WP__I_`_ 13 0 POLICY RESEARCH WORKING PAPER 1830 Private Ownership Data on mid-sized firms In three transition economies and Corporate Performance provide strong evidence that private ownership - except . . ~~~~~for worker ownership - Some Lessons from Transition dr oves dramratically Improves Economies corporate performance. And the privatized firms' superior Roman Frydman ability to generate revenues Cheryl W. Gray allows those firms to sustain Marek Hessel or expand employment. Andrzej Rapaczynski The World Bank Development Research Group September 1997 | POLICY RESEARCH WORKING PAPER 1830 Summary findings Using a large sample of data on mid-sized firms in the superior ability to generate revenues, rather than Czech Republic, Hungary, and Poland, Frydman, Gray, competence at cost-cutting, that allows them to sustain Hessel, and Rapacynski compare the performance of or expand employment. This is why privatization is the privatized and state firms in the environment of the dominant strategy for expanding employment in postcommunist transition. transition. They find strong evidence that private ownership - Outsider-owned firms perform better than insider- except for worker ownership - dramatically improves owned firms on most performance measures, but there is corporate performance. enough difference between employee- and manager- They find no evidence of the "privatization shock" that owned firms to suggest that putting all insiders under a was supposed to afflict the behavior of firms undergoing common umbrella is unjustified. Although the effects of rapid changes in ownership. Instead, they observe a managerial ownership are ambiguous, putting employees severe shock from marketization, affecting both state and in control appears to offer no advantages over state privatized firms - but a shock for which private ownership on any measure and creates a distinct ownership provides a powerful antidote. disadvantange in terms of employment performance. Among their other findings: Among outsider owners, privatization funds seem to Private ownership is most effective in improving a do as well at revitalizing the privatized companies as do firm's ability to generate revenues, an area in which other outsider owners; in particular, the authors find no entrepreneurship seems to be required. Ownership also evidence that funds are less effective than "strategic" affects a firm's ability to remove the rather obvious cost investors. And foreign investors provide perhaps less of inefficiencies inherited from the past, but this effect is an edge than might have been expected; their impact less pronounced, as both state and privatized firms appears no stronger than that of major domestic engage in significant cost restructuring. outsiders. Most important, privatized firms generate significantly more employment gains than state firms. It is their This paper - a product of the Development Research Group - is part of a larger effort in the Bank to explore issues of corporate governance in transition economies. The study was funded by the Bank's Research Support Budget under research project "Corporate Governance in Central Europe" (RPO 678-42). Copies of this paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Bill Moore, room N9-038, telephone 202-473-8526, fax 202-522-1155, Internet address gmoorel@worldbank.org. September 1997. (38 pages) The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent. Produced by the Policy Research Dissemination Center PRIVATE OWNERSHIP AND CORPORATE PERFORMANCE: SOME LESSONS FROM TRANSITION ECONOMIES' Roman Frydman,2 Cheryl Gray,3 Marek Hessel,4 Andrzej Rapaczynski5 The authors are grateful to the CEU Foundation, the Open Society Institute and the World Bank for supporting research on this paper. CV Starr Center for Applied Economics at New York University has provided additional support for Roman Frydman's research. None of these institution are responsible for opinions expressed in this paper. The authors would like to thank Joel Turkewitz for his contributions to the design and implementation of the survey instrument, and Mihaela Popescu for her extraordinary assistance in the analysis of the data. The authors also thank Sarbajit Sinha for computer support in the initial stages of research. 2 Department of Economics, New York University, and CEU Privatization Project 3 The World Bank 4 Graduate School of Business, Fordham University, and CEU Privatization Project 5 Columbia University School of Law and CEU Privatization Project 1. INTRODUCTION This paper is part of a larger study which deals, on different levels, with two issues: the impact of ownership on corporate performance in the environment of the postcommunist transition, and the more general role of ownership in a market economy. What connects the two is the special significance of the transition environment for the study of the importance of private ownership. This environment, because of its far-from-equilibrium characteristics, offers a unique testing ground for a number of hypotheses concerning the more universal economic significance of private ownership. In more stable market economies, the impact of ownership on performance is more difficult to test. As Demsetz and Lehn (1985) point out, a firm's ownerships structure is itself subject to market pressures, so that in the long-run each firm that is capable of surviving competition will end up with an essentially close to optimal ownership structure. As a result, the impact of ownership on performance will remain largely unobservable in any society in which this process has been operating for some time. To be sure, state ownership is much less subject than private ownership to the operation of this "market for ownership structure." Nevertheless, the state may also be under pressure to stay out of areas in which the performance of state-owned firms is most manifestly inadequate (and the most difficult to legitimize), and confine itself to those in which access to large amounts of capital (which the state can provide or guarantee) is relatively important, where competition is less demanding, and where alternative governance arrangements may to some extent substitute for monitoring by owners. Furthermore, the presence of a set of institutionalized routines and governance arrangements may partly compensate, at least in the short term, for the attenuated presence of the owners of capital in some firms, and thus obscure the long-term effects of the ownership itself The most important characteristics of firms in which such compensatory arrangements can to a considerable extent substitute for private ownership are size and maturity. If a state firm operates in an area in which entry is costly, and technological change is slow and relatively predictable, and if the firm has a well established product and market, competent management should permit it to preserve its position and profitability, allowing it to perform at levels not so manifestly inferior to private firms as to create serious legitimation problems. Car manufacturers like Renault, state telecom companies prior to the recent technological revolution, as well as state railroads and airlines could for a while perform at levels not so manifestly inferior to private firms as to create serious pressure for denationalization. Even then, state firms usually deteriorate over time due primarily to political depredations and special-interest demands that state bureaucrats find very difficult to resist. Absent such political pressures, however, state enterprises in some areas may hold their own and withstand a certain amount of competition.1 When ownership is likely to matter most and governance substitutes are likely to be much less effective, is when a firm's environment is characterized by a high degree of uncertainty and The fact that state firms tend to operate in more capital intensive and less competitive sectors often increases the possibility of waste and political mismanagement. The creation of a competitive environment can then sometimes improve performance as much as privatization. See Vickers and Yarrow (1988). 