Introduction. The New State Capitalism (3) Empirical Studies on State-owned Enterprises, Privatization and Nationalization.

Transcription

1 Introduction The New State Capitalism (3) Empirical Studies on State-owned Enterprises, Privatization and Nationalization Carsten Sprenger Higher School of Economics Moscow Februrary 2017 / CES Lectures, Munich In this third lecture, I try to give a condensed overview of empirical studies on state ownership its determinants and consequences, mostly in terms of firms performance but also for the cost of debt and their acquisition strategies. The amount of academic work is huge, especially on privatization, so that I had to make (probably subjective) choice. I emphasize papers that have made some methodological innovations or tested new types of hypotheses. Overview 1 Cross-sectional comparisons between SOEs and private firms Cross-sectional comparisons between SOEs and private firms

2 Boardman and Vining, 1989 Boardman and Vining, 1989 Boardman and Vining, 1989 compare the performance of state-owned, privately owned and mixed enterprises in a competitive environment. Previous studies often had problems to disentangle the effects of ownership and regulation in non-competitive markets. They use a sample of 500 large non-us companies and regress several accounting measures of performance on dummy variables for mixed and state ownership, and control for firm size, market share, industry, the level of competition in each industry and country where the headquarter is located. Mixed and SOEs are found to perform substantially worse than similar private companies. Boardman and Vining acknowledge that their results may potentially suffer from the endogeneity of the ownership structure : The government could, for example, choose to nationalize poorly performing companies to prevent massive layoffs. In other words, the subsample of state-owned and mixed firms might not be random. Kole and Mulherin, 1997 Dewenter and Malatesta, 2001 This problem is avoided in the study by Kole and Mulherin, 1997 who use a presumably random sample of SOEs, namely those subsidiaries of enemy-owned firms that were seized during second world war and subsequently held under custodianship by the US government. The authors find that the government owned companies do not perform significantly different from the private companies in the same industry. Dewenter and Malatesta, 2001 repeat the analysis of Boardman and Vining, 1989 for a larger sample and their control variables include firm size, GDP growth and country dummies. The authors find that SOEs are less profitable (in terms of return on sales, return on assets and return on equity), employ more labour per unit of sales and are more leveraged.

3 Bozec et al., 2002 Bartel and Harrison, 2005 An important issue in the comparison of performance between public and private companies is the role of noncommercial objectives of SOEs that would lower their profitability in most cases. Bozec et al., 2002 study Canadian SOEs and distinguish between those with non-commercial objectives and those with the explicit objective of profit maximization. They find that SOEs that maximize profits perform similarly to their private peers, while only the SOEs with political and social objectives perform worse. This is consistent with the finding that performance of SOEs often improves before their privatization (e.g., Dewenter and Malatesta, 2001). Bartel and Harrison, 2005 go a step further in exploring the institutional environment of companies in order to provide a meaningful comparison between companies under public and private ownership. For a sample of Indonesian firms, they control for the access to soft loans from the government, for domestic and international competition as well as for foreign investment in the industry. The authors use Total Factor Productivity (TFP) as a performance measure and address the issue of possible endogeneity of ownership and government financing. Bartel and Harrison, 2005 Two studies on China The authors conclude that state ownership alone does not lower TFP. However, when state ownership interacts with a soft budget constraint, protection from external competition and foreign ownership, then it does lower TFP. In short: The environment where an SOE operates matters for the performance. Both Tian and Estrin, 2008 and Chen et al., 2009 conclude that state ownership can have a positive effect on corporate performance in a weak institutional environment such as in China. Tian and Estrin, 2008 find a positive effect of state ownership if it exceeds a certain threshold level (approx. 25 per cent if performance is measured by Tobin s Q). Chen et al., 2009 differentiate between different types of SOEs. They find that SOEs affiliated to the central and local governments perform better than private companies, which in turn outperform state asset management bureaus that are operated at a regional level.

