Study of the Relationship between Oil Revenues, Budget Deficit, Money Supply and Inflation in Iran Sara Farhangdoost1,*, Seyyedeh Fatemeh Naseri 2 and Mohammad Ali Falahi 3 1

MS.c Student, Faculty of Economics and Administrative Sciences, Ferdowsi University of Mashhad, Mashhad, Iran 2 MS.c Student, Energy Economics, Faculty of Economics, University of Tehran, Tehran, 3 Associate Professor, Department of Economics, Faculty of Economic and Administrative Sciences, Ferdowsi University of Mashhad, Mashhad, Iran 

ABSTRACT In this paper, the relationship between oil revenues, money supply, budget deficit, and inflation in Iran’s economy, during the seasonal interval of 1990-2011, has been examined using the vector autoregressive (VAR) model. After examining the stasis of the variables, it has been attempted to fit Johansen Cointegration Test, Vector Error Correction Model (VECM) and Granger Causality Test. To analyze the model dynamics,the two criteria of impulse response function and analysis of variancehave been used. In this study, the results show that in the seasonal interval of 1990-2011budget deficit in Iran has a significant relationship withmoney supply growth and oil revenues, while it has no significant relationship with theinflation rate. This result emphasizes the finding that the way of financing the budget deficit is a highly decisive factor which should be heeded.

KEYWORD Budget deficit , Inflation , Money supply, vector autoregressive (VAR) model

INTRODUCTION

I

n order to achieve their goals- including economic, social, political and cultural goals -governments need financial resources. However, it should be noted that due to the unpredictable nature of economic conditions and states and exogenous factors in financing, governments are usually faced with the lack of resources concerning the level of costs for achievement of their goals (Rostami, 2005). Budget shows a government’s position in economy and, in other words, the limits of the government’s intervention in the society and correct management of government finance plays a substantial role in improvement of the economic and administrative system or management of the government or

*Corresponding Author: Sara Farhangdoost E-mail r: 09155253563 Telephone Number r: [email protected]

Fax. Number r:

public sector and through these expenses the government intends to achieve certain goals and this leads to government budget deficit since the government’s financial resources do not determine the level of its expenses. The importance of the budget deficit becomes clearer when the developing and third-world countries are discussed, since in the other countries (developed countries) governments act as systems supervising and guiding economy and based on the mechanism of prices and the function of the market, while in developed countries the governments are usually too large and interfere in all issues, which adds to the budget deficit. From among the budget deficit causes in Iran, the following factors can be mentioned: Exogenous government expenses, lack of elasticity of tax revenues, excessive expenses, massiveness of the government, inflationary nature of the government’s activities and existence of time delays in obtaining tax revenues.Economists believe that temporary and moderate budget deficits cause no problem for economy and during commercial periods, having deficit during recession which is neutralized by smaller deficits or surplusesin boom, is also quite appropriate, so in common policies of funding for meeting economic and social requirements and needs the budget deficit method is one of the common methods, specifically for developed countries in order to encourage total demand during recessionand avoiding budget deficit in this state leads to more adverse effects and consequences. Financing the budget deficit in developing countries is usually done by borrowing from the central bank and leads to continuous and sustained budget deficits which are harmful for economy and eventually lead to inflation. However, the statistical evidence related to the empirical research on the two variables of inflation rate and budget deficit shows that the relationship between these two is not the same for all the countries. In fact, study of the results of different studies shows that the method of financing and macroeconomic conditions greatly affect the quality of impact of budget deficit on economic variables. The main 69