2 the conduct of business calls for the exercise of entrepreneurship. If technological innovation, general economic dislocation, or other exogenous shocks produce a need for rapid reorientation and risk taking, state ownership -- we suggest -- will be particularly ineffective, even if state firms operate in a competitive environment.2 The postcommunist transition environment, ripe with such shocks and characterized by far-from-equilibrium conditions, offers a unique testing ground for a number of hypotheses concerning the role of ownership in corporate behavior and performance. The following factors are particularly relevant in this respect: 1. The transition economies are characterized by a very large number of state and privatized firms in all areas of economic activity which start and operate in essentially similar conditions. This allows meaningful performance comparisons between state and private firms, as well as comparisons among various forms of private ownership, 92. More generally, the environment of the postcommunist transition contains a wide variety of ownership forms in the early stages of their evolution, with many probably dysfunctional forms of ownership present and the distribution of the various forms largely unrelated to their effectiveness for the particular type of business. This "far from equilibrium" condition makes it possible to examine the relative effectiveness of various forms of ownership before the less appropriate ones are weeded out. 3. Many of the mechanisms that attenuate -- and perhaps even substitute for -- the role of owners in mature economies are absent from the transition environment. Many firms are still shielded from external competition. Markets for managerial talent are very imperfect.3 Securities markets are in their infancy, and the degree to which they exercise any disciplining effect on management is still quite negligible. Corporate governance institutions, such as boards of directors, are still largely ineffective in substituting for direct monitoring by owners. 4. Not surprisingly, therefore, ownership tends to be extremely concentrated in most Central and East European firms.4 Owners, who cannot afford to rely on other institutions and arrangements designed to monitor and discipline managerial performance, have both the power and incentives to exercise their rights. Thus, although the separation of ownership and management is common in the region, agency problems are likely to be tempered by 2 The results of our empirical work comparing the different ways in which state and privatized firms respond to competitive pressures will be presented in a separate paper. That competition in general, which only forces firms to do no worse than others, cannot substitute for ownership is argued in Frydman, Phelps, Rapaczynski, and Shleifer (1993) 3The results of our study relating to managerial turnover and its causes, attitudes of managers and business strategies will be presented in a separate paper. 4Even m the Czech Republic the intial fears of widely dispersed holdigs did not materialize and more than half of the voucher points distributed in the mass privatization program ended up with a handful of investinent funds. See Frydman, Rapaczynski, and Turkewitz (1997) and Coffee (1996). 3 the direct involvement of owners in many important decisions and day-to-day monitoring. 5. Finally, the environment of transition is characterized by a high degree of uncertainty and deep institutional change. Under these conditions, standard rules-driven behavior brings relatively meager competitive advantages. Success calls for spontaneity and innovation that can only be provided by the owner - indeed, not just any owner, but a private owner who is actively engaged in the running of the business. If, therefore, different types of owners differ in their ability to improve their firms' performance, the environment of Central and Eastern Europe should magnify such effects and allow for their more accurate measurement and analysis. 2. METHODOLOGY Evaluating the impact of ownership There are two ways in which the relation between ownership and performance can be evaluated. The first is to compare pre- and post-privatization performance of selected privatized firms. Although intuitively appealing, this "historical" approach can suffer from the difficulty of sorting out a potential variety of causal factors. Pre- vs. post-privatization comparisons are more meaningful if they involve privatization programs through which significant numbers of state companies are transferred to the private sector. Such programs, however, usually come as parts of broader shifts in economic policies, often including, in addition to privatization, such changes as tighter financial discipline and a reduction in subsidies. As a result, it is difficult if not impossible to separate the effect of new ownership from the impact of the new policies.5 Moreover, traditional privatizations frequently involve significant injections of funds into the firms to be privatized and are usually preceded by extensive preparations during which management and organizational structures are revamped.6 The impact of these preparations may well account for some or all post-privatization improvements and cloud the impact of ownership change on company performance.7 5This is also true in transition economies, where, as a result of variety of policy changes, restructuring of operations is common to privatized and state firms alike and thus cannot be attibuted to ownership changes alone. 6 British Steel, for example, is reported to have reduced its employment by 40% prior to privatization while keeping revenues constant; British Airways reduced its labor force by the same percentage, while actually increasing the number of flights. Djankov and Pohl (1977). 7Among the "historical" studies using privatized firms in predominantly market economies, the most important is Megginson et al. (1994), who compare the pre- and post-privatization peiformance of 61 firms in 18 countries and find that privatization increases profitability, output per employee, investment, and employment, as well as leading to lower leverage and higher dividend payouts. But the study does not control for changes in the econonic environment or most fonns of pre-privatization restructuring. Moreover, although the authors try to reject the possibility that the improvements they noted were a result of significant injections of funds into the privatized companies, the rejection is based on a lack of reports of such injections in a few Western newspapers, and as such may not be overly reliable. Most of the remaining pre- 4 The second approach to evaluating the relation between ownership and performance focuses on comparing the performance of state and private firms operating under reasonably identical conditions: at the same time, in the same markets, within the same environment. This "synchronic" approach confronts its own difficulties. In order to capture the effect of ownership cleanly, a study must assure the absence of a selection bias. Such a bias might occur, for example, if state ownership was introduced in the first place to shore up firms that could not compete in the market (so that state firms were handicapped from the beginning). Or, if one wants to avoid this problem by comparing the performance of privatized: (rather than originally private) firms to that of state companies, selection bias may occur