4 (1) (2) Borisova and Megginson, 2011 and Borisova et al., 2015 analyze the effect of government ownership on the cost of debt (bond spreads) The first paper uses variation in retained government shares in partially or fully privatized European firms, while the second looks at increases in government shares in listed firms around the world. In principle, state ownership can affect bond spreads via (at least) four channels: Implicit government guarantee Performance differences between state-owned and private companies, threat of government expropriation Uncertainty surrounding ownership change Bondholder-shareholder conflicts, especially if preference is given to new shareholders in privatization. Borisova and Megginson, 2011 find that a 1 percentage point decrease in government ownership is associated with an increase in the credit spread by 0.75 basis points. However, fully privatized companies exhibit lower credit spreads than partially privatized firms (indicating costs of a lengthy privatization process). Borisova et al., 2015 generally find that cost of debt increases with higher government ownership, consistent with state-induced investment distortions. However, during the financial crisis and for financially distressed firms, this effect is reversed. This means that implicit government guarantees become dominant in these situations. Sapienza, 2004 Berger et al., 2005 find for Argentinean banks in the 1990s that state-owned banks have poorer long-term performance and that privatized banks significantly improve their performance after privatization. Dinc, 2005 studies government-owned banks. The author finds that government-owned banks across countries engage in politically motivated actions: They increase lending in election years relative to private banks. Also, they find that such banks in emerging markets hold a 50 per cent higher share of their assets in government securities, i.e., instead of financing projects that the private sector does not want to finance, they finance the government itself. Sapienza, 2004 uses a detailed database of loans of Italian banks to investigate how government ownership affects bank lending. The sample period is In 1995, 58% of banking assets were held by state-owned banks. The author aims to distinguish between different theories of state ownership the social view, agency view and the political view. Since the mix of banking activities could change under state ownership, looking just at overall bank performance might not allow to shed light on these theories.

5 Sapienza, 2004 Sapienza, 2004 The comparison is made cleaner since all Italian banks use the same information system that provides banks with financial and economic information about Italian companies and computes a credit score to identify the risk profile of these companies. It turns out that state banks lend at a lower interest rate (the difference is 44 basis points) if two sets of companies with identical scores that borrow either from state- or privately owned are compared. Companies in the south benefit more from borrowing from state-owned companies than companies in the north, even after controlling for the presence of financial constraints. State-owned banks are also more inclined to lend to larger firms. While these results are consistent with the social view (channeling funds to more depressed areas) and the agency view (channelling funds to larger firms), the third results points to the presence of patronage (the political view) Lending behavior of the state-owned banks is affected by electoral results of the party affiliated with the bank: the stronger the political party in the area where the bank is lending, the lower are the interest rates charged. This result is robust to the inclusion of bank and firm fixed effects, which excludes that results are driven by (time-invariant) omitted bank and firm characteristics. The literature on the effects of privatization, primarily in Central and Eastern Europe and the former Soviet Union has been summarized in three excellent surveys by Megginson and Netter (2001), Djankov and Murrell (2002) and Estrin et al., (2009). While in Central and Eastern Europe, privatization typically brought improvements in technical efficiency (labor productivity, total factor productivity) and financial performance, the effects for the countries of the former Soviet Union are less clear.

6 Brown and Earle, 2006 Brown and Earle, 2016 Early studies often had small samples and short time series. Using a much larger database, Brown and Earle, 2006 find positive effects of privatization on total factor productivity in Romania, Hungary and Ukraine. The effect for Russia is positive or negative depending on specifications, but in any case smaller than in the other three countries. In addition, any positive performance effects of privatization need more time to materialize in Russia than in the other countries. Brown and Earle, 2016 revisit the evidence of the performance effects of privatization and ask why the estimated effects vary so much across countries and time. Is this due to different designs of privatization programs, types of firms privatized, types of data, different outcome variables, estimation methods? The authors find evidence that the type of post-privatization controlling owner matters: Foreign owners perform better than domestic large shareholders, which are better than management-employee buyouts and voucher privatizations. Brown and Earle, 2016 Bai et al., 2009 Little evidence that firm characteristics (initial quality, complexity, capital intensity) matter. Finally, the economic environment (inflation, GDP growth) play an important role. In a study on a large sample of privatization deals in China, Bai et al., 2009 find that privatization of China s SOEs had almost no effect on employment in the medium-run, but it increased labour productivity and firm profitability. These changes were more pronounced when state ownership was reduced to a minority position. It is not straightforward to reconcile these results with the results from Tian and Estrin, 2008 and Chen et al., One possibility is that privatized firms are different from private firms in that the former are receiving preferential treatment from the government even after privatization, leading to better performance indicators than in newly established private firms.

