Are Complementary Relationship between Public Physical Capital Formation and Private Physical Capital Formation truly Exist and stay unchanged in Malaysia? ANDERSON SENGLI Department of Economics, Faculty of Economics and Business, Universiti Malaysia Sarawa, 94300 Kota Samarahan, Sarawa, MALAYSIA anderson_sengli@yahoo.co.u THURAI MURUGAN NATHAN Department of Economics, Faculty of Business and Finance, Universiti Tunu Abdul Rahman, 31900 Kampar, Pera, MALAYSIA thurai@utar.edu.my Abstract This paper investigates the existence of complementary relationship between public and private physical capital investment in Malaysia. Empirical result reveals that the complementary relationship exists in a reverse form of public capital hypothesis, where the growth of public physical capital investment is led by the growth of private physical capital investment. The complementary relationship remains unchanged even the instability occurs to the private capital investment after 1997 Asian financial crisis. To achieve the public capital hypothesis, public sector investment should be further strengthened. Keywords: Public Physical Capital Formation, Private Physical Capital Formation, econometric, ARDL, Unit Root Test 1. INTRODUCTION A major challenge for countries that are struggling to achieve higher and more sustainable medium term growth is in finding ways to revitalize private investment [4] which is regarded as an investment that can potentially transmit significantly positive effect on private sector output, productivity, and capital formation. Public investment may also increase aggregate output and thus enhance the physical and financial resources in the economy as well as public spending on infrastructure such as roads, highways, education, sewer and water systems, and power plants, which in turn reduces the burden of the private sector [7]. [7] added such infrastructure investments complement private investment and raise the productivity of private capital at the same time. Public and private investments may be lined by a complementary relationship if public capital exerts positive stimulus on the private sector [9]. Based on our nowledge, the complementary relationship is considered to exists if the growth of public capital directly influence the growth of private capital or reversely, when the growth of private capital indirectly influence the growth of public capital. In Malaysia, the volume of public and private investment in the forms of public physical capital formation (PCPCF) and private physical capital formation (PVPCF), were increasing annually until the strie of the 1997 Asian financial crisis (1997 AFC). After the catastrophic crisis, the growth of PVPCF became unstable while PCPCF continued to grow. The complementary relationship between PCPCF and PVPCF might have been affected by the presence of economic crisis or even PCPCF itself. According to [7] there are two cases in which public investment may negatively affect private investment. Firstly, the public sectors may have been competing with the private sector to obtain the same resources in the economy, and thus crowds out private capital investment. Secondly, the public investment may have substituted private investment when both are producing goods and services that are in direct competition in a maretplace, and 199
especially when the public production is subsidized by the government 1. A budget constraint faced by public sector may further exacerbate distortions in the economy and increase the cost of inputs, leading to an adverse effect on expected output growth and private investment [10]. The interest in evaluating the impact of public capital formation on private investment was spared by [1] and [2], who concluded that public infrastructure or public capital formation had very strong positive effect on private sector productivity in his own study. [1] claimed that the positive influence of public investment on private investment can be explained by the public capital hypothesis. According to this hypothesis, an increase in public investment results in an increase in private investment. This is due to the availability of economic and social infrastructure that have brought conducive conditions for private decision to invest, most prominently by offering short-run and long-run essential services to the production system [9]. To test whether the complementary relationship between public and private capitals in Malaysia exists in accordance with the public capital hypothesis or otherwise, this paper has been produced to provide a piece of empirically new evidence, obtained from different econometric approaches, sample sizes, and countries. 