Savings and Economic Growth Nexus in a Small Opening Economy: The Case of Financial Repression versus Financial Liberalization Episodes in Nigeria

Transcription

1 IJE : Volume 8 Number 1 June 2014, pp Savings and Economic Growth Nexus in a Small Opening Economy: The Case of Financial Repression versus Financial Liberalization Episodes in Nigeria Kazeem Bello Ajide * Abstract: The study seeks to provide empirical evidence on the relationship between savings and economic growth under the two financial regime episodes namely: financial repression and liberalization in Nigeria. The Granger causality test and ARDL methodology was used to explore such relationship over the period 1970 to The bivariate causality showed a unidirectional relationship between the two while the error correction term of for repression and for liberalization indicated that the latter reverted back to long run equilibrium quickly before the former thus validating McKinnon-Shaw Hypotheses. A few relatable suggestions include: due recognition should be accorded to savings mobilization, provision of good infrastructural facilities, discouraging unfettered profligate spending by government and impediments that may slow down the full implementation of reforms policies should be removed. Keywords: Financial repression, Financial Liberalization, Savings, Economic Growth JEL Classification: G38, E21, O11, 1. INTRODUCTION The traditional growth theories have identified savings as one of the effective transmission channels for achieving the mantra sustainable economic growth mostly sought by most economies. It has been asserted that countries that save more also tend to grow faster. The direction of causality between the two however, has generated intense theoretical and empirical debate in the literature. The impact of savings on growth, albeit indirect, mostly occurs through the intervening link of productive investment goods. The intermediating role of financial institutions, to a very large extent, have been critical in this regard by channelling scarce financial resources from surplus (savers) to deficit (borrowers) units in return for an agreed rate of interest. The interest rate, a link, between savings and investment decisions presents a fundamental point of departure between financial repression and liberalization theses. Conceptually, different conclusions are arrived at, in saving-growth nexus when the two opposing views are upheld. While it was argued that financial repression promotes shallowness of financial system since it is characterized by low or falling real monetary aggregates, low levels of national income and wealth and often negative real interest rates, which eventually retards economic growth. Financial liberalization, on the other hand, directly reverses the features of repression by driving the interest rates higher, thereby encouraging more savings and * Department of Economics, University of Lagos, Lagos, Nigeria, kazeemjide@gmail.com

2 64 Kazeem Bello Ajide subsequently promotes economic growth. 1 Despite the benefits that are advanced as likely to be derived from financial liberalization, it has also been criticized as causing problems of increased cost of borrowing, increased non-performing loans, financial instability, economic recession, even at the extreme, attract surge of capital inflows which further exacerbates macroeconomic instability. Nonetheless, the two financial episodes have found profound applications in most industrialized and developing economies alike, dictated largely by economic imperatives and desires for change in ideological paradigms. Nigeria, just like other country within the sub- Saharan region has a history of financial repression. However, in 1986 structural adjustment programme was adopted as part of a global trend towards economic liberalization. The country therefore undertook measures to liberalize its financial systems by moving towards marketdetermined interest rates, trade and foreign exchange markets liberalization coupled with changes in the statutes governing the financial sector as well as adoption of an indirect instrument of monetary policy. These changes in policy provide an opportunity to assess the issues involved in the debate about financial repression and liberalization most especially within the context of savings-growth relationship. This study therefore is basically concerned with assessing the impact of savings on economic growth under the episodes of financial repression and liberalization theses within the Nigerian context. Though extensive empirical studies has been conducted on savings-growth nexus but majority of these studies focused mainly on the impact, nature and direction of causality between the two (examples of such studies include Caroll and Weil, 1994; Sinha, 1996; Anoruo and Ahmad, 2001; Mavrotal and Kelly, 2001; Baharumshah et al., 2003) and are cross-country in nature 2 thus providing an additional impetus making gravitation towards country-specific cases experiences imperative. This paper is important for a number of reasons. First, this study improves on the previous studies by accounting for the structural break in the data set by dividing the study period into pre and post reforms as well as giving consideration to the entire coverage period, thus ensuring more robust conclusive evidence. Second, the study employs an advanced econometric methodology of bound testing approach. The use of this methodology presupposes that the underlying regressors are all integrated of order one (Pesaran et al., 2001). This is necessary because in the presence of a mixture of stationary series and series containing a unit root, standard statistical inference based on conventional likelihood ratio tests is no longer valid. Third, the study also establishes the direction of causality between savings and economic growth as a way of revalidating or refuting the contentious results earlier established for Nigeria. The rest of the paper is organized as follows: Section 2 reviews the relevant literature on saving and economic growth nexus, Section 3 attempts a terse presentation of the stylized fact about savings-growth relationship in the Nigerian context while section 4 presents the methodology and estimation techniques. Section 5 analyses the estimated savings models. Section 6 concludes with a summary of major findings, policy implications of results and major recommendations. 2. LITERATURE REVIEW The section presents a short synopsis of theoretical and empirical review on savings-growth nexus so as to provide a context for subsequent discussions.

