3. Do Entrepreneurs Make Predictable Mistakes?

CHAPTER3

---

Do Entrepreneurs Make Predictable Mistakes? Evidence From Corporate Divestitures^†

With Sandra K. Klein

---

After a brief lull in the early 1990s, the market for corporate control became increasingly active toward the end of the decade. Both 1996 and 1997 set new records for the number of US merger filings, and 1998, 1999, and 2000 brought high-profile “mega-mergers” in financial services, energy, telecommunications, pharmaceuticals, and automobiles. In banking alone, for example, a wave of mergers over the last decade has led to widespread industry restructuring and consolidation. While total industry activity continues to expand, the number of US banks and banking organizations both fell by almost 30 percent between 1988 and 1997 (Berger, Demsetz, and Strahan, 1999).

Like other business practices that do not conform to textbook models of competition, mergers, acquisitions, and financial restructurings have long been viewed with suspicion by some commentators and regulatory authorities. However, the academic literature clearly suggests that corporate restructurings do, on average, create value. Event studies of acquisitions consistently find positive average combined returns to acquirer and target shareholders. As summarized by Jensen (1991, p. 15), “the most careful academic research strongly suggests that takeovers—along with leveraged restructurings prompted by the threat of takeover—have generated large gains for shareholders and for the economy as a whole.” These gains, historically about 8 percent of the combined value of the merging companies, “represent gains to economic efficiency, not redistribution between various parties” (Jensen, 1991, p. 23).¹

At the same time, however, several studies have found a sharp divergence between market participants’ pre-merger expectations about the post-merger performance of merging firms and the firms’ actual performance rates. Ravenscraft and Scherer’s (1987) large-scale study of manufacturing firms, for example, found that while the share prices of merging firms did on average rise with the announcement of the proposed restructuring, post-merger profit rates were unimpressive. Indeed, they find that nearly one-third of all acquisitions during the 1960s and 1970s were eventually divested. Ravenscraft and Scherer conclude that acquisitions, particularly diversifying acquisitions, typically promote managerial “empire building” rather than efficiency. While acknowledging that product and capital markets eventually discipline poorly performing firms, forcing divestitures and other restructurings, Ravenscraft and Scherer (p. 217) argue for tighter government restrictions on mergers, particularly diversifying acquisitions and acquisitions financed by stock: “When the roads are strewn with wrecks, government officials cannot rest content because the tow trucks, ambulances, and hearses are doing a good job removing the remnants and clearing the right-of-way.”²

Implicit in this criticism is the idea that divestitures of previously acquired assets expose past errors, and that these errors should have been foreseen (and perhaps prevented, if regulators had been sufficiently empowered). Certain types of acquisitions, it is claimed, are more likely to be later divested, so managers should generally avoid them. If such acquisitions occur, this is then cited as evidence for widespread agency problems. In this sense, the takeover wave of the 1980s is typically understood as an “undoing” of the earlier, conglomerate merger wave of the 1960s and early 1970s. According to conventional wisdom, the 1980s was a period of respecialization or “refocus,” showing the failures of unrelated diversification. The three decades from 1960 to 1990 thus represent a “round trip of the American corporation” (Shleifer and Vishny, 1991, p. 54).

This view is based partly on evidence from studies of the conglomerate period by Rumelt (1974, 1982), Ravenscraft and Scherer (1987, 1991), Porter (1987), Kaplan and Weisbach (1992), and others who find no evidence that unrelated diversification brought long-term benefits to the firms that diversified.³ Combined with evidence of negative stockmarket returns to diversification during the 1980s (Bhagat, Shleifer, and Vishny, 1990; Lang and Stulz, 1994; Berger and Ofek, 1995; Comment and Jarrell, 1995), many observers conclude that unrelated diversification is per se inefficient, and that the conglomerate era is best understood as an agency phenomenon.

The conventional wisdom on conglomerate sell-offs can be challenged on at least four grounds. First, divestitures of previously acquired assets do not necessarily show that the original acquisitions were failures. Weston (1989) argues that divestitures occur for a variety of reasons, such as changes in corporate strategies and antitrust rules, and not necessarily poor performance. Kaplan and Weisbach (1992) studied 217 large acquisitions completed between 1971 and 1982 and found that while 43.9 percent had been divested by 1989, only about a third of those divestitures were responses to poor post-merger performance.⁴ Thus the mere fact that many acquisitions are later divested does not prove widespread managerial misconduct.

