Impact of Privatization on Financial & Operating Performance of Developing Country SOEs, Apuntes de Finanzas

¡Descarga Impact of Privatization on Financial & Operating Performance of Developing Country SOEs y más Apuntes en PDF de Finanzas solo en Docsity! The Financial and Operating Performance of Newly Privatized Firms: Evidence from Developing Countries NARJESS BOUBAKRI and JEAN-CLAUDE COSSET* ABSTRACT This paper examines the change in the financial and operating performance of 79 companies from 21 developing countries that experienced full or partial privatiza- tion during the period from 1980 to 1992. We use accounting performance mea- sures adjusted for market effects in addition to unadjusted accounting performance measures. Both unadjusted and market-adjusted results show significant in- creases in profitability, operating efficiency, capital investment spending, output, employment level, and dividends. We also find a decline in leverage following pri- vatization but this change is significant only for unadjusted leverage ratios. Our results are generally robust when we partition our data into various subsamples. PRIVATIZATION HAS TURNED INTO a major worldwide phenomenon, in both de- veloped and developing countries. Over the last decade, privatization of state- owned enterprises ~SOEs! has been occurring at an increasing rate, particularly in developing countries. The share of these countries in global privatization revenues has actually risen from 17 percent in 1990 to 22 percent in 1996 ~The Economist, March 22, 1997!. This paper examines how privatization in developing countries affects the financial and operating performance of the former SOEs. This issue is important because developed and developing coun- tries are not equally endowed with those factors likely to inf luence the suc- cess of a privatization program. Indeed, the privatization efforts of most developing countries are inhibited by embryonic financial markets, a weak * The authors are with the Faculty of Administrative Sciences, Université Laval, Québec. We gratefully acknowledge the financial support of the Programme d’analyses et de recherches économiques appliquées au développement international ~PARADI!, a Center for Excellence funded by the Canadian International Development Agency ~CIDA!. Many government agen- cies, embassies ~particularly, the Canadian embassies!, and companies around the world de- serve our gratitude for their assistance in collecting the data. We would like to thank Benoît Boyer, John Cockburn, Lucie Courteau, Hubert de La Bruslerie, Bernard Decaluwé, Klaus Fis- cher, Jean-Marie Gagnon, Campbell Harvey, Edward Kane, Jocelyn Martel, Seyed Mehdian, John Norsworthy, René Stulz ~the editor!, Jean-Marc Suret, Adrian Tschoegl, Luc Vallée, and an anonymous referee for valuable comments and suggestions. We have also benefited from the comments of seminar participants at the University of Fribourg ~Switzerland!, the University of Ottawa, Université Laval, the 1995 PARADI Conference, the 1996 Academy of International Business meeting, the 1996 Financial Management Association meeting, and the 1997 French Finance Association international meeting. Finally, we are indebted to Patrick Chamberland, Jacqueline Chavez, and Josée Maltais for excellent research assistance. THE JOURNAL OF FINANCE • VOL LIII, NO. 3 • JUNE 1998 1081 regulatory capacity, and a public sector that accounts for a large share of the GDP.1 Moreover, many of these countries, particularly those with relatively low per capita income, lack some of the main ingredients for a successful privatization, such as competent managers, entrepreneurs, and capital ~Vernon- Wortzel and Wortzel ~1989!!. On the other hand, some of these countries have large markets and a fast rate of economic growth, which makes the success of government divestiture more likely ~Galal et al. ~1994!!. Analysis of the privatization phenomenon in a wide set of developing countries should allow us to determine whether privatization is beneficial in the economic environments and the institutional settings specific to these countries. Few empirical studies have examined the privatization phenomenon in developing countries. The existing empirical literature has focused on the experience of developed countries with the notable exceptions of Galal et al. ~1994! and Megginson, Nash, and van Randenborgh ~1994; henceforth MNR!.2 In a World Bank study, Galal et al. ~1994! assess the welfare gains or losses resulting from the privatization of 12 companies operating mostly in non- competitive markets and in four countries: Chile, Malaysia, Mexico, and the United Kingdom. The authors report net welfare gains in 11 of the 12 cases, finding no cases in which workers show an overall loss from privatization. However, as pointed out by the authors, their sample is small and unrepre- sentative of the universe of privatized firms in developing countries, and any generalization of their evidence should be avoided. A major study by MNR compares the pre- and postprivatization financial and operating performance of 61 firms from 18 countries ~12 industrialized and 6 developing! and 32 industries for the period from 1961 to 1990. The authors present strong evidence that, following privatization, their sample firms become more profitable, increase their real sales and investment spend- ing, and improve their operating efficiency. Furthermore, these companies significantly lower their debt levels and increase dividend payments. Per- haps more surprisingly, they find that employment levels increase after pri- vatization. Their results are generally unchanged when they partition the data into smaller subsamples. As impressive as MNR’s sample of privatized firms is, it includes only a small number of firms headquartered in devel- oping countries ~from 3 to 12, depending on the financial and operating per- formance measure!. Therefore, the sample size is too small to be considered as indicative of the actual experience of privatizations in developing countries. 