Environmental Liability and Privatization in Central and Eastern Europe: Toward an Optimal Policy

Environmental and Resource Economics 17: 335–352, 2000. © 2000 Kluwer Academic Publishers. Printed in the Netherlands.

335

Environmental Liability and Privatization in Central and Eastern Europe: Toward an Optimal Policy RANDALL A. BLUFFSTONE1 and THEODORE PANAYOTOU2 1 University of Redlands and 2 International Environment Program, Harvard Institute for

International Development, Harvard University Received 27 May 1999; accepted 3 September 1999 Abstract. It has been estimated that the costs of remediating contaminated sites in Central and Eastern Europe (CEE) will be very high. These contingent environmental liabilities have emerged precisely at the time CEE countries are attempting to privatize their capital stocks, creating major challenges for privatization agencies. It is also generally agreed that an inadequate handling of these liabilities in the past has reduced sales of state-owned enterprises in Central and Eastern Europe. This paper uses an analytical model to identify optimal policies for minimizing the damage to privatization processes from environmental liabilities. Policy simulations are conducted which evaluate the effects of the use of liability indemnifications and environmental audits on privatization sales, prices and government revenues net of environmental costs. The major finding of the paper is that goals to mitigate the effects of environmental liability on privatizations and maximize government net revenues from privatization sales require equivalent environmental liability policies. Both goals require optimal provision of information and indemnifications to investors, and moreover it is found that the two policies are complementary rather than substitutes. Key words: Central and Eastern Europe, environmental liability JEL classification: Q28, P31

I. The Problem Since the countries of Central and Eastern Europe (CEE) experienced political and economic revolutions in the late 1980s and early 1990s, they have been faced with a variety of economic challenges simultaneously. Reforming and reorienting economies away from the priorities and methods of planning is undoubtedly the most important of these challenges, and privatization is the cornerstone of that reform process. Privatization is, of course, a priority for economies throughout the world, and indeed much of the world can be said to be in “transition.” The purpose of this paper is to discuss one area of the CEE privatization experience – the interaction between environmental and privatization issues – in the hope that this experience will prove useful elsewhere. Several privatization vehicles have been utilized in CEE countries, including direct sales to strategic investors, auctions, voucher sales and management/

336

RANDALL A. BLUFFSTONE AND THEODORE PANAYOTOU

employee buyouts. With more than five years of hindsight, however, it is becoming clear that the issues of corporate governance and control of insider behavior are crucial. Without assuring that incentives for efficiency are created, and that infusions of expertise and finance accompany transfers of ownership, the true goal of privatization – improved enterprise performance – may not be reached (Gray 1996; Frydman and Rapaczynski 1994). Actual sales of enterprises, rather than distributing them free-of-charge to citizens, is therefore considered to be a particularly important mechanism for accomplishing “true” privatization (Linderfalk 1997; Baldwin 1994). It is for this reason that our paper focuses only on this case. At the same time privatization is moving forward, countries are faced with often severe site contamination problems left over from previous regimes. As Goldenman et al. (1993) pointed out, these contamination problems are likely to be concentrated in particular industries such as chemicals, oil refining, smelting, waste oil processing, etc. Privatization and site cleanup are related, because toxic dumps can be anywhere and investors therefore face the risk that they are purchasing major liability problems when they participate in privatization markets.1 As Boyd (1996) correctly points out, the policy problem is to identify instruments that allow the identification and allocation of these site remediation costs in a way that minimizes distortions in privatizations, site remediation decisions and incentives for future safety. The remainder of this section discusses these three issues, though the primary focus will be on distortions in the privatization market itself. That environmental liability is a concern of investors, and particularly foreigners, has been known for a long time. In a study of 1000 large North American and West European firms, of those that actually evaluated sites in CEE countries, half rejected them partly on environmental grounds (The Economist 1993). Liability for past practices and inherited contamination risks then ranked as the most important environmental concern among surveyed firms (World Bank/OECD 1992). Though it is not known how much foreign investment has been lost due to environmental factors, anecdotal evidence from the region suggests that poor handling of possible environmental liabilities certainly complicates, and probably has deterred, foreign investment in CEE countries (Simons 1994). For example, early in the process of privatizing former East German enterprises, the German privatization agency, the Treuhandanstalt, was forced to address the problem. As Rotschke (1996) noted, “The Treuhandanstalt . . . had to carry out privatization in the new German lander under extreme time pressure . . . The problem of hazardous contaminations on industrial and commercial sites soon appeared as one of the obstacles to quick privatization.” Dodds and Wachter (1993) put it more prosaically at the start of their section on environmental issues. “Prospective purchasers looking at their first acquisition on the other side of the Wall may be inclined to panic when the issue of environmental damage first comes up . . . clearly there are concerns to address when buying an industrial facility that meets 1950s standards when 1990s rules apply.”

