A Review of Wagner's Law by Using Total Government Expenditures and Its Subsets ( )

Transcription

1 A Review of Wagner's Law by Using Total Government Expenditures and Its Subsets ( ) Sajedeh Akbari Nouri Department of economic, college of management, Arak Branch, Islamic Azad university, Arak, Iran. Abstract Over time and by elaboration of different theories and models in regard to government expenditures and economic growth, Wagner s law is considered one of the first methods of explaining public sector. In this study, reliability of Wagner s theory has been examined in Iran in the period of by using annual data on the components of government expenditure through autoregressive distributed lag (ARDL) model. The obtained results indicate that there is only a longterm relationship between security and public order spending, public spending, municipal spending, and spending on education and spending on social welfare, and national income. Wagner s theory states that the elasticity of government expenditures should be greater than one in proportion to its income; while, this requirement only exists for the municipal spending and spending on social welfare, and Wagner s law is confirmed for these two components of state expenditures. Key words: government expenditures, national income, Wagner s law, autoregressive distributed lag model Introduction From among the factors effective in determining gross domestic product (GDP) growth, government expenditures are of special importance. Empirical evidence has indicated the effect of government expenditures on GDP growth. When GDP regresses on government expenditures, researchers insist that the causal relationship is from the latter to the former; this theory is consistent with Keynes (causal relationship from government expenditure to GDP growth), or Wagner s law (causal relationship from GDP growth to government expenditure) and/or there is twoway causal relationship between these two variables (Thornton, 1998). Governments spend a considerable portion of their revenues on consumption expenditure. Understanding and the way of mutual influence of important variables such as state expenditures and economic have always been taken into account by economists and policymakers. Government expenditure can on one hand be considered an exogenous factor and affect economic growth in the form of political instruments (Keynes s viewpoint) and on the other hand, such expenditures, as an exogenous factor, may be the result of the growth (Wagner s law).government revenues as a factor affecting government expenditures is one of the features of modern global economy. The increasing trend of state expenditures in sectors such as social security, education and social welfare shows the role and share of these expenditures in human communities. Nevertheless, today, sustainable economic growth will not be achieved except by enhancing the level of welfare and health of the society which requires increasing government expenditures in these sectors. Review of government revenues will indicate that these two sectors are directly moving towards each other. We intend to try to be able to step forward in order to adopt a sound and consistent plan for economic growth by examining the aforementioned model in our country; therefore, this study intends to assist in development and improvement of our country by examining new aspects of the effects of country s revenue on state expenditures. The aim of the present study is to examine the relationship between components of government expenditures and GDP of Iran by using the data related to 1960 to 2006 and finally, estimates the elasticity of components of government expenditures in proportion to Iran s GDP. Literature Review Assessment of the Size of Public Sector A standard framework for assessing the size of public sector and the factors determining it was proposed by German economist Adolf Wagner in Based on Wagner s law, if income increases in one economy, the relative size of public sector will increase as well. Wagner provides several reasons for this observation. Wagner believed that the need for making new contracts and rules will increase by the growth of industry. As a result, a type of administrative and judicial system is required for attending to above matters. On the other hand, industrialization trend itself leads to urbanization and expansion of cities, and thus, more need is felt for state public services. Therefore, government should act in order to meet such needs. Then again, when real revenues of the community increase, public services spending such as education, healthcare and culture, 255

