The enormous growth in government spending over the past century was by and large promoted and defended in the name of economic and social progress. While economic progress can be defined as a rise in per capita income or gross domestic product (GDP), social progress, or improvement in the quality of life, usually refers to improvements in physical conditions of life such as life expectancy and infant mortality. A broader notion of social progress could encompass the development of political institutions such as democracy and the rule of law.

Government Spending and Economic Progress. In a previous NCPA study I showed that government spending can promote economic progress - up to a point.² Government spending on such things as roads, education and criminal justice positively affects per capita GDP. But beyond some level the tax burden necessary to finance government becomes a drag on economic growth and reduces per capita GDP below what it otherwise would be. By this standard, I found the optimal level of U.S. taxation to be 21 percent of GDP. Yet taxes as a share of GDP have not been in this range since 1949. Had the United States stuck with the 1949 level of taxation, our compound rate of economic growth over the period 1950-95 would have been 4.8 percent, compared to the actual growth rate of 3.4 percent.

Today average government spending in developed countries is 45 percent of GDP. In fact, for decades government spending in almost all developed countries has far exceeded the level at which it increases the rate of economic progress. It continues at levels that reduce growth rates below what they otherwise would be.

Government Spending and Social Progress. This paper examines the effects of government spending on social progress. The spending we are concerned with is government consumption expenditures: the portion of government spending that actually buys goods or services such as school buildings and teachers' salaries. In most countries, this is 40 percent to 60 percent of total government spending. The rest of government expenditures go for transfer payments or payments on government debt. Transfer payments simply redistribute income after it is earned and taxed away by government. Much of the increase in total government spending in developed countries in recent years has been for transfer payments and debt service. But government consumption expenditures have also increased, and even if they remained constant as a portion of GDP the absolute level of government spending has increased as GDP has grown.

It has been assumed that government consumption spending buys goods and services that improve the quality of life. But remarkably, very little is known about what has been achieved for the tax dollars spent. The only paper I have found that addresses quality of life gains from the growth of public spending is a 1995 study by Vito Tanzi and Ludger Schuknecht.³

Tanzi and Schuknecht examined the progress of certain social indicators over more than a century and compared this progress with the growth in public spending. They used noneconomic indicators - such as infant mortality, the death rate and years of primary schooling - as well as economic indicators such as the share of national income going to the lowest 40 percent of all households. They found that as total government spending as a portion of GDP grew from 1870 to 1960, it had a positive effect on economic growth and social progress but "that the growth in spending after 1960 may not have brought about significantly improved economic performance or greater social progress."⁴

Moreover, they found that in recent years social progress in some newly industrialized economies (Chile, Hong Kong, South Korea and Singapore) and in industrialized countries with governments that spent less than 40 percent of GDP ("small" government countries such as Japan, Switzerland and the United States) is no worse than in countries that spend more than 50 percent of GDP ("big" government countries such as Sweden and the Netherlands) or those that spend 40 to 50 percent of GDP ("medium-sized" government countries such as Canada, Germany and Ireland).

Clear inferences from their paper are that public spending for social progress in industrialized countries has reached a level of diminished returns and that a prudent reduction in the size of government can occur without social harm.⁵

Evaluating the Results of Government Spending. This study examines the effects of government expenditure on certain noneconomic measures of social progress, using 1995 data on 112 countries. It addresses these issues more precisely than Tanzi and Schuknecht by asking and answering two basic questions: (1) at what level of government expenditures per capita does spending cease to have any positive impact (the marginal benefit is zero) and (2) beyond what level of expenditures does spending cease to be worth its cost (the point at which the marginal benefit equals the marginal cost)? The answer to the first question tells us at what level of public spending an increase in expenditures results in no additional improvement in social progress. Hence, any government expenditure beyond this level is unnecessary for advancing the quality of citizens' lives. The answer to the second question tells us at what point a dollar increase in public spending buys an additional dollar's worth of social progress.

Spending above the level at which the marginal benefit is zero is a clear waste of resources. Spending at the level for which the marginal benefit is zero represents the maximum possible social progress. However, that maximum level is not the optimal level because as spending approaches the level at which the marginal benefit to social progress equals zero, we must spend much more than a dollar to achieve a dollar's worth of improvement. The optimal level of government spending is the point at which marginal benefit equals marginal expenditure, where an additional dollar of spending will yield a dollar of social progress.

