The current issue and full text archive of this journal is available at www.emeraldinsight.com/0144-3585.htm JES 37,6 658 Savings and investment in South Asia Evidence from likelihood ratio based panel cointegration Abu N.M. Wahid Received 11 November 2008 Accepted 20 October 2009 Tennessee State University, Nashville, Tennessee, USA Mohammad Salahuddin Southeast University, Dhaka, Bangladesh, and Abdullah M. Noman American International University, Dhaka, Bangladesh Abstract Purpose – This paper seeks to contribute to the study of the relationship between savings and investment in a panel of five South Asian countries. Design/methodology/approach – A number of unit root tests such as Levin, Lin, and Chu or LLC, Breitung, Im, Pesaran, and Shin or IPS, Fisher-type tests using ADF and Fisher-type tests using PP tests are conducted that confirm the non-stationarity of data. Then the paper applies maximum likelihood-based panel cointegration method to examine the relationship between savings and investment using data on investment and savings for five South Asian developing countries, namely, Bangladesh, Pakistan, India, Nepal, and Sri Lanka over the period 1973-2007 compiled from the World Development Indicator (WDI) Database 2008 CD-ROM. Findings – The results obtained suggest that savings and investment are cointegrated, which implies that the Feldstein-Horioka (F-H) puzzle does not hold in this region. It is also found that most of these countries have maintained an international solvency condition. Originality/value – This is another contribution that would enrich the existing literature on the F-H puzzle. The paper includes data that involve the longest sample period. No other study, as of now, has employed the panel cointegration method to study the savings investment relationship in this region. Keywords Savings, Investments, South Asia Paper type Research paper Journal of Economic Studies Vol. 37 No. 6, 2010 pp. 658-666 q Emerald Group Publishing Limited 0144-3585 DOI 10.1108/01443581011086684 1. Introduction The positive relationship between savings and investment is now an empirically established fact and well documented in the economics literature. The studies of recent times also corroborate this relationship (see for example, Salahuddin and Islam, 2008). In addition, in the modern era of financial and economic deregulation, capital markets across the world are expected to be highly integrated with each other. Increased integration among financial markets would enhance the set of opportunities available to participants of financial markets, and, thereby, stimulate business and economic growth. However, the relationship between savings and investment is not beyond JEL classifications – C32, F21, F32 dispute. Keynes’ well-known “paradox of thrift” insists that policies to encourage savings by raising investment and growth might in fact prove to be futile, thus, undermining the link between savings and investment. It is likely that the developed nations have undertaken more measures to deregulate their financial markets than the developing nations. Consequently, capital is expected to be more mobile in those markets when compared to capital markets of developing nations, as the speed and magnitude of deregulation are slower in the latter case. Contrary to this theoretical expectation, the seminal work of Feldstein and Horioka in 1980 (Feldstein and Horioka, 1980) on savings investment correlation found that domestic savings and investment were highly correlated implying low capital mobility across 16 organization for economic cooperation and development (OECD) countries. Using a cross sectional analysis, they showed that 85-95 percent of domestic savings were transformed into investment in the domestic economies. The authors interpreted the results as evidence of poor capital mobility, which was contrary to what one would expect given the level of deregulation and capital markets integration measures undertaken in the OECD member countries. Subsequently, this has come to be known as the Feldstein-Horioka puzzle (or F-H puzzle, in short) (Obstfeld and Rogoff, 2000). Since this groundbreaking work, the interplay between these two important macroeconomic variables has drawn researchers’ and policymakers’ special attention that has resulted in voluminous research and intellectual debate on