IllIHBII^M AllI'IE IN PRESS

HOSTED BY

ELSEVIER

Available online at www.sciencedirect.com

ScienceDirect

Review of Development Finance xxx (2017) xxx-xxx

Review

Review of

DEVELOPMENT

FINANCE

www.elsevier.com/locate/rdf

Financial liberalization and growth in African economies: The role of policy complementarities

Ousmanou Njikam

Faculty of Economics & Management, University of Yaoundé II, Cameroon

Abstract

This paper examines whether the effect of financial liberalization on economic growth depends on reform complementarities. A non-linear growth regression specification that interacts a proxy of financial liberalization with proxies of reform complementarities is estimated using a panel of 45 Sub-Saharan Africa (SSA) countries. The cross-country, panel-data evidence shows no clear relationship between financial liberalization and growth. The study however finds that financial liberalization is more likely to positively and significantly increase growth across the SSA region if the following complementary reforms are undertaken e.g. improvement in educational attainment, macroeconomic and external stability, and overall governance.

© 2017 Africagrowth Institute. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

JEL classification: O11; O16; O38; O55

Keywords: Financial liberalization; Growth; Economic reform; Policy complementarity; Africa

Contents

1. Introduction................................................................................................................00

2. Model specification and data.................................................................................................00

2.1. Model specification...................................................................................................00

2.2. Data................................................................................................................00

3. Results....................................................................................................................00

4. Conclusion.................................................................................................................00

Acknowledgements.........................................................................................................00

Appendix..................................................................................................................00

References.................................................................................................................00

1. Introduction

Financial liberalization is an important catalyst for economic growth (Stiglitz, 2000). At the domestic front, financial liberalization affects growth through the following channels: it reduces quantitative controls, allows interest rates to be market determined; produces higher interest rates and allocates capital toward higher return projects. Market determination of interest rates results in positive real interest rates, which improve

E-mail address: onjikam2002@yahoo.fr

financial intermediation and provide incentives for borrowers to invest in more productive activity (Gibson and Tsakalotos, 1994). At the external front, financial openness creates large capital flows and larger financial markets mean cheaper and more credit (Özdemir, 2014). The inflows of foreign savings augment domestic investment and capital-poor economies can free themselves from a binding constraint on economic growth e.g. lack of capital (Levine, 2001). Financial liberalization is also associated with increased domestic competition and technology transfers with a potential positive influence on economic growth (Levine, 2006).

http://dx.doi.org/10.1016/j.rdf.2017.02.001

1879-9337/© 2017 Africagrowth Institute. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

ARTICLE IN PRESS

O. Njikam / Review of Development Finance xxx (2017) xxx—xxx

But, critics assert that its benefits have not been demonstrated mainly because countries have lagged behind in terms of reform complementarities (Williamson, 1993; Rodrik and Subramanian, 2009). For example, the growth effect of financial liberalization can be disappointing if implemented in a context of poor human capital unable to welcome foreign capital inflows and the associated technology (Rodrik, 1999; Edison et al., 2002; Abed, 2003).

Since 1980, growth in Sub-Saharan Africa (SSA) has declined at almost 1% per annum (Collier and Gunning, 1999).1 The choice of bad policies e.g. financial repression seems to be the main cause of this slow growth (Azam et al., 2002). In order to achieve higher levels of economic growth, financial openness became the appropriate development strategy of SSA countries.2 Contrary to expectations, after nearly three decades, financial liberalization is not growth enhancing in SSA economies (Reinhart and Tokatlidis, 2003).

The empirical studies on financial openness in SSA focused on its design, difficulties, implementation, and growth effect (Chang and Mendy, 2012; Ahmed, 2013). But, less attention has been given to the role of complementary reforms in the financial liberalization-growth nexus. The aim of this article is twofold: investigate empirically the impact of financial liberalization on economic growth over the sample period of 1970-2010 and then examine whether the growth effect of financial liberalization is contingent on complementary policies. This article is the first study to tackle these questions focusing on the SSA countries. Our findings show no relationship between financial liberalization and growth while the growth effect of financial liberalization is significantly positive if accompanied by improvement in schooling, macroeconomic and external stability, and overall governance.

