THE EFFECT OF TRADE OPENNESS AND INCOME ON THE SIZE OF A GOVERNMENT.

• Author: Turan, Taner

1. Introduction

   Government size, generally measured as government expenditures as percentage of GDP, has remarkably increased in many countries especially after 1970s. This increase is not only limited to developed countries. Some developing countries have also experienced a considerable increase in the government expenditures over time. For example, this ratio increases from 11.5% in 1960 to 26.1% in 2013 for Turkey, and from 18.6% in 1970 to 30.2% in 2011 for South Korea.

   The determinants of government size have been extensively examined for a long time. In this context trade openness and economic development have attracted much interest from researchers. There is almost an agreement that both trade openness and economic performance can have a positive effect on government size. The relationship between trade openness and government size is first discussed by Cameron (1978, pp. 1249-1250), but developed and extended from a different perspective by Rodrik (1998, pp. 997-998) who argues that trade openness has a positive impact on government size due to higher exposure to external risk, called compensation hypothesis in the literature. The other possible determinant suggested by Wagner (1883, pp. 1-15) states that the government size or expenditure rises with the level of income and development, known as the Wagner's law or hypothesis.

   This paper examines the effect of trade openness and per capita GDP on the size of government for Turkey and South Korea by means of ARDL approach to co-integration. There are two main reasons for us to restrict our analysis to only these two countries. First, these two countries were relatively similar in terms of some economic indicators, such as per capita income and the size of government, in 1960-1970s among OECD member countries. But they differed economically as the time passed, and now South Korea is a developed country, whereas Turkey is an emerging economy. It is clear that during the period examined South Korea has had a higher growth rate and openness level than Turkey even though these two countries have experienced a significant rise in the government size in the period observed. Openness and income growth seem to initiate very different mechanisms in these countries. Since there is a considerable increase in the government size, per capita GDP and trade openness in both Turkey and Korea, our analysis can be very helpful to reveal the relationship among these variables by means of time series methods. As pointed out by some studies, such as Islam (2004, p. 996), analyses utilizing time series approach would be more helpful to shed light on the subject in question. Secondly, although we also thought to enlarge our sample by including some OECD countries similar to Turkey in some aspects, such as Mexico, we could not obtain a long enough dataset for the government total expenditures, the best proxy for government size regarding the matter in hand in our opinion, to perform robust econometric analysis.

   Our results imply the importance of using a proper proxy for government size. We also found that per capita GDP has a positive and significant effect on government size for both Turkey and Korea in the long run while it has a negative impact for Korea in the short run. However, trade openness has a negative (positive) effect in Turkey (Korea) whereas its impact is significant in the short run only for the Turkish case. There are two unique contributions to the relevant literature in this study. First, to employ several different proxies of the government size is vitally important because not all measures of government size are appropriate to be used in the estimates. The other contribution is that, while testing the Wagner's hypothesis for Turkey and Korea, we consider the impact of trade openness on the nature of the relationship between per capita income and government size.

   This paper is organized as follows: section 2 reviews the literature, section 3 presents the dataset and unit roots tests, section 4 describes the empirical methodology and reports estimation results, and finally, section 5 concludes.

2. Literature review

   Cameron (1978, pp. 1249-1250), who is the first to point out the positive effect of trade openness on government size by drawing attention to the substitution between external and internal goods, mentions as a possible role for the government to lessen the vulnerabilities of an open economy. However, Rodrik (1998, pp. 997-1032) examines this relationship in a more comprehensive way and makes it a popular idea. He highlights the role of government in providing insurance as a response to higher external risks arising from greater trade openness as the driving force for the relation and shows that this is not restricted to only developed countries but also holds for other ones. In other words, trade openness associated with higher volatility leads to an increase in the government size as the government tries to alleviate the effect of higher income volatility. Epifani and Gancia (2009, pp. 629-630) suggest that openness can have a positive effect on government size mainly through two factors: terms of trade externality and higher demand for insurance. The former is related to lowering the cost of domestic taxation while the latter is associated with increasing public transfers due to higher risks arising from the greater openness.

   There are many studies trying to explain empirically the relationship between trade openness and government size, especially since 1990s. Some studies lend evidence for a positive relationship between trade openness and government size. In an influential study, Rodrik (1998, p. 998) reports a positive correlation between trade openness and government size for a large sample of countries and initiates a new literature. Alesina and Wacziarg (1998, pp. 305-306) provide some evidence for the positive relationship between the openness and size of government transfers, and point out the importance of the size of the countries. However, Ram (2009, pp. 213-218), using data for more than 150 countries, presents some evidence against the argument of Alesina and Wacziarg (1998, pp. 305-306). He reports a positive and significant effect of openness on the size of government and determines that openness has a negative effect on per capita GDP. In a comprehensive study, Shelton (2007, pp. 1230-1231), using data for a large sample of countries, finds a positive relationship between government size and trade openness but provides only a limited evidence for the risk mitigating role of the government size on social protection mechanisms such as social transfers and public employment. Epifani and Gancia (2009, pp. 631-646) report a strong and positive relationship between openness and government consumption for 143 countries while per capita income has a negative effect. Lin, Li and Sim (2014, pp. 783-808) conclude that trade openness has a positive effect on government size, measured as government consumption expenditures, for small developing countries. Also, Abounoori and Ghaderi (2011, p. 170), for MENA countries, and Shahbaz, Rehman and Amir (2010, p. 114), for Pakistan, report a positive relationship between trade openness and government size.

   However, using the bounds testing approach to co-integration, Islam (2004, p. 996) finds that a significant relationship exists for only the US and Canada but not for Australia, England, Norway and Sweden. He reports a negative (positive) relationship between openness and government size for the US (Canada) while a positive relationship between per capita GDP and government size is found for all countries examined. Molana, Montagna and Violato (2004, pp. 2-3), using time series data for 23 OECD countries, do not find any strong evidence for the positive relation for the majority of the countries examined but some weak evidence is reported for a few countries. Garen and Trask (2005, pp. 533-534), using non-budgetary items, conclude that the government size is larger in less open economies. But, they report a positive relationship using the government expenditure as a measure of government size for 96 countries. They also point out the importance of controlling per capita GDP that has a positive impact when examining the effect of openness on the government size.

   A number of studies fail to find a positive relationship between the government size and openness. For example, Liberati (2007, pp. 215-247) for 16 developed countries, Benarroch and Pandey (2008, pp. 157-159), using panel data for 96 countries, Benarroch and Pandey (2012, pp. 239-241), using disaggregated data for 119 countries, conclude that openness does not have a positive impact on the size of government. Moreover, Benarroch and Pandey (2008, pp. 239) report a negative casual effect running from government size to openness. Aydogmus and Topcu (2013, pp. 321-322), using data for Turkey over the period 1974-2011, fail to find any effect of openness...