Three empirical issues need careful handling. First, what is social progress? Second, how much weight should be given to each of the attributes of social progress? And third, how can we avoid bias in measuring social progress?

Issue: Defining Social Progress. Social progress or improvement in the quality of life may be a subjective concept that arises from one person's life experience relative to others. If so, it would be hard to measure. If social progress is treated as an objective concept, as is done here, it involves a finite set of attributes that can be measured and compared across environments and through time.

While per capita gross national product is an obvious choice for a single measure of well-being, it has been criticized as too narrow to capture "basic needs."⁶ The most widely employed measure of basic needs is the physical quality of life index (PQL), an average of literacy, infant mortality rates and life expectancy. The PQL is one of the aggregate indices of social progress used here. But the PQL index also has been criticized as too narrow in scope to fully capture differences in quality of life. Hence, I have constructed a multidimensional index that contains 16 social progress indicators. Using such a large number of indicators, many of which are intercorrelated, precluded the omission of relevant variables that would significantly affect the results. Income or GDP per person is an important measure of the ability of a society to make expenditures for social progress, but it does not measure social progress per se. Thus it is an input to, rather than an outcome of, social progress. By contrast, the social progress indicators used here measure outcomes.

Issue: Assigning Weights to the Attributes of Social Progress. How much weight should be given to each of the attributes of social progress in constructing an aggregate measure? The conventional practice is to weight them equally. However, while a representative individual might have an equal preference for any or all of the attributes of social progress, it is not likely. And it is not possible to say in advance exactly what the preferences of society are and how much society is willing to pay for them. Accordingly, two other ways to objectively weight the attributes also are employed. Using all three weighting schemes gives some assurance that the results are not distorted by the peculiarities of the weighting method.

Issue: Avoiding Measurement Bias. Any functional form imposed on the data is arbitrary. To ensure that the empirical results are not largely due to the particular mathematical function used, several nonlinear models were estimated.

In what follows, this paper discusses the attributes of social progress, the optimal levels of government expenditures that were found by performing some statistical tests and the conclusions that can be drawn from the analysis.

The first attributes of social progress I use are widely recognized measures of human welfare: literacy, infant mortality and life expectancy. These three attributes make up the Physical Quality of Life index (PQL), a widely accepted index of social progress. Testing the PQL index separately, I am able to compare it to the broader index and to see whether the additional attributes that are included in the multidimensional index add more information. Furthermore, using 1995 data for the PQL index from 112 nations, and data for the other attributes from 85 countries permits confirmation that the results are not skewed by the omission of some countries.⁷ Including the three PQL attributes in the multidimensional index ensures that the broader index is inclusive. In addition to the three PQL attributes, the multidimensional index includes 11 other attributes chosen by the author and two composite indices - the Index of Economic Freedom and the Freedom House index of political and civil rights.⁸

The Physical Quality of Life Index. The variables in the PQL index are widely accepted as measures of human welfare. Life expectancy among the 112 countries in the PQL index ranges from 37.4 to 79.25 years, infant mortality from 4.3 per 1,000 live births to 148.6, and the literacy rate from 11.05 to 100 percent.⁹ [See Table I for the range of values of all 16 attributes.]

The Multidimensional Index of Social Progress. Economist Amartya Sen, among others, has criticized the PQL index as too narrow.¹⁰ To more fully capture differences in quality of life across countries, a more multidimensional approach is warranted. To accomplish this, I combined the three attributes from the PQL index with 13 other measures to construct an index with 16 attributes. (For a detailed discussion of the meaningfulness of the attributes, see the author's previous joint research.)¹¹

For example, a fourth attribute is average number of persons per household. Economist Gary Becker has argued that in modern societies children are a consumption good and that as family size shrinks "child quality" is substituted for quantity of children.12 Thus a family with fewer children can make a greater investment of household resources per child and declining household size is an advance in social progress.

A fifth attribute is military manpower per 1,000 inhabitants. While ambiguous as a measure of social progress, since troops can be used to oppress the domestic population, on the whole more troops represent greater protection from external aggression. A sixth attribute is energy use per capita in kilograms. A seventh is miles of road per square mile of land area, an indicator of infrastructure development and facility of physical movement within a country. An eighth is persons per telephone, another indicator of economic development and ease of communication.