goal of this study is analysis of the relationship between budget deficit and inflation in Iran. Thus, the paper is organized in this way: in the first part, the theoretical discussions on the relationship between budget deficit and inflation are presented and in the second part empirical research on the effect of budget deficit on inflation is reviewed. Also, the third part is devoted to the analysis of variables and presentation of tests and estimates and the last part engages in conclusion. THEORETICAL FOUNDATIONS In economic literature, there is no precise answer with regard to the relationship between budget deficit and money growth and inflation, since the economic effects of the government budget deficit will depend on its formation, method of financing and macroeconomic conditions. Friedman (1968) believes that monetary authorities can control the inflation rate, specifically in the long-term, through money supply regulation. In other words, budget deficit can cause inflation but this will be realized when budget deficit is financed by the change in money supply. Thus, the budget deficit that leads to the increase of money supply will be inflationary. On the other hand, if the budget deficit is financedthroughthe bonds, it may be noninflationary (Buchanan & Wagner, 1977). Inflationary or non-inflationary nature of financing the budget deficit through the bonds depends on the method used by monetary authorities for financing the budget deficit. If monetary authorities stabilize the interest rate, then financing of the budget deficit throughthe bonds will be inflationary, since this act requires the increase of money supply, which will eventually lead to the increase of the prices. Sargent and Wallas (1981) maintain that the central bank has to finance the budget deficitsoon or late. Miller (1983) believes that the budget deficit, regardless of its financing method, is inflationary. Barro (1978 & 1979) proposed a hypothesis based on which budget deficit is the result of inflation and not the cause of inflation. He maintains that budget deficit is in fact a change in the nominal value of the bonds in the hands of people. If the predicted inflation increases, then the nominal value of the bonds must increase so that the real value of those bonds can be protected. Moreover, monetarists believe that there is a positive relationship between budget deficit and the growth of monetary variables. So, if a greater amount of budget deficit is financedthrough the bonds, greater pressure will be made for increase of the interest rate and the government bonds, since the central bank intends to pave the way for the changes of the interest rate. Thus, it will have to increase the money supply (Darrat, 1985). Furthermore, Sargent and Wallas (1981) showed that if the time path of the government expenses and taxes is exogenous, financing of the budget deficit through the bonds will not be stable since it causes the interest rate to greatly increase and the central bank will have to eventually increase the money supply in order to finance the budget deficit. This act increases the money supply and leads to inflation in the long term.

Meltezer (1989) proposed the monetarists’ method for financing the budget deficit and referred to this point that budget deficit affects inflation. He believes that Argentina, Bolivia and Brazil are good examples and instances of inflationary economies and inflation was caused in these countries due to the increase of money supply for financing the government expenses in 1980s. Furthermore, he points out that the experiences of most developed countries do not support this view that budget deficit increases the money supply and eventually inflationsoon or late. Based on these issues, it can be concluded that the inflationary effect of the government budget deficitsdepends on the financing method and the effect of budget deficit on total demand. If the government finances the budget deficit through the bond issue, this can be justified in such a way that the relationship between budget deficit and inflation depends on money creation. Thus, inflation is generally considered as a monetary phenomenon. In other words, increase of money supply is a factor that determines the increase of prices during the rate time. Moreover, it should be noted that there is a direct relationship between the government’s needs for borrowing and money creation, and this is subject to borrowing being done from the central bank and trading banks. Monetary authorities try to finance budget deficit by the sale of government bonds to non-bank private sectors such as households, companies and financial institutions. In this case, supply doesn’t change since the assets are transferred from the private to the government or public sector. Thus, regardless of the financing method, the effect of budget deficit on money supply mostly depends on the behavior of monetary authorities; that is to say, the collaboration and treatment of monetary authorities regarding the increase of budget deficit is crucial, in such a way that the financing method of budget deficit through increase of money supply or other methods is highly important. Therefore, based on this issue, it can be concluded that from a theoretical perspective, there is a close relationship between budget deficit and money growth, on the one hand, and inflation, on the other hand. A REVIEW OF EMPIRICAL STUDIES Researchers that have tested the relationship between inflation and budget deficit have not reached similar results. A broad range of results has been thus obtained which are sometimes in conflict. Studies (AghAvla& Khan) (1978) for developing countries show that the government expenses are adjusted far more rapidly than the government revenues in the face of inflation. As a result, inflation in these countries increases the government budget deficit. Heler (1980) shows that the results of the effect of inflation on budget deficit are not predictable in general. Based on his studies, almost in half of the countries under study the government expenses during inflation have increased far more rapidly than its revenues. Based on the results, the same way that the inflation rate is accelerated, the government expenses are adjusted far more rapidly than the taxes, but when the inflation rate is stabilized at a new

