Transparency and Health

Transparency and Health as Complementary Public Goods

James Stodder and Peter W. Schroth

Rensselaer Polytechnic Institute*

Presented at the XIII World Congress of the International Economics Association;

Lisbon, Portugal, September 2002.

ABSTRACT: The social basis of corruption helps resolve a paradox: that a rationally selfish state has the incentive to limit corruption, but may not have the means. "Transparency" (lack of corruption plus openness to public scrutiny) cannot be provided by government alone. Transparency depends upon other basic public goods that, like transparency itself, require voluntary inputs by private households. Health is a "socially produced" public good, at least in part. While transparency has already been shown to promote public health outcomes, we present evidence that the reverse causality also holds, so that health and transparency are mutually reinforcing. It is argued that anti-corruption incentives in poor countries must be linked with (a) investment in basic public infrastructure and (b) retooling business and government institutions toward traditional forms of social-reciprocal enforcement of honesty, such as rotating savings and credit associations (ROSCAs).

Journal of Economic Literature codes: K420, F230, O170

1. Introduction

We do not want to know if life improves when Togo becomes Denmark; we want to know if life improves when a poor Togo becomes a richer Togo.

William Easterly (1999)

There is widespread agreement that corruption is bad for growth; why, then, are many countries that desire growth nevertheless so corrupt? Identifying the self-interests of kleptocratic elites (Charap & Harm 1999, Olson 2000) provides only a partial answer, because there are many more parties to corruption than the governmental and business elites. For large scale corruption to be effective in its own terms, there must also be widespread cooperation by middle class professionals, small business owners, and probably even most urban families.

Corruption is not merely the strong taking from the weak; it is a public bad, a culture of corruption maintained by daily actions, and by millions of willing participants. This much is well known to most people, but what is not so obvious is the logical connection between the publicness of the corruption bad and the difficulty in devising effective incentives against it, especially in developing countries.

One aim of this paper is to explain why anti-corruption incentives are so difficult to implement. We begin with the paradox that rationally self-interested state actors will wish to end widespread corruption, which is demonstrably a terrible barrier to growth. Corruption, therefore, should be strictly limited even by a self-interested state, one that is, in the memorable formulation of Mancur Olson (2000), a "stationary bandit," maximizing the economic surplus it can extract. It is elementary to show that decentralized and uncontrolled graft makes a state "weakly corrupt," i.e., too corrupt for its own selfish good: charging multiple corruption "taxes" that are too high to maximize its total tax revenue.[1] The obvious fact that such states are too weak to reform themselves – too corrupt to rationalize their own corruption – is a hint that the sources of corruption may be beyond direct state control.

Transparency (the relative absence of corruption) is a public good analogous to clean air or water (the relative absence of pollution). As such, its provision can be expected to be closely tied to the success of a society in providing other basic public goods, such as education, health, and economic security. The new political economy literature models the causality between corruption and these other goods as simple and unidirectional: corruption slows down growth, while both corruption and slower growth damage these pubic goods. We demonstrate that some of these public goods and bads are bound together in a simultaneously determining relationship, part of a socio-cultural whole. The way in which they are connected has important implications for development and for anti-corruption policy.

To be clear, we agree with the dominant anti-corruption literature (Gupta et al. 1998, 2000): transparency directly promotes health, education, and economic security, even after accounting for its indirect benefit through faster growth. But that is only half the story. We further show that the reverse is also true, that improvements in at least one of these public goods, health, also encourage transparency. Thus, causality is not unidirectional; it is simultaneous. The econometric methodology for doing this is a straightforward instrumental variable approach, the same as that used by Gupta and many others in the literature.

We argue that human capital investment into traditional community networks is an effective way of preventing "corruption" within the networks themselves; i.e., to prevent cheating on fellow community members. A key aspect of the problem of corruption in modern bureaucracies, in business and government, is the impersonalization and separation of these structures from traditional community networks.

Corruption can be prevented only by large scale investment in the public goods of enforcement and monitoring, by traditional social network monitoring, or, in most cases, by institutions that have evolved out of compromises between the two. The Securities and Exchange Commission (SEC) in the United States, for example, grew out of the self-policing organizations of stockbrokers themselves.

It is possible to find traditional social and modern public forms of providing these public goods. The Grameen Bank "micro-financing" institution (see www.grameen-info.org) is perhaps the best-known traditional analogy to the modern SEC example, but a host of other examples have emerged in developing countries, and now constitute a veritable micro-finance revolution (Beasley et al. 1993, De Soto 2000). The World Bank now supports a network of micro-financing programs around the world.[2] These programs seek to legalize and securitize the vast informal sector in all the cities of the developing world, where De Soto estimates as much as half their population earns its living. To the extent that modernization destroys traditional social networks, however, the process of monitoring corruption is made that much more difficult. We show that corruption can be made worse if other basic public goods are in short supply. The old sources of such goods, networks of traditional reciprocity, have been undermined by modernization. At the same time, their modern replacements, the state and market, have not had much success providing such public goods.

2. The Paradox of Persistent, Immiserizing Corruption

That corruption is negatively correlated with growth is easy to show. The "new economics of growth" literature shows a strong positive correlation between economic growth and most indicators of human happiness: health, education, democracy, and transparency (Barro 1997). Another intellectual current, "the new political economy of corruption" literature, has shown that corruption is negatively correlated with growth, and hence also inimical to the progress of human happiness. (See Mauro 1995; Lopez & Mitra 2000; Gupta et al. 2000.)

There is now a substantial body of published research into the effects of corruption. For instance, Maoro (1997, p. 91) found a strong negative correlation between perceived corruption, on one hand, and GDP and investment, on the other:

This means that if a given country were to improve its corruption “grade” from 6 out of 10 to 8 out of 10, its investment-GDP ratio would rise by almost 4 percentage points and its annual growth of GDP per capita would rise by almost half a percentage point.

Generally similar conclusions were reached by Burki & Perry (1998), Hines (1995), Wei (2000), and Rose-Ackerman (1978, 1997, 1999).[3]

These findings are particularly important when linked with recent estimates by Reisen & Soto (2001) on the relation between foreign investment and growth. Using a panel data set of 44 developing countries over the period 1986 to 1997, these authors find that both foreign direct investment (FDI) and foreign portfolio investment correlate positively and significantly with long term GDP growth. As we shall see, panel data sets allow us to estimate directly the effects of change within countries over time, not just differences between countries, at more or less the same time.

Despite the conclusion these of studies – that a government curtailing corruption should reap large payoffs in growth and welfare – effective anti-corruption incentives are weak to non-existent in much of the developing world. After persistent prodding by the Clinton administration, the Organization of American States (OAS),[4] the Organization for Economic Cooperation and Development (OECD),[5] and the European Union[6] have moved toward institutionalizing anti-corruption standards similar to those in the United States Foreign Corrupt Practices Act (FCPA) of 1977.[7]

With the notable exceptions of the OAS Convention and the Council of Europe’s Criminal Law Convention, there has been little anti-corruption action by developing countries, but very recent events include hopeful signs. For example, no African country has or ever had a law criminalizing foreign bribery, as far as we have been able to determine (Schroth & Sharma 2002), but the Prevention of Corruption Bill, introduced by the Government in South Africa in April 2002, not only will reach that but, if adopted in its current form, will be, in many respects, the harshest and most sweeping anti-bribery law of any country in the world.[8] No African country is a party to any anti-bribery convention, but South Africa is discussing possible accession to the OECD Convention. Also, one of the last acts of the Organization of African Unity, before it was replaced by the African Union, was a proposed OAU Convention on Combating Corruption; although it does not appear to be a priority at present (and has not yet appeared on the new African Union web site), with a bit of help from the OECD, it may in time be signed and ratified.

The OECD Convention is designed for investor countries – the payers of bribes – not for the host countries in which they are paid, so most of Africa doesn’t qualify. Why then do most countries find it so hard to take measures that are in the economic interests of the great majority? And why is this especially true for the poorest of the poor?