The rest of the paper is as follows. Section 2 presents the model and data. Section 3 presents and discusses the results and Section 4 concludes.

2. Model specification and data

2.1. Model specification

Financial liberalization is not growth enhancing in the presence of distortions. To test this prediction, we estimate the following equation,

Ayit = c + p0yn-1 + P1FLU + fcCRit + fo(FLu x yit—1) +P4(FLit x CRU) + /M + n + £it (1)

1 Though economic performance in SSA has markedly improved since the mid-1990s, the low and unsustained level of growth in SSA remains a major challenge to economists.

2 Beginning in the 1990s, many SSA countries liberalized interest rates, phased out directed credit, moved to indirect monetary policy instruments, restructured and privatized banks, and reinforced banking sector supervision and microfinance. At the external front, they abolished controls on international capital movements. See among others Mehran et al. (1998) for a compact description of the financial reforms in SSA countries.

where the subscripts i and t represent country and time period, respectively; y is the log of GDP per capita; yt-1 is the level of per capita GDP at the start of the corresponding period; CR is a set of complementary reform variables; FL represents financial liberalization i.e. a dummy indicator or the Chinn and Ito (2008) index to capture the impact of domestic and external financial liberalization, respectively; ¡xt and m denote unobserved time-and country-specific effects, respectively; and e is the error term. To allow the growth effect of financial liberalization to vary with the complementary reforms, FL is interacted with CRit one at a time. This strategy allows the derivation of different set of coefficients for 'without' and 'with' policy complementarity. Hence, the coefficient ¡¡1 measures the growth effect of financial liberalization while the pattern of complementarity is captured by ¡} 1 + ¡¡4 when using the dummy and ¡} 1 + ¡¡4 x CRit when the Chinn-Ito's index is the measure of financial openness. We expect ¡¡1 + ¡¡4 x CRit to be positive for the variable whose increase indicates progress and negative for the variables whose decline denotes improvement (e.g. inflation, size of government, external debt). If ¡¡4 is significant, then the effect of financial liberalization on growth depends on what happens to a given complementary reform variable. The rate of conditional convergence when using the 0-1 dummy or Chinn-Ito's index is measured by ¡ 0 + ¡ 3 and ¡ 0 + ¡ 3 x FLit, respectively. If ¡ 3 is significant and positive, then a greater financial liberalization contributes to a faster rate of conditional convergence.

We used the system-GMM estimator i.e. we estimate a system of equations in differences and in levels (see Arellano and Bover, 1995; Blundell and Bond, 1998). For the equation in differences and variables measured as period averages e.g. inflation, public infrastructure, government size, trade openness, external debt, and governance index, the instrument is the average of period t — 2, while for the variables in initial values e.g. per capita GDP and secondary school enrolment, the instrument is the observation at the start of period t — 1. For the equation in level, the instruments are given by the lagged differences.3

2.2. Data

We follow the convention for the studies in this area and focus on the SSA countries, setting aside the North African countries.4 The sample consists of an unbalanced panel dataset that comprises 45 SSA countries in 1970-2010. To control for business cycles, the dataset is structured as a panel with observations for each country consisting of non-overlapping 5-year averages for each variable. Thus, each country has 8 observations - the averages for 1970-1974, 1975-1979, 1980-1984, 1985-1989, 1990-1994, 1995-1999, 2000-2004, and 2005-2010.5

3 The system GMM estimates are obtained using the STATA command -xtabond2 - which provides the first-order, second-order, and the Sargan/Hansen test statistics.

4 The main reason is that the North African countries e.g. Algeria, Egypt, Libya, Morocco and Tunisia are part of a different regional economy - the MENA - with its own distinctive set of economic issues.

5 The last panel rather consists of a 6-year average spanning the 2005-2010 period. Appendix Table A provides the full list of countries in the sample.

