 point about higher Make Allowances, I think we adjusted 25 Make Allowances the last -- if I'm remembering correctly, 26 so my memory may be fuzzy on this, but I remember we 27 adjusted Make Allowances at a hearing in 2011 that was 28 probably based on 2010 data, and we hadn't adjusted it 3693 1 since then. 2 As somebody -- I think Ms. Hancock pointed out, 3 there's a host of other things going on in California that 4 probably had a -- have a bigger impact on decisions 5 than -- than the Make Allowance. 6 And, you know, in particular the cheese plant that 7 I know that ceased production of cheese, they were located 8 in Southern California, and part of the issue there is the 9 diminishing milk supply in that region and the higher cost 10 of hauling in there, which makes it tough for a 11 manufacturing plant to compete. 12 Q. Okay. I'll just note this is a request for at 13 some point when perhaps IDFA puts together your official 14 notice list, you cite the BLS California wage rate that 15 you used, but if you could provide us with a -- 16 A. Data link? 17 Q. Yes, that would be helpful to make sure we're 18 looking at the same thing that you are looking at. 19 And on that wage rate that you used, because that 20 was just for -- oh, gosh, I forget the term you used. It 21 wasn't dairy specific I guess. 22 A. Correct. It was manufacturing wage rate. 23 Q. Okay. And you think -- I don't -- this is -- I -- 24 I don't know the answer to this question. I like to know 25 my answers sometimes, but not this one. 26 Was that like a comparable position to what you 27 find in dairy manufacturing plants? I mean -- 28 A. Yeah. 3694 1 Q. -- what does that encompass? 2 A. I don't know that it is the actual rate in those 3 plants. But I do believe that changes in that rate, 4 because of the competitive nature of the labor market, 5 would be reflected in the changes in labor rates and dairy 6 plants as well. 7 Q. And is the labor rate -- I know there's been some 8 talk about it, but just try to make it a little bit 9 clearer. That wage rate, does that, in your mind, account 10 for efficiencies gained in labor that would also be 11 reflective of plant investments at the time, so you need 12 less labor to run your -- or you get more product out of 13 your plant, maybe you invest it to your labor costs? 14 A. Yeah. So you're picking up with that model, you 15 know, it's -- that's the difference between sort of an 16 indexing approach, where you're just assuming, okay, we 17 have got this cost factor that's going up, and so 18 therefore, the cost associated with that factor is going 19 to rise by the same amount. 20 The regression analysis, because it's regressing 21 against the actual cost, is going to pick up those 22 efficiencies over time. And I -- I don't know if I am 23 answering your question. 24 Q. Sure. 25 A. If that was at least the start of the direction. 26 Is that what you were asking? 27 Q. Yeah, sort of. I mean, we've heard a lot of 28 discussion over the past few weeks about investments and 3695 1 what decisions have been made to invest or not invest 2 because of the -- attributable somewhat to some of the 3 manufacturing -- well, to the manufacturing allowance that 4 we currently have in the formulas. So I'm trying to see 5 if somehow that -- whatever investments were made in 6 California plants during that time, assuming there were 7 some, that that's picked up in your cost estimates. 8 A. So it would not be picked up -- if it's happened 9 since 2016, it would not be picked up. 10 Q. Okay. So does your model -- do you have the 11 ability to calculate prediction intervals for your 12 estimates? You gave us a number -- 13 A. Sure. 14 Q. -- right, but somehow there's probably -- 15 A. So -- 16 Q. -- a range -- 17 A. -- sort of a confidence interval on the forecast? 18 Q. Uh-huh. 19 A. The answer is there may be -- because of how the 20 model is constructed, it's fairly easy to do a confidence 21 interval on any of the individual regression models 22 because you have got confidence intervals on each of the 23 parameter estimates, and so you can go from there. 24 How you build that up to a total, I -- I would 25 have to do a little more work to understand how to do that 26 correctly, and I don't. 27 But I can tell you, you know, confidence intervals 28 are oftentimes a function of -- you know, a function of 3696 1 how much explanatory power is in the model, but also a 2 function of the number of observations you have. So, you 3 know, when I did my dissertation work, I had like 50 years 4 of data, of here we have got 13, 14, 15, I don't know. So 5 they would probably be pretty wide confidence intervals on 6 the forecast. 7 Q. Okay. I know you cited in your testimony some 8 other witnesses that have talked about their cost 9 increases -- 10 A. Uh-huh. 11 Q. -- to say that what you have -- the model has 12 predicted is kind of within the range of what people have 13 testified to. 14 Did you share your results with your member plants 15 of Dairy Institute to see from their perspective if 16 they -- the results are kind of what they see? 17 A. I have had some conversations, just -- you know, 18 these are what the estimates are I'm coming in for; are 19 they in the ballpark? You know, people I have asked if -- 20 nobody said, boy, you are really way off. But, again, 21 that's not a -- that's not a -- what's the word -- 22 rigorous -- rigorous survey of the data. That's just from 23 asking a few people. But, yeah, I haven't sent it out and 24 asked for people to report back or anything like that. 25 Q. Okay. And you didn't get any notable feedback 26 either then? 27 A. That -- that -- no. 28 MS. TAYLOR: Okay. 3697 1 CROSS-EXAMINATION 2 BY MR. WILSON: 3 Q. Hello, Dr. Schiek. Todd Wilson, Dairy Programs. 4 A. Nice to see you, Mr. Wilson. 5 Q. Good to see you. 6 I'm going to read the question because it's 7 probably better. 8 A. Okay. 9 Q. Looking at the regression results in Table 5 -- 10 I'm sorry -- down at the bottom of page 5, not the 11 regression table -- on the bottom of page 5 on 12 Exhibit 180, you describe including the Producer Price 13 Index for intermediate goods. 14 Since BLS publishes many different PPIs at 15 different stages in the supply chain for different 16 grouping of commodities, can you clarify which specific 17 PPI you used in your analysis? 18 A. This would be a higher level PPI, not a specific 19 industry PPI. So it's an overall U.S. PPI for 20 intermediate goods. 21 Q. Okay. Thank you. All right. 22 I got one more. Sorry. 23 A. So could I add on to my answer for that? 24 Q. Absolutely. 25 A. So my thinking there is that dairy plants are 26 buying inputs that aren't -- not all of which are unique 27 to the dairy industry. You know, they are buying 28 packaging materials. They are buying, you know, certain 3698 1 ingredients, maintenance and supply, equipment for the 2 plant. And so they are probably procuring these things, 3 you know, from vendors or manufacturers in other parts of 4 the country. And therefore I thought that the more 5 general number was appropriate for -- for that kind of -- 6 what I'm trying to represent, which is sort of the other 7 cost materials and how those costs are changing. 8 Q. Yes. So when you -- can you provide the link -- 9 A. Yes. 10 Q. -- official notice on that information? 11 Okay. One more question. On Table 3, page 9, 12 there is a parameter with an estimate for spot electricity 13 that's negative. 