TOWARD AN OPTIMAL POLICY

337

In a number of cases government privatization agencies and investors were unable to conclude agreements or reached inferior arrangements because of environmental liability problems. For example, the Hungarian privatization agency has reportedly had difficulty selling the mineral oil processing and trading firm, Eptek, because of soil contaminated by oil during several years when the site was rented by a private company. Because of uncertainties around the assignment of liabilities, the privatization agency was also forced to substantially discount the $4 million estimated value of the firm (Baka 1995). In another case, the electronics firm, Phillips, Inc, negotiated a joint-venture with Czech Tesla Strasnice, but it refused to accept liability for soil contamination on the site. In the end Phillips chose to buy another building one hundred meters away from the initial site (Crosby 1992). Perhaps the most striking example of the potential importance of environmental liability is when an investor refuses to take a facility for free. Such a case indeed occurred when the German film manufacturer Agfa refused to accept restitution of its own previously confiscated plant because of suspected site contamination (Simons 1994). Making sure that environmental liabilities do not get in the way of privatization is, of course, not the only goal. Cleanup may also be necessary to avoid serious current and future risks to human health and the environment, and it is therefore recognized that optimizing and financing site remediation must be considered part-and-parcel of the environmental liability management problem. This process should, of course, allow more effort to be focused on some sites and less on others, and include the possibility that some projects are simply not worth the resources that must be expended on them. A particularly poignant example where such effort was not optimized occurred when the Swedish electronics firm Electrolux purchased the Hungarian firm, Lehel Refrigerator. When the company was sold, the privatization agency indemnified the new owner for all costs related to past pollution. It also placed 60% of the purchase price in escrow for use by Electrolux to clean up the site. No risk-based cleanup priorities were specified, however, and after some time the money was exhausted and Electrolux asked for more resources to complete the remediation work. The privatization agency refused, saying that the site was clean enough, and the matter had to be resolved in court (Baka 1995). Finally, liability allocations should not reduce incentives for future safety. For example, applying retroactive liability is problematic, because it creates potential claims against the assets of firms that may be on the verge of bankruptcy. If the financial situation is precarious enough, retroactive liability can therefore significantly reduce incentives for current safety, and redirect efforts into rent-seeking (Boyd 1993, 1996).2 To examine possibilities for addressing these problems, the next section discusses some instrument options. Section III presents a simple analytical model of the privatization process where privatization agencies can provide site specific environmental information (“audits”) to investors, indemnify investors against

338

RANDALL A. BLUFFSTONE AND THEODORE PANAYOTOU

environmental liability claims and also discount the purchase prices of enterprises to account for expected environmental liabilities. The main finding from this model is that from a privatization agency’s perspective, indemnification and price discount policies are equivalent. In Section IV the model is extended in what we consider to be critical ways, and we find that price discounts and indemnifications are no longer equivalent. It is also found that it is in the interest of privatization agencies to seemingly overinvest in environmental audits. In addition, it is found that indemnifications and information are complementary policies rather than substitutes. Section V concludes the paper and offers suggestions for future work.

II. Instrument Choice Uncertainty about future liabilities is just a risk tax on privatization markets, and anecdotal evidence suggests that particularly foreign investors quite heavily discount enterprises to account for that risk (Thomas 1994). The example of Eptek in Hungary is a case-in-point. Even though the former renter Kordax was legally responsible for cleanup, buyers still assumed the worst and were scared off. Perhaps the most powerful instrument for reducing this risk is information. As Thomas (1994) points out, defining cleanup requirements based on objective standards is the most basic and important information needed. To-date, however, such standards have not been adequately articulated anywhere in the region. Environmental audits are the primary tool used to define cleanup requirements, with the first step being a checklist and desk study to see if further investigation is warranted. If a problem is likely, sampling is conducted, followed by site remediation if required (Cowley 1996). Because of the use of strict liability, in North America and Western Europe audits are considered to be one of the important precursors to investment in industrial and even residential sites.3 While investing in information reduces risks to investors and allows privatization markets to work more efficiently, it is also only after some site analysis that privatization agencies can take control of cleanup decisions. The Treuhandanstalt and its successor agency, the Budesanstalt fur vereinigungsbedingte Sonderaufgaben used environmental audits particularly effectively to keep costs down. In the early 1990s cleanup costs in Eastern Germany were expected to run into the hundreds of billions of dollars (Lari 1992). As of 1996, privatization was virtually complete and only $6.4 billion had been spent on cleanup, largely because remediation activities were prioritized based exclusively on health risks.4 It must be noted, however, that environmental audits are imperfect tools and cannot be expected to accurately inventory all pollution. It is therefore a reality that investors and governments will probably have substantially less than perfect information when enterprises are privatized. Audits can also be expensive, with relatively preliminary investigations costing up to $100,000 per large industrial site in the US (Boyd 1993). Large, complex industrial operations may require as much as $500,000 for an initial audit, with an additional $1,000,000 necessary