ARTICLE IN PRESS

O. Njikam/ Review of Development Finance xxx (2017) xxx—xxx 3

As the dependent variable, we used GDP per capita growth rates i.e. the log difference of real GDP per capita in PPP terms, data are from the Penn World Table 7.1 (PWT7.1). For the independent variables, initial GDP per capita is the initial level of log GDP per capita at the start of each 5-year period while financial liberalization is measured using a 0-1 dummy for dates of financial liberalization events or the Chinn-Ito's index.6 We also include gross domestic investment to GDP ratio in constant prices and annual population growth rate, and data are from PWT7.1.

For policy complementarity variables, education is the total secondary enrolment, regardless of age, to the population of the age group that officially corresponds to that level of education, observations are from the World Bank World Development Indicators (WDI).7 We follow Savvides (1995) and use the growth in the share of government spending in GDP, observations are from the PWT7.1. Inflation rate is the percentage change in the consumer price index (CPI). Following Chang et al. (2009) and in order to take into account the fact that very low or negative inflation rates are also signs of macroeconomic instability, we take the absolute value of annual inflation minus 3%. Public infrastructure is proxied by the number of main telephone lines per capita and electricity production in Kwh. The external stability is measured by the annual growth rate of the foreign debt/GDP ratio. Data on foreign debt, inflation, and public infrastructure are from the WDI. Trade openness is measured as the ratio of exports plus imports to GDP from PWT7.1. Governance is captured using the indicators of civil liberties and political rights from Freedom House and range from 1 (maximum rights) to 7 (minimum rights). Both indicators are combined into a composite indicator of governance as follows: [14 — (Civil Liberties + Political Rights)]/14.8

Summary statistics of variables as well as the percentage difference in means before and after financial liberalization are in Table 1. The growth rate of GDP per capita, Chinn-Ito's index and control variables vary widely, justifying the use of the panel estimation techniques. The post-liberalization era is not statistically different from the pre-liberalization period in terms of growth rate of GDP per capita, initial GDP per capita, primary school enrolment, growth rate of government size, gross domestic investment or openness to international trade. The remaining differences are statistically significant. Compared to the pre-liberalization, the following variables are significantly higher in

6 See Appendix Table B for the dates indicating the switch toward a more liberal domestic regime in SSA countries. The Chinn-Ito's index is available online at http://www.ssc.wisc.edu/~mchinn/research.html. It is based on four convertibility restrictions reported in the IMF's Exchange Arrangement and Agreements: restrictions on payments for capital account transactions, surrender or repatriation requirements for export proceeds, and the existence of multiple exchange rates. The index is calculated based on the first standardized principal component of the four variables above, and it ranges between —2.66 (full capital controls) and 2.66 (complete liberalization). This indicator of financial liberalization is used in much of the empirical work e.g. Yalta and Yalta (2012) among others.

7 The stability of the results will be checked using the primary school enrolment ratio.

8 See Helliwell (1994) for details.

■' <0 -t

I „8

Fig. 1. Correlation between financial liberalization and economic growth in SSA, 1970-2010.

the post-liberalization period e.g. Chinn-Ito's index (19.8%), initial secondary enrolment (60.1%), inflation rate (38.8%), main telephone lines per capita (90.8%), electricity (108.4%), and the quality of governance (13.2%). Conversely, the post-liberalization growth rates of foreign debt to GDP and population are respectively 15 times and 22% significantly lower than the pre-liberalization ones.

Before turning to econometric results, it is important to recall the main stylized fact that characterizes financial liberalization and growth nexus in SSA. Fig. 1 presents the simple correlation between economic growth and financial liberalization and reveals no apparent relationship between financial liberalization and economic growth. This is a key piece of evidence that is the focus of the evaluation of whether the effect of financial liberalization on growth can be significantly enhanced if some complementary reforms are undertaken.