2 A Review of Wagner's Law etc. will increase because of increased demand, and since income elasticity of demand for these commodities and services is greater than one, these expenditures will increase in higher proportion by the increase of income. Wagner has examined the public sector growth in European countries, America and Japan, and his theory is based on these experiences. Another justification exists for public sector growth. The first case is based on fiscal illusion. For example, Wagner (1967) argued that consumers estimate the true spending of public expenditures at less than the correct amount, and thus, their demand for public goods is higher than the amount they are willing to pay. Fiorina and Noll (1978), Niskansen (1971) and Romer and Rosenthal (1978) argued that in electoral democracies, individual voters may not be able to control the exclusive power of employees looking for selfinterest. Therefore, the government can take advantage of this weakness and increase its spending. Another justification was presented by Peacock and Wiseman (1967). They believed that national state of emergency (such as wars) increases public tolerance for imposing higher taxes, and it appears as a ratchet effect, in such a way that higher taxes will remain after the state of emergency. Another interpretation was provided by Meltzer and Richard (1983) which was based on the distributive role of government expenditures in exemplar democracies. Based on this idea, low demands for redistribution of income is reflected more profoundly by increase in income gap. Theories of the Relationship between Economic Growth and Government Expenditures Theory of the relationship between economic growth and government expenditures is one of the complicated economic matters which was first proposed by Habs in 1965 and many discussions and studies has been conducted by economists of different countries in this regard. The results of the conducted studies on the effect of size of government on economic growth fall under two categories; the first category includes studies which consider the effect of the size of government on economic growth to be positive. These studies believed that the positive effect of the size of government is due to the performance of the government in areas such as property rights, establishment of a sustainable monetary system, development of infrastructures, production of public goods, national security and establishment of just judicial system which pave the way for sustainable growth. The second category includes studies which consider the effect of increased government expenditures on economic growth to be negative and point out three important factors for this matter; the first factor is tax increase and government borrowing to finance its expenditures which leads to the transfer of resources of private sector to the government and finally, decrease of investment in private sector and economic growth. The second factor refers to the effect of the increase of share of public sector in economy and diminishing return due to government expenditures. By increase of government expenditures after financing prioritized public goods, the resources will be allocated to providing commodities in production of which the government is not involved and in this case, diminishing return of government expenditures will result in decreased growth. The third factor refers to this matter that political mechanisms of the government for handling the affairs have less dynamism compared to market process and thus increase of government size can have a negative effect on economic growth. Wagner s Theory Wagner is a 19 th century German economist who has had special attention to matters related to government s economy and public spending. His theory states that if per capita income increases in one economy, the relative size of public sector will increase as well. If GNI and per capita income increase, government expenditures will also increase and the rate of increase in government expenditures will be higher than the rate of increase of national income. Wagner has examined the public sector growth in European countries, America and Japan, and his theory is based on these experiences.wagner did not present his opinion in a form of a law, but next economists have called Wagner s opinion the Wagner s law and have made it widely known. The Wagner s law can be stated as follows: with the increase of per capita income in economy, relative size of public sector will increase and this increase is higher than the increase of per capita income (Pazhooyan, 1993). Viewpoints and Methods Derived from Wagner s Model Even though Wagner s model is one of the pioneers in examining the size of the government and the reasons for its growth, there have been criticisms and analyses about it, the most important of which are presented as follows. Bird s Viewpoint Bird (1971) believed that three conditions are necessary for implementing Wagner s law. 652