TOWARD AN OPTIMAL POLICY

339

if extensive soil and ground water analyses are required (Goldenman et al. 1993). These costs of information cannot and should not be ignored by often cash-strapped CEE governments. Once estimated, environmental liabilities can either be transferred to investors or retained by privatization agencies. If the former strategy is chosen, reductions in enterprise sale prices to account for estimated cleanup costs, including a risk premium for the uncertainty involved, will have to be negotiated with investors. An alternative is for the government to retain the responsibility for past polluting activities. As Goldenman et al. (1993) discussed in detail, contractual instruments for doing this include indemnifications and releases. Indemnifications allow privatization agencies to reimburse investors for specific cleanup activities necessary to remediate sites to achieve well-defined goals.5 A release is a functionally similar instrument, except that the privatization agency is responsible for arranging site remediation. Whether indemnifications/releases or price discounts are used, it is clear that information is an important complementary policy. As the case of Lehel Refrigerator showed, this is particularly true when a privatization agency covers some or all of the cleanup costs, because there are significant possibilities for moral hazard when such insurance is given. Perhaps not the worst example is the tendency to blur “luxury” and “necessary” remediation measures (Zylicz and Lehoczki 1992). Even when price discounts are used, information remains of primary importance. For example, when the US firm, Powergen, began negotiations to purchase the Cspel Power Plant in Hungary, it was initially agreed that the firm would receive a $10 million price discount to cover site cleanup. After a detailed audit and risk analysis, however, it was determined that a sufficient level of cleanup defined by the government would cost approximately $1 million. Powergen received this level of discount along with a detailed list of required remediation steps, and the privatization agency saved itself $9 million.6 An important factor complicating the use of any of these tools is a heterogeneity of interests and an asymmetry of information between enterprise managers and the two transaction parties, privatization agencies and investors. Indeed, as argued by Frydman and Rapaczynski (1994), CEE governments cannot even be considered sole owners of state enterprises, because managers and workers also claim de facto and even de jure rights to firms. They point out that monitoring managers can be extremely difficult and therefore state enterprises are often run with very little supervision. In a prelude to a full chapter on insider behavior they note: “The incentives of people in control of the larger economic units are clearly at odds with the interest of state enterprises and the health of the economy as a whole. This state of affairs leads to all kinds of attempts at wild appropriation . . . ” (Frydman and Rapaczynski 1994, p. 117). For example, managers may want to keep enterprises in state hands to preserve their own positions or they may hope to purchase enterprises themselves or in conjunction with employees.7 They may also have concerns about investor quality

340

RANDALL A. BLUFFSTONE AND THEODORE PANAYOTOU

and intentions. As was pointed out by Goldenman (1997), during the first few years of the Polish privatization program the most common method of privatization was liquidation. With asset stripping – that may or may not be socially optimal – a common practice, it is reasonable that managers and employees would use any insider knowledge they have to fight privatizations. At the very least, employees and managers are likely to be faced with the prospect of enterprise restructuring and job losses when firms are privatized to foreigners. Managers may therefore hide or distort key information useful to privatization agencies and investors, making evaluations of enterprises more difficult. They may also use their often substantial influence to remove enterprises from privatization lists, meaning that some of the best enterprises may not even be in the privatization pool.8 III. A Simple Model of Privatization with Environmental Liabilities This section presents a simple model that examines the interaction between one investor and a government privatization agency negotiating the purchase and sale of a particular enterprise when on-site contamination and potential environmental liabilities are important. Privatization agencies in CEE countries are charged with selling enterprises and maximizing their net value, which is a goal that is rather distinct from the those of ministries of environment, which focus on environmental protection.9 Emphasizing this difference is important, because it helps us understand why site contamination issues can easily be ignored. The following assumptions underlie the model: BACKGROUND

ASSUMPTIONS

1. The investor in the model is a foreigner and therefore concerned about environmental liabilities; 2. The firm in question is “high quality” in that actual quality is above the average of all firms in the same industry in the privatization pool. If we define actual quality as q∗ , q∗ > qave q∗ ∈ (qmin , qmax ) 3. All information on the distribution of enterprise qualities within particular industries is common knowledge; 4. The costs of cleaning up contaminated sites are imperfectly observed. The investor is risk averse and therefore discounts the firm depending on the degree of risk perceived.