A ninth attribute is hospital beds per 10,000 population and the tenth is population per physician, both indicators of the quality of health care. The eleventh is daily caloric consumption per person, a measure of the food supply. The twelfth is persons per radio receiver, a measure of the extent to which the population can be informed as well as entertained.

Attribute thirteen is the crime rate per 100,000 persons. As long as the ratio of reported to unreported crimes is stable across countries, it is a fairly reliable measure of security of person and property.

Other Indices Used as Attributes. Personal freedom and choice (political, civil and economic) are important attributes of social progress. The Freedom House index of political and civil rights¹³ measures the extent to which a population is free to choose its leaders and is governed by the rule of law.¹⁴

The Index of Economic Freedom attribute is a summary measure from James Gwartney and Robert Lawson.¹⁵ This measure is based on an average of up to 17 attributes (including a number of the economic indicators used by Tanzi and Schuknecht): (i) money and inflation (money supply growth, inflation variability, foreign exchange, bank account ownership), (ii) government operations (public consumption share, state enterprise, price controls, business environment, legal system, lack of negative interest rates), (iii) takings (transfers and subsidies, top marginal tax rate, conscription), and (iv) international sector (trade taxes, black market exchange premium, size of trade sector, foreign capital transactions).

Other Possible Indicators. Perhaps some other attributes such as the female labor force participation rate, unemployment rate, poverty rate or fraction of low-income population might be added.¹⁶ But there are real problems with the meaning of some of these attributes and the extent of their coverage. Unemployment and underemployment mean different things in advanced and backward countries; female labor force participation may mean liberation in advanced countries but forced family work in rural, agricultural, less-developed countries. Poverty is difficult to measure since reported income may not measure economic well-being,¹⁷ particularly in less-developed countries with large informal sectors and extensive barter systems.

Weighting the Attributes.gh it is standard practice to equally weight the indicators in constructing an aggregate measure of social progress, two objective methods of obtaining different weights are available: weighting by the variances through principal components analysis and weighting by regression coefficients through an instrumental variable or hedonic approach. The latter technique has the virtue of measuring the implicit value assigned to the indicator.¹⁸

Both the PQL index and the multidimensional indices used here were weighted with equal, principal component and hedonic weights. Thus a total of six indices of social progress were constructed. These indices turn out to be highly but not perfectly intercorrelated. The range in the simple correlations is from .8610 to .9725. [See Appendix Table I.]

General government consumption is government spending on wages, salaries, materials and supplies; it is the public spending that directly absorbs resources. Such expenditures are usually 40 percent to 60 percent of total government spending; the remainder consists mainly of transfer payments and interest payments on accumulated public debt. (Among the 112 countries in the sample, per capita government consumption spending ranges from $44.80 to $5,307.50 annually.)

The Relationship between Government Spending and the Quality of Life. As expected, we find increases in social progress as government expenditures increase. But social progress increases at a diminishing rate. [See Appendix for discussion.] There is strong evidence of diminishing returns to government expenditure for both the PQL index and the multidimensional index, no matter which weighting scheme is used. For example, the plot in Figure I of the Physical Quality of Life (PQL) index (equal weights) against per capita expenditures for the 112 nations in the sample clearly shows that the relationship between the PQL index and per capita government expenditure is nonlinear - and thus that the index will increase at a slowing rate as expenditures rise. The estimated relationship is drawn in Figure II. It appears that at a per capita spending level of around $3,000 to $4,400, no further gains in social progress occur. The profile of the data for the PQL index and the other indices seems to indicate a "saturation level" beyond which per capita government consumption expenditures have no effect on social progress. [See Appendix for discussion.]

Level of Expenditures at Which There Is No Marginal Benefit. At what level of per capita government expenditures is the marginal benefit of social progress, as measured by these indices, exhausted? In general, the results are not very sensitive to the weighting scheme used in constructing the indices. We find, using a polynomial model, that the level of per capita government expenditures at which the marginal benefit is equal to zero averages about $4,130.¹⁹ Based on this estimate:

• At most, a country can spend $4,130 per person on government consumption before it derives no gain in social progress from an additional dollar of spending.

• Countries that spend between $4,130 and the observed maximum of $5,307.50 on government consumption - which includes some of the world's most-developed economies - appear to be wasting up to $1,177.50 per person per year in spending that yields no improvement in social progress.