3. Results

The results of the basic regression with no interaction terms are reported in Table 2. In the basic regression, the Ramsey's RESET test fails to reject the null hypothesis of no omitted variables.9 In all cases, the Hansen and second order serial correlation tests show that the null hypothesis of correct specification cannot be rejected. This lends support to the estimation results. The results in column [1] show that as expected the initial GDP per capita carries a negative albeit insignificant coefficient. The coefficient on the secondary school enrolment is insignificant corroborating the findings by Savvides (1995).10 As predicted by the Solow model, the growth of population contributes negatively (and significantly) to economic growth. For example, an increase in the annual growth rate of population by one percentage point reduces real annual per capita income growth by 0.6%. Also, external debt and inflation have negative and highly

9 The Ramsey's RESET test rejects with a p-value of 0.057 the null hypothesis of no omitted variables when the public infrastructure variables e.g. main telephone lines per capita and electricity production are included in the estimation model.

10 This finding should not be interpreted as a lack of importance of human capital accumulation in African growth. Indeed, the secondary enrolment ratio is not a significant determinant of economic growth possibly because of the poor quality of enrolment data and/or the possible collinearity of enrolment indices with physical capital accumulation. The unreported results, but available from the authorusing the initial primary school enrolment ratio yield a positive (1.141) and insignificant (standard error of 1.045) coefficient.

-ai—jâw» -, rmthw. __

LKrtr -----—^

«Buundl •ZwffU) »Cot» il**«

t&SSL. -TO*

m ConootMnehaaa)

in proxied by the (log of) Chinn-Ito's index coeff=-0.1841, (robust) 86=0.4606, t-0.38 I

ARTICLE IN PRESS

4 O. Njikam / Review of Development Finance xxx (2017) xxx—xxx

Table 1

Descriptive statistics (1970/1974 - 2005/2010).

(i) Descriptive statistics for all SSA countries

Variable Mean Standard deviation Minimum Maximum

Chinn-Ito's index -0.681 1.055 -1.864 2.439

Growth rate of GDP per capita (%) 0.690 4.297 -22.451 36.198

Initial GDP per capita (in logs) 7.106 0.912 4.033 10.166

Secondary enrolment (in logs) 2.892 0.855 0.055 4.794

Primary enrolment (in logs) 4.191 0.748 -1.561 5.146

Growth rate of government size (%) 0.137 5.369 -18.810 35.737

Inflation (in logs [inflation rate - 3%]) 1.630 1.380 -1.087 7.606

Growth rate of trade openness (%) 0.624 5.747 -21.342 31.041

Main telephone lines per capita (in logs) -0.617 1.349 -3.963 3.402

Electricity production (in kwh, in logs) 21.338 1.885 14.914 26.253

Growth rate of foreign debt/GDP ratio (%) 1.627 20.770 -70.105 203.917

Investment to GDP ratio (in logs) 2.773 0.682 0.379 4.241

Growth rate of population (%) 2.079 0.869 -2.182 8.141

Governance 0.309 0.211 0.000 0.857

Number of observations 319

(ii) Mean t-test comparison

Pre-liberalization Post-liberalization Percentage change in meana

Mean Standard deviation Mean Standard deviation

Chinn-Ito's index -0.794 0.926 -0.596 1.137 19.84**

Growth rate of GDP per capita (%) 0.461 3.445 0.863 4.843 40.16

Initial GDP per capita (in logs) 7.086 0.843 7.121 0.962 3.57

Secondary enrolment (in logs) 2.549 0.930 3.150 0.692 60.09***

Primary enrolment (in logs) 4.198 0.753 4.187 0.747 -1.09

Growth rate of government size (%) 0.065 3.928 0.192 6.249 12.69

Inflation (in logs [inflation rate - 3%]) 1.851 1.266 1.463 1.441 -38.79***

Growth rate of trade openness (%) 0.178 6.479 0.960 5.121 78.17

Main telephone lines per capita (in logs) -1.135 1.146 -0.227 1.361 90.83***

Electricity production (in kwh, in logs) 20.757 1.844 21.841 1.782 108.44***

Growth rate of foreign debt/GDP ratio (%) 10.811 26.717 -4.139 13.057 -1495.04*

Investment to GDP ratio (in logs) 2.722 0.759 2.812 0.618 9.04

Growth rate of population (%) 2.205 0.657 1.985 0.991 -22.00***

Governance 0.233 0.180 0.365 0.215 13.20***

Number of observations 137 182

a A difference of means test between the post- and pre-liberalization periods. Denotes significant at 5% level. *** Denotes significant at 1% level.

