SAVING-INVESTMENT CONNECTION: EVIDENCE FROM THE ASEAN COUNTRIES.

Anoruo, Emmanuel

Abstract

The relationship between saving and investment has been sharply debated in the literature following the pioneering work of Feldstein and Horioka (1980). This paper extends this debate to the ASEAN countries by using cointegration procedure in time-series analysis. Specifically, three analyses are conducted. First, saving and investment are tested to determine the order of integration using both the Dickey-Fuller (DF) and augmented Dickey-Fuller (ADF) approaches. Second, the long-run equilibrium relationship between saving and investment is explored by utilizing the cointegration tests proposed by Johansen and Juselius (1990). Third, Granger-causality tests based on vector error-correction models (VECM) are undertaken to ascertain the direction of causality between the two series. The results indicate that saving and investment are integrated of order one [I(1)]. Based on the cointegration results, saving and investment are found to share long-run equilibrium association. The Granger-causality tests reveal tha t investment causes saving in the cases of Indonesia and Singapore. For the Philippines, causality runs from saving to investment. As for Malaysia and Thailand, the results suggest bi-directional causality between saving and investment.

1. INTRODUCTION

The theoretical finding by Feldstein and Horioka (1980) that saving and investment are highly correlated has generated intense debate in the economics literature. Feldstein and Horioka interpreted the high correlation between saving and investment as evidence of imperfect capital mobility across national boundaries. This finding is inconsistent with the conventional wisdom, which stipulates that in the absence of financial controls, capital should flow between countries in search of the highest rate of return. However, if capital movements were restricted, domestic saving and investment will be highly correlated. The existence of strong association between saving and investment has been validated by Feldstein (1983), Summers (1988), Baxter and Crucini (1993), Dooley, Frankel, and Mathieson (1987), Caprio and Howard (1984), Feldstein and Bacchetta (1989), Miller (1988), and Tesar (1991). In general, these studies contend that capital is not internationally mobile. Under this assumption, increases in domestic saving, all things being equal, will stimulate domestic investment since capital is not flowing across country borders in search of the highest return.

However a number of researchers including Murphy (1984), Obstfeld (1986), Finn (1990), Stockman and Tesar (1991), and Barkoulas, Filizetkin, and Murphy (1996) have challenged the existence of high correlation between saving and investment. These authors surmise that capital is internationally mobile. Under this hypothesis, foreign capital flows to countries with higher real interest rates. Perfect capital mobility, has important policy implications especially for small open economies. In the event that capital were internationally perfectly mobile, increases in domestic saving do not necessarily translate into higher domestic investment because foreign savings generally flow to countries with higher real interest rates.

Although the previous studies in the literature have furnished insights with regard to the relationship between saving and investment, the conceptual and methodological approaches utilized in these studies present a number of concerns. First, most of the studies used single equation ordinary least squares (OLS) regression method to examine the relationship between saving and investment. These studies are likely to suffer from simultaneous equation bias leading to fallacious conclusions. Second, studies that employed OLS regression analysis did so without first examining the time series properties (unit roots) of saving and investment. Nelson and Plosser (1982) have shown that most macroeconomic time series data are nonstationary in their levels but stationary when differenced. Third, a number of studies used cross-section data, which makes it difficult, if not impossible, to apply their findings to any particular country. Fourth and finally, most of the studies in the literature concentrated on the relations hip between saving and investment in the developed countries [mainly for the members of the Organization for Economic Cooperation and Development (OECD)] with little or no attention devoted to countries with nascent economies.

This paper attempts to overcome these weaknesses by using recent advances in cointegration techniques and country-specific time series data to examine the long-run equilibrium relationship between saving and investment for the ASEAN countries specifically, Indonesia, Malaysia, Philippines, Singapore, and Thailand. In particular, we conduct unit root tests using both the Dickey-Fuller (DF) and augmented Dickey-Fuller (ADF) to determine the order of integration, since we must only include variables with the same order of integration in the cointegrating equation. In addition, we undertake the cointegration tests utilizing the maximum-likelihood procedure suggested by Johansen and Juselius (1990) and Johansen (1991) to ascertain the long-run relationship between saving and investment. [1] Finally, the Granger-causality tests based on the vector error correction models (VECM) are conducted to determine the direction of causality between saving and investment series.

