2. Types of corruption

If corruption is after all damaging to growth, why did it not seem to impair the growth of corrupt East Asian countries? One possible response is that East Asia would have grown even faster had levels of corruption been lower. Another possibility is that East Asia suffered from a particular kind of corruption that may have been less damaging for growth. It has often been suggested that all corruption is not alike, that different kinds of corruption have different effects. This section presents a classification of types of corruption and some tests of whether they indeed have different effects on levels of investment.

Table 5.6 shows different types of corruption that have been suggested in the literature. It turns out that they are not mutually exclusive but that each addresses a particular aspect of the corruption problem.

The first distinction, drawn by Goudie and Stasavage (1997), is between political and bureaucratic and administrative corruption. Most observers would, for example, agree that when a customs officer demands a bribe for letting a product enter the country, this is corruption. The question is complicated if the exchange is in kind and perhaps involves long-term mutual favors or gifts. However, this definitional problem can be overcome, and in fact many governments establish rules on the size of gifts and the amount of "wining and dining" that public officials may accept. The question becomes more difficult when corruption is mixed with political patronage or favoritism for electoral reasons. Is it corruption when a politician votes on whether or not to devalue his nation's currency based solely on whether such a move will suit his political supporters? The general definition of corruption as "the use of public office for private gain" (e.g., Bardhan 1997) would suggest that this is indeed corruption. But most people would agree that the situation of the politician cannot be compared with that of the customs officer and that the two types of corruption should not be combined in one single indicator. However, it is not clear from Goudie and Stasavage (1997) whether this distinction is relevant in terms of economic performance.

A related distinction is that between large-scale and petty corruption. The typical example of petty corruption is the police officer who demands a bribe for not enforcing a speeding ticket or the public official who takes a "tip" for accelerating a procedure. Large-scale corruption typically occurs in government procurement or tax offices. It seems useful to distinguish between the two forms mainly because they may potentially have different economic effects. Small-time corruption may act more like a tax or a service fee while large-scale corruption may cause more significant distortions and uncertainties. The implication is that large-scale corruption would be more damaging to economic performance than petty corruption.

Another distinction is that drawn by Shleifer and Vishny (1993), who distinguish between corruption without theft and corruption with theft. The former could be said to happen if an official charged a bribe in addition to, for instance, an import license fee, but then passed the license fee on to the state treasury. In the latter case the official would just let the goods enter the country without paying duty in exchange for a bribe. Obviously this distinction is mostly relevant from the point of view of the treasury.

Table 5.6. - Different types of corruption

Political corruption	versus	Bureaucratic and administrative corruption
Large-scale	versus	Petty corruption
Corruption with theft	versus	Corruption without theft
Extortive corruption	versus	Collusive corruption
Corruption that is predictable	versus	Corruption that causes uncertainties

In Brunetti and Weder (1998) we draw a distinction between extortive and collusive corruption. Extortive corruption is one-sided and can occur when the official uses her discretionary power to extract a payment, for instance by threatening to apply a specific interpretation of the law.⁶

6. Klitgaard (1991) provides examples of both types of corruption in his case study of the Philippine tax system.

Collusive corruption is two-sided; that is, the official uses his discretionary power to offer a deal that benefits both himself and the client. In collusive corruption the private agent cooperates in the corrupt act and always pays the bribe. The relevance of this distinction is that collusive corruption is much more difficult to fight than extortive corruption because the arrangement is beneficial for both parties and neither has an incentive to expose it.

A last distinction is between corruption that is predictable and corruption that causes uncertainties. This distinction was already briefly discussed in Chapter 1 because it is probably the most useful from the point of view of economic performance.⁷ The implication is that predictable corruption is less damaging than corruption that causes uncertainty because of the fundamental difference between a certain cost and an uncertain cost. For the private investor there is not much difference between a transaction tax and a bribe she knows she has to pay. A transaction tax enters the investment decision as an additional cost and reduces the net present value of the project. If the net present value of the project remains positive the investment will go forward. However, if there is uncertainty about the costs and the investment decision involves at least some irreversibility, any uncertainties may have a significant impact. The theory of irreversibility has shown that uncertainty about the future creates an option to wait without committing resources and therefore reduces levels of investment much more dramatically.