1 Although the total privatization proceeds in developing countries increased from $2.6 bil- lion in 1988 to $24.5 billion in 1993, note that the percentage share of SOEs in GDP for these countries has remained constant at around 11 percent since 1978 ~The World Bank, 1995!. 2 We limit our review of the empirical literature to these two important studies because they are not restricted to a single country. A discussion of the early empirical literature on the relative performance of state-owned versus privately owned firms appears in Boardman and Vining ~1989! and Vickers and Yarrow ~1991!. More recent single-country studies appear in Menyah and Paudyal ~1996! for the United Kingdom and in La Porta and López-de-Silanes ~1997! for Mexico. Readers are also referred to Dewenter and Malatesta ~1997! and Jones et al. ~1996! for recent studies of the determinants of privatization initial returns. 1082 The Journal of Finance Of the 203 newly privatized companies, many could not be included for the following reasons: they were privatized after 1992 and no postprivatization data were available ~23 companies!; the pre- and postdivestment data for some companies privatized through direct sale to another company were not comparable because only consolidated financial statements were available for the postprivatization period ~2 companies!; data were missing for at least one year of the ~22, 12! window ~68 companies!; finally, failure to comply with our request for annual reports ~31 companies!. Our final sample consists of 79 newly privatized firms from 21 countries that are among the most active privatizers in the developing world. Of these 79 companies, 34 answered our questionnaire. For the remaining 45 firms, we gathered as much information as possible from annual reports, prospec- tuses, and secondary sources. The Appendix provides the following descrip- tive information on these companies: the type of industry in which each firm operates, the date of government sale, the number of employees in the year of privatization, the value of each issue in U.S. dollars, and the government shareholdings before and after privatization. The sample is well diversified, exhibiting a wide geographical dispersion and different levels of country de- velopment. The firms also belong to different types of industries and market structures ~competitive and noncompetitive! and vary in size. II. Methodology The analysis conducted in this paper seeks to determine whether the pri- vatization of SOEs in developing countries is truly desirable and lives up to the expectations of governments and development agencies for the perfor- mance of newly privatized firms. In particular, the study tries to determine whether, following privatization, the firms increase ~1! their profitability, ~2! their operating efficiency, ~3! their capital expenditures, and ~4! their out- put. We also examine the impact of privatization on employment levels, cap- ital structure, and dividend policies. We use the same ratios as in MNR to allow a constructive comparison be- tween their sample of mostly developed countries and our sample of devel- oping countries. We first compute empirical proxies for each company for a seven-year period ~from three years before to three years after privatization!. We then compute means of each variable for the preprivatization ~years 23 to 21! and postprivatization ~years 11 to 13! periods. However, it is important to note that we included companies in our sample as long as we had observa- tions from at least year 22 to year 12. Furthermore, the year of the privat- ization, year 0, is excluded from the analysis because it includes both the public and private ownership phases of the firm. The date of privatization is the date on which the government sold, for the first time, a certain amount of shares. We then use the two-tailed Wilcoxon signed-rank test to test for significant changes in the variables, as well as a proportion test to determine whether the proportion ~ p! of firms experiencing a change in a given direction is greater than what would be expected by chance, typically testing whether p 5 0.5. Performance of Newly Privatized Firms 1085 Because most of the privatizations covered in this study took place over a relatively short period of time ~1989–1992!, the sample firms’ performance is likely to be inf luenced by economy-wide changes. To take account of this possibility and to isolate the effect of privatization on the financial and op- erating performance of SOEs from other determinants of this performance, we also use market-adjusted accounting performance