if better firms were chosen for privatization.8 A reliable study of the effects of ownership on performance must combine the two approaches, comparing state and private firms operating at the same time, but using historical data, in addition to the synchronic comparisons, to eliminate the possibility of selection bias. This is the route we follow in this paper. We focus on comparisons of performance between state and privatized firms, leaving out the issue of the differences between either of these types of 1989 literature, predominantly based on small samples and lacking statistical significance, is listed and classified in Boardman and Vining (1989). An earlier review, together with an argument that private ownership is more efficient, may be found in De Alessi (1980). Among more recent studies, Caves (1990) surveys the impact of British privatization policies of the 1980; in a case-type study, Galal et aL (1994) examine the pre- and post-privatization performance of 12 firms in developing countries. Most of the literature on the postcommunuist transition and enterprise restructuring also follows the historical approach and is by now very large. Frydman and Rapaczynski (1994) provide references to the initial discussions which focused on the merits and demerits of different privatization programs. The first performance-focused studies of firms in transition dealt with so-called "small" privatization" (i.e., privatization of shops and other small establishments), which raises different issues from those discussed here. See Earle, Frydman, Rapaczynski, and Turkewitz (1994). Perfornance- focused studies of "large" privatization tend to concentrate on individual countries, particular owners, or specific problems, and usually involved comparisons of pre- vs. post-privatization performance. Some works of interest in the context of this paper are: Pinto, Belka and Krajewski (1993) and Pinto and van Wijnbergen (1994) (on the restructuring of state enterprises in Poland); Blanchard, Commander and Coricelli (1994) (on unemployment and restructuring in Eastern Europe); Djankov and Pohl (1997) (on restructuring of large firms in Slovakia); Balcerowicz, Gray and Hashi (1997) (on employment and output downsizing in the 200 largest manufacturing finns in the Czech Republic, Hungary, and Poland); Earle and Estrin (1996) (on employee ownership in transition); Claessens, Djankov and Pohl (1997b) (on the governance performance of Czech investment funds). Russian privatization, with its fairly unique problems, also produced a large number of studies: Lieberman and Nellis (1994) contains a number of essays on the restructuring of privatized firms. s The most important "synchronic" study of performance of state, private and mixed enterprises in market economies is Boardman and Vining (1989), who compare the performance of the largest 500 non-US industrial corporations and find that private firms perform better along several dimensions than both state-owned and mixed firms, and that mixed firms do no better, and often worse, than state-owned companies. Although the Boardman and Vining control for sectoral and country influences, such controls are insufficient to exclude the selection bias hypothesis. Among the still rare synchronic comparisons of state and privatized firms in transition economies, the most important is the study by Pohl et aL. (1997), who examine the performance of a large number of state and privatized firns in seven transition economies between 1992 and 1995. The study by Pohl et al, which appeared as this paper was going to print, begins to build a very powerful case in favor of the effectiveness of privatization. But the study fails to deal with the possibility of selection bias and is thus potentially open to the objection that the superior performance of privatized firrms it reports may be due to factors other than the ownership itself. 5 companies and new private firms.9 To establish that pre-privatization selection bias does not affect post-privatization results, we evaluate the pre-privatization performance of privatized firms relative to state firms and show that, originally, they did not differ significantly from each other (see part 5 below).10 To establish the later effect of ownership changes, we evaluate the post- privatization performance of privatized firms, again relative to state firms. Based on these comparisons, we offer strong empirical evidence that private ownership, except for worker ownership, dramatically improves the most essential aspects of corporate performance in the countries undergoing the postcommunist transition. We find no evidence of any "privatization shock" afflicting the behavior of the firms undergoing rapid ownership changes; we observe instead a severe shock of marketization, which affects both state and privatized firms, but to which private ownership provides a powerful antidote. Moreover, the effect of privatization is the strongest where entrepreneurial acumen matters most -- it is the ability to stem revenue losses and often generate revenue increases that most strongly distinguishes privatized from state firms. Ownership also affects a firm's ability to remove rather obvious cost inefficiencies inherited from the past, but this is less pronounced, with both state and privatized firms engaging in significant cost restructuring. Most importantly, we find that privatized firms are also more successful at moderating employment losses. Since it is the revenue-creating capability -- rather than cost-cutting restructuring -- that allows a firm to grow, privatization turns out to be the dominant employment strategy in transition. The paper is organized as follows: the remainder of this section describes the sample firms, discusses the performance measures used in the analysis, and presents the two equations we use throughout the paper. The first equation allows us to estimate the general impact of privatization across all types of private owners, which we examine in Section 3. The second equation allows us to evaluate the performance differences among various private owners, which we discuss in Section 4. We complete our analysis in Section 5, where we evaluate the pre-privatization performance of privatized firms relative to state firms and reject the possibility that selection bias or moral hazard effects might affect our results. The final section contains concluding remarks and previews the remaining areas of our study. The supplement provides the regression statistics (Appendix A) and a detailed sample description (Appendix B.) The sample This paper analyzes performance implications of privatization of mid-sized firms in the Czech Republic, Hungary and Poland over the initial period of transition, from 1990 to mid-1994. 9 Although startup firms are clearly the most dynamic and entrepreneurial businesses in Central and Eastem Europe, they are usually quite small and operate against very different background conditions (the most important of which is the absence of the pervasive inheritance from the socialist past), so that comparisons to both state and privatized firms raise extremely difficult methodological problems. 10 Moreover, the overwhelming majority of Central and East European privatizations (and in nearly all cases of medium-sized companies we examine m this study) privatization was not preceded by any special preparations or other pre-privatization restructuring that would further distinguish the privatized firms from their state-owned counterparts. 