14 A. Yeah. Yes. 15 Q. Can you -- 16 A. That's -- that was the result of the regression 17 equation, which is not what you would expect, right? 18 And I think what we're -- what we found, 19 generally, with utility cost model estimates is they had 20 the poorest fit and, you know, behaved in -- you know, 21 there's one where it behaved in a way that we didn't 22 expect. I think it's a -- it's a combination of the 23 amount of forward buying and hedging people do of costs. 24 I think I'm using an industrial electricity price, 25 which a lot it would apply to a lot of plants. I do know 26 that there are some dairy plants, some large dairy plants, 27 that are in, for lack of a better term, metropolitan 28 utility districts that have very different cost parameters 3699 1 than the heavy industry in the rest of the state. So, you 2 know, states -- the two big utilities are Southern 3 California Edison and Pacific Gas and Electric, and I know 4 there are some big dairy plants that operate in 5 metropolitan utility districts that have very different 6 rates. 7 CROSS-EXAMINATION 8 BY MS. TAYLOR: 9 Q. So for the lay listener -- 10 A. Yeah. 11 Q. -- and me, who is also a lay listener, can you 12 explain how to interpret that negative number? 13 A. I would -- I would basically interpret it as -- as 14 that is the number that the model estimated as the best -- 15 as the parameter when you include electricity that would 16 give you the best fit. So I think it's -- it's a -- it's 17 a consequence of probably the variables I'm using in that 18 particular case are not really good proxies, they are not 19 the best proxies for electricity, for, example for dairy 20 plants in California. 21 Q. Okay. So we have heard testimony kind of 22 throughout this week and last about how 2022 is -- I don't 23 know if I want to say an outlier. We don't know what 24 normal is anymore. But with the supply chain issues and 25 inflation, as you have spoken to yourself, do you see 26 those things have moderated some? Should we look at 2022 27 as an outlier and not necessarily want to bake that result 28 into some Make Allowances that we would set for an 3700 1 undetermined amount of time? 2 A. Well, we have some moderation. I think, you know, 3 fuel prices have come down, and that's explained an awful 4 lot of the reduction in the -- in the general headline 5 numbers. I didn't see what the CPI was today. Maybe 6 somebody saw that. I don't know whether it was up from 7 last month in terms of increases but -- so -- so I don't 8 know going forward. None of us have a great crystal ball. 9 Certainly inflation spiked in 2021 and 2022, and 10 it's been coming down in '23, the rate of inflation 11 anyway. I don't know that the price levels necessarily 12 have been coming down, but the rate of inflation has. 13 The other thing I would point out -- and we used 14 to look at this when we were at the CDFA hearings too, is 15 that when you have gone a long time without an adjustment, 16 there's been a shortfall in manufacturing costs relative 17 to, you know, where the Make Allowance is for an extended 18 period of time. And if -- if we were a little higher for 19 a short period of time, that -- that may not be 20 necessarily such a bad thing from a -- you know, allowing 21 plants to get healthy and make the kind of investments 22 they need. 23 I think, you know, the reality is we shouldn't -- 24 we shouldn't be updating Make Allowances once every 25 15 years or whatever it is. We should have this 26 discussion more regularly, I think, to keep things 27 current. And, you know, that would be my solution. I 28 don't know that I would say shy away from current costs 3701 1 just because it might be a bad year. I think come back 2 when the -- you know, some -- somebody should petition 3 and ask for a hearing, you know, when things change and 4 say, hey, we need to adjust this again. 5 Q. So speaking of inadequate Make Allowances as have 6 been described in this hearing, can you speak about how 7 your Dairy Institute members are coping with that? If 8 they are so inadequate, how are they able to keep 9 operating? 10 A. Well, generally, I think there are several things 11 that can happen. If you have the ability to delay 12 reinvestment -- I kind of look back. My -- my father and 13 his brother grew up on a dairy farm, and I kind of looked 14 at how they handled new equipment purchases. And the 15 reality is if they could keep something running and 16 keep -- you know, not have to buy the new equipment, they 17 would do it as long as they could. 18 Q. Duct tape and baling wire? 19 A. Yeah, basically that's the analogy, right. So 20 you -- you defer maintenance -- or you defer investment 21 and just keep things going and hope at some point you will 22 be able to upgrade. So that's one way to -- 23 Q. For 15 years? 24 A. Yeah. Well, I don't -- I don't know for 15 years. 25 Obviously, in California we did have some Make Allowance 26 increases since then. But for the -- in terms of a lot of 27 the specific strategies plants are doing in California to 28 deal with that, I -- I don't know. We -- you know, we 3702 1 just -- on cheddar, we don't have that many plants 2 anymore. And on nonfat dry milk and butter, those are 3 generally not members, those -- I don't have many of those 4 in my membership. 5 Q. Okay. 6 MS. TAYLOR: I think that's it from AMS. Thank 7 you. 8 THE WITNESS: Okay. Yep. 9 THE COURT: Redirect. 10 REDIRECT EXAMINATION 11 BY MR. ROSENBAUM: 12 Q. Just a few follow-ups. 13 On this question of whether 2022 was a year of 14 relatively high inflation, okay. 15 A. Uh-huh. 16 Q. I mean, whether it was relatively high or not, the 17 costs are what the costs were in 2022, correct? 18 A. Correct. 19 Q. I mean, and Federal Orders don't adjust based upon 20 projections of future inflation, correct? 21 A. That's not been my understanding that they have 22 ever done that. 23 Q. I mean, in other words, in order for the costs of 24 manufacture to be less than as surveyed in 2022, you would 25 have to actually have deflation, correct? 26 A. You would have to have deflation in those cost 27 factors, right. 28 Q. Okay. So I mean, even though general inflation 3703 1 may have dropped somewhat in 2023 compared to 2022, 2 generally things are still more expensive in 2023 than 3 they were in 2022, correct? 4 A. Yeah. I would say that's correct. And, you know, 5 when you look at the trend lines, we generally don't see 6 things like labor go backwards in terms of cost. 7 Q. Okay. 8 A. May get more efficiencies in a plant, but the 9 trend line on labor cost is pretty consistently upward. 10 Q. So I want to understand a little bit more about 11 the trend line. So you were -- you were -- when you're 12 looking at the California data starting in twenty -- 13 strike that. 14 When you are looking at the California data that 15 starts in 2002, which was your first year, through 2016, 16 your -- the costs that the California Department of Food 17 and Agriculture is calculating are based upon how many 18 pounds of product you produced at what cost, correct? 19 A. Correct. 20 Q. That's how you get to a cost per pound, correct? 21 A. Correct. 22 Q. And if you have engaged in efficiencies during 23 that 16-year period, then by 2016 you're going to have 24 lower costs based upon those efficiencies. You may have 25 higher costs based upon other things, but to the extent 26 that there were efficiencies, you have picked that up, 27 correct? 