3 Savings and Economic Growth Nexus in a Small Opening Economy: 65 One of the fundamental economic issues that have received widespread attention as well as diverse interpretations in the economic literature centers on what drives economic growth? To date, myriad of reasons have been advanced as probable factors, but somewhat research findings remains inconclusive. The importance of savings however, has been stressed extensively in the literature as one of the potent conduits of stimulating economic growth. To date, both traditional economic theorists and empiricists are divided on the direction of causation between savings and economic growth and so also that different positions are upheld about saving and growth relationship under financial repression and liberalization episodes. On the theoretical ground, growth models have clearly delineated on the role of savings in the growth process. According to Solow- Swan model, the saving rate influences only steadystate and impact on growth rates of output only temporarily. That is, a change in the saving rate changes the economy s balanced growth path and hence per capita output in steady state, but it does not affect the growth rate of output per worker on the balanced growth path. In the alternative approach by Romer, the impact of the saving rate is not on steady-state output, but on the growth rate of output directly. In his model, technology is endogenised; hence, an increase in the saving rate not only increases per capita output in steady state but also increases the growth rate of per capita output. On financial episodes front, McKinnon (1973) and Shaw (1973) maintained and advocated the existence of complementarity effect between savings and investment because the former is believed to be channeled into the latter regardless of whether saving goes either into the accumulation of money balances or the accumulation of physical capital. Thus, capital accumulation or physical investment is therefore seen as the proximate source of economic growth. In the light of the foregoing, they offered support for financial liberalization thesis. Financial repression, on the other hand, however, argues that savings are not necessarily channelled into investment and that the development of a monetary sector could be damaging. With the introduction of money balances, agents face the choice of allocating resources not used for consumption either to the purchase of physical capital, or to money balances. Since it is physical investment that is the source of economic growth, if money balances are not made available for investment, but rather held as a stock of purchasing power, the equilibrium growth path of an economy will occur at a lower level of per capita output than before. Also, in terms of direction of causation between savings and economic growth, the argument is deeply rooted in classicalism and keynesianism schools of thought. The former posits that the direction of causation runs from savings to investment and finally to growth since savings is a pre-requisite for investment. The latter on the other hand, emphasized the importance of profit prospects and the elastic supply of credit to the private sector rather than prior savings. Unarguably, savings-growth relationship has provoked intense theoretical discourse among researchers and policy makers alike, thus injecting a large dosage of empirical studies into the pool of savings literature. The cubicle of knowledge on savings-growth nexus that has received wide empirical assessment to date can be aptly partitioned into the investigation of the direction of causality, causal relationship and cross-country case studies. On the basis of the foregoing, therefore, a terse presentation of the empirical evidence on the savings-growth nexus is forewarned in what follows.