6 These firms were part of a sample of over 500 mid-size firms, employing between 100 and 1,500 persons; the median 1993 employment in the sample is about 360 full-time employees, and the median 1993 sales are just short of US$ 6 million. The size restriction brings into focus the region's more vibrant and dynamic manufacturing sector, which retains a lion's share of output and employment in transition economies. Its performance has significant implications for economic stability, budgetary policies, and political developments in the region; the ability of these firms to cope with the new environment is of vital importance to the transition process. The size restriction also leaves out the industrial "dinosaurs" of the communist era, the privatization of which raises special political and social problems. The sub-sample used in this paper includes only state and privatized firms (i.e., it excludes firms which were never state-owned.) By a state firm we mean a firm in which private parties do not have a blocking power: its legal form may be that of a non-corporatized state enterprise, a corporatized state enterpn'se, or a privatized firm in which the combined shareholdings of private owners fall short of blocking power. "' By a privatized firm we mean an enterprise (partially or totally) privatized.through a privatization of a predecessor state-owned company (or its part) in which the combined holdings of private parties give them a blocking power. 12 To allow for analysis of post-privatization performance, we limit our sample to firms privatized in 1990, 1991 and 1992 only"3. The number of these firms used in the analysis varies slightly with the particular aspect of performance under examination, as not all firms provided data complete enough to meaningfully track their showing-4 in general, however, the sample includes about 190 firms. These firms are split about evenly between privatized and state companies; 35% of them are in the Czech Republic, 39% in Hungary, and 26% in Poland. They operate both in consumer goods (food and beverages, clothing, and furniture) and in industrial goods sectors (non-ferrous minerals, chemicals, textiles and leather), with roughly 58% of I1 We consider pnvate parties to have blocking power if they control the percentage of votes formally sufficient to block major decisions at the general shareholder meeting. If this is the case, we classify the frm as non-state even if the state retains majority holdings, the situation arising in 15% of the privatized firms in our sample. Although the blocking power requirement is different from majority holding, the high concentration of holdings in our sample makes the difference academic. (For the identity of the largest owners and their holdings in the privatized finns in the sample, see Table B.2, Appendix B.) 12 However, we have excluded from our sample all firms privatized through leasing. (Ihere were 25 such fims out of 506 in our sample, with 11 providing data complete enough to evaluate their perfornance.) The nature of leasing in Poland and ESOPs in Hungary made it difficult for us to categorize unambiguously the leased firms according to their ownership, especially when they were not employee-owned. Still, the inclusion of those firms produces no significant changes in the results reported in this paper. 13 There were no annual post-privatization data for sample finms privatized in 1993 or early 1994. 14 We have no reason to believe that the incompleteness of data for certain firms (which relates primarily to lack availability or an obvious misunderstanding of the meaning of certain questions, such as those concerning the initial period for which data were to have been provided) introduces any systematic bias "in favor" or "against" any group of firms. 7 privatized firms and 48% of state firms in consumer goods sectors. (Distributions of sample firms by type, country and industrial sector are provided in Appendix B.) All privatized firms in the sample have highly concentrated ownership: except for privatization funds, the average holdings of private parties in the position of the largest owner are majority holdings." The sole exception - privatization funds - is due to legal limitations. All of these institutions in our sample are investment funds in the Czech Republic, whose holdings in individual firms were legally capped at 20 percent. Even then, the combined holdings of different investment funds in a single firm typically add up to a majority. This degree of concentration not only identifies the typical ownership structure as a domi- nant-owner arrangement, but also comes sufficiently close to a single-owner structure to allow us to ignore the other owners and identify the firm's ownership type with that of the largest shareholder."6 The most frequent among those owners in our sample are foreign investors (the largest shareholders in nearly 30% of privatized firms), followed by managerial or non-managenral employees (over 20% of privatized firms.) (See Appendix B) Performance measures We focus much of our attention on the revenue performance of a firm. Economists tend to think of profit as the measure of performance that best captures both the creativity (the revenue side) and the discipline (the cost side) required for survival in a market economy. In the long run this is true. But in the short term, profits may be extremely volatile and subject to a number of accounting decisions, especially with respect to costs, that bear little relation to long-term performance. They are also sensitive to cross-country differences in accounting methods, making comparability that much more difficult. These vulnerabilities are particularly true in the initial stages of transition, which are characterized by dramatic changes in accounting systems and practices, imperfect disclosure systems, and short reporting histories. But there is a more important reason why we take special interest in revenue performance: because of its relation to entrepreneurial success, we find the revenue performance of privatized firms to best reflect the advantages of private ownership. If the cost side of the profit equation seems the bread and butter of most managers, it is because costs relations tend to be known to them (or other company insiders) with a high degree of certainty and because cost-cutting measures are often a matter of relatively standard procedures. Revenues are not only less subject The ownership distribution in our sample is not fully representative of the population of firms from which it was drawn, since the sample was stratified by the type of owner. Thus, if certain owners (such as foreign trade investors) tend to acquire larger stakes, stratification may bias the sample in favor of greater concentration. Simnilarly, if a firm was chosen in order to include a certain type of owners, the firm so chosen would probably be one in which that owner was represented at a significant enough level (thus adding to an overall bias in favor of greater concentration). 16 State as the largest owner provides an important exception. As we discuss later in the paper, in privatized firms when the state remains the largest owner, the second owner (or, more generally, the largest private owners) is likely to be important. 