Of course, many of the countries in the sample are so poor they would have to spend many times their per capita GNP to reach the point at which the marginal benefit is exhausted. For instance, Zambia, with a per capita GDP of about $200, would have to spend 20 times its GDP to achieve parity with the developed nations. (Of course, long before Zambia reached the point of devoting 100 percent of GDP to social progress, it would have raised the tax burden to the point at which economic growth would have ceased!)²⁰ Thus it is more relevant to examine developed countries that are able to spend enough to exhaust the marginal benefit of social progress.

Results for Some High-Income Countries. For each of 22 high-income countries, Table II shows the percentage of GNP per capita that these governments would consume if they were at the average values of MB = 0 and MB = ME obtained above for the 112-country sample.²¹ The countries are mainly from the Organization for Economic Cooperation and Development and are ordered from high to low 1995 per capita GNP. The table also shows the average for each country of all six indices of social progress (rescaled to be in the interval 0 to 100), government consumption spending and the actual 1995 value of total government outlays as a share of GDP.

The results are consistent with the findings of Tanzi and Schuknecht, and all of these countries have high values f

Advanced countries can achieve maximum social progress, in the sense of marginal benefit equal to zero, with "maximum" per capita government consumption spending in the range of $3,970 to $4,380 - 20.2 percent to 22.3 percent of GNP, or "minimum" expenditures of $2,300 to $3,980 - 11.7 percent to 20.3 percent of GNP.

Optimal government consumption spending per capita, where the marginal benefit of social progress is equal to marginal government expenditure, is much lower, in the range of $605 to $1,830 or 3.1 percent to 9.3 percent of GNP. Thus:

• Advanced countries do not benefit, in terms of social progress, from government consumption spending beyond $3,650 per person or 18.6 percent of GNP.

• The optimal level of per capita government spending is $1,105 or 5.6 percent of GNP.

• With the exception of Singapore (and Hong Kong, if the data were available), the world's most developed countries receive no gains in social progress for much of the GNP consumed by government.

Therefore, there is considerable scope for shrinking the size of the fiscal state without doing harm to social progress.

Of course, government is much larger than is necessary to maximize social progress or to spend resources to the point at which the marginal social progress equals the marginal government expenditure necessary to achieve it.²⁵ On average, total public spending in the high-income countries is 45 percent of GDP. Almost all categories of real government budgets have grown, including government consumption, over the last four or five decades. But the high-growth sectors have been transfers and subsidies, social expenditures and interest payments on accumulated public debt.

Many explanations for the growth in the size of government have been offered: Wagner's law or Baumol's disease,²⁶ interest group lobbying, insights from public choice theory and bureaucratic expansionism.²⁷ Whatever the explanation, twenty-first century demographics make its current size unsustainable. Domestic population growth is below (often well below) replacement levels, resulting in aging populations that collect state-sponsored pensions for a long time and use a high fraction of national health resources. As fewer and fewer workers support more of society's members on at least one universal benefit, pressure builds to import more young workers from abroad (immigration on the required scale has implications for national identity), cut universal benefits or make work and investment more attractive by cutting taxes.

Some shrinkage in the size and scope of the fiscal and regulatory state seems inevitable. Appropriately fashioned, such shrinkage will not diminish social progress as measured by the indices constructed here, but redistributing the benefits and costs associated with shrinking the size of government implies considerable political difficulty. Only two governments to date have tackled the problem with some success:

• Ireland reduced public spending as a share of GDP from 47.2 percent in 1988 to 40.6 percent in 1989 and has more or less held the line since then.

• New Zealand shrank central government public spending from 40.4 percent of GDP in 1992 to 34.9 percent in 1995.

• However, the United Kingdom, where public spending was reduced to 40.7 percent of GDP in 1989, has since increased spending to about 45 percent.

Several European nations recently trimmed government expenditures to meet the criteria set by the Maastricht agreement for participation in the European single currency market, but the policy changes are marginal and are not part of long-term strategies to shrink the fiscal size of the states.

And what are the prospects of shrinking the size of the U.S. fiscal state? Citizens hate taxes but appear to love government expenditures more. Both political parties have grasped the reality that to get elected and to stay in office they must cater to the middle class, the largest recipients of public spending. While ideas do matter, rebellion and revolution or crisis are the only sources of dramatic change. The prospects for cutting government expenditure through full or partial privatization of social security, medical savings accounts, school vouchers, curtailment of farm subsidies and other corporate welfare and so on have never been better. Yet politicians and citizens alike prefer tinkering at the margin rather than fundamental reform.