4 66 Kazeem Bello Ajide Broadly, three strands of empirical literature can be sieved from a pool of studies on savingsgrowth relationship. First, are those that believed that the relationship runs from savings to growth and second, supports the reverse causality running from growth to savings. While the third empirically established mixed results. Belonging to the first are authors like Bacha (1990), Otani and Villanueva (1990), DeGregorio (1992), and Jappelli and Pagano (1994), Romm,(2003), Alguacil, Candros and Ort, (2004), who found empirical evidence to support that higher savings rate led to higher economic growth. The inclusion of investment variable in the savings model by Krieckhaus (2002) also lent credence to supporting savings running to growth. Included in the second category are Sinha and Sinha, (1998), Bhaharumshah et al., (2003) and Abebiyi, (2005), they also were able to establish the causal link running from economic growth to higher savings. The findings of the last group of researchers are largely mixed and these include Carol and Weil, (1994), Anoruo and Ahmadi, (2001) and Mohan, (2006). Further, majority of these studies are cross-country in nature. For instance, Caroll and Weil (1994) conducted five year averages for OECD countries; Sinha and Sinha (1998, 1999) considered both Mexico and Sri lanka; Bhaharumshah et al. (2003) did use Singapore, South Korea, Malaysia, Philippines and Thailand; Mohan (2006) determine whether the direction of causality in savings-growth based on income class namely: low income, low middle income, upper middle income, and high income countries respectively where he used 25 countries altogether. Also, in the Nigerian context, though, quite substantial studies have been conducted on savings-related issues but majority of the studies mostly focused on determinants of private savings (Nwachukwu and Egwakhide, 2007), direction of causality (Adebiyi, 2005; Abu, 2010); private saving-growth nexus (Johnson, 2011) but to mention a few. Variations also exist in terms of methodologies that have been adopted by previous studies conducted so far in estimating the relationship between savings-growth. Such methods include: granger causality, ( Sinha,1996; Mavrotas and Kelly 2001); least square method, (Bacha,1990,; Otani and Villanueva,1990; DeGregorio, 1992; and Jappelli and Pagano, 1994; panel data analysis, error correction modeling approach and VAR (Adebiyi, 2005), VEC (Anoruo and Ahmad, 2001). Thus, given the theoretical and empirical ambiguities surrounding the argument, it would perhaps be unrealistic to assume that clear-cut conclusions will emerge from Nigeria s experiences as it relates to savings-growth nexus under financial liberalization and repression periods. From a brief literature review, it is abundantly clear that, this study improves on previous studies on a number of fronts which include (i) according consideration to financial repression and liberalization episodes using Nigerian experience (ii) advancing a more robust econometric approach (iii) adding to empirical literature on country-specific studies which are scarcely available to date. 3. STYLIZED FACTS: SAVINGS AND ECONOMIC GROWTH IN NIGERIA Table 1. depicts declining trends in the annual average growth rates of real GDP beginning from with 5.78% but later dropped to 4.05% during 1976 to Between 1981 and 1985, the percentage annual average growth rates further plunged into negative value of This can be explained, in part by the unfettered government profligate spending of national

5 Savings and Economic Growth Nexus in a Small Opening Economy: 67 resources. Similar trends are observed in savings during the same periods except that no negative value was recorded. Table 1 Savings and Economic Growth (Bidecadal averages, ) Annual Averages Growth Rates Growth Rates Deposit Rates of Real GDP of Savings * 2.8 Source: Authors Calculation from CBN Statistical Bulletin, Note*: Moving averages were used in generating savings values for 2009 and 2010 due to non-availability A major contributory factor during this period to these developments can be likened to financial repression which fully accommodated government overbearing function in the economy. Infact, all features that characterizes repressive regime like negative interest rates, low levels of national income, imposition of interest-rate ceilings, directed credit allocation policies and high reserve requirements and foreign exchange regulations are fully operational during the period. The country heaves a sigh of relief when structural adjustment programme was adopted in 1986 and brought in its wake several policy reforms of which austerity measures form the key component part. One spectacular dimension of this was observed when the annual average of real GDP rose substantially to 5.42% between 1986 and 1990 but this could not be sustained as it later fell precipitously to 2.5% over the period, 1991 to Upward trends are continued to be maintained from 1996 up to Saving, on the other hand, also picked up during the SAP period and rose to an all high increase of 31.0% between 1991 and 1995 and from this period; downward trends continue to be experienced, though marginally. It was also observed from the table that for the most part of the 1970s, the deposit rates were never up to 5% thus suggesting financial repressive nature of the economy. However, between 1981 and 1985 deposit rates rebounded from its sluggish level of 4.6% to 8.0% and the upward trends continued into the mid 1990s thus suggesting the application of the reforms measures embedded in the structural adjustment program adopted in the late 1980s. But beginning from late 1990s up to 2010, there have been a consistent decline in the rates. 4. EMPIRICAL MODEL, DATA AND ECONOMETRIC METHODOLOGY The study is anchored on a simplified Solow growth model that relates income (output) as a function of savings. He argued that higher savings preceded economic growth. Thus, we specify growth model with some modifications to allow for the inclusion of some intervening variables as follows G = f(s, other var) (1)