2. Data, Methodology, and Result The annual data of public GFCF (Gross Fixed Capital Formation) and private GFCF from 1970 to 2011 were taen from various publications of the Malaysian Ministry of Finance. Both data were transformed into natural logarithms. Augmented Dicey-Fuller (ADF) [5] and Phillips-Perron (PP) (Phillips and Perron, 1988) unit root tests were initiated to ensure that the data used were not stationary in the form of second difference. Table 1 shows the results of unit root tests. The time series being tested was stationary at level I(0) and in the first difference I(1) form only. As such, the autoregressive distributed lag (ARDL) test was deemed appropriate since the variables were in a mixed order of I(0) and I(1), but without I(2). TABLE 1: Unit Root Test Result Augmented Dicey-Fuller (ADF) Series Level First Difference Intercept Intercept and Trend Intercept Intercept and Trend LPCPCF -2.022 (1) -2.726 (3) -3.531 (0)** -3.793 (0)** LPVPCF -2.853 (0)*** -2.389 (3) Phillips-Perron (PP) Series Level First Difference Intercept Intercept and Trend Intercept Intercept and Trend LPCPCF -2.309 (2) -2.073 (2) -3.576 (1)** -3.849 (1)** LPVPCF -2.733 (3)*** -3.314 (3)*** Notes: Asteris (**) and (***) indicate statistically significant at 5 percent and 10 percent level, respectively. Number in the bracet represents optimum lag length. By using the ARDL approach as proposed by [12] the existence of long-run relationship between PCPCF and PVPCF can be captured. The two bound test models are as represented in equation 1 and equation 2 respectively, denoted as F PCPCF (PCPCF PVPCF) and F PVPCF (PVPCF PCPCF). LPCPCF t = ρ 0 + i=1 τ 1,i LPCPCF t-i + i=0 τ 2,i LPVPCF t-i + τ 3 PCPCF t-1 + τ 4 PVPCF t-1 + ε t (1) 1 This statement was supported by [10] who stated that, public enterprises may also produce goods and service which compete directly with the private sector, maing the two investments into substitutes. 200
LPVPCF t = σ 0 + i=1 γ 1,i LPVPCF t-i + i=0 γ 2,1 LPCPCF t-i + γ 3 PVPCF t-1 + γ 4 PCPCF t-1 + ε t (2) where is the first-difference operator; is the lag lengths; τ 1,i and τ 2,i (Eq. 1) as well as γ 1,i and γ 2,i (Eq. 2) represents the short-run dynamics of the model; τ 3 and τ 4 (Eq. 1) as well as γ 3 and γ 4 (Eq. 2) represents the long-run relationship and ε t is a white noise error term. The null hypothesis in the ARDL is as follows: H 0 :τ 3 =τ 4 =0 ; γ 3 =γ 4 =0, (3) H 1 :τ 3 τ 4 0 ; γ 3 γ 4 0. (4) Besides that, this ARDL model also taes the error correction factors of previous periods into account. These error correction terms, EC t, and lag difference terms can test both short-term and long-term relationships [15]. The general error correction model is as follows: LPCPCF t = ρ 0 + i=1 τ 1,i LPCPCF t-i + i=0 τ 2,i LPVPCF t-i + ϑec t + ε t (5) LPVPCF t = σ 0 + i=1 γ 1,i LPVPCF t-i + i=0 γ 2,1 LPCPCF t-i + υec t + ε t (6) where ϑ and υ are the speeds of the adjustment parameter and expected to be negative as well as statistically significant. It indicates how fast the current differences in a dependent variable respond to the error correction term in disequilibrium within the previous period. EC t represents the residuals obtained from the estimated cointegration model [15]. In this study, the cointegration relationship was examined using the F-test in the ARDL framewor. The computed F-test value was compared with the critical value. In this ARDL framewor, if the F-test value is higher than the upper bound of the critical value, then the null hypothesis is rejected and it can be concluded that there is a long-run relationship among the variables. If the F-test value is less than the lower critical value, then the null hypothesis of the cointegration relationship is accepted, which means that no cointegration relationship exists among the variables. On the other hand, if the F-test value is between the lower and upper bound value, it means that the results are inconclusive [12]. However, there are a few steps that have to be taen care of before the cointegration test can be conducted. Firstly, the optimum lag should be identified based on the Schwarz-Bayesian criterion (SBC), [14] and/or Aaie Information criterion (AIC) [3]. Then, the ordinary least square (OLS) technique