the issue. Some researchers have questioned the validity of their interpretation of the high savings – investment correlation as evidence of poor capital mobility and offered alternative explanations for it, while others tried to rehabilitate the puzzle. The main contribution of this paper is as follows: It estimated the F-H relationship using the longest sample period (1973-2007). We also employed recently developed likelihood ratio based panel cointegration to examine savings-investment relationship in South Asia. 2. Literature review The savings investment correlation puzzle (the F-H puzzle) still remains a puzzle and it continues to draw attention of economists around the world. This is evident from the large and growing body of literature on the issue. There are basically two strands of literature that deal with this puzzle. The first strand of literature attempts to establish the validity of F-H puzzle as a means of measuring the degree of capital mobility by extending the sample period and by accounting for different exchange rate regimes, structural breaks due to the abolition of exchange controls, and temporary and permanent shocks to savings and investment (De Vita and Abott, 2003). The empirical literature includes both cross section and time series investigations. Among cross sectional studies, the findings of Feldstein (1983), Penati and Dooley (1984), Dooley et al. (1987), Vos (1998) and Tesar (1991) suggest a high correlation between savings and investment. A number of studies employed cointegration techniques. Miller (1988) was the pioneer to apply cointegration technique to study the dynamic savings investment relationship. Hoffmann (1998) stressed that in intertemporal optimization models of current account dynamics, the budget constraint would lead to high correlation and hence cointegration between savings and investment. De Vita and Abott (2002, 2003) found cointegrated relationship between savings and investment for USA and UK data. Ang (2007) also established a Savings and investment in South Asia 659 JES 37,6 660 robust cointegrated relationship between savings and investment for Malaysia. However, the results of some of the studies (for example, Schmidt, 2003; Narayan, 2005) failed to find any cointegrated relationship between savings and investment. Another strand of literature offers alternative hypotheses to explain high savings investment correlation. It argues that high savings investment correlation has nothing to do with capital mobility rather it is due to other macroeconomic factors such as productivity shocks (Obstfeld, 1986) country size (Baxter and Crucini, 1993), current account solvency (Coakley et al., 1996), long run current account targeting in economic policy (Summers, 1988; Artis and Bayoumi, 1989) and financial structure (Kasuga, 2004.) Some of the studies employ panel estimation techniques (see inter alia: Coakley and Kulasi, 1997; Coakley et al., 1996; Coiteux and Olivier, 2000; Jansen, 2000; Corbin, 2001; Kim, 2001; Ho, 2002; Coakley et al., 2004; Kollias et al., 2008) to study savings investment relation. The results are mixed. Kool and Keijzer(2009) in a study establish a structural link between three puzzles in international macroeconomics and finance; the F-H puzzle, the equity-home bias puzzles (EHB) and the trade home bias puzzle. Another study (Georgopoulos and Hejazi, 2009) demonstrates that the estimates of the F-H coefficient under standard framework are biased upward in the presence of a high positively correlated inward and upward capital flows. To shed further light on the relationship between savings and investment and contribute to the existing pool of literature on the topic, we re-examine the relationship between savings and investment using the most recent data covering the period 1973-2007 for a panel of five South Asian countries namely Bangladesh, India, Pakistan, Sri Lanka, and Nepal. To this end, we have employed panel cointegration technique to assess the long run relationship between savings and investment. 