6 68 Kazeem Bello Ajide Where G stands for economic growth and S for savings while othervar represents other intervening variables. The intervening variables considered are both domestic investment and foreign direct investment since it is believed that the impact of savings on growth usually work through capital formation of investment expenditure. Thus, equation (1) is rewritten as: LNGDP 0 1LNSAV 2LNINV 3LNFDI t (2) LNGDP = Logarithm of Real gross domestic product LNSAV = Logarithm of National savings LNINV = Logarithm of Gross fixed capital formation LNFDI = Logarithm of Foreign direct investment a Also, 1 3 are coefficients of estimated variables and t is the error term. and 1, 2, 3 > 0. All the variables were log transformed so that the problem of heteroskedasticity can be reduced since it compresses the scale in which the variables are measured, thereby reducing a tenfold difference between two values to a twofold difference (Gujarati, 1995). While some variables were not log transformed because they are already in their rates examples include inflation and real exchange rate Estimation Technique The study adopts an Auto-Regressive Distributed Lag (ARDL) bounds testing approach developed by Pesaran et al. (2001) to modelling the long run determinants of domestic private investment. This approach has some econometric advantages over the Engle-Granger (1987) and maximum likelihood based approach proposed by Johansen and Juselius (1990) and Johansen (1991) cointegration techniques. Firstly, the bounds test does not require pre-testing of the series to determine their order of integration since the test can be conducted regardless of whether they are purely I(1), purely I(0), or fractionally integrated. Second, endogeneity problems and inability to test hypotheses on the estimated coefficients in the long-run associated with the Engle-Granger (1987) method are avoided. According to Pesaran and Shin (1999), modeling the ARDL with the appropriate lags will correct for both serial correlation and endogeneity problems. Jalil et al. (2008) argue that endogeneity is less of a problem if the estimated ARDL model is free of serial correlation. In this approach, all the variables are assumed to be endogenous and the long run and short run parameters of the model are estimated simultaneously (Khan el al., 2005).Third, as argued in Narayan (2005), the small sample properties of the bounds testing approach are far superior to that of multivariate cointegration (Halicioglu, 2007). The approach, therefore, modifies the Auto-Regressive Distributed Lag (ARDL) framework while overcoming the inadequacies associated with the presence of a mixture of I(0) and I(1) regressors in a Johansen-type framework. Four, Secondly, the long and short-run parameters of the model in question are estimated simultaneously. Lastly, The ARDL has superior small sample properties as compared to the Johansen and Juselius (1990) cointegration test (Pesaran and Shin, 1998). An ARDL representation of equation (1) can be specified as follows: q q q q LNRGDP LNRGDP LNSAV LNINV LNFDI 0 1 t i 2 t i 3 t i 4 t i i 1 i 1 i 1 i 1