Few researchers including Barkoulas, Filizetkin, and Murphy (1996), Jansen and Schulze (1996), Taylor (1996), and Miller (1988) have examined the relationship between saving and investment using cointegration techniques. Again, these authors focused mainly on the OECD countries. As a contribution to the literature, the current study employs cointegration-based techniques to extend the saving-investment debate to the ASEAN countries. Such an analysis is worthwhile given that the economic experiences of the ASEAN countries are arguably very different from those of OECD countries. Notably, the ASEAN economies are often plagued with inefficient public enterprises, deficient infrastructure, tight trade controls, restrictive regulations in the financial sector, poor corporate governance and political uncertainty. Under these conditions, the macroeconomic dynamics that govern the relationship between savings and investment in developing countries could be very different from those that are witnessed in the OECD gro up of countries.

The selection of the ASEAN countries also deserves justification from another perspective. It has been documented that East Asian countries including, Indonesia, Malaysia, the Philippines, Singapore, and Thailand have experienced high saving and investment rates. Table 1 indicates that between 1960 and 1996, saving as a percentage of gross domestic product (GDP) increased in all of the countries under consideration with the exception of the Philippines. Similarly, Table 2 reveals that gross domestic investment as a ratio of GDP increased in all of the sample countries. A number of explanations have been advanced in the literature in relation to this phenomenon. Adams and Prazmowski (1996) explain the high saving and investment rates for the region in the context of virtuous cycle. According to the authors, increases in saving and investment engender economic growth, which in turn, leads to increases in domestic saving and investment. World Bank (1993) reports that most of the East Asian countries restricted the outflow of capital during the periods of their high economic growth in order to encourage domestic investment. However, it is important to point out that most of the financial markets in the region were liberalized beginning in the early 1970s to late 1980s. In light of these characteristics, the sample countries provide estimable avenues to explore the long-run equilibrium relationship between saving and investment.

Following the present introduction to the paper, section 2 examines the data and methodology of the study. Section 3 discusses the empirical results. Finally, section 4 summarizes the findings of the study.

2. DATA AND METHODOLOGY

The study utilizes annual data on gross domestic saving and investment for Indonesia, Malaysia, the Philippines, Singapore, and Thailand. The data cover the period 1960 through 1996. All the data are collected from the World Bank, World Development Indicators 1998.

In conducting cointegration tests, the time series are required to be nonstationary in their levels. Moreover, it is important that all time series in the cointegrating equation have the same order of integration. Consequently, the study first ascertains the time series properties of domestic saving and investment by employing both the DF and ADF tests for stationarity. The equation estimated for the ADF test takes the form: [2]

[delta][X.sub.t] = [[alpha].sub.0] + [[beta].sub.1][X.sub.t-1] + [delta]t + [[[sigma].sup.m].sub.i=1][[theta].sub.i][delta][X.sub.t-1] + [[epsilon].sub.t] (1)

where, [delta] is the first-difference operator, t is the time trend, and [epsilon] is the stationary random error, and m is the maximum lag length.

To determine whether saving and investment are cointegrated, the Johansen cointegration procedure is utilized [see Johansen (1991) and Johansen and Juselius (1990)]. The procedure involves the estimation of VECM in order to obtain the likelihoodratios (LR). The VECM used for cointegration is as follows:

[delta][Y.sub.t] = [[theta].sub.0] + [[[sigma].sup.k=1].sub.i=1][delta][Y.sub.t-i] + [alpha][beta]'[Y.sub.t-k] + [[epsilon].sub.t] (2)

where [delta] is the difference operator, [delta][Y.sub.t] is ([delta][GDS.sub.t], [delta][GDI.sub.t]), [[theta].sub.0] represents the intercept, and [epsilon] represents the vector of white noise process. The matrix [beta] consists of r (r [less than or equal to] n -1) cointegrating vectors. On the other hand, the matrix [alpha] contains the error-correction parameters. In equation (2), the null hypothesis is that the matrix (II = [alpha][beta]') has a reduced rank of r [less than or equal to] n - 1. The alternative hypothesis, on the other hand, is that the matrix (II = [alpha][beta]') has full rank. [3]

Johansen procedure of cointegration provides two statistics. These include the value of the LR test based on the maximum eigenvalue of the stochastic matrix and the value of the LR test based on the trace of the stochastic matrix. Under the trace test, the null hypothesis is that II has zero rank (r = 0) and the alternate hypothesis is that r [less than or equal to] 1. However, the null hypothesis for the maximum eigenvalue test is that r = 1 while the alternate hypothesis is that r = 2. The existence of at least one cointegrating vector in the system indicates the presence of causality between saving and investment.