significant coefficients, indicating that external and domestic instabilities are harmful to SSA economic growth. An increase in the annual growth rate of foreign debt/GDP ratio by 10% is associated with a decrease of 0.3% in the average annual growth rate of per capita GDP. A similar increase in the average annual inflation rate depresses average annual GDP per capita growth by 6.5%.

The coefficient of investment is positive and highly significant. This indicates the importance of physical capital accumulation for SSA countries: an increase in the investment/GDP ratio by 10% is associated with an increase of 17.6% in the average annual growth rate of real per capita GDP. The coefficient on the institutional quality is positive and statistically significant. This indicates that better institutions have a beneficial impact on economic growth across the SSA region: a 1% increase in the governance index tends to increase annual per

capita economic growth by 4.7%. Finally, financial liberalization, proxied by the dummy variable or the Chinn-Ito's index, is associated with slower growth, but the effect is statistically significant only in the former case.

Table 3 Panels A-B presents the results with interaction. The results in column [1] Panel A reveal that the coefficient of the interaction between initial GDP per capita and the dummy is positive and significant, indicating that SSA economies experience more rapid conditional convergence following financial openness. The coefficients on the interaction between financial liberalization and secondary schooling in column [2] and governance index in column [7] are positive and strongly significant. This indicates that financial liberalization boosts economic growth in SSA when accompanied with a stronger human capital basis and good institutions. For example, the results using the financial liberalization dummy show that while governance

HlllPHWI^M AIIIILE IN PRESS

O. Njikam /R, Table 2 Basic growth regression system-GMM estimation results. eview of Development Finance xxx (2017) xxx—xxx 5

Variable Dummy [1] Chinn-Ito's index [2]

Constant 9.194 (7.085) 7.116 (8.892)

log(initial GDP per capita) -1.888 (1.381) -1.468 (1.749)

log(initial secondary enrolment) -0.385 (1.380) -0.791 (1.600)

Growth rate of government size -0.03 1(0.033) -0.032 (0.037)

log(deviation of inflation rate from 3%) -0.649*** (0.254) -0.561*** (0.244)

Growth rate of external debt/GDP ratio -0.025** (0.014) -0.024** (0.013)

Growth rate of trade openness 0.026 (0.032) 0.022 (0.032)

log(investment/GDP ratio) 1.764*** (0.698) 1.764*** (0.743)

Growth rate of population -0.545*** (0.251) -0.560*** (0.256)

Governance 4.660*** (1.517) 4.292*** (1.501)

Dummy financial liberalization -1.460** (0.795) /

log(Chinn-Ito's index) / -0.021 (0.454)

RESETa 0.738 0.742

2nd order serial correlation test (p-value) 0.226 0.237

Hansen test (p-value) 0.205 0.206

# observations 280 280

a Ramsey RESET test, p-value for H0: no omitted variables (based on OLS estimation). The dependent variable is the growth rate of real GDP per capita. Numbers in parentheses are the corresponding robust standard errors. ** Denotes significant at 5% level. *** Denotes significant at 1% level.

