Data and Discovery Issues

• Pages: 61-94

61

CHAPTER III

DATA AND DISCOVERY ISSUES

A. Introduction

Data are the critical inputs in econometric analysis. Even the most

scientific statistical analysis cannot glean reliable inferences from bad

data, and without specific types of data, certain econometric methods

may be unavailable. Data collection, meanwhile, is a time-consuming

and expensive project, whether involving the expenditure of thousands of

dollars to purchase commercial databases from third parties, the

numerous client hours required to obtain data from internal sources, or

attempts to use compulsory process to obtain data from adversaries or

third parties.

Although data issues are frequently dry and highly complex, it is

important that lawyers understand them, along with the advantages and

disadvantages of particular types and sources of data. Not only will that

allow them to counsel their clients as to the likely utility of particular

data collection efforts, but it will allow the lawyers—who are frequently

the primary interface between the econometricians and the data source,

whether a commercial provider or an internal client resource—to

communicate effectively and prioritize the experts’ data requests.

Informed lawyers should also be better able to help experts understand

what types of data are available from the client, how they are kept, and

where there are problems in the data set that could raise questions as to

the reliability of ensuing analyses.

Of course, lawyers are also intimately involved in the often

voluminous exchange of data and other econometric-related information

during discovery. The effective use of pretrial discovery is crucial to the

successful use of and defense against econometric evidence at trial.

Beginning with the mandatory disclosures required of all parties, and

continuing with document demands and interrogatories as necessary,

discovery should be tailored to obtain the data and information on

models and methods necessary to allow one’s own econometric expert to

Econometrics in Antitrust

62

replicate the opposing expert’s substantive results. The documents

obtained through discovery should also form the foundation for

depositions of the opposing party’s economic experts, aimed at revealing

potential data problems, violations of statistical assumptions, and prior

inconsistent results, as well as the expert’s qualifications (or lack thereof)

as an econometrician. That information can be fertile ground not only

for arguments at a Daubert hearing, but also for cross-examination and

rebuttal at trial. Conversely, the lawyer for a party receiving a request

for data should be aware that the data often will be used by the

government or opposing party in an econometric analysis and must be

mindful of the limitations of the data when they are produced.

This chapter discusses the twin issues of data and discovery. It will

not obviate the need for close consultation with econometric experts, and

is not intended to provide a comprehensive guide to the law of pretrial

discovery or evidence, or to serve as a substitute for the Antitrust

Section’s handbooks on those subjects,1 or replace the many treatises on

federal practice. It aims, rather, to highlight some of the issues most

relevant to lawyers dealing with econometric evidence.

The chapter opens with brief descriptions of some of the varied types

of data commonly used in econometric analysis, followed by a

discussion of four common data problems: multicollinearity, improper

aggregation, missing observations, and influential observations. It then

discusses the disclosures required of expert econometricians, potential

sources of data for econometric analysis (including special issues related

to discovery from non-parties and the government), and the

consequences of inadequate compliance with discovery obligations.

Finally, the chapter closes with some sample deposition questions that,

when appropriately modified, can serve as the starting point for the

effective cross-examination of opposing econometric experts at trial.

B. Data Types

1. Time-Series Price and Quantity Data

Perhaps the most useful types of data for econometric methods are

data that measure price and quantity over time, preferably including data

1. ABA SECTION OF ANTITRUST LAW, ANTITRUST DISCOVERY HANDBOOK

(2d ed. 2003); ABA SECTION OF ANTITRUST LAW, ANTITRUST EVIDENCE

HANDBOOK (2d ed. 2002).

Data and Discovery Issues

63

from all competitors in the relevant market. These sorts of data are best

for measuring cross-elasticities between competing products, and

therefore provide a critical input for the merger simulation methods

described in Chapter XI.

There are significant problems in the collection of time-series price

and quantity data. First, the data frequently are in the possession of

competitors. This is a significant problem in the merger context where

the merging parties lack pre-litigation subpoena power. This problem

continues in litigation because courts frequently are unwilling to allow a

firm to obtain competitively sensitive price and quantity information

from a competitor.2

Because time-series price-quantity data is difficult to obtain,

econometricians frequently are forced to use merger simulations that

estimate price effects from market share or cost data in the parties’

possession. To the extent that these methods may rely upon assumptions

that may not prove to be accurate, they may not estimate cross-

elasticities and merger effects as accurately as merger simulation based

upon time-series price-quantity data. These issues are discussed in

greater detail in Chapter XI.

The government has a significant advantage over private parties

because it can obtain price-quantity time-series data from competitors

through the use of pre-litigation subpoenas and civil investigative

demands. Because data obtained in such manner are confidential and

exempt from discovery under the Freedom of Information Act, there are

significant limitations on the government’s ability to share data with the

merging parties during the merger review process.3

2. Commercial Databases

Commercial entities accumulate and sell time-series price and

quantity databases, frequently purchasing bills from consumers and then

accumulating these data in the form of a database. Because the data were

collected by a person not under the party’s control, and frequently not

subject to cross-examination, there may be substantial questions about

the reliability of these data.

When purchasing databases from commercial vendors, whether bill-

harvesting databases or other types, it is extremely important to

2. See part E(1) of this chapter.

3. See part E(2) of this chapter.