Second, the market for corporate control is already highly regulated, and it is difficult to draw from current and recent experience strong conclusions about how unhampered capital markets would work. For example, Ravenscraft and Scherer (1987, 1991), Porter (1987), and other critics propose particular sequences of inefficient and efficient restructurings: diversifying, empire-building acquisitions in the 1960s and early 1970s, and then efficient divestitures in the 1980s. But why did entrepreneurs make systematic mistakes during the conglomerate period, but not later? Can changes in the legal, political, and regulatory environments account for clusters of errors during specific periods?

Third, even if divestitures are seen as revealing prior mistakes, the failure of a particular acquisition does not necessarily indicate a failure of the acquisition strategy. Certain kinds of acquisitions—for example, acquisitions of firms in knowledge-intensive, high-technology industries—may be inherently riskier than others. If the returns from a successful integration of the target’s activities with the firm’s existing activities are sufficiently high, then the acquisition has positive expected value, even if it is more likely to fail than a safer acquisition. Matsusaka (2001) offers this kind of interpretation of corporate diversification. Diversifying acquisitions represent experiments, as firms try various combinations of businesses, seeking those that match their capabilities (in the sense of Penrose, 1959; Nelson and Winter, 1982; and Wernerfelt, 1984). After learning their capabilities, firms divest acquisitions that turn out to be poor matches. In this sense, divestitures reflect successful experiments—the acquirer has learned that the target’s industry is not a good match for its capabilities. Such “match-seeking” firms will actively acquire and divest over time.⁵

This chapter elaborates a fourth, “Austrian” interpretation of corporate divestitures, one that builds on the three just mentioned. Austrian writers view market competition as a dynamic, rivalrous process that unfolds gradually through time—a “discovery procedure,” in Hayek’s (1978) famous phrase. The future holds genuine surprises, not merely a closed set of events whose probabilities are unknown. From this perspective, the long-term success of an acquisition, like any entrepreneurial action, cannot be “predicted.” Entrepreneurs rely on judgment, or what Mises (1949) calls understanding. Understanding is intuitive, subjective, and qualitative, and thus inherently imperfect. For this reason, divestitures of underperforming subunits may be seen as efficient responses to unforeseen changes in industry and regulatory conditions, or more generally, to poor judgments by profit-seeking entrepreneurs. Ex post viability is not a good indicator of ex ante efficiency.

We begin with the theory of entrepreneurship proposed by Mises (1949), posing it as a challenge to the agency view of divestitures. We then present empirical evidence that the long-term performance of corporate acquisitions cannot, in general, be predicted by measures of agency conflicts. Instead, divestitures of previously acquired assets are more likely when firms are experimenting, learning, and otherwise trying to deal with uncertainty about future conditions. We also show that mistaken acquisitions are more likely under certain circumstances, namely during periods of intense, industry-specific regulatory activity. Our own research on restructuring (Klein and Klein, 2008) shows that significantly higher rates of divestiture follow mergers that occur in a cluster of mergers in the same industry. As argued by Mitchell and Mulherin (1996), Andrade, et al. (2001), and Andrade and Stafford (2004), mergers frequently occur in industry clusters, suggesting that mergers are driven in part by industry-specific factors, such as regulatory shocks. When an industry is regulated, deregulated, or re-regulated, economic calculation becomes more difficult, and entrepreneurial activity is hampered. It should not be surprising that poor long-term performance is more likely under those conditions.

This last result is consistent with the view, expressed repeatedly in the Austrian literature, that entrepreneurial error is associated with government intervention in the marketplace—in particular, with government ownership of property and interference with the price system. Mises (1920) famously showed that economic calculation is not possible without private property in all markets, especially markets for factors of production. The Austrian business-cycle literature (Mises, 1912; Hayek, 1935; Garrison, 2000) suggests that entrepreneurial errors are more likely under government-sponsored credit expansion. This chapter makes a related argument: entrepreneurial decisions to make acquisitions that will later be regretted, and divested, are more likely in the wake of government intervention in particular industries.