7 Savings and Economic Growth Nexus in a Small Opening Economy: 69 LNRGDP LNSAV LNINV LNFDI (3) 1 t 1 2 t 1 3 t 1 4 t 1 t Where is the first difference of a variable. LN represents a natural logarithms 0 is a constant q is a maximum lag order, 1,... 4 represent the short-run coefficients (error correction dynamic), 1,... 4 correspond to the long-run relationship, i time trend, and, µ t is the white noise error. The implementation of the ARDL approach involves two stages. First, the existence of the long-run nexus (cointegration) between variables under investigation is tested by computing the F-statistics for analyzing the significance of the lagged levels of the variables. (Pesaran, Shin, and Smith, 1999) and (Narayan, 2004) have provided two sets of appropriate critical values for different numbers of regressors (variables). This model contains an intercept or trend or both. One set assumes that all the variables in the ARDL model are of I(0), and another assumes that all the variables are I(1). If the F-statistic lies above the upper-bound critical value for a given significance level, the conclusion is that there is a non-spurious long-run level relationship with the dependent variable. If the F-statistic lies below the lower bound critical value, the conclusion is that there is no long-run level relationship with the dependent variable. If it lies between the lower and the upper limits, the result is inconclusive. The general form of the null and alternative hypotheses for the F-statistic test is as follows: H : 0 0 LNRGDP LNSAV LNINV LNFDI H : 0 1 LNRGDP LNSAV LNINV LNFDI Secondly, if the cointegration between variables is identified, then one can undertake further analysis of long-run and short-run (error correction) relationship between the variables. The error correction representation of the series can be specified as follows: q q q q LNRGDP LNRGDP LNSAV LNINV LNFDI ECM 0 1 t i 2 t i 3 t i 4 t i t 1 t i 1 i 1 i 1 i 1 Where the speed of adjustment process and ECM is the residuals obtained from equation (1) while other variables remain as earlier defined. The coefficient of the lagged error term correction term is expected to be negative and statistically significant to further support the existence of a cointegrating relationship Data Source The study employs secondary sources of data spanning 1970 through 2010 where four keys variables are used in estimating savings-investment relationship. The choice of the study period is limited by the data availability consideration on most of the variables used in the study. The data is drawn mainly from the Central Bank of Nigeria Statistical Bulletin,2010 and Central Bank Annual Reports (various issues).

8 70 Kazeem Bello Ajide 5. DISCUSSION OF RESULTS Unit root tests were conducted for all variables in the model and the results are presented in table.1. Both the Augmented Dickey-Fuller (ADF) and Phillips Perron (PP) unit root tests were conducted (including both the intercept without trend as well as with intercept and trend respectively) and both tests reject the null hypothesis of no unit root in all the variables except for foreign direct investment and real GDP that were found to be significant at levels in both ADF and PP tests. All other variables were found to be significant at first differences thus suggesting they are integrated of order 1. Table 1 Unit Root Tests Variable Augmented Dickey Fuller Phillip Perron Intercept Without Intercept with Intercept Without Intercept With Trend Trend Trend Trend LNRGDP *** LNGDP *** *** *** LNSAV LNSAV *** *** *** *** LNINV LNINV *** *** *** *** LNFDI ** LNFDI *** *** *** Note: (***)**(*) indicates significant at the 1%, 5% and 10% levels. The mere fact that some variables are of order zeros, makes it difficult to apply Engle- Granger and Johansen cointegration techniques. This is because both techniques require all variables to be integrated of order 1 before cointegration tests can be applied. An alternative approach of Autogressive Distributed Lag Framework (ARDL) does not impose such restrictions. This approach allows for the inclusion of variables integrated of order 0 and 1 in the cointegration equation. Having established the presence of variables of order 0 and 1in the model further provides a justification for applying this method for the study Granger Causality Test Results This test is normally performed in order to measure the linear causation between savings and economic growth. It is basically concerned with the use of past information in a variable to be able to predict the value of the other. The study applies causality test developed by Granger (1969). The study therefore specifies pairwise causality test of the form: 1 2 (6) SAV SAV RGDP t 0 1i t 1 2i t 1 1t i 1 i 1 m1 m2 (7) RGDP RGDP SAV t 0 1i t 1 2i t 1 2t i 1 i 1

9 Savings and Economic Growth Nexus in a Small Opening Economy: 71 Where 1, 2, m 1 and m 2 are the optimal lag length, 1t and 2t are white noise error terms which are identically and independently normally distributed with mean zeros and constant variance. The Granger causality test statistics results are presented in Table.2. The results show that the null hypothesis that savings does not granger cause real GDP can be rejected thus implying that savings does cause real GDP. The same argument fails to hold from real GDP to savings. By implication the relationship between them is unidirectional in nature with causality running from savings to growth only. This result is consistent with that of Olajide (2009) for Nigeria. Table 2 Bivariate Granger Causality Test Results Sample: Null Hypothesis F-Statistics P-Value SAV does not granger cause RGDP RGDP does not granger cause SAV Cointegration tests presented in table.2 shows whether a long run relationship exist between the variables by computing the F-statistic for the joint significance of lagged levels of variables. We find evidence of cointegration between real GDP and savings as well as other control variables. Specifically, the F-statistics falls above the critical values of 99 percent significance levels in all financial regimes. Therefore, we can conclude that a long run relationship exists and we can proceed to obtaining long run coefficients. Financial Regimes Table 3 F-Statistics for Cointegration Tests F-Statistics Financial Repressed Period ( ) *** Financial Liberalization Period ( ) *** Entire Period ( ) *** Note: The critical value bounds are from Table F in Pesaran and Pesaran (1997) (with an intercept and no trend). They are at the 90% significance level at the 95% significance level and at the 99% significance level. Explanatory Variables Table 4 Long Run Coefficients from ARDL Estimation of Growth Equations Dependent Variable: LN REAL GDP Financial Repressed Financial Liberalization Entire period: Period: Period: Constant (0.7110) (6.5527)*** (8.7847)*** LNSAV (1.8974)* (4.5126)*** (4.0929)*** LNINV ( ) (2.1669)** ( )** LNFDI (0.3726) (0.9503) (0.6875) Note : (a). (***)**(*) indicates significant at the 1%, 5% and 10% levels. (b) Figures in parenthesis ( ) are t-ratios.