should be used with the selected model to find out the F-test value. If this value is higher than the critical value, the null hypothesis of no long-run relationship among the variables is rejected. Finally, the long-run relationship and error correction model (ECM) can be estimated using the selected lags. The ECM term should have a negative sign and should be statically significant. TABLE 2: Bound Test Result Model Dependent Variable Lag a F-statistic Critical Value Bound b F PVPCF (PVPCF PCPCF) LPVPCF 1 3.925 10.0% 4.04-4.78 5.0% 4.94-5.73 F PCPCF (PCPCF PVPCF) LPCPCF 1 4.837* 2.5% 5.77-6.68 1.0% 6.84-7.84 Note: Asteris (*) indicates variables cointegrated at 10 percent significance level. a Optimal lag determined by Schwarz Bayesian Criterion [14] and fix at 1 for both models. b Critical values obtained 201
from Table CI(iii) Case III: Unrestricted intercept and no trend reported in [12]. Based on the bound test result (see Table 2), there is no evidence of long-run relationship in F PVPCF (PVPCF PCPCF). The F-statistic for this model is less than the lower bound critical value. Hence, PCPCF does not influence the growth of PVPCF in the long-run. However, when PCPCF becomes the dependent variable, a long-run relationship emerged where the F-statistic for F PCPCF (PCPCF PVPCF) becomes greater than the upper bound critical value. This means that PVPCF can influence the growth of PCPCF in the longrun. To validate this result, an estimation of F PCPCF (PCPCF PVPCF) long-run coefficient and error correction model (ECM) was done. The estimation result of F PCPCF (PCPCF PVPCF) long-run coefficient and ECM (see Table 3) has validated the results of the bound test. The long-run (LR) coefficient of PVPCF is highly significant and positive, which directly indicates that a 1 percent increase in PVPCF can increase the PCPCF by roughly 1.00 percent as well. Meanwhile, the significant and negative ECT coefficient indicated that deviation from the long-term PCPCF has to be corrected by 26 percent over the following year and the long-run causality exists from PVPCF towards PCPCF. TABLE 3: Estimated Long-Run Coefficient and Error Correction Model (ECM) Panel A: Estimated Long-Run Coefficient Bound Test Model Lag a Dependent LR Regressor Variable Coefficient b t-statistic [Prob] F PCPCF (PCPCF PVPCF) 1, 0 LPCPCF LPVPCF Intercept 1.058* -0.868 13.781 [0.000] -0.645 [0.523] Panel B: Error Correction Model (ECM) Bound Test Model Lag a Dependent ECT Regressor Variable Coefficient c t-statistic [Prob] F PCPCF (PCPCF PVPCF) 1, 0 LPCPCF LPVPCF Intercept ECT 0.273* -0.224-0.258* 3.380 [0.002] -0.586 [0.561] -3.957 [0.000] Note: Asteris (*) indicates that the coefficients are significant at 1 percent level. a Lag length was determined automatically by Schwarz Bayesian Criterion [14]. b Long-run coefficient. c Error correction term coefficient. [Prob] refers to the probability. The stability of F PCPCF (PCPCF PVPCF) estimated coefficient was checed by using the cumulative of sum (CUSUM) and cumulative sum of square (CUSUMSQ) stability tests under the ARDL approach. The plots of CUSUM and CUSUMSQ statistics, as presented in Figure 1, are respectively within the 5 percent critical bounds line, indicating that all parameters and variances in the estimated ECM model are stable over the sample period. CUSUM FIGURE 1: Stability Test Result CUSUMSQ 202
After a long-run relationship had been identified, the [8] Causality test was used to investigate the causal relationships among the variables. This was done because the cointegration test earlier could only detect the long-run relationship, not the direction of causality [11]. According to[6], if the past value of X t is able to predict the Y t value, then X t is taen as the Granger cause of Y t. Else, it can be said that Y t is the Granger causes of future X t values. The null hypothesis of the Granger causality test is that X t does not Granger cause Y t and vice versa. The Granger causality test based on ARDL framewor (Wald test), also nown as the short-run causality test, was used to find out the causality effect among the variables. The causality relationships were discovered by using a common factor that could restrict the lags of the variables coefficient, which should be zero in this case. If the null hypothesis of no causality can be rejected, it means that the variable are Granger caused (less than 5% of significance level). The results of bivariate Granger s causality test are summarized in Table 4. The results showed that there is a bi-directional causal relationship from PVPCF to PCPCF in the short-run where the chi-square statistic of the Wald test is highly significant. TABLE 4: Granger Causality Test Result Wald Test Bound Test Model Null Hypothesis Chi-square p-value F PCPCF (PCPCF PVPCF) PVPCF not Granger-cause PCPCF 129.363* 0.000 Note: Asteris (*) indicates chi-square statistic significant at 1 percent level or rejection of the null hypothesis at 1 percent level. P-value refers to the probability. 