increases per capita growth before financial liberalization by 6.7 times, the increase for the post-liberalization period is 9.8 times. Also, the secondary schooling has no significant effect on growth before financial liberalization and has significant effects on the financial liberalization-growth link.11 With respect to Chinn-Ito's index, the coefficient on the interaction between financial liberalization and secondary school enrolment shows that a one-standard deviation increase in schooling is associated with a 2.7 percentage points increase in growth rate. Likewise, the coefficient estimate of the interaction between financial liberalization and governance index shows that a one standard deviation increase in the institutional quality is associated with an increase in the GDP per capita growth rate of 15.3 percentage points.12 Another interesting pattern of reform complementarity emerges in columns [4] and [6]. Here again, the coefficients on the interaction between financial liberalization and inflation or foreign debt are statistically significant and with the correct sign, that is negative since a decline in both inflation and growth rate of foreign debt denotes improvement. Hence, the growth effect of financial liberalization depends positively on the progress made in lowering price inflation and reducing the growth rate of foreign debt. In the case of dummy financial liberalization, while inflation decreases per capita growth in the pre-liberalization sample by 3% annually, the decrease for the post-liberalization is 7.5% annually. Also, while the growth rate of foreign debt depresses annual per capita economic growth by 0.8% before

11 The unreported results using primary schooling are statistically insignificant.

12 From Eq. (1), the effect of a given complementary reform is given by

AGrowth = ¡¡crv + Pint x CRV, where Pcrv and Pint are the estimated regression coefficients on a given complementary reform variable and on the interaction between financial liberalization and a given complementary reform variable (CRV), respectively.

financial liberalization, the depression following financial liberalization is 10.3% annually. In the case of the Chinn-Ito's index, the quantitative effect is also quite strong. An increase in the inflation rate of 1.4 percentage points (equivalent to a one-standard deviation change) is associated with a 7.9 percentage points drop in the GDP per capita growth rate, while a one standard deviation increase in the growth rate of foreign debt tends to depress annual per capita economic growth by 4.1 percentage points.

Finally, we conducted two robustness checks. Because of space constraint, only the results on the different interactions are presented. First, it is argued in the literature that the growth effect of financial liberalization depends on the countries' level of development as proxied by the level of income. Hence, we include the interaction between financial liberalization proxy and initial GDP per capita in each of the regressions. Table 4 Panels A-B presents results. We continue to observe pattern of complementarity between financial liberalization and schooling, inflation, foreign debt, and governance. This is consistent with the expectation that the beneficial impact of financial liberalization on economic growth is larger when, educational attainment and institutional quality are improved, external environment as measured by the growth rate of foreign debt is stable, and inflation rate is lower.

In the second check, we use the least squares dummy variable corrected (LSDVC) estimator which has been proposed as a suitable dynamic panel data technique in the case of small samples (Bruno, 2005).13 Table 5 Panels A-B presents the results. All the interaction terms are basically the same: education and governance still exert a positive and significant effect on the financial liberalization-growth link. The coefficients on the interactions

13 The LSDVC estimates are obtained using the Stata commands - xtlsdvc.

ARTICLE IN PRESS

O. Njikam / Review of Development Finance xxx (2017) xxx—xxx

Table 3

Economic growth, financial liberalization and reform complementary in SSA.

Panel A

Interaction of dummy financial liberalization with

[1] Initial GDP per capita

[2] Schooling

[3] Government size

[4] Inflation

Control variables

log(initial GDP percapita) —1.717(1.689)

log(initial secondary enrolment) —0.495(1.731)

Growth rate of government size —0.033(0.034)

log(inflation) —0.617**(0.334)

Growth rate of external debt/GDP ratio —0.024**(0.014)

Growth rate of trade openness 0.025(0.033)

log(investment/GDP ratio) 1.797***(0.723)

Growth rate of population —0.543***(0.250)

Governance 4.539***(1.710)

interactions

log(initial GDP percapita) x Dummy 0.124**(0.063) log(initial secondary enrolment) x Dummy Growth rate of government size x Dummy log(inflation) x Dummy

-1.392(1.921)

-0.810(1.645)

-0.036(0.036)

-0.547**(0.300)

-0.024**(0.013)

0.022(0.033)

1.762***(0.774)

-0.567***(0.267)

4.133**(2.055)

0.134 (0.070)

-1.392(1.510)