7 Recent nationalizations in Russia, China, Argentina, Bolivia, Venezuela They might follow different motivations and patterns, so one-country studies might be appropriate if a sufficiently large sample of state takeovers can be found This seems to be the case in Russia. As with privatization, it is crucial to correct for selection and to find an appropriate control group. So far, there is one study on the choice of targets for nationalization in Russia by and Huang et al., 2015 on re-nationalization in China. focuses on the incentives of the government to acquire particular firms using a sample of 153 firms that were privately owned in 2003 among the 200 largest Russian companies in that year. Out of these firms, 26 had been targeted for nationalization at the end of 2008, and 19 nationalizations actually occurred by that time. Sectors include oil and gas, auto-making, machine-building, aviation, titanium production, media, banking. The Russian government did not issue any official explanations or a nationalization program. The author finds that nationalization in Russia has been driven by political factors and not by economic factors. Belonging to one of the strategic sector (according to a list in a law that came into effect in 2008) increases considerably the likelihood of privatization. The law introduced restrictions for foreign takeovers of companies in such strategic sectors. Also, the vast majority of these companies had been owned by domestic owners and had been privatized previously, so we actually talk about re-nationalizations here. Profitability, market share, employment and other economic factors are not systematically related to the likelihood of nationalization.

8 , "The Reversal of Privatization: Determinants and Consequences" documents large-scale reversal of privatization in China in the period Local governments were re-possessing controlling ownership stakes in previously privatized firms. Among the economic determinants, the authors find that firms are nationalized to lessen the burden of local unemployment, a key yardstick for social harmony. Larger firms whose workforce accounts for a significant fraction of the local labor market and those located in regions with higher unemployment rates are more likely to be taken over by the local government. But there is an additional political economy factor at work. Many secretaries of the Chinese Communist Party in a province belong to one of three influential factions within the party linked to either 1 Hu Jintao, former Secretary General of the Party coming from the Youth League of the Party, 2 his predecessor Jiang Zemin, former Party Secretary of Shanghai, 3 and the Princelings, descendants of prominent and influential older generations of Party officials. It turns out that provincial party leaders without strong support and protection from one of these factions initiate renationalization policies more often. The immediate good social outcomes such as lower unemployment rates could help guarding their status. In the second part, the authors study the impact of large-scale re-nationalization on firm performance. Since re-nationalization is not a random event, an instrumental variable approach is used. Instruments for re-nationalization are regional variables (unemployment rate, share of the state sector) and whether the Party leader is newly appointed and is affiliated with a particualr political faction by this work experience (for the first two) or by birth (for the third one). The initiation of factional relationship has been always before the leader would eventually become the supreme leader of the Party. The authors try to exclude direct channels of influence of factional affiliation on firm performance or fiscal support of the province from the central government. The IV regressions show that re-nationalization leads to a sharp drop in firms profitability and labor productivity. On the provincial level, unemployment rates indeed go down initially, but not in the long run. Provincial GDP growth rates fall after re-nationalization, suggesting that the cost of temporarily stabilizing the labor market is lost efficiency and overall economic growth.

