The Decision to Privatize: Finance, Politics and Patronage

Transcription

1 The Decision to Privatize: Finance, Politics and Patronage Serdar Dinç 1 and Nandini Gupta 2 April 17, Sloan School, MIT 2 Kelley School of Business, Indiana University

2 Why privatize? Increase e ciency of state-owned rms Performance improves after privatization (Megginson & Netter, 2001; Gupta, 2005) Raise revenues from privatization proceeds > $1 trillion raised from (Megginson & Netter, 2001) Develop stock markets Market value of privatized SOEs increased from <$50B in 1983 to $2.4T in 1999 (10% of world stock market capitalization) (Megginson et al., 2004)

3 Yet... governments can t seem to let go Governments retain majority ownership in SOEs in many countries (Megginson, 2005): Boubakri, et al. (2005) nd that government is controlling shareholder in 40% of recently privatized rms. Majority ownership stakes privatized in 30% of companies across 34 countries (Bortolotti et al., 2004). La Porta et. al (2002) show that governments control 48% of assets of privatized banks across 92 countries. In India only 50 of 280 federal government owned rms have sold some equity to private owners. Question: Do politicians career concerns and private bene ts a ect the decision to privatize?

4 Does politics matter in the privatization decision? If I do it immediately, I get into trouble. I get trouble from the workers. I get trouble from the political parties. I get trouble from the general public. Prime Minister Narasimha Rao, 3/11/94 If you face immediate political problems - elections in four states - it is hard to push ahead. We had to worry about the prospects of unemployment if public sector units faced closure. Finance Minister Manmohan Singh, 4/8/97 The Indian voters were rejecting the NDA (National Democratic Alliance) government, which, as one poll slogan had it, stood for the National Disinvestment Agency. The Hindu, 5/14/2004

5 The Political Economy of Privatization Bene ts from privatization are dispersed Revenues go to the state E ciency improvements are captured by consumers and tax payers Costs of privatization are concentrated Layo s of surplus workers Loss of private bene ts for politicians Costs may be geographically concentrated Special interests may capture the policy process

6 The Political Economy of Privatization Bene ts from privatization are dispersed Revenues go to the state E ciency improvements are captured by consumers and tax payers Costs of privatization are concentrated Layo s of surplus workers Loss of private bene ts for politicians Costs may be geographically concentrated Special interests may capture the policy process

7 Political Factors Political Competition Hypothesis: Adverse e ects of privatization likely to be concentrated in region where rm operates so government may lose political support in that region. E ect of voter backlash may be greater if governing and opposition parties in a close race. Will the government delay privatization in states where it faces a close race with opposition parties?

8 Political Factors Political Competition Hypothesis: Adverse e ects of privatization likely to be concentrated in region where rm operates so government may lose political support in that region. E ect of voter backlash may be greater if governing and opposition parties in a close race. Will the government delay privatization in states where it faces a close race with opposition parties?

9 Preview: Privatization and Political Competition

10 Political Factors Political Strength Privatize in region where governing party has strong support to minimize e ects of a backlash. Or Reward supporters in region where governing party has strong support. Political Patronage Hypothesis: Politicians get rents from SOEs such as jobs for political supporters; corruption (Shleifer and Vishny, 1994; Boycko, et al. 1996). Will the loss of rents matter more if the rm is located in the home state of the politician?

11 Political Factors Political Strength Privatize in region where governing party has strong support to minimize e ects of a backlash. Or Reward supporters in region where governing party has strong support. Political Patronage Hypothesis: Politicians get rents from SOEs such as jobs for political supporters; corruption (Shleifer and Vishny, 1994; Boycko, et al. 1996). Will the loss of rents matter more if the rm is located in the home state of the politician?

12 Political Factors Political Strength Privatize in region where governing party has strong support to minimize e ects of a backlash. Or Reward supporters in region where governing party has strong support. Political Patronage Hypothesis: Politicians get rents from SOEs such as jobs for political supporters; corruption (Shleifer and Vishny, 1994; Boycko, et al. 1996). Will the loss of rents matter more if the rm is located in the home state of the politician?