-0.665(1.382)

-0.111(0.104)

-0.576***(0.241)

-0.024**(0.013)

0.025(0.032)

1.699***(0.744)

-0.584***(0.243)

4.214***(1.488)

0.096(0.120)

-1.633(1.524)

-0.745(1.373)

-0.027(0.033)

-0.030(0.026)

-0.023*(0.014)

0.017(0.032)

1.797***(0.740)

-0.537***(0.245)

4.358***(1.493)

-0.045 (0.025)

2nd order serial correlation test (p-value) 0.229

Hansen test (p-value) 0.122

# Observations 280

0.242 0.133 280

0.188 0.109

0.157 0.205 280

Panel A

Interaction of dummy financial liberalization with

[5] Trade openness

[6] Foreign debt

[7] Governance

Control variables log(initial GDP per capita) log(initial secondary enrolment) Growth rate of government size log(inflation)

Growth rate of external debt/GDP ratio Growth rate of trade openness log(investment/GDP ratio) Growth rate of population Governance

interactions

Growth rate of trade openness x Dummy Growth rate of external debt/GDP ratio x Dummy Governance x Dummy

-1.352(1.473)

-0.093(1.702)

-0.007(0.042)

-0.651***(0.219)

-0.026**(0.015)

0.356(0.247)

2.013***(0.870)

-0.206***(0.033)

3.818***(1.243)

0.080(0.051)

-1.177(1.535) -0.731(1.372) -0.030(0.032) -0.469**(0.230) -0.008(0.008) 0.021(0.028) 1.628***(0.674) -0.463***(0.203) 3.863***(1.424)

-0.095 (0.032)

-1.753(1.402)

-0.768(1.364)

-0.031(0.033)

-0.643***(0.251)

-0.023**(0.014)

0.024(0.032)

1.853***(0.745)

-0.551***(0.254)

6.708***(2.402)

3.113*(1.904)

2nd order serial correlation test (p-value) Hansen test (p-value) # Observations

0.242 0.352 280

0.195 0.141

0.221 0.303 280

Panel B

Interaction of Chinn-Ito's index of financial liberalization with

[1] Initial GDP per capita

[2] Schooling

[3] Government size

[4] Inflation

Control variables log(initial GDP per capita) log(initial secondary enrolment) Growth rate of government size log(inflation)

Growth rate of external debt/GDP ratio Growth rate of trade openness log(investment/GDP ratio) Growth rate of population Governance log(Chinn-Ito's index)

-1.894(1.665)

-0.754(1.506)

-0.033(0.037)

-0.652***(0.247)

-0.022* (0.013)

0.017(0.032)

1.701***(0.691)

-0.564***(0.251)

4.783***(1.623)

-8.861***(3.674)

-0.996(1.508)

-1.265(1.408)

-0.031(0.039)

-0.563***(0.233)

-0.024**(0.014)

0.024(0.032)

1.664***(0.743)

-0.568***(0.268)

4.338***(1.543)

-2.676***(1.236)

-3.335(2.769)

-3.474(2.641)

-0.285**(0.158)

-0.986***(0.439)

0.034(0.029)

0.009(0.070)

2.627***(1.116)

-0.405***(0.155)

9.963**(5.003)

-0.906(0.934)

-1.326(2.227)

-0.848(1.662)

-0.031(0.039)

-0.546**(0.290)

-0.025**(0.014)

0.027(0.054)

1.723**(0.792)

-0.592**(0.301)

4.473***(1.600)

0.280(1.723)

МНИЯ^^И ARTICLE IN PRESS

O. Njikam/ Review of Development Finance xxx (2017) xxx—xxx 7

Table 3 (Continued))

Panel B Interaction of Chinn-Ito's index of financial liberalization with

[1] Initial GDP per capita [2] Schooling [3] Government size [4] Inflation

Interactions log(initial GDP per capita) x Chinn-Ito log(initial secondary enrolment) x Chinn-Ito Growth rate of government size x Chinn-Ito log(inflation) x Chinn-Ito 1.230* **(0.536) 0.941**(0.529) -1.758** (0.965) -0.247**(0.138)