9 Karolyi and Liao, 2016 Karolyi and Liao, 2016, "State capitalism s global reach: Evidence from foreign acquisitions by state-owned companies" examine the motives for and consequences of cross-border acquisitions led by government-controlled acquirers over the period from 1990 to A government-controlled acquirer is defined as a government agency, investment fund (e.g., SWF) or SOE with an ultimate government ownership of at least 50%. The total number of announced deals in the data is 127,786, of which 4,759 (3.7%) are government-led. Only in about 40% of the cases, deal values are reported. Total (reported) deal value is 9,040 billion (constant 2000) US$, of which 593 billion (6.6%) are for government-led deals. Number and total deal value of cross-border acquisitions of government-led acquirers by year Total deal value of government-led acquisitions by country of acquirer

10 Total deal value of government-led acquisitions by country of target firms Karolyi and Liao, 2016 First set of estimations is at the country level. The authors construct a variable "acquisition intensity" for any country pair i (home) j (host) for government-led deals, A G ijt, defined as the number of government-led deals with acquirer from i and target in j, normalized by all government-led deals in target country j in a given year t. The same variable is computed for corporate (private)-led deals, A C ijt. The dependent variable is then the difference A G ijt AC ijt, i.e., the excess fraction of the number of deals between government-controlled acquirers and corporate acquirers. Karolyi and Liao, 2016 Karolyi and Liao, 2016 Government-led acquirers are more active when their home currency appreciates, are more likely to come from emerging market economies, countries with worse governance and transparency indicators, and more autocratic countries... countries with higher levels of foreign currency reserves, larger domestic acquisition programs and a more dissimilar industry structure (higher potential to diversify). These results speak to the possible objectives of governments in their foreign acquisitions. Results are mostly confirmed in deal-level estimations logistic regressions of the probability that a firm is targeted by a government-controlled acquirer in a given year. In addition, government-led acquirers are more likely to acquire larger firms, with higher market-to-book ratios and sales growth (two measures of growth opportunities). When evaluating the consequences of the presence of government-led acquirers, the author control for the selection bias by applying a Heckman two-stage procedure. Instruments are the economic and political determinants of government presence in cross-border acquisitions just described.

11 Karolyi and Liao, 2016 Karolyi and Liao, 2016 The authors concentrate on capital market outcomes: cumulative abnormal returns (CARs) for different event windows (-10,+10), (-5,+5) and (-1,+1), completion rates and likelihood of 100% control change transactions. In univariate comparisons, government-led acquirers have lower CARs, lower deal completion rates, are less likely to engage in a 100% control change transaction. When controlling for country, firm and deal-specific factors, the differences in CARs and deal completion rates become insignificant, complete control is still less likely. Results change drastically when controlling for selection: Government acquirers have now higher 11-day accouncement-window CARs and higher likelihood of complete control. Certain observed and unobserved attributes that increase the likelihood that a cross-border deal is led by a government-controlled acquirer (the political, economic, or social objectives of governments) are also associated with lower target CARs. Thus, it can be argued that government-controlled acquirers strategically identify acquisition deals that offer less value added for target shareholders than what would accrue if they chose targets with the same attributes as corporate cross-border acquirers. Internationalization of SOEs Internationalization of SOEs There has been increased regulatory concern about foreign acquisitions with government-controlled acquirers. The equal failure rates and higher announcement returns after correcting for selection seem to indicate that such investments are evaluated positively by stock markets, which might mitigate such concerns. In the Karolyi and Liao paper, there is also a short discussion of the role of sovereign wealth funds (SWFs) who are a subset of their government-led acquirers and have been caught more attention in the literature. Still, many questions are unanswered, for example the role of politically connected managers or boards in both SOEs and private firms and the effect on their acquisitions, the role of different types of government agencies, long-term operational and financial performance, the role of restrictions to foreign investment in host countries, in particular with respect to SOEs.

12 Conclusion (1) Conclusion (2) In general, the concept of state capitalism could be even generalized from state ownership to a wider range of political influence (as in the broad definition of Mussaccio and Lazzarini in Lecture 1). This would include a variety of mixed ownership forms such as combinations of local government and private ownership in China, politically connected managers and board members, revolving doors between business and politics, limited power of the government on SOEs (e.g., SOEs sometimes refuse to pay dividends to the government), government support through preferential treatment or subsidies. Also, we know little about corporate governance within SOEs managerial compensation, the composition of boards of directors, especially when private investors hold equity stakes in these firms as well as about many decisions in these firms (investment, employment, wages, capital structure, dividends etc.) In short, both on the theoretical and empirical side, there is ample scope for future research!