13 Financial Factors Firm Size Information costs are lower for larger rms (Chemmanur and Fulghieri, 1995; Pagano et al., 1998). More likely to privatize large rms rst. Firm Pro tability For higher sale proceeds and to build public support sell better rms rst (Dewenter and Malatesta, 1997; Megginson et al., 2004; Gupta et al., 2007). But e ciency improvements highest for least e cient rms (Claessens et al., 1997; Frydman et al., 1999). Trade-o between proceeds versus e ciency.

14 Preview of results A rm is less likely to be privatized if its main operations are located in a state where the party in power is in a close race with opposition parties. where the party in power did not win a large share of the seats. A rm is more likely to be privatized if it is large and pro table. No rm located in the home state of the Cabinet minister in charge of that rm is privatized.

15 Preview of results A rm is less likely to be privatized if its main operations are located in a state where the party in power is in a close race with opposition parties. where the party in power did not win a large share of the seats. A rm is more likely to be privatized if it is large and pro table. No rm located in the home state of the Cabinet minister in charge of that rm is privatized.

16 Preview of results A rm is less likely to be privatized if its main operations are located in a state where the party in power is in a close race with opposition parties. where the party in power did not win a large share of the seats. A rm is more likely to be privatized if it is large and pro table. No rm located in the home state of the Cabinet minister in charge of that rm is privatized.

17 Di erences in political institutions may explain broad patterns in privatization: Conservative governments privatize more (Biais and Perotti, 2002). Privatization more likely in majoritarian political systems (Bortolotti and Pinotti, 2006). Firm level factors a ect likelihood of privatization: Pro table rms more likely to be sold early (Gupta, et al., 2008), and in public capital markets (Megginson, et al., 2004). Unpro table banks more likely to be privatized (Clarke and Cull, 2002). Financial reforms implicitly motivated by political incentives: Kroszner and Strahan (1999); Chari and Gupta (2008). Large literature nds positive e ect of privatization on rm performance (Megginson and Netter, 2001): We show selection of rms is not random and correct for endogeneity of privatization decision. Literature

18 State-Owned Firms in India Federal government-owned rms account for > 40% of capital stock. Half of the rms are chronically loss making. Subsidies to 300 rms > 5% of GDP. Oversta ed with overpaid workers average wages twice that of private rms (Panagariya, 2007). Disinvestment launched with 1991 economic reforms.

19 Political System British-style parliamentary government. Lower house or Lok Sabha are directly elected representatives. Representatives elected from 550 districts across 35 states and union territories. Anti-incumbency bias - 4 out of 5 elections governing party voted out. Considerable ideological heterogeneity in governing and opposition parties.

20

21 Firm Level Data Firm level balance sheet data on federal government owned rms from FY from CMIE Both privatized rms and rms that remain fully government-owned 259 manufacturing and non- nancial sector rms - 92% of rms owned by the federal government. Location of main operations of rm (CMIE and hand collected) If multiple locations, main operations de ned as plant with largest asset base. Privatization data (hand collected)

22 Summary Statistics (Table 1) Variables Privatized Fully Govt Owned All Firms Sales 10,969*** 3,701 4,372 (12,516) (7,767) (8,579) Number of Firm years Number of Firms Assets 15,204*** 7,870 8,548 (20,313) (15,839) (16,437) Number of Firm years Number of Firms Profit/Sales (%) 15.5*** (22.3) (68.7) (66.2) Number of Firm years Number of Firms Wages/Sales (%) 14.7*** (16.7) (54.8) (53.1) Number of Firm years Number of Firms

23 Political Data Observe the following electoral data for 35 states and union territories (Election Commission of India) All federal (5) and state assembly elections (107) held from Number of seats contested and won and votes of all 450 political parties participating in the elections. Political alliances identi ed from press sources and election websites. Governing party is de ned as the political party alliance that is in power. Opposition party is the party alliance with second largest number of seats. Match each rm by location of main operations to state political measures. Data on Cabinet ministers and their jurisdiction (hand collected) Match each rm by location of main operations to home state of minister in charge.