10 72 Kazeem Bello Ajide From the table, we observed that all the explanatory variables conformed with the a priori hypothesized signs except for investment which had negative signs during the financially repressed era but carries the expected positive sign during financially-reformed period. In sum, the overall effects of the negativity of the repression far overshadowed that of the reform period thus making the entire coverage period being negative. Of all the variables in the financially repressed model, savings is the only variable that is positive and statistically significant at a 10% level thus contradicting the McKinnon-Shaw belief that savings tend to fall during the period due to the negative real interest rates. Thus, a 1% increase in savings rate tends to increase economic growth by 1.09 percent point. The result however, find evidence to support McKinnon-Shaw Liberalization Theses which states that savings generally tends to increase more during liberalized periods should there exist positive interest rates. From the table while savings appears significant at 1% during liberalization, but only significant at 10% level during the period of repression. A study by Okpara (2010) for Nigeria corroborates the usefulness of liberalization era over repression in Nigeria. Also, investment variable fails to conform with a priori expectation and also not statistically significant at any conventional levels. Foreign direct investment though carries positive sign but insignificant as well. Most FDI attracted into the country have discriminately been channelled towards extractive industries whose extent of linkages with rest of the sectors in the economy has been narrower and sowehat sectionalized. Different pictures emerge during the financially- liberalized era ( ) in which all the explanatory variables except the FDI were statistically significant at different levels. Infact, a 1% increase in both savings and investment tend to increase the level of economic activities by 1.12 and 0.10 percentage points respectively. Similar pattern in variables were also observed for the entire coverage period, except that investment was carrying negative value. The plausible rationale behind the negative sign of the investment may not be unconnected to dilapidated nature of the public investment goods like bad roads, epileptic power supply, poor health care delivery and a host of retardatory factors. In order to see the short run dynamics, the estimates of the error correction model are presented in table 5. Just like what was observed in the long run, we equally observed that some variables like real GDP, Savings and investment contradict theoretical hypothesized signs except for FDI whose sign conform with expectation. Interestingly however, is that majority of the variables are statistically significant in the short run. The previous value of real GDP is positive and significantly related to current changes in the value of real GDP prior to the liberalized period. The situation however, turns sour during the reform era simply because a large scale of corruption and mismanagement of public funds was perpetrated by the then administrators thus contributing to the poor performance of the economy. The effects of such profligate spending on the part of the government during the reform era largely overwhelm that of the repressive period. The current savings has the expected sign that is positive and significantly impacted on the growth of real GDP in the short run just like what occurs in the long run. It also has a positive impact at the period when the country adopted reform. This implies that savings does achieve the purpose for which it was partly adopted. One of the plausible reasons for this may be likened to increase in the number of financial institutions in the country coupled with the innovation