28 A. Yeah, I have. 3704 1 Q. And so when you then use that data to project 2 forward from 2016 to 2022, which is what you did, are 3 you -- by using the 2002 through 2016 data to create your 4 formulas, are you capturing that phenomenon? 5 A. So I think you are capturing the phenomenon. The 6 question really comes is, you are picking up a rate of 7 gain in efficiency, right, as part of -- during the sample 8 period. So the only thing you wouldn't be picking up is 9 if outside the sample period there was a change in that 10 gain. 11 Q. But let's just be clear about that. As long as 12 the trend in efficiency from 2016 through 2022 has been 13 the same as it was from 2002 to 2016, you are going to 14 have captured that, correct? 15 A. That would be accounted for, yes. 16 Q. It would only be if there were some substantially 17 different kind of change in efficiency since 2016 that you 18 would have missed that; is that fair? 19 A. I would say yes. 20 Q. And are you aware of some vast change that's 21 happened since 2016 that is hugely different than the 22 trend that existed from 2002 to 2016? 23 A. I am not. 24 Q. Now, you were asked some questions about -- 25 well -- and you were asked a number of questions about 26 whether we should be using California data to set minimum 27 milk prices. I mean, that's -- that's a bridge that USDA 28 has crossed two or three times already, correct? 3705 1 A. Yes. They have already used California data in 2 constructing the Make Allowances. 3 Q. Okay. And we'll hear some more about that history 4 tomorrow. I won't try to recite it -- potentially 5 tomorrow. We'll see how fast we move. 6 But the other topic I want to just cover quickly 7 is, in terms of Dr. Stephenson's cost survey, there was 8 some questions about, you know, more data is better than 9 less data and, you know, sample size can make a difference 10 and things of that nature, right? Correct? 11 A. Correct. 12 Q. All right. We'll have this exact testimony when 13 Mr. -- when Mr. Brown testifies, but I just want you to 14 assume these calculations are correct. 15 But let me just start by saying did you -- did 16 you -- were you here when I asked Dr. Stephenson how one 17 would go about calculating the percentage of total cheddar 18 cheese, whey, nonfat dry milk, and butter production in 19 the United States, how one can calculate what percentage 20 of that is covered by his survey? 21 A. I'm not positive I was, actually. 22 Q. Then no reason to rehearse that. 23 Just take these numbers as what those -- that 24 information establishes. This will be the subject of 25 testimony later. 26 So for nonfat dry milk, assume with me that the 27 plants that participated in the survey produce 91.2% of 28 all nonfat dry milk produced in the country in 2022. 3706 1 Okay? Do you have -- I mean, what would that mean to you 2 as an economist as to the representativeness of the sample 3 data? 4 A. That would be -- I would assume that would be very 5 highly representative of the population as a whole. 6 Q. I mean, there's -- there's a survey and there's a 7 census; is that a phrase people use in the -- 8 A. Yeah. 9 Q. -- economic world? A census is where you cover 10 everything, I guess? 11 A. Yes. 12 Q. I mean, 91% is sort of getting into census 13 territory, isn't it? 14 A. It is. 15 Q. And let's take butter. Assume that 16 Dr. Stephenson's 2022 cost survey included plants that 17 collectively produced 80.1% of all the butter produced in 18 the United States in 2022. 19 Now, it's not as high as 91, but -- but what would 20 be your reaction to that in terms of the likelihood that 21 this was reflective of the actual costs of producing 22 butter in 2022? 23 A. I would assume it was pretty reflective. 24 Q. And for cheddar cheese and whey, the numbers are 25 somewhat lower. Cheddar cheese is 55.6%. Whey is 50.8%. 26 What is your general view as to those levels of the 27 percentage of total production that is reflected in the 28 plants that participated in the survey? 3707 1 A. I think I would also expect those to be pretty 2 representative of the population as a whole. 3 MR. ROSENBAUM: Your Honor, at this point I would 4 just simply like to move my exhibits into evidence, which 5 are Exhibits 180 through 195. 6 THE COURT: Mr. Miltner has -- 7 MR. MILTNER: No objection. Just additional 8 questions. 9 THE COURT: Let's -- well, okay. I'll tell you 10 what, let's take a break. It's been quite a while. 11 Any objections to any of this exhibits coming into 12 evidence? 13 Seeing none. Let me get out my list. 14 Okay. With that, Exhibit Numbers marked for 15 identification as 180, 181, 182, 183, 184, 185, 186, 187, 16 188, 189, 190, 191, 192, 193, 194, and 195, are all 17 entered into the record of this proceeding. 18 (Thereafter, Exhibit Numbers 180 through 195 19 were received into evidence.) 20 THE COURT: Okay. Let's -- ten minutes. It is 21 late in the day, ten minutes. Let's come back at 4:15, 22 and Mr. Miltner, it will be your turn. 23 (Whereupon, a break was taken.) 24 THE COURT: Back on the record. 25 I understand that Mr. Miltner has waived any 26 further questions of this witness. So unless there are 27 objections from someone, we'll let this witness step down 28 from the stand. 3708 1 Thank you, Doctor. 2 MR. ROSENBAUM: Your Honor, we would call as our 3 next witness, Mr. James DeJong. 4 THE COURT: Raise your right hand. 5 JAMES DEJONG, 6 Being first duly sworn, was examined and 7 testified as follows: 8 THE COURT: Mr. Rosenbaum, your witness. 9 DIRECT EXAMINATION 10 BY MR. ROSENBAUM: 11 Q. Good afternoon, Mr. DeJong. Can you please state 12 your full name for the record and provide your mailing 13 address, your business mailing address? 14 A. Sure. My name is James DeJong. And my mailing 15 address for work is 121 4th Avenue South, Twin Falls, 16 Idaho, 83301. 17 Q. And, Mr. DeJong, have you prepared two different 18 written testimonies that you are going to deliver today? 19 A. Yes, or as much as I can. 20 MR. ROSENBAUM: Your Honor, I -- the first of them 21 is labeled as IDFA Exhibit 22, and we would ask that that 22 be marked with the next Hearing Exhibit number. 23 THE COURT: IDFA Exhibit 22 is marked for 24 identification Exhibit 196. 25 (Thereafter, Exhibit Number 196 was marked 26 for identification.) 27 MR. ROSENBAUM: And, your Honor, IDFA Exhibit 41, 28 we would ask that to be marked as Hearing Exhibit 197. 3709 1 THE COURT: So marked. 2 (Thereafter, Exhibit Number 197 was marked 3 for identification.) 4 BY MR. ROSENBAUM: 5 Q. Mr. DeJong, can you please read Hearing 6 Exhibit 196? 7 A. My name is James DeJong, and I am currently the 8 Senior Director of Dairy Economics, Risk Management, and 9 Sales Planning for Glanbia Nutritionals (GN) for short, 10 whom I am representing today. I work out of GN’s 11 corporate office at 121 4th Ave South, Twin Falls, Idaho 12 83301. 13 I have worked for GN the last five years. My main 14 responsibilities include market and industry intelligence, 15 milk pricing analysis, hedging dairy commodity price risk, 16 and balancing our internal supply and demand for whey 17 proteins. Prior to that, I worked for Hilmar Cheese for 18 four and one-half years and at Rabobank for three years. 19 At Hilmar Cheese, I worked as their Dairy 20 Economist, dairy commodity and energy price risk manager, 21 and also as their Strategic Planner. For Rabobank, I 22 worked for their Food and Agricultural Research and 23 Advisory division as an Agricultural Analyst. There I 24 specialized in dairy industry economics, general 25 California agricultural economics, U.S. row crops, and 26 economics of North American forest products. I have a 27 bachelor’s degree in social science and a master’s degree 28 in public administration from California State University 3710 1 Stanislaus. 