measures. Firms op- erating in the financial and real sectors exhibit different operating and fi- nancial profiles. Consequently, for banks and insurance companies, we use a market index comprising financial corporations. For other firms, the market index consists of nonfinancial corporations. We estimate the market index for each year as the median ratio for all financial or nonfinancial firms in the country on the Disclosure databases ~excluding all privatized firms, whether in our sample or not!. Hence, the market-adjusted accounting per- formance measure of a firm is the difference between its accounting perfor- mance measure and the market median accounting performance measure of its country.4 III. Empirical Results In the following sections, we present and discuss the empirical results using unadjusted performance measures for the whole sample of 79 privat- ized firms, and market-adjusted performance measures for the 39 privatized firms for which market-adjusted ratios could be computed ~Tables I and II!.5 At the end of each section, we also discuss the market-adjusted and un- adjusted results for the following partitionings of our data. First, we deter- mine whether the effect of privatization varies according to the prevailing market structure, since competition can greatly improve monitoring possi- bilities and hence incentives for productive efficiency ~Vickers and Yarrow ~1991!!. To this end, we partition the full sample into competitive and non- competitive firms. Second, to determine whether postprivatization perfor- mance varies with the level of economic development, we partition our sample of developing countries into two subsamples: upper-middle-income countries and low-income and lower-middle-income countries. Kikeri, Nellis, and Shir- ley ~1992! argue that a market-friendly policy framework and a relatively well-developed regulatory policy, two factors prone to favor the success of privatization, are correlated with income. Privatization in high- or middle- 4 We also intended to consider an industry performance benchmark. However, we were un- able to perform industry adjustments for more than 13 firms, mainly because several sample firms were monopolies and only a limited number of firms from developing countries were available in the databases. Therefore, we do not report results with this alternative perfor- mance benchmark. However, our analysis of this sample of 13 firms reveals that the results with market-adjusted and industry-adjusted performance measures are similar. For a thorough study of the impact of British Airways’ privatization on the airline industry, see Eckel, Eckel, and Singal ~1997!. 5 To assess possible differences in the findings across samples, Table II reports market- adjusted and unadjusted results for the “market-adjustment” sample of 39 companies. We ob- tain very similar unadjusted results using the entire sample of 79 companies ~Table I! and the subsample of 39 companies ~Table II!. 1086 The Journal of Finance income countries should therefore yield greater benefits. Third, we contrast the postdivestiture performance of totally and partially privatized firms. In this regard, Boardman and Vining ~1989! argue that partial privatization may not be the best strategy for a government wishing to move away from reliance on state ownership. Fourth, we contrast the postdivestiture perfor- mance of control privatizations, in which governments surrender voting con- trol, and revenue privatizations, in which governments sell a minority ownership stake and do not surrender voting control. Boycko, Shleifer, and Vishny’s ~1996! model suggests that, in order to facilitate restructuring, both cash f low rights and voting control rights should pass from governments to private hands. For each of these partitions, we examine whether each group of firms records significant changes in the values of variables following the privatization. We also verify whether the value changes for the two subsam- ples are significantly different. For the sake of space, we only report the unadjusted results for the subsamples of competitive and noncompetitive firms, and privatized firms operating in upper-middle-income countries and in low-income and lower- middle-income countries ~Tables III and IV!. Although they are discussed in the text, we do not report the unadjusted results for the other two partition- ings. Furthermore, we discuss the market-adjusted results for the subsam- ples but we do not report them. The tables of the unreported results are available from the authors upon request. A. Profitability As firms move from public to private ownership, their profitability should increase. Privatization typically transfers both control rights and cash f low rights to the managers who then show a greater interest in profits and ef- ficiency than did the politicians ~Boycko et al. ~1996!!. We measure profit- ability by the return on sales ~net income to sales!, return on assets ~net income to total assets!, and return on equity ratios ~net income to equity!.6 Table I reports all three ratios, whereas the subsequent tables rely only on the return on sales ratio. We focus on this ratio because it is based on two f low measures that are less sensitive to inf lation and to accounting conven- tions than are the two other profitability ratios.7 6 The tax figure reported on a firm’s annual reports may include tax credits or carryforwards that do not pertain to the current year’s performance. Therefore, we compute profitability and efficiency ratios with income before and after taxes. For the 48 companies for which we have data on both net income measures, we obtain qualitatively similar results. 