TOWARD AN OPTIMAL POLICY

R ESTRICTIONS

341

ON BEHAVIOR

5. The privatization agency can invest in information by conducting an environmental audit. Information gathering helps the agency prioritize cleanup options and therefore reduces costs. Environmental auditing also reduces investors’ risk and therefore their risk premia. The investor does not conduct environmental audits;10 6. The privatization agency can take responsibility for any or all site cleanup costs, thus indemnifying the investor; 7. Privatization provides an opportunity for site cleanup that would not otherwise exist, and the privatization agency does not incur cleanup costs if it does not sell an enterprise. This assumption can be thought of as the privatization agency needing the quasi-rents from the sale of the firm for site cleanup;11 8. Investors maximize the expected present discounted value of profits, but investor rents are eliminated during negotiations with the privatization agency. 9. The privatization agency maximizes net revenues to the treasury after covering cleanup and information costs.12 Assuming that the firm in question is the investor’s best investment option, s/he is willing to buy the firm if the present discounted value of expected profits is nonnegative. If 5 is that value, P is the price paid for the enterprise, C is the present discounted value of cleanup costs, m is the fraction of those costs indemnified by the privatization agency,13 and π is the present discounted value of profits from ongoing operations (i.e. revenues minus economic costs), this implies the following: E(5) ≥ 0 iff P ≤ E[π − (1 − m)C]

(1)

In this model only C is uncertain, meaning that the condition for purchase can be expressed as P ≤ π − (1 − m)E(C). The expectation of C, including a risk premium, is assumed to depend on the amount of information (i) provided by the privatization agency, proxied by expenditures on environmental audits, and the distribution of qualities of firms in a particular industry (e.g. basic chemicals, tanneries, etc.) in the privatization pool. This distribution is summarized by the mean quality, which is the same sufficient statistic Akerlof (1970) used in his “lemons” example which focused on the used car market with imperfect information. Because of significant differences across industries and the questionable utility of comparing refineries and flower shops, average qualities are industry specific and not directly comparable across industries. E(C) is now defined as C(i, qave ). i ≥ 0, ∂C < 0, ∂i ∂C < 0, ∂qave

qave > 0 ∂ 2C >0 ∂ 2i ∂ 2C >0 ∂ 2 qave

(2)

342

RANDALL A. BLUFFSTONE AND THEODORE PANAYOTOU

Diminishing returns to information and improvements in average quality are imposed on C(i, qave ). This assumption simply takes into account the likelihood that reductions in costs will be less than proportionate to increases in information provided on a particular firm, and the same will be true for improvements in average quality in a firm’s industry. These features are summarized in (2). The privatization agency knows the investor’s inverse demand function and as noted in assumptions 3 and 4, as much about the site in question and environmental policy as the investor. The privatization agency also has a relatively limited area of focus, and its optimization problem therefore boils down to a maximization of the difference between the sale price of the enterprise, the present value of indemnified cleanup costs and expenditures on information. In equation 3, N is the present value of this difference and Pi is the price of a unit of environmental audit. It maximizes this present value of net revenues from privatization by choosing levels of indemnification (m) and information (i) optimally. This maximization is subject to the investor participation constraint given in (1) and the privatization agency’s own break even constraint. The privatization agency therefore does not sell any enterprise at a price less than indemnified cleanup costs.14 N = P − mC(qave , i) − Pi i s.t. P = 5 − (1 − m)C(qave , i) s.t. P ≥ mC(qave , i)

(3)

If we substitute the first constraint into the objective function, we get the following: N = 5 − C(i, qave ) − Pi i

(4)

We therefore find that m drops out, implying that the degree of indemnification does not matter when information is symmetric. The intuition behind this result is straightforward. Whether investors are indemnified by the privatization agency or reduce their reservation prices to cover expected cleanup costs is immaterial from the privatization agency’s perspective. In either case the agency picks up the tab, because a government cannot unilaterally determine the price of an enterprise and force an investor to buy it too. This result leaves only i as a choice variable. Maximizing (4) with respect to i yields the standard first order condition that investments in information should be made until the marginal benefit equals the marginal cost (i.e. until ∂C/∂i = Pi ), but this result can only potentially offer something new if we apply an explicit form for C(i, qave ) that reflects the reality of the relationship between C, i and Q. The form chosen should fulfill the constraints imposed in Equation 2 and be bounded. Ideally, the function would also have the following features: ∂C/∂i > ∂C/∂qave , and ∂ 2 C/∂qave ∂i > ∂ 2 C/∂i∂qave .

TOWARD AN OPTIMAL POLICY

343

The idea underlying these additional constraints is that information about a particular firm under discussion is a more powerful risk-reducing and prioritization tool than news that firms in a particular industry are, on the whole, not as bad as was originally thought. The second derivative condition notes that any improvement in quality is probably more effective if better information is given, but improvements in the overall privatization pool probably do not make the provision of information more effective. One function that combines these features reasonably well is C = a + b∗q−i ave , where a and b are technological parameters related to the way increases in qave and i translate into cost reductions. Substituting this function into (4) and maximizing with respect to i gives the first order condition in (5): bq−i ave ∗ lnqave − Pi = 0

(5)

Solving this equation for i yields the following reduced form solution. i=

lnb + ln(lnqave ) − lnPi lnqave

(6)

