Summary
upholding agency's determination of a prevailing wage rate, where the agency relied on certain statistical sampling methods to compute the rate and the plaintiff argued that those methods were arbitrary and capricious, because the agency "made difficult decisions" and "applied reliable—if limited and imperfect—statistical models" and the challenger had "failed to show that [the agency] made sufficiently egregious statistical or mathematical errors to warrant setting that decision aside"
Summary of this case from Ariz. Health Care Cost Containment Sys. v. Ctrs. for Medicare & Medicaid Servs.Opinion
No. 1:19-cv-03180-SMJ
03-02-2020
ZIRKLE FRUIT COMPANY, a Washington Corporation, Plaintiff, v. UNITED STATES DEPARTMENT OF LABOR; Patrick Pizzella, in his official capacity as Acting United States Secretary of Labor; John P. Pallasch, in his official capacity as Assistant Secretary of Labor, Employment & Training Administration, United States Department of Labor; Cheryl M. Stanton, in her official capacity as Administrator of the Wage & Hour Division, United States Department of Labor; Defendants.
Brendan Victor Monahan, Jaime Cuevas, Jr., John-Philip J. Schroeder, Sarah Lynn Wixson, Stokes Lawrence Velikanje Moore & Shore, Yakima, WA, Lance A. Pelletier, Stokes Lawrence PS, Seattle, WA, for Plaintiff. John T. Drake, Rudolf J. Verschoor, Vanessa Ruth Waldref, U.S. Attorney's Office, Spokane, WA, for Defendants.
Brendan Victor Monahan, Jaime Cuevas, Jr., John-Philip J. Schroeder, Sarah Lynn Wixson, Stokes Lawrence Velikanje Moore & Shore, Yakima, WA, Lance A. Pelletier, Stokes Lawrence PS, Seattle, WA, for Plaintiff.
John T. Drake, Rudolf J. Verschoor, Vanessa Ruth Waldref, U.S. Attorney's Office, Spokane, WA, for Defendants.
ORDER DENYING PLAINTIFF'S MOTION FOR DECLARATORY JUDGMENT AND INJUNCTION AND GRANTING DOL DEFENDANTS' MOTION FOR PARTIAL SUMMARY JUDGMENT
SALVADOR MENDOZA, JR., United States District Judge
Before the Court, without oral argument, are Plaintiff Zirkle Fruit Company's ("Zirkle") Motion for Declaratory Judgment and Injunction, ECF No. 87-1, the United States Department of Labor (DOL) Defendants' Response to Zirkle's motion and Cross-Motion for Partial Summary Judgment, ECF No. 88, and Defendant-Intervenor State of Washington Employment Security Department's (ESD) Response to Zirkle's motion, ECF No. 90.
The Court regards ECF No. 87-1, a corrected version of Plaintiff's motion, as the operative filing and disregards the original filing, ECF No. 81.
The DOL Defendants include the United States Department of Labor; Eugene Scalia, in his official capacity as United States Secretary of Labor; John P. Pallasch, in his official capacity as Assistant Secretary of Labor, Employment & Training Administration, United States Department of Labor; and Cheryl M. Stanton, in her official capacity as Administrator of the Wage & Hour Division, United States Department of Labor. See ECF No. 48 at 1.
The Court is essentially presented with cross-motions for summary judgment on Zirkle's claim under the Administrative Procedures Act (APA) regarding the prevailing wage rate (PWR) for blueberry harvesting in Washington certified by DOL during the 2019 harvest. See generally ECF No. 48. For the reasons that follow, the Court finds Zirkle has failed to establish either that the methods ESD employed in calculating the 2019 blueberry PWR, or the process DOL employed in certifying that result, were arbitrary or capricious. As such, the Court denies Zirkle's motion for declaratory and injunctive relief and grants Defendants' motions for partial summary judgment.
BACKGROUND
Zirkle is a farming company based in Selah, Washington. ECF No. 48 at 3. Zirkle grows a variety of crops and is responsible for approximately ten percent of the state's annual blueberry production. Id. at 7. Zirkle contends it is an outlier in the blueberry industry in that it harvests its annual crop entirely by hand, relying on a combination of domestic and foreign laborers. Id. at 3. Each year, many of Zirkle's foreign laborers—2750 of them, for the 2019 blueberry harvest—arrive by way of the H-2A program, which authorizes visas for temporary agricultural workers when there is a shortage of domestic laborers in a particular region. See id. at 10; see also Hispanic Affairs Project v. Acosta , 901 F.3d 378, 382 (D.C. Cir. 2018) (citing 8 U.S.C. § 1101(a)(15)(H)(ii)(a) ).
To ensure that foreign laborers hired through the H-2A program do not depress the wages of domestic laborers in the same agricultural activity, H-2A employers are obligated to pay H-2A laborers the highest of four potential wages: the adverse effect wage rate, any collectively-bargained wage, the applicable state or federal minimum wage, or the prevailing hourly or piece wage rate ("prevailing wage rate" or "PWR"). 20 C.F.R. § 655.120(a). Only the PWR is at issue in this case.
A. The PWR
The PWR is intended to reflect the average wage paid to domestic laborers engaged in a given agricultural activity in the same region in which H-2A laborers will be employed. 20 C.F.R. § 653.501(c)(2)(i). DOL administers the H-2A program and is ultimately responsible for setting the PWR. Id. But the regulations expressly delegate to state "workforce agencies" (SWAs), like Washington's ESD, the task of gathering prevailing wage data and calculating the PWRs for their state. Id.