3. Transparency as a Public Good

a. A Socially Produced Public Good

That private spillovers, externalities, can have public consequences for good or ill is one of the staples of economic theory (Heal & DasGupta 1980). Insofar as property or liability rights have not been established to adequately cover and encourage market-based voluntary transfers among these spillover goods and bads, the efficiency implication is very clear. Receiving no market sanction, negative spillovers will be overproduced; receiving no market reward, positive spillovers will be underproduced – in each case, relative to an efficient level of provision. Although policy prescriptions differ widely, this basic argument is entirely uncontroversial among economists, even staunch libertarians (Friedman 1982).

Transparency is self-evidently a public good. It is a good because most of the public benefits from transparent institutions in the form of lower transaction costs and reduced risk. It is a classically public good because my use of this benefit does not “use up” the amount of it available to anyone else, and thus will not affect its price. In fact, my use of transparent procedures may, in some cases, even increase the amount of transparency available to everyone else. In particular, this public good of transparency is created primarily as the positive spillover of millions of “private” acts. That is, transparency is usually enforced by accounting standards and by laws. But it can be maintained only by the largely voluntary or habitual compliance of individuals, including accountants, lawyers, and official legal guardians.

It is virtually always the case that the transparent act confers benefits extending beyond the individuals who practice it. It not only sets standards and expectations among others who will benefit from the good provided, but in some cases may directly and significantly reduce corruption costs to many other people. The conferring of external benefits becomes problematic when the private cost to the individual who produces the transparent practice outweighs the private benefit accruing to him or her. This may be the case because, when transparency is not well established, insistence on such procedures may entail substantial economic and personal risks.

Transparency is far from unique in being a public good that is primarily produced by voluntary provision within long-term social networks, instead of by direct governmental production, in the way that a government builds, say, a hydro-electric dam. [9] Most public goods can be and are produced as by voluntary acts within social networks and communities, both for self-interested reasons of long-term reciprocity and for emotional commitments to values and to other human beings. (See Frank & Scitovsky 1992; Axelrod & Cohen 2000.)

The implications for anti-corruption policy are straightforward in theory, if difficult in their policy implications. If transparency is a public good produced by both governmental provision and voluntary social networks, it should be correlated with the society's success in providing other public goods meeting basic human needs, especially those also produced partially by diffuse social networks. This paper examines the correlation between corruption and another such (partially) public good: health care. Health care is indeed positively correlated with transparency, at least in this cross-section of countries. (We also tested the relationship between measures of Education and Transparency, and found an empirical correlation quite similar to that between Health and Transparency. Our results were short of statistical significance, however, and will not be presented here.)

If health care is both a product of, and a contributor to, the “social capital” accumulated within voluntary social networks, then its provision should not only be correlated with the provision of other public goods, such as transparency, but also should help determine the levels of their provision. This is the argument about causality that we address in the econometric methods of within-country panel estimates, discussed in the next section.

First, let us look at a simple model of cost-complementarity in the provision of health care (H) and transparency (T), and two individuals (1, and 2). The demand for H, a straightforward public good, is found by the summing of the demands by both 1 and 2:

Ph1 = a - bH, Ph2 = c - dH => Ph = Ph1 + Ph2 = (a+c) - (b+d)H.

The demand for T is formed similarly:

Pt1 = e - fT, Pt2 = g - hT => Pt = Pt1 + Pt2 = (e+g) - (f+h)T.

H and T are cost complements in the sense that the more of one is produced, the cheaper it is to produce the other. Simple cost functions would be as follows:

C(H) = j + kH + mH² - nHT => MCh = (k -nT) + 2mH, and

C(T) = q + rH + sH² - tHT => MCt = (r -tH) + 2sT.

Below is a simple graph of the supply and demand for transparency:

Note that for parameters such as shown above, the relation between the provision of health care (H) and transparency (T) is:

a) If there is free riding, but with full provision of H, then T will be undervalued by society and move from E1 to E3, even though its cost of production is correct.

b) If H falls, but with no free riding (so we have both layers of demand), this pushes T provision from E1 to E2 – i.e., from an intercept as low as r-tH, to one as high as r. Transparency is now being correctly valued by demand, but has become too costly to provide adequately.

c) If there is both free riding and non-provision of the complementary good H, then transparency may disappear altogether, moving from E1 to intercept e. Transparency is both not adequately valued by societal demand, and has become too costly to supply.

The policy recommendations of this argument are fairly straightforward. For a set of public goods which are cost complements, “free riding” is not just one good at a time. Under-provision of any one of the interlinked public goods can have drastic consequences on many others – even if there is no free riding on these other public goods by themselves – as in case b) above. An anti-corruption strategy, therefore, cannot concentrate on corruption alone, and certainly not just on punishing it. It must rely on building up the traditional social networks that can and do maintain, at least to some degree, both transparency and other public goods, such as health. And it must recognize that, to the extent modernization and economic growth dismantle and dissolve these traditional values and networks, modernization will promote rather than diminish corruption.

This debilitating backwash of modernization is now taken seriously by development economists and anthropologists studying the social capital invested in traditional organizations such as Rotating Savings and Credit Associations (“ROSCAs”). ROSCAs are found throughout the developing world (Beasley et al. 1993, Stodder 1995, De Soto 2000), and there is an impressive record of projects like Grameen Bank that have built upon them, rather than trying to surpass them. Grameen Bank has a rate of loan repayment better than that of many commercial banks in the developed world. Therefore, a policy of anti-corruption should concentrate on:

a) providing basic needs public goods, such as health, education, economic security, and public safety, both directly and through traditional social networks.

b) shoring-up those traditional social networks themselves, by promoting business, governmental and communal forms that are based on these forms of accumulated social capital, rather than predicated upon their dismantling and dissolution.

These points are re-examined in the conclusion.

b. Variation Within (as Opposed to Between) Countries: Income versus Transparency and the Rule of Law

With the exception of the Reisen & Soto study, the papers cited above are based on cross-sectional data tests. A widely-cited paper by Maoro (1997), for example, is sub-titled, "A Cross-Country Analysis." To test the emerging consensus on growth, Easterly (1999) constructed a large international panel data set on income growth and standard indicia of progress under development – transparency, health, and education.

Easterly argues that it is highly misleading to assume that this variation between rich and poor countries provides a map for income growth within any single country:

The bad news is that other key modernization or development indicators like democracy, good institutions, human rights, years of schooling, school enrollment ratios, and life expectancy do not robustly improve with income controlling for country effects and sometimes even have the wrong sign. (Easterly 1999, p. 16.)

The present paper builds upon Easterly's findings. We use his panel data to show how corruption interacts with other quality-of-life variables, such as health. Testing for within country variation, we show that improvements in health can improve transparency, independently of the negative effect of income upon transparency. That is, even after accounting for the positive relation of income upon the health, the effect of higher income upon freedom from corruption (transparency) remains negative.[10]

Why does it not work, this seemingly reasonable use of cross-sectional variation to map the historical trajectory of development? One depressing possibility is that variations between countries that are largely fixed and immune to policy even in the reasonably long run may explain most cross-country variation in development. Easterly & Levine (1997), Sachs & Warner (1997), and other empirical investigators have suggested that fixed factors (such as access to the sea, location in the tropics, ethno-linguistic fragmentation, and resource abundance) may explain more between-country income variation than any policy-relevant variables.

It is also true that transparency and health are higher in the richer countries, and their correlation can be shown to be highly significant. This leads to the “virtuous circle” argument based on cross sectional evidence. But, as will be seen from the panel data estimates, the within-country variation between income and transparency is also significant, and strongly negative.[11]

4. Empirical Estimates: The Problem of the Direction of Causality; Health Care and Corruption

Here we wish to assess accurately the effect of both health and income on corruption. Easterly shows in Tables 1 and 2 (and we do not repeat his regressions showing) that:

1) Higher income improves life expectancy between countries, but harms it within countries. Higher income tends to improve infant mortality, both between and within countries.

2) Income growth improves transparency and rule of law between, but harms them within, countries.

In contrast to 2), we show that higher life expectancy and lower infant mortality improve both transparency and rule of law, both within and between countries. Because income pushes healthcare up but transparency down, the interesting question is their joint effect on transparency. Here we follow the standard practice of the literature (Pritchett & Summers 1996), and model income as more determinative of health care than vice versa. A truly complete model would test for co-determination here, and our estimation procedure in no way rules this out. Lacking appropriate data "instruments" to determine income growth, however, our instrumental variable approach will look merely at "one side," the effect of income on health care.