70

level, the differences between the adjustment rate of expenses and government income will be eliminated. To test the relationship between the government budget deficit and macroeconomic variables such as 1952-1987 U.S. inflation, Diver (1982) used the vector autoregressive (VAR) model. His results confirm this hypothesis that from among the predictable changes of the government debt to people the effect of wealth has not been observed. Thus, the government debts do not affect inflation. MacCalom(1984) used Competitive Equilibrium Model for the theoretical study of monetarists’ hypothesis. This hypothesis is in such a way that if deficit can be financed instead of increase of money supply in the method of bond issue, a budget deficit without inflation can be obtained. Aking& Miller(1985) modeled budget deficit, money supply growth and 1950-1980 U.S. inflation by means of the trivariate autoregressive process. The research results show that there is a bilateral causal relationship in 1950s and 1970s between the government budget deficit and inflation. Thus, it seems that budget deficit in 1950s and 1970s has been inflationary but has been non-inflationary in 1960s. Darrat (2000), by applying the error correction model, tested the effect of budget deficit on inflation in the time interval of 1957-1993 for Greece. The results of this study showed that there is a positive and significant relationship between the variable of budget deficit and inflation in Greece. He concluded that in addition to money growth, budget deficit has a more powerful and significant effect on the inflationary process of Greece compared with the other variables. JafariSamimi (1990), in a study of the bilateral relationship between the government budget deficit and inflation in Iran’s economy during the period of 1973-1990, concluded that there is a bilateral relationship between the government budget deficit and inflation. Increase of budget deficit through the increase in the monetary base and money supply increases the inflation rate and increase of inflation rate in turn leads to the increase of nominal budget deficit of the government. Taqipour&Nofarasti (1999), in analysis of different models of the reaction of monetary policy in Iran, concluded that during this period monetary authorities have willingly or unwillingly applied a coordinated monetary policy. Its reason is the financing of the government budget deficit by borrowing from the banking system and partly dependence of government revenues on oil, since the sudden change of oil revenues changes the monetary base without paying attention to the general level of the prices and under the above conditions the monetary policy follows the financial policy and the financial policy cannot controlthe price level in the economy. Farahbakhsh (2010) believes that most of Iran’s foreign exchange revenues result from oil revenues. Thus, the country’s economy is not capable of regulating the annual income and also method of its cost the way any dynamic

economy in the world deserves. Therefore, study of dimensions of the current costs of the government is of great importance. In his study, the factors affecting the government’s current expenses are analyzed during the years 1966-1999. PRESENTATION OF MODEL, FIT, INTERPRETATION OF RESULTS At first, stasis of the variables is analyzed using the Dickey-Fuller Test. In the next step, for estimation the vector autoregressive (VAR) model is used. In this model, the optimal level is determined, the model estimate results are analyzed for the three variables and the Granger Causality Test analyzes the causality between the variables, and for the results related to the vector autoregressive (VAR) model the impulse response functions and analysis of variance have been used. Also, through Johansen Test, the model convergence has been analyzed so that long-term equilibrium relations between the variables can be determined, and subsequently for the study of the short-term fluctuations in the model the vector error correction model (VECM) will be used. 1) Introduction of the Data Used in the Model In this study on the time series database of the central bank (1990-2011), the data has been used seasonally and in billionrials, and the variables used are as follows: BUDIt refers to deficit, overpayment of income and surplus, additional revenues from payments. Since 2002, in seasonal levels, deficit or surplus has been calculated by calculating the petty cashof certain treasury of provinces, salary and continuous benefits of the employees of provinces, cost credits and the petty cash of acquisition of capital assets. INFInflation rate for whose calculation the consumer price index has been used. LCPI=log(CPI) →