The causal relationship between gross saving and investment is explored with Granger-causality test based on VECM. This procedure is particularly attractive over the standard VAR because it permits temporary causality to emanate from (1) the sum of the lagged coefficients of the explanatory differenced variables and (2) the coefficient of the lagged error-correction term. In addition, the VECM allows causality to emerge even if the lagged differences of the explanatory variables are not jointly significant [see Granger (1988), Miller and Russek (1990), Miller (1991), and Garcia and Zapata (1991)]. It must be pointed out that the standard Granger-causality test omits the additional channel of influence ([z.sub.t-1]) In this study, the causality tests are based on the following VECM:

[delta][X.sub.t] = [alpha][z.sub.t-1] + [[[sigma].sup.a].sub.i=1][[beta].sub.i][delta][X.sub.t-i] + [[[sigma].sup.b].sub.j=1][[phi].sub.j][delta][Y.sub.t-1] + [[micro].sub.t] (3)

[delta][Y.sub.t] = [varphi][z.sub.t-1] + [[[sigma].sup.c].sub.i=1][[theta].sub.i][delta][Y.sub.t-1] + [[[sigma].sup.d].sub.j=1][[lambda].sub.j][delta][X.sub.t-1] + [[epsilon].sub.t] (4)

where, [z.sub.t-1] represents the error correction term lagged by one period, [4] X is the gross domestic saving (GDS) as a ratio of GDP, Y stands for gross domestic investment (GDI) scaled by GDP, a, b, c, and d represent the optimal lag lengths obtained from the Akaike Information Criterion (AIC). In equation (3) the rejection of the null hypothesis that gross domestic investment does not Granger-cause saving requires that (i) the [phi]j's conjointly be statistically significant and/or (ii) the error correction term ([z.sub.t-1]) be statistically significant. Similarly, in equation (4) the null hypothesis that gross domestic saving does not cause investment is rejected provided that the [lambda]j's are jointly statistically significant and/or the error-correction term ([z.sub.t-1]) is significant.

3. EMPIRICAL RESULTS

The results of both the Dickey-Fuller (DF) and the augmented Dickey-Fuller (ADF) unit root tests are presented in Table 3. The null hypothesis of nonstationarity of saving and investment is tested against the alternative hypothesis of stationarity. The results indicate that both saving and investment are not stationary in their levels. However, after first differencing, the null hypothesis of no unit root is rejected in all of the cases. In all, the results indicate one order of integration [I(1)] for saving and investment series. The nonstationarity of the time series in their levels calls for the application of cointegration procedure to avoid the problem of spurious regression.

Having determined the order of integration, we next apply the Johansen procedure to ascertain whether investment and saving are cointegrated for each of the countries under consideration. The results of the Johansen cointegration tests are presented in Table 4. The null hypothesis of no cointegration between saving and investment (i.e. r = 0) is rejected by both the trace and maximal eigenvalue ([[lambda].sub.max]) tests at the 5 percent significance level in all of the cases. Nevertheless, the null hypothesis that r [less than or equal to] 1 could not be rejected for all of the sample countries. The fact that saving and investment are found to be cointegrated suggests that capital is immobile internationally relative to the sample countries. This finding that investment and saving are cointegrated, although suggestive, is consistent with Feldstein and Horioka (1980), Caprio and Howard (1984), Feldstein and Bacchetta (1989), Baxter and Crucini (1993), and Dooley, Frankel, and Mathieson (1987).