8 to manipulation and more transparent to an outside observer, but also more future-oriented and unpredictable on the basis of past history.17 To capture new sources of revenues or to regain the disappearing ones is a matter of entrepreneurial, risk-taking activity even in mature economies. As The Financial Times put it, "costs are essentially static, since they encapsulate the past history of the company. Revenues are dynamic, reflecting the ebb and flow of economic activity, customer preference and pricing signals."'8 These characteristics of costs and revenues are particularly true in the environment of transition, where the gross cost inefficiencies of the past are easily identifiable, as are, most often, their most obvious remedies. On the other hand, skills required to restructure most companies -- when their old markets have collapsed, imports have introduced overnight competition from the most advanced world producers, and buyers have become more careful and demanding -- are not that different from those needed to start a new business; within some additional constraints (such as the existing labor force or the already available machinery), the post-communist firms must reinvent their products and find markets in which they can be sold The evaluation of revenue performance of state and privatized firms alike must recognize the significant downward pressure that the early stages of transition put on revenues and employment. First, the very size of these firms was often not dictated by economic calculation;"9 second, the loss of some markets (such as COMECON) and/or the marketization of the remaining activities forced most of them to discontinue slow-selling products, spin off less productive assets, shed inefficient operations, and cut employment.20 This restructuring is common to state and privatized firms, and we view the resulting drop in revenues and employment as a necessary consequence of past distortions: we refer to it as the transition or marketization effect that all 17 We analyze differences in the way state and privatized finns approach uncertainty and risks inherent in restructuring of their activities in a forthcoming paper. 18 "It's the revenues, stupid," The Financial Times, Dec. 27, 1996. The focus on revenues could distort our performance comparisons if the superior revenue.performance of privatized firms could be attributed to mergers or acquisitions. There is, however, no evidence of any mergers or acquisitions in our sample. We examined the annual revenue changes of all privatized firms in the sample, and for all those that increased their revenues by over 25% within a single year, we evaluated the employment changes. The highest annual employment increase among those firms was about 18%, and it was a small firm that increased its employment from 176 to 209 employees -- hardly a merger-generated growth. The next highest annual employment increases among the same group of finms were 14.5% and 2.5%, respectively, indicating no major mergers or acquisitions in our sample. Another distortion could have been introduced if state firms were more likely than privatized companies to split or otherwise contribute a part of their assets to other entities. On this question we have direct evidence and it excludes such possibility: 20 privatized firms and only 11 state firms contributed some portion of their assets to other entities, and the effects of these contributions on employment and revenue measures used in our equations were approximately similar for state and privatized firms. 19 Enterprises in Eastern Europe tend, on average, to be larger than those in comparable developed market economies; see Komai (1992). 20 Balcerowicz, Gray and Hashi (1997) document the staggering decline of output and employment in the initial stages of transition. 9 firms must cope with. It is against this common effect that we evaluate the performance of privatized firms. We assume that the difference between the state and privatized firms -- which we label the privatization effect -- reflects the success (or failure) of privatized enterprises in dampening the transition effect and accelerating the expected conversion to a "normal" market economy. The privatization effect -- across all types of owners of privatized firms as well as for particular types of owners (foreign or domestic, individual or institutional, insider or outsider) -- is the focus of this paper. Although we emphasize the revenue performance most, we also contrast other performance indicators of privatized and state firms, beginning with their employment behavior. Differential abilities of privatized and state firms to generate revenue growth have potentially significant impact on their ability to sustain or generate employment. This impact is of particular interest to policymakers in the transition economies, where labor layoffs are among the most feared aftermaths of-- and the most socially and politically threatening deterrents to -- large-scale privatization. We augment the analysis of revenue and employment changes with an evaluation of revenue per employee performance, which proxies for labor productivity effect. Finally, we discuss the cost effects of privatization, with labor and material costs per unit of revenues as the measure of performance. Performance evaluation For each of the four performance measures -- revenues, employment, revenue per employee and costs per unit of revenues -- we use rates of growth to compare the performance of privatized and state firms. To smooth out the year-to-year variations of these rates, we annualize growth over the entire period of interest.2" For privatized firms, the growth rate is annualized bet- ween the year of a firm's privatization and 1993 (the final sample year),22 so that it reflects the post-privatization performance. Since the year of privatization varies among privatized firms, the rate annualizes performance of different "vintages" of privatized firms over different time horizons. To assure comparability with performance of state firms, we annualize the performance 21 That is, if PERF, denotes the value of a firm's performance measure in year t, the annualized rate of growth of performance, ARGPERF, is an imputed rate which satisfies PERFT/PERF, = (1 + ARGPERF)(T-') over the appropriate time interval (T - t), T > t. 22 By the "year of privatization" we mean the year in which the new owners assumed a defacto control of a firm, rather then the year in which the firm's shares were officially distributed. [The survey question pertaining to the date of privatization asked the CEO of a privatized firm to provide the date when "(a) the voting power ofprivate parties in the company rose above the percentage formally sufficient to block major decisions, or (b) if a significant portion of the company shares was privatized through vouchers, when new owners became active in company affairs (even if before their shares were distributed). "I The difference is of importance in the Czech Republic, where new owners could assume control in 1992 (when the auctions in the first wave of privatization were undertaken) even though the shares were distributed later. 10 of the latter over all possible time periods (1990-93, 1991-93, and 1992-93).23 The comparisons are based on a multivariate analysis which evaluates the performance of privatized firms through an additive performance contrast to state firms. The analysis proceeds in two stages. We begin with estimating the general effect of privatization, i.e., the average performance contrast across all categories of private owners and all countries.24 Here, we relate growth of a firm's performance, ARGPERF, to the initial value of its performance, INIPERF, and the its ownership type, PRV (with PR V= I if the firm is privatized and 0 otherwise): (1) ARGPERF = a + goINIPERF + /JPRV + e where ,B measures the incremental performance effect specific to privatized firms. The use of additive effect for privatized firms allows us to separate the transition and privatization effects.25 We interpret p = /JPRV + e2 as the the (average) privatization effect specific to private ownership. The remaining part of equation (1), r = a + a0INIPERF + e1 can then be interpreted as the performance element common to all firms, i.e., the transition effect. In equation (1), this effect consists of two components: the mean effect a and the initial position effect ao INIPERF. (Note that for revenue and employment regressions, the initial performance gauges the initial size of the firm.) Since equation (1) evaluates only