3. Data and methodology This paper includes data on investment and savings for the five South Asian developing countries, namely, Bangladesh, Pakistan, India, Nepal, and Sri Lanka over the period of 1973-2007 compiled from the World Development Indicator (WDI) Database 2008 CD-ROM. For the purpose of estimating the model of this paper, domestic investment is calculated as a share of GDP (i.e. the investment-GDP ratio) and domestic savings as a share of GDP (i.e. the saving-GDP ratio). The application of the Feldstein-Horioka puzzle (relationship between saving and investment) involves estimation of the following regression equation: I t ¼ a þ b S t þ ut ð1Þ where, I is the ratio of gross domestic investment to GDP, and S is the ratio of gross national savings to GDP, and u is the error term. Panel unit root tests Before we can test for the presence of cointegrating relationship between investment and savings in South Asian nations, time series properties of the panel data need to be examined. To this end, a number of panel unit root tests are applied here; Levin et al. (2002), or LLC, Breitung (2000), Im et al. (2003), or IPS, Fisher-type tests using ADF (Maddala and Wu, 1999), and Fisher-type tests using PP tests (Choi, 2001), and Hadri (2000). Among them, the LLC, Breitung and Hadri tests assume common unit root process across all units, while the other three tests allow individual unit root process. The null hypothesis of all panel unit root tests, with the exception of Hadri, is that the variable under investigation is nonstationary or has a unit root. On the other hand, Hadri test considers presence of no unit root under the null hypothesis against the alternative hypothesis of a unit root in the data. Moreover, except for the Fisher-type tests using ADF proposed by Maddala and Wu (1999), all of the tests are asymptotic in nature. The IPS test accounts for a higher degree of heterogeneity in the cross-sectional data as compared to the other tests. The panel cointegration test The paper applies the panel cointegration technique for the whole sample at hand. A motivation of considering the panel cointegration test is to have more reliable results from a panel than from the individual countries as the former contain more information than the latter (Baltagi, 2005). We choose to apply the Larsson et al. (2001) maximum likelihood based panel cointegration test that is straightforward to formulate and superior in performance to many other tests (Karaman Örsal, 2008). Larsson et al. (2001) assumes that the data generating process for each of the cross-sections can be represented by an ECM specification as in the following model: DY i;t ¼ Pi Y i;t21 þ n X k¼1 Gik DY i;t2k þ ui;t ð2Þ The estimation of the above model is carried out individually for each cross-section using maximum likelihood methods. The null and the alternative hypotheses of the test are as follows: H0 : rankðPi Þ ¼ r i # r H A : rankðPi Þ ¼ p for all i ¼ 1; . . . . . . ; N ð3Þ where, p is the number of variables that we use in order to test for possible cointegration among them. The trace statistic for each cross-sectional unit and their average are then obtained. Now, the standardized panel cointegration rank trace test-statistic (denoted by YLR) is given by: pffiffiffiffi N ½LRNT~ 2 EðZ k Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Y LR ¼ N ð0; 1Þ ð4Þ Var ðZ k Þ where, LRNT is the average of the trace statistic obtained from each cross-sectional calculations, and E(Zk) and Var (Zk) are the mean and variance of the asymptotic trace statistic reported in Larsson et al. (2001). 4. Estimation results Table I presents results of the panel unit root tests. As seen from the Table, the LLC and Breitung tests cannot reject the null hypothesis of unit root both for the investment and savings series on the level. However, when we consider the first difference of the variables, the null hypothesis is rejected for both of the series. Similarly, for IPS and Savings and investment in South Asia 661 (I/Y)t p-value 0.700 0.804 (0.758) (0.789) 2 9.270 * 2 4.798 * 2.197 6.108 2.862 (0.986) (0.806) (0.984) 6.901 * (0.000) Stat D(I/Y)t p-value JES 37,6 662 Table I. Panel unit root tests results Null: unit root (assumes common unit root process) LLC t * Breitung t-stat Null: unit root (assumes individual unit root process) IPS W-stat ADF-Fisher Chi-square PP-Fisher Chi-square Null: no unit root (assumes common unit root process) Hadri Z-stat Stat (S/Y)t D(S/Y)t p-value Stat p-value Stat (0.000) (0.000) 1.264 2 0.705 (0.897) (0.240) 29.391 * 23.717 * (0.000) (0.000) 2 8.875 * 89.101 * 110.211 * (0.000) (0.000) (0.000) 0.961 12.839 13.955 (0.831) (0.232) (0.175) 