SWAs are guided in the process of calculating the PWR in part by a DOL publication known as Handbook 385. See generally AR 1 –25. Handbook 385 sets out standards both for conducting prevailing wage surveys, see AR 2 – 5, and for calculating PWRs from those results, see AR 5 –7. After the SWA has calculated what it believes to be the PWRs, it submits those findings to DOL, which reviews the information and "determines whether the survey results may be validated." ECF No. 24 at 3–4. If so, the PWR is published, and H-2A employers must pay it immediately, even if the change comes mid-harvest. 20 C.F.R. § 655.120(b).
References to the Administrative Record (AR), ECF Nos. 62, 62-1, 62-2, 62-3 & 62-4, are to the page number provided, rather than the corresponding ECF page number.
Certain aspects of these processes are disputed in this litigation and explained in more detail below. Insofar as aspects of the process are not relevant to the Court's decision, they are not recited here.
1. PWR for Blueberries in Washington
The PWR for blueberries in Washington is of relatively recent vintage. See ECF No. 23 at 4. In 2016, the first year one was calculated, it was $0.47/lb.; in 2017 it was $0.50/lb. with a guarantee of $9.47 per hour; and in 2018 it was $11.00 per hour with no piece-rate wage. Id. at 4–5; see also ECF No. 23-4. In March 2019, before ESD completed that year's work on the prevailing wage determination, DOL approved Zirkle's application to hire 2750 foreign laborers for the blueberry harvest at a piece-rate wage of $0.50/lb. See AR 791–94, 863–66.
The 2019 PWR survey was conducted by online survey, telephone calls, and forms sent through the mail. See AR 721; see also ECF No. 23 at 10. Before ESD began the survey, it met with "stakeholders" in the Washington agricultural community and previewed the survey form it planned to use. ECF No. 23 at 5–6. Although Zirkle was invited to this presentation, it is unclear if it attended. Id. at 18–19. ESD did, however, specifically solicit Zirkle's feedback on the form of the survey, and Zirkle indicated it had no concerns. Id. at 7.
The data used to calculate the 2019 PWR for blueberries was collected between October 2018 and January 2019. ECF No. 23 at 7. For clarity, the Court refers to this as the 2019 PWR survey.
ESD interviewed laborers in person after the survey ended but did not include that information in setting the PWR. ECF No. 23 at 10.
Washington economist Joshua Moll oversaw the 2019 survey and calculated the 2019 PWR for blueberries, among other crops. Id. at 1–3. He estimated that 5622 laborers worked for 214 growers during the 2019 blueberry harvest's busiest week. Id. at 13. Moll derived this estimate using three statistical models. First, he used a "capture-recapture" algorithm, a statistical model used to estimate the size of a population for which no definitive census exists after at least two "capture occasions," such as, in this case, an employer's response or lack thereof to an annual survey. See ECF No. 88 at 3. Next, Moll drew on data from the state's unemployment insurance database and applied a "K-means test" to categorize employers into "small," "medium," and "large" categories. Id. Finally, Moll applied a "raking algorithm" to the data and calculated "survey weights" based on the estimated likelihood of employers from each category responding to the survey. Id. Moll applied these survey weights to responses from nineteen employers, only one of which fell within the "large" category, and determined that the survey, which had generated actual wage information for 1786 domestic blueberry laborers, represented roughly one-third the total estimated population. ECF No. 88 at 34; ECF No. 24-2 at 7.
ESD received responses from fifty-four blueberry growers, but only nineteen employed a piece-rate wage structure suitable for consideration in determining the PWR. See ECF No. 88 at 34 n.9.
Zirkle did not respond to the survey. ECF No. 23 at 17; ECF No. 31 at 12. Once the survey period closed, ESD calculated the new PWR for blueberries as $0.75/lb. ECF No. 23 at 17. After the survey was complete, ESD again met with stakeholders and previewed the updated PWRs for them. ECF No. 23 at 18–19. ESD then reported its findings to DOL. Id. Nelson Patterson, a DOL analyst, confirmed that the sample size of ESD's survey was adequate, reviewed ESD's methodology and calculation of the PWR, and published the result. ECF No. 24 at 4–5; ECF No. 91-3 at 6–7. On July 24, 2019—in the seventh week of an approximately fifteen-week harvest, see ECF No. 4 at 9—DOL notified Zirkle of the increased PWR, which Zirkle was required to immediately pay to its H-2A employees. AR 788.
2. Procedural History
Zirkle sued DOL less than two weeks later, seeking a temporary restraining order enjoining DOL from enforcing the increased PWR. ECF Nos. 1, 4. Roughly one month later, the Court granted Zirkle's motion for a preliminary injunction, temporarily prohibiting DOL from enforcing the $0.75/lb. PWR until the merits of Zirkle's APA claims were resolved. ECF No. 49 at 24–25. Zirkle subsequently filed a First Amended Complaint challenging the updated PWR for high-density apple harvesting, as well. ECF No. 48. On January 9, 2020, the Court held a scheduling conference and set an expedited briefing schedule on Zirkle's APA claims with regard to the blueberry PWR only. See ECF No. 66 at 3–4. The Court subsequently granted Zirkle's motion to take the deposition of Moll, the ESD economist who oversaw the 2019 survey and PWR calculation process, and of the then-unidentified DOL analyst responsible for certifying the 2019 blueberry PWR. ECF No. 79 at 6–11. The Court also ordered the production of the underlying survey responses from which ESD calculated the PWR, with limited redactions. Id. at 11–13.