24 Comparing Privatized to Govt-Owned Firms (Table 3) Variables Privatized Fully Govt Owned All Govt Seat Share (%) 50.6* (32.9) (33.7) (33.7) Opposition Seat Share (%) (29.7) (26.6) (26.8) Seat Share Difference (%) (55.7) (50.7) (51.1) Abs Seat Share Difference (%) 54.9*** (25.3) (29.3) (29.1) Govt Vote Share (%) 39.3* (13.3) (14.0) (13.9) Opposition Vote Share (%) (16.0) (14.5) (14.6) Vote Share Difference (%) (22.1) (20.3) (20.4) Abs Vote Share Difference (%) (16.0) (14.0) (14.2) Number of Firm years

25 Empirical Speci cation: h it = h 0 (t) + exp(βx it 1 + γ Political Factors i,t ) Cox proportional hazard analysis for the time until privatization Privatization de ned as rst sale of equity Errors are clustered at the state level Industry dummies are included i denotes rm. Privatized rms included until rst sale Null Hypothesis: Privatization decision not related to the governing party s political strength or the political competition it faces in the state where a rm s main operations are located.

26 The Decision to Privatize (Table 4) (1) (2) (3) (4) (5) Ln(Sales) 0.798*** 0.805*** 0.774*** 0.807*** 0.775*** (0.162) (0.176) (0.169) (0.177) (0.169) Profit/Sales 1.997** 1.912** 1.854** 1.912** 1.854** (0.898) (0.950) (0.898) (0.949) (0.894) Wages/Sales 3.463** 3.181** 3.273** 3.171** 3.267** (1.480) (1.446) (1.451) (1.456) (1.458) Govt Seat Share 0.723** 0.742** (0.347) (0.351) Abs Seat Share 0.793* 0.809* Difference (0.425) (0.438) State Govt Majority (0.332) (0.329) Industry FE Yes Yes Yes Yes Yes Number of Firm Years

27 Politics and Privatization The party in power is more likely to privatize a rm located where it is strong From 25th to 75th percentile in Govt Seat Share =) 70% faster rate of privatization Less likely to privatize a rm where it faces a close race From 25th to 75th percentile in Absolute Seat Share Di erence =) 63.5% faster rate of privatization

28 Robustness checks Results are robust to Alternative political measures State legislature elections Political ideology and state level characteristics Privatization method Alternative speci cations

29 Alternative Political Measures (Table 6A) (1) (2) (3) (4) (5) (6) Ln(Sales) 0.823*** 0.817*** 0.810*** 0.827*** 0.832*** 0.815*** (0.166) (0.171) (0.161) (0.151) (0.151) (0.153) Profit/Sales 2.042** 1.975** 1.951** 2.073** 2.043** 1.980** (0.948) (0.962) (0.929) (0.957) (0.982) (0.993) Wages/Sales 3.507** 3.323** 3.314** 3.547** 3.419** 3.408** (1.507) (1.478) (1.493) (1.492) (1.504) (1.509) Opposition Seat Share 0.601* (0.344) Seat Share Difference 0.401** (0.191) Govt Vote Share (1.301) Opposition Vote Share 1.488* (0.876) Vote Share Difference 1.340** (0.633) Abs Vote Share 1.947*** Difference (0.623) Industry FE Yes Yes Yes Yes Yes Yes Number of Firm Years

30 State and Federal Elections (Table 6B) (1) (2) (3) (4) (5) (6) (7) Ln(Sales) 0.803*** 0.818*** 0.813*** 0.775*** 0.829*** 0.831*** 0.814*** (0.170) (0.163) (0.166) (0.169) (0.149) (0.148) (0.152) Profit/Sales 1.942** 2.029** 1.986** 1.874** 2.080** 2.065** 1.949* (0.937) (0.930) (0.946) (0.912) (0.954) (0.974) (0.999) Wages/Sales 3.300** 3.508** 3.395** 3.324** 3.518** 3.396** 3.391** (1.463) (1.497) (1.485) (1.464) (1.490) (1.504) (1.507) Govt Seat Share 0.532* (0.276) Opposition Seat 0.443* Share (0.259) Seat Share 0.298** Difference (0.136) Abs Seat Share 0.684* Difference (0.372) Opposition Vote 1.494* Share (0.881) Vote Share 1.249** Difference (0.636) Abs Vote Share 1.975*** Difference (0.594) Industry FE Yes Yes Yes Yes Yes Yes Yes Number of Firm Years

