The Political Economy of Income Inequality in Iran (unedited first draft) Naseraddin Alizadeh 1 There are different studies that aim to shed light on different aspects of inequality and distribution. These studies range from descriptive to empirical or theoretical ones that scrutinize causative relations between different factors and phenomena like poverty and disparity. Among all, economists are especially interested in main factors and the degree these factors affect economic inequality. A. Kaasa (2003) lists these factors as economic development, demographic, political, historical-cultural-natural and macroeconomic factors. Pioneering work of Kuznets (1955) was one of the attempts to explain nonlinear relation between inequality and economic development. He maintained that there are an inverse U shape relation between economic growth and inequality: in the first steps thriving economies may experience declining disparity that erase out after a while and is replaces by inverse trend. According to Kuznets up to a point two factors contribute to inequalities in opposite directions: on the one hand savings may be in the hand of limited wealthy people and on the other hand due to industrialization and urbanization the labor and profession patterns may change. As a result, inverse U pattern results from opposing and then matching relations between aforementioned factors. Macroeconomic factors are among the most important contributors to inequality. These factors are mainly related to government intervention bringing about redistribution of resources and rent-seeking opportunities. Starting from 1960s Tullock, Fordon, Buchanan, Krueger and Tollison among others analyzed some of these redistribution policies by using new approaches that formed rent-seeking studies. Most of these pioneering rent-seeking studies dealt with issues related to democratic systems in west but in order to study rent seeking in less democratic cases new considerations were needed. Although some of these less democratic countries are rich in natural resources but witness low growth rates, 1 - Ph.D candidate of Economics at Ankara university Email: na.alizadeh2005@gmail.com 1
poverty and gap between poor and rich. Theories like natural resource curse attempt to explore these contradicting aspects of less democratic resource rich countries. These study aims to examine the effects of macroeconomic policies on income inequality in Iran as an oil rich country by taking into account sociocultural and socioeconomic cleavages. Namely, this paper try to study inequality among different social groups in Iran when the exposed to different economic interventions. Government Intervention and Rent Seeking in Iran Iran is an oil rich country and has its special political structure characterized by restriction to provision of opposition parties and media. Iran is also a mosaic of different ethnics and has a Sunni minority in a Shia theocracy beside other socioeconomic and sociopolitical cleavages that any country may face all around the world. Petrol incomes always account for important economic fluctuations and form important part of annual government budget. It is expected that these petrol revenues beside the special polity and social cleavages pave way to uneven income and welfare distribution among different social groups. Rent-seeking not only can hinder efficient resource allocation and waste physical and human capital but also can exacerbate income disparity in expense of most vulnerable and minority groups income decline. Different researchers have taken different proxies to measure rent-seeking in Iran and different parts of world. These proxies involve government size, foreign exchange rate fluctuations, inflation, interest rates, oil revenues, tariffs, economic openness and subsidies among other factors. According to annual statistics during 2005-2012 petrol incomes account for about %60 of government revenues, %35 of GDP and %80 of foreign exchange from exports (Bahrami, 2008:3). Also petrol composed about %70 of Iran exports during 2001-2011 and %70 of imported goods foreign currencies came from petrol dollars. Other statistics show that during 2001-2011 government budget composed about %70 of GDP in Iran (Annual Statistics of Iran, 2001-2012). Other important source of rent has been the difference between market and official exchange rates. For instance graph (1) shows the huge difference between market and official exchange rates during 1990-2010. Market exchange rates were 17 times more than 2
official one in 1980s where the groups or individuals who have access to these sources can get windfall wealth. Graph (1): different exchange rate in Iran (in Riyal). Source: Central Bank of Iran Official statistics shows that liquidity and inflation increased faster than nominal exchange rate during 2002-2010. As graph (2) depicts liquidity and consumer index grew five and three times more than exchange rate respectively. 6 5 4 3 2 Dollar CPI Money 1 0 2002 2003 2004 2005 2006 2007 2008 2009 2010 Graph (2): Dollar stand for exchange rate, CPI for consumer index and money for liquidity. Model and Variables This study will regress different panel data models to estimate the effect of six factors on income inequality among Persian and non-persian ethnic groups and Shias and Sunnis (both 3