On February 14, 2020, Zirkle moved for a permanent injunction prohibiting enforcement of the 2019 blueberry PWR and a declaratory judgment that DOL's adoption of the PWR was arbitrary and capricious. ECF No. 81. The DOL Defendants and ESD individually filed responses, and DOL also moved for summary judgment in its response. ECF Nos. 88, 90. To ensure the expeditious resolution of Zirkle's claims and consistent with the Court's Scheduling Order, Zirkle has not filed a reply. See ECF No. 66 at 4.
The Court construes Zirkle's "Objections to Defendants' ECF 89," which requests the Court "disregard" Defendants' Joint Statement of Undisputed Facts, ECF No. 89, as a motion to strike that filing. ECF No. 92 at 2. That motion is granted, and the Court has considered nothing set forth in Defendants' Joint Statement of Undisputed Facts in ruling on the cross-motions for summary judgment. See LCivR 56(i) ("The procedures described in LCivR 56(c)(1) do not apply to administrative record review cases ....").
LEGAL STANDARD
Under the Administrative Procedures Act (APA), a court must invalidate "agency action, findings, and conclusions" that are "arbitrary, capricious ... or otherwise not in accordance with law." 5 U.S.C. § 706(2)(A).
Review under the arbitrary and capricious standard is narrow, and [the Court does] not substitute [its] judgment for that of the agency. Rather, [it] will reverse a decision as arbitrary and capricious only if the agency relied on factors Congress did not intend it to consider, entirely failed to consider an important aspect of the problem, or offered an explanation that runs counter to the evidence before the agency or is so implausible that it could not be ascribed to a difference in view or the product of agency expertise.
League of Wilderness Defs. Blue Mountains Biodiversity Project v. Allen , 615 F.3d 1122, 1130 (9th Cir. 2010) (quoting Lands Council v. McNair , 537 F.3d 981, 987 (9th Cir. 2008) ). Judicial "deference is highest when reviewing an agency's technical analyses and judgments involving the evaluation of complex scientific data within the agency's technical expertise." Id. Moreover, agency action is not arbitrary and capricious simply because it relies on a "dataset [that] was less than perfect." Dist. Hosp. Partners, L.P. v. Burwell , 786 F.3d 46, 61 (D.C. Cir. 2015).
However, "[a]n agency's unannounced departure in practice from a written regulation is a distinct form of agency action that is challengeable, separate and apart from adoption of the regulation itself." Acosta , 901 F.3d at 387. Thus, although a plaintiff cannot prevail simply by showing that an agency reached an imperfect conclusion, if the agency followed a "general policy by which its exercise of discretion [was] governed, an irrational departure from that policy"—rather than "an avowed alteration of it"—may be arbitrary and capricious independent of the result it reaches. I.N.S. v. Yueh-Shaio Yang , 519 U.S. 26, 32, 117 S.Ct. 350, 136 L.Ed.2d 288 (1996).
DISCUSSION
Zirkle's motion for declaratory judgment and permanent injunction ascribes error to ten separate aspects of ESD and DOL's actions in certifying the 2019 blueberry PWR. See ECF No. 87-1. Each of these ten claims falls within one of three broad arguments: (1) that ESD unjustifiably departed from the requirements of Handbook 385, (2) that DOL failed to adequately scrutinize ESD's findings, and (3) that ESD used flawed and inconsistent statistical models, and therefore reached flawed conclusions, in calculating the PWR. The Court addresses each of these arguments in turn.
A. ESD did not improperly deviate from the requirements of Handbook 385
Zirkle first argues that ESD erred in failing to adhere to several of Handbook 385's requirements and that DOL's decision to certify the resulting PWR was therefore arbitrary and capricious. See ECF No. 87-1 at 7–13. The fact that ESD did not strictly observe all of Handbook 385's requirements is clear from the record—for example, Handbook 385 dictates that wage surveys be conducted using a substantial number of in-person employer interviews, while the 2019 prevailing wage survey included no in-person interviews. See AR 5. This conclusion does not, however, automatically entitle Zirkle to relief, because Defendants contend the requirements of Handbook 385 have been amended by guidance from DOL over time. See ECF No. 88 at 11–12. The success or failure of this aspect of Zirkle's claims therefore hinges on a threshold inquiry: whether DOL was entitled to amend or revoke the provisions of Handbook 385 and, if so, whether it did so with respect to the provisions Zirkle claims were violated.
1. DOL properly amended Handbook 385's procedures over time
As the Court concluded at the preliminary injunction stage, Handbook 385 is the primary—and certainly the original—source of authority for prevailing wage surveys. See ECF No. 49 at 3 n.1. However, with the benefit of a more robust administrative record, limited extra-record discovery, and additional briefing, it has become clear that Handbook 385 is not the sole source of authority governing the PWR survey process. Instead, Handbook 385 must be interpreted in light of subsequent policy pronouncements from DOL.
