UNITED STATES DISTRICT COURT SOUTHERN DISTRICT OF NEW YORK
November 21, 2008
THE CITY OF NEW YORK, PLAINTIFF,
GROUP HEALTH, INC., ET AL., DEFENDANTS.
The opinion of the court was delivered by: Ronald L. Ellis, United States Magistrate Judge
OPINION AND ORDER
On April 20, 2007, this Court issued an Opinion and Order denying, without prejudice, the City's motion to compel Defendant HIP to produce certain documents responsive to the City's document requests, namely cost and experience documents relating to its accounts, with this information broken down by employer/customer (hereinafter "HIP data" or "data"). (Opinion and Order, April 20, 2007 ("Op. & Order") at 4; see Plaintiff's Memorandum of Law in Support of Motion to Compel Under Federal Rule of Civil Procedure 37(a), October 25, 2007 ("Pl.'s Mem., Oct. 2007") at 1.) Pending before this Court is the City's renewed motion for an order compelling HIP to produce this data. Id. For the reasons set forth below, the City's motion to compel production of these documents is GRANTED.
A. April 20, 2007, Opinion & Order
In its previous motion, the City indicated that the requested information was needed so that it could perform an analysis to determine "whether, and if so by how much, HIP will be able to raise the premium for its HMO product to the City following the [proposed] merger [of GHI and HIP]."*fn1 (Plaintiff's Memorandum of Law in Support of Motion to Compel Under Federal Rule of Civil Procedure 37(a), February 13, 2007 ("Pl.'s Mem., Feb. 2007") at 6-7.) The Court found that, because the information requested would be insufficient standing alone to calculate a post-merger cost analysis, the City's request did not adequately relate to its stated justification for obtaining such information. (Op. & Order at 5-6.) Based on the record presented at the time, the Court concluded that only upon obtaining a substantial amount of similar data from third-party health insurance consumers located within the same market-base would the City be able to perform a comparative analysis with any probative value. Id. at 5. At that time, the City did not have, and had no plan to obtain, the supplemental information from third parties the Court found necessary to this analysis. Id. The Court noted that if the City could articulate a plan for obtaining third-party information and specify which "commercial entities would be necessary for a calculation of the merged entity's profitability of raising prices to have any predictive validity, and that it has the ability to obtain such information from the identified third parties[,]" the Court would allow for reconsideration. Id. at 6.
B. City's Actions in Response to Court's April 20, 2007, Opinion and Order
Subsequent to this Court's April 20, 2007, Opinion and Order, further dispute arose between the parties regarding the City's Second and Third Requests for Documents and a Second Interrogatory Request. (Declaration of John R. Low-Beer, October 25, 2007 ("Low-Beer Decl., Oct. 2007") at 5.) On August 28, 2007, the City filed a second motion to compel "the production of documents and responses to interrogatories, and for reconsideration" of the Court's April 20, 2007, Opinion and Order. The information requested by the City at that time was distinct from that requested in the initial motion to compel. Id. at 4. Subsequently, the parties engaged in negotiations which resulted in substantial accommodation on nearly all issues pertaining to the Second and Third Document Requests, and the City agreed to withdraw its August 28, 2007, Motion to Compel without prejudice. Id. at 5.
On October 25, 2007, the City filed the instant motion, asking the Court to reconsider the April 20, 2007, Opinion and Order, and to compel production of the documents requested in their initial discovery request. (Pl.'s Mem., Oct. 2007 at 1.) In support of this motion, the City reports "the 'record before the Court' has evolved significantly from that before the Court in the Spring." (Low-Beer Decl., Oct. 2007, ¶ 16.) The City asserts it has a "plan to obtain information from third parties . . . and has served subpoenas on a number of large employers and other insurers."
Id. ¶ 3. The City additionally presents different economic models which they argue minimize, if not entirely eliminate, the need for third-party data to predict the likelihood and profitability of HIP raising its premiums post merger. (See Pl.'s Reply at 1.)