the post-privatization performance of privatized firms, any contrasts with state firms could potentially reflect (i) a selection bias in the form of a choice (deliberate or otherwise) of better firms for privatization and/or (ii) a moral hazard effect in the form of a deliberate decision by insiders to bring down the pre-privatization performance of a 23 Ideally, the performance of each "vintage" of privatized finms ought to be compared with the performance of state firms annualized over the matching period. However, our data set is not large enough to allow separate comparisons for each vintage of privatized firms. The smallest "vintage," that of 1990, consists of 15 fums only; even the 1992 vintage, with 58 firms, becomes quite small when divided between three countries, various ownership categories, etc. This forces us to group all vintages of privatized fims together. As a result, frms within the single "privatized fimns" category have their post-privatization performance evaluated over different periods of time: those privatized in 1990 between 1990 and 1993; those privatized in 1991 between 1991 and 1993, and those privatized in 1992 between 1992 and 1993. This raises the question of the period over which the performance of state finns should be evaluated so as to "match" the evaluation of privatized finns. Clearly, no such matching is possible within any single period of time. Instead, we estimate the post- privatization performance of privatized finns against the performance of state finns evaluated, alternatively, between 1990- 93, 1991-93, 1992-93. These estimations produced three sets of estimates, depending upon which of these three periods was used to evaluate the performance of state firms. In virtually all cases, our results are invariant with respect to the time period used: the estimates retain their signs and significance across all three sets. 24 Complete definitions of all dependent and explanatory variables are provided in Appendix A. 25 The choice of state firms to provide a nonm (or benchmark) against which privatized finms are evaluated is not arbitrary. While the performance of state finns falls within a relatively narrow range, the performance of privatized firms shows too much variability to serve as a clear performance standard. Indeed, the gamut of finm-level actions it emancipates is one of the most striking (and telling) aspects of privatization, which deserves a separate treatment. We discuss this issue in a forthcoming paper. 11 state firm to enhance the likelihood of acquiring it. To ensure that neither of these biases drives our results, we compare the pre- and post-privatization performance of the same set of state and privatized firms, including and excluding firms which ended up in the hands of insiders. We also subject equation (1) to a number of tests to assure that the privatization effect does not proxy for other factors. These factors -- country and industrial sector in which a firm operates, time of pri- vatization, etc. -- are typically defined as 0-1 dummy variables, and injected into equation (1) as additive and/or interactive terms. Where these factors are significant, we modifyr the estimates of the transition and privatization effects accordingly The privatization effect in equation (1) is the average effect across all owners of privatized firms. It may, of course, vary with a particular type of ownership; indeed, it need not be positive for all types of owners, and there is a considerable discussion concerning the effectiveness of different private owners in the transition environment. In the second part of the analysis, we replace the privatization variable PRV in equation (1) with the dumnmy variable OWNRi, which identifies the largest owner of a privatized firm.26 This allows for evaluation of privatization effects for different types of owners -- the ownership effects, so to say. Since there are pronounced differences in the preponderance of different types of private owners across the three countries, we also augment the equation with country variables (HU= 1 for Hungary and 0 otherwise, PL = 1 for Poland and 0 otherwise). These changes yield equation (2) below: (2) ARGPERF= a + aoaINIPERF + E, BfOW RNR, + y,HU+ y2PL + e With 8i measuring the incremental performance effect specific to a privatized firm with a particular type of largest shareholder, pi = Ei,8i OWN'Ri + e2 is the privatization effect for the i-th type of owner. The remainder of equation (2), rc = a + aoINPERF + y,HU + y2PL + * represents the transition effect, which now includes country effects, y, HU or y2PL, in addition to the mean and initial position effects. All equations in this paper are estimated by ordinary least squares (OLS) with the White heteroscedasticity consistent estimator (1980) estimating the asymptotic covariance matrix. All regressions were run using L1MEP 6.0 econometric software. 3. TRANSITION, PRIVATIZATION AND PERFORMANCE We begin by evaluating the average effect of privatization in the early stages of transition. Bivariate summary measures of revenue and employment performance of state and privatized firms, shown in Figure 1 below, provide the background. (New private firms are included for comparative purposes.) Over 80% of state and over 56% of privatized firms lost revenues in each 26 As explained earlier, firms in our sample have very concentrated ownership. Accordingly, we identify a firm's ownership type with its largest owner. With few exceptions (to be noted in the text), we found the identity of other owners to be generally insignificant 12 year between 1990 and 1993, with most of them losing over 30% of their sales between 1990 and 1991 alone. The labor market story is even starker, with nearly all state and a majority of privatized firms losing employment over the entire sample period. It is a telling indication of the magnitude of the transition effect that even the few state firms with growing revenues had to decrease employment during the first four years of the transition to reduce the endemic overmanning of the old socialist enterprises. Figure 1 Decline and growth of firms in transition -0OFOETION OFFPlUMS-1 DI$IECLINING REVENUES 550-AN RATES OF REVENUE DECLINE OF F RMS WITH DECLINING RtEVENUIES 100] 94_ 80~~~~~8 SO * ElS4 rL '-20 -21 : 40 ,20 - l 10 4 36 _ t990-91 1991-92 1:92-93 1990-91 1991-92 1992-93 i ' _ _ _ _ _ __st7d 2 Nepwe[ t * P 1 [] Ndn PROPORTION OF FIRU6 WIąTH bECLINING EMPLOYMENT PROPORTION OF FIRMS WITH EMPLOYMENT INCREASES 0ETWEEN INITIAL AN0 EN.O.. YEAR sao 79 20 2 W 055. aRe -1 - 5 40 _ _ _ _ _ _ 1 L i, tle:9 911 20 ~Pvtzd- 1990-91 1891.92 1992-93 0 20 40 50 so 100 But if Figure 1 shows the strength of the transition effect, it also offers a glimpse into the power of the privatization effect. The percentage of privatized firms losing revenues in each of the sample years is considerably lower than the percentage of state firms; among revenue-losers, privatized firms tend to arrest the decline markedly faster than state firms; and nearly half of privatized firms managed to create new employment over the entire period.27 It might be objected that our results are skewed by a possible survival (exit) bias, since there may be more liquidations among privatized fums. There is no evidence, however, that this is the case; there have been many closures of state finrns under state enterprise laws (particularly in Poland) and relatively few closures (as opposed to reorganizations) of mid-sized and large state or privatized finns under bankruptcy laws in the three countries (Balcerowicz, Gray and Hashi, 1997). It is also worth noting that had such a bias toward the exit of non-viable private firms existed, it would have 13 Indeed, when the transition and privatization effects are isolated, the performance