The Court's analysis begins with the fundamental proposition that "not all agency policy pronouncements which find their way to the public can be considered regulations enforceable in federal court." Rank v. Nimmo , 677 F.2d 692, 698 (9th Cir. 1982) (quoting Chasse v. Chasen , 595 F.2d 59, 62 (1st Cir. 1979) ). Rather, only "legislative rules" have the force and effect of law in the same way as would a congressionally enacted statute. Chrysler Corp. v. Brown , 441 U.S. 281, 295, 99 S.Ct. 1705, 60 L.Ed.2d 208 (1979). To constitute a legislative rule, an agency pronouncement must "(1) prescribe substantive rules—not interpretive rules, general statements of policy or rules of agency organization, procedure or practice—and, (2) conform to certain procedural requirements." Rank , 677 F.2d at 698 (citing Chrysler Corp. , 441 U.S. at 301, 99 S.Ct. 1705 ). A rule is substantive when it affects "individual rights and obligations." Id.
An agency may not amend or revoke a legislative rule without public notice and comment. Clean Air Council v. Pruitt , 862 F.3d 1, 9 (D.C. Cir. 2017) (quoting National Family Planning and Reproductive Health Association, Inc. v. Sullivan , 979 F.2d 227, 234 (D.C. Cir. 1992) ). But non-legislative rules, which may be enacted without the notice and comment process, may also be amended or revoked without those procedures. Put more simply, the APA only requires "agencies use the same procedures when they amend or repeal a rule as they used to issue the rule in the first instance." See Perez v. Mortg. Bankers Ass'n , 575 U.S. 92, 101, 135 S.Ct. 1199, 191 L.Ed.2d 186 (2015) (citing F.C.C. v. Fox Television Stations, Inc. , 556 U.S. 502, 515, 129 S.Ct. 1800, 173 L.Ed.2d 738 (2009) ).
The Court concludes Handbook 385 is a statement of agency practice or procedure, and not a legislative rule. For one thing, Handbook 385 was not promulgated through the notice and comment rulemaking process. More importantly, Handbook 385 does not prescribe "individual rights and obligations," the hallmark of a legislative rule. Rank , 677 F.2d at 698. Rather, individual rights and obligations associated with the H-2A program—including, most importantly, the requirements for what agricultural employer participants must pay non-immigrant foreign laborers—are set out in the Code of Federal Regulations. See 20 C.F.R. § 655.120(a). Handbook 385, by contrast, simply dictates the procedure SWAs, as DOL's delegees, are to follow in calculating the PWR. See AR 1. In short, Handbook 385 is a statement of agency procedure or practice and therefore, DOL was not required to undergo the notice and comment process before amending its requirements. See Rank , 677 F.2d at 698 ; Perez , 575 U.S. at 101, 135 S.Ct. 1199.
In granting Zirkle's motion for a preliminary injunction, the Court observed that DOL was contemporaneously engaged in notice and comment rulemaking for regulations intended to supplant Handbook 385. ECF No. 49 at 3 n.1 (citing 84 Fed. Reg. 36168 ). However, this does not indicate that DOL was required to undergo notice and comment rulemaking in authorizing departures from Handbook 385's provisions that were issued without this rulemaking process. Rather, the current notice and comment process appears to be part of a broader initiative to modernize DOL's regulations governing the H-2A program overall. See 84 Fed. Reg. 36168–36301. Moreover, nothing prohibits DOL from engaging in a more onerous process than necessary to implement a revised approach to the PWR process.
Nevertheless, if DOL "announce[d] and follow[ed] ... a general policy by which its exercise of discretion w[ould] be governed, an irrational departure from that policy (as opposed to an avowed alteration of it)" may be arbitrary, capricious, or an abuse of discretion. See I.N.S. v. Yueh-Shaio Yang , 519 U.S. 26, 32, 117 S.Ct. 350, 136 L.Ed.2d 288 (1996) (quoting 5 U.S.C. § 706(2)(A) ). Accordingly, even if DOL was entitled to alter the requirements of Handbook 385 without public notice or comment, an "unannounced departure in practice" from a settled and consistent approach to PWR surveys could still be arbitrary and capricious. Acosta , 901 F.3d at 387.
2. ESD was not required to conduct in-person interviews
Zirkle first contends DOL acted in an arbitrary and capricious manner by certifying ESD's findings even though the survey process included no in-person interviews. Handbook 385 provides that:
All wage surveys must include a substantial number of personal employer interviews. Survey information obtained from employers may be supplemented to a limited extent by telephone or mail contacts. Under certain conditions, employer contacts by mail or by telephone may be made, in lieu of personal field contacts, but the State agency must assure itself that information gathered in this manner is representative of the rates being paid in the crop activity.
AR 5. Defendants concede that ESD's 2019 PWR survey included zero in-person interviews. See AR 721 (listing method of employer contact as "web/phone/mail".)
Nevertheless, the record makes clear that DOL's decision to not require in-person interviews was neither irrational nor unannounced. In 2013, DOL eliminated from form ETA-232, which SWAs use to report PWR findings, the column used to provide the number of worker interviews conducted, observing that "most states no longer conduct field surveys due to reduced funding" and were instead "opting for mail, fax, or telephone surveys thereby making the worker interview process obsolete." AR 222.
Moreover, every year DOL issues "Training and Employment Guidance Letters" (TEGLs) providing "programmatic guidance and financial information to SWAs to support state foreign labor certification activities" in the coming year. See, e.g. , AR 88. Since at least 2016, DOL directed SWAs to "prioritize ... limited resources" to conduct PWR surveys in a manner consistent with Handbook 385 that would "yield statistically valid wage findings." AR 154. In early 2017, DOL expressly recognized that the PWR survey could be conducted using a variety of survey means including "telephone survey." AR 134. Both the 2018 and 2019 TEGLs contained the same direction. AR 75, 106.