The remainder of the chapter is organized as follows. The first section reviews the Austrian literature on entrepreneurship, uncertainty, and economic calculation, suggesting that the ex post success of entrepreneurial actions cannot be forecasted based on generally available information. The second section introduces recent theory and evidence on the reasons for mergers and divestitures, contrasting two opposing views of sell-offs: empire-building and experimentation. The third section reviews some empirical evidence on the pre-merger causes of divestiture, challenging the generally accepted, empire-building explanation. The final section concludes.

Entrepreneurship, Profit, and Loss

As discussed in chapter 2 above, Misesian entrepreneurship is a fundamentally dynamic phenomenon, unfolding in calendar time. Entrepreneurial profit or loss is the difference between eventual, uncertain revenues and initial outlays, less the general rate of interest. Of course, Mises’s entrepreneur-promoter is absent from textbook models of competitive general equilibrium in which uncertainty is defined away, replaced by probabilistic risk. In these models, it is possible to anticipate which actions, on average, would be profitable. In a world of “true” (structural, rather than parametric) uncertainty, however, profit opportunities do not exist “out there,” waiting to be realized by anyone willing to take a specified action. Instead, profit opportunities are created through entrepreneurial action. As I read Mises, entrepreneurial skill is not simply luck or “alertness,” the ability to recognize profit opportunities that appear, ex nihilo, to the discoverer. Rather, entrepreneurship is judgment: “Alertness is the mental quality of being on the lookout for something new; judgment is the mental process of assigning relevance to those things we already know” (High, 1982, p. 167). In this context,

promoter-entrepreneurs are those who seek to profit by actively promoting adjustment to change. They are not content to passively adjust their... activities to readily foreseeable changes or changes that have already occurred in their circumstances; rather, they regard change itself as an opportunity to meliorate their own conditions and aggressively attempt to anticipate and exploit it. (Salerno, 1993, p. 123; see also Hülsmann, 1997)

All entrepreneurs, particularly those who operate in financial markets, use economic calculation as their primary decision-making tool.⁶ By economic calculation, Mises means simply the use of present prices and anticipated future prices to compare present costs with expected future benefits. In this way the entrepreneur decides what goods and services should be produced, and what methods of production should be used to produce them. “The business of the entrepreneur is not merely to experiment with new technological methods, but to select from the multitude of technologically feasible methods those which are best fit to supply the public in the cheapest way with the things they are asking for most urgently” (Mises, 1951, p. 110). To make this selection, the entrepreneur must weigh the costs and expected benefits of various courses of action, and for this he needs the cardinal numbers provided by money prices. Monetary calculation, then, requires private property and market prices.

As we saw in chapter 1, Mises’s famous 1920 essay on economic calculation under socialism is not so much about socialism per se; it is an argument about the role of prices for capital goods (Rothbard, 1993, pp. 547–78). Entrepreneurs make decisions about resource allocation based on their expectations about future prices and the information contained in present prices. To make profits, they need information about all prices, not only the prices of consumer goods but the prices of factors of production. Without markets for capital goods, these goods can have no prices, and therefore entrepreneurs cannot make judgments about the relative scarcities of these factors. In short, resources cannot be allocated efficiently. In any environment, then—socialist or not—where a factor of production has no market price, a potential user of that factor will be unable to make rational decisions about its use. Stated this way, Mises’s claim is simply that efficient resource allocation in a market economy requires well-functioning asset markets.

Despite Mises’s explicit focus on entrepreneurship, much of modern production theory—indeed, the entire neoclassical theory of the firm— focuses not on entrepreneurs, but managers. The traditional theory of profit maximization is nearly always told from the perspective of the manager, the agent who operates the plant, not that of the owner, who supplies the capital to fund the plant. Yet owners control how much authority to delegate to operational managers, so capitalists are, in an important sense, the ultimate decision-makers. To understand the firm, then, we must focus on the actions and plans of the suppliers of financial capital, the capitalist-entrepreneurs.

