THE EFFECTS OF BUDGET DEFICIT ON NATIONAL SAVING IN MALAYSIA 1 Fatimah Wati Ibrahim Asmawi Hashim ABSTRACT This paper analyses the effect of government budget deficits on national saving in Malaysia utilizing annual time series data from 1970-2000. A model that includes budget deficits, money supply, real interest rate and real exchange rate to explain national saving is developed. After examining the time series properties of the data, an error correction model is estimated. With regards to the effect of budget deficits on national saving in Malaysia, the result indicates that there exist a negative relationship between budget deficit and national saving. Intuitively, this implies that decreasing the government budget deficits will tend to increase the national savings. Our finding using Malaysian data is parallel with Pradhan and Upadhyaya (2001) in their study for the USA case. The finding also gives empirical evidence for Malaysia case that the relationship between the variables budget defic it and national saving in Malaysia are consistence with the theoretical arguments against Ricardian equivalent proposition. The other variables, namely, the money supply and real interest rate have positively contributes to an increase in national saving. However, the real exchange rate has a negative relationship with the national saving. The findings presented in this study show that a range of variables, rather than a single policy variable, explains the differential saving performance. Specifically, prudent fiscal policy may be the core policy instruments that boosted saving rates in Malaysia, but policies that improved the framework in which saving decisions are made, including macroeconomics stability and financial market development, are clearly important as well. Therefore, by understanding the effect of budget deficits on national saving, policy maker is able to analyze the behavior of these variables in the past for the purpose of improving the performance of national saving in the future. INTRODUCTION Public debt and budget deficits have long been a debatable issue over their consequences on the economy. The consequences of budget deficit would include impact on macroeconomic variables as well as its effect on long-term economic growth. From the neoclassical model of the debt perspective, its stresses that when the government initiates a project, whether financed by taxes or borrowing, resources are removed from the private sector. One usually assumes that when tax finance is used, most of the resources removed come at the expense of consumption. On the other hand, when the government borrows, it competes for funds with individuals and firms who want the money for their own investment projects. Hence, it is generally assumed that debt has most of its effect on private investment. To the extent that these assumptions are correct, debt finance leaves the future generation with smaller capital stock, ceteris paribus. Its members therefore are less productive and have smaller real incomes than otherwise would have been the case. Barro (1974), on the other hand, provides a major departure from the view that budget deficits impose burdens on future citizens. His view is based on the Ricardian equivalence proposition that theorizes that public debt and taxation exert equivalent effects on the economy. Barro argues that savings rise in anticipation of higher future tax burdens implied by public debt. 1 Paper to be presented at the 1 st International Conference of the Asian Academy of Applied Business Narrowing the Competitive Gap of Emerging Markets in the Global Economy, jointly organized by the Asian Academy of Applied Business and the School of Business and Economics, Universiti Malaysia Sabah, 10-12 July 2003, Kota Kinabalu, Sabah. 238
The flow of Malaysian budget for the past three decades have shown that most of the years Malaysia stayed in deficit condition. Malaysia had the surplus budget condition only for the fiscal years 1988, 1990, 1991, 1993, 1995, 1996 and 1997. With regard to saving, over the last three decades Malaysia saved an average of 24 percent of gross national product (GNP) a year, a level amongst the highest in the world. In 1986 to 1990, the ratio was high at 31 percent. The level of gross national savings in term of GNP in Malaysia has been fluctuating from 30 percent in 1982 to 40.4 percent in 1987, subsequently declining to 30 percent in 1990, with a further drop to 28 percent in 1991. However, the level of savings increased to 32.3 percent in 1992. The purpose of this paper is to examine the effects of budget deficits on national saving in Malaysia. In general, the main objective is to provide some probable explanations on the determinants of national saving function in Malaysia. Then, the study examines the impact of budget deficits on the economy to see whether the empirical evidence for Malaysia case consistence with the theoretical arguments in favour or against Ricardian equivalent proposition. The rest of the paper is organized as follows. Section 2 reviews the existing literature on budget deficits and national saving. Section 3 presents the model used in explaining the relationship between budget deficits and national saving in Malaysia; and also describes the data used. Section 4 discusses the empirical results and statistical analysis. Section 5 concludes with a summary of the main findings. REVIEW OF RELATED LITERATURE Theoretically, economists do not encourage large budget deficits because