We begin applying instrumental variables to our health care variables – life expectancy and infant mortality. To begin with, we must model the determination of life expectancy as dependent on both income and other instrumental variables: length of paved highways, and cars per capita. That ordinary least squares (OLS) regression is shown in Table 1 below:

TABLE 1. Dependent variable: LIFE EXPECTANCY, Observations = 107

Ordinary Least Squares (OLS) Estimates

Mean dependent variable	62.857	R-Squared	0.742
Standard Error Regression	10.421	Adjusted R-Squared	0.729

Variable		Coefficient	T-statistic	P-value
Constant		46.654	31.617	[.000]
Sanitation		0.165	6.878	[.000]
Highway Distance		2.08E-02	0.950	[.344]
Cars per Capita		-8.738	-0.904	[.368]
Per Capita Income		1.40E-03	3.407	[.001]
CO² per Capita		-0.257	-0.950	[.344]

Note that there is likely multicollinearity or functional dependency between variables here: for example, between Cars per Capita and CO² per Capita. If we were concerned about estimates of the coefficients, this would be a problem. Because we are interested only in generating a "fitted" form of life expectancy, however, multicollinearity is of no concern.

Next, we show the estimates of the between-country and within-country effects of life expectancy upon transparency. The between- and within-country equations are shown here as (1a) and (1b), relating to the Tables 2a and 2b, respectively:

Transparency _i = Constant + b*(Life Expectancy _i ) + c*(Per Capita Income _i) + ε_i (1a)

Transparency _it = Constant_i+ b*(Life Expectancy _it ) + c*(Per Capita Income _it) + ε_it (1b)

where i = (1, 2, …, C) the number of countries, and t = (1, 2, ..T), the number of time periods observed. Note that in the following data, the numbers of countries is at most 106, and the maximum number of times periods is 2 (1980 and 1990).

TABLE 2: Dependent variable: TRANSPARENCY, O = 107, C = 70, T = 2, P = 37,

where O = # Observations, C = # Countries, T = Maximum # of Time Periods, P = # of Paired observations (number of countries with both time periods observed.

TABLE 2a. BETWEEN COUNTRY Estimates (OLS on Country means, 70 Observations):

Mean dependent variable	3.400	R-Squared	0.543
Standard Error Regression	1.648	Adjusted R-Squared	0.530

Variable		Coefficient	T-statistic	P-value
Constant		-1.987	-0.947	[.347]
Life Expectancy (fitted)		7.63E-02	1.976	[.052]
Per capita INCOME		1.23E-04	1.647	[.104]

TABLE 2b. WITHIN COUNTRY Estimates (Fixed Effects, 107 Observations):

Mean dependent variable	3.327	R-Squared		0.943
Standard Error Regression	1.753	Adjusted R-Squared		0.827
Durbin-Watson Statistic		2.00	[1.00]
Variable		Coefficient		T-statistic	P-value
Constant _i		–		–	–
Life Expectancy (fitted)		0.110		2.298	[.028]
Per Capita INCOME		-3.63E-04		-2.654	[.012]

Hausman test of H0: RE vs. FE: CHISQ(2) = 13.337, P-value = [.0012]

In Table 2b's within-country estimates, the coefficient on Life Expectancy is positive and significant, while that on the Income term has turned significantly negative. Comparing this with Table 2a, note that the positive effect of Life Expectancy upon Transparency is seriously understated in these between-country estimates. The coefficient on Life Expectancy in Table 2b is almost twice as high as in 2a, as well as having a higher t-statistic. Similarly, the coefficient on per-capita Income in Table 2b is not only of the opposite sign from that in Table 2a, but is also far more significant.

In the within-country estimates of the effect of income alone on transparency (not shown here), income showed a negative correlation with transparency. The present estimates strengthen that basic conclusion, by adding the "omitted variable" of health care. Because life expectancy is positively correlated with transparency, while income itself is negatively correlated with both transparency and life expectancy, a biasing of the results can be expected to hold when life expectancy is omitted from the equation. Income's negative influence on transparency looks smaller when “flying solo,” without the positive offset of life expectancy, than when life expectancy is included (2b). The coefficient values are |-2.24 E-04| < |-3.63 E-04|, respectively.

Along with the data on transparency, Easterly's data set has observations on the "Rule of Law," both assembled by the IRIS Country Risk Database (University of Maryland). Now it might be thought that transparency and respect for the rule of law are virtually the same thing, and indeed the two series are highly correlated. But there are interesting exceptions that make the difference worth studying.

An authoritarian state may be relatively uncorrupted without having much rule of law. The highly militarized states of Algeria and Israel, for example, show respectable Transparency scores of 4 and 5, but fall to 2 and 1 for their Rule of Law scores, respectively (both scores can take values between 0 and 6). At the other extreme, states may have strong traditions for the rule of law, and still be relatively open to corruption. Portugal and Bulgaria seem to fit this pattern, with the high scores for Rule of Law (6 and 5), but unimpressive levels of Transparency (both scoring 3).

The equations on rule of law have functional forms similar to (1a), (1b):

RuleLaw _i = Constant+ b* Transparency _I + c*(Income _i) + ε_i (2a)

RuleLaw _it = Constant _i+ b* Transparency _it + c*(Income _it) + ε_it (2b)

TABLE 3: Dependent variable: RULE OF LAW, O = 107, C = 70, T = 2, P = 37,

where O = # Observations, C = # Countries, T = Maximum # of Time Periods,

P = # of Paired observations (number of countries with both time periods observed).

TABLE 3a. BETWEEN COUNTRY Estimates (OLS on Country means, 70 Observations):

Mean dependent variable	2.961	R-Squared	0.604
Standard Error Regression	1.800	Adjusted R-Squared	0.592

Variable		Coefficient	T-statistic	P-value
Constant		-0.497	-0.233	[.817]
Life Expectancy (fitted)		3.65E-02	0.931	[.355]
Per capita INCOME		2.39E-04	3.145	[.002]

TABLE 3b. WITHIN COUNTRY Estimates (Fixed Effects, 107 Observations):

Mean dependent variable	2.935	R-Squared		0.977
Standard Error Regression	1.881	Adjusted R-Squared		0.932
Durbin-Watson Statistic		2.00	[1.00]
Variable		Coefficient		T-statistic	P-value
Constant _i		–		–	–
Life Expectancy (fitted)		7.40E-02		2.301	[.027]
Per Capita INCOME		-2.59E-04		-2.811	[.008]

Hausman test of H0: RE vs. FE: CHISQ(2) = 27.407, P-value = [.0000]

Interpreting these results in terms of elasticities, we can take the regressions in Tables 2b and 3b, and evaluate them at the sample means for Transparency, Rule of Law, Life Expectancy and Per Capita Income (3.327, 2.935, 62.86 years, and $4,917, respectively):

Table 4: Life Expectancy versus Income – Effects on Transparency and Rule of Law

Effects of	Coefficient	Elasticity
Life Expectancy on Transparency	0.110	2.078
Income on Transparency	-3.63E-04	-0.537

Life Expectancy on Rule of Law	7.40E-02	1.585
Income on Rule of Law	-2.59E-04	-0.434

These results are rather striking. A one percent increase in Life Expectancy yields better than a 2.0 percent and 1.5 percent improvement in Transparency and Rule of Law, respectively. Those elasticities are almost 4 times as great as the elasticities on income, and in the opposite direction. We conclude that improving life expectancy may be a cost-effective way to improve other public goods linked to corruption, much more so than an increase in income itself.