2 As to the background of our company, GN is part of 3 Glanbia PLC, a global nutrition company based in Ireland. 4 Glanbia PLC includes GN (business to business sales only), 5 Glanbia Performance Nutrition (business to consumer brands 6 such as Optimum Nutrition), and our Joint Ventures (which 7 include Southwest Cheese and MW cheese/whey plants). You 8 can see our basic company organization below. 9 I am here to represent GN and our 50% ownership 10 interest in the two Joint Venture cheese/whey plants. Our 11 partners in our Joint Venture plants, Dairy Farmers of 12 America and Select Milk Producers, are not represented in 13 this testimony. 14 GN is a diversified nutrition solutions company 15 that specializes in custom pre-mix solutions, bioactive 16 ingredients, flavors, micronutrients, plant-based 17 nutrition solutions, bakery ingredients, as well as 18 American-style cheeses and high concentrate whey proteins. 19 Specifically, to the dairy segment of our 20 business, GN fully owns four dairy plants in Idaho that 21 process a combined 12 million milk pounds a day and turn 22 that milk into barrel cheese, block cheese, high 23 concentrate whey proteins, proprietary protein blends and 24 lactose. Our Idaho plants operate outside the Federal 25 Milk Marketing Order (FMMO) system. 26 Our Joint Venture plants in New Mexico (FMMO 126) 27 and Michigan (FMMO 33) process a combined 22 million 28 pounds of milk per day and turn it into American-style 3711 1 block cheese and high concentrate whey proteins. Our 2 combined output between our fully-owned and Joint Venture 3 plants makes us the largest American-style cheese 4 manufacturer and the largest whey-based nutritional 5 solutions provider in the US. 6 Further, although not all our plants fall within 7 the FMMO marketing areas, we still have a substantial 8 stake in the maintenance and proper functioning of the 9 FMMO system. This is especially true in the case of the 10 Class III milk price, on which my testimony will focus. 11 Our plants make the type of cheddar cheese 12 represented in the Class III formula, compete locally and 13 nationally with other dairy manufacturers that rely on the 14 FMMO pricing system, and ourselves and our patron milk 15 suppliers utilize the risk management tools that are 16 linked to the FMMO pricing system. 17 Headline: Proposals 8 and 9: Make Allowances 18 proposed by Wisconsin Cheese Makers Association and 19 International Dairy Foods Association. 20 GN supports the Make Allowance proposals from 21 Wisconsin Cheese Makers Association (WCMA) and the 22 International Dairy Foods Association (IDFA). The 23 WMCA (sic) and IDFA proposals use an average of the Dr. 24 Schiek study (which uses the 2016 California Department of 25 Agriculture audited manufacturing cost study adjusted with 26 inflation indexes) and the last manufacturing cost survey 27 from Dr. Mark Stephenson using 2022 plant survey data. 28 Why the WMCA/IDFA Make Allowance proposal should 3712 1 be adopted: 2 We believe the data from these studies should be 3 used because there is a higher degree of transparency, and 4 USDA has precedent for using similar studies in past FMMO 5 decisions. IDFA's testimony discusses past USDA precedent 6 for using high quality and data driven research to 7 establish Make Allowances. 8 Further, as the largest processor of cheddar 9 cheese in the US, all five of our cheddar plants 10 participated in the last 2022 Stephenson cost study, which 11 includes our Joint Venture plants as well. 12 GN supports the $0.0015 per pound marketing 13 allowance cost addition: 14 GN supports the $0.0015 per pound marketing cost 15 addition to the WMCA and IDFA Make Allowance proposal. On 16 one hand, marketing costs have risen like other costs due 17 to inflation. On the other hand, one could also argue 18 that industry consolidation has reduced the amount of 19 resources needed to sell cheese domestically. In balance, 20 we ask that the $0.0015 per pound marketing cost be 21 included in the final Make Allowance as it was in the 22 previous FMMO Make Allowance decision. 23 Why Make Allowances need to be maintained: 24 GN believes FMMO Make Allowances must be 25 maintained to reflect reality. The FMMO system relies on 26 these Make Allowances to set minimum pricing and 27 distribute pool revenues, while the industry uses these 28 prices to make investment decisions, set the pricing of 3713 1 milk, and are heavily used in CME and USDA risk management 2 tools. 3 However, when these Make Allowances are not 4 maintained, as they haven't been in 15 years, we can 5 expect to see market distortions and further real-world 6 variances versus the USDA announced Class prices. 7 Looking at USDA published data, we can see 8 declining mailbox milk prices versus uniform milk prices 9 at test (Figures 2 through 5). The analysis in these 10 figures takes the USDA mailbox milk prices from four 11 states/regions, then subtracts the order's uniform price 12 at the order's weighted average milk components. The 13 purpose of the analysis is to illustrate how actual 14 producer milk prices have changed over time versus the 15 regulated price at real world milk components. 16 For example, in Wisconsin the mailbox milk price 17 from October 2008 to September 2010 averaged $14.42 per 18 hundredweight, while the uniform milk price at test (using 19 the $1.70 zone PPD) averaged $13.54 per hundredweight. 20 This equals an $0.88 per hundredweight positive variance 21 versus the uniform price at test. 22 However, from May 2021 to April 2023 (last 23 available data), the Wisconsin mailbox price averaged 24 $21.78 per hundredweight while the uniform milk price at 25 test (again using the $1.70 zone PPD) averaged $22.21 per 26 hundredweight. This equals a $0.43 per hundredweight 27 negative variance versus the uniform price at test and a 28 $1.31 per hundredweight negative total swing over this 3714 1 period. 2 What this data shows is that there is a "bumping 3 up" of the mailbox price against FMMO uniform prices; in 4 other words, the market is trying to take the actual pay 5 price below the FMMO minimum price. That is a sign that 6 the minimum price is too high, and that the price is too 7 high in large part because of inaccurate Make Allowances. 8 While other factors, like higher milk hauling 9 costs, changes in checkoff program amounts, or variances 10 in milk components will cause noise in the analysis, the 11 trendline is unmistakable. 12 Further, the other three regions analyzed (Figures 13 3 through 5) that are inside FMMOs show the same pattern 14 of collapsing milk premiums versus the FMMO uniform 15 prices. We believe a good portion of this collapse is 16 attributed to extremely outdated Make Allowances. There 17 is a summary of the total swing in mailbox prices versus 18 the uniform price at test for the four areas in the 19 Appendix section. 20 Milk premiums take over when FMMO milk prices are 21 below competitive levels: 22 We believe there is more industry risk when 23 regulated milk prices are set too high versus too low. 24 When the FMMO milk prices are set too low in a milk shed, 25 historically speaking, market premiums over the Class III 26 prices take hold. 27 Looking at Figures 3 through 5 again, in the early 28 years following the 2008 Make Allowance change, mailbox 3715 1 prices were relatively strong versus the uniform prices at 2 test in multiple regions. In this case, dairy processors 3 had extra margin over the FMMO Class prices that was then 4 diverted to pay for premiums. 