A screen is a simple test that identifies likely candidates for a broader and more complex economic analysis. The use of screens in economic analysis typically involves comparing an observation of a variable to a reference value, threshold or benchmark. For example, a measure of the dispersion of a group of prices may be compared to its historical level. Screens are often formalized into statistical tests.
Screens are used in a variety of different contexts. For example, current prices may be compared to levels observed during periods known to be competitive. If that comparison indicates prices are abnormally high, that may trigger a price-fixing investigation. As another example, financial data may be examined to see if they follow an expected pattern or distribution. If certain financial data deviate from the pattern they have followed historically, that anomaly may trigger further investigation. Suppose the data commonly follow Benford’s Law, an empirical regularity describing the behavior of the leading digits of the numbers in a data set, but do not follow that pattern in a given period. Other evidence may then be examined to determine whether the anomaly reflects market manipulation or other misconduct.
Competition agencies often use screens to help identify market situations that raise antitrust concerns and justify further scrutiny. Government agencies have limited resources and the use of screening tools contributes to efficient antitrust enforcement. Screens may help detect past anticompetitive behavior. For example, certain percentage price increases in an industry or a lack of price variability can be interpreted as possible manifestations of collusion. Screens also may indicate the likelihood that anticompetitive conduct will develop in the future. For example, since 1982, the U.S. Department of Justice, the Federal Trade Commission, and the state Attorneys General use the post-merger Herfindahl-Hirschman Index (“HHI”) to assess the likelihood of anticompetitive effects from a merger. This screen is based on interpreting market concentration as a signal of a merger’s likely competitive effects. More recently, and for the same purpose, the Upward Pricing Pressure (“UPP”) test based on diversion ratios, gross profit margins and efficiencies has been proposed to predict the direction of post-merger price changes in prospective mergers involving differentiated products.
The use of screens in economic analysis is justified to the extent that a single structural index or statistic offers a good indicator or signal of a particular economic condition. The strength of a screen, however, depends on the strength of the theory from which it is derived, as well as the proper empirical application of that theory. For example, the strength of a threshold based on a market concentration index to analyze market power depends on the link between market structure and economic conduct and on whether the market is properly defined.
Designing and selecting a screen typically presents a tradeoff between the efficiency and simplicity that justify the use of the screen in the first place and its effectiveness. Screening tools that are data-intensive or difficult to implement can be better applied as part of a full investigation. For example, the data requirements of the UPP test may be hard to satisfy at the initial stages of a merger inquiry. Conversely, screening devices that oversimplify the analysis can be easy to implement but fail to serve their purpose. An effective screen will minimize two types of errors: not finding anticompetitive conduct when it is there (false negatives) and detecting anticompetitive conduct when it is not there (false positives).
Well-designed screens can be misused if practitioners fail to recognize their limitations. Screens are by design not dispositive pieces of evidence but indicators that are calculated with limited resources. In some instances, a test may be insufficient because the range of competition concerns extends beyond the information conveyed by the screen. In other instances, a test may flag cases where a necessary condition for conduct of concern is met, but that condition is not sufficient to conclude that such conduct exists. In many cases, overcoming these limitations calls for a more extensive economic inquiry.
For example, recent litigation concerning manipulation of the Libor (London Interbank Offer Rate)—a reference interest rate derived from individual banks’ reported borrowing costs—alleges that concerted action by banks to under-report their borrowing costs led to an abnormally low Libor. In this case, a variety of indicators can be designed to flag anomalies or departures from historical patterns. The strength of these tests will depend on how well they can detect instances of collusive behavior. For example, detection of collusion may be guided by a test designed to flag periods when borrowing costs among banks appear abnormally uniform, under the assumption that such lack of variability is inconsistent with presumed differences across risk profiles. Such uniformity, however, may result from other determinants of individual banks’ borrowing costs. Thus the uniformity is not a sufficient basis to conclude that collusion occurred. Banks may have similar expectations about future monetary policy during periods of stability, and this similarity is likely to contribute to homogeneous borrowing costs. The breakdown of a customary price pattern that resulted from legal economic conduct does not necessarily imply that the anomalous price pattern resulted from illicit conduct.
In sum, screens can be useful in a wide variety of economic applications, including applications in antitrust and finance. Overreaching interpretations of the scope and strength of an economic screen, however, may lead to erroneous conclusions. As a result, screens are best utilized when considered together with other pieces of economic information and analysis.