Turning to another measure of general health, infant mortality, the results are very similar. Infant mortality is a bad, so the negative association of income mortality with income, both between and (not shown) within countries, is clearly a good thing. The OLS result is here:

TABLE 5. Dependent variable: INFANT MORTALITY, Observations = 105

Ordinary Least Squares (OLS) Estimates

Mean dependent variable	61.676	R-Squared	0.755
Standard Error Regression	46.718	Adjusted R-Squared	0.743
Variable		Coefficient	T-statistic	P-value
Constant		133.811	21.102	[.000]
Sanitation		-0.838	-7.821	[.000]
Highways		-0.065	-0.680	[.498]
Cars Per Capita		10.784	0.255	[.799]
Per Capita INCOME		-3.51E-03	-1.925	[.057]
CO² Per Capita		-0.147	-0.118	[.906]

TABLE 6: Dependent variable: TRANSPARENCY, O = 105, C = 69, T = 2, P = 36,

where O = # Observations, C = # Countries, T = Maximum # of Time Periods,

P = # of Paired observations (number of countries with both time periods observed)

TABLE 6a. BETWEEN COUNTRY (OLS on Country means, 69 Observations) Estimates:

Mean dependent variable	3.442	R-Squared	0.586
Standard Error Regression	1.697	Adjusted R-Squared	0.574

Variable		Coefficient	T-statistic	P-value
Constant		3.905	5.047	[.000]
Infant Mortality (fitted)		-0.018	-2.403	[.019]
Per capita INCOME		1.29E-04	2.021	[.047]

TABLE 6b. WITHIN COUNTRY (Fixed Effects, 105 Observations) Estimates:

Mean dependent variable	3.371	R-Squared		0.944
Standard Error Regression	1.783	Adjusted R-Squared		0.830
Durbin-Watson Statistic		2.000	[1.000]
Variable		Coefficient		T-statistic	P-value
Constant _I		–		–	–
Infant Mortality (fitted)		-2.00E-02		-2.190	[.035]
Per Capita INCOME		-3.10E-04		-2.381	[.023]

Hausman test of H0: RE vs. FE: CHISQ(2) = 12.251, P-value = [.0022]

Now consider the relationship between Infant Mortality and the Rule of Law variable.

TABLE 7: Dependent variable: RULE OF LAW, O = 105, C = 69, T = 2, P = 36,

where O = # Observations, C = # Countries, T = Maximum # of Time Periods,

P = # of Paired observations (number of countries with both time periods observed)

TABLE 7a. BETWEEN COUNTRY Estimates (OLS on Country means, 69 Observations):

Mean dependent variable	3.047	R-Squared	0.615
Standard Error Regression	1.851	Adjusted R-Squared	0.604

Variable		Coefficient	T-statistic	P-value
Constant		2.440	2.998	[.004]
Infant Mortality (fitted)		-9.22E-03	-1.173	[.245]
Per capita INCOME		2.36E-04	3.495	[.001]

TABLE 7b. WITHIN COUNTRY Estimates (Fixed Effects, 105 Observations):

Mean dependent variable	3.005	R-Squared	0.979
Standard Error Regression	1.917	Adjusted R-Squared	0.935
Durbin-Watson Statistic		2.000	[1.000]
Variable		Coefficient	T-statistic	P-value
Constant _i		–	–	–
Infant Mortality (fitted)		-1.39E-02	-2.241	[.032]
Per Capita INCOME		-2.21E-04	-2.562	[.015]

Hausman test of H0: RE vs. FE: CHISQ(2) = 25.660, P-value = [.0000]

Similar to the elasticity comparisons of Life Expectancy and Income (Table 5), we can do a computation with Infant Mortality, Income, and their effects on Transparency and Rule of Law. The sample means for these variables are 61.67 (per thousand births), $5,016, 3.371, and 3.005, respectively. Using the coefficients in Tables 6b and 7b, this leads to the following elasticities:

Table 8: infant mortality versus Income – Effects on Transparency and Rule of Law

Effects of	Coefficient	Elasticity
Infant Mortality on Transparency	-2.00E-02	-0.366
Income on Transparency	-3.10E-04	-0.461

Infant Mortality on Rule of Law	-1.39E-02	-0.285
Income on Rule of Law	-2.21E-04	-0.369

The signs on Infant Mortality and Income are in the same direction: both are bads from the point of view of Transparency and the Rule of Law. We see that their elasticities are also comparable, with Income somewhat more “corrupting” than Infant Mortality.

As a final point on our empirical work, we should note that we performed similar estimates on the public goods of Education (years of schooling) and income Inequality (the Gini coefficient, as a public "bad" serving as proxy for lack of income security). The results on Inequality were very far from statistical significance. Tests on our Education variables, however, were just short of statistical significance, and all of the same sign as the tests on our Health variables. These results will not be presented here; the variables themselves are given in the Appendix.

4. Summary and Conclusions

The argument of this paper is that transparency is a public good created largely by capital in social networks, rather than by direct governmental production alone. This hypothesis is supported by within-country estimates, based on extensive panel data, demonstrating a determining influence of health care upon transparency, so that its greater provision encourages greater transparency and rule of law. This does not in any way deny the reverse causality established by earlier literature, i.e., that greater transparency and rule of law also encourage better health care.

Thus, there is a positive determining influence for health care upon corruption, at least in the structural sense of exogeneity. But there still may be no very direct causal link: both improvements might be caused by some third term. So how might better public health improve transparency and the rule of law? There is evidence that good health is highly associated with strong community and family networks (Castles 1994, Cox et al. 1997). Health is a good most people prize highly, so that much leisure time and disposable income are invested in promoting it. The same social networks that promote good health may also promote greater long-term relationships of trust, which is the social basis of transparency and the rule of law.

Mutual aid delivered through community health care may also allow developing countries to leverage traditional forms of social capital used to promote honesty and transparency. The example of Grameen Bank has been widely noted, and is built upon the informal ROSCA networks found throughout the developing world (Beaseley et al. 1993). Literacy and numeracy can only aid such leverage. Another example may be the use of modern communication technologies to connect village school programs with university programs, such as that of the University of South Africa (UNISA), which has been involved in distance education since 1946, and is also used to promote public health. It is now using the Internet to reach out to the world and in the year 2000 had over 8000 students outside South Africa, 6000 of them from other African countries (Schroth & Sharma 2002).[12]

The findings of this paper support of a policy that not only strives to improve the level of basic public goods like education and health, but also attempts to complement and strengthen the social networks responsible for their provision in traditional societies. These traditional values and networks in developing countries now appear, rather than standing as the great roadblocks to transparency and capital investment, as their most secure foundation (Beasley et al. 1993, De Soto 2000).

A program of strengthening these social networks and allying them with the market becomes all the more important in light of the within-country evidence in Easterly (1999) that corruption is encouraged by income growth. Capitalist modernization has been based largely upon dismantling and degrading traditional values and networks. To the extent it has been so based, it has delayed the establishment of transparent institutions, and thus hindered its own development.

Whenever within-country effects of income growth are qualitatively at odds with the cross-sectional evidence, as in Easterly (1999) and the present paper, there are arguments that the cross-sectional evidence remains more fundamental. Recall that our fixed effects estimates explain most of the between country differences by individual country differences, and that the time period of change is typically very short: in the case of this paper, just one decade. It is uncertain, therefore, whether these estimates really capture the most of the socioeconomic changes wrought by sustained income growth. It may be that the “corruption enhancing” effects of growth and poor health care are strictly short-term. Easterly's results would become a mild qualification to the prevailing pro-growth optimism: the negative backwash of growth would then be only a minor friction on the upwardly spiraling “virtuous circle” of income growth.

There are many more questions than can be answered, or even seriously posed, by a paper like this. A closing thought, however, on why social networks should not be ignored: modern capitalism in Western Europe and in Japan was not built upon the simple destruction of traditional social networks. Rather, what made these societies so successful was precisely their ability to mobilize these traditional forms and transform them toward capitalism (Polanyi 1962; Rosser & Rosser 1999). Polanyi, a student of Max Weber, stressed that capitalist social relations in Europe rest upon an underpinning of “bourgeois community values” (in the non-Marxist and entirely approving sense of that phrase), built up over centuries of medieval trade and town life. Without such a broad social basis, the institutions of a well-functioning capitalism are impossible. Development programs must fight and punish corruption, of course. But they must also nurture the social basis of transparency. That social basis will be found in traditional social networks and the norms of reciprocity that support the vast informal sector.

REFERENCES

Axelrod, Robert & Michael Cohen (2000), Harnessing Complexity: Organizational Implications of a Scientific Frontier, New York: Free Press.

Barro, Robert (1997), Determinants of Economic Growth: A Cross-Country Empirical Study, Cambridge (MA): MIT Press.

Besley, Timothy, Stephen Coate & Glenn Loury (1993), “The Economics of Rotating Savings and Credit Associations,” American Economic Review, September 83(4), pp.792– 810.

Burki, Shahi & Guillermo Perry (1998), Beyond the Washington Consensus: Institutions Matter

"Viewpoints," Washington, DC: World Bank.

