آیا اقتصادهای باز تر دولت بزرگتر دارند ؟ نگاهی دیگر

We reconsider the evidence of a positive relationship between openness and the size of government recently established by Rodrik [Rodrik, D., 1998. Why do more open economies have bigger governments? Journal of Political Economy 106, 997–1032] and others. While the existing literature focuses on expenditure based measures of government, we consider non-budgetary measures such as government ownership of enterprises, price controls, asset expropriation. Each of these may have little impact on government expenditures, yet can make the role of government sizable. We demonstrate that the scope of government is much larger in less open economies when considering non-budgetary measures. Additionally, we show that higher levels of non-budgetary government are positively correlated with trade barriers. Finally, we provide further evidence on the hypothesis that larger governments provide social insurance in open economies subject to terms of trade shocks. Although we find that greater terms of trade risk is weakly associated with larger government in all forms, these interventions include price controls and greater risk of asset expropriation which are hard to reconcile with a pure insurance motive.

The relationship between the size of government and openness to trade has been examined extensively in the recent literature. Conventional wisdom suggests that smaller governments will be associated with more openness since both indicate less intervention in the economy. In addition, open economies may have smaller governments as intervention is either less effective in more open economies or interventionist policies reduce competitiveness on global markets. However, government size could increase in response to greater openness. Larger governments may occur in more open economies to compensate those groups which lose from trade.1 Empirically, the existence of a strong, positive correlation between openness and the size of government among the OECD countries has been established in the literature for some time.2 More recently, a series of papers by Rodrik, 1997a, Rodrik, 1997b and Rodrik, 1998 demonstrates that this correlation exists for a broad sample including both developed and developing countries. These papers have generated renewed interest in the literature as authors have sought explanations for the correlation.3 In the present paper, we pursue a different route and investigate whether or not the correlation between openness and the size of government is robust to the measure of government. Although there are a variety of ways in which the size of government is measured in the literature, they are almost exclusively expenditure based. However, the influence of government in an economy goes beyond its spending and tax collections. State ownership of enterprises, price controls, mandates, and restrictions on competition are examples of government intervention that can have profound effects on an economy yet have little impact on the government budget. Thus, government may be “large” with a relatively small expenditure.4 In fact, our prior expectations are that non-budgetary aspects of government are very important in understanding the link between government size and openness because, in a broad sample of countries, differences in institutions and the relative size of the budgetary and non-budgetary components of government are likely to be significant. In the current paper, we examine the relationship between openness and both budgetary and non-budgetary measures of the size of government. Our results indicate that focus on strictly budgetary measures of government misses much of the picture regarding the relationship between the scope of government and openness. We demonstrate that, in many respects, the scope of government is much larger is less open economies. In addition, we show that much of the association between the various forms of government and openness is explained by per capita GDP. Lower income economies tend to have more non-budgetary intervention and more closed economies while high income countries are more open and have greater government expenditure. As a corollary, we demonstrate that higher levels of non-budgetary government, in all forms, are positively correlated with trade barriers, providing evidence of the link between types of government and degree of openness. The motives of government may be important in understanding the association between government size and openness. Indeed, Rodrik (1998) presents evidence to support the hypothesis that larger governments provide social insurance in more open economies facing higher terms of trade risk. As Hillman (2003) indicates, this can arise with a government benevolently maximizing a social welfare function that encompasses risk aversion.5 If openness is associated with greater risk, we expect greater openness to be related to greater ex post public expenditures to provide greater ex ante social insurance. However, in contrast to this idea, another of our principal findings in this paper is that less openness and higher terms of trade risk are related to non-budgetary forms of government such as greater use of price controls, asset expropriation, and contract repudiation. This is hard to reconcile with an insurance motive. It is more consistent with politically motivated government use of trade barriers and other non-budgetary policies to cater to special interests.6 It is possible that governments simultaneously provide benevolent social insurance and also pursue political objectives to favor special interests with non-budgetary policies. Indeed, Milanovic (2000) finds evidence of this where median voters in wealthy countries redistribute income to poorer households through budgetary transfers while non-budgetary policies favor special interests. The remainder of the paper is organized as follows. After describing the data in Section 2, in Section 3 we present our empirical analysis. We begin by examining some basic correlations between openness and various measures of government to gain a sense of the plausibility of our priors that non-budgetary aspects of government vary in much different ways that budgetary aspects. We show that in fact, the relationship of openness to government varies depending on the measure of government. We present evidence that less open economies are highly interventionist and their governments “large” as measured by non-budgetary forms of government. Using expenditure measures of government, generally we find a positive association between government and openness. In examining the influence of income, we show that controlling for per capita GDP explains a large proportion of both the positive association of openness to expenditure based measures of government and its negative association with the non-budgetary measures. We find very similar results for another form of government-trade barriers. Trade barriers are negatively related to government expenditure and GDP per capita, but positively related to a range of non-budgetary measures of government. These findings dovetail with our other results: trade barriers are an aspect of government intervention that occur together with other forms of intervention, a less open economy, less government expenditure, and lower GDP per capita. In Section 4, we examine the relationship between the various measures of the size of government, openness, and the risk of terms of trade shocks. As argued in Rodrik (1998), if larger governments provide social insurance against terms of trade risk, one expects greater government in countries with both greater openness and greater terms of trade risk. We do find some evidence of this effect with government consumption, consistent with Rodrik (1998). However, in our analysis, we find little evidence of this effect with government expenditures. Additionally, aspects of government involvement such as government ownership, government expropriation of assets, government repudiation of contracts, and trade barriers are associated with greater terms of trade risk. While it is possible that some of these government activities could be used to provide insurance against terms of trade risk, it would seem that they have major distortionary effects and serve special interests. Section 5 concludes the paper. Our findings imply that understanding the relationship between government and openness is complex. One needs to understand the positive relationship between openness and government expenditure and the negative association of openness and other forms of government. A clear conclusion of our paper is that focus on budgetary measures of government misses much of the picture regarding the role of government and its relationship to international trade. Any explanation of the association of government and openness should include non-budgetary aspects of government; otherwise it will be seriously incomplete.

Our examination of the data gives us a different perspective and conclusion than Rodrik (1998), who states, “The scope of government has been larger, not smaller, in economies taking greater advantage of world markets.” In contrast, we find that, in many respects, the scope of government is much larger in less open economies. Less open economies tend to have less government expenditure, but have a great deal more government in other forms. The lower government expenditure gives a misleading view; the role of government is large in these countries. Moreover, the nature of the non-budgetary forms of government seems to indicate an arbitrary control by the state that the budgetary forms do not. Thus, governments with high values of the non-budgetary measures are large, but also different from governments with high levels of expenditure. Further empirical analysis leads us to question that the provision of insurance is the sole explanation for the correlation of openness and government expenditure. Government expenditure is not higher for countries with greater exposure to trade shocks. Government consumption is higher for these countries, but so are trade barriers, government ownership, use of price controls, and the chances of expropriation and of repudiation of contracts. Although governments may be providing social insurance in some respects, it is difficult to argue that this set of government activities provides value-increasing insurance for the citizens of those countries. The inter-relationships uncovered by Rodrik (1998), various authors in the literature, and in this paper are still not completely understood. It is clear, however, that focus on budgetary measures of government misses much of the picture regarding the role of government and its relationship to international trade. This is particular true since non-budgetary measures of government tend to be inversely correlated with budgetary measures. Any explanation of the association of government and openness should include non-budgetary aspects of government; otherwise it will be seriously incomplete.