7 In a high-inf lation environment, unadjusted historical-cost information may not be meaning- ful ~White, Sondhi, and Fried ~1994!!. For example, rates of return on assets may be overstated because of undervaluation of fixed assets and underprovision for depreciation. The high- inf lation countries of Latin America require companies to use general price level accounting ~also known as constant dollar accounting!. In countries with moderate inf lation rates, such as Jamaica, Nigeria, and Turkey, financial statements are partially adjusted, except for those of AIICO, a Nigerian firm. However, the results of the tests for the full sample and the subsam- ples excluding AIICO were unchanged. Performance of Newly Privatized Firms 1087 Table II Comparison of Market-Adjusted and Unadjusted Results for the Sample of Market-Adjusted Firms This table presents, for the sample of firms for which we could compute market-adjusted accounting performance measures, both unadjusted and adjusted results. Market-adjusted proxies are obtained by subtracting the market median empirical accounting proxy from the firm’s accounting proxy in our sample. The market proxies are calculated for a sample of control firms available on the Disclosure database and which were not privatized. The table provides, for selected accounting performance measures, the mean and median values for the three-year period before and after privatization, the number of available observations, and the change in mean and median values. It also presents the Wilcoxon Z statistic for the difference in medians, the proportion of firms that performed as predicted by governments and the models discussed in the text, and a test of significance for this proportion. Variable Mean Before ~Median! Mean After ~Median! N Mean Change ~Median! Z-Statistic for Difference in Medians ~After–Before! Proportion of Firms That Performed As Predicted ~%! Z Statistic for Significance of Proportion Change Profitability Return on sales 5 Net income0sales Unadjusted 0.0421 0.1260 38 0.0839 3.241* 68.42 2.274** ~0.0475! ~0.1055! ~0.0381! Adjusted 20.0175 0.0695 38 0.0870 2.951* 71.05 2.598* ~20.0409! ~0.0294! ~0.0638! Efficiency Sales efficiency 5 Real sales0employees Unadjusted 0.8126 1.2570 16 0.4443 3.464* 93.75 3.500* ~0.8180! ~1.2347! ~0.4234! Adjusted 0.0379 0.2477 16 0.2097 1.758** 75.00 2.000** ~0.0825! ~0.2475! ~0.3275! Capital Investment spending Capital expenditures0sales Unadjusted 0.1612 0.5567 11 0.3954 1.689** 72.72 1.514*** ~0.1132! ~0.1548! ~0.0416! Adjusted 0.0093 0.0362 11 0.0269 1.333*** 63.63 0.908 ~0.0033! ~0.0190! ~0.0614! 1090 T h e J ou rn al of F in an ce Output Real sales 5 Nominal sales0consumer price index Unadjusted 0.9651 1.2514 39 0.2863 3.656* 69.23 2.374* ~0.9268! ~1.1541! ~0.2146! Adjusted 20.0204 1.1484 39 1.1688 1.828** 64.10 1.762** ~20.0118! ~0.1530! ~0.1511! Employment Number of employees Unadjusted 11584 12185 12 601 1.412*** 63.63 0.946 ~3217! ~3685! ~149! Adjusted 8490 8851 12 361 1.568** 66.66 1.156 ~2930! ~3431! ~154! Leverage Total debt0total assets Unadjusted 0.5627 0.4890 33 20.0737 22.188** 63.63 1.566*** ~0.5555! ~0.4378! ~20.0336! Adjusted 0.1080 0.0623 33 20.0457 21.223 60.60 1.218 ~0.1091! ~0.0291! ~20.0340! Dividends Dividend0sales Unadjusted 0.0371 0.0593 31 0.0221 2.821* 77.41 3.080* ~0.0116! ~0.0416! ~0.0193! Adjusted 0.0067 0.0226 31 0.0159 1.842** 67.74 1.993** ~20.0119! ~0.0001! ~0.0125! *, **, *** Significant at the 1, 5, and 10 percent levels, respectively. P erform an ce of N ew ly P rivatized F irm s 1091 Table III Summary of Results for the Noncompetitive versus Competitive Privatized Firms This table presents a comparison between the unadjusted performance changes of noncompetitive privatized firms versus competitive privatized firms. It provides, for selected performance measures, the mean and median values for the three-year period before and after privatization, the number of available observations, and the change in mean and median values. It also provides the Wilcoxon Z statistic for the difference in medians, the proportion of firms that performed as predicted by governments and the models discussed in the text, and a test of significance for this proportion. Variable Mean Before ~Median! Mean After ~Median! N Mean Change ~Median! Z-Statistic for Difference in Medians ~After–Before! Proportion of Firms That Performed as Predicted ~%! Z Statistic for Significance of Proportion Change Profitability Return on Sales 5 Net income0sales Noncompetitive 0.0464 0.1091 37 0.0627 2.278** 62.16 1.482*** ~0.0380! ~0.0591! ~0.0181! Competitive 0.0519 0.1104 41 0.0585 2.235** 63.41 1.719** ~0.0464! ~0.0900! ~0.0193! Efficiency Sales efficiency 5 Real sales0employees Noncompetitive 0.9189 1.1361 30 0.2171 3.506* 80.00 3.296* ~0.9348! ~1.1135! ~0.2041! Competitive 0.9263 1.2098 26 0.2834 3.314* 80.76 3.138* ~0.8671! ~1.1452! ~0.2875! Capital Investment spending Capital expenditures0sales Noncompetitive 0.1116 0.1749 26 0.0632 1.231 57.69 0.784 ~0.0535! ~0.0970! ~0.0078! Competitive 0.0977 0.3116 22 0.2138 1.996** 68.18 1.715** ~0.0723! ~0.1093! ~0.0181! 