Nelson Patterson, a DOL program analyst, also testified that SWAs are not trained by DOL to require in-person interviews but are instead directed to "leverage" their funding by utilizing the most efficient survey method that would result in reliable PWR findings. See ECF No. 91-3 at 65. To that end, form ETA-232 directs the SWA to identify the "method(s) of contact with the employers (i.e. , personal interviews, telephone, mail, etc.)." See AR 26, 29. Paterson also testified in departing from the strict requirement of in-person interviews, DOL relied in part on the 2013 ruling by an Administrative Law Judge that the failure to conduct in-person interviews did not necessarily render a PWR finding invalid. AR 551–52 (noting DOL stopped training SWAs to conduct in-person interviews in 2006).
Handbook 385 must be interpreted in light of subsequent guidance issued by DOL to SWAs for the administration of prevailing wage surveys. The record is clear that DOL recognized budgetary constraints made in-person interviews impractical and therefore stopped requiring them well before ESD began the 2019 PWR survey. As such, DOL's decision to certify the 2019 blueberry PWR despite the absence of in-person interviews was neither irrational nor unannounced and was therefore not arbitrary or capricious. Yueh-Shaio Yang , 519 U.S. at 32, 117 S.Ct. 350.
3. ESD validly considered wage data for more than a one-week period
In a similar vein, Zirkle contends ESD erred in relying on wage data from a period of nine weeks despite Handbook 385's direction that "surveys should normally be completed within 3 days" and "the survey should not exceed 1 week." See ECF No. 87-1 at 10–11; AR 4. DOL responds that ESD had discretion to define the "period of peak activity" as to given agricultural activities and did not err in considering wage data from more than a one-week period. ECF No. 88 at 12–13. It notes that while Handbook 385 directs SWAs to estimate the beginning and end of the harvest for each crop and the "period of peak activity," SWAs need not include that information in reporting PWR results. Compare AR 11 with AR 20–25. And like the requirement that SWAs conduct in-person interviews, recent TEGLs no longer direct SWAs to identify the period of "peak activity," nor even the anticipated start and end dates for the harvest of each crop, but simply direct them to provide an anticipated timeframe for the PWR survey. See AR 88–115.
As with the in-person interview requirement, the record establishes that DOL's decision to certify the PWR based on wage data representing a period exceeding one week was neither an irrational nor an unannounced departure from longstanding practice. Rather, the record establishes DOL made a considered judgment in light of state budgetary constraints to depart from the requirement that wage surveys be limited to one week. The Court cannot set aside that judgment as arbitrary and capricious. See also AR 551 (Administrative Law Judge noting that due to low response rate, expanded survey timeframe "appears to have actually enhanced its potential for accuracy").
4. ESD did not err in disregarding geographical or agricultural factors
Zirkle also contends DOL acted in an arbitrary and capricious manner by failing to calculate separate PWRs applicable to different blueberry harvesting activities—such as between organic and conventional blueberries—or between different geographic regions in Washington. Defendants respond that Handbook 385 dictates that separate reporting is required only when the SWA identifies a resulting significant variation in the prevailing wage. See ECF No. 88 at 13; see also AR 2. Indeed, DOL's recent TEGLs have directed SWAs to do so only where such factors result in significant variation in the PWR. See AR 75 ("In circumstances where substantial dissimilarities in crop or related conditions exists in different parts of the state, the state agency may use sub-state reporting areas.").
More importantly, however, Moll testified that ESD sought input from agricultural stakeholders including Zirkle regarding whether there were such factors it should consider in collecting PWR data. ECF No. 91-1 at 133–34. While Zirkle provided significant information about factors affecting the prevailing wage in harvesting crops such as apples, cherries, and pears, it was silent as to factors affecting the PWR for blueberries. Id. Nor has Zirkle persuasively identified such factors in its briefing; instead, it appears to contend that ESD's failure to affirmatively consider the possibility that such factors existed was, standing alone, arbitrary and capricious. See ECF No. 87-1 at 10–11. However, Moll testified he did consider how certain factors may affect the PWR but did not find them reflected in the data before him. ECF No. 91-1 at 41 ("In terms of analysis, if it is not reported or a differing commodity activity which would include some sort of factor that may affect the wage rate is not reported, then I do not analyze something that does not get reported."). Put simply, it was not arbitrary and capricious for ESD to certify a single, undifferentiated PWR without regard for the role of geographic or agricultural factors where survey data indicated no influence from such factors.
5. ESD did not err by failing to note an increase in the PWR
Finally, Zirkle contends that ESD erred in failing to identify the fifty-percent increase in the blueberry PWR when reporting its findings to DOL on form ETA-232. ECF No. 87-1 at 11. Both ESD and DOL contend that because the 2018 blueberry PWR was expressed in terms of an hourly wage, and the 2019 PWR was a piece-rate wage, there was no increase ESD could have reported. ECF No. 90 at 21; see also See AR 31–32 (directing SWA to explain "[i]ncrease or decrease in prevailing rate from comparable period of previous year " (emphasis added)).
In weighing the equities while considering Zirkle's motion for a preliminary injunction, the Court noted the dramatic jump in wages Zirkle would be required to pay H-2A laborers mid-harvest—that is, a sudden increased in the effective wage rate Zirkle was obligated to pay. See ECF No. 49 at 15–16. While the reality facing Zirkle was a fifty percent increase in the effective wage rate it was required to pay, form ETA-232 only directs the SWA to note an increase in the wage rate year-over-year, and SWAs are expressly prohibited from equating different wage rates, as between an hourly and a piece-rate wage. See AR 6. Thus, from ESD's perspective there was no increase in the blueberry PWR it could report, and the failure to do so was not arbitrary or capricious. See ECF No. 23 at 4–5; AR 31–32.