5 Given milk cooperatives control about 85% of all 6 the milk in the U.S., this places them in an extremely 7 strong position to bargain for premiums above the FMMO 8 Class prices, providing enough value is being generated 9 from dairy products in that milk shed. 10 If Make Allowances were set too high in some milk 11 sheds, market principles will take over and premiums will 12 again become common. 13 Importance of Make Allowances for pooling dollar 14 distribution: 15 In the case of FMMO pooling revenue distribution, 16 when the Class III and IV Make Allowances are not 17 reflective of reality, a situation can be created where 18 pool revenues are not distributed in a fair or economical 19 justifiable manner. 20 For example, if the Class III Make Allowances were 21 too low (creating an artificially high Class III price), 22 but set too high for Class IV (creating an artificially 23 low Class IV price), Class IV milk handlers would have an 24 unfair advantage because pool dollars flow to the lowest 25 Class value of milk. 26 In this case, the Class IV handlers could be 27 financially strong while also pulling in extra pool 28 revenue, while the Class III milk handler could be 3716 1 struggling while not getting any pool revenue (or worse, 2 paying into the pool). The opposite situation could exist 3 between Class III and IV depending in which direction the 4 Make Allowances were distorted. 5 In the end, the point proves USDA needs to 6 maintain accurate Make Allowances to ensure the FMMO 7 pooling system is functioning equitably for producers. 8 Failing to correct Make Allowances with the best available 9 data, or delaying their implementation, will create 10 disorderly marketing. 11 Impact of higher manufacturing costs on GN: 12 GN's costs have gone up considerably since the 13 Class III Make Allowances were last changed in 2008. Our 14 Twin Falls, Idaho plant, which processes about 2.5 million 15 milk pounds per day, is our best plant to compare costs 16 over time since it only makes American-style cheese 17 (mostly cheddar), does not dry any whey, and has been 18 minimally changed over the years. Our other plants have 19 seen major expansions or whey processing investments over 20 the years that make them more difficult to compare versus 21 2008. 22 For our Twin Falls, Idaho plant from 2008 to 2022, 23 we have seen some costs like energy only go up slightly 24 (lower natural gas cost combined with energy efficiency 25 projects), items like direct labor and packaging go up 26 about 30%, and some items have gone up considerably more, 27 like plant insurance, which was up over 70%. Overall, we 28 have seen total costs from 2008 to 2022 increase at a 3717 1 similar rate as reflected in the Stephenson and Schiek 2 cost studies. 3 Additionally, we have also seen higher costs arise 4 on the regulatory and sustainability front. For example, 5 regulatory costs related to the Food Safety Modernization 6 Act have produced massive increases in testing and 7 analysis requirements. 8 Sustainability-related costs have also 9 skyrocketed. We have invested in more sustainable 10 packaging, plant upgrades that reduce carbon output and 11 waste, $2.5 million per unit water polishers that allow 12 water to be reused many times over (often multiple 13 polishers are required per plant), and investment in 14 personnel who monitor dairies and enforce on-farm 15 sustainability requirements. It is extremely difficult to 16 extract market premiums for our regulatory and 17 sustainability efforts. It is often looked at as the cost 18 of doing business today. 19 Many of our 2023 costs will be even higher than 20 2022 given the persistent inflation in the broader 21 economy. That includes items like labor, where we see 22 fierce competition for workers with other manufacturers, 23 but also the cost to replace dairy processing equipment. 24 We estimate the cost to build the 8 million milk 25 pound per day MWC cheese and whey plant with our Joint 26 Venture Partners, which was completed in late 2019 and 27 early 2020, would have gone from about $470 million 28 originally to about $600 to $700 million if it was built 3718 1 today. If $650 million is used as the midpoint, this is a 2 38% increase in just a few years. This increase in plant 3 equipment costs is reflected in things like replacement 4 silos, electric motors, water polishers, various 5 electrical equipment, and countless other parts that keep 6 a cheese plant running. 7 GN fights to keep manufacturing costs low: 8 While our manufacturing costs have undoubtably 9 increased over the years, we also go to extreme lengths to 10 try to keep costs as low as possible. This includes 11 negotiating with vendors and various suppliers to get the 12 most competitive pricing, while also investing heavily in 13 plant equipment and technology to control costs. 14 For example, since the last Make Allowance 15 adjustment in 2008, we have spent countless millions of 16 dollars on projects such as recovering biogas from lost 17 milk components in wastewater, heat exchange systems that 18 take cold water from the milk and use it to cool other 19 systems in the plant, automation projects that reduce 20 labor costs, and right-sizing of equipment (for example, 21 doing analysis to determine the minimum pump size needed). 22 Further, our newest Joint Venture Plant, MWC in Michigan, 23 incorporates a lot of the latest efficiency learnings into 24 its design. 25 New cheese plant investors working around 26 regulated system: 27 Cheese processing growth outside of FMMO 28 regulation is creating additional cheese capacity that 3719 1 competes directly with manufacturers regulated under 2 Federal Orders. These plants have been able to attract 3 the milk needed at prices outside the FMMO minimums, 4 making it harder for many regulated plants to compete for 5 cheese sales at the price that generates margins 6 sufficient to pay the regulated price. This can 7 contribute to disorderly marketing where pooled plants 8 would be at a financial disadvantage to those who don't 9 pool or operate outside the system. 10 Cheese manufacturers cannot raise prices to 11 recover losses: 12 For most industries, raising prices is one of the 13 most common ways to offset higher costs. However, raising 14 prices for dairy products that are reported in the NDPSR 15 survey creates a feedback loop. For example, if over the 16 course of a few years cheddar cheese manufacturers raised 17 their overages versus the CME cheese price by $0.01 per 18 pound, this would then be fed back into the Class III 19 protein price and increase the price of milk 20 commensurately. In this case the manufacturer has not 21 gained anything, but nonetheless must still increase their 22 overage over the CME spot market or risk falling behind 23 the NDPSR price in Class III. 24 Without Make Allowance increases, the only way for 25 a manufacturer of NDPSR reported products to recover 26 higher manufacturing costs is to pursue ruthless 27 efficiency, look for opportunities outside the NDPSR 28 reported products, look for escape valves out of the 3720 1 Class III price, invest outside the FMMO regulated dairy 2 industry, or invest outside of dairy. 