1092 T h e J ou rn al of F in an ce Output Real sales 5 Nominal sales0consumer price index UMIC 0.9324 1.2616 43 0.3291 4.648* 81.39 4.130* ~0.9047! ~1.1376! ~0.2416!†† LILMIC 1.0141 1.1734 35 0.1595 2.473* 68.57 2.210** ~0.9627! ~1.0931! ~0.1118! Employment Number of employees UMIC 9073 9205 35 132 0.638 51.42 0.169 ~4645! ~3965! ~94! LILMIC 13215 13365 22 150 1.834** 68.18 1.715** ~2487! ~2686! ~118! Leverage Total debt0total assets UMIC 0.5215 0.4342 32 20.0873 22.767* 71.87 2.485* ~0.5112! ~0.4181! ~20.1301!†† LILMIC 0.5766 0.5611 33 20.0155 20.679 54.54 0.521 ~0.6175! ~0.5658! ~20.0045! Dividends Dividend0sales UMIC 0.0201 0.0385 35 0.0184 3.341* 77.14 3.230* ~0.0039! ~0.0349! ~0.0185! LILMIC 0.0374 0.0684 32 0.0310 2.758* 75.00 2.840* ~0.0181! ~0.0246! ~0.0106! *, **, *** Significant at the 1, 5, and 10 percent levels, respectively. †, ††, ††† The difference between the value changes for the two subsamples of firms is significantly different at the 1, 5, and 10 percent levels, respectively. P erform an ce of N ew ly P rivatized F irm s 1095 Our results show significant improvements in profitability after divesti- ture. For example, the mean ~median! unadjusted return on sales goes from 4.9 percent ~4.6 percent! before privatization to 11 percent ~8 percent! after privatization. This significant increase ~at the 1 percent level! is achieved by 63 percent of the sample firms. As in MNR, the results are more significant for the return on sales ~at the 1 percent level! and return on assets ~at the 10 percent level! ratios. The increase in the market-adjusted return on sales ratio is equally significant for 71 percent of the firms. Our subsamples of firms operating in both competitive and noncompeti- tive industries, of firms from upper-middle-income countries, of partially privatized firms, and of both control and revenue privatizations show a signifi- cant ~at the 5 or 1 percent level! increase in return on sales following dives- titure ~for both unadjusted and market-adjusted ratios!. For the subsamples of fully privatized firms, profitability also increases significantly at the 10 percent level for unadjusted return on sales and at the 5 percent level for the market-adjusted return on sales ratio. Only for firms from low-income and lower-middle-income countries is the increase in profitability insignifi- cant ~for both unadjusted and market-adjusted return on sales ratios!. One possible explanation for the increase in profitability is that govern- ments “manage” reported earnings. For example, governments may have the incentive to decrease reported earnings prior to privatization in order to convince employees that privatization is the only viable solution. In this regard, as the timing of asset acquisitions and sales can affect reported earn- ings, our measure of net income excludes income from discontinued opera- tions and extraordinary items. Similarly, the choice of depreciation methods can affect reported earnings. For instance, choosing accelerated depreciation instead of straight-line depreciation is income-decreasing. Consequently, we reran the tests for the 46 companies for which we had data on operating income before depreciation. We obtained qualitatively similar results.8 B. Operating Efficiency Following privatization, firms should employ their human, financial, and technological resources more efficiently because of greater stress on profit goals and a reduction of government subsidies ~Kikeri et al. ~1992!, and Boycko et al. ~1996!!. We measure operating efficiency with the sales effi- ciency ~real sales per employee! and net income efficiency ~net income per employee! ratios. The sales efficiency ratio shows a significant median increase following privatization for both raw ~at the 1 percent level! and market-adjusted ~at the 5 percent level! performance analyses. This increase is achieved by 80 per- 8 However, firms might be able to manipulate their operating income by increasing their inventory and their accounts payable. Similarly, the decision of managers to tighten credit terms to decrease sales and hence accounts receivable could lead to a decrease in operating income. A preliminary analysis of inventories, accounts payable, and accounts receivable shows no evidence of manipulation of this kind. 1096 The Journal of Finance cent and 75 percent of the sample firms, respectively. For example, un- adjusted real sales per employee go from an average ~median! of 92.2 percent ~90.6 percent! of the year 0 level during the preprivatization period to 117 percent ~112.6 percent! of the year 0 level during the postprivatization pe- riod. Unadjusted net income per employee also increases dramatically ~63 per- cent ~mean! and 43.6 percent ~median!!. Overall, divested firms seem to improve their operating efficiency and thus to reach the objective most fre- quently put forward by governments launching privatization programs. How- ever, these improvements seem to be less significant ~at the 5 percent level rather than the 1 percent level! when adjusted for market effects. The increase in sales efficiency is also very significant for all the subsam- ples, except for firms operating in low-income and