The ultimate issue in this antitrust action is whether the merger of GHI and HIP will substantially lessen competition when the entities cease to compete with each other, and as a result, whether HIP will have the ability to profitably raise its prices. (SeeDefendants' Memorandum of Law in Opposition to Motion to Compel the Production of Cost and Experience Documents, March 6, 2007 ("Defs.' Mem., Mar. 2007") at 2; see also Pl.'s Mem., Feb. 2007 at 7.) The City argues that customer-specific cost information is relevant because it is necessary to calculate the relative profitability of the City's business compared to HIP's other large-group customers, in order to calculate the net profitability to HIP of a given increase in its premiums after the merger. (Pl.'s Mem., Oct. 2007 at 6.) In response to the City's first motion on this issue, the Court concluded that, with adequate data from third parties, the documents at issue here could be applied to a comparative analysis, the results of which would present a prediction of the effect of the merger on insurance premiums. (Op. & Order at 5-6.) The Court determined that under such conditions, the controversial HIP data could lead to information relevant to the instant litigation. The question before the Court here is whether the City now has adequate information to engage in an analysis of the profitability to HIP of raising its premiums post-merger.
In support of its contention that such analysis would be possible with the various data available to them and the HIP data here at issue, the City has submitted declarations by Frank R. Lichtenberg, an expert in the economics of health care. (See generally Declaration of Frank R. Lichtenberg, August 28, 2007 ("Lichtenberg Decl., Aug. 2007") & Declaration of Frank R. Lichtenberg, November 16, 2007 (Lichtenberg Decl., Nov. 2007").) In response, Defendants have provided the declaration of their own expert, Lawrence Wu, who specializes in antitrust and competition policy, with expertise in the area of health care, including health insurance. (See generally Declaration of Lawrence Wu, November 9, 2007 ("Wu Decl., Nov. 2007").) Defendants challenge the predictive value of the data available to the City and assert that the information the City has access to is inadequate to perform the necessary analysis using any of the proposed models the City plans to use, namely the critical loss analysis*fn2 and merger simulation models.*fn3
The City's expert avers that the data from HIP is "highly relevant" to an antitrust inquiry, (Lichtenberg Decl., Aug. 2007, ¶ 5), and the City argues that HIP put raising their premiums at issue by repeatedly asserting that they would lose too many commercial customers if they were to raise their prices. (Low-Beer Decl., Oct. 2007, ¶ 20.) Defendants argue that the City has failed to comply with the parameters established by the Court in the April 2007, Opinion and Order. (Defs.' Mem., Nov. 2007 at 2.) Defendants note that they were provided with eighteen CDs of documents that the City compiled in response to various subpoenas or from insurers in response to a recent bid solicitation, and they argue these documents do not "provide the data necessary to perform with any predictive validity either the critical loss analysis that Professor Lichtenberg originally stated that he intended to perform, or the merger simulation that he now claims would have more relaxed data requirements." Id. at 5. The City replies and requests that "[t]his Court, looking on the record now before it, should draw its conclusions from that record, while also bearing in mind that the present record remains fluid and will be supplemented with additional materials." (Pl.'s Reply, Nov. 2007 at 5.)
This Court previously determined that the information from HIP may be relevant to a determination of whether and, if so, by how much HIP will be able to raise its premiums if coupled with adequate third-party data for use in a comparative analysis. The Court further cautioned that mere "random samples of other customers may not yield a statistically appropriate example." (Op. & Order at 6.) The Court now must determine whether the City complied with, and made the showing required by, the April 2007 Opinion and Order, and if the Court should reconsider compelling production of the HIP data.
1. Information Acquired by City
As of November 2007, the City had succeeded in obtaining employer-specific pricing and enrollment information from a total of four of the fourteen employers and all four insurers that it had subpoenaed. (Pl.'s Mem., Oct. 2007 at 3.) Through discovery requests and interrogatories the City now has information regarding HIP and GHI's "enrollments and premium rates over time broken down by their largest customers" as well as "information about plans offered by other health insurers" in the New York region. (Lichtenberg Decl., Nov. 2007, ¶¶ 16-17.) As supplemental sources of information, the City notes that it has plans to utilize data received from certain public sources, including the New York State Department of Insurance ("NYDOI") (Pl.'s Mem., Oct. 2007 at 8), the New York State Department of Civil Service (Declaration of June R. Buch, November 16, 2007 ("Buch Decl., Nov. 2007"), ¶ 4), and two open-source public databases (Lichtenberg Decl., Nov. 2007, ¶¶ 14-15). Lichtenberg indicates he has also reviewed additional information received by the City, including "enrollment and premium data from Oxford/United Health Care on its 20 largest (ranked by number of subscriber months) New York Groups over the last three years . . . and data obtained from Con Edison on enrollment, premiums, and characteristics of all of its health plans for the period of 2002-2007." Id. ¶ 13. Finally, the City represents that it is still receiving discovery and data relevant to their proposed comparative analysis. (Buch Decl., Nov. 2007 at 2.)