of their detrimental effect to long-term growth and stability of an economy. A large budget deficit leads to lower national savings, lower investment, and a trade deficit 2. In the long run, large deficits could lead a smaller capital accumulation and larger foreign debt. Many researchers found that budget deficits give a negative impact on national savings. For example, a study by Swanstrom (1989), shows that budget deficits have a negative relationship with national savings. He suggests that the most direct way to increase national saving would be to reduce the federal budget deficits. Evans (1986) also looked at these two variables relationship in his study. He argues that higher budget deficits tend to raise domestic consumption. Higher consumption will tend to reduce the private saving and lastly will disturb the national saving. Higher consumption falls on both domestic and imported goods, which in turn lead to higher domestic interest rates with respect to their counterpart abroad. In a recent paper by Pradhan and Upadhyaya (2001), the empirical analysis suggests that an increase in government budget deficits tend to reduce national saving. They utilize annual time series USA data from 1967 to 1996. In contrast, a study by Robert Eisner (1994) using USA data for the 1972-1991 period, show that real budget deficits have increased rather than decreased national saving. He argues that this effect on national saving is consistent with the hypothesis that federal deficits can increase national saving by stimulating more employment, consumption and investment through a Keynesian expansion 3. In his model he examines how national saving is related to the federal budget deficits, changes in the money supply and changes in the real exchange rate. Barro (1974) in his study argues that federal deficits are irrelevant to the level of national saving because increase in private saving will neutralize federal budget deficits. Under the Ricardian 2 Based on Economics Review published by the Economics Division, Public Bank Bhd (Malaysia), Oct 2002. 3 Keyne proposed solution to high unemployment was to have the government increase its purchases of goods and services, thus raising the demand for output. He argues that this policy would reduce unemployment because to meet higher demand for their product, business would have to employ more workers. Keyne suggests that the newly hired workers would have more income spend creating another source of demand for output that would raise employment further. 239
equivalence view, deficit policy is a matter of indifference, since an increase in government debt leads to a future increase in taxes and thus it is not an addition to private wealth. This fact has no effect on consumption, interest rates, and aggregates demand. Cebula, Hung and Menage (1995), in their study on Ricardian equivalent, budget deficits and saving in US found that structural deficits 4 elicit increased saving but cyclical deficit do not. Thus, findings indicate support for a partially Ricardian equivalent world: saving only partially offsets budget deficits. Gramlich (1989) and Mixon and Wilkonson (1999), note that in the early 1980 s, there was a huge increase in federal deficits in USA. According to Barro s hypothesis, private saving should increase as well while the fed deficits increase. But, in their study private saving relative to net national product actually fell. Of course, this finding is only suggestive because other factors beside the deficits also affect national saving. However, the general finding by Berheim (1989) found that even after taking other variables into account, saving does not increase enough to offset increased deficits. METHODOLOGY The Method In analyzing the effect of government budget deficits on national saving, we follow the model proposed by Pradhan and Upadhyaya (2001). Factors that are likely to influence national saving in Malaysia includes the size of federal government budget deficit, the level of money supply, the real exchange rate and real interest rate. Thus, the national saving is a function of these variables, and mathematically can be written as follows: NS = ƒ (DEF, MS, RER, RIR) (1) The following represents the regression model: NS = α + β1def + β2ms + β3rer + β4 RIR + e (2) Where NS is the national saving, DEF is the federal budget deficit as a percentage of GDP, MS is the money supply as a percentage of GDP, RER is the real exchange rate, RIR is the real interest rate and e is the random error term. Before carrying out the estimation of Equation 2, to avoid spurious regression, a Dickey Fuller test is conducted on the data series to ensure the stationary of the data. This involves estimating the following regression and carrying out unit root tests. n?x t = a +? t + ß X t-1 + S? i?x t-i X + e t ( 3) i=1 Where X is the variable under consideration,? is the first difference operator, t is a time trend and e is a stationary random error term. After establishing the stationary of the data, we used Johansen (1986) and Johansen and Juselius (1990) approaches to examine the test for a long run equilibrium relationship among the variables. This involves the test of co-integrating vectors. Consider a p dimensional vector autoregression, k Xt = S?txt-i + c + et (4.1) i=1 4 The structural deficits has been used for measuring the impact of deficits on their important macroeconomics variables such as interest rates (Barth, iden and Russek, 1985) and national saving (Eisner, 1994) 240