lower-middle-income coun- tries, for which it is significant only at the 10 percent level. This result for privatized firms in developing countries is similar to that of privatized firms in developed countries ~MNR!, and suggests that divested firms utilize their resources in a more efficient way than in the past. The results also indicate that the value changes for the partitions based on income are significantly different. First, the value changes in both prof- itability and efficiency are significantly larger for firms from upper-middle- income countries than for those from low-income and lower-middle-income countries at the 5 and 1 percent levels, respectively. This evidence lends support to Kikeri et al.’s ~1992! argument that high-income or upper-middle- income countries should be in a better position to privatize successfully.9 Second, the value changes in both profitability and efficiency are signifi- cantly larger for control privatizations than for revenue privatizations at the 10 percent level. The evidence is consistent with Boycko et al.’s ~1996! model, suggesting that in order to facilitate restructuring and improve performance governments should surrender voting control. Once we adjust for market conditions, we also document an increase in the sales efficiency ratio for all of our subsamples. A high percentage of the small number of observations making up each subsample shows such an increase. For the three subsamples including at least ten observations ~non- competitive industry firms, control privatizations and firms operating in upper- middle-income countries!, the increase in the sales efficiency ratio is significant at conventional levels for the last two partitionings. Many of the privatization programs in developing countries reserve a frac- tion of the shares issued ~from 5 to 20 percent! for the firm’s employees. We partition the sample into firms sponsoring ~44! or not sponsoring ~26! em- ployee stock ownership plans ~ESOPs! and we compare the performance of these two groups of firms. Our findings ~not reported here in the interest of 9 Since a country with a fairly developed capital market is likely to have a market friendly framework that favors the success of privatization, we partition our sample using the median of the 1985–1990 average stock market capitalization per GDP. We find that privatization yields greater benefits for countries with a fairly developed stock market than for those with less developed stock markets. Performance of Newly Privatized Firms 1097 As far as the subsamples are concerned, the unadjusted increase in the number of employees is very significant ~at the 1 percent level! for revenue privatizations. It is likewise significant ~at the 5 percent level! for partial privatizations and for firms operating in low-income and lower-middle- income countries and is significant ~at the 10 percent level! for noncompet- itive firms. The employment results also indicate that the difference between the value changes in the number of employees is significantly larger for noncompetitive firms and control privatizations than for competitive firms and revenue privatizations, respectively. As expected, newly privatized firms operating in competitive industries are more inclined to reduce employment levels. With the exception of the subsample of upper-middle-income coun- tries for which the increase in the employment level is significant at the 10 percent level, the small number of observations ~less than 10! for each sub- sample of newly privatized firms does not allow us to draw strong conclu- sions regarding the impact of privatization on the employment level, once we adjust for market effects. However, with the exception of the fully privatized firms, we observe an increase in the employment level for each subsample. F. Leverage The switch from public to private ownership should lead to a decrease in leverage because the government’s removal of debt guarantees will increase the firms’ cost of borrowing and because firms will have increased access to public equity markets ~MNR!. We measure leverage by the total debt to total assets and by the long-term debt to equity ratios. As do MNR, we focus on the first leverage measure. We document a significant decrease ~at the 5 per- cent level! in the debt to assets ratio and in the long-term debt to equity ratio for the unadjusted data.12 The average ~median! declines in the total debt0total assets ratio of 5.1 percentage points ~1.6 percentage points!, and in the long-term debt0equity ratio of 44.1 percentage points ~5.7 percentage points!, are important.13 However, the decrease in leverage is no longer sig- 12 A particular feature of the privatization programs in some countries, particularly in Latin America, is the prior debt restructuring of the firms upon privatization ~López-de-Silanes ~1997!!. By writing off the debt or part of the debt of the firm to be privatized, the government can increase the attractiveness of the firm to be sold. A thorough investigation of this issue shows that debt restructuring occurs in six of our sample companies operating in the following coun- tries: Argentina ~1!, Brazil ~4!, Trinidad and Tobago ~1!. We rerun the tests excluding these firms to determine whether debt restructuring drives the leverage decrease of our sample firms; we obtain virtually identical results. For a study of other restructuring methods accompanying privatizations, see Barberis et al. ~1996!, who provide evidence from a sample of privatized Russian shops. 