211.432 * 112.126 * 138.408 * (0.000) (0.000) (0.000) 0.683 (0.247) 6.121 * (0.000) 0.160 (0.436) Notes: Probabilities for Fisher tests are computed using an asymptotic Chi-square distribution. All other tests assume asymptotic normality. Optimum bandwidth for Hadri Z-test and Fisher PP test are selected using Bartlett kernel and for all other tests lags are selected using SIC two Fisher-type tests (ADF and PP), one cannot reject the null hypothesis of the unit root on the level and the null hypothesis is rejected on the data in first difference, both for investment and savings. These tests confirm that both investment and savings series are I ð1Þ. The Hadri Z-test (with null hypothesis of no unit root) results at the bottom part of Table I shows that the null hypothesis is rejected for both the variables. The null hypothesis is, however, not rejected when first difference-data are used. Different test statistics reported in Table I unequivocally indicate that the variables under investigation are unit root process. This allows us to test for cointegration between the variables. The results of the Larsson et al. (2001) panel cointegration test are presented in Table II. Table II shows results of Johansen cointegration test trace statistic for individual countries. The VAR lag is selected using Swartz Information Criteria (SIC) from a maximum lag of 4. The results for individual countries show that ranks are either 0 (India and Pakistan) or 1 (Bangladesh, Nepal, and Sri Lanka). The panel test statistic (the last row in bold face) is reached at by utilizing the method described earlier. The null hypothesis of no cointegration is rejected for r ¼ 0 (as the test statistics are larger than the critical value of 1.96) but cannot be rejected for r ¼ 1. This establishes the highest common rank of the panel to be one. This indicates that savings and investment are cointegrated for the South Asian panel as a whole. Similar results were obtained in Murthy (2007, 2009) for a panel of Latin America and the Caribbean countries, which use the same method as ours. The panel results are more reliable on the grounds of enhanced power of rejecting the null hypothesis of no cointegration in the presence of more information as compared to the time series evidence. The results obtained suggest that most of these countries have maintained international solvency condition. Second, as cointegration is found in the panel as a whole, the F-H puzzle does not hold for this region. Savings and investment in South Asia 663 5. Summary and conclusions This paper aims to examine the savings-investment relationship for some South Asian countries, namely Bangladesh, India, Pakistan, Sri Lanka, and Nepal by employing the likelihood ratio based panel cointegration test proposed by Larsson et al. (2001). A panel cointegration test has been conducted in order to take benefit of increased amount of LRiT (H(r)jH(2)) Country Bangladesh India Nepal Pakistan Sri Lanka LRNT E(Zk) Var (Zk) YLR-test r ¼ 0 Lag 1 1 4 1 1 16.732 10.917 17.417 4.9817 15.920 13.193 6.086 10.535 4.896 r ¼ 1 [0.031] [0.220] [0.024] [0.809] [0.042] 0.27830 2.0799 1.1980 1.4868 1.1778 1.244 1.137 2.212 0.161 Rank (ri) [0.598] [0.149] [0.274] [0.223] [0.278] 1 0 1 0 1 Notes: The values for E(Zk) and Var(Zk) are obtained from Larsson et al. (2001); VAR lag was selected using Schwartz-Bayesian criterion (SIC) with maximum lag ¼ 4 Table II. Panel cointegration test results JES 37,6 664 information embedded in a panel. The results of the panel cointegration test indicate that savings and investment are cointegrated in South Asia as a whole. With reference to the F-H puzzle, the panel test results indicate that the puzzle does not hold good for the South Asian panel under investigation. The findings may not be interpreted, however, to prove or disprove cross-boarder capital mobility for South Asian countries, as there are many other empirically established factors that account for high or low savings investment correlation in these developing countries. Further research may be initiated to specifically examine those other factors that might potentially explain the relationship between savings and investment in the South Asian nations. References Ang, J.B. (2007), “Are saving and investment cointegrated? The case of Malaysia (1965, 2003)”, Applied Economics, Vol. 