Which can be written as: k?xt = S? i? X t-k?x t-k+ c + et (4.2) i=1 Where? i = -I +? 1 +? 2 +.+? t (4.3) i = 1,2, k-1 and? = I-? 1 -? 2 -.-? k (4.4) Where p is equal to the number of variables under consideration. The matrix? captures the long run relationship between p variables, can be decomposed into two matrices, A and B, such that? = AB. A is the vector error correction parameter and B as cointegrating vectors. This procedure is used to test the existence of a long run relationship between NS, DEF, MS, RER and RIR in Equation 2. After establishing the long run relationship between national saving and the explanatory variables we turn to examine the short run dynamics of the relationship. We utilized the error correction model by Engle and Ganger (1987). It involves estimating the model in the first difference form and adding an error correction term as another explanatory variable. The error correction model is implies as follows:? NS = bo + b1? DEF + b2? MS + b3? RER + b4? RIR + b5? EC + v (5) Where EC is the error correction component and v is the random error term. The error correction term is the lagged values of the error term that has been derived from the estimation of Equation 2. Description of Data NS is the conventional measure of national saving, defined as the difference between national income and the sum of private and government consumption. This measure is not affected by different assumptions about the financing decisions of firms. The sources of national saving is composed of three components, that is, the government or public savings that represent the difference between current income and current expenditure of various levels of government operations. Second, is the corporate or business savings that is made up of retained profits and the depreciation allowances of the profit-seeking private enterprises. And the third is the household or personal savings that is determined by using disposable income approaches, that is, the difference between disposable personal income of household and private personal consumption expenditure. DEF is the federal budget deficit as a percentage of GDP. Deficit occurs when the government s total expenditure exceed total revenue. MS is the money supply as a percentage of GDP. In this study the broader definition of money supply is used, that is, M2 comprises of currency, demand and time deposits. RER is the real exchange rate, an index, defined as the multilateral trade weighted value of the Malaysia ringgit adjusted by the changes in the domestic consumer price and the foreign price index. It is the value of current exchange rate divided by GDP. RIR is the real interest rate, defined as the nominal interest rate minus the inflation rate. In this study we used the annually compounded interest rate for saving deposit. Data on national savings and the four explanatory variables (budget deficits, money supply, real exchange rate and real interest rate) were collected for the period 1970-2000. The sources of data are available from the Malaysian Economic Report, Bank Negara Report and the Malaysia Key Indicator. 241
EMPIRICAL RESULTS AND DISCUSSION Test of Stationary The unit root test of the series is crucial for cointegration and causality analysis. It is examines by the standard augmented Dickey-Fuller (ADF). All series have been log-transformed before the analysis. The test results are reported in Table 1. TABLE 1 Unit Root Test by Augmented Dickey-Fuller Test Variable Level First Difference τ lags τ µ lags NS -3.573 2-3.696* 2 DEF -3.685 2-4.417* 2 MS -3.857 2-4.338* 2 RER -3.685 2-3.612* 2 RIR -2.971 2-4.394* 2 Note: Asterisk (*) denotes the significant at 1 percentage level. The results indicate that all the series are stationary in the first difference at 1 percentage level. It also shows the augmented Dickey-Fuller test for all series involved in the analysis in logarithmic form in levels and first-differenced. The results indicate that non-stationarity cannot be rejected for the levels at the 5 percent significant level based on the ADF test. When the series are differenced, nonstationarity can be rejected for all series. The ADF statistics suggest that all five series are integrated of order one, whereas the first-differenced are integrated of order zero. Therefore, all series is best characterized as difference-stationary process instead of trend-stationary process. The Error-Correction Model After establishing the long run relationship between national saving and the explanatory variables we turn to examine the short run dynamics of the relationship. For this purpose as mentioned earlier, following Engle and Granger (1987), an error correction model was estimated. Results of the errorcorrection model estimated are reported in Table 2. Variable Constant DEF MS RIR RER ECT (-1) TABLE 2 Result of Error-Correction Model Coefficient -0.001 (-0.05) -0.028 (-2.064)** 0.801 (2.546)* 0.184 (1.822)** -0.349 (-2.036)** -0.016 (-1.907)** 0.51 1.93 Adj. R 2 DW Note: refers to first difference of variables. Asterisk (*) denotes statistically significant at 5 percent level and (**) at 10 percent level. 242
