Home

Dissertation (Précis)

September 13, 2003
The fiscal contract: The state, taxes and social welfare

Few issues in the social sciences have received more attention than the question of why some governments perform “better” than other governments, particularly with respect to the level of basic public goods and the protection of property rights, two defining features of the modern state. Basic public goods and secure property rights are widely considered to be building blocks for social and material advancement, whether measured in life expectancy, infant mortality, or purchasing power. Despite decades of research and billions of dollars in foreign aid, however, responsible government remains elusive in many parts of the world. Not only do many governments fail to provide even a minimal level of public services, leading to countless premature deaths that could easily have been prevented, but many more governments fail to provide secure property rights, stymieing growth and the establishment of material prosperity. The World Health Organization (WHO), for example, estimates that US$66 billion per year in well targeted spending in developing countries could save some 8 million lives per year and generate six-fold economic benefits by 2015.

My dissertation seeks to explain the wide variation in government performance by returning to the study of fiscal systems pioneered by Weber. Building on the tax morale literature and recent developments in game theory, I argue that the foundations of responsible government reside in the tax system, which Cicero called the sinews of the state. My argument is that if citizens respond to government demands for taxes based on their evaluation of government performance; and assuring tax compliance purely through force is costly; states and citizens can strike a series of efficiency enhancing bargains, whereby states exchange services in return for revenue. Reaching the efficient equilibrium hinges on the existence of non-state mechanisms for overcoming the free-rider problem, including tax morale and social sanctions. The critical assumption I make is that different groups of citizens want government to do different things. Whereas lower income groups have an especially strong desire for the provision of basic public goods, upper income groups have an especially strong desire for the provision of property rights. The theoretical rationale is simple: the level of basic public goods has a tremendous impact on the well-being of average citizens, especially lower income groups, while property rights have a tremendous impact on the returns earned by asset holders, especially the wealthy.

If different citizens want different things, taxpayers resist taxation without compensation, and free-riding can be overcome, states have incentives to match the distribution of the tax burden with the distribution of public benefits, limiting the amount they redistributes to themselves and between taxpayers. This logic leads to the prediction that “regressive” taxation will coincide with “progressive” spending, but not property rights, while “progressive” taxation will coincide property rights, but not basic public goods.

I test these hypotheses using a variety of statistical models and data from approximately 90 countries from 1975 to 1999. The selected sample should be sufficient to make valid generalizations since it includes countries at different levels of economic development, from different geographical regions and with different political characteristics. My results demonstrate that there is a strong positive relationship between basic public goods and the amount raised from regressive taxes, but not total government revenue or the total amount from progressive taxes. Furthermore, there is a strong positive relationship between property rights protection and the amount of revenue raised from progressive taxes, but not total government revenue or the amount raised from regressive taxes. These results suggest that the extent of redistribution is over-stated; in most countries, the people who pay for the state obtain the bulk of its benefits. States that incorporate lower income groups into the tax system provide more social benefits, but not property rights; states that incorporate the wealthy into the tax system provide property rights, but not extensive social benefits; state that incorporate both groups into the tax system provide both property rights and social benefits. These results hold within democracies and non-democracies, vis-à-vis alternative arguments (such as ethnolinguistic fractionalization, income inequality and globalization), and are not particularly sensitive to changes in the sample.

These quantitative findings are buttressed by an extensive case study of Brazil, where I conducted fieldwork for 9 months in 2002 on a Fulbright scholarship. In many ways, Brazil is an important outlier in the cross-national statistics. Not only has the Brazilian state been able to raise more money than one might expect given its performance, but it did so by taxing lower income groups quite intensely. Nevertheless, Brazil fits quite well with the model I develop. Repeated defections by the state have generated enormous tax evasion and forced it to earmark almost all new tax revenue for specific purposes, notably health care, education and pensions. From the 1960s onward, the Brazilian central government acquired income by relinquishing greater control over spending, showing that even the most recalcitrant Leviathan can be bound by the fiscal contract.