B. DOL did not blindly "rubber stamp" ESD's PWR findings
Zirkle next contends that DOL acted arbitrarily and capriciously in "rubber stamp[ing]" ESD's PWR findings, particularly because those findings were "riddled with red flags and question-marks." ECF No. 87-1 at 7. In evaluating this allegation, the Court begins with the proposition that "an agency's reliance on a report or study without ascertaining the accuracy of the data contained in the study or the methodology used to collect the data "is arbitrary agency action, and the findings based on [such a] study are unsupported by substantial evidence." City of New Orleans v. S.E.C. , 969 F.2d 1163, 1167 (D.C. Cir. 1992).
But this standard does not require an agency to engage in a de novo determination of another entity's findings. Quite the contrary, a federal agency's reliance on a state agency's analysis without duplicating the state's evaluative process is not arbitrary or capricious, particularly where the pertinent statute or regulation specifically envisions cooperation between the state and federal governments. See City of Carmel-By-The-Sea v. United States Dep't of Transp. , 123 F.3d 1142, 1162 (9th Cir. 1997) (citing Laguna Greenbelt, Inc. v. United States Dept. of Transp. , 42 F.3d 517, 524 n.6 (9th Cir. 1994) ) (federal agency's reliance on state's assessment of environmental impact was not arbitrary or capricious).
Rather, it is only an agency's wholly unexplained acceptance of another entity's conclusions, with no apparent effort to ensure the reliability of those conclusions or the evaluative process that produced them, that is arbitrary and capricious. For example, in Home Health Care, Inc. v. Heckler , the D.C. Circuit held it was error for the decisionmaker to rely on a AAA study of average car costs reported in a local newspaper without any effort to validate the study's findings. 717 F.2d 587, 592 (D.C. Cir. 1983). Likewise, the D.C. Circuit held the S.E.C.'s unexplained acceptance of a utility's analysis that a proposed merger would result in cost savings to ratepayers was arbitrary and capricious where the agency gave no indication it had validated those findings or the methodology the utility employed. City of New Orleans , 969 F.2d at 1167 ("If the Commission knew of the methodology used by Entergy in collecting its plant replacement data, we can find no evidence of its knowledge in the record.").
With these principles in mind, the Court finds DOL gave sufficient scrutiny to ESD's findings and that the agency's decision not to duplicate ESD's analysis or to insist on the raw survey data was not arbitrary or capricious. Patterson testified that upon receipt of ESD's findings, he validated those findings to ensure both that they were derived from a sufficient sample size and were valid under the so-called "40 percent" and "51 percent" rules. ECF No. 91-3 at 5–7, 9. He testified this process took him four to five minutes. Id. Were this the extent of DOL's review—verifying that ESD's findings were sufficiently representative and arithmetically sound—the Court might be able to find DOL's review process insufficient. But Patterson testified the review process did not end there. Rather, Patterson met with Shawm Ahmed, another DOL analyst, and together they discussed certain aspects of ESD's methodology and report, though they did not particularly focus on the 2019 blueberry PWR. Id. at 6. Patterson said this second level of review was focused on "trying to understand why things were done a certain way as a whole." Id. at 7, 9–10.
DOL was entitled to rely on the validity of ESD's calculations and, after reasonably assuring itself that ESD's results were reliable, to certify those findings without engaging in a de novo review of the underlying survey data or statistical models ESD employed. See City of Carmel-By-The-Sea , 123 F.3d at 1162. This is particularly true given the applicable regulation's explicit requirement of DOL's cooperation with state workforce agencies like ESD. 20 C.F.R. § 653.501(c)(2)(i). The record is clear that after receiving ESD's report, Patterson and Ahmed engaged in a substantial review process and concluded ESD's findings were reliable. As such, DOL did not act in an arbitrary or capricious manner.
C. ESD's choice of statistical models was not arbitrary or capricious
Finally, Zirkle argues that DOL erred in certifying the 2019 blueberry PWR despite significant methodological flaws by ESD. Specifically, Zirkle contends ESD relied on statistical models ill-suited to the tasks to which they were applied and which relied on contradictory assumptions. Furthermore, Zirkle contends the result of these algorithms—estimates of the relevant employer and employee populations, categorization based on employers' size, and differing "weights" applied to responses from employers in each category—were without a sound statistical basis.
In attempting to invalidate the 2019 blueberry PWR on the basis of ESD's statistical approach, Zirkle bears a heavy burden. It is well settled that an agency's decision is not arbitrary and capricious simply because it relies on an imperfect dataset or statistical model. See Dist. Hosp. Partners, L.P. v. Burwell , 786 F.3d 46, 61 (D.C. Cir. 2015) ("[E]ven if this dataset was less than perfect, imperfection alone does not amount to arbitrary decision-making."); Appalachian Power Co. v. E.P.A. , 249 F.3d 1032, 1052 (D.C. Cir. 2001) ("That a model is limited or imperfect is not, in itself, a reason to remand agency decisions based upon it."); Allied Local and Reg'l Mfrs. Caucus v. EPA , 215 F.3d 61, 71 (D.C. Cir. 2000) ("We generally defer to an agency's decision to proceed on the basis of imperfect scientific information, rather than to ‘invest the resources to conduct the perfect study.’ " (quoting Sierra Club v. E.P.A. , 167 F.3d 658, 662 (D.C. Cir. 1999) ); see also In re Polar Bear ESA Listing , 709 F.3d 1, 13 (D.C. Cir. 2013) ; North Carolina v. FERC , 112 F.3d 1175, 1190 (D.C. Cir. 1997) ; Chemical Mfrs. Ass'n v. EPA , 28 F.3d 1259, 1265 (D.C. Cir. 1994). Nevertheless, "[w]hile courts routinely defer to agency modeling of complex phenomena," the agency must ‘explain[ ] the assumptions and methodology used in preparing the model and provide[ ] a complete analytic defense should the model be challenged.’ " In re Polar Bear ESA Listing , 709 F.3d at 13 (quoting Appalachian Power Co. , 249 F.3d at 1053–54 )).
