3. SOME EMPIRICAL RESULTS CONCERNING PREDICTABILITY

Let us now quickly turn to some empirical studies that can be interpreted as evidence for the rucial importance tance of predictability. As an example for the importance that should be attached to the issue of predictability with regard to the general functioning of market economies, we cite a cross-country study by Brunetti et al. (1997). Empirical research concerning the effects of merger policy is very rare. Most of the existing studies deal with the effects of US antitrust policy. We will refer to two of these studies, Shughart and Tollison (1991) and Bittlingmayer (2001). Empirical evidence concerning the economic consequences of European merger policy is virtually non-existing and definitely a desideratum.

Over the last number of years, an entire cottage industry dealing with the economic effects of formal institutions as, e.g., constitutions and legislation has developed. Rather than providing an overview of that literature, we confine ourselves to the presentation of one such study here (a short survey can be found in chapter five of Voigt 2002). It has been selected because it does explicitly exclude the material content of rules and concentrates exclusively on the perceived credibility of rules.

In a study commissioned by the World Bank, Brunetti, Kisunko, and Weder (1997) use what could be called “subjective uncertainty indicators”. They asked local entrepreneurs in some five dozen countries whether they believed the announcements of their governments to be credible. If the locals do not believe their government, then a government announcing secure private property rights, low tax rates, etc., will not induce investment. There is thus a very close relationship between credibility and predictability. Notice that neither concept focuses on the content of the rules but just on the formal issue of whether some policy (in the form of legislation or the like) announcement is believed to be implemented. Brunetti et al. (1997) surveyed more than 2,500 entrepreneurs in less developed countries and approximately 200 entrepreneurs in OECD member states. The partial indicators “security of persons and property rights” and “predictability of legislation” are most closely correlated with economic growth in this sample of 58 countries. The partial indicators “corruption”, “perceived political instability” and “predictability of judicial decisions” closely correlate with investment rates.

Let us now turn to the empirical studies interested in the economic effects of antitrust policy. Shughart and Tollison (1991) are interested in ascertaining the employment effects of the Sherman and Clayton Acts, i.e., those two laws that are commonly believed to be the core of US antitrust policy. They conjecture that unanticipated strengthening of antitrust enforcement, i.e., low predictability of the actions of antitrust agencies, can force firms to adopt inefficient production technologies. As firms are forced to lower their current and planned production quantities, they might have to lay off some of their employees as a consequence. Suppose enforcement of antitrust legislation is unexpectedly strengthened in a particular industry. Mergers might be prevented and divestitures ordered. Suppose this is accompanied by unchanged production technology, which would mean that from a production oriented point-of-view, the optimal size of firms in the industry has remained unchanged. Yet, the optimal size of firms in that specific industry is reduced due to increased enforcement effort. Shughart and Tollison argue that this will lead firms to lower their current and planned production as well as to lay off some of their employees. If other industries with similar structures observe this, unemployment may also increase there. For the period between 1947 and 1981 (the starting point was chosen because 1947 is the first year for which extensive data concerning the U.S. labour market are available), they find a positive and significant relationship between unanticipated increases in antitrust enforcement activities and the unemployment rate. This study can thus be read as empirical evidence supporting the claim that the absence of predictability – here called unanticipated increases in antitrust enforcement activities – can have real economic costs, namely an unemployment rate that is higher than if merger policy was predictable.

Bittlingmayer (2001) is interested in determinants of investment levels. In empirical studies seeking to identify the factors determining investment, only economic variables such as interest rates or the prices of capital goods have traditionally been taken into consideration. Bittlingmayer now argues that these regressions always leave a fair amount of variation in investment unexplained and suspects that political uncertainty could be one such factor. He is then in need of a proxy for political uncertainty and chooses antitrust enforcement as that proxy. Based on data for the period between 1947 and 1991, he comes to the conclusion (ibid., 321) that “statistically, each extra case [of antitrust enforcement] is thus associated with a total decline in investment of about $1.7 billion.” It is worth noting that the major adverse effects on investment do not arise within the industry of the firms against which a suit has been brought. Using an industry classification made up of 21 industries, an extra case filed is associated with a decline in investment in each of these 21 industries, where decreases are between $34 million and $110 million.

It can thus be concluded that unpredictable merger policy can have real costs in terms of increased unemployment and decreased investment levels. We now turn to the question whether European Merger Policy is predictable.