It is true, of course, that when capitalist-entrepreneurs supply resources to firms, they usually delegate to managers the day-to-day responsibility for use of those resources. The resulting possibility for managerial discretion is of course the focal problem of the modern literature on corporate finance and the theory of the firm. As discussed in chapter 2 above, the literature on corporate governance identifies a variety of mechanisms by which shareholders can limit this discretion, such as establishing boards of directors, using performance-based pay, adopting the “M-form” structure, and exploiting competition among managers and potential managers. Outside the firm, competition in product, labor, and capital markets helps align managers’ interests with those of shareholders.⁷

We should therefore be cautious in attributing the eventual divestiture of many acquisitions, particularly during the 1960s and 1970s, to managerial motives. Moreover, the claim that the acquisitive conglomerates of the 1960s and 1970s were inefficient is inconsistent with recent evidence that during those years, diversifying acquisitions—particularly those that created internal capital markets—tended to increase the market values of the acquiring firms (Matsusaka, 1993; Hubbard and Palia, 1999; Klein, 2001). In light of this evidence, “[t]he simple view that the 1980s ‘bustups’ were a corrective to past managerial excesses is untenable” (Matsusaka, 1993, p. 376). In short, both theory and empirical evidence cast doubt on the conventional wisdom that corporate managers made systematic, predictable mistakes by acquiring (often unrelated) subunits during the 1960s and 1970s, and that financial-market participants made systematic mistakes by approving these acquisitions.

Mergers, Sell-offs, and Efficiency: Theory and Evidence

Why, in general, do firms expand and diversify through merger? Why do they sometimes retreat and “refocus” through divestiture? The theory of merger is a subset of the theory of the optimal size and shape of the firm, a relatively undeveloped area in the Austrian literature. Chapters 1 and 2 above argue for a modified Coasian, or contractual, view of firm boundaries, in which the limits to organization are given by the need to perform economic calculation using prices generated in external markets. Other writers see the Austrian approach as more congenial to the resourcebased theory of the firm, defining firms’ capabilities in terms of Hayekian tacit knowledge (Langlois, 1992, 1994a; Minkler, 1993a). In either case, we can think of merger or takeover as a response to a valuation discrepancy: acquisition occurs when the value of an existing firm’s assets is greater to an outside party than to its current owners. Put differently, a merger can be a response to economies of scope, in that the value of the merging firms’ assets combined exceeds their joint values separately.⁸ As with any voluntary exchange, the transaction is (ex ante) advantageous to both parties, and should thus be welfare-enhancing.

New combinations of corporate assets can generate efficiencies by replacing poorly performing managers (Jensen and Ruback, 1983; Mitchell and Lehn, 1990), creating operating synergies (Weston, Chung, and Hoag, 1990, pp. 194–95), or establishing internal capital markets (Alchian, 1969; Williamson, 1975; Gertner, et al., 1994; Stein, 1997). In particular, considerable evidence suggests that the market for corporate control disciplines incumbent management. For example, Morck, Shleifer, and Vishny (1988) found that firms with lower Tobin’s q-ratios are more likely to be targets of takeovers. Tobin’s q measures the ratio of the firm’s market value to the replacement cost (or book value) of its assets. Because firms with low market-to-book ratios have low expected cash flows relative to the amount of invested capital, the market-to-book ratio can be interpreted as a measure of the firm’s investment opportunities (Smith and Watts, 1992; Gaver and Gaver, 1993), or as a measure of managerial inefficiency or agency conflict within the firm (Lang, Stulz, and Walkling, 1991). Low-q firms are the most likely takeover targets.

Given the benefits of takeovers, why are many mergers later “reversed” in a divestiture or spin-off? Here we distinguish between two basic views. The first, which may be termed empire-building, holds that entrenched managers make acquisitions, often paying with the acquiring firm’s (inflated) stock, primarily to increase their own power, prestige or control. These acquisitions produce negligible efficiency gains, and are thus more likely to be divested ex post. Most important, because the acquiring firm’s motives are suspect, such acquisitions are ex ante inefficient; neutral observers can predict, based on pre-merger characteristics, that these mergers are unlikely to be viable over time. By permitting these acquisitions, capital-market participants are also guilty of systematic error. Admittedly, in the empire-building view, markets did eventually correct these mistakes with the restructurings of the 1980s.⁹ Still, farsighted regulators could have reduced social costs by limiting such acquisitions in the first place.