The results for the error-correction based Granger-causality tests are presented in Table 5. These tests are conducted with residuals from the cointegration equations. The results indicate that for Indonesia and Singapore, causality runs from investment to saving. The null hypothesis that investment does not Granger-cause saving is rejected because the error-correction terms are statistically significant for these countries. This finding implicates the error-correction terms as the only channel of influence since the lagged differences of saving are not conjointly significant. In the case of the Philippines, the null hypothesis that saving does not cause investment is rejected since the error-correction term is significant. Again, the error-correction term emerges as the sole channel of influence. For Malaysia and Thailand, the results in Table 5 show that causality runs from saving to investment and vice versa. Specifically, a bi-directional causality is detected between the two series. In the case of Malays ia, causality emerges from both the significant error-correction terms and the lagged differences of domestic investment. Again, for Thailand, the two channels of causality are present. We observe in Table 5 that the error-correction term is statistically significant relative to equation (2). Similarly, the lagged differences of gross domestic saving and investment are statistically significant.

[delta][GDS.sub.t] = [alpha][z.sub.t-1] + [[[sigma].sup.a].sub.i=1][[beta].sub.i][delta][GDS.sub.t-i] + [[[sigma].sup.b].sub.j=1][[phi].sub.j][delta][GDI.sub.t-1] + [[micro].sub.t]

[delta][GDI.sub.t] = [xi][z.sub.t-1] + [[[sigma].sup.c].sub.i=1][V.sub.i][delta][GDI.sub.t-1] + [[[sigma].sup.d].sub.j=1][[lambda].sub.j][delta][GDS.sub.t-1] + [[epsilon].sub.t]

where [z.sub.t-1] represents the error-correction term lagged by one period, GDS is the ratio of gross domestic savings to GDP, GDI is the ratio of gross domestic investment to GDP, a, b, c, and d represent the optimal lag lengths determined by Akaike Information Criterion. The level of significance for the error-correction terms ([z.sub.t-1]) is determined by the standard t-statistics.

4. CONCLUSIONS

This paper has used cointegration procedure to examine the long-run equilibrium relationship between saving and investment in the ASEAN countries namely--Indonesia, Malaysia, the Philippines, and Thailand. Consistent with a number of earlier studies in the literature, saving and investment are found to be cointegrated in the long run for all of the countries under consideration. The conclusion that saving and investment are cointegrated implies that there is a strong internal association between the two series. Simply, the results indicate that long-term capital is not mobile internationally for all of the countries under study. Furthermore, the error-correction based Granger-causality tests indicate that causality runs from investment to saving for Indonesia and Singapore. For the Philippines, a unidirectional causality from saving to investment was found. Finally, in the cases of Malaysia and Thailand, causality runs in both directions.

Although this paper has established the existence of long-run relationship between saving and investment for the countries under study, there is still room for further research. Interest rate could be included as a mediating variable in the examination of the long-run relationship between saving and investment. The inclusion of interest rate will alleviate the possibility of distortion of causality inferences resulting from the omission of a relevant variable.

(*.) Department of Management Science and Economics, Coppin State College, Baltimore, Maryland 21216-3698. The author is grateful to Sanjay Ramchander, Habtu Braha, and an anonymous referee for their invaluable comments and suggestions that helped to improve the quality of this paper. The author assumes full responsibility for all errors in this study.

Notes

(1.) Furthermore, Granger and Newbold (1974) and Engle and Granger (1987) have demonstrated that the use of cointegration technique avoids spurious regression results.

(2.) For standard DF unit root test, we exclude the summation from the right hand side of equation (1).

(3.) In the present study, full rank would imply that r is equal to 2, since saving and investment are the only two time series in the model.

(4.) The cointegration equation used in this study is as follows: [y.sub.t] = [alpha][x.sub.t] + [[epsilon].sub.t], where, [y.sub.t] and [x.sub.t] each represents GDS and GDI.

References

Adams, F. G. and P. Prazmowski (1996) "Why Are Saving Rates in East Asia So High? Reviving the Life Cycle Hypothesis." Economics Research Unit: University of Pennsylvania: Philadelphia.

Barkoulas, John, Alpay Filizetkin and Robert Murphy (1996), "Time Series Evidence On the Saving-Investment Relationship," Applied Economics Letters, 3, 77-80.

Baxter, Marianne and Mario Crucini (1993) "Explaining Saving-Investment Correlations," American Economic Review, 83, 416-436.

Bayoumi, Tamin (1990), "Saving-Investment Corrections: Immobile Capital, Government Policy, or Endogenous Behavior?" IMF Staff Papers, 27, 360-387.