3 1 The increase of per capita income is placed in the economy; 2Some kind of technological and organizational changes are occurred. 3 There are democratic grounds in the concerned society and within the political frameworks. Bird emphasizes three elements for justifying the growth of government. One is performing internal and external administrative and protective tasks by the public sector. Second is government s assurance of appropriate performance of market forces and finally, the third is organizing or procuring social and cultural products by the public sector. Peacock & Wiseman s Viewpoint Peacock & Wiseman based their analysis on the political theory determining public expenditures. These theories are stated in this way that governments tend to spend more, while people do not like to pay higher taxes and the governments inevitably pay attention to demands of the people (Pazhooyan, 1993).Peacock & Wiseman consider voters to be people who enjoy the benefits of public goods and services. But they do not like to pay taxes, as a result, when the government makes decision about its expenditures, it should pay enough attention to the opinions of the voters for compensating for the expenditures. According to this theory, they believed that an acceptable level of taxes exists which acts as a restriction for the government s behavior. It means that the government knows that it should not incur expenditures which require it to inevitably exceed the boundary of acceptable taxation. In Peacock & Wiseman s opinion, taxes create a restriction for government expenditures. As a result, when the government cannot raise tax rates, increase of its income only is possible through raising its national income. With the growth and increase of national income, tax revenues increase at a fixed rate as well, and as a result, this allows that government expenditures increase along with the growth and increase of GNI. Therefore, under normal circumstances, government expenditures will have gradual increase. However, in undesirable social conditions, this regular trend in increase of government expenditures will be disrupted. Undesirable social conditions such as war, famine or incidents like floods and earthquakes which require rapid increase of government expenditures, create an obligation to raise taxation levels. Ram s Viewpoint It is possible that test results be different when time series data are used from when crosssectional data are used. For instance, Ram concluded that Wagner s law can be supported by using time series analyses but if, finally, in our opinion, application of analyses is in a crosssectional form, it cannot be supported. Sometimes, the word law for the study of Wagner cannot be compatible in all cases and situations. Review of Related Literature Yak (2005), by examining the England s economy in lingterm ( ), believed that Wagner s law is defendable in case of that country.ahmed (2005) examined the relationship between size of government and per capita income for the D8 group (Bangladesh, Egypt, Indonesia, Iran, Malaysia, Nigeria, Pakistan and Turkey) for the period Research results indicated that there is Granger causality between the size of government and per capita income only in Iran, and there is no such relationship in other countries. Narayan et al. (2006) tested Wagner's law of increasing state activity using panels of Chinese provinces. The paper's main methodological contribution is in that they employed for the first time in the literature on Wagner's law a panel unit root, panel cointegration and Granger causality testing approach. Overall, they found mixed evidence in support of Wagner's law for China's central and western provinces, but no support for Wagner's law for the full panel of provinces or for the panel of China's eastern provinces.sinha (2007) examined Wagner s law in six different situations for Thailand in the time period of He concluded that the convergence between government expenditures and GDP is significance at the level of 90%. Rehman et al. (2007) examined Wagner s law by using time series data in the period in Pakistan and concluded that Wagner s law is compatible for Pakistan s economy. Abul Kalam & Aziz (2009), by using series data in the period , concluded that Wagner s law is defendable for Bangladesh.Hussain et al. (2010), in studying the effects of economics, population and export on government expenditures in Pakistan s economy for the period of , concluded that economic growth, population and export have positive and significant effect on government expenditures both in longterm and shortterm. Afzal and Abbas s (2010) study indicated that it is only compatible for the Pakistan in the period Akpan (2011) examined Wagner s theory by using time series data for for the Nigeria s economy and concluded that Wagner s law is defendable in Nigeria.Komeijani and Nazaru (2009), in their study in the time period of by employing the vector autoregression (VAR) model, concluded that government expenditures 652

4 A Review of Wagner's Law have a positive effect on economic growth.shafeei et al. (2006), in their study in the time period of by employing autoregressive distributed lag (ARDL) model, concluded that construction costs have a direct and significant impact on economic growth, while consumption expenditure does not have significant impact on this variable. Research Methodology The methodology of the study has been descriptive, analytical and econometric. First, library method has been employed for theoretical topics and the related literature has been gathered and presented. Then, after collecting data from Statistics Center website, and with the help of MicroFit software, econometric analyses as well as required estimations have been carried out. The present statistical population is Iran.Since it has been attempted in this study to investigate the