Castles, Ian (1994), How Australians Use Their Time, Australian Bureau of Statistics, Catalog No. 4153.0, Canberra: Australian Government Printing Service.

Charap, Joshua & Christian Harm (1999), “Institutionalized Corruption and the Kleptocratic State,” IMF Working Paper WP/99/91, www.imf.org/external/pubs/ft/wp/1999/wp9991.pdf

Council of Europe (1999), Criminal Law Convention on Corruption, 27 Jan. 1999, Europ. T.S. No.173.

Cox, Donald, Emmanuel Jimenez & Wlodek Okrasa (1997), "Family Safety Nets and Economic Transition: Worker Households in Poland," Review of Income and Wealth, June: 191-209.

De Soto, Hernando (2000), The Mystery of Capital: Why Capitalism Triumphs in the West and Fails Everywhere Else, New York: Basic Books.

Easterly, William (1999), "Life During Growth: A Compendium of Political, Social and Environmental Indicators of What Gets Better and What Gets Worse from Low to High Income," Journal of Economic Growth 4(3), September, pp. 239-276.

Easterly, William and Ross Levine (1997), "Africa's Growth Tragedy: Policies and Ethnic Divisions," Quarterly Journal of Economics, 112(4), November, pp. 1203-50.

Eliot, Kimberly Ann, ed. (1997), Corruption and the Global Economy, Washington, D.C.: Institute for International Economics.

Engle, Robert F. and David F. Hendry (1983), "Exogeneity," Econometrica, 51(2), March: 277-304.

Frank, Robert H. & Tibor Scitovsky (1992), The Joyless Economy: The Psychology of Human Satisfaction, Oxford: Oxford University Press.

Friedman, Milton (1982), Capitalism and Freedom, Chicago: University of Chicago Press.

Gupta, Sanjeev, Hamid Davoodi & Erwin Tiongson (2000), "Corruption and the Provision of Health Care and Education Services," IMF Working Paper, Fiscal Affairs Department, www.imf.org/external/pubs/cat/longres.cfm?sk=3652.0

Gupta, Sanjeev, Hamid Davoodi & Rosa Alonso-Terme (1998), "Does Corruption Affect Income Inequality and Poverty?," IMF Working Paper, Fiscal Affairs Department, www.imf.org/external/pubs/cat/longres.cfm?sk=2629.0

Heal, Geoffrey & P.S. DasGupta (1980), Economic Theory and Exhaustible Resources, Cambridge: Cambridge University Press.

Hines, Jr., James R. (1995) “Forbidden Payment: Foreign Bribery and American Business After

1977,” NBER Working Paper 5266.

Inter-American Convention Against Corruption (1997) OEA/Ser. K/XXXIV.1, CICOR/doc. 14/96 rev. 2 (29 Mar. 1996), 35 ILM 724 (1996).

Lopez, Ramon & Siddhartha Mitra (2000) "Corruption, Pollution, and the Kuznets Environment Curve," Journal of Environmental Economics and Management 40, 137-150.

Maoro, Paolo (1995), “Corruption and Growth,” Quarterly Journal of Economics, 110(3): 681-712.

___________ (1997), “The Effects of Corruption on Growth, Investment, and Government Expenditure: A Cross-Country Analysis,” in Kimberly, Ann Elliott, ed., Corruption and the Global Economy, Washington, D.C.: Institute for International Economics.

OECD (1998) Convention on Combating Bribery of Foreign Public Officials in International Business Transactions, OECD/DAFFE/IME/BR(97)16/FINAL (18 Dec. 1997), 37 ILM 1 (1998), S. Treaty Doc. No. 105-43.

Olson, Mancur (2000), Power and Prosperity: Outgrowing Communist and Capitalist Dictatorships, New York: Basic Books.

Organization of African Unity (2001), “Draft OAU Convention on Combating Corruption,” Expt/Draft/OAU/Conv/Comb/Corruption (I).

Pritchett, Lant & Larry Summers (1996), “Wealthier Is Healthier,” Journal of Human Resources, 31(4): pp. 841-868.

Polanyi, Karl (1962), The Great Transformation, Boston: Beacon Press.

Reisen, Helmut, and Marcelo Soto (2001) "Which Types of Capital Inflows Foster Country Growth?" International Finance, 4(1).

Republic of South Africa (2002), Prevention of Corruption Bill, Government Gazette No. 23336, 18 April 2002.

Rose-Ackerman, Susan (1999), Corruption and Government: Causes, Consequences and Reform, Cambridge (UK): Cambridge University Press.

___________________ (1997) “The Political Economy of Corruption,” in Kimberly, Ann Elliott, ed., Corruption and the Global Economy, Washington, D.C.: Institute for International Economics.

___________________ (1978) Corruption: A Study in Political Economy, New York: Academic Press.

Rosser, J. Barkley, Jr. & Marina Rosser (1999), "The New Traditional Economy: A New Perspective for Comparative Economics?" International Journal of Social Economics, 26(6) pp. 763-778.

Sachs, Jeffrey D. & Andrew Warner (1997) “Fundamental Sources of Long-run Growth,” American Economic Review, Papers and Proceedings, 87, No. 2: 184-88, May.

Schroth, Peter W. (2002a), "43 Years of Transnational Law Against Corruption (of Which 40 in the United States Alone)", in David Birch & Jonathan Batten, eds., Governance and Corporate Social Responsibility in the New Millennium: Proceedings of the 2001 Conference, Burwood, VIC, Australia: Deakin University.

_______________ (2002b), "The United States and the International Bribery Conventions," American Journal of Comparative Law, 50(supp.) pp.593-622.

_______________ (2002c), “Why Have the Corruption Treaties’ Provisions on Accounting and Control Failed to Overcome the Business Judgment Rule?,” Academy of International Business, Proceedings of the 2002 Northeast Conference, Salisbury, MD: Salisbury University.

_______________ (2003), "Fostering Informed and Responsible Management: The Failure of the Corruption Treaties: Provisions on Accounting and Control," Research in International Business and Finance,17, pages ___-___ (forthcoming).

Schroth, Peter W. & Preeti Sharma (2002) "Fighting Corruption in Africa: Transnational Law and New Communication Technologies," in 2002 Proceedings of the International Academy of African Business and Development: Expanding the Horizons of African Business and Development, Port Elizabeth, South Africa: University of Port Elizabeth pp. 549-556.

Stodder, James (1995) "The Evolution of Complexity in primitive Exchange," Journal of Comparative Economics, Vol. 20 (in two parts): "Theory," Issue 1, February, pp. 1-31 and "Empirical Tests," Issue 2, April, pp. 190-210.

____________ (1993) "Ex-Communist Economics: the Logic of Capital Reform," New York Economic Review, 24(4).

Wei, Shang-Jin (1997) "Why is Corruption So Much More Taxing Than Tax? Arbitrariness Kills," NBER Working Paper No. w6255.

____________ (2000) “How Taxing Is Corruption on International Investors?” Review of Economics & Statistics, 82(1).

Appendix (A): Country Data in the Easterly Data Set, Basic Indicators

The Data in this table are the basic ones used in this study; the instrumental variables used to derive the fitted estimates can be downloaded by going to www.worldbank.org/research/peg/, then clicking on the link for Working Paper Number 17. Below there are C = 106 Countries, P = 76 with T = 2 time period observations, and S = 30 with only one observation. O = 182 observations in all. The variable definitions are as follows:

Table A1: Data Sources

VARIABLE		Data Sources
NAME	Name of Country
YEAR	Year of Observation
TRANSP	Transparency	IRIS Int'l Country Risk data (Note: This variable is labeled 'Corruption' in the Easterly Data, even though it is increasing as corruption decreases. We have chosen 'Transparency' as less confusing.
RULELAW	Rule of law	IRIS Int'l Country Risk data
INCOME-PC	Income per capita	Summers, R. & A. Heston. 1991. "The Penn World Table (Mark 5): an expanded set of international comparisons, 1950-88." Quarterly Journal of Economics 106 (2): 327-68 (with 1993 on line update from NBER).
NSMORT	Mortality - Infant	WRI (World Resources, Inc.)
NSSCHL	Average schooling years in the total population over age 25	Barro-Lee
NSLIFEX	Life expectancy at age zero	Barro-Lee
GINIDC	Gini coefficient (decade avg)	Gini coefficient (decade avg)