Caprio, Gerald, Jr. and David H. Howard (1984) "Domestic Saving, Current Accounts, and International Capital Mobility," International Finance Discussion Paper 244 (Washington: Board of Governors of the Federal Reserve System).

Dickey, D. A., and W. A. Fuller (1979) "Distribution of Estimates of Autoregressive Time Series With Unit Root," Journal of the American Statistical Association, 427-431.

Dooley, Michael, Jeffrey Frankel, and Donald Mathieson (1987), "International Capital Mobility in Developing Countries versus Industrialized Countries: What Do Savings-Investment Correlations Tell Us?" IMF Staff Papers, 34, 503-529.

Engle, R. F. and Granger, C. W. J. (1987), "Cointegration and Error Correction: Estimation and Testing," Econometrica, 55, 251-276.

Feldstein, Martin and Charles Horioka (1980), "Domestic Saving and International Capital Flows," Economic Journal, 90, 314-329. Feldstein, Martin (1983), "Domestic Saving and International Capital Rows," European Economic Review, 21, 129-151.

----- and Baccheta (1989) "National Saving and International Investment," NBER Working Paper 3164 (Cambridge, Massachusetts: National Bureau of Economic Research, November).

Finn, Mary G. (1990), "On Saving and Investment Dynamics in a Small Open Economy," Journal of International Economics, 29, 1-21.

Garcia, Philip and Hector Zapata (1991), "Co-integration, Error-correction and the Fisher effect: A Clarification" Applied Economics, 23, 1367-1368.

Granger, Clive W. J. (1969) "Investigating Causal Relations by Econometric Models and Cross-Spectral Methods" Econometrica, 37, 424-438.

----- (1983) "Co-Integrated Variables and Error-Correcting Models." Working Paper 83-13. University of California, San Diego.

----- (1988) "Some Recent Developments in a Concept of Causality" Journal of Econometrics, 39, 199-211.

----- and P. Newbold (1974) "Spurious Regression in Econometrics," Journal of Econometrics, 34,111-120.

Jansen, W. J. and G. G. Schulze (1996) "Theory--Based Measurement of the Saving Investment Correlation with an Application to Norway." Economic Inquiry, 34, 116-132.

Johansen, S. (1991) "Estimation and Hypothesis Testing of Cointegration Vectors in Gaussian Vector Autoregressive Models," Econometrica, 59, 1551-1580.

----- and K. Juselius (1990) "Maximum Likelihood Estimation and Inference on Cointegration--with Applications to the Demand for Money," Oxford Bulletin of Economics and Statistics, 52, 169-210.

Miller, Stephen (1988) "Are Saving and Investment Cointegrated?" 27, 31-34.

----- (1991) "Monetary Dynamics: An Application of Cointegration and Error-Correction Modeling," Journal of Money, Credit, and Banking, 23, 139-154.

Miller, Stephen and Frank S. Russek (1990) "Cointegration and Error-Correction Models: The Temporal Causality between Government Taxes and Spending," Southern Economic Journal, 221-229.

Murphy, Robert, G. (1984) "Capital Mobility and the Relationship between Saving and Investment in OECD Countries," Journal of International Money and Finance, 10, 327-342

Nelson, C. R. and C. I. Plosser (1982) "Trends and Random Walks in Macroeconomic Time Series: Some Evidence and Implications," Journal of Monetary Economics, 10, 139-162.

Obstfeld, Maurice (1986) "Capital Mobility in the World Economy: Theory and Measurement," Carnegie-Rochester Conference Series on Public Policy, Spring, 24, 55-104.

Stockman, Allen C. and Linda Tesar L. (1991) "Tastes and Technology in a Two-Country Model of the Business Cycle: Explaining International Co-movements," Rochester Center for Economic Research Working Paper No. 255, University of Rochester.

Summers, Lawrence H. (1988) "Tax Policy and International Competitiveness," in J. Frenkel, ed., International Aspect of Fiscal Policies, Chicago: University of Chicago press, 349-375.

Taylor, Alan M. (1996) "International Capital Mobility in History: The Saving-Investment Relationship," NBER Working Paper 5743 (Cambridge, Massachusetts: National Bureau of Economic Research, September).

Tesar, Linda L. (1991) "Savings, Investment and International Capital Flows," Journal of International Economics, 31, 55-78.

World Bank (1993) "The East Asian Miracle: Economic Growth and Public Policy" New York: Oxford.