effect of GDP on the components of government expenditure as well as the total government expenditures, first, augmented unit root Dickey Fuller test has been used to measure the reliability and unreliability of the variables. Afterwards, ARDL model has been used to examine the relationship between variables in shortterm, estimate the coefficients of the longterm model and estimate error correction mode (ECM), and the significance of the variables has been determined. The reason for adopting this method among other methods is that, according to Wagner s theory, at the right side of the equations only the income of the whole country and/or per capita income exists, while it might be possible that other variables affect government expenditures; but Wagner s theory does not allow other variables other than income and/or per capita income to be inserted at the right side. There has been no variable other than income or per capita income at the right side of the equations in all the empirical studies in regard to Wagner s law. So we may make a mistake in specifying the model because of removing other variables affecting different forms of government expenditures. But this problem is solved by using ARDL, because this method allows the lags of dependent and independent variables to be inserted at the right side. Estimation of Empirical Model and Data Analysis Test of Stationary Even though there is no restriction on the degree of variable accumulation in ARDL method, existence of variables with degree of accumulation greater than 2 can lead to disruption of the results. Therefore, first, variables have been tested in terms of reliability. In order to measure the stationarity and nonstationarity of the variables, augmented unit root Dickey Fuller test has been adopted and the null hypothesis about the existence of unit root is tested. For this purpose, the optimal lag length is selected based on Schwartz Bayesian Criterion (SBC) and test statistic and critical values are compared at the level of 5 percent. The results of this test are presented in table 1. Variables Level First Difference Including y Including y Including y Including yintercept intercept intercept and trend intercept and trend LSS LMS LCS LES LO LHS LSW LH LE LS LD LPS LM LTE LNI The critical value of the test at the level of 95% in the model including yintercept is 2.92 and in the model including yintercept and trend is

5 Dickey Fuller test has been carried out in the model including yintercept as well as yintercept and trend. The null hypothesis is that variables have unit root in level values and thus they are nonstationary; while, alternative hypothesis states that variables do not have unit root in level values and thus they are not stationary. Number of lags are selected automatically based on Schwartz Bayesian Criterion. In the model including yintercept at the level all variables are nonstationary because the obtained statistic value is less than the critical value of the table (2.92) at the level of 95%, but after obtaining the first difference for all the variables, they have been turned into stationary variables. It means that accumulated time series are of order one. We know that in ARDL method, we do not need the stationarity of variables to estimate the model; they should not only have degree of accumulation greater than 2. Effect of GDP on Government Expenditures Separately for Public SpendingAffairs ShortTerm Estimation Of the Effect of National Income on Public Spending In order to estimate the shortterm or dynamic relationship between variables, first, the number of optimal lags of the model should be determined by four criteria of adjusted coefficient of determination (R2), Akaike information criterion (AIC), Schwartz Bayesian Criterion (SBC) and HannanQueen (HQ); the selected criterion for estimation of the model is SBC. The reason for selecting this criterion is that it allows us to estimate the coefficients with minimum lag. The results of estimation are presented in table 2. Table 2. Results of estimation of the effect of national income on public spending (dependent variable LPS) Wagner s theory Significant Acceptable Probability level t statistic 3.15 Variable coefficient Rejected Acceptable Acceptable R 2 = 0.67 DW= 2.21 F(2.42)=43.81 (0.000) Source: research findings Variable Public spending with one lag National income yintercept Variable LPS(1) Public spending with one time lag has had a positive and significant effect on public spending in the current period. The coefficient of public spending with one time lag is equal to In other words, if public spending with one time lag increases by 1 percent, public spending in the current period will increase by 0.46 percent. The sign of the above variable is as per the theoretical expectations and it is acceptable at the level of 95 %. The coefficient of national income variable is acceptable statically and in terms of the sign, and it indicates that if national income increases by 1 percent in the current period, public spending will increase by 0.20 percent. The coefficient of this variable is acceptable at the level of 95%.Durbin Watson statistic in the