A second view, which we term entrepreneurial market process, acknowledges that unprofitable acquisitions may be mistakes ex post, but argues that poor long-term performance does not indicate ex ante inefficiency. In the market-process perspective, a divestiture of previously acquired assets may mean simply that profit-seeking entrepreneurs have updated their forecasts of future conditions or otherwise learned from experience. As Mises (1949, p. 582) puts it, “the outcome of action is always uncertain. Action is always speculation.” Consequently, “the real entrepreneur is a speculator, a man eager to utilize his opinion about the future structure of the market for business operations promising profits. This specific anticipative understanding of the conditions of the uncertain future defies any rules and systematization” (Mises, 1949, p. 582, emphasis in original).¹⁰

As discussed above, this notion of entrepreneurial decision making under uncertainty squares with recent theories of acquisitions as a form of experimentation (Mosakowski, 1997; Boot, Milbourn, and Thakor, 1999; Matsusaka, 2001). In these models, profit-seeking entrepreneurs can learn their own capabilities only by trying various combinations of activities, which could include diversifying into new industries. Firms may thus make diversifying acquisitions even if they know these acquisitions are likely to be reversed in a divestiture. This process generates information that is useful for revising entrepreneurial plans, and thus an acquisition strategy may be successful even if individual acquisitions are not. In these cases, the long-term viability of an acquisition may be systematically related to publicly observable, pre-merger characteristics associated with experimentation, but not characteristics associated with managerial discretion.

To explain the particular pattern of mergers and acquisitions observed over the last several decades, market-process explanations must appeal to changes in the legal, political, competitive, or regulatory environments that affect the ability of entrepreneurs to anticipate future conditions. Why, for instance, was it particularly difficult for entrepreneurs to forecast the success of acquisitions in the 1960s and 1970s? Why did entrepreneurs feel a greater need to experiment with various combinations of businesses during those years?

One possibility is that complex organizations with active internal capital markets were necessary in the 1960s, but became less important after capital markets were deregulated in the 1970s. The investment community in the 1960s has been described as a small, close-knit group where competition was minimal and peer influence strong Bernstein (1992). As Bhidé (1990, p. 76) puts it, “internal capital markets... may well have possessed a significant edge because the external markets were not highly developed. In those days, one’s success on Wall Street reportedly depended far more on personal connections than analytical prowess.” During that period, the financial markets were relatively poor sources of capital. In 1975, the SEC deregulated brokerage houses and removed its rule on fixed-price commissions. The effect of deregulation, not surprisingly, was to increase competition among providers of investment services. “This competitive process has resulted in a significant increase in the ability of our external capital markets to monitor corporate performance and allocate resources” (p. 77). As the cost of external finance has fallen, firms have tended to rely less on internal finance, and thus the value added from internal-capitalmarket allocation has fallen. Consequently, firms have adopted simpler, more “focused” structures that rely more heavily on external capital markets and outsourcing, possibly explaining some of the divestitures observed in the last two decades.

Are Divestitures Predictable? Evidence from a Duration Study

This section summarizes some ongoing research on the causes of divestiture (Klein and Klein, 2008). If acquisitions are most often symptoms of managerial empire-building, as suggested by Ravenscraft and Scherer (1987), Porter (1987), and other critics, then pre-merger characteristics associated with high levels of managerial discretion should be systematically related to the long-term failure and reversal of these acquisitions. In the market-process view, by contrast, long-term performance should be correlated only with pre-merger characteristics associated with experimentation, rapidly changing environments, or knowledge-intensive industries. Our empirical research finds little support for the empire-building hypothesis, and much stronger support for the market-process view. Specifically, we find that most characteristics typically associated with empire-building are poor predictors of merger duration. The evidence also suggests that divestiture is more likely when the original acquisition is driven by industry-specific competitive or regulatory shocks.

Klein and Klein (2008) study 222 pairs of firms that merged during the 1977–1983 period. Of the 222 acquisitions, 64, or almost 30 percent, had been divested by July 1995. We used a duration or “hazard” model to study the effects of pre-merger characteristics on the time to divestiture. Duration models help explain how exogenous factors, unobserved factors, and time itself affect the average duration until some discrete event (in our case, divestiture) occurs. Duration analysis allows us to see, historically, how characteristics of the acquiring and acquired firms affect the likelihood that the acquired firm will later be divested.

To study the effects of pre-merger characteristics on average merger duration, we estimated a hazard regression of duration (measured as the natural logarithm of the number of days) on a constant and on a series of potentially exogenous factors. For assessing the empire-building hypothesis, we included three characteristics associated with high levels of managerial discretion: relatedness of target and acquirer, differences in price-earnings ratios, and the medium of payment. Relatedness addresses the common view that managers deliberately pursue unrelated targets to expand their control or make themselves more valuable to the firm. Following Kaplan and Weisbach (1992), we examined this claim by constructing a dummy variable equal to one if the acquiring and target firm share at least one two-digit SIC code and zero otherwise.