 Gross Domestic Saving (as % of GDP)
Country 1960-1970 1970-1980 1980-1990 1990-1995 1996
Indonesia 8.55 26.16 30.82 32.26 33.15
Malaysia 24.11 30.02 33.34 35.72 41.90
Philippines 18.84 24.83 20.73 16.62 14.40
Singapore -1.90 29.42 41.97 47.56 50.47
Thailand 18.88 22.31 27.11 35.57 35.31
Source: World Bank, World Development Indicators 1998.
 Domestic Investment (as % of GDP)
Country 1960-1970 1970-1980 1980-1990 1990-1995 1996
Indonesia 10.30 21.85 27.20 29.75 31.83
Malaysia 19.45 25.92 30.81 36.47 41.24
Philippines 19.71 27.19 22.64 22.66 24.19
Singapore 21.50 41.00 41.65 34.93 35.07
Thailand 20.98 26.11 30.50 41.28 41.00
Source: World Bank, World Development Indicators 1998.
 Unit Root Tests (1960-1996)
Country/ DF
Period Series Level
Indonesia
1960-1996 GDS [T.sub.[micro]] = -1.61
 [T.sub.[tau]] = -1.64
 GDI [T.sub.[micro]] = -1.98
 [T.sub.[tau]] = -1.91
Malaysia
1960-1996 GDS [T.sub.[micro]] = -1.70
 [T.sub.[tau]] = -2.03
 GDI [T.sub.[micro]] = -0.83
 [T.sub.[tau]] = -1.96
Philippines
1960-1996 GDS [T.sub.[micro]] = -1.17
 [T.sub.[tau]] = -1.79
 GDI [T.sub.[micro]] = -1.67
 [T.sub.[tau]] = -1.67
Singapore
1960-1996 GDS [T.sub.[micro]] = -1.90
 [T.sub.[tau]] = -1.90
 GDI [T.sub.[micro]] = -2.22
 [T.sub.[tau]] = -1.80
Thailand
1960-1996 GDS [T.sub.[micro]] = -0.69
 [T.sub.[tau]] = -1.92
 GDI [T.sub.[micro]] = -0.88
 [T.sub.[tau]] = -2.13
Country/ ADF
Period Difference Level
Indonesia
1960-1996 [T.sub.[micro]] = -6.70 [**] [T.sub.[micro]] = -1.64
 [T.sub.[tau]] = -7.63 [**] [T.sub.[tau]] = -1.54
 [T.sub.[micro]] = -5.98 [**] [T.sub.[micro]] = -2.17
 [T.sub.[tau]] = -6.78 [**] [T.sub.[tau]] = -1.76
Malaysia
1960-1996 [T.sub.[micro]] = -7.02 [**] [T.sub.[micro]] = -0.71
 [T.sub.[tau]] = -6.89 [**] [T.sub.[tau]] = -2.47
 [T.sub.[micro]] = -4.27 [**] [T.sub.[micro]] = -1.21
 [T.sub.[tau]] = -4.20 [**] [T.sub.[tau]] = -3.06
Philippines
1960-1996 [T.sub.[micro]] = -7.40 [**] [T.sub.[micro]] = -0.54
 [T.sub.[tau]] = -7.59 [**] [T.sub.[tau]] = -1.18
 [T.sub.[micro]] = -4.16 [**] [T.sub.[micro]] = -2.22
 [T.sub.[tau]] = -4.10 [**] [T.sub.[tau]] = -2.19
Singapore
1960-1996 [T.sub.[micro]] = -4.26 [**] [T.sub.[micro]] = -1.46
 [T.sub.[tau]] = -4.30 [**] [T.sub.[tau]] = -1.68
 [T.sub.[micro]] = -4.77 [**] [T.sub.[micro]] = -2.27
 [T.sub.[tau]] = -5.18 [**] [T.sub.[tau]] = -1.83
Thailand
1960-1996 [T.sub.[micro]] = -7.79 [**] [T.sub.[micro]] = -0.26
 [T.sub.[tau]] = -7.91 [**] [T.sub.[tau]] = -1.42
 [T.sub.[micro]] = -5.50 [**] [T.sub.[micro]] = -0.84
 [T.sub.[tau]] = -5.45 [**] [T.sub.[tau]] = -2.19
Country/
Period Difference Lag Order
Indonesia
1960-1996 [T.sub.[micro]] = -4.13 [**] 2
 [T.sub.[tau]] = -4.85 [**] 2
 [T.sub.[micro]] = -3.62 [**] 2
 [T.sub.[tau]] = -4.33 [**] 2
Malaysia
1960-1996 [T.sub.[micro]] = -4.40 [**] 2
 [T.sub.[tau]] = -4.29 [**] 2
 [T.sub.[micro]] = -3.82 [**] 4
 [T.sub.[tau]] = -3.78 [**] 4
Philippines
1960-1996 [T.sub.[micro]] = -4.45 [**] 1
 [T.sub.[tau]] = -4.72 [**] 1
 [T.sub.[micro]] = -4.48 [**] 1
 [T.sub.[tau]] = -4.42 [**] 1
Singapore
1960-1996 [T.sub.[micro]] = -4.14 [**] 1
 [T.sub.[tau]] = -4.31 [**] 1
 [T.sub.[micro]] = -2.96 [**] 1
 [T.sub.[tau]] = -3.25 [**] 1
Thailand
1960-1996 [T.sub.[micro]] = -3.77 [**] 2
 [T.sub.[tau]] = -4.02 [**] 2
 [T.sub.[micro]] = -4.22 [**] 2
 [T.sub.[tau]] = -4.20 [**] 2