13 Unlike a secondary offering where the government keeps the proceeds and no capital f lows into the firm, privatization through a primary offering raises new capital. We partition our SIPs into ~1! pure secondary offerings and ~2! at least partially primary issues. On the basis of our analysis of questionnaires, prospectuses, and annual reports, we are able to identify 15 pure secondary offerings and 21 at least partially primary issues. For both offerings, we observe a significant decrease in the debt to assets ratio. This suggests that primary issue privatizations do not drive leverage decreases. 1100 The Journal of Finance nificant when we adjust for market effects. This suggests that the leverage of newly privatized firms does not decline in comparison with other firms in their national markets in the postprivatization period.14 As for our subsamples, the unadjusted decrease in leverage after privatiza- tion is significant for competitive and noncompetitive firms, for firms oper- ating in upper-middle-income countries, for partially privatized firms, and for revenue privatizations, but the market-adjusted decrease in leverage is significant only for firms headquartered in upper-middle-income countries. G. Dividend Policy Following privatization, dividend payments should increase because, un- like governments, private investors generally demand dividends, and divi- dend payments are a classic response to the atomized ownership structure to which most privatization programs lead ~MNR!. The dividend to sales ratio and the dividend payout ratio ~dividend payments divided by net income! show significant increases for both unadjusted and market-adjusted data. For example, the average ~median! unadjusted dividend to sales ratio in- creases from 2.8 percent ~0.9 percent! before privatization to 5.3 percent ~3 percent! after privatization. Using unadjusted and market-adjusted data, 76 percent and 68 percent of the privatized firms, respectively, distribute more dividends as a proportion of their sales. This evidence combined with that of MNR suggests that, whatever the level of development, newly pri- vatized companies increase their dividend payments markedly. We also observe an increase in the dividend to sales ratio for all of our subsamples. The unadjusted increase in dividend to sales ratio is highly significant ~at the 1 percent level! for all the subsamples except of revenue privatizations, for which it is significant at the 10 percent level. Once we adjust the dividend to sales ratio for market effects, we still note an increase in dividend payments for all subsamples, although it is significant at the 5 percent level only for competitive firms, control privatizations, and firms headquartered in upper middle-income countries. For all of our subsamples, a fair majority of newly privatized firms ~at least 60 percent! experience an increase in the market-adjusted dividend to sales ratio. IV. Conclusion This study examines the financial and operating performance of newly privatized firms in developing countries. A large sample of newly privatized firms is drawn from a wide set of developing markets. To take account of the possibility that some of the differences between preprivatization and postpriva- 14 A possible explanation for the insignificant decrease in the market-adjusted leverage is provided by Singh ~1995! who gives evidence that in recent years, developing countries have tended to use more equity than debt to finance their growth. We would thus expect the market median to exhibit a decrease in leverage and hence to reduce the possibility that the market- adjusted debt to asset ratios of the newly privatized firms decline significantly. Performance of Newly Privatized Firms 1101 tization performance could be due to economy-wide factors, we use account- ing performance measures adjusted for market effects in addition to unadjusted accounting performance measures. For our full sample, both unadjusted and market-adjusted results show significant increases in profitability, operat- ing efficiency, capital investment spending, output ~adjusted for inf lation!, employment level, and dividends. We also observe a decline in leverage fol- lowing privatization but this change is significant only for unadjusted le- verage ratios. It is of interest to note that the results are generally less significant when we adjust the performance ratios for market effects. Our results are generally robust when we partition our data into various subsamples. In particular, in the case of unadjusted performance measures for which there are a sufficient number of observations, we note strong per- formance improvements for companies operating in competitive and noncom- petitive environments, for firms from upper-middle-income countries, for full and partial privatizations, for control privatizations where governments sur- render voting control, and for revenue privatizations where governments sell a minority ownership stake. We also present evidence of weaker perfor- mance improvements for firms from low-income and lower-middle-income countries. Furthermore, we find that