If we simulate (6) for changes in qave we find that the choice of i is virtually unaffected by changes in qave . The only shifters of this information demand function are therefore the technical parameters a and b. Thus, the demand for information by the privatization agency depends wholly on the price of information and technology, but is unaffected by whether the average quality in a particular industry is high or low. With regard to particular industries, this result suggests a very simple rule of thumb for optimizing investments in information; regardless of whether most firms are clean or dirty, the degree of information purchased to clarify liabilities and calm investors’ nerves should depend on its cost and the technological effectiveness of that information.15 IV. Extending the Model The simple model is now extended to include the possibility that qave is probably not fixed and indeed should be considered endogenous. The idea that qave is endogenous is closely related to the following question: “With quality uncertain, will the average quality of firms in a particular industry be the same near the end of the privatization process as it was at the start?” Furthermore, we can ask whether there will be a systematic evolution of quality during the privatization process. There are at least two reasons why the quality of the privatization pool in a particular industry should deteriorate as privatization progresses, implying an inverse relationship between privatizations and average quality. First, privatization agencies are partly judged by the degree to which they can actually get enterprises out the door, particularly through sales to West Europeans and other foreigners who can bring expertise and financial resources. These agencies typically want to promote direct foreign investment and worry about their programs’ images. To the extent possible, privatization agencies therefore often try to prime markets by

344

RANDALL A. BLUFFSTONE AND THEODORE PANAYOTOU

focusing early promotion efforts on obviously best enterprises.16 In other words, reputation matters in the world of direct foreign investment, and therefore the probability of selling off firms in an industry is probably a positive function of past privatization activity. As these highest quality firms are sold off, the average quality of the remaining privatization pool, of course, declines. The second reason qave should fall as privatization progresses is adverse selection by managers and employees. We know that this group can be a powerful political force17 and that it possesses better knowledge about on-site conditions than either of the transaction parties. In earlier joint work with Vladislav Balaban (Balaban et al. 1994), it was shown that when quality is imperfectly observable any above-average enterprise will be under-valued by the market. Managers and employees will, of course, know their firms are undervalued by investors, and it seems reasonable to suppose that in some cases they will worry about investor quality and intentions. As was already noted, for example, asset stripping by buyers is more than just a possibility. Indeed, in Poland during the first years of economic transition the major form of privatization was the liquidation of assets (Goldenman 1997). Depending on individual firms’ circumstances, these groups may strongly support privatization as a way to stay afloat or they may put up barriers to privatization. In either case they may use their superior information and influence to advantage, either by pushing forward poor quality firms that are going nowhere anyway or by trying to take undervalued enterprises off the market. With quality imperfectly observable, privatization agencies have to reduce prices to sell more enterprises (Balaban et al. 1994). Above average enterprises therefore become more undervalued as privatization progresses, increasing incentives to remove high quality enterprises from the market.18 A model in which privatization agencies attempt to sell the best quality enterprises first is not consistent with one where alliances of managers and workers try to pull those same firms off the market. These are potentially countervailing forces, and which effect dominates is a political economy and empirical issue. The bottom line for our purposes here is that both effects point to the same relationship between the average quality in an industry and the extent of privatizations. Both effects cause average quality to decline as privatization progresses, resulting in a relationship like that presented in Figure 1. The important question is how will the results be altered once this feature is incorporated, and what inferences can be drawn for privatization policy. To examine this issue, we need to look at the number of enterprises transacted in a particular industry rather than just the marginal firm. As shown in Figure 1, there is also a one-to-one relationship between average quality and the number of transactions in an industry. Choosing a level of privatizations is therefore equivalent to choosing qave . We will therefore talk in terms of qave as a choice variable. C(qave , i) in this case is defined as the expected average cleanup per firm purchased. Equation 1 is altered to reflect the fact that investors who purchase

345

TOWARD AN OPTIMAL POLICY

Figure 1.

a firm have an effect on the overall quality of the privatization pool, and therefore drive up the expected average cleanup cost when an additional enterprise is purchased. The altered optimal purchase condition is given in Equation 7. P ≤ 5 − (1 − m)(C(qqve , i) +

∂C ∗ qave ∂qave

(7)

The privatization agency again takes this inverse demand as given and maximizes its expected revenues net of cleanup costs. Its choice variables are its supply of firms (i.e. its level of qave ), the percentage of cleanup costs that will be indemnified (m), and the level of information (i) the privatization agency will provide on average for each firm being considered. Substituting (7) into (3) and simplifying gives (8). Of immediate note is that the variable m no longer drops out as in the simple model, implying that indemnifications and price discounts are no longer equivalent. N = C(qave , i) − (1 − m)

∂C ∗ qave − Pi i ∂qave

(8)

Maximizing (8) with respect to the three choice variables gives three first order conditions (FOCs). These equations are given in (9). ∂ 2C ∂C −(1 − m) ∗ qave − − Pi = 0 ∂qave ∂i ∂i  2  ∂ C ∂C ∂C −(1 − m) − ∗ qave + = 0 ∂qave ∂qave ∂qave ∂C qave = 0 ∂qave

(9)