11 Savings and Economic Growth Nexus in a Small Opening Economy: 73 Explanatory Variables Table 5 Error Correction Representation for ARDL Model of Growth Equations Dependent Variable: LN REAL GDP Financial Repressed Financial Liberalization Entire period: Period: Period: Constant (0.1932) (1.9415) (0.3180) (LNRGDP(-1)) (3.5398)*** ( )*** ( )*** (LNSAV) (2.0847)** (2.1931)** (2.1665)** (LNSAV(-1)) (1.1063) (0.9634) (0.2104) (LNINV) ( )** ( )** ( )** (LNINV(-1)) ( )** ( ) ( ) (LNFDI) (0.2152) (1.9121)* (1.8987)* (LNFDI(-1)) (0.2134) (0.3287) (0.2141) ECM(-1) ( )** ( )*** ( )*** R ADJ R F [0.0000] [0.0000] [0.0000] Note : (a). (***)**(*) indicates significant at the 1%, 5% and 10% levels. (b) Figures in parenthesis ( ) are t- ratios. introduced in their financial product lines. In addition, the lag values of savings followed the same pattern in terms of the hypothesized signs in all the regimes but none appears significant. Investment variables, albeit significant, but have contradictory theoretical signs. This further reinforces the earlier assertions that decaying infrastructural development in the country may have been responsible for this. The FDI significantly impacted on the growth of real GDP but this occurs during the period of reforms. This may possibly be explained in terms of free allowance granted to foreign participants in the economy. Unlike pre- liberalization era where indigenization policy was fully operational thereby limiting the extent of foreign participation in the economy. The error correction terms are negatives in all the estimated equations and they are statistically significant, thus corroborating the results of the cointegration tests of the existence of a stable long run relationship between the variables. The error correction term is and respectively indicating that 15%, 31% and 28% of the previous year s deviations from the long run equilibrium values will be restored within one year. Interestingly, financial liberalization era reverted back to long run equilibrium before financially- repressed era as indicated by their error term values. Finally, the structural stability of the long-run and short-run relationships of the ARDL model for the entire period is examined by the cumulative sum (CUSUM) and the cumulative sum of squares (CUSUMSQ) of the recursive residual test which proposed by (Brown et al,

12 74 Kazeem Bello Ajide 1975). The null hypothesis of these tests is that the regression equation is correctly specified. These two tests are presented in figure 1 and 2. The pair of straight lines is each figure indicates the 5 per cent significant level and if the plotted CUSUM and SUSUMSQ graphs remain inside the straight lines the null hypothesis of correct specification of the model can be accepted. Otherwise, the null hypothesis is rejected and it can be concluded that the regression equation is miss-specified. The two figures reveal that the plots of CUSUM and SUSUMSQ stay within the lines, and, therefore, this confirm the equation 1 is correctly specified and stable. Also, since we are interesting in the relationship between savings and growth under different financial episodes, it would be useful to know if there are structural breaks in the equation parameters. Accordingly, Chow s breakpoint test has been conducted using 1986 has the breakpoint year since it was a year in which considerable changes were effected into the Nigerian economy. The results show that there are breakpoints thus rejecting null hypothesis of no break in 1986 as indicated by all statistics. By implication, a breakpoint exists in equation 1 in Table 6 Chows Breakpoint Test F-ststistic Prob (6, 28) Log likelihood ratio Prob (6) Wald Statistics Prob (6) Source: Computed CUSUM 5% Significance Figure 1: Plotting of CUSUM Statistics for Stability Test

13 Savings and Economic Growth Nexus in a Small Opening Economy: CUSUM of Squares 5% Significance Figure 2: Plotting of CUSUMSQ Statistics for Stability Test 6. CONCLUSION AND POLICY MESSAGES The study empirically assesses the impact of savings on economic growth in Nigeria under two different financial regimes episode namely repression and liberalization. The outcome of granger causality test revealed a unidirectional relationship running from savings to economic growth thus suggesting that savings granger causes economic growth. This result found evidence to support the earlier studies by like Olajide (2009) for Nigeria that showed the causality running from savings to economic growth. Interesting also, were the results of ARDL models that evidently showed that savings both in the long and short runs had positive and significant impacts but found to be more pronounced under financial liberalization than repression thus validating McKinnon-Shaw liberalization theses position that savings is more likely to impact on economic growth under liberalization than repression owing largely to positive interest rates. The policy messages that emanate from the findings are as follows: (i) government should accord due recognition to savings mobilization as one of the major means of stimulating economic growth (ii) more enabling investment climate as well as good infrastructural facilities should be provided as poor and decaying infrastructural development has negatively impacted on economic growth (iii) unfettered profligate spending of government should be curtailed so as to allow such resources to be diverted to the productive sector of the economy and (iv) the likely impediments (like undue bureaucratic delays) which may not allow the full benefits of reforms realizable should be removed.