fitted model is greater than 2 which indicates lack of autocorrelation in the fitted model. In order to examine the disproof or proof of classic hypotheses in the estimated model, serial correlation test, correct specification test, and heteroscedasticity test were carried out as well, and the results indicated the acceptance of above hypotheses. In other words, it can be said that the research model is well fitted. Review of Wagner s Law Wagner s law states that causal relationship is from the national income to government expenditures and since national income is the independent variable, the sole significance of national income in estimated models confirms this part of Wagner s law, but the next part of Wagner s law states that increase of expenditures is higher than increase of national income, and this only can be confirmed if the elasticity of national income is greater than one in the estimated models, or in other words, its coefficient is greater than 1. Therefore, in general, Wagner s law is established when, firstly, national income coefficient or its lags are significant in the relationship of government expenditures and secondly, its coefficient is greater than one. Banerjee, Dolado and Mestre s Convergence Test After estimation of the shortterm equation, first, the existence of longterm relationship must be ensured. In order to examine the existence or nonexistence of the longterm relationship, Banerjee, Dolado and Mestre s convergence test has been used. In this test, if the sum of the coefficients of the dependent variable is greater than 1, shortterm relationship between variables of the model will not be established towards a longterm equilibrium. The test statistic of this theory can be calculated as follows: NI L C 652

6 A Review of Wagner's Law p i 1 p i 1 i S i 1 The calculated statistic is equal to Because this number is greater than critical value of the Banerjee, Dolado and Mestre s table (3.28) in terms of absolute value, null hypothesis is rejected and the alternative hypothesis about the existence of longterm relationship at the level of 95 percent is accepted. Estimation of LongTerm Relationship After passing through above stages, longterm coefficients of the estimated model can be estimated through autoregression method; if it is observed in estimation of model through autoregression method that the long term effect of the variables of government public spending is not possible through shortterm estimation, it is because values with dependent variable lag are effective in the dependent variable itself. Besides, some variables have appeared in the model with lag, and we need longterm estimation for their definite effect. Results of longterm estimation are presented in table 3. Table 3. Results of longterm estimation for LPS Wagner s theory Result Level of probability t statistic Variable coefficient Variable Rejected Acceptable (0.000) LNI Acceptable (0.061) c Source: research findings Given the results of longterm estimation, it is seen that national income variable is statistically significant and has the expected sign. It indicates that 1 percent increase in national income; public services expenditures will increase by Error Correction Model ECM coefficient indicates that what percentage of the shortterm imbalance is adjusted in order to reach longterm. Adjustment coefficient or error correction coefficient in the fitted mode is equal to 0.53 which indicates that 53% of the imbalances existing in one period in the aforementioned relationship are adjusted in the next period every year. Review of the Effect of National Income on Other Components of Government Expenditures Given the similarity of calculations process for other components, the calculations are not mentioned and the related results are presented in table

7 Error correction coefficient 31% of the imbalances existing in one period are adjusted in the next period 17% of the imbalances existing in one period are adjusted in the next period 23% of the imbalances existing in one period are adjusted in the next period 36% of the imbalances existing in one period are adjusted in the next period Table 5. Effect of national income on other components of government expenditures Results of longterm relationship National income with the coefficient of has significant effect and Wagner s law is rejected National income with the coefficient of 1.18 has significant effect and Wagner s law is confirmed in longterm National income with the coefficient of 0.88 has significant effect nd Wagner s law is rejected in longterm National income with the coefficient of 1.64 has significant effect and Wagner s law is confirmed in longterm Banerjee, Dolado and Mestre s convergence test Rejection of longterm coefficient of 0.11 Confirmation of longterm coefficient of 3.34 Rejection of longterm coefficient of 2.88 Rejection of longterm coefficient of 1.87 Confirmation of longterm coefficient of 4.22 Confirmation of longterm coefficient of 3.57 Confirmation of longterm coefficient of 4.10 Rejection of longterm coefficient of 2.26 Rejection of longterm coefficient of 2.83 Rejection of longterm coefficient of 2.66 Rejection of