In the table below, names are in bold if they occur without a second period observation:

Table B1: Basic Data

NAME	YEAR	TRANSP	RULELAW	INCOME-PC	NSMORT	NSSCHL	NSLIFEX	GINIDC
ALGERIA	1980	2	2	$2,758	112	1.868	58.4	38.73
ALGERIA	1990	4	2	2,777	74	2.394	61.5	#N/A
ARGENTINA	1980	3	2	6,506	41	6.63	69.3	#N/A
ARGENTINA	1990	4	3	4,706	32	6.676	70.2	#N/A
AUSTRALIA	1980	6	6	12,520	13	10.076	74.5	34.267
AUSTRALIA	1990	5	6	14,445	9	10.241	75.8	41.72
AUSTRIA	1980	6	6	10,509	17	6.224	72.7	28.669
AUSTRIA	1990	5	6	12,695	9	6.642	73.7	#N/A
BANGLADESH	1980	0	1	1,085	137	1.681	47.9	35.723
BANGLADESH	1990	0	1	1,390	119	1.974	49.9	28.27
BELGIUM	1980	6	6	11,109	14	8.794	73.2	24.461
BELGIUM	1990	5	6	13,232	9	9.152	74.3	26.968
BOLIVIA	1980	1	1	1,989	131	3.971	49.9	#N/A
BOLIVIA	1990	2	1	1,658	98	4.285	52.2	42.04
BRAZIL	1980	4	4	4,303	79	2.976	62.8	53.858
BRAZIL	1990	4	4	4,042	63	3.486	64.3	#N/A
BULGARIA	1990	3	5	6,203	14	#N/A	#N/A	27.797
BURKINA FASO	1990	4	3	511	126	#N/A	46.4	#N/A
CAMEROON	1980	3	3	1,194	102	1.758	53	49
CAMEROON	1990	3	3	1,226	74	2.23	55.1	#N/A
CANADA	1980	6	6	14,133	12	10.158	74.7	31.569
CANADA	1990	6	6	17,173	7	10.369	76.2	30.52
CHILE	1980	4	5	3,892	47	5.986	69.5	54.607
CHILE	1990	3	4	4,338	18	6.45	71.3	52.35
CHINA	1990	4	3	1,324	32	#N/A	68.8	36.033
COLOMBIA	1980	3	2	2,946	59	4.233	63.1	47.9
COLOMBIA	1990	3	1	3,300	40	4.533	65	50.023
CONGO	1990	3	2	2,211	84	3.143	57.7	#N/A
COSTA RICA	1980	5	4	3,717	30	4.814	71.7	44.42
COSTA RICA	1990	5	4	3,499	16	5.333	73.4	#N/A
COTE D'IVOIRE	1990	3	2.5	1,213	#N/A	#N/A	51.7	#N/A