Differences in price-earnings ratios are commonly seen as another indicator of merger motives. Merger critics have often suggested that acquiring firms grow and prosper by “bootstrapping.” This refers to the practice whereby bidding firms seek targets with low P/E ratios to boost their reported earnings per share. It is trivially mathematically true that when a firm with a high P/E multiple acquires a firm with a low P/E multiple and pays with its own stock, the acquirer’s earnings per share will rise, simply because the combined earnings of the two firms will then be divided by a smaller number of total shares outstanding. Hence, it is argued, acquiring firms can expand rapidly, with market approval, as managers exploit this accounting opportunity.¹¹

Of course, this argument assumes market participants could be systematically fooled by a simple algebraic trick. Admittedly, much more complicated financial and balance-sheet manipulations are often used in corporate-control transactions: bidders in the 1960s and 1970s sometimes financed acquisitions with convertible bonds, convertible preferred stocks, and other unique instruments. Although investors could (and eventually did) require that earnings be reported on a “fully diluted” basis, to take account of these manipulations, Malkiel (1990, p. 61) reports that “most investors in the middle 1960s ignored such niceties and were satisfied only to see steadily and rapidly rising earnings.” However, there is no evidence that bootstrapping was either prevalent or successful (Barber, Palmer, and Wallace, 1995; Matsusaka, 1993).

Nonetheless, we included a measure of relative P/E ratios to see if acquirers that choose targets with lower P/E ratios are historically more likely to divest those same targets, implying that bootstrapping does tend to fool investors, as increases in reported earnings disguise inefficient acquisitions.

We also included a series of dummy variables to represent the medium of payment used in the merger. Several theories suggest that how an acquisition is financed can affect performance. First, the bootstrapping technique described above works only for mergers financed by stock swap. Second, Jensen (1986) holds that financing takeovers by issuing debt serves to discipline the acquiring firm’s management by reducing post-merger discretion in the use of free cash flow. If true, we would expect entrenched managers to avoid making acquisitions using debt, opting for stock swaps instead.

In the market-process or experimentation view, by contrast, divestitures occur when the acquirer receives new information about the target after the merger has taken place. Plausibly, some relevant characteristics of potential target firms can be learned only by experience, forcing entrepreneurs to revise their plans accordingly. What kinds of targets are most likely to have unknown characteristics? Large firms engage in more activities than smaller firms, so potential acquirers have more to learn. On the other hand, smaller firms are less likely to have been written about in the business press, so one could plausibly argue that private information is a bigger problem for small targets.¹² We included target size in our regressions to see if either aspect is relevant. Firms in rapidly changing, knowledge-intensive industries are also likely to have characteristics hidden from potential acquirers. To capture this effect, we created a dummy variable based on two-digit SIC codes. The dummy takes a value of one if the target is in any of the following industries: computers (systems, software, and services), medical products, communications, aerospace, and miscellaneous high-tech industries. We also included target R&D because R&D is difficult to value, especially to outsiders, and thus firms in R&D-intensive industries are likely to have hidden characteristics. Potential acquirers would know potential targets’ R&D expenditures (which is reported), but this may not give the acquirers much information on the quality of the research or even the content of the R&D.

Finally, unrelated acquisitions, as discussed above, may be a form of experimentation, as firms try new combinations of activities to find those that best fit their existing capabilities. Although match-seeking acquisitions are more likely to be divested, they can still be part of a value-maximizing acquisition strategy. Our relatedness variable, described above, can proxy for match-seeking behavior. However, the best way to identify matchseeking firms is to look directly at historical patterns of acquisitions and divestitures. A firm with a history of repeated acquisitions and divestitures, especially acquisitions into unrelated industries, is likely to be a match-seeker, and thus any current acquisition is more likely to be divested. Unfortunately, our merger sample is too small to compile detailed acquisition and divestiture histories on individual acquirers. As a first approximation, we searched our sample for acquiring firms with at least one previous acquisition that was later divested within a few years of the acquisition. We created a dummy variable equal to one if the acquirer in a particular merger met these criteria, and zero otherwise.¹³

As noted above, periods of intense merger activity in particular industries may be responses to industry-specific competitive or regulatory shocks. Mitchell and Mulherin (1996), Andrade, et al. (2001), and Andrade and Stafford (2004) argue that mergers tend to occur in industry clusters, suggesting that industry-specific factors are important. Mitchell and Mulherin argue that corporate takeovers are often the most cost-effective way for industries to respond to these shocks. Moreover, they add, “because takeovers are driven in part by industry shocks, it is not surprising that many firms exhibit volatile performance following takeovers, with actual failures following some negative shocks” (Mitchell and Mulherin, 1996, p. 195).