1. ESD's use of the capture-recapture model was not arbitrary
Zirkle first alleges ESD's reliance on the capture-recapture model to estimate the number of blueberry growers in the state was inappropriate. ECF No. 87-1 at 14–15 (citing ECF No. 84-5). Zirkle relies on the expert of opinions of Stephen Bronars, Ph.D., who opines that the capture-recapture model was "a curious choice," and notes there is no evidence in the record establishing whether ESD considered its viability for use in the PWR survey. ECF No. 84-5 at 5. From this Dr. Bronars concludes "it was unreasonable to rely upon the blueberry grower and blueberry-harvesting worker counts estimated by ESD." Id. at 6. While Dr. Bronars's report opines at length about the limitations of the capture-recapture model, it does not appear to propose a superior alternative. ECF No. 84-5 at 9–14.
Moll, the ESD economist responsible for the 2019 PWR survey, testified that the capture-recapture model is a commonly used approach to estimate the size of populations for which there is no definitive census. ECF No. 91-1 at 29. Moll further testified that ESD selected the model after assuring itself, based on data from the state unemployment insurance database, that the population of blueberry growers was relatively stable, which served to assure him of the viability of the capture-recapture model. Id. at 29–30. ESD also represents it considered eight different versions of the capture-recapture model and selected the one "which accounts for variations in response rate over time, including differing response rates among small, medium, and large employers, as the best "statistical fit" for the blueberry dataset. See id. ; see also ECF No. 90 at 10.
The Court finds Zirkle has established, at most, that the capture-recapture model has limitations. Nevertheless, ESD and DOL have offered a sufficient analytic defense of the model, and Zirkle has failed to show that it was entirely inappropriate to estimate the size of Washington's blueberry grower population. Furthermore, even if Zirkle could show that the capture-recapture model was an entirely inappropriate selection, that would not by itself entitle it to relief. Handbook 385 requires the PWR be derived from a sufficiently representative sample of Washington's domestic agricultural laborers , not from a representative number of the businesses that employ them. As such, establishing ESD's estimate of the grower population was flawed would not automatically invalidate the resulting PWR. Zirkle nevertheless contends the choice of a capture-recapture model is relevant in two ways.
First it contends ESD's shifting estimates of the grower population over time is circumstantial evidence that its calculations—and by extension the model used to produce those calculations—are unreliable. Zirkle compares ESD's 2015 estimate (111 growers and 12,774 laborers) with ESD's 2018 estimate (214 growers and 5,622 laborers). ECF No. 87-1 at 5. That ESD's estimate of the number of employers "halved" while its estimate of the number of laborers "doubled," Zirkle argues, indicates that ESD's methods were flawed. But as DOL and ESD respond, the capture-recapture model becomes more accurate with each successive "capture occasion," as reflected by ESD's 2017 estimate—which Zirkle omits from its argument—of 182 employers and 5,377 laborers, well within the 2018 estimate's 95% confidence interval. ECF No. 88 at 10. This conclusion is buttressed by the declaration of Zirkle's own witness Alan Schreiber, Executive Director of the Washington Blueberry Commission. Mr. Schreiber, in support of Zirkle's motion for a temporary restraining order, testified that "[i]t takes about 5,000 – 6,000 pickers to harvest Washington's fresh blueberry crop." ECF No. 8 at 5. Thus, if anything, the disparity between ESD's 2015 and 2019 estimates of Washington's grower population gives the Court greater confidence in the reliability of the capture-recapture model's result in the 2019 PWR.
Second, Zirkle contends the choice of a capture-recapture model rendered ESD's findings unreliable because a fundamental assumption underlying the model—that all members of the relevant population are equally likely to respond—is incompatible with the raking algorithm used to estimate what percentage of the labor market was reflected in the survey responses, which assumes variation in the likelihood members of the population will respond. Defendants concede the assumptions underlying the specific capture-recapture model and the raking methodology are "conceptually" inconsistent. See ECF No. 88 at 21. They also concede there were capture-recapture models more conceptually aligned with the raking algorithm. Id. But they maintain Moll, "[i]n the exercise of his training, experience, and professional judgment as a research economist consciously elected the "conceptually imperfect" pairing because the capture-recapture models better suited to use with the raking algorithm were "extremely poor fits for the blueberry dataset." Id. ; ECF No. 91-1 at 135–36.