are two important sects of Islam; Shias are dominant religious majority in Iran). Doing so, statistics of 28 provinces of Iran during 2000 and 2013 will be used. Dependent variable is the difference between provinces income per capita and average income per capita of all country. Six independent explanatory variables used to capture the effect of macroeconomic factors, five of them can be used as proxy of rent-seeking. Below are the independent variables (all are real variables with 2011=100 or rates): 1- Economic openness: measured by the ratio of import and export on GNP. It is expected that less open countries provide opportunity for rent-seeking by restricting competition. Also this variable can measure the effect of economic sanctions that increased after 2005. 2- The difference between market and official exchange rate: As already explained the huge difference between market and official exchange rate paved way to rentseeking efforts. 3- The government investment budget per capita for each province: It is expected that they who are more organized or belong to majority Shia sect and Persian ethnics can get more investment budgets. 4- Oil income per capita: an increase in oil revenues can improve the welfare of dominant Shia-Persian group compared with other groups. 5- Inflation of each province: inflation can considered as a hidden tax and can cause contradicting effects on different groups. 6- Income per capita for each province: as explained above Kuznets was the first one who put forth the invers U shape relation between inequality and income per capita. All statistics come from Central Bank of Iran and Statistical Center of Iran. Eviews8 program is used to estimate model and test variables. Four variables out of seven have unit root and by one differentiation all turn to stationary variable (I(1) unit root). Therefore, to avoid unit root problems and autocorrelation the first differential of variables will be used. By using first differential we will lose 28 observations and intercept term, but slope estimators are as same as original model. Suppose model (1) with one independent and one dependent variables, x it and y it respectively. Here i stands for provinces (individuals), t for years, u it random error term and D 1i and D 2i are dummy variables stand for two different Persiannon-Persian and Shia-Sunni groups: 4
y it = c 1 D 1i + c 2 D 2i + β 1 (D 1i x it ) + β 2 (D 2i x it ) + u it (1) For example if we suppose D 1i =1 for Persian and D 1i =0 for non-persian provinces and D 2i = 1 for non-persian provinces and D 2i = 0 for Persian provinces we will have two models (one for Persian and another for non-persians): For Persians: For non-persians: y it = c 1 D 1i + β 1 (D 1i x it ) + u it y it = c 2 D 2i + β 2 (D 2i x it ) + u it Now by rewriting (1) for y i(t 1) and finding y it = y it y i(t 1) : y it = β 1 (D 1i (x it x i(t 1) ) + β 2 (D 2i x it x i(t 1) ) + u it y it = β 1 (D 1i x it ) + β 2 (D 2i x it ) + u it (2) Equation (1) and (2) shows that estimated β 1 and β 2 from variables and their differential are same but we will lose intercept term in (2). Model Estimation and Results Equation (2) with six explanatory variables instead of one is the main equation estimated in this paper. First of all it should be tested if all observations can be used as pooled data to estimate an OLS model with common intercept and coefficients. F-limer is a test that shows whether we can use data from all provinces as pooled data or we must consider different features of provinces. The value of F-limer calculated for both models are 475 that shows data cannot be used to estimate an OLS model. In the second step Hausman test help us to choose from Random and fixed effects method. H 0 hypotheses in Hausman test is random effect is best and H 1 is random effect is not best estimation method. In both of our estimation random effect were not rejected in favor of fixed effect and estimations were done by random effect approach. The results show that estimated coefficients for Persian and non-persian model are significant for both group but the sizes are different. Table one illustrate these results: 5
Variables Coefficient (Persian) Coefficient (non-persian) Petrol income per capita -0,563-0,54 Economic openness 19,12 15,09 inflation 10,44 8,18 The difference between market and official exchange rates 0,078 (0,02) 0,09 Government investment per capita -1,47-0,39 (0,08) Income per capita 0,92 0,68 Table (1): Persian and non-persian model The results show that all factors except the difference between market and official exchange rates have slightly severe effect on non-persian ethnics (these will be discussed in second draft) The second estimation for Shia and Sunni provinces also shows slightly different effects of changes in macroeconomic factors on income per capita differences from average income per capita of country. Table (2) summarize the results for this model. Pay attention that only two variables are significant for Sunni sect income per capita difference from average country s income per capita. Variables Coefficient (Shia) Coefficient (Sunni) -0,588-0,625 Petrol income per capita 17,40 Economic openness 15,89 11,11 4.16 inflation (0,40) The difference between market and 0,093 0,06 official exchange rates (0,31) -0.89-0,423 Government investment per capita (0,58) 0,904 0,889 Income per capita Table (2): Shia and Sunni model 6
Pay attention that only two variables are significant for Sunni sect income per capita difference from average country s income per capita. The results will be discussed in second draft. References Kaasa, A. (2003), Factors Influencing Income Inequality in Transition Economics, University of Tartu Faculty of Economics and Business Administration, www.tyk.ut.ee, Order No. 207. Krueger, A.O., (1974). The political economy of the rent-seeking society. The American Economic Review 64, 291-303 Kuznets, S. (1955), Economic Growth and Income Inequality, American Economic Review, Vol. 45, PP. 1-28. Tullock, G., (1967). The Welfare Costs of tarrifs, monopolies, and theft. Western Economic Journal 5, 224-232. B.Ang, James (2008), Finance and Inequality: The Case of India, CAMA working paper series, ss.1-25. 7