My findings—both quantitative and with the case study—offer insight into a number of important policy debates. Consider, first, the role of the state. To a large extent, questions of what “the state” should do hinge on the efficiency of government spending. My results show that the quality of public expenditure hinges the existence of a fiscal contract: when states have to bargain with taxpayers for revenue, they have to provide something in return; hence, the quality of public expenditure will depend on the source of that revenue. More specifically, my theory predicts (and I show empirically in Chapters 3-6) that states that raise revenue from lower income groups are more likely to have high quality social expenditures, while states that raise revenue from the wealthy are more likely to have high quality spending on the judicial system. States that receive exogenous revenue, in contrast, are likely to spend wantonly, a point best illustrated by the plethora of resource rich countries with abysmal human development indicators and poor property rights protection.

The second important policy issue is whether regressive but efficient taxes hurt or help lower income groups. To increase government revenues, the IBD, IMF and World Bank have encouraged the increased use of value added taxes with very limited, or even without, exemptions for basic foodstuffs. Such taxes would clearly increase the tax burden on the poor, but it would also increase government revenue. If governments were to spend that money on basic health care and education, then the poor would almost certainly be better off than they would be in the absence of both the tax and the benefit. My results suggest that proponents of the tax are right in probabilistic terms. States that are able to collect more revenue from lower income groups are far more likely to cater to their wishes.

The third policy debate concerns the establishment of property rights. A number of property rights scholars have advocated the universal adoption the English legal system, which is perceived as providing better property rights. My results show that there is no such thing as free lunch. Although countries with the English legal system provide more secure property rights, they systematically raise more money from capital than countries with other legal systems. In other words, the English legal system is not an unrequited panacea; if the wealthy want to secure their assets, they still need to pay for it.

The fourth important policy issue is globalization. Some critics of globalization claim that economic integration forces governments to cater to mobile factors of production, increasing the tax burden of lower income groups and undermining the state’s ability to pay for social services. My results suggest that the latter claim is patently untrue. Although globalization appears to have increased the tax burden on lower income groups, I can find no evidence that increased integration has decreased the services received by the poor. In point of fact, my results indicate that the opposite is true. Countries that have increased their intake from regressive taxes over the past quarter century—due to globalization or other factors—have increased their output of basic public goods far more than countries that obtain revenue from other sources. More importantly, the relationship between regressive taxes and public goods holds even before globalization. In other words, globalization may facilitate the contract between lower income groups and the state, but it is not the underlying cause of the contract.

Besides these important policy debates, my dissertation is also important for abstract theoretical reasons. A unified theory of the state is perhaps the Holy Grail in the social sciences; for at least two hundred years, however, the debate has revolved around a number of contending perspectives, which I label “capture,” “autonomous” and “contractual” based on their underlying logic. These perspectives have served as paradigms for generations of scholars and continue to be sources of great intellectual debate, but they have not been tested systematically or compared head to head in any rigorous fashion. By putting together cross-country evidence about tax revenue, tax compliance rates and state performance, I help untangle these stories. Using both cross-sectional and time-series cross-sectional regressions, my results show that state capture and autonomous theories have virtually no empirical support—though they may be short-run equilibriums—while contractual ones are strikingly robust. Not only is there ample support for the fiscal contract, but there is also considerable support for Mancur Olson’s and Martin McGuire’s “encompassing” interest model of democracy, which predicts that democracies should be more efficient with taxpayer money: for any given level of revenue, I show that democracies produce more basic public goods.

The dissertation is organized into seven chapters. Chapter 1 provides an overview of the dissertation. It outlines the argument, details the academic contribution, and summarizes of the main points of each chapter.

Chapter 2 sets forth the important theoretical issues and specifies the research design. It presents the theory of the fiscal contract in more detail and compares it to rival explanations for the provision of public goods. Using the iterated Prisoner’s Dilemma as an analytic device and recent developments in game theory, it shows that different theories of the state operate with different causal logics, generating different hypothesis about the relationship between the distribution of the tax burden and the distribution of public benefits. By comparing tax receipts with government performance in a variety of dimensions, it establishes a framework for testing these competing perspectives. Chapter 2 also presents empirical evidence about tax compliance and tax enforcement strategies. This main point of this evidence is that coercion cannot account for tax compliance in many cases, even with high levels of risk aversion. Given the existing penalty parameters in the United States, for example, optimizing individuals should report no income, something that rarely occurs in practice.