impact of privatization is quite dramatic. The table below shows the transition and privatization effects estimated from the revenue, employment, revenue per employee, and cost per unit of revenue versions of equation (1).2" TABLE 1 TRANSITION AND PRIVATIZATION EFFECTS OVER 1990-93 PERIOD Transition effect (%) Performance measure Privatization Mean Initial value effect (%) como net component Revenue .15.82* -0.25* 19.01* Employment 7.47* 0.40** 5.88* Revenue per employee -5.21 * -0.28** 11.17* Cost/revenue 30.06* -0.31* -3.00 *p<0.05 **p<0.10 Note first the enormous downward pressure exerted by the transition: the mean component of the transition effect depresses (annual) revenue growth by almost 16% and employment growth by over 7% . Larger firms are hit harder: the initial value component indicates that revenue and employment losses increase with the firm's size29. Not surprisingly, the pressure extends to the excess of revenues over costs: the mean component of transition effect increases the costs' share of revenues at the annualized rate of 30%. (Here, the initial value component works in the opposite direction, as gross inefficiencies tend to be easier to remove.) The transition effect should, of course, be understood as a temporary phenomenon: the failure to generate revenue growth or sustain employment cannot be forever attributable to the change of economic regime. Moreover, the (mean component of the) transition effect represents our interpretation of the average performance of state firms, and no firm can survive revenue provided a strong, independent argument in favor of privatization as an efficient resource reallocation mechanism - even while biasing our results. 2S These estimates result when performance of state finns is evaluated over the entire sample period, 1990-93. Evaluating this performance over the remaining two periods (1991-93 and 1992-93) affects neither the significance nor the order of magnitude of the estimnates. We therefore use the first set of estimates throughout the paper; full regression statistics for the remaining two sets are reported in Table A.2, Appendix A. 29 It is noteworthy in this connection that the initial revenue of a firm is a very powerful determinant of its later revenue growth, with larger firms suffering sharper revenue declines. The reasons for this are not apparent from our survey, but are not hard to conjecture. The larger the firm, the more political clout it used to have, and the less business justification was needed for its operations, under the old regime. It was also more likely to have been in a quasi- monopolistic position and to have produced goods for export to the COMECON countries, and less likely to be able to respond to a fast-changing environment. 14 declines of double-digit magnitude over long periods of time.30 Thus, the privatization effects in Table 2 suggest that privatization may the most efficient antidote to the shock of transition: privatized firms outperform state firms on all performance measures, and on all except costs per dollar of revenue the margins are very large and highly significant. This is particularly true of revenue growth, where the estimated privatization effect (+19.0 1) is at least as strong as the mean component of the transition effect (-15.82). Not only does the restructuring undertaken by privatized firns allow them to moderate significantly their revenue losses vis-a-vis their state counterparts, but it actually allows about half of them to generate revenue gains: estimates of equation (1) imply that a privatized firm with initial revenues of about US$ 6 million (approximately the median initial revenue of state and privatized finns in the sample) would overcome the transition effect, experiencing a modest 1.5% annualized growth between 1990 and 1993.31 In marked contrast, throughout the same period state firms would be losing revenues regardless of their size, with the median-sized firm suffering an annualized decline of 17.5% per year.32 Differences of this magnitude testify to the strength of the privatization effect, as well as to the odds that older firms must overcome during the early years of the transition. Our estimates belie the existence of any tangible "privatization shock" which afflicts the behavior of the firms that undergo swift ownership changes. On the contrary, what we observe is a very severe shock of marketization, which affects the performance of both types of firms, and to which privatization provides a powerful remedy It would be quite surprising if the superior ability of privatized firms to generate revenue growth (or moderate revenue declines) did not affect their employment behavior -- and indeed it does. The average employment effect of privatization is nearly 6%, significantly above zero.33 30 If some state firms survive nevertheless while continuing to show inferior performance, their inferiority will have to be ascribed to some sort of "state-ownership effect," rather than blamed on the difficulties of transition. 31 The expected value of the transition effect - the sum of the mean and initial revenue components - for firms with initial revenues of $6 mln is -17.5%, while the expected value of the privatization effect is 19%. 32 As startling as these estimates may appear, they reflect bona fide differences in the ways state and privatized furms restructure their activities: only 6 of the 89 state firms in the sample managed to increase their revenues over the entire 1990-93 period, while 33 of the 81 privatized firms did so in the wake of their privatization. This inferior performance of state firms holds regardless of the time period: between 1991 and 1993 only 11 state firms increased their revenues, and only 12 did so between 1992 and 1993 (See Figure I above). It must be remembered that this effect compares the employment of privatized firms with their state counterparts, and does not gauge absolute changes. These, on average, were negative even among privatized firns, a robust indication of the extent of overmanning of their communist predecessors. With the mean component of the transition effect at -7.47%, and the size component lowering it by .4% for each additional 100 workers, the transition constitutes a drag that is very difficult to overcome. Our estimate of the transition effect is quite consistent with the results of Pinto and van Wijnbergen (1996), who report a 27% drop in employment among large (average 1992 employment of about 3,000 15 As we shall see shortly, this effect is somewhat affected by the behavior of insider-dominated firms; however, with the exception of worker-owned firms, it does not come at the expense of labor productivity. This suggests that the increased revenues of privatized firms make up for what would otherwise have to be greater employment reductions (as compared with state firms) Our data indicate quite clearly that privatization arrests employment decline and often creates new employment. Privatization is not a matter of trading off long-term employment benefits for short-term costs - the positive employment impact of privatization is practically immediate, showing that privatization is an antidote to the deadly cycle of revenue falls, followed by layoffs, followed by further revenue falls, etc.34. So far as we can see, there are simply no costs to privatization in this respect: once the macroeconomic reforms are in place and the system of pervasive subsidization is stopped or significantly reduced, privatization is the dominant employment strategy in transition.35 Revenue per employee results are, of course, a firm-level version of the story foretold by the average revenue and employment effects. But the measure also proxies for the labor productivity, and therefore provides an independent estimate