As shown in the table, the error-correction term is strongly significant and above 10 percent value in McKinnon s (1991) table, thus providing support for the hypothesis of co-integration. With regards to the effect of budget deficits on national saving in Malaysia, the result indicates that there exist a negative relationship between budget deficit and national saving. Intuitively, this implies that decreasing the government budget deficits will tend to increase the national savings. Our finding using Malaysian data is parallel with Pradhan and Upadhyaya (2001) in their study for the USA case. The result also gives empirical evidence for Malaysia case that the relationship between the variables budget deficit and national saving in Malaysia are consistence with the theoretical ar guments against Ricardian equivalent proposition. With regards to the other variables, namely, the money supply variable, the result shows a positive relationship with the national saving. It is significantly different from zero at the 5 percent level. This implies that the role of financial development has a positive effect on national savings in Malaysia. This relation also holds for the variable, real interest rate where it shows a positive relationship with the national savings. But, it is significantly different from zero at the 10 percent level. This means that if the interest rate of saving increased, it will raise the national saving. The finding can also provide some evidence to support the classical theory that personal saving responds positively to change in the real interest rate (Bonskin 1978). If this relation hold true, the coefficient of the real interest rate would carry a positive sign. On the other hand, more sophisticated examinations of the data suggests that the real interest rate has little effect on personal saving (hall, 1988; Campbell and Mankiw, 1989). The variable, real exchange rate shows a negative relationship with the national savings. Generally, the intuition is, if our exchange rate increased, it will tend to reduce the private saving and public saving because individuals and government tend to increase their consumption and less saving. Pradhan and Upadhyaya (2001) argue that a lower real exchange rate would be expected to increase national savings. Since net exports are negative function of the real exchange rate a lower real exchange rate raises the demand for net exports, which in turn raises national savings. SUMMARY AND CONCLUSION This study has analyzed the relationship between dependent variable, national saving and the explanatory variables: government budget deficits, money supply, the real interest rate and the real exchange rate in Malaysia utilizing annual time series data from 1970-2000. After ensuring the stationary of the data, a cointegration test is conducted to establish the long run relationship between national saving and the other variables. Generally, the data series were found stationary only in the first difference level and the hypothesis of no cointegration is rejected. Therefore, an error-correction model was estimated. The error-correction term carries a negative sign for the government budget deficits. Our finding using Malaysian data is parallel with Pradhan and Upadhyaya (2001) in their study for the USA case. The result also gives empirical evidence for Malaysia case that the relationship between the variables budget deficit and national saving in Malaysia are consistence with the theoretical arguments against Ricardian equivalent proposition. The results presented in this study als o show that a range of variables, rather than a single policy variable, explains the differential saving performance. Specifically, prudent fiscal policy may be the core policy instruments that boosted saving rates in Malaysia, but policies that improved the framework in which saving decisions are made, including macroeconomics stability and financial market development, are clearly important as well. Therefore, by understanding the effect of budget deficits on national saving, policy maker is able to analyze the behavior of these variables in the past for the purpose of improving the performance of national saving in the future. 243
REFERENCES Apergis N. (1998). Budget Deficits and Exchange Rate: Further Evidence from Cointegration and Causality Tests. Journal of Economic Studies, vol.25, Number 3, pp.161-178. Blanchard, O.J. (1985). Debts, Deficits and Finite Horizons, Journal of Political Economy 93: pp 1045-1076. Cebula R. J., Chao-Shun Hung and Neela D. Manage (1996). Ricardian Equivalence, Budget Deficits and Saving in the US, 1955: 1-1991:4. Applied Economics 1996,3,525-528. Edward M. Gramlich. (1989). Budget Deficit and National Saving: Are Politicians Exogeneous? Journal of Economics Perspectives -volume 3, pp 23-35 Eisner, Robert. (1984). Deficits, Saving, and Economic Policy, American Economist, vol.38 Engle, R.F and B.S. Yoo. (1987). Forecasting and Testing in Cointegrated Systems. Journal of Econometrics 35: 143-159 Hoover, Dennis K. (1992). Low US Saving: Increase it by Rreducing the Federal Deficits. Business Economics, Jan 1992, vol.27 Kraay A. (1997). Household Saving in China, The World Bank Masson P. R., Tamim Bayoumi and Hossein Samiei (1998). International Evidence on the Determiants of Private Saving, The World Bank Economic Review Vol. 12. Norashikin Abd. Hamid and Prof Dr. M.G. Kanbur (1993). Saving In Malaysia. School of Economics and Public Administration, Universiti Utara Malaysia. Pradhan G. and Kamal P. Upadhyaya. (2001). The Impact of Budget Deficits on National Saving in the USA. Applied Economics, 33,1745-1750. Robert A. Johnson, Commentary: What Do Budget Deficits Do? Vickrey W., The Need for a Budget Deficit. http://wfhhummel.cnchost.com /vickrey.html William D. Nordhaus. (1996). Budget Deficits and National Saving. March-April/Challenge 45-49 244