3 Moving on to Proposal 7: Make Allowances proposed 4 by National Milk Producers Federation. 5 GN supports the Make Allowance proposal brought 6 forth by Wisconsin Cheese Makers Association and the 7 International Dairy Foods Association because it is 8 well-supported by studies (studies which I understand were 9 shared before the start of this hearing). 10 In contrast, the National Milk Producers 11 Federation (NMPF) proposal lacks transparency. While NMPF 12 clearly acknowledges the need for updated Make Allowances 13 in their petition, they offer no methodology to their 14 approach other than to say their, "…Make Allowance 15 increases represent a fair balance between the producer 16 impact of higher Make Allowances and the processor impact 17 of Make Allowances." 18 This statement, and similar ones later, imply they 19 are asking USDA to ignore a scientific approach to setting 20 minimum FMMO minimum prices and instead use what appears 21 to be a politically negotiated number. 22 Since the Class III and IV minimum milk pricing 23 series started in the year 2000, USDA has relied on 24 empirical studies to set Make Allowances. Specifically, 25 they have relied on audited manufacturing cost studies 26 from the California Department of Agriculture (CDFA) and 27 non-audited studies, which are similar to Dr. Stephenson’s 28 recent manufacturing cost studies. 3721 1 Furthermore, the Make Allowances proposed by NMPF 2 is even lower than from the last available audited CDFA 3 study from 2016 for cheese ($0.24 per pound proposed 4 versus $0.2454 per pound in CDFA 2016). Since 2016, we 5 have nearly seven years of cheese manufacturing cost 6 inflation that has not been accounted for. 7 To conclude this topic, we urge USDA to adopt the 8 data driven approach to Make Allowance estimates as 9 proposed by WMCA and IDFA. 10 Proposal 3: Elimination of cheddar cheese 11 500-pound barrels from protein price. 12 GN opposes the elimination of 500-pound barrels 13 from the protein price and maintains that the status quo 14 is a better system. While we sympathize with the view 15 that the unstable relationship between block and barrel 16 prices in Class III have caused a variety of problems for 17 the industry, removing the price series from Class III 18 protein would create other, even greater problems. 19 First, moving Class III to a 100% block weighting 20 would greatly complicate milk pricing for manufacturers 21 that make barrel cheese. Barrels produced in the U.S. are 22 almost always sold based on the CME spot barrel price, 23 while Proposal 3 would essentially disconnect Class III 24 milk pricing from the CME barrel (Figure 1). 25 The resulting disconnect between revenue and the 26 Class III milk price could drastically increase margin 27 volatility and ability to compete for milk – even for 28 barrel manufacturers outside the FMMOs. 3722 1 Our barrel plant in Gooding, Idaho, which is 2 outside the FMMO system, frequently uses a basis to 3 Class III to buy/sell milk for plant balancing purposes, 4 while most milk handlers and dairy farmers also use 5 Class III as a competitive benchmark in Idaho. 6 The removal of barrels from the protein price 7 would essentially put barrel manufacturers and their milk 8 suppliers on an island and disconnected from the Class III 9 price surface. This would be a major strategic risk for 10 our Idaho business, which produces a lot of barrel cheese. 11 While we realize the unpredictable relationship 12 between block and barrel prices in Class III has created 13 challenges in the industry, removing barrels from the 14 protein formula will create more significant industry-wide 15 challenges. 16 If this issue is going to be further explored, we 17 believe it should be done outside the FMMO system. For 18 example, there has been a discussion in the industry about 19 eliminating the CME barrel market. Such a solution would 20 negate the need to remove barrels from the NDPSR since 21 barrels would likely become a reflection of the block 22 market. 23 Proposal 4: Addition of 640-pound cheddar cheese 24 blocks to protein price. 25 GN opposes the addition of the 640-pound blocks of 26 cheese into the protein price. The first reason we oppose 27 it is because we believe it will not add new information 28 to the survey. In our experience, 640-pound cheddar 3723 1 blocks are virtually always priced off a basis to the CME 2 block cheddar price, so I would expect any NDPSR 640-pound 3 cheddar survey to track virtually perfectly with the 4 current NDPSR 40-pound block cheddar price. 5 The second reason we oppose adding 640-pound 6 blocks to the Class III price is the risk CME would add a 7 640-pound cheddar spot market, much like the current CME 8 cheddar block and barrel spot markets. All NDPSR dairy 9 markets currently have a corresponding CME spot market, so 10 it is not a stretch to assume CME would also add a 11 640-pound blocks. 12 The problem with a 640-pound CME block market is 13 the fact there is a smaller pool of buyers and sellers 14 versus the more liquid 40-pound block market on the CME. 15 A small number of buyers and sellers could more easily 16 sway a CME 640-pound block market in ways that are not 17 helpful to the larger industry or dairy producers linked 18 to Class III. 19 Basically, 640-pound blocks on the CME spot market 20 could become "barrels 2.0" in the Class III price with 21 unpredictable and volatile relationships to the current 22 40-pound block price, which would then feed into the 23 Class III protein formula. In future hearings, 24 petitioners could be asking to take out 640-pound blocks 25 from the Class III protein price for the same reasons we 26 are discussing taking out barrels today. 27 In summary, we would ask USDA to reject 28 Proposal 4. 3724 1 Proposal 6: Addition of mozzarella to the protein 2 price. 3 GN opposes the proposal to add mozzarella to the 4 Class III protein price for several reasons. First, the 5 mozzarella price would be difficult to incorporate into 6 the Class III protein price formula. Mozzarella has very 7 different fat, solids-nonfat, and moisture levels compared 8 to a very standard cheddar cheese, which is the current 9 foundation of the Class III protein formula. 10 To integrate mozzarella into the protein price 11 would require a separate and unique protein formula that 12 is weighted into the current cheddar-based protein 13 formula. Depending on the weightings of cheddar versus 14 mozzarella in a new NDPSR price survey, the protein 15 formula would be constantly changing. 16 Second, mozzarella has many different 17 specifications, some of which are made to order for 18 specific customers. Unless one specification was 19 identified as accurate to use in the protein formula, even 20 more protein formulas would be needed to account for the 21 different product compositions. In this case, USDA would 22 need to survey a broad spectrum of the mozzarella price 23 surface and weight many different protein formulas, that 24 fluctuate with surveyed weightings, to get an accurate 25 price. Chaos would ensue. 26 In addition, for the current Class III and IV 27 Make Allowances from the 2007 decision, the CDFA 28 Make Allowances data sets and the 2019 and 2022 Stephenson 3725 1 studies only use cheddar cheese in their analysis. A new 2 robust cost study would need to be created for mozzarella 3 and its many variations before it could be integrated into 4 a new Class III protein price formula. This would be very 5 challenging from a time perspective to integrate into the 6 Final Decision since the petitioners have presented no 7 such study. 