the differences in the increase of prof- itability and efficiency are significantly larger for control privatizations and firms from upper-middle-income countries than for revenue privatizations and firms from low-income and lower-middle-income countries, respectively. In any event, ownership seems to be important. Indeed, privatization brings with it private owners who place greater emphasis on profit goals and also carry out new investments that lead to increased output and employment. As a result, efficiency improves and profitability follows. These improve- ments cannot be attributed to the prevailing market structure or the terms of share issue privatizations. Nevertheless, our results, combined with those of MNR, suggest that, both in developing and developed countries, newly privatized firms improve their performance. Furthermore, as far as devel- oping countries are concerned, it appears that privatization yields greater benefits for companies headquartered in countries with high income per capita. 1102 The Journal of Finance Malaysia Malaysia Airlines System Berhad Airlines Oct. 1985 10,632 77.21 100 70n Malaysia International Shipping Corp. Shipping Jan. 1987 2,578 53.97 100 30n Cement Manufacturer Sarawak Berhad Cement Dec. 1988o 2.39 100 84.40 Edoran Otomobil Nasional Bhd. Automobile June 1990 1,941 57.11 100 70m Syaricat Telekoms Malaysia Bhd. Telecommunication Oct. 1990* 28,000 872 100 76.10m Apr. 1992 Cement Industries of Malaysia Berhado Cement 1990 540 74.03 58.86 0.18 Perusahaan Otomobil Nasional Berhad Automobile Jan. 1992 3,141 278.34 75.20 52.60m Tenaga Nasional Berhad Electricity Mar. 1992 22,752 1192.65 100 78m,n Mexico Compañía Mexicana de Aviación Airlines Sept. 1989* 140 50.83 40.60 1992 35.2 4i Teléfonos de Mexico Telecommunication Dec. 1990* 49,912 2557 56 26.10 May 1991 2368 26.10 9.60 May 1992 1361 9.60 4.60 Banamex Banking Aug. 1991 31,964 3100 71 0 Bancomer Banking Oct. 1991* 36,414 2907 78.53 22.53 May 1993 22.53 0 Nigeria AIICO Insurance 1989 364 0.465 0 Crusader Insurance Insurance 1989 202 45.52 0 Unipetrol Petroleum 1991 113 5.33 100 40 Pakistan Pak International Airlines Airlines 1990 11 10i Millat Tractors Tractor Jan. 1992 568 4.9 51 0p Philippines Hotel Enterprises of the Philippines Hotel 1988 484 Philippine National Bank Banking May 1989* 84.97 1992 117.6 67 57 Portugal Uniao cervejeira Brewing Apr. 1989* 1,558 60.2 100 51 June 1990 90.58 51 0 Banco Totta y Açores Banking July 1989* 4,200 181.3 100 51q July 1990 159.03 51 20 P erform an ce of N ew ly P rivatized F irm s 1105 Appendix—Continued Government Shareholdings ~%! Company Name Industry Date of Sale Number of Employees Value of Issue ~U.S. $millions! Before Privatization After Privatization Portugal ~continued ! Companhia de seguros Tranquilidade Insurance Dec. 1989* 1,525 172.03 100 51q Oct. 1990 141.49 51 0 Jornal de Noticias Newspaper May 1990 701 50.6 86.26 0q Central de Cervejas Brewing Nov. 1990 2,700 260.98 100 0q Banco Portugues do Atlantico S.A. Banking Dec. 1990* 372.4 100 67q May 1992 378.82 67 49.40 July 1993 181.27 49.40 24.40 Mar 1994 89.91 24.40 0 Sociedade Financeira Portuguesa ~Banco Mello! Banking May 1991 106 106.8 100 0q Banco Espirito Santo e Commercial Banking July 1991* 6,729 406.0 100 60q de Lisboa Feb. 1992 632.13 60 0 Banco Foncecas & Burnay Banking Sept. 1991* 3,000 250.72 100 20 July 1992 71.41 20 0 Mundial Confiança Insurance Apr. 1992 239.54 100 0q Petroleos de Portugal Oil & Energy June 1992 3,982 320.05 100 75 Imperio Insurance Nov. 1992 177.24 100 0q Credito Predial Portugues Banking Dec. 1992 278.17 100 0q Uniao de Bancos Portugueses Banking Dec. 1992 3,607 166.39 61.11 43.11q Singapore Sembawang Shipyard Ship Repairing & Apr. 1973 100 75 Building 1988* 1,956 75 '30 Keppel Corp Shipbuilding Sept. 1980* 6,600 47.29 100 55r 1988–89 68 '30 Neptune Orient Lines Ltd. Shipping Apr. 1981* 1,632 66.4 100 67.70s Apr. 1984 67.70 62.30 Dec. 1987 88 62.30 52.80 Nov. 1988 154.1 52.80 30 1106 T h e J ou rn al of F in an ce Singapore International Airlines Airlines Nov. 1985* 14,963 233 77t 63 June 1987 284 63 53.50 DBS Land Real Estate Sept. 1987 325 150.11 100 48.80u Taiwan China Steel Corp. Steel 1989* 9,715 100 97.20 Apr. 1991 97.20 91.10 May 1992v 904 91.10 76.50 Thailand Thai Airways International Airlines Mar. 1992 19,286 237.5 100 92.7 w Tunisia Société Industrielle de Textile Textile 1989 1,750 4.42 43.40 0 Turkey Arcelik Electronics Apr. 1990 3,434 19.89 13.32 7.49 Bolu Cimento Cement Apr. 1990 8.27 35.33 24.95 Celik Halat Metal Products Apr. 1990 7.75 29.28 16.03 Cukurova Elektrik Electricity Apr. 1990* 1,475 38.83 18.65 13.20 Feb. 1993x 81.1 11.50 0.25 Eregli Edemir Iron & Steel Apr. 1990 8,595 53.11 52 45.72 Kepez Elektrik Electricity Apr. 1990* 9.39 43.68 35.54 Feb. 1993x 41.3 35.2 0 Petkim Petro Kymia Petrochemical Products June 1990 7,192 150.62 99.97 91.88 Adana Çimento Sanayii Cement Feb. 1991 819 27.96 47.28 12.96 Migros Turk Supermarket Chain Feb. 1991 5.61 42.22 5.82 Tofas Otomobil Ticaret Automobile Feb. 1991* 229 13.07 39 23y June 1991 0.97 23 21.64 1992 0.9 21.64 20.64 1993 5.3 20.64 13.64 Oyse-Nidge Çimento Cement May 1991* 2.65 99.83 87.11 Mar. 1992z 22.5 87.11 Petrol Ofisi Petroleum May 1991* 7,871 14.39 100 95.98 1992 2.5 95.98 95.48 1993 2.5 95.48 94.48 P erform an ce of N ew ly P rivatized F irm s 1107