39 No. 17, pp. 2167-74. Artis, M. and Bayoumi, T. (1989), “Saving, investment, financial integration, and the balance of payments”, Working Paper 89/102, International Monetary Fund, Washington, DC. Baltagi, B.H. (2005), Econometric Analysis of Panel Data, John Wiley & Sons Ltd, Chichester. Baxter, M. and Crucini, M.J. (1993), “Explaining saving-investment correlations”, American Economic Review, Vol. 83 No. 3, pp. 416-36. Breitung, J. (2000), “The local power of some unit root tests for panel data”, in Baltagi, B. (Ed.), Advances in Econometrics, Vol. 15: Nonstationary Panels, Panel Cointegration, and Dynamic Panels, JAI Press, Amsterdam, pp. 161-78. Choi, I. (2001), “Unit root tests for panel data”, Journal of International Money and Finance, Vol. 20 No. 2, pp. 249-72. Coakley, J. and Kulasi, F. (1997), “Cointegration of long span saving and investment”, Economic Letters, Vol. 54 No. 1, pp. 1-6. Coakley, J., Fuertes, A.M. and Spagnolo, F. (2004), “Is the Feldstein-Horioka puzzle history?”, The Manchester School, Vol. 72 No. 5, pp. 569-90. Coakley, J., Kulasi, F. and Smith, R. (1996), “Current account solvency and the Feldstein-Horioka puzzle”, Economic Journal, Vol. 106 No. 436, pp. 620-7. Coiteux, M. and Olivier, S. (2000), “The saving retention coefficient in the long run and in the short run: evidence from panel data”, Journal of International Money and Finance, Vol. 19 No. 4, pp. 535-48. Corbin, A. (2001), “Country specific effect in the Feldstein-Horioka paradox: a panel data analysis”, Economics Letters, Vol. 72 No. 3, pp. 297-302. De Vita, G. and Abott, A. (2002), “Are savings and investment cointegrated? An ARDL bounds testing approach”, Economics Letters, Vol. 77 No. 2, pp. 293-9. De Vita, G. and Abott, A. (2003), “Another piece in the Feldstein-Horioka puzzle”, Scottish Journal of Political Economy, Vol. 50 No. 1, pp. 69-87. Dooley, M., Frankel, J. and Mathison, D.J. (1987), “International capital mobility: what do saving-investment correlations tell us?”, IMF Staff Papers, Vol. 34 No. 3, pp. 503-29. Feldstein, M. (1983), “Domestic saving and international capital movements in the long run and the short run”, European Economic Review, Vol. 21, pp. 129-51. Feldstein, M. and Horioka, C. (1980), “Domestic saving and international capital flows”, Economic Journal, Vol. 90 No. 358, pp. 314-29. Georgopoulos, G. and Hejazi, W. (2009), “The Feldstein-Horioka puzzle revisited – the home bias much less?”, International Review of Economics and Finance, Vol. 18, pp. 341-50. Hadri, K. (2000), “Testing for stationarity in heterogeneous panel data”, Econometric Journal, Vol. 3 No. 2, pp. 148-61. Ho, T.W. (2002), “The Feldstein-Horioka puzzle revisited”, Journal of International Money and Finance, Vol. 21 No. 4, pp. 555-64. Hoffmann, M. (1998), “Long run capital flows and the Feldstein-Horioka puzzle: a new meansure of international capital mobility and some historical evidence from Great Britain and the United States”, European University Institute urn: hdl: 1814/675, European University Institute. Im, K.S., Pesaran, M.H. and Shin, Y. (2003), “Testing for unit roots in heterogeneous panels”, Journal of Econometrics, Vol. 115 No. 1, pp. 53-74. Jansen, W.J. (2000), “International capital mobility: evidence from panel data”, Journal of International Money and Finance, Vol. 19, pp. 507-11. Karaman Örsal, D.D. (2008), “Comparison of panel cointegration tests”, Economics Bulletin, Vol. 3 No. 6, pp. 1-20. Kasuga, H. (2004), “Saving-investment correlations in developing countries”, Economics Letters, Vol. 83 No. 3, pp. 371-6. Kim, S.H. (2001), “The saving-investment correlation puzzle is still a puzzle”, Journal of International Money and Finance, Vol. 20 No. 7, pp. 1017-34. Kollias, C., Nikolaos, M. and Paleologou, S-M. (2008), “The Feldstein-Horioka puzzle across EU members: evidence from the ARDL bounds approach and panel data”, International Review of Economics and Finance, Vol. 17 No. 3, pp. 380-7. Kool, J.M. and Keijzer, M. (2009), “International capital mobility: linking the Feldstein-Horioka puzzle to the trade and equity home bias puzzles”, Cambridge Journal of Regions, Economy and Society, Vol. 2 No. 2, pp. 211-27. Larsson, R., Lyhagen, J. and Lothgren, M. (2001), “Likelihood-based cointegration tests in heterogeneous panels”, Econometrics Journal, Vol. 4 No. 1, pp. 109-42. Levin, A., Lin, C.F. and