8 Further, the latest cheddar Make Allowance data 9 sets have a certain level of history and trust built into 10 them which makes them easier to sense check. A new 11 mozzarella study would probably need to be audited, like 12 the past CDFA cheddar studies, to create some level of 13 confidence in the industry. 14 Lastly, the petitioners imply there are lavish 15 profits associated with the production and sale of 16 mozzarella. Specifically, they point to a competitive 17 USDA bid for consumer packaged mozzarella string cheese, 18 which was awarded at $3.56 to $3.89 per pound as evidence 19 of excess profits. 20 The first issue is that this was a solicitation 21 for packaged consumer product, not for FOB bulk wholesale 22 product, as is collected through the NDPSR for milk 23 pricing. As we know, there can be large price differences 24 between bulk commodity wholesale products and consumer 25 packaged products. 26 The second issue is that, upon searching for 27 generic brand mozzarella string cheese online for pickup 28 at a local Kroger, at the time of this writing, the price 3726 1 was $4.49 per 12 ounce package ($5.99 per pound). USDA 2 appears to have gotten a bargain. 3 Third, cheese makers are smart for the most part, 4 so if there were extreme profits associated with 5 mozzarella production, huge amounts of investment would 6 follow. Along these lines, there are already cheese 7 makers with plants that can flex production between 8 cheddar and mozzarella to maximize profits. 9 Based on our experience watching markets, these 10 manufacturers do flex their production based on expected 11 returns. Overall, mozzarella does not appear to be as 12 lucrative as the petitioners claim and adding it into the 13 Class III protein price would create chaos. 14 We ask USDA to reject this proposal. 15 Proposals 10 and 11: Increase butterfat recovery 16 in Class III to 93% and eliminate Class III farm-to-plant 17 shrink. 18 GN opposes the proposed increase in butterfat 19 recovery and elimination of farm-to-plant shrink. We 20 support the status quo until audited plant cost studies 21 can be completed that show real world yields, shrink, and 22 dairy solids recovery. This issue is very complex with 23 broad ranges for fat recovery in the industry based on 24 plant age and processing techniques. 25 While in our experience many modern plants can 26 achieve 93% cheddar fat recovery (as the petitioner 27 contends) and probably see relatively low farm-to-plant 28 shrink (but not 0%), we believe the proposals only focus 3727 1 on price-enhancing aspects of the Class III formula while 2 ignoring the parts that overvalue milk within Class III. 3 For example, the current Class III formula 4 incorrectly assumes all excess fat from the cheese making 5 process is recovered. Specifically, at 2.9915% protein 6 and 3.5% fat (standard Class III test), the current 7 formula stipulates 90% of fat goes towards cheese making, 8 with the remaining 10% being recovered as sweet cream, 9 which is valued using the NDPSR Grade AA butter price. 10 The 90% cheese fat recovery plus the 10% sweet cream fat 11 recovery add to 100% recovery. 12 The first problem here is that there is no such 13 thing as a lossless manufacturing system. All plants lose 14 milk solids, which in our case go into wastewater (and 15 often recovered as biogas). While we do not measure 16 farm-to-plant losses, for simplicity, we do measure total 17 losses from farm through our entire manufacturing system, 18 primarily through the measurement of milk solids in our 19 wastewater. 20 Even with highly efficient plant equipment and 21 mostly full milk tanker loads, in our experience modern 22 cheese plants are expected to lose about 1.5% of the 23 purchased milk solids. 24 Specifically for fat, about 1.5% of farm test fat 25 ends up in wastewater primarily because of equipment 26 clean-outs and the milk ultrafiltration process prior to 27 entering the vat. This lost fat is completely 28 unmarketable. To quantify the impact to Class III at 3728 1 standard components (2.9915% protein, 3.5% fat), using 2 $2.3475 per pound butter (the same ten-year markets as 3 used in the petitioner's analysis), and the current 4 Make Allowance and butter yield factors, this loss would 5 equal $0.14 per hundredweight of milk (see Figure 6). 6 The second problem is that the Class III formula 7 values the remaining 10% of the fat not going into cheese 8 (which is called whey cream) using the NDPSR Grade AA 9 butter price. Per USDA regulations, butter with a whey 10 flavor would be assigned as Grade B butter. As such, we 11 see about 20% discounts or more for whey fat versus the 12 Grade A sweet cream due to its limited marketability. 13 This discrepancy can easily overvalue Class III fat 14 another $0.17 per hundredweight (see Figure 6). 15 Further, included in Figure 7 is an algebraically 16 simplified version of the current Class III protein price 17 and fat value explanation that may make this topic easier 18 to understand. 19 In summary, we urge USDA to reject Proposals 10 20 and 11 regarding cheese fat retention and farm-to-plant 21 shrink. The confounding factors identified above would 22 decrease Class III by a combined $0.31 per hundredweight 23 versus the $0.12 per hundredweight increase Proposals 10 24 and 11 would bring (using the petitioner's ten-year 25 average market analysis). 26 Given the vast complexity of these issues, 27 differences in plant equipment and operations, and the 28 fact critical parts of the Class III formula overvalue 3729 1 milk, we should wait for a USDA audited cost study to be 2 completed so we can accurately measure real world yield 3 factors across a variety of plants. 4 Q. I somewhat hate to ask you to do this, but could 5 you please turn to your other exhibit, Hearing 6 Exhibit 197, and please read that into the record. It is 7 only three pages, so hopefully not too hard to do that as 8 well. 9 You don't have to introduce yourself. 10 A. Thank you. Thank you. 11 USDA should not delay FMMO reform due to risk 12 management. 13 NMPF and IDFA agree that timely increases in 14 Make Allowance are needed: 15 Petitions and testimony coming from both the 16 National Milk Producers Federation (NMPF) coalition and 17 the International Dairy Foods Association (IDFA) coalition 18 have made it abundantly clear that: 19 (1) Outdated Make Allowances are a source of 20 disorderly marketing; 21 (2) There is an urgent need for reform. 22 Specifically, NMPF states in their petition that, 23 "There are consequences to setting Make Allowances too low 24 relative to the actual cost of manufacturing under a 25 system of PPFs. Inadequate Make Allowances challenge 26 manufacturing operations' abilities to pay minimum 27 announced milk prices and still operate their facilities 28 at a reasonable rate of return. This discourages the 3730 1 plant investment needed to provide market demand on a 2 daily, seasonal, and annual basis." 3 Further, NMPF quotes USDA in its 1999 Final 4 Decision on FMMO reform in saying, "The importance of 5 using minimum prices that are market-clearing for milk 6 used to make cheese and butter/nonfat dry milk cannot be 7 overstated. The prices for milk used in these products 8 must reflect supply and demand and must not exceed a level 9 that would require handlers to pay more for milk than 10 needed to clear the market and make a profit." 