Chapter 3 explores the fiscal contract and lower income groups. Using World Bank data from approximately 90 countries from 1975 to 1999, I show that regressive taxes strongly correlate with a variety of public health indicators, including infant mortality, life expectancy, measles and DPT immunization rates, and public health spending. Because of the problem of serial correlation, most of the tests are cross-sectional, using OLS with robust standard errors. The results are compelling. To take one example: controlling for national income and other relevant factors, a country that raised the maximum level of revenue from “regressive taxes” in 1975 would have had 24 fewer infant deaths and an average life span 6.1 years greater than a nearly identical country that garnered the minimum level of revenue from “regressive” taxes. That same year, a country that acquired the maximum revenue from “progressive” taxation would have had no fewer infant deaths and an expected life span 6.7 years shorter than a country that raised the minimum from capital. Total government revenue, meanwhile, had no effect. These results are quite consistent across time and countries and hold with a variety of control variables.

Chapter 4 examines the fiscal contract and upper income groups. Using data from approximately 80 countries and a variety of measures of property rights, I show that progressive taxes strongly correlate with property rights protection. To take one example: using the Heritage Foundation and Wall Street Journal’s measure of property rights, a country that raised the maximum level of revenue from progressive taxes in 1995 would have had a 1-point increase in property rights protection (roughly one-standard deviation) compared to a nearly identical country that garnered the minimum level of revenue from progressive taxes. Neither total revenue, nor regressive taxes had a systematic effect. As with the public health measures, these results hold controlling for national income and other relevant factors, including legal systems.

Chapter 5 focuses on the fiscal contract in democracies, using OLS, a generalized estimating equation (GEE) and panel-corrected standard errors (PCSE). It shows that democracies also convert revenue from regressive taxes into basic public goods and revenue from capital into property rights—falsifying the median-voter model. The main exceptions are India and failed democracies, where there seems to be no relationship between regressive taxes and the output of basic public goods. My results also show that democracies are more efficient: for any level of revenue, democracies have lower infant mortality, longer life expectancies and more public health spending, per the encompassing interest model of democracy.

Chapter 6 traces the evolution of the tax system in Brazil, where I show that earmarking has been a solution to abysmal state performance. In order to obtain revenue, successive Brazilian governments—both non-democratic and democratic—have increasingly relinquished control over spending; earmarked taxes have gone from less than 15 percent of central government revenue in 1960 to around 50 percent in 2000, as the state has had to make credible commitments that it will spend new revenue in accord with taxpayer preferences. In the case study, I use public opinion and tax incidence data to show that while lower income groups pay a higher proportion of their income in taxes than other income groups, they also think they are getting a better deal. They are the least cynical about public spending and the most likely to access public services, according to independent data about consumption. Finally, I present data showing that the Brazilian states that have increased their tax receipts the fastest have also been the ones that have increased spending on health and education the most.

Chapter 7 concludes. It summarizes the important points and sets out the research agenda that follows from my findings. The main tasks are as follows: First, my dissertation makes strong claims about the extent (or lack thereof) of redistribution, which I claim is not a long-run equilibrium. My results provide strong evidence to support that claim: using a variety of dependent variables, I show that cross-class redistribution is quite limited in practice—even in democracies; to a certain extent, however, these findings conflict with the findings of people who study only OECD countries, where strong partisan effects have been found, especially with respect to spending and transfers (though not necessarily taxes, a point noted in my literature review). The critical question is whether these partisan effects are ephemeral (which is perfectly plausible in my model), or enduring. I am in the process of recoding data from the Luxembourg Income Studies, which would allow me to formally test hypotheses of this nature. Second, my dissertation controls for economic integration (e.g., trade), but there is room for a more in-depth analysis of the effects of globalization on changes in the tax system.