longterm coefficient of 2.66 Rejection of longterm statistic of 1.89 Rejection of longterm statistic of 1.24 Effective variables in the shortterm relationship at the level of 95% First lag of ministries and institutions with the coefficient of 0.973, national income with the coefficient of and second lag of national income with the coefficient of (Wagner s law is rejected) First lag of public order and security spending with the coefficient of and national income with the coefficient of (Wagner s law is rejected in shortterm) First lag of public defense with the coefficient of and national income with the coefficient of (Wagner s law is rejected in shortterm) First lag of social security spending with the coefficient of (Wagner s law is rejected in shortterm) First lag of municipal spending with the coefficient of and national income with the coefficient of (Wagner s law is rejected in shortterm) First lag of spending on education with the coefficient of and national income with the coefficient of (Wagner s law is rejected in shortterm) First lag of spending on social welfare with the coefficient of and national income with the coefficient of (Wagner s law is rejected in shortterm) First lag of expenditure on recreation and culture with the coefficient of and national income with the coefficient of0.339 (Wagner s law is rejected in shortterm) First lag of health expenditure with the coefficient of (Wagner s law is rejected in shortterm) First lag of housing expenditure with the coefficient of (Wagner s law is rejected in shortterm) First lag of economic expenditures with the coefficient of (Wagner s law is rejected in shortterm) First lag of other government expenditures with the coefficient of and national income with the coefficient of 1.63 (Wagner s law is confirmed in shortterm given the coefficient of national income variable) First lag of total government expenditures with the coefficient of and national income with the coefficient of 0.410; second lag of national income with the coefficient of (Wagner s law is rejected in shortterm) Effect of national income Ministries and affiliated institutions expenditures Public order and security spending Public spending defense Social security spending Municipal spending Spending on education Spending on social welfare Expenditure on recreation and culture Health expenditure Housing expenditure Economic expenditures Other government expenditures total government expenditures Source: research calculations As it is seen in the above table, Wagner s law is only established for the components of municipal spending and spending on social welfare in longterm and for other government expenditures in shortterm. 622

8 A Review of Wagner's Law Conclusion and Suggestions Peacock and Wiseman s model has been used in this study to test Wagner s theory. Results indicate that: The effect of national income on public spending in shortterm is positive and significant. Wagner s theory is rejected and not acceptable in shortterm. Besides, longterm relationship between national income and public spending of the government is established but Wagner s theory is not confirmed.the effect of national income on ministries and affiliated institutions expenditures in shortterm is positive and significant. Wagner s theory is rejected; Banerjee, Dolado and Mestre s convergence test indicated that longterm relationship is not established between these variables. The effect of national income on security and public order spending is positive and a significant relationship is established in shortterm between these variables. Wagner s theory is unacceptable in shortterm; Based on Banerjee, Dolado and Mestre s convergence test, longterm relationship is established but Wagner s theory is rejected. The effect of national income on social security spending is positive and a significant relationship in shortterm is established between these variables. Wagner s theory is unacceptable in shortterm; Based on Banerjee, Dolado and Mestre s convergence test, longterm relationship is not established. The effect of national income on municipal spending in shortterm is positive and significant. Wagner s theory is rejected in shortterm and not acceptable. Besides, longterm relationship between national income and public spending of the government is established and Wager s theory is confirmed. The effect of national income on spending on education in shortterm is positive and significant. Wagner s theory is rejected in shortterm and not acceptable. Besides, longterm relationship between national income and public spending of government is established but Wagner s theory is not confirmed. The effect of national income on spending on social welfare in shortterm is positive and significant. Wagner s theory is rejected in shortterm and not acceptable because the income elasticity is smaller than 1. Besides, longterm relationship between national incomes is greater than 1 and Wagner s theory is confirmed. There is positive and statically significant and acceptable relationship between other government expenditures and national income in shortterm. Besides, Wagner s theory is confirmed in shortterm