NAME	YEAR	TRANSP	RULELAW	INCOME-PC	NSMORT	NSSCHL	NSLIFEX	GINIDC
CYPRUS	1990	3	4	8,368	11	7.128	75.4	#N/A
CZECHOSLOVAKIA, FORMER	1990	5	5	4,095	#N/A	#N/A	#N/A	24.557
DENMARK	1980	6	6	11,342	9	10.139	74.1	26.554
DENMARK	1990	6	6	13,909	8	10.331	74.9	29.063
DOMINICAN REPUBLIC	1980	3	3	2,343	#N/A	3.709	63.3	47.608
DOMINICAN REPUBLIC	1990	3	3	2,166	#N/A	4.177	65.2	49
ECUADOR	1980	4	4	3,238	82	5.401	63.1	44.53
ECUADOR	1990	3	4	2,755	63	5.584	65	43
EGYPT	1980	1	1	1,645	131	#N/A	57.1	#N/A
EGYPT	1990	2	2	1,912	65	#N/A	59.6	32
EL SALVADOR	1980	2	1	2,014	87	3.283	57.3	#N/A
EL SALVADOR	1990	2	1	1,824	59	3.574	60.2	#N/A
FINLAND	1980	6	6	10,851	9	9.609	73.2	25.545
FINLAND	1990	6	6	14,059	6	9.491	75	23.549
FRANCE	1980	6	6	11,756	11	5.968	74.9	32.05
FRANCE	1990	5	5	13,904	8	6.523	76.6	#N/A
GABON	1980	2	2	4,797	122	#N/A	48.4	#N/A
GABON	1990	2	2	3,958	103	#N/A	51.1	#N/A
GAMBIA, THE	1990	3	3	799	143	0.837	42.4	#N/A
GERMANY, FORMER FDR	1980	6	6	11,920	#N/A	8.457	73.2	#N/A
GERMANY, FORMER FDR	1990	5	6	14,341	#N/A	8.544	74.3	#N/A
GHANA	1980	1	1	976	103	2.382	51.6	36.32
GHANA	1990	4	3	902	90	3.216	53.2	33.94
GREECE	1980	3	3	5,901	25	6.556	74.3	35.738
GREECE	1990	5	3	6,768	11	6.727	75.9	#N/A
GUATEMALA	1980	2	1	2,574	82	2.342	58	58.66
GUATEMALA	1990	2	1	2,127	59	2.591	60.8	#N/A
GUINEA	1990	4	3	767	145	#N/A	#N/A	#N/A
GUINEA-BISSAU	1990	2	1	689	151	0.56	38.7	56.12
GUYANA	1980	1	0	1,927	67	4.603	65.3	#N/A
GUYANA	1990	1	1	1,094	56	5.109	65.9	40.22
HAITI	1980	0	0	1,033	121	1.455	51.9	#N/A
HONDURAS	1980	2	1	1,519	90	2.698	60.2	54.98
HONDURAS	1990	2	2	1,377	68	3.561	63.2	53.726
HONG KONG	1980	6	6	8,719	#N/A	6.734	74.1	41.493
HONG KONG	1990	5	4	14,849	#N/A	7.511	75.8	45
HUNGARY	1990	5	5	5,357	17	10.745	69.9	27.874
ICELAND	1980	6	6	11,566	9	7.398	#N/A	#N/A
ICELAND	1990	6	6	13,362	6	7.889	#N/A	#N/A
INDIA	1980	3	3	882	129	2.715	53.7	31.428
INDIA	1990	2	1	1,264	96	3.046	56.7	31.628
INDONESIA	1980	1	1	1,281	105	3.086	54.7	33.957
INDONESIA	1990	0	2	1,974	75	3.75	58.5	34.592
IRAN, ISLAMIC REPUBLIC OF	1980	1	1	3,434	#N/A	2.323	60.6	42.9
IRAN, ISLAMIC REPUBLIC OF	1990	3	1	3,392	#N/A	3.281	#N/A	#N/A
IRAQ	1980	2	1	7,242	84	2.456	62.4	#N/A
IRELAND	1980	6	5	6,823	15	7.605	72.6	35.804
IRELAND	1990	5	5	9,274	8	8.008	73.5	#N/A
ISRAEL	1980	5	1	7,895	18	9.135	72.8	#N/A
ISRAEL	1990	5	1	9,298	11	9.41	75	#N/A
ITALY	1980	3	5	10,323	18	5.83	74.7	31.853
ITALY	1990	4	5	12,488	10	6.276	76.3	28.302
JAMAICA	1980	2	1	2,362	26	3.602	71.3	50.73
JAMAICA	1990	2	2	2,545	17	4.162	73.5	39.825
JAPAN	1980	6	6	10,072	9	8.166	76	35.2
JAPAN	1990	5	5	14,331	5	8.458	77.3	35
JORDAN	1980	3	1	3,384	65	2.933	62.7	36.968
JORDAN	1990	3	2	2,919	44	4.308	65	40.66
KENYA	1980	3	2	911	88	2.444	54.9	57.3
KENYA	1990	3	3	911	72	3.093	57.1	54.39
KOREA, REPUBLIC OF	1980	3	3	3,093	#N/A	#N/A	#N/A	35.335
KOREA, REPUBLIC OF	1990	2	2	6,673	#N/A	#N/A	#N/A	#N/A
KUWAIT	1980	3	1	20,018	34	4.294	70.8	#N/A
LIBERIA	1980	1	1	927	167	1.345	51.5	#N/A
LUXEMBOURG	1990	6	6	16,280	9	#N/A	#N/A	#N/A
MADAGASCAR	1990	4	3.5	675	120	#N/A	52.7	43.44
MALAWI	1980	5	1	554	177	2.271	43.7	58.3
MALAWI	1990	4	2	519	149	2.577	45.4	#N/A
MALAYSIA	1980	6	4	3,799	34	4.489	66.9	48.168
NAME	YEAR	TRANSP	RULELAW	INCOME-PC	NSMORT	NSSCHL	NSLIFEX	GINIDC
MALAYSIA	1990	4	3	5,124	17	5.361	68.9	#N/A
MALI	1990	2	2	531	169	0.819	46.3	#N/A
MEXICO	1980	2	4	6,054	58	3.512	66.6	52.453
MEXICO	1990	3	3	5,827	41	4.418	68.1	50.31
MONGOLIA	1990	4	3	1,842	68	#N/A	#N/A	#N/A
MOROCCO	1980	1	1	1,941	110	#N/A	57.3	43.763
MOROCCO	1990	3	2	2,151	82	#N/A	59.8	39.2
MOZAMBIQUE	1990	4	2	760	155	1.077	47	#N/A
MYANMAR	1980	1	3	505	114	1.464	#N/A	#N/A
NAMIBIA	1990	3	2	2,854	80	#N/A	#N/A	#N/A
NETHERLANDS	1980	6	6	11,284	10	8.199	75.7	29.17
NETHERLANDS	1990	6	6	13,029	9	8.572	76.5	29.163
NEW ZEALAND	1980	6	6	10,362	14	12.141	73.2	35.314
NEW ZEALAND	1990	6	6	11,513	11	12.039	74.4	40.21
NICARAGUA	1980	3	1	1,853	97	2.735	58.4	#N/A
NICARAGUA	1990	5	2	1,294	71	3.776	61.9	50.32
NIGERIA	1980	1	1	1,438	124	#N/A	47.7	36.26
NIGERIA	1990	2	1	995	105	#N/A	50.1	39.31
NORWAY	1980	6	6	12,141	9	10.316	75.7	27.036
NORWAY	1990	6	6	14,902	8	10.382	76.5	28.906
PAKISTAN	1980	1	1	1,110	130	1.737	53	33.586
PAKISTAN	1990	2	1	1,394	109	1.92	55.5	31.15
PANAMA	1980	3	1	3,392	32	5.982	70.3	51.97
PANAMA	1990	2	2	2,888	23	6.3	71.6	#N/A
PAPUA NEW GUINEA	1990	3	3	1,425	59	1.654	53.2	#N/A
PARAGUAY	1980	3	1	2,534	53	4.627	66.3	#N/A
PARAGUAY	1990	0	2	2,128	49	4.704	66.7	#N/A
PERU	1980	2	2	2,875	105	5.442	58	45.023
PERU	1990	3	1	2,188	88	5.786	60.3	41.935
PHILIPPINES	1980	0	1	1,879	62	6	61	44.027
PHILIPPINES	1990	2	1	1,763	45	6.481	62.8	45
POLAND	1990	5	4	3,820	17	8.412	71.2	26.685
PORTUGAL	1980	3	6	4,982	30	3.23	71.4	35.353
PORTUGAL	1990	5	5	7,478	14	3.827	72.9	34.442
SAUDI ARABIA	1980	3	1	13,750	75	#N/A	59.7	#N/A
SENEGAL	1980	3	1	1,134	112	1.989	45.2	#N/A
SENEGAL	1990	3	2	1,145	87	2.393	46.7	54.12
SIERRA LEONE	1990	2	3	901	154	1.721	40.3	#N/A
SINGAPORE	1980	6	6	7,053	13	3.691	71.5	42.41
SINGAPORE	1990	4	5	11,710	8	4.553	72.5	#N/A
SOUTH AFRICA	1990	5	1	3,248	58	4.955	59.4	62.3
SPAIN	1980	6	5	7,390	16	5.152	74.9	26.378
SPAIN	1990	4	4	9,583	8	5.585	76.3	#N/A
SRI LANKA	1980	3	2	1,635	44	5.183	68	43.868
SRI LANKA	1990	3	0	2,096	28	5.371	69.8	30.1
SUDAN	1980	1	1	866	131	0.64	46.7	#N/A
SUDAN	1990	2	1	757	108	0.909	49	#N/A
SWEDEN	1980	6	6	12,456	8	9.469	75.9	30.287
SWEDEN	1990	6	6	14,762	6	9.447	76.4	28.877
SWITZERLAND	1980	6	6	14,301	10	9.666	75.8	34.313
SWITZERLAND	1990	6	6	16,505	7	9.088	76.7	#N/A
SYRIAN ARAB REPUBLIC	1980	1	1	4,467	#N/A	3.106	61.6	#N/A
SYRIAN ARAB REPUBLIC	1990	2	2	3,897	#N/A	3.987	64	#N/A
TAIWAN, CHINA	1980	5	6	4,459	#N/A	6.365	72.4	29.442
TAIWAN, CHINA	1990	4	5	8,063	#N/A	6.999	72.8	30.139
TANZANIA	1980	1	3	480	125	2.425	49.7	#N/A
THAILAND	1980	4	3	2,178	56	3.765	62.2	47.037
THAILAND	1990	3	4	3,580	32	5.081	63.8	50.264
TOGO	1980	2	2	731	117	1.623	49.5	#N/A
TOGO	1990	2	2	641	94	2.132	52	#N/A
TRINIDAD AND TOBAGO	1980	2	4	11,262	#N/A	6.599	68.2	41.72
TRINIDAD AND TOBAGO	1990	3	4	7,764	#N/A	6.503	69.4	#N/A
TUNISIA	1980	3	3	2,527	88	1.918	61.9	43.143
TUNISIA	1990	3	2	2,910	49	2.478	64.4	40.493
TURKEY	1980	3	3	2,874	120	2.616	61	43.945
TURKEY	1990	2	2	3,741	68	3.294	63.1	#N/A
UGANDA	1980	1	1	534	114	1.641	46	33
UGANDA	1990	3	1	554	108	1.917	47.3	40.78
UNITED ARAB EMIRATES	1980	3	1	31,969	38	#N/A	68.2	#N/A
NAME	YEAR	TRANSP	RULELAW	INCOME-PC	NSMORT	NSSCHL	NSLIFEX	GINIDC
UNITED KINGDOM	1980	6	6	10,167	14	8.345	73.8	29.621
UNITED KINGDOM	1990	5	5	13,217	9	8.648	74.7	32.35
UNITED STATES	1980	6	6	15,295	14	11.888	73.7	36.878
UNITED STATES	1990	5	6	18,054	10	11.787	74.9	37.158
URUGUAY	1980	3	3	5,091	42	5.8	70.1	#N/A
URUGUAY	1990	3	3	4,602	24	6.454	70.8	#N/A
VENEZUELA	1980	2	3	7,401	43	4.93	68.5	42.965
VENEZUELA	1990	3	4	6,055	36	5.368	69.6	49.12
ZAIRE	1980	1	1	476	115	1.738	49.2	#N/A
ZAMBIA	1980	1	3	971	94	3.459	50.1	#N/A
ZAMBIA	1990	2	1.5	689	86	4.349	52.1	43.51
ZIMBABWE	1980	3	1	1,206	86	2.395	55	#N/A

It is noteworthy that all examples of countries with scores of 5 or 6 for both Transparency and Rule of Law were from the countries of North-West or Central Europe or peopled largely by their immigrants (e.g., Australia), with the following exceptions: Taiwan, Hong Kong, Singapore, and Japan.

Appendix (B): Examining the Regression Outliers

It is often instructive to check which observations the regressions do not able handle particularly well. This was done here, and the inconsistencies exhibit a rather striking consistency of their own. All of the countries with absolute error terms greater than 0.5 (on a 7 point scale) are either "Latin" in their culture or colonial legacy, or they are African. We have no theories on the reasons behind this pattern, but it is certainly interesting.