346

RANDALL A. BLUFFSTONE AND THEODORE PANAYOTOU

The last equation, which is the derivative of (8) with respect to m, does not satisfy the first derivative test for a maximum and is therefore a minimum. This means that we cannot get a reduced form solution, because we have too few equations. The equation is still of interest, however, because it tells us that a strategy that drives qave to zero (and therefore maximizes privatizations) will not be an optimum. N is also minimized when ∂C/∂qave equals zero, which we know from (2) occurs only at very high levels of average quality (i.e. at very low levels of privatizations). Maximizing or minimizing the level of privatizations is therefore expected to be suboptimal, implying that very low quality enterprises should go unsold. We now examine the other two FOCs to better understand the nature of an optimal liability policy. Solving the second FOC for the variable m, we look at the condition for m to be 1 (full indemnification). It turns out that the term ∂C/∂qave must again equal zero, which we know from the last equation in (9) is not an optimal policy. Full indemnification is therefore not optimal. What about a policy of no indemnifications, where liability is addressed solely through price adjustments? When is such a policy optimal? After a little manipulation we find that m = 0 only if the following condition holds: ∂C 1 ∂ 2C =− 2 ∂qave 2 ∂ qave

(10)

Can this condition hold? The answer is yes, but given the functional forms believed to represent C(qave , i), only at very low levels of quality (i.e. when the privatization process has progressed very far or the pool is simply of very bad quality) will the right hand side ever be double the left in absolute value. At that point a policy of zero indemnifications will be optimal, because the cleanup costs for those low quality properties will simply be too high. Those firms will either be sold without indemnification, continue to be run by the privatization agency or will be abandoned. For most situations, the optimal policy will be to have a policy which offers investors a less than complete, non-zero level of indemnification. This result occurs when the left hand side of (10) is greater than the right hand side. Figure 2 graphically shows the menu of equilibrium relationships between m and qave . The first FOC in (9) sheds additional light on the question of optimal indemnification, and its relationship with information policy. Solving the equation for m yields the following: ∂C + Pi ∂i m= +1 ∂C ∗ qave (∂qave ∂i)

(11)

Given the assumptions in (2) and also those embedded in an explicit form such as (5), this equation indicates that providing more information (i.e. decreasing the

TOWARD AN OPTIMAL POLICY

347

Figure 2.

absolute value of ∂C/∂i) is consistent with higher equilibrium levels of indemnification. Information provision and indemnification are therefore complementary policies in this model, with full indemnification occurring when the numerator of the first term in (11) equals zero. Of course, this is the maximum economically feasible amount of information, and is also the condition for optimal investment in information that resulted from the simple model. That complementarity exists is intuitive, because more information shifts the expected cleanup function to the left, implying that there is less cleanup to indemnify. The complementarity of information and indemnifications in this model is summarized in Figure 3. But how does the equilibrium level of investment in information compare with that in the simple model? To answer this question we solve each of the first two FOCs in (9) for the variable qave and set these equations equal to each other. We then solve for ∂C/∂i + Pi , which we know equaled zero in the simple model. As shown in (12), ∂C/∂i + Pi is now unambiguously positive, meaning that in absolute value Pi > ∂C/∂i. There is therefore a greater investment in information than in the simple model. ∂ 2C ∂C (2 − m) ∂C ∂qave ∂qave ∂i + Pi = ∂i ∂ 2C (m − 1) ∂qave

(12)

This result occurs, because there is an additional cost present in this more complex model, namely that for a given level of information, the sale of one firm makes it harder to sell the next firm. All else equal, therefore, there exists

348

RANDALL A. BLUFFSTONE AND THEODORE PANAYOTOU

Figure 3.

an increased incentive for the privatization agency to invest in information. Mathematically, we see that everything depends on ∂ 2 C/∂qave ∂i. This term is nonnegative, but if it is zero the equilibrium devolves to the same one as in the simple model, meaning that C/∂i + Pi = 0. A lack of separability of qave and i in the function C(qave , i) is therefore of importance for getting this result, implying that information must change investor perceptions of the functional relationship between the cost of cleanup and the average quality of firms in an industry. This is, of course, the whole point of providing information, suggesting that it is perhaps reasonable to advocate what may at first glance seem like an excessive investment in information. A lack of separability is not, however, crucial for the existence of an equilibrium value of the variable m. If an explicit form for C(qave , i) is used that is separable in qave and i, the level of indemnification still does not drop out. This result suggests that it is the endogeneity of qave that eliminates the symmetry between price discounts and indemnification.

V. Conclusions and Directions for Further Research This paper presented a modest attempt to unravel some of the policy issues associated with large-scale privatization and environmental liability. Our results confirm the conventional wisdom that policy makers should recognize that past environmental mistakes are sunk costs that must not be allowed to distort current privatization efforts. Indeed, in the simple model investors just reduced their equilibrium offers to account for the lack of indemnification.