(*.)and (**.)10 and 5 percent significance levels, respectively. [T.sub.[micro]] = without trend. DF Dickey-Fuller statistic, ADF = augmented Dickey-Fuller statistic, [T.sub.[tau]] = with trend. GDS = ratio of gross domestic savings to GDP and GDI = ratio of gross domestic investment to GDP. The critical values at the 5% level of significance are -2.96 and -3.57, respectively for without trend and with trend. The 10% critical values for without trend and with trend are -2.26 and -3.20, respectively. The lag orders are determined by Akaike Information Criterion (AIC).

 Johansen Cointegration Test Results
 Null: r = 0
Country Trace [[lambda].sub.max]
Indonesia 26.86 [**] 21.66 [**]
Malaysia 25.42 [**] 20.66 [**]
Philippines 71.11 [**] 67.42 [**]
Singapore 34.80 [**] 31.50 [**]
Thailand 30.92 [**] 24.69 [**]
 Null: r
 [less than or equal to] 1
Country Trace [[lambda].sub.max]
Indonesia 5.20 5.20
Malaysia 4.76 4.76
Philippines 3.69 3.69
Singapore 3.30 3.30
Thailand 6.23 6.23

(**.)indicates the rejection of the null hypothesis at the 5% significance level. The critical values for the trace test hypotheses r [less than or equal to] 1 and r [less than or equal to] 0 are 11.54 and 18.33 respectively. The critical values for the [[lambda].sub.max] test hypotheses r [less than or equal to] 1 and r [less than or equal to] 0 are 11.54 and 23.83 respectively. The critical values are obtained from the Microfit 4.0 program.

 F-Statistics for Bivariate Causality
 Tests Based on VECM
 Indonesia Malayasia
Panel A: Saving Equation ([delta]GDS)
[z.sub.t-1] 9.13 [*] 5.04 [*]
[sigma][delta]GDI 1.86 6.34 [*]
Panel B: investment Equation ([delta]GDI)
[z.sub.t-1] 0.08 7.47 [*]
[sigma][delta]GDS 0.23 1.58
 Philippines Singapore
Panel A: Saving Equation ([delta]GDS)
[z.sub.t-1] 0.56 7.79 [*]
[sigma][delta]GDI 1.21 0.18
Panel B: investment Equation ([delta]GDI)
[z.sub.t-1] 7.12 [*] 0.12
[sigma][delta]GDS 1.88 0.13
 Thailand
Panel A: Saving Equation ([delta]GDS)
[z.sub.t-1] 0.42
[sigma][delta]GDI 3.12 [**]
Panel B: investment Equation ([delta]GDI)
[z.sub.t-1] 15.38 [*]
[sigma][delta]GDS 2.73 [***]
(*.), (**.), and (***.)represent rejection
of the hypotheses at 1%, 5%, and 10% level
of significance. The estimated error-
correction models are as follows: