1 Why Do Government Banks Perform Worse? —A Political Interference View Chih-Yung Lin Chung-Hua Shen.

1

Why Do Government Banks Perform Worse?

—A Political Interference View

Chih-Yung Lin Chung-Hua Shen

2

Outline:

1. Introduction

2. The Data

3. Methodology and Empirical Results

4. Conclusions

3

1. Introduction

Main Question:

Why the government-owned banks (GOBs) perform worse than private-owned banks (POBs)?

44

GOB effect GOB effect : Government-owned banks

under-perform the private-owned banks.

Prior empirical studies confirm this argument in different :

1. profitability measures 2. regions 3. sample periods

See Beim and Calomiris (2001), Megginson (2005) , Mian (2003), Iannotta et al. (2007), Cornette et al. (2008).

55

Three views of GOB effect Sapienza (2004) propose three views, the

social, agency and political views, to explain GOB effect.

1. Social and agency views: suggest that the government banks are created to maximize social welfare, not to maximize profit.

2. Political view: suggests that GOBs are a mechanism for pursuing the individual goals of politicians.

we focus on this view

6

Difficulty of studying political interference Lack of any operational definition:

In the literature,

1. election years: are often used to measure the political interferences on GOBs

For example, see Brown and Dinç (2005), Dinç (2005) and Micco et al. (2007).

2. the political connections with the ruling party: See Sapienza (2004)

The definition of political interference The definition : the executives of government banks are

replaced within 12 months after the presidential elections

(hereafter executive turnover).

the executive: includes the CEO or the chairman of the

board in a bank.

7

8

< Political Interference Hypothesis > H1:

This hypothesis firstly suggests that once government banks are undertaking political interference, their financial performance declines.

That is, 1. Political banks : the worst performance Government banks undertaking political interference

2. Non-political banks : the middle performance Government banks undertaking no political interference

3. Private banks : the best performance Private banks are served as our benchmark.

9

< Political Interference Hypothesis >

H2: These influences of political interferences are

much larger in developing countries than in developed countries.

H3: If the interference is indeed the reason that

causes the performance declines of government banks, then the declines should be disappear if we remove these political interferences.

10

<The implications of our hypothesis> Three implications:

1. They imply that the performance of two similar government banks may be saw different because they bear different degrees of political interferences.

2. Government banks may have the chance to perform similar as those of private banks if they bear no political interferences.

3. our study can explain the finding of Micco et al. (2007) that government banks underperform private banks in developing countries but not in developed countries.

That is, in developing countries, government banks undertake more political interferences while in developed countries they do not.

11

12

<Our contributions in the literature> Three views:

1.The election year is the macro-level proxy for political interference.

The election year affects all government banks even though only part of banks are requested to increase lending.

We combines the both macro and micro political interference, which checks the executive turnover of each government bank within 12 months after the presidential elections on a case-by-case basis.

13

2. We discuss whether the worse performances are coming from

macro (elections) or macro-micro (executive turnovers during the election years) factor.

Moreover, usually, in countries with high corruption levels, the performance of GOBs is usually worse than that of POBs (Sapienza, 2004; Dinç, 2005; Micco et al., 2007).

So we test whether our results are due to the level of corruption.

All the empirical results support our political interference hypothesis again.

14

<Our contributions in the literature> 3. to ensure that this GOB effect is not due to

the sample selection,

we collect worldwide 329 government banks in 100 countries for the longest sample period 1993 to 2007.

To the best of our knowledge, our study is the most complete collection of government banks.

15

2. The data

1616

Government banks Data

1. Government ownership data : above 20%. Bankscope, world bank, and privatization databases.

2. after considering countries with GOBs: initial sample contains 329 GOBs and 5,501 POBs

from 100 countries.

17

Political Interference Data 1. We first collect the dates of all presidential

elections in 100 countries from 1993 to 2007.

2. Before 2003: Bankscope provides only a limited number of name lists

of GOB executives. Thus, we have to collect the name lists from company websites, local newspapers, Wall Street Journal, and Factiva database.

3. After 2003: Bankscope provides more complete names of

the most of the government banks.