of efficiency of state and privatized firms. The negative mean component of the productivity effect indicates that heavy employment layoffs among state firms fail to arrest the decline in labor productivity; privatized firms, on the other hand, tend to reverse the slide despite relatively smaller labor reductions. Thus, while state finns seem to deal with the shock of transition primarily by downsizing, privatized firms are more successful at regaining their markets and laying foundations for future growth. The cost side of operations -- as measured by the decline of costs' share of revenues -- shows the least contrast between state and privatized firms. Although the privatization effect is negative (at -3.00), suggesting that privatized firms are gaining on state firms in terms of cost efficiency, the average effect across all privatized firms is not consistent enough to be significant. This confirms our prior expectations of cost cutting in transition environment. Most communist enterprises had worked under a "cost plus" system, in which prices of their products had been employees) Polish state enterprises, and Balcerowicz, Gray and Hashi (1997), who report 32, 47 and 33% decline in employment of the largest 200 firms in the Czech Republic, Hungary, and Poland, respectively, between 1989 and 1993. 34 It would be of considerable interest to estimate the employment impact of revenue changes. However, including revenue changes in the employment version of equation (1) would expose the estimates to a simultaneous equation bias, and our data set did not include enough instruments to allow simultaneous estimation. We have therefore opted for another way of gauging the relation between employment and revenues of the firms in our sample, namely estimating the ratio of the latter to the former. The results, presented below, confirm that firms suffering revenue declines lower employment while firms increasing employment tend to enjoy revenue growth. 35 This suggests that if employment subsidies are deemed desirable, they would be more effective if directed to private (including privatized) rather than state finns. 16 determined by their production costs, however irrational and uneconomic these may have been.36 As a result, most enterprises had bloated expenditures and failed to follow even elementary cost- saving production procedures. Given these obvious inefficiencies, their removal after the switch to a market regime tends to be a rather straightforward, more a matter of political will than entrepreneurial skill.37 Before turning to the evaluation of performance differences among different types of private owners, we would like to reiterate the average impact of privatization across all owners by illustrating the costs of delayed privatization. Figure 2 below illustrates the strength of revenue, employment and labor productivity privatization effects estimated from equation (1). The graphs show the relative performance of privatized and state firms with identical initial values of each performance measure. In each case, the vertical axis measures the ratio of the performance measure of a privatized firm to that of a state firm. The horizontal axis is set where the ratio equals 1, i.e., at the point where the privatization effect is 0. (Shaded areas mark the confidence intervals around the mean values.) In Section 5, we test these results for the presence of any such bias or moral hazard effects which could handicap state firms -- and find none. We also verified that the privatization effect is not limited to a particular country, industrial sector or a particular vintage of privatized firm.38 It is interesting to note that inclusion of country effects has no significant effect on either the sign or the magnitude of the privatization effect. The superior performance of privatized firms reflects the strength of the privatization effect alone: it is common to all three countries (despite some pronounced differences in their privatization programs) and it does not significantly depend on sectoral or temporal differences. It does, however, vary between various types of private owners; we turn now to this dependence. 36 Hungary had in theory abandoned the "cost plus" system of pricing during the period of its economnic reform in the 1980's. In fact, however, most pnces were controlled, and enterprises were able to bargain to make sure that they could cover their costs. Kornai (1990). 37 Cost are, to be sure, related to revenues. When the latter fall precipitously, it may be very difficult to lower the former quickly enough to make up for lost sales, given that a significant portion of the costs is fixed. But we do not have in our data sufficient information concerning fixed costs to be able to measure any lag in cost adjustment this may entail. In fact, in our discussion of costs, we are always referring to labor and material costs, exclusive of the cost of capital, depreciation, etc. Still, we do observe the persistence of the transition effect in the costs-per-unit-of-revenues over the first few years of the transition, which means that finns do not make immediate cost adjustments to their failing revenues. This may be due to two factors: (1) the reduction in most firm's employment figures lags somewhat behind revenue losses, especially at the beginning of the transition, when the firms have not yet learned that they cannot expect their losses to be temporary or to be made up by government assistance; (2) at the beginning of the transition, the prices of most material inputs are severely distorted and they are adjusted only over time. This means that the rising costs of a firmns' inputs percolate over time in a way that, except when substitution of imports in an option, is next to impossible for the firm to anticipate. 38 To avoid presenting an overwhelmingly large number of regression results, we do not report the results of these tests here. They are available from the authors upon request. 17 Figure 2 Relative performance of privatized and state firms REVENUE EMPLOYMENT REVENUE PER EMPLOYEE PERFORMANCE PERFORMANCE PERFORMANCE 2-22T :1 j~~~~~177 F , / 3 9 $.1 .. . , , ~~~~~~~..,...,.,...,.... -- - - - - - - flogo 1091 1902 199, logo 1th1 1992 1993 1990 1991 1992 1993 Privatzd f1. Ste firs 4. OWiNERSIH1P EFYECTS The privatization effect estimated from equation (1) is the average effects across all owners of privatized firms. Equation (2) allows its estimation for particular categories of private owners.39 However, because of a relatively small number of observations within some categories (see Table B.2, Appendix B), the estimates of some ownership effects -- even though of considerable magnitude -- are not precise enough to allow statistically defensible inferences. We therefore augrnent these results by estimating the ownership effects for all outsider-owned firms and all insider-owned firms taken together. Full regression statistics for all estimates are given in Tables A. 1 and A.2, Appendix A. Table 2 below provides the estimated privatization effects and indicates their significance levels. 39 Recall that in the context used here, "owner" means the largest shareholder, tae equivalency made possible by the high concentration of holdings in the sample. 18 TABLE 2 OWNERSHIP EFFECTS Privatization effect when the largest shareholder is Outsider Insider All outsider Foreign Prvatiza- Domestic individual State All insider Mag | Workers owners company tion fund nonfman. owner owners - ~~~~~~~~~company Revenue 18.30* 18.84* 27.09* 8.54** 16.61* 15.88* 14.51** 25.14* 4.65 Employment 4.01* 7.26* 0.96 3.99 3.10 1.42 10.59* 6.86* 13.76* Revenueiemployee 10.37* 8,.29** 16.25* 11.20* 19.32 17.61* -0.59 11.39 -7.76 Cost/revenue -5.37 . -3.16 -5.23 2.86 -18.07* -5.70** 2.17* 4.98 0.18 *p