To capture the effects of industry-specific shocks such as regulatory and tax changes, we included measures of industry clustering, both for acquirers and for targets, in our regressions. We did find substantial clustering in our sample. For example, slightly more than half of the mergers during the 1977–83 period occurred in only five of the thirty-seven (two-digit) SIC industries in our sample, both for acquirers and targets, with three-quarters occurring in ten industries. We constructed clustering variables by counting the number of mergers in each two-digit SIC category in each year and creating variables for each merger corresponding to the number of mergers occurring (a) within one year of the merger under observation (including the year before the merger, the year of the merger, and the year after the merger, to span a three-year window), (b) within two years of the merger (a five-year window), and (c) within the entire sample (a seven-year window). If risky acquisitions or increased levels of experimentation tend to appear during periods of industry-specific shocks, then these clustering variables will have negative and significant effects on the length of time to divestiture.

Our results challenge the empire-building hypothesis and offer evidence more consistent with the market-process view. Of the variables associated with agency problems, only relatedness has a statistically significant effect on merger duration. Neither P/E differences nor the medium of payment has a statistically significant coefficient in any of several specifications. The coefficient on relatedness is positive, meaning that related acquisitions are less likely, on average, to be divested (as in Kaplan and Weisbach, 1992). However, the coefficient on relatedness is also consistent with the market-process explanation. Of the five variables associated with this view, the coefficient on relatedness is positive and significant, the coefficient on target R&D is negative and significant, and the coefficient on the match-seeking indicator is negative and significant, all as expected. The coefficients on target size and the indicator for high-technology industries have the expected signs (positive and negative, respectively) but are not statistically significant.

Overall, our findings suggest that the divestitures in our sample are not, on average, the predictable result of unwise acquisitions. Rather, divestitures follow experimentation and learning, healthy characteristics of a market economy. Moreover, industry clustering appears to have a regular effect on average merger duration. The coefficients on our acquirer-clustering variables are consistently negative and statistically significant. (The coefficients on the target-clustering variables, by contrast, are not statistically significant.) This suggests that volatile performance does follow shocks, as suggested by Mitchell and Mulherin (1996).

This finding is consistent with the view that firms make acquisitions when faced with increased uncertainty (see Spulber, 1992, pp. 557–59). Regulatory interference could be a major cause of such uncertainty. As discussed above, government intervention makes economic calculation more difficult, and can ultimately render calculation impossible. When faced with increased regulatory interference, firms apparently respond by experimenting, making riskier acquisitions, and consequently making more mistakes, ex post.

Conclusions

Do entrepreneurs make predictable mistakes? Theory and evidence suggest otherwise. Contrary to the conventional wisdom on mergers and sell-offs, divestitures of previously acquired assets do not necessarily indicate that the original acquisitions were mistakes. Indeed, empire-building motives do not seem to be systematically related to long-term merger performance. Instead, divestitures are more likely to be associated with experimentation, learning, and other socially beneficial activities.

Acquisitions are uncertain endeavors, and the entrepreneur-promoter —along with the manager to whom he delegates authority—is a speculator. If the consequences of his actions were determinate, he would not be an entrepreneur, but rather, as some economic theories seem to treat him, “a soulless automaton” (Mises, 1949, p. 582). However, the future can never be known with certainty; long-term profit and loss cannot be predicted based on current information. As Mises explains:

What distinguishes the successful entrepreneur and promoter from other people is precisely the fact that he does not let himself be guided by what was and is, but arranges his affairs on the ground of his opinion about the future. He sees the past and the present as other people do; but he judges the future in a different way. (p. 585)

As discussed above, it is hardly surprising that these judgments are less than perfect. The relevant question for policy is whether there is a feasible alternative to market-based corporate governance. Our reading of the political-economy literature leaves us doubtful that such an alternative exists.