Zirkle appears to contend that the conceptual incongruity between the capture-recapture model and the raking algorithm standing alone renders the 2019 PWR arbitrary and capricious. But Dr. Bronars does not appear to contend that Moll should have relied on a capture-recapture model that was ill-suited to the blueberry wage dataset simply to ensure greater conceptual agreement. See generally ECF No. 84-5. The Court's deference to ESD and DOL's decisions is at its apex when reviewing "technical analyses and judgments involving the evaluation of complex scientific data within the agency's technical expertise." Allen , 615 F.3d at 1130. Mindful of the principle that just because "a model is limited or imperfect is not, in itself, a reason to remand agency decisions based upon it," the Court declines to find this a sufficient basis to invalidate the 2019 blueberry PWR. Appalachian Power Co. , 249 F.3d at 1052.
2. ESD's application of the raking algorithm was not arbitrary
Zirkle also contends that ESD arbitrarily divided Washington's growers into "small," "medium," and "large" categories and assigned arbitrarily weights to survey responses. ECF No. 87-1 at 12–13. Specifically, Zirkle ascribes error to ESD's choice to assign a "six-fold multiplier" to the single responding "large" grower—an employer that, had it employed two fewer laborers, would not have qualified as "large," reducing ESD's estimated sample size below the required 15% threshold. Id. Had Moll defined employer categories or chosen sample weights at random, Zirkle could likely establish that the PWR was premised on an arbitrary methodology.
But the record does not support that conclusion. The criteria by which ESD assigned responding employers to categories based on their size was anything but arbitrary—rather, those criteria derived from a statistical model known as a K-means test that relied on data from the state's unemployment insurance database. ECF No. 91-1 at 33. Insofar as Zirkle contends ESD erred in basing this conclusion on data regarding all berry growers, and not blueberry growers specifically, this argument is fruitless—as ESD points out, there is no dataset distinguishing growers by the type of berry they harvest and at most this renders the dataset imperfect. See Dist. Hosp. Partners, L.P. , 786 F.3d at 61.
Nor were the sample weights ESD applied to responding employers chosen arbitrarily. Rather, as Zirkle and Dr. Bronars recognize, the sample weights were a product of the raking algorithm, which as Moll testified "optimizes" the sampling weights applied to employer responses. Id. at 14–16. Indeed, Handbook 385 itself expressly authorizes SWAs to apply sampling weights to employer responses. See AR 25, 31& 35.
To be sure, ESD made difficult decisions in determining how best to calculate the PWR based on the limited data it received from Washington's growers. Because far fewer than all Washington's blueberry growers responded to the survey, ESD was in the unenviable position of determining what percentage of Washington's overall grower community was represented in those responses. To that difficult task the record establishes ESD applied reliable—if limited and imperfect—statistical models. Those limitations are not a sufficient basis to invalidate DOL's action. See Appalachian Power Co. , 249 F.3d at 1052 ("That a model is limited or imperfect is not, in itself, a reason to remand agency decisions based upon it."). This Court must "defer to an agency's decision to proceed on the basis of imperfect scientific information," and Zirkle has failed to show that ESD made sufficiently egregious statistical or mathematical errors to warrant setting that decision aside. Allied Local and Reg'l Mfrs. Caucus , 215 F.3d at 71.
Notably, Zirkle—purportedly the state's largest blueberry grower—declined to participate in the voluntary survey, foregoing the opportunity to dramatically increase the dataset on which ESD's findings were made and—if Zirkle in fact pays less than $0.75/lb. to domestic laborers—potentially reduce the PWR.
--------
CONCLUSION
Agency action is not arbitrary or capricious simply because it is imperfect. Nor are agencies required to delay or forego their delegated duties simply because they lack a perfect dataset from which to undertake them. In this case, Congress delegated to DOL—and, by extension, ESD—the important task of calculating the PWR to ward off serious damage to the domestic labor market that could result from an influx of foreign labor paid below-market wages. ESD applied its limited funding to the imperfect data it was capable of gathering and relied on sophisticated, if imperfect, statistical models to calculate the PWR. DOL evaluated those findings and assured itself they complied with the relevant standards. At most, Zirkle has shown this process was limited and produced an imperfect result. Under the APA, that is not enough. Zirkle's motion for summary judgment is denied. The DOL Defendants' motion for partial summary judgment is granted.
Accordingly, IT IS HEREBY ORDERED :
1. Plaintiff's Motion for Declaratory Judgment and Injunction, ECF No. 87-1 , is DENIED .
2. The DOL Defendants' Response to Plaintiff's motion and Cross-Motion for Partial Summary Judgment, ECF No. 88 , is GRANTED .
3. Within thirty days of the entry of this Order, Zirkle shall remit payment of the wages held in escrow pursuant to the Court's Order, ECF No. 49 at 24–25 , to those laborers employed during the 2019 blueberry harvest from whom the wages were withheld after the Court entered the preliminary injunction in this matter.
A. Zirkle shall withhold all customary taxes and other deductions from these wages and shall remit them as required by law as if the
wages were paid at the time they were earned.
B. Zirkle shall employ all commercially reasonable efforts to locate and remit payment of the wages to those laborers now entitled to them. Within seven days after the expiration of the thirty-day period set out in item 3. above, Zirkle shall file a status report updating the Court on the status of its efforts to remit the wages. Zirkle shall explain (1) whether it has been unable to locate any laborer entitled to payment, and (2) if so, what efforts it has made, or is continuing to make, to locate each laborer.
C. If there are laborers whom Zirkle has been unable to locate to remit payment, the Court will evaluate Zirkle's efforts and proceed accordingly.
IT IS SO ORDERED. The Clerk's Office is directed to enter this Order and provide copies to all counsel.