2. Adequacy of Information Now Available to City
In his original declaration, accompanying the City's First Motion to Compel, Lichtenberg indicated the relevance of the HIP data, concluding that financial data by individual employees is necessary to accurately predict the effect of the merger on insurance premiums. (See Op. & Order at 4.) By declaration submitted with the City's withdrawn Second Motion to Compel Production of these documents, Lichtenberg asserted that, in light of the other data the City then had and was in the process of receiving from GHI and HIP, the requested HIP data alone would be adequate to enable the City to calculate a lower-bound estimate of the amount by which HIP could profitably raise its prices after the merger. (See Lichtenberg Decl., Aug. 2007, ¶ 7.) Lichtenberg references two types of models, critical loss analysis and Bertrand models, which he opines are less demanding with respect to the input information required. Id. He asserts that such models are commonly used for estimating the effects of a merger on product cost, and require only pre-merger cost, pricing and other data from the merging parties. See id.; (Pl.'s Mem., Oct. 2007 at 8.) Lichtenberg elaborates, stating that "[d]ata from GHI and HIP, together with more limited information on the general level of pricing and sales of other insurers, are sufficient to predict price increases post-merger. Additional data, including data that plaintiff has sought through subpoenas and data available from public sources, will make the economic models more accurate and more powerful." (Lichtenberg Decl., Aug. 2007, ¶ 13.)
In responding, Wu argues that both critical loss analysis and Bertrand models require virtually the same data. (Defs.' Mem., Nov. 2007 at 3; Wu Decl., Nov. 2007, ¶¶ 4, 33, 35-38.) Defendants assert that to perform either analysis properly it is necessary to possess nearly all of the same data, including estimates of the own-price elasticity of demand*fn4 and the cross-price elasticity of demand.*fn5 Furthermore, in the context of health insurance markets there are two components to each calculation: 1) the employer level and 2) the employee level. (Defs.' Mem., Nov. 2007 at 2-3; Wu Decl., Nov. 2007, ¶ 10.) Wu proposes that to "accurately and reliably estimate the total enrollment loss[,]" reliably estimate the demand elasticities, or determine market definition using critical loss analysis more systematic data collection would be necessary. (Wu Decl., Nov. 2007, ¶¶ 6, 31.) Defendants also argue that specific information not obtained by the City is necessary for calculating both levels of demand elasticity and this data must come from many employers, span many years, and include many other data factors. (Defs.' Mem., Nov. 2007 at 7; Wu Decl., Nov. 2007, ¶¶ 25-28.) Nevertheless, Lichtenberg maintains that the data from third parties is adequate to estimate elasticities of demand and other necessary components to the equations for which the City moves to compel production of HIP's data. In his reply affidavit, he argues that Wu's comments address only estimated-demand models and not calibrated-demand models,*fn6 and that the latter require considerably less data. He notes, moreover, that some estimated-demand models require very few parameters. (Lichtenberg Decl., Nov. 2007, ¶ 5.) Concomitantly, the City outlines its plan for collection of third-party information and notes that the process is ongoing. (See Pl.'s Mem., Oct. 2007 at 3-4.)
Defendants and Wu also challenge the City's plan, questioning whether the City's data from four insurers and fourteen New York City employers would supply sufficient third-party information for analysis, and further critiquing the City's limited success with receiving data from even those parties. (Defs.' Mem., Nov. 2007 at 6-7; see generally Wu Decl., Nov. 2007, ¶¶ 5-9, 19-31.) Wu recognizes that the data produced to date has some value, but argues that it is inadequate to "make the necessary and complex estimations required for a demand elasticity study." (Wu Decl., Nov. 2007, ¶ 9.) Wu challenges Lichtenberg's proposed models of analysis by indicating that Lichtenberg makes many assumptions, which must first be tested for their factual validity. Id. ¶ 14. Wu further argues the public information available from NYDOI is not sufficiently comprehensive to provide the information necessary to estimate demand elasticities. Id. ¶ 32. Wu concludes that Lichtenberg does not have the data necessary to reliably predict the impact of the merger. Id. ¶ 33.