and based on Banerjee, Dolado and Mestre s convergence test, longterm relationship is not established between these variables. There is positive and significant relationship between total expenditures and national income in shortterm but Banerjee, Dolado and Mestre s convergence test, longterm relationship is not established between these variables and Wagner s theory is not established in shortterm. The overall conclusion is presented as follows based on an overall view to the above results.from among the selected variables of the study, there is only longterm relationship between security and public order spending, public spending, municipal spending, spending on education and spending on social welfare, and national income. As it was mentioned, Wagner s theory states that elasticity of government expenditures should be greater than one, and this relationship is only established between municipal spending and spending on social welfare, and national income; that is coefficient of national income is estimated to be greater than 1 in regard to these variables and Wagner s theory is confirmed.in future studies, the effect of factors such as human development index and degree of economic openness on accuracy of Wagner s law in developing and developed countries shall be examined. Since data used in this study belonged to Iran s economy, the same study can be conducted for other countries.if the need is felt for examining these variables in a study for several countries, this process can be studied by using other dynamic models such as vector autoregression (VAR) or in the form of multiple countries through panel data. 626

9 References 1. Borcherding, T. E. (1985). The causes of government expenditure growth: a survey of the US evidence. Journal of Public Economics, 28(3), Connolly, S., & Munro, A. (2009). Economics of the public (M. Sameti, Trans.). Noor Elm Publications. 3. Central Bank of IRI (2010). Annual report of Central Bank. 4. Dao, M. Q., & Esfahani, H. S. (1995). A competitive model of the growth of government. Journal of Economic Studies, 22 (2), Dogan, E., & Tang, T. C. (2006). Government expenditure and national income: causality tests for five south East Asian countries. International Business & Economic Journal, 5(1), Fakhraee, E., & Mansouri, S. A. (2008). Estimation of longterm consumption function through cointegration and calculation of the relationship of shortterm consumption in Iran. Journal of Quantitative Economics. 7. Fiorina, M. P., & Noll, R. G. (1978). Voters, bureaucrats and legislators: a rational choice perspective on the growth of bureaucracy. Journal of Public Economics, 9 (2), Gujarati, D. (1995). Basics of econometrics (H. Abrishami, Trans.). Tehran: Tehran University Press. 9. Haji, Gh. (2011). A study of the relationship between size of government and economic growth in Iran during Research Program of IAU of Arak. 10. Huang, CJ. (2006). Government expenditure in china and Taiwan: Do they follow Wagner s law? Journal of Economics Development, 31(2), Khodaparast Mashahdi, M. (2012). A study of the reliability of Wagner s law from Keynes s viewpoint for Iran s economy. Journal of Economic Studies. 12. Meltzer, A., & Richard, S. (1983). A rational theory of the size of government. Public Choice, 41(3), Mohammadi, H., Cak, M., Cak, D. (2007). Wagner s hypothesis: New evidence from Turkey using the bounds testing approach. Journal of Economic Studies, 35(1), Musgrave, R., & Musgrave, P. (1994). Public finance in theory and practice (M. Mohammadi & Y. Ebrahimifar). Tehran: Iranian Management and Planning Organization Press. 15. Musgrave, R. A. (1969). Fiscal systems. Yale University Press, New Haven, CT. 16. Narayan, P., Nielsen, I., & Smyth, R. (2008). Panel data, cointegration, causality and Wagner s law: Empirical evidence from Chinese provinces. China Economic Review, (9), Nazari, R., & Dadgar, Y. (2011). Testing Wagner s law in selected countries and Iran ( ). Research Academic Journal of Economic Development and Growth. 18. Pazhooyan, J. (1993). Economics of the public sector. Tehran University s Jihad Daneshgahi Press. 19. Peacock, A. T., & Wiseman, J. (1967). The growth of public expenditure in the United Kingdom. London: Allen and Unwin. 20. Peacock, A. T., & Wiseman, J. (1979). Approaches to the analysis of government expenditure growth. Public Finance Quarterly, 7(1), Peltzman, S. (1980). The growth of government. Journal of Law and Economics, 23 (2), Pesaran, M. H., Shin, Y., & Smith, R. J. (2001). Bounds testing approaches to the analysis of level relationships. Journal of Applied Econometrics, 16(3), Romer, T., & Rosenthal, H. (1978). Political resource allocation, controlled agendas and the status quo. Public Choice, 33 (4), Sameti, M. (1993). The effect of government expenditures on economic growth (Doctoral dissertation). 25. Statistical Center of Iran. Statistical yearbook for different years. 26. Tobin, D. (2005). Economic liberalization, the changing role of the state and Wagner s Law : Chinas development experience since World Development, 33 (5), Yousefi, M., & S. Abizadeh, (1992). Wagner s law: New evidence. Atlantic Economic Journal, 47(2),