Chu, C. (2002), “Unit root tests in panel data: asymptotic and finite-sample properties”, Journal of Econometrics, Vol. 108 No. 1, pp. 1-24. Maddala, G.S. and Wu, S. (1999), “A comparative study of unit root tests with panel data and a new simple test”, Oxford Bulletin of Economics and Statistics, Vol. 61, S1, pp. 631-52. Miller, S.M. (1988), “Are saving and investment co-integrated?”, Economics Letters, Vol. 27 No. 1, pp. 31-4. Murthy, N.R.V. (2007), “Feldstein-Horioka puzzle in Latin American and Caribbean countries: evidence from likelihood-based cointegration tests in heterogeneous panels”, International Research Journal of Finance and Economics, Vol. 11, pp. 111-21. Murthy, N.R.V. (2009), “The Feldstein-Horioka puzzle in Latin American and Caribbean countries: a panel cointegration analysis”, Journal of Economics and Finance, Vol. 33 No. 2, pp. 176-88. Narayan, P.K. (2005), “The saving and investment nexus for China: evidence from cointegration tests”, Applied Economics, Vol. 37 No. 17, pp. 1979-90. Obstfeld, M. (1986), “Capital mobility in the world economy: theory and measurement”, Carnegie-Rochester Conference Series in Public Policy, Vol. 24, pp. 55-104. Obstfeld, M. and Rogoff, K. (2000), “The six major puzzles in international macroeconomics: is there a common cause?”, NBER Working Paper No. 7777, Cambridge, MA. Penati, A. and Dooley, M. (1984), “Current account imbalances and capital formation in industrial countries 1949-1981”, International Monetary Fund Staff Papers, Vol. 31, pp. 1-14. Savings and investment in South Asia 665 JES 37,6 Salahuddin, M. and Islam, R. (2008), “Factors affecting investment in developing countries: a panel data study”, The Journal of Developing Areas, Vol. 42 No. 1, pp. 21-37. Schmidt, M.B. (2003), “Savings and investment in Australia”, Applied Economics, Vol. 35 No. 1, pp. 99-106. Summers, L.H. (1988), “Postwar developments in business cycle theory: a moderately classical perspective: comment”, Journal of Money, Credit and Banking, Vol. 20 No. 3, pp. 472-6. Tesar, L. (1991), “Savings, investment and international capital flows”, Journal of International Economics, Vol. 31 Nos 1-2, pp. 55-78. Vos, R. (1998), “Savings, investment and foreign capital flows: have capital markets become more integrated?”, Journal of Development Studies, Vol. 24 No. 3, pp. 310-34. 666 Further reading Bayoumi, T. (1990), “Saving-investment correlations: immobile capital, Government policy on endogenous behaviour?”, IMF Staff Papers, Vol. 37 No. 2, pp. 360-87. Blanchard, O. and Giavazzi, F. (2002), “Current account deficits in the euro area: the end of the Feldstein-Horioka puzzle?”, Brookings Papers on Economic Activity, Vol. 2, pp. 147-86. De Haan, J. and Siermann, C.L.J. (1994), “Saving, investment and capital mobility: a comment on Leachman”, Open Economies Review, Vol. 5 No. 1, pp. 5-17. Elliott, G., Rothenberg, T.J. and Stock, J.H. (1996), “Efficient tests for an autoregressive unit root”, Econometrica, Vol. 64 No. 4, pp. 813-36. Engle, R.F. and Granger, C.W.J. (1987), “Co-integration and error correction: representation, estimation, and testing”, Econometrica, Vol. 55 No. 2, pp. 251-76. Georgopoulos, G. and Hejazi, W. (2005), “Feldstein-Horioka meets a time trend”, Economics Letters, Vol. 86 No. 3, pp. 353-7. Harvey, A. and Collier, P. (1977), “Testing for functional misspecification in regression analysis”, Journal of Econometrics, Vol. 6 No. 1, pp. 103-19. Hoffman, M. (2004), “International capital mobility in the long run and the short run: can we still learn from saving-investment data?”, Journal of International Money and Finance, Vol. 23 No. 1, pp. 113-31. Jiranyakul, K. and Brahmasrene, T. (2008), “Cointegration between investment and saving in selected Southeast Asian countries: ARDL bounds testing procedure”, unpublished report, Purdue University, West Lafayette, IN and National Institute of Development Administration. Pesaran, M.H., Shin, Y. and Smith, R.J. (2001), “Bounds testing approaches to the analysis of level relationships”, Journal of Applied Econometrics, Vol. 16 No. 3, pp. 289-326. World Bank (2008), World Bank Development Indicators CD-ROM, World Bank, Washington, DC. To purchase reprints of this article please e-mail: reprints@emeraldinsight.com Or visit our web site for further details: www.emeraldinsight.com/reprints View publication stats