TOWARD AN OPTIMAL POLICY

349

We then incorporated the notion that the privatization pool is likely to change as privatization progresses. Several results were altered in the move to the extended model, suggesting that privatization agencies would do well to set policies to account for this reality. First, the symmetry of price reductions and indemnifications disappeared, with the optimal indemnification percentage generally being non-zero, but not 100%. Second, if certain assumptions hold, as shown in (12), a greater investment in information is optimal when privatizations and average quality are related. This result occurs, because better information reduces investors’ risks and therefore increases their willingness-to-pay, creating an incentive for privatization agencies to invest more in information than the simple model would suggest. Finally, as suggested by (11), the model indicates that information provision and indemnification are complementary policies, not substitutes as one might expect. This is because in an environment of imperfect information it is the ex ante liability perception on the part of both the privatization agency and the investors that matters. Offering information therefore reduces investors’ risk and provides privatization agencies with an opportunity to better define costs. Ceteris paribus, this better definition reduces the privatization agency’s expected burden, and it is therefore willing to fund a larger percentage of the cleanup. This paper has generated several testable hypotheses that point to the need for empirical work. First, to date there has been virtually no empirical research which has attempted to confirm or refute some of the presumptions common in the region regarding the effect of differential environmental liability policies on privatization markets. For example, a basic untested conventional wisdom exists – reinforced by our results – that information and indemnification are empirically important at all. It should also be possible to test assumptions and predictions of this paper and to estimate the effects of the policy instruments analyzed. For example, the prediction that indemnifications and price reductions are not symmetrical should be tested. With sufficient data it should also be possible to place a marginal value on environmental audit expenditures by CEE governments. Acknowledgement The authors would like to thank two anonymous reviewers for their helpful comments, as well as participants in various conferences where earlier versions of this paper were presented. Notes 1. Liabilities are defined as payments when the property rights of others are violated. In the case considered here, most liabilities are “contingent” in that a government agency or a third party must take action against the violator before the liability results (Goldenman 1997). 2. See Goldenman et al. (1993) and Larson (1995) for more in-depth discussions of the importance of future environmental performance.

350

RANDALL A. BLUFFSTONE AND THEODORE PANAYOTOU

3. For example see the following commercial web sites focusing on environmental due diligence in the UK and the US respectively. http://www.avnet.co.uk/contaminated/#DDE_LINK1 and http://www.ceo-online.com/kogut/art1.htm 4. The Treuhandanstalt has also been criticized for focusing too much on costs and too often choosing containment rather than site remediation measures (Goldenman 1995). 5. As discussed by Boyd (1996), on the financing side environmental escrow accounts have been extensively used in the region. National liability pools are more efficient mechanisms, because funds are less narrowly earmarked. 6. As discussed by Boyd (1996), however, exclusive reliance on price discounts is undesirable because they do not clearly specify the level of purchaser liability. As such, they invite adverse selection. 7. As Murphy and Foushee (1995) note, in Lithuania most privatizations of large state-owned enterprises have indeed been management buyouts. In Latvia, managers are able to delay privatizations by leasing enterprises for five years. With such tools available, it is not surprising that as of 1995 Latvia had virtually no hard currency privatizations. In Poland, the law on privatization allows employees to purchase up to 20% of their employing companies at half price (Frydman and Rapaczynski 1994, p. 115). 8. For example, in 1994 a deal to sell the largest hotel in Vilnius, Lithuania to an American hotel chain was scuttled for exactly this reason (Baldwin 1994). 9. We would like to thank an anonymous reviewer for helping us clarify this point. 10. Though investors often will conduct their own audits, this assumption is imposed to focus on privatization agency policies and actions. 11. Indeed, the only places where significant sight remediation has occurred is where sale proceeds have been specifically allocated for cleanup. Particularly if an enterprise proves to be uneconomic and is closed, cleanup is likely to be deferred far enough into the future that it can be viewed as “indefinite.” 12. In most or perhaps all countries in the region there are no policies that would define explicit requirements for remediation of particular sites or even particular industries. Requirements are therefore typically very vague. 13. This is the most common way indemnification is structured. In Germany, for example, the policy was for the privatization agency to cover 90% of all cleanup costs (Goldenman 1995). 14. It is assumed for simplicity – without altering the results – that the opportunity cost of selling the enterprise is zero. The second constraint therefore defines the participation constraint. 15. Of course, investment in information should vary systematically across industries, with dirtier industries receiving more attention. 16. A case in point occurred in Lithuania in May, 1993 when Klaipeda Tobacco, Inc. was sold to Philip Morris, Inc. for a total package valued at $42 million. This gave Philip Morris a valuable foothold in the Baltic region located in an important port city (Murphy and Foushee 1995). 17. For example, until recently the former communists formed the government in Lithuania. This party was strongly supported by the national council of trade unions, which has called for the right to approve all privatizations, among other conditions (Baltic News Service, 12/17/97). 18. We also found that adverse selection results in an inefficient level of privatizations and a poorer quality privatization pool than would have otherwise occurred.