3. Concluding Remars The aim of this paper is to test whether the complementary relationship between public and private capital investment in Malaysia, in terms of public physical capital formation (PCPCF) and private physical capital formation (PVPCF), exists in accordance with the public capital hypothesis (where an increase in public investment results in an increase in private investment) or reversely (where an increase in private investment result in an increases in public investment). In addition, this study was carried out to ascertain that the instability of PVPCF after the 1997 Asian financial crisis did not affect the hypothesized or reverse complementary relationship between PCPCF and PVPCF. From the results, it is found that the complementary relationship between PCPCF and PVPCF in Malaysia exists in a reversed form in both longrun and short-run. The existence of a reversed complementary relationship along the period of 1970-2011 indicated that PCPCF was crowded in by PVPCF, as such supported the notion that private investment preceded public investment. In other words, the public sector had acted in providing a complementary physical capital for most physical capital investments made by the private sector. The result also proved that the reverse complementary relationship remained unaffected by economics crisis or other factors. From a policy perspective, we believe that the hypothesized complementary relationship is better for developing countries such as Malaysia. To achieve the public capital hypothesis, public sector investment should be further strengthened. This cannot be done by improving physical capital solely; it must be baced with financial and human capital improvement. References: [1] Aschauer, D.A. (1989a). Is public expenditure productive? Journal of Monetary Economics, 23, 177-200. [2] Aschauer, D.A. (1989b). Does Public Capital Crowd Out Private Capital? Journal of Monetary Economics, 24(2), 171-188. [3] Aaie, H. (1987). Factor analysis and AIC. Psychometria, 52, 317-332. [4] Coutinho, R.M., & Gallo, G.M. (1991). Do Public and Private Investment Stand in each other s Way? WDR Bacground Paper, World Ban. [5] Dicey, D.A., & Fuller, W.A. (1981). Lielihood Ratio Statistics for Autoregressive Time Series with a Unit Root, Econometrica, 49, 1057-1072 203
[6] Engle, R.F., & Granger, C.W.J. (1987). Cointegration and Error Correction: Representation, Estimation and Testing. Econometrica, 55, 251-276. [7] Erden, L. & Holcombe, R.G. (2006). The Linage between Public and Private Investment: A Cointegration Analysis of a Panel of Developing Countries. Eastern Economic Journal, 32(3), 479-492. [8] Granger, C. W. J. (1969). Investigating Causality Relationships by Economics Models and Cross- Spectral Models. Econometrica, 37, 424-438. [9] Hassan, S., Othman, Z., & Abdul Karim, M.Z. (2011). Private and Public Investment in Malaysia: A Panel Time-Series Analysis. Internati_onal Journal of Economics and Financial, 1(4), 199-210. [10] Khan, M. & Kumar, M. S. (1997). Public and Private Investment and the Growth Process in Developing Countries, Oxford Bulletin of Economics and Statistics, 59(1), 69-88. [11] Oztur, I., & Acaravci, A., (2010). The Causal Relationship between Energy Consumption and GDP in Albania, Bulgaria, Hungary and Romania: Evidence from ARDL Bound Testing Approach. Applied Energy, 87, 1938-1943. [12] Pesaran, M. H., Shin, Y. & Smith, R. J. (2001). Bounds Testing Approaches to the Analysis of Level Relationship. Journal of Applied Econometrics, 16(3), 289-326. [13] Philips, P.C.B. & Perron, P. (1988). Testing for a Unit Root in Time Series Regression. Biometria, 75(2), 335-346. [14] Schwarz, G. (1978). Estimating the Dimension of a Model. Annals of Statistics, 6, 461-464. [15] Wang, Y. S., (2009). The Impact of Crisis Events and Macroeconomic Activity on Taiwan s International Inbound Tourism Demand. Tourism Management, 30, 75-82. 204