INF=LCPI-LCPI(-1)

(1) OILINThis income includes crude oil sales, sales of oil products and in some years sales of natural gas. Also, in the years 1980s and 1990s it includes the income from the foreign exchange sales difference in the market rate. MLiquidity which is the sum of money (including money and coins in the hands of people and non-government sector sight deposits by the banks and non-bank credit institutions) and quasi money (including non-government sector nonsight deposits by the banks and non-bank credit institutions) and has become a fixed price by means of an implicit deflator. 2) Study of Stasis of Variables by means of DickeyFuller Test

71

Tab. 1. Results of stasis test of the model variables Results of unit-root tests Model with intercept but not a Model with intercept and trend trend ADF ADF BUD in levels -1.8336 -4.4323 BUD in first differences -9.4424 -9.3819 OILIN in levels 2.1211 .57107 OILIN in first differences -8.9079 -9.4503 INF in levels -1.3863 -1.9918 INF in first differences -11.5570 -11.4754 LM in levels -.95227 -2.5290 LM in first differences -13.5638 -13.6992

The results show that none of the variables are at a static level but that all the variables have become static with firstorder differencing. In other words, all the variables are filled with first-orderI(1). It should be noted that in analyzing the stasis of variables the number of optimal lag has been determined based on Schwartz statistic. The obtained results have been presented in the table. Also, the level of confidence in this test is 5%. 3) Results of Multivariate Time Series Model In this section, by applying the vector autoregressive (VAR) model,the relationship between budget deficit, inflation and money supply is analyzed. The variable of oil revenues is used as the exogenous variable in this model. 3.1) Determination of the Optimal Lag Length of VAR Model Before estimating the vector autoregressive (VAR) model, the optimal lag length must be primarily determined. The results of determination of the optimal lag length have been reported in the following table. The criteria of adjusted LR and Schwartz (SC) have reported the optimal lag of 2; so the optimal lag length of 2 will be considered. Tab. 2. Determination of the optimal lag length Oorder 0 1 2

LL -547.1806 -507.6108 -475.4884

AIC -553.1806 -522.6108 -499.4884

SBC -560.1728 -540.0913 -527.4572

Adjusted test

LR

173.7596[.000] 127.9419[.001] 90.7475[.067]

3.2) Results of VAR Model Estimate for the Variable of Budget Deficit In the following table, the coefficients of the variables have been presented. At the 5% level, the coefficients of variables of budget deficit with the lag of 1 and 2, money supply with

the lag of 1, oil revenues and intercept are significant. In fact, these variables affect budget deficit in the present time. Tab.3. Results of VAR model estimate for the variable of budget deficit Regressor

Coefficient

Standard Error

T-Ratio[Prob]

DLINF(-1)

-28938.5

60097.2

-.48153[.632]

DLINF(-2)

29322.0

55921.9

.52434[.602]

DBUD(-1)

-.44149

.10120

-4.3628[.000]

DBUD(-2)

-.020987

.091661

-.22896[.820]

DLM(-1)

152889.1

55960.1

2.7321[.008]

DLM(-2)

8133.6

61355.4

.13256[.895]

DOILIN

.60744

.085411

7.1120[.000]

-11019.2

6172.4

-1.7852[.078]

C

3.3) Results of VAR Model Estimate for the Variable of Inflation Rate At the 5% level, inflation rate with the lags of 1 and 2 and also the money supply growth with the lag of 1 are significant. Tab. 4. Results of VAR model estimate for the variable of inflation rate Regressor

Coefficient

Standard Error

T-Ratio[Prob]

DLINF(-1)

-.34523

.096035

-3.5948[.001]

DLINF(-2)

-.48358

.089363

-5.4114[.000]

DBUD(-1)

.3274E-7

.1617E-6

.20246[.840]

DBUD(-2)

.3625E-7

.1465E-6

.24747[.805]

DLM(-1)

.34502

.089424

3.8583[.000]

DLM(-2)

-.12788

.098046

-1.3043[.196]

DOILIN

-.4584E-8

.1365E-6

-.033584[.973]

C

-.012651

.0098635

-1.2826[.204]

3.4) Results of VAR Model Estimate for the Variable of Money Supply Growth At the 5% level, inflation rate with the lag of 2, money supply growth with the lags of 1 and 2 and intercept are significant. Tab. 5. Results of VAR model estimate for the variable of money supply growth Regressor

Coefficient

Standard Error

T-Ratio[Prob]

DLINF(-1)

-.14892

.097855

-1.5218[.132]

DLINF(-2)

-.53806

.091056

-5.9092[.000]

DBUD(-1)

-.2225E-6

.1648E-6

-1.3505[.181]

DBUD(-2)

-.2150E-6

.1492E-6

-1.4402[.154]

DLM(-1)

-.21810

.091118

-2.3936[.019]

DLM(-2)

.38804

.099903

3.8841[.000]

DOILIN

-.2610E-6

.1391E-6

-1.8767[.064]

C

.050414

.010050

5.0162[.000]

72

3.5) Impulse Response Function for Budget Deficit (BUD) Impulse response function shows the shock resulting from the impulse as much as a standard deviation on the present and future values of endogenous variables. If the impulse notices mth variable, while the effect of the impulse notices the same variable, it is also transferred to other endogenous variables in the system through the dynamic system. In order to analyze the relationship between inflation, budget deficit and money supply we also refer to impulse response function. Diagram 1 shows the shock that we lapply to the budget deficit. At first, when the shock is applied the budget deficit curve drastically increases, and the function fluctuations become less and less over time until they become fixed.

Tab. 6. Results of ANOVA test for budget deficit Horizon

DDLINF

DBUD

DLM

0

.4946E-4

1.0000

.0070664

1

.0026176

.94915

.070722

2

.0048201

.92806

.092588

3

.018320

.89313

.11654

4

.025055

.86890

.13510

5

.026660

.85170

.15106

6

.027755

.84063

.16125

3.7) Results of Granger Causality Test for Inflation (INF), Budget Deficit (BUD) and Money Supply (M) In order to study the causality between budget deficit, inflation and money supply, Granger Causality Block Erogeneity Test has been used. Based on this test, H0 hypothesis according to which budget deficit and inflation are not the cause of money supply, is rejected. Also, H0 hypothesis according to which budget deficit and money supply do not Granger cause inflation, is also rejected. For the variable of budget deficit and money supply also H0 hypothesis according to which budget deficit and money supply do not Granger cause inflation, is also rejected. The results have been shown in the table below. Tab. 7. Results of Granger Causality Test

Fig. 1.Impulse response functionof budget deficit

The second diagram shows the reaction of the two variables of inflation rate and money supply to the impulse applied to the budget deficit. The diagram shows that the money supply and inflation rate also start fluctuating under the influence of this shock (however, it should be noted that the shock applied to the budget deficit of the diagram affects the money supply more than inflation rate, until its fluctuations decrease over time and lean towards zero).

Null Hypothesis

CHSQ( 4)

BUD and INF do not Granger Cause M

35.8714[.000]

BUD and M do not Granger Cause INF

27.2890[.000]

3.8)Long-term Cointegration and Error Correction Model To conduct the Johansen Test, λMax and λTrace tests are used and by comparing the test statistics and the critical values at the 5% level of confidence, the results are reported separately in the following tables. In both tests the H0 hypothesis according to which there are two convergence vectors, is rejected and there are practically 3 convergence vectors. In fact, in this state where n=r, our vector process is reliable and all the variables have the convergence level of zero and the issue of cointegration is no longer the case in this state. Tab. 8. Results of λTrace test Null

Alternative

Statistic

95%Critical Value

r=0

r>= 1

195.4212

34.8700

r= 2

81.6807

20.1800

r