References Akerlof, A. G. (1970), ‘The Market for “Lemons”: Quality Uncertainty and the Market Mechanism’, Quarterly Journal of Economics (August) 84.

TOWARD AN OPTIMAL POLICY

351

Baka, Eva (1995), Implementation of Environmental Assessment and Performance Audits at the Hungarian State Property Agency. Final Report to the Harvard Institute for International Development. Balaban, Vladislav, Randall A. Bluffstone and Theodore Panayotou (1994), ‘Lemons and Liabilities: Privatization, Foreign Investment and Environmental Liability in Central and Eastern Europe’, Environmental Impact Assessment Review (May). Baldwin, Andrew (1995), EU-Phare Privatization Project, Vilnius, Lithuania, Private Communication, August 17. Baldwin, Andrew (1994), Enterprise Privatization Progress in the Baltic States – Phare Assistance 1993/94. EU Privatization Project, Tallinn, July. Baltic News Service (1997), ‘Trade Unions Demand Inclusion in the Privatization Process’, Vilnius, 17 December. Boyd, James (1996), ‘Environmental Liability Reform and Privatization in Central and Eastern Europe’, European Journal of Law and Economics 3(1). Boyd, James (1993), Issues and Proposals for Environmental Liability Reforms in Central and Eastern Europe. Resources for the Future Working Paper ENR 93-09, January. Cowley, Christine (1996), ‘Cost-Benefit Issues in Clean-Up Decisions’, Paper Prepared for the third OECD/Phare Workshop to Address Environmental Issues in Privatization and Economic Reconstruction. Budapest, November. Crosby, Alan (1992), ‘Skoda’s Board of Directors Resigns’, The Prague Post, October 13–19. Dodds, P. and G. Wachter (1993), ‘Privatization Contracts with the Treuhandanstalt: and Insiders’ Guide’, The International Lawyer (Spring) 27(1). The Economist (1993), ‘Environmental Liabilities: Dirty Legacy’, September 18. Frydman, Roman and Andrzej Rapaczynski (1994), Privatization in Eastern Europe: Is the State Withering Away? London: Central European University Press. Goldenman, Gretta (1997), “Privatization and Environmental Liability in Central and Eastern European Countries and the NIS: Reviewing Progress Since the 1992 Warsaw Conference’, Prepared for the OECD Experts’ Workshop on Privatization and the Environment in Central and Eastern Europe, September. Goldenman, Gretta (1995), ‘Legal, Economic and Environmental Dimensions of Privatization of Industrial Enterprises in Central and Eastern Europe’, Prepared for the Conference on Environmental Aspects of Privatization in the Period of Economic Transition, Organized by the Government of Poland and the Harvard Institute for international Development, April. Goldenman, Gretta et al. (1993), Environmental Liability and Privatization in Central and Eastern Europe. A Report for the Environmental Action Program for Central and Eastern Europe, The World Bank, April. Gray, C.W. (1996), ‘In Search of Owners: Privatization and Corporate Governance in Transition Economies’, The World Bank Research Observer (August) 11(2). Lari, Eugenio (1992), Statement by the Director for Europe and Central Asia of the World Bank. International Conference on Privatization, Foreign Direct Investment and Environmental Liability, May. Warsaw, Poland. Larson, Bruce, ed. (1995), The Harvard Institute for International Development. Proceedings of the Seminar “Environmental Liability Reform in Transition Economies,” Riga, February 27– March 1. Linderfalk, Bjorn (1997), ‘Baltic Privatization’, The Baltic Times, April 3–9, 15. Menell, Peter (1992), ‘The Limitations of Legal Institutions for Addressing Environmental Risks’, Journal of Economic Perspectives (Summer). Murphy, Brian and Scott Foushee (1995), ‘Baltic States: Challenges and Opportunities’, in Privatization Yearbook 1995. London: Privatization International Ltd. Rotschke, H. (1996), ‘Treuhandanstalt’s “Quick Assessment” Method for Estimating Cleanup Costs for past Environmental Damage in the New Lander’, Paper Prepared for the third OECD/Phare

352

RANDALL A. BLUFFSTONE AND THEODORE PANAYOTOU

Workshop to Address Environmental Issues in Privatization and Economic Reconstruction. Budapest, November. Simons, Marlise (1994), ‘Pollution Blights Investment, Too, in East Europe’, The New York Times, May 13. Thomas, Randall (1994), ‘The Impact of Environmental Liabilities on Privatization in Central and Eastern Europe: A Case Study of Poland’, University of California at Davis Law Review 28(1), 165–217. World Bank/OECD (1992), Corporate Survey on Western Direct Investment and Environmental Issues in Central and Eastern Europe. World Bank. Zylicz, Tomasz and Zsuzsa Lehoczki (1993), Towards Environmental Recovery: The Czech Republic, Hungary, Poland and Slovakia. Warsaw University Mimeo.