31 Ideology and State-Level Factors (Table 7) (1) (2) (3) (4) (5) (6) Largest States Ln(Sales) 0.821*** 0.788*** 0.818*** 0.782*** 0.842*** 0.787*** (0.193) (0.181) (0.172) (0.164) (0.243) (0.235) Profit/Sales 1.945** 1.889** 1.962** 1.895** 3.068*** 2.933*** (0.986) (0.933) (0.943) (0.895) (0.823) (0.871) Wages/Sales 3.117** 3.205** 3.200** 3.295** (1.507) (1.521) (1.438) (1.425) (1.822) (1.857) Govt Seat Share 0.694* 0.821** 0.985*** (0.368) (0.338) (0.369) Abs Seat Share Difference 0.727* 0.965* 1.437*** (0.408) (0.522) (0.350) Ln (Per Capita Income) (0.458) (0.440) Per Capita Income Growth (1.990) (1.956) Literacy rate 0.020** 0.019** (0.008) (0.008) Leftist Seat Share (0.748) (0.763) Industry FE Yes Yes Yes Yes Yes Yes Number of Firm Years

32 Control Transfer Privatizations (Table 8) (1) (2) (3) (4) (5) (6) (7) (8) Ln(Sales) 0.207* * * 0.205* 0.221** (0.114) (0.124) (0.107) (0.123) (0.121) (0.114) (0.106) (0.101) Profit/Sales (0.556) (0.580) (0.510) (0.574) (0.568) (0.519) (0.588) (0.440) Wages/Sales (0.818) (0.813) (0.742) (0.782) (0.839) (0.773) (0.721) (0.796) Govt Seat Share (0.572) Opposition Seat 1.734* Share (0.922) Seat Share Difference (0.447) Abs Seat Share Difference (0.589) Opposition Vote Share (1.282) Vote Share 2.568* Difference (1.580) Abs Vote Share 3.855** Difference (1.640) Industry FE Yes Yes Yes Yes Yes Yes Yes Yes Number of Firm Years

33 Alternative Speci cations (Table 9) With Sales Quintile Dummies Controlling for Listing Requirements Firm Specific Variables not Winsorized Ln(Sales) 0.694*** 0.621*** 0.876*** 0.817*** 0.610*** 0.591*** (0.204) (0.210) (0.177) (0.156) (0.109) (0.108) Profit/Sales 1.717* 1.650* 2.982*** 2.888*** (1.035) (0.976) (1.091) (1.050) (0.031) (0.031) Wages/Sales 2.730* 2.781** 6.963*** 7.116*** 4.151*** 4.135*** (1.456) (1.418) (2.595) (2.514) (1.328) (1.349) Govt Seat Share 0.659* 0.830** 0.670** (0.354) (0.388) (0.328) Abs Seat Share 0.814* 1.125*** 0.901** Difference (0.433) (0.402) (0.451) Industry FE Yes Yes Yes Yes Yes Yes Number of Firm Years

34 The Role of Political Patronage (Table 5) Privatized Main Operations in Home State No Yes Total No Yes Correlation 0.118*

35 E ect of Privatization on Firm Performance (Table 10)

36 Summary of Results A rm is less likely to be privatized if its main operations are located in a state where The governing party faces a close race with opposition parties. The governing party did not win a large share of the seats. No rms located in the home state of the Cabinet Minister in charge of the rm are privatized. Correcting for endogeneity of the privatization decision using political variables as IVs we nd that privatization has a positive e ect on rm performance.

37 Questions?

38 Political Competition by State (Table 3)