11 Dairy cooperative members have also spoken to the 12 need for urgent action. Rob Vandenheuvel of CDI, in his 13 written testimony, noted, "The issue of establishing 14 appropriate manufacturing cost allowances, hereafter 15 Make Allowances, in the Federal Order formulas is of 16 critical importance to CDI…" 17 He further noted, "…the immediate adjustments 18 reflected in Proposal Number 7 in this hearing process are 19 also a critical need for the industry. The risk of 20 inaction or delayed action is simply too great to put the 21 issue off any further." 22 Other witnesses, such as Karl Rasch of Michigan 23 Milk Producers Association in his written testimony, also 24 acknowledged the need for "urgent" action. 25 USDA has been telegraphing for years reform could 26 be coming. 27 USDA has changed FMMO regulation and pricing 28 formulas multiple times over the decades. With USDA 3731 1 commissioning the 2021 Stephenson Cost Study, which 2 collection efforts began in earnest in 2020, should have 3 clued in market participants that change could be coming. 4 Further, the existence of this hearing, and a 5 likely Final Decision roughly not expected until late 6 2024, should act as another indicator for market 7 participants that risk factors are changing. 8 Stakeholders acknowledge this. In the Chicago 9 Mercantile Exchanges (CME) last annual Form 10-K filing, 10 which provides a comprehensive view of a publicly traded 11 business financial condition, they specifically 12 acknowledge regulatory change is a risk for their business 13 model. 14 CME crush traders/arbitrage traders will adjust 15 their models to deal with Class III/IV change risk. So 16 called "crush traders" or "arbitrage traders" will often 17 take short or long positions in cheese, whey, butter, 18 NFDM, Class III milk, and Class IV milk derivatives to 19 profit off the mathematical relationships. 20 For example, if the combination of selling $1.21 21 NFDM futures and $2.20 butter futures created an implied 22 Class IV futures price of $17.47 per hundredweight, but 23 the Class IV futures could be bought at $17.35 per 24 hundredweight, an arbitrage trader could execute these 25 available derivatives to lock in a $0.12 per hundredweight 26 profit. Whether the market goes up or down, their margin 27 is secure as long as the milk formula remains constant. 28 The same principles apply to Class III and its market 3732 1 components. 2 If, for example, Make Allowances were to change, 3 the relationship between the market prices and the milk 4 prices would also change. In this case, the arbitrage 5 traders would change their buy/sell formulas to reflect 6 the IDFA or NMPF Make Allowance changes for the beginning 7 of 2025, or near that time period. 8 The Make Allowances they would choose in their 9 models (IDFA vs. NMPF) would depend on what gave them the 10 larger margin cushion depending on what side of the trade 11 they were on. 12 Given the Make Allowances proposed by IDFA 2025 13 and the NMPF only differ by $0.19 per hundredweight for 14 Class III milk and $0.15 per hundredweight for Class IV 15 milk, this should give arbitrage traders a reasonable 16 level of confidence to adjust their risk models 17 accordingly. While USDA could technically set 18 Make Allowances substantially outside what the major 19 industry groups are petitioning for, the chances seem low. 20 The dairy industry can hedge with individual 21 commodities, not just Class III and IV milk derivatives. 22 In a worst case where dairy producers, for 23 example, were having a hard time finding liquidity to sell 24 Class III or IV milk futures or options due to lack of 25 arbitrage trader's liquidity, they could also hedge with 26 individual commodity prices directly. 27 In fact, GN's Idaho direct ship producers 28 typically hedge directly by selling CME CSC cheese futures 3733 1 (settles to NDPSR cheese price). Given the cheese price 2 is typically the vast majority of their milk pay price, 3 the hedges are effective. This allows them to correlate 4 their mailbox price to the CME derivative regardless of 5 Make Allowance changes. Producers in different orders can 6 hedge with more NFDM, butter, or dry whey to reflect their 7 mailbox milk price. 8 Figures 1 through 3 show the USDA mailbox milk 9 price correlations for risk management for individual 10 dairy commodities versus hedging the Class III and IV milk 11 prices. The analysis shows that effective hedges can be 12 created using only the commodity futures/options. Risk 13 management brokers or the producer's milk handler can 14 easily provide guidance on appropriate weightings and 15 volumes of the commodities the dairies should hedge with. 16 CME makes money by transaction counts, is 17 sensitive to needs of arbitrage trading community. 18 One reason CME's testimony is sensitive to 19 "liquidity providers" is due to the amount of fee revenue 20 they generate. In their last Form 10-K filing, CME 21 states, "Our revenue is substantially derived from fees 22 for transactions executed and cleared in our markets." 23 Given that crush traders are taking multiple parts 24 of dairy markets and figuratively "crushing" them together 25 requires multiple transactions to accomplish. For 26 example, crushing a Class III milk contract could involve 27 buying a Class III milk contract and selling cheese, dry 28 whey, and butter derivatives at the same time. This is 3734 1 four transactions the CME benefits from. For dairy 2 farmers managing their risk using only cheese derivatives, 3 or maybe only one or two additional commodities, there are 4 less transactions involved. 5 While this section of the testimony is not meant 6 to say CME is nefarious for charging for their valuable 7 services, or that CME market liquidity is not very 8 important for the industry, it is meant to point out their 9 interests are not always aligned with the broader dairy 10 industry. 11 Conclusion: We urge USDA to not delay reform 12 implementation due to risk management concerns. The 13 industry knows change is coming, within a reasonable level 14 of certainty in scope, and dairy producers should still be 15 able to hedge. The CME's concerns about liquidity impacts 16 are worth noting, but their concerns are not necessarily 17 rooted in the health of the broader industry. 18 If USDA ultimately decides to delay the 19 implementation, GN would support skipping the IDFA 20 proposed four-year phase-in approach to Make Allowance 21 reform and instead move straight to the maximum 2028 22 proposed levels. 23 That concludes my testimony. 24 MR. ROSENBAUM: Your Honor, I may have one or two 25 questions, but it's -- we're after five o'clock, so I 26 would suggest we break for the day and start tomorrow 27 morning. 28 THE COURT: Yeah, I think so. We'll start with 3735 1 you with this witness and then have cross-examination. 2 Let's go off the record. 3 (Whereupon, the proceedings were concluded.) 4 ---o0o--- 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 3736 1 STATE OF CALIFORNIA ) ) ss 2 COUNTY OF FRESNO ) 3 4 I, MYRA A. PISH, Certified Shorthand Reporter, do 5 hereby certify that the foregoing pages comprise a full, 6 true and correct transcript of my shorthand notes, and a 7 full, true and correct statement of the proceedings held 8 at the time and place heretofore stated. 9 10 DATED: October 10, 2023 11 FRESNO, CALIFORNIA 12 13 14 15 16 MYRA A. PISH, RPR CSR Certificate No. 11613 17 18 19 20 21 22 23 24 25 26 27 28