14 76 Kazeem Bello Ajide References Abu, N. (2010), Saving-Economic Growth Nexus In Nigeria, : Granger Causality And Co- Integration Analyses. Volume 3, Issue 1, pp , June Review of Business and Economics Studies. Adebiyi, M.A. (2005), Saving-Growth Relationship in Nigeria: An Empirical Evidence, African Review of Money, Finance and Banking, Alguacil, M., A. Cuadros, and V. Orts, (2004), Does Saving really matter for Growth? Mexico ( ). Journal of International Development, March, 16, 2: Anoruo, E and Y. Ahmad (2001), Causal Relationship between Domestic Savings and Economic Growth: Evidence from Seven African Countries, African Development Bank, Blackwell publisher, Oxford. Bacha, E.L. (1990), A Three-Gap Model of Foreign Transfers and the GDP Growth Rate in Developing Countries, Journal of Development Economics, Vol. 32, Baharumshahl et al. (2003), Savings Dynamic in Asian Countries, Journal of Asian Economics, Vol. 13: Carroll, C. D., and D. N. Weil (1994), Saving and Growth: A Reinterpretation, Carnegie-Rochester Conference Series on Public Policy, 40, DeGregorio, J. (1992), Economic Growth in Latin America, Journal of Development Economics, Vol. 39: Engle, R.F., and C.W.J. Granger, (1987), Cointegration and Error Correction: Representation,Estimation and Testing, Econometrica 55: Gujarati, D. N., (1995), Basic Econometrics, 3rd ed., New York: McGraw-Hill. Japelli, T. and M. Pagano, (1994), Savings, Growth and Liquidity Constraints, Quarterly Journal of Economics, 109: pp Johansen, S and K. Juselius (1990), Maximum Likelihood Estimation and Inference on Cointegration with Applications to Demand for Money. Oxford Bulletin of Economics and Statisticis, Vol. 52, May, Krieckhaus, J., (2002), Reconceptualizing the Developmental State: Public Savings and Economic Growth, World Development Vol. 30 No. 10: Mavrotas, G. and R. Kelly. (2001), Old Wine in New Bottles testing Causality between Savings and Growth, The Manchester School, Vol. 69: McKinnon, R. I. (1973), Money and Capital in Economic Development, Brookings Institution. Mohan, R. (2006), Causal Relationship Between Savings and Economic Growth in Countries with Different Income Levels, Economics Bulletin, Vol. 5, No. 3 pp Narayan, P. K., (2004), Reformulating Critical Values for the Bounds F-statistics Approach to Cointegration: An Application to the Tourism Demand Model for Fiji, Department of Economics, Discussion Papers, No. 02/04, University of Monash. Olajide S. O. (2009), Does Saving Really Matter for Growth in Developing Countries? The Case of a Small Open Economy, Cited online at Otani, I and D. Villannueva, (1990), Long Term Growth in Developing Countries and Its Determinants: An Empirical Analysis, World Development, Vol. 18:

15 Savings and Economic Growth Nexus in a Small Opening Economy: 77 Pesaran, M. H. and Y. Shin, (1998), An Autoregressive Distributed-Lag Modeling Approach to Cointegration Analysis, in Econometrics and Economic Theory in the 20th Century: The Ragnar Frisch Centennial Symposium, Steinar Strom (ed), Cambridge University Press, New York, 1998, pp Pesaran, H.M.; Y, Shin and R.Smith, (1999), Bounds Testing Approaches to the Analysis of Long- Run Relationship, Department of Applied Economics, University of Cambridge, 199. Okpara,G. C. (2010), Relative Potency of Financial Repression and Liberalization on Financial Development and Economic Growth: An Empirical Survey. American Journal of Scientific and Industrial Research. ISSN: X. Romm, A. T. (2003), The Relationship between Savings and Growth in South Africa: An Empirical Study. TIPS/DPRU FORUM, Shaw, E. S. (1973), Financial Deepening in Economic Development, Oxford University Press. Sinha, D. and T. Sinha (1998), Cart Before the Horse? The Saving Growth Nexus in Mexico, Economics Letters, 61, pp

16