In response to Defendants' critique of the City's plan to obtain data and the adequacy of what it had received, the City indicates the process is ongoing and that, even in the week between Defendants' Response and the City's Reply, significant additional data collection had taken place, including, inter alia, data from GHI and HIP, data from private and public third-party employers, and data from additional third-party health insurers. (Buch Decl., Nov. 2007, ¶¶ 3-5.) Lichtenberg concludes that "[a]s the amount of relevant data available increases, so too will the power and accuracy of the models discussed here to predict the effect of the merger and the defendants' ability to raise prices." (Lichtenberg Decl., Nov. 2007, ¶ 18.)
Lichtenberg also indicates that some of the figures Wu alleges must be determined for use in the merger simulation models have already been estimated, such as the diversion ratio, which Wu argues is developed from demand-elasticity figures. Id. ¶ 6; (see Wu Decl., Nov. 2007, ¶ 37.) Further, he posits that "[w]hen the data required to implement an estimated-demand simulation model are not available, a calibrated-demand simulation model, which is based on additional assumptions about the nature of demand, can be used." Id. ¶ 7. Such models, he identifies, include the logit demand model and the Proportionality-Calibrated Almost Ideal Demand System ("PCAIDS"). Id.; see also id. ¶ 9. Lichtenberg explains that PCAIDS "requires only information on market shares and reasonable estimates of two elasticities. Estimates of these elasticities often can be obtained from marketing information, or when appropriate, through demand estimation." Id. ¶ 9. Finally, Lichtenberg asserts that "[t]he models can be used to review the potential antitrust exposure resulting when unilateral effects issues are raised but sufficient information is not available to estimate reliably a full set of cross-price elasticities." Id. As this information was provided in the City's Reply Declaration, Wu did not specifically address the models identified therein by Lichtenberg. From a review of PCAIDS, as presented by Epstein & Rubinfeld,*fn7 it appears there may be sufficient data available, with the production of the HIP data here at issue, to utilize this model. The question remains whether the resulting calculations have sufficient predictive value.
In presenting this motion to compel, both the City and Defendants rely heavily, if not exclusively, on the declarations of their experts to prove or dispute the relevancy and utility of the HIP data at issue. They frame the relevancy discussion in such a way that leaves the resolution requiring a determination of the weight to be given to the respective experts. This is a matter best left for the trier of fact.*fn8 For discovery purposes, the basic question is whether the requested data "appear reasonably calculated to lead to the discovery of admissible evidence." FED. R. CIV. P. 26(b)(1). Essentially, Defendants' expert challenges whether the data would allow for an ideal analysis, where the resulting calculation would be highly predictive. (See Defs.' Mem., Nov. 2007 at 6 (challenging whether "meaningful analysis of customer reaction to any hypothetical post-merger price increase" can occur with the data available to the City at that time) (emphasis added).) While the City's expert makes no grand assertions about the predictive value of the analysis based on the available information, he demonstrates that an analysis may be done with the third-party data, the GHI & HIP documents already provided to the City, the publicly available data, and the HIP data here at issue.
Based on the record now before the Court, the data will lead to information relevant to the instant antitrust litigation.*fn9
B. Balance of Interests for Production of HIP Data
A final determination the Court may consider is whether a balance of interests would favor precluding the City from discovering the HIP documents. While Defendants do not focus on the adverse impacts disclosure of the HIP data would have, and do not raise such as grounds for protecting discovery of this data in the instant motion,*fn10 in the April 2007, Opinion & Order, the Court acknowledged that HIP's data is sensitive, and indicated it may "consider the burden the production would have on the business integrity of the party from whom the information is sought." (Op. & Order at 6.)
Weighing the circumstances and evidence presented here, the Court finds: 1) it would not be unduly burdensome for HIP to produce the data, as the data has already been compiled; 2) the detailed confidentiality order in place in this litigation protects against concerns about supplying any commercially sensitive information or interfering with Defendants' business integrity; and 3) any concerns about the City learning information that may have an adverse impact on their own retention of HIP as an insurance-provider ought to be similarly assuaged by the confidentiality order in place. Therefore, the balance of interests favors production of HIP's data to the City, as the City has made a substantial showing that the data is relevant.
For the foregoing reasons, the City's motion to compel is GRANTED.