Table B1: Transparency Regressed on Life Expectancy

	Estimated	Standard
Variable	Coefficient	Error	t-statistic	P-value
Constant i	-	-	-
Income-PC	-3.63E-04	1.37E-04	-2.65416	[.012]
LifeEx (fitted)	0.109551	0.047682	2.29755	[.028]

The following table shows those regressions where the estimated Transparency differed from the Observed by more than 0.5 points.

Table B2: Outliers for Regression of Transparency on Life Expectancy

		Coefficient	Variable	Variable	Estim.	Observ.	Estim. minus	Bias
Year	Country	Constant	Income-PC	LifeEx (fitted)	Transp.	Transp.	Observ. Transp.	Estim.
1980	ECUADOR	-1.62245	3238	55.3	3.26	4.0	-0.7	under
1990	ECUADOR	-1.62245	2755	58.0	3.74	3.0	0.7	over
1980	GHANA	-3.23589	976	52.7	2.19	1.0	1.2	over
1990	GHANA	-3.23589	902	58.2	2.81	4.0	-1.2	under
1980	ITALY	-0.63614	10323	75.0	3.84	3.0	0.8	over
1990	ITALY	-0.63614	12488	76.0	3.16	4.0	-0.8	under
1980	PANAMA	-3.09236	3392	58.5	2.09	3.0	-0.9	under
1990	PANAMA	-3.09236	2888	64.3	2.91	2.0	0.9	over
1980	PARAGUAY	-4.37733	2534	65.5	1.88	3.0	-1.1	under
1990	PARAGUAY	-4.37733	2128	57.2	1.12	0.0	1.1	over
1980	PHILIPPINES	-5.01573	1879	61.1	1.00	0.0	1.0	over
1990	PHILIPPINES	-5.01573	1763	60.7	1.00	2.0	-1.0	under
1980	SPAIN	0.16239	7390	70.7	5.23	6.0	-0.8	under
1990	SPAIN	0.16239	9583	73.8	4.77	4.0	0.8	over
1980	SUDAN	-3.73478	866	51.1	1.55	1.0	0.6	over
1990	SUDAN	-3.73478	757	49.8	1.45	2.0	-0.6	under
1980	UGANDA	-3.68546	534	50.0	1.59	1.0	0.6	over
1990	UGANDA	-3.68546	554	57.4	2.41	3.0	-0.6	under
1980	ZAMBIA	-4.57559	971	59.6	1.60	1.0	0.6	over
1990	ZAMBIA	-4.57559	689	56.8	1.40	2.0	-0.6	under

We can also consider the direction of change in Transparency, according to our regressions. For each pair of observations on a country here, we have bolded the larger of the two estimates and observations on the dependent variable. When the bolded values are in the same year, the regression correctly predicts the direction of change.

We will see very much the same pattern of outliers when we use Child Mortality variable, rather than Life Expectancy, as the dependent variable.

Table B3: Transparency Regressed on Child Mortality

	Estimated	Standard
Variable	Coefficient	Error	t-statistic	P-value
Constant i	-		-	-
Income-PC	-3.10E-04	1.30E-04	-2.38085	[.023]
Child.Mort. (fitted)	-0.020462	9.34E-03	-2.18983	[.035]

Table B4: Outliers for Regression of Transparency on Child Mortality

		Coefficient	Variable	Variable	Estimate	Observ	Estim. minus	Bias of
Year	Country	Constant	Income-PC	Mortality (fitted)	Transp.	Transp.	Observ. Transp.	Estimate
1980	ECUADOR	6.29551	3238	99.4	3.26	4.0	-0.7	under
1990	ECUADOR	6.29551	2755	83.2	3.74	3.0	0.7	over
1980	GHANA	4.68724	976	107.0	2.20	1.0	1.2	over
1990	GHANA	4.68724	902	78.4	2.80	4.0	-1.2	under
1980	ITALY	7.19094	10323	10.5	3.78	3.0	0.8	over
1990	ITALY	7.19094	12488	5.1	3.22	4.0	-0.8	under
1980	PANAMA	4.83749	3392	82.4	2.10	3.0	-0.9	under
1990	PANAMA	4.83749	2888	51.1	2.90	2.0	0.9	over
1980	PARAGUAY	3.58555	2534	46.7	1.84	3.0	-1.2	under
1990	PARAGUAY	3.58555	2128	86.6	1.16	0.0	1.2	over
1980	PHILIPPINES	2.93285	1879	66.0	1.00	0.0	1.0	over
1990	PHILIPPINES	2.93285	1763	67.8	1.00	2.0	-1.0	under
1980	SPAIN	8.04508	7390	27.5	5.19	6.0	-0.8	under
1990	SPAIN	8.04508	9583	13.3	4.81	4.0	0.8	over
1980	SUDAN	4.17514	866	116.4	1.53	1.0	0.5	over
1990	SUDAN	4.17514	757	120.5	1.47	2.0	-0.5	under
1980	UGANDA	4.22259	534	119.6	1.61	1.0	0.6	over
1990	UGANDA	4.22259	554	81.1	2.39	3.0	-0.6	under
1980	ZAMBIA	3.34266	971	70.7	1.60	1.0	0.6	over
1990	ZAMBIA	3.34266	689	84.3	1.40	2.0	-0.6	under

Note that these results for Rule of Law are almost identical to those for Transparency. The Error terms (in the second to last column) are identical to the first decimal place in every case. There-fore the same pattern holds, of over-estimating the Transparency of the Latin countries, while under-estimating that of the African countries shown.

Next we turn to the regressions for Rule of Law.

Table B5: Rule Law Regressed on Life Expectancy

	Estimated	Standard
Variable	Coefficient	Error	t-statistic	P-value
Constant i	-		-	-
Income-PC	7.40E-02	3.22E-02	2.30E+00	[.027]
LifeEx (fitted)	-2.59E-04	9.21E-05	-2.811	[.008]

Table B6: Outliers for Regression of Rule of Law on Life Expectancy

		Coefficient	Variable	Variable	Estimated	Observed	Estim. minus	Bias of
Year	Country	Constant	Income-PC	LifeEx (fitted)	Rule Law	Rule Law	Obs. Rule Law	Estimate
1980	GHANA	-1.86147	976	52.7	1.79	1.0	0.8	over
1990	GHANA	-1.86147	902	58.2	2.21	3.0	-0.8	under
1980	PARAGUAY	-2.43742	2534	65.5	1.75	1.0	0.8	over
1990	PARAGUAY	-2.43742	2128	57.2	1.25	2.0	-0.8	under
1990	PERU	-2.11651	2875	56.5	1.32	2.0	-0.7	under
1980	PERU	-2.11651	2188	59.0	1.68	1.0	0.7	over
1990	ZAMBIA	-1.84248	971	59.6	2.32	3.0	-0.7	under
1990	ZAMBIA	-1.84248	689	56.8	2.18	1.5	0.7	over

Once again, all the outliers are Latin or African. A similar pattern will be seen in the regressions using Child Mortality.

Table B7: Transparency Regressed on Child Mortality

	Estimated	Standard
Variable	Coefficient	Error	t-statistic	P-value
Constant i	-		-	-
Income-PC	-3.10E-04	1.30E-04	-2.38085	[.023]
Child.Mort. (fitted)	-0.020462	9.34E-03	-2.18983	[.035]

Table B8: Outliers for Regression of Rule of Law on Child Mortality

		Coefficient	Variable	Variable	Estimated	Observed	Estim. minus	Bias of
Year	Country	Constant	Income-PC	Child Mortality (fitted)	Rule Law	Rule Law	Observ. Rule Law	Estimate
1980	GHANA	3.49653	976	107.0	1.79	1.0	0.8	over
1990	GHANA	3.49653	902	78.4	2.21	3.0	-0.8	under
1980	PARAGUAY	2.94254	2534	46.7	1.73	1.0	0.7	over
1990	PARAGUAY	2.94254	2128	86.6	1.27	2.0	-0.7	under
1990	PERU	3.23503	2875	92.0	1.32	2.0	-0.7	under
1980	PERU	3.23503	2188	77.0	1.68	1.0	0.7	over
1990	ZAMBIA	3.51119	971	70.7	2.31	3.0	-0.7	under
1990	ZAMBIA	3.51119	689	84.3	2.19	1.5	0.7	over

Once again, the pattern of error terms is almost identical in Tables B6 and B8, with nearly identical terms in the second to last column -- except in the case of Paraguay.