18

Data of Political Interference 1. we obtain about only 15% of the name lists

for 329 GOBs from 1993 to 2002

2. 80% of the names lists during 2003~2007 from Bankscope and other sources.

3. To avoid the small sample problem, the latter sample size (2003~2007) is used.

4. We identify 80 political GOBs and 249 non-political GOBs

19

20

21

22

23

24

3. Methodology and Empirical Results

25

26

27

Empirical Results 1. we study whether the GOB effect exists.

2. we study whether GOB effect exist in developing countries but not in developed countries.

3. we study the Political Interference Hypothesis.

Through our paper, the proxies for performance variable include ROA, ROE, NIM and NPL.

28

1. GOB Effect (1993~2007) First, we study whether the GOB effect exists.

Confirm: ROA, ROE, and NPL Not confirm: NIM

29

30

2. GOB Effect in DCs versus LDCs(1993~2007) Next, we study whether GOB effect exist in developing

countries but not in developed countries.

DCs: Confirm: NIM Not confirm: ROA, ROE, and NPL

LDCs: Confirm: ROA, ROE, and NPL Not confirm: NIM We find GOB effect exists in LDCs but not in DCs.

31

32

3. Political Interference Hypothesis

Political banks: α 8 and α10

Non-political banks: α 7 and α9

GOB effect in Political banks LDCs: Confirm: ROA, ROE, and NIM Not confirm: NPL =>H1: This hypothesis firstly suggests that once government

banks are undertaking political interference, their financial performance declines

DCs: Not confirm: ROA, ROE, NIM, and NPL =>H2: These influences of political interferences are much

larger in developing countries than in developed countries. 33

34

GOB effect in Non-Political banks DCs: Confirm: NIM Not confirm: ROA, ROE, and NPL

LDCs: Confirm: NPL Not confirm: ROA, ROE, and NIM

=>H3: the declines should be disappear if we remove these

political interferences.

35

36

The Election year

1. The underperformances of GOBs are enlarging during the presidential elections.

=> consistent with the literature

2. after controlling the election variable : the political interference hypothesis is more pronounced

in LDCs than in DCs by using political banks. => consistent with Political Interference Hypothesis .

37

38

The Level of Corruption

1. after controlling to the level of corruption : the political interference hypothesis is more pronounced

in LDCs than in DCs by using political banks.

That is, the GOB effect is not just due to the level of corruption but it is due to our political interference proxy.

=> consistent with Political Interference Hypothesis .

39

40

Robustness Testing 1. Using Non-Political Banks

2. Using GOB ownership of 50%

3. Country Governance

4. Bank Fixed Effect with Clustering Standard Error

=> consistent with Political Interference Hypothesis

41

42

43

44

4. Conclusion

45

4. Conclusion

This study shows why government-owned banks perform not worse than private-owned banks from 100 countries during 1993~2007.

We propose a Political Interference Hypothesis to explain this fact.

1. we study whether the GOB effect exists.

Confirm: ROA, ROE, and NPL Not confirm: NIM

We find GOB effect exists.

46

2. we study whether GOB effect exist in developing countries but not in developed countries

DCs: Confirm: NIM Not confirm: ROA, ROE, and NPL

LDCs: Confirm: ROA, ROE, and NPL Not confirm: NIM

We find GOB effect exists in LDCs but not in DCs.

47

GOB effect in Political banks In developing countries, we find GOB effect exists in

Political banks.

LDCs: Confirm: ROA, ROE, and NIM Not confirm: NPL

=>H1: This hypothesis firstly suggests that once government

banks are undertaking political interference, their financial performance declines

48

GOB effect in Political banks In developed countries, we find GOB effect exists in

Political banks.

DCs: Not confirm: ROA, ROE, NIM, and NPL

=>H2: These influences of political interferences are much

larger in developing countries than in developed countries.

49

GOB effect in Non-Political banks We do not find GOB effect exists in Non-political banks.

DCs: Confirm: NIM Not confirm: ROA, ROE, and NPL

LDCs: Confirm: NPL Not confirm: ROA, ROE, and NIM

=>H3: the declines should be disappear if we remove these

political interferences.

50