7. Recall that Shleifer and Vishny (1993) suggested that whether corruption is predictable or uncertain depends on the internal organization of the bureaucracy. A bureaucracy in which corruption is organized in a centralized fashion behaves like a monopolist. It charges the optimal tax rate, that is, the tax that maximizes total bribe income. In order to accomplish this goal, rules for sharing bribe proceeds within the bureaucracy must be established and individual officials must be prevented from trying to demand additional private bribes on the side. In contrast, in a bureaucracy in which corruption is disorganized the individual officials all act independently and end up competing against one another for bribe returns. The total bribe return may be higher in the case of a monopolistic organization, but the level of uncertainty of private actors is certainly higher in the case of disorganized officials.

The empirical problem is that most corruption data are aggregate and do not allow one to distinguish among different types of corruption. Moreover, it is not clear what is subsumed within these aggregates. They might well be comparing apples with oranges, and this might be a reason why overall corruption measures show ambiguous results in cross-country growth regressions. The only data set that tries to distinguish among different forms of corruption is that of the WDR +, which includes a number of questions aimed at measuring unpredictability in corruption.⁸

8. See Chapter 2 for a detailed discussion of these questions.

What do these indicators say about the type of corruption in East Asia? Is it true that in East Asian countries the uncertainty element in corruption is lower, even though the overall level of corruption might be high? There is little evidence of this. Rather it seems that in those countries with a high level of corruption the level of uncertainty also tends to be high.

Table 5.7. - Corruption uncertainty in investment regressions

Dependent variable: Average investment rate, 1972-95

Independent variable	(1)	(2)	(3)	(4)
Constant	15.04 (3.25)	11.38 (2.55)	1.69 (0.36)	2.05 (0.40)
GDP70	-0.0001 (-0.31)	1.03 x 105 (0.02)	-0.0002 (-0.50)	-0.0002 (-0.44)
Sec70	23.03 (3.89)	22.45 (3.66)	21.60 (3.63)	23.52 (3.94)
Uncertainty of corruption frequency	-4.08 (-1.27)
Predictability of bribe size		-066 (-0.48)
Risk of blackmailing			2.13 (1.71)
Discretion of bureaucrats				2.38 (1.47)
N Adjusted R	52 0.49	51 0.47	51 0.50	51 0.50

How about in larger country samples? Based on the argument advanced previously, these indicators of corruption uncertainty should be very closely associated with economic performance - even more so than the indicators of the overall level of corruption. As shown in Chapter 3, however, they were only very weakly correlated with growth. This leaves open the question of whether they might be more closely associated with investment levels. After all, the indicators of the overall level of corruption have been shown to be more closely associated with investment than with growth. Moreover, the theories of irreversible investment are investment theories, not growth theories. Table 5.7 presents the results from minimal and extended investment regressions on four corruption uncertainty indicators, and Table 5.8 summarizes the results.

Table 5.8. - Summary test results for different corruption uncertainty measures in investment regressions

Variable	Coefficient has expected sign?	Level of significance in minimal specification	Level of significance in extended specification	N
Uncertainty of corruption	Yes	-	-	50
Predictability of bribe size	No	-	-	50
Blackmailing	Yes	*	**	50
Discretion of bureaucrat	Yes	-	-	51

Note: Coefficient from growth regressions that include as further control variables the initial level of GDP per capita, the initial rate of secondary school enrollment (minimal specification), the contribution of government consumption to GDP, the contribution of imports plus exports to GDP, and the average inflation rate (extended specification). ***, Significant at the 1 per cent level; **, significant at the 5 per cent level; *, significant at the 10 per cent level. -, Not significant at the 10 per cent level. If there are two entries in the expected sign column, the first one refers to the minimal specification and the second to the extended specification.

The results are not very encouraging for the proposition that corruption uncertainty is more detrimental to investment than other forms of corruption. Out of four indicators tested, only one was significant and of the correct sign. However, the most plausible conclusion is that the limitations of available data sets are too large to allow for proper testing of the hypothesis. For instance, Wei (1997a, b) obtains somewhat stronger results by using corruption uncertainty measures in regressions with foreign direct investment as the dependent variable. Overall, it appears that the question of whether different types of corruption have systematically different effects on economic performance remains open to further research.