2nd order serial correlation test (p-values) Hansen test (p-value) # Observations 0.285 0.319 280 0.217 0.103 280 0.218 0.125 280 0.291 0.202 280

Panel B Interaction of Chinn-Ito's index of financial liberalization with

[5] Trade openness [6] Foreign debt [7] Governance

log(initial GDP per capita) log(initial secondary enrolment) Growth rate of government size log(inflation) Growth rate of external debt/GDP ratio Growth rate of trade openness log(investment/GDP ratio) Growth rate of population Governance log(Chinn-Ito's index) Growth rate of trade openness x Chinn-Ito Growth rate of external debt/GDP ratio x Chinn-Ito Governance x Chinn-Ito -1.464(1.736) -0.693(1.604) -0.034(0.037) -0.541***(0.244) -0.024** (0.013) 0.021(0.031) 1.679*** (0.750) -0.596***(0.253) 4.524***(1.480) 0.057(0.046) -0.065(0.044) -1.540(1.740) -0.792(1.578) -0.035(0.037) -0.571*** (0.243) -0.022(0.014) 0.022(0.032) 1.793***(0.746) -0.537***(0.259) 4.366***(1.564) 0.012(0.014) -0.009***(0.003) -1.488(1.738) -0.780(1.585) -0.033(0.037) -0.553***(0.244) -0.024**(0.013) 0.023(0.032) 1.754***(0.743) -0.562***(0.257) 4.375***(1.538) 0.103(0.174) 0.356***(0.146)

2nd order serial correlation test (p-value) Hansen test (p-values) # Observations 0.190 0.116 280 0.241 0.200 280 0.235 0.266 280

The dependent variable is the growth rate of real GDP per capita. Numbers in parentheses are the corresponding robust standard errors. * Denotes significant at 10% level. ** Denotes significant at 5% level. *** Denotes significant at 1% level.

between financial liberalization and inflation, and growth of foreign debt in columns [4] and [6], respectively remain of the same sign, same statistical significance, albeit of different magnitude. This corroborates the expectation that the lower the inflation rate and the growth rate of foreign debts, the more SSA countries benefit from financial liberalization.

4. Conclusion

In this paper, we have rigorously looked at how complementarities affected the link between both domestic and external financial liberalization and economic growth in 45 Sub-Saharan Africa (SSA) countries in 1970-2010. The empirical analysis used a non-linear growth regression specification that interacts the proxy of financial liberalization (a 0-1 dummy variable and the Chinn-Ito's index for domestic and external financial liberalization, respectively) with proxies of reform complementarities e.g. schooling, inflation stabilization, fiscal policy, trade openness, external stability, and governance.

The empirical work finds no significant effect of financial liberalization on SSA economic growth. Second, human capital

positively and significantly affects the financial liberalization-growth link. Finally, the growth impact of financial liberalization is also conditional on macroeconomic and external stability, and a sound institutional framework.

From a policy point of view, the main message of the paper is that financial liberalization could result in higher economic growth across SSA if accompanied by improvement in secondary educational attainment, macroeconomic stability i.e. inflation, stable external environment i.e. foreign debts, and overall governance.

Acknowledgements

We are grateful for the comments received from participants at the different African Economic Research Consortium (AERC) workshops. We also wish to thank all the AERC resource persons, and in particular, Prof. Isaac Otchere for helpful suggestions and comments on the early drafts. We gratefully acknowledge the financial support from the AERC under the Grant No RC14507.

та i

. t F n

i о n

Table 4

Robustness check to level of development.

Panel A

Interaction of dummy financial liberalization with

[1]Initial GDP per capita [2]Schooling

[3]Government size [4]Inflation

[5]Trade openness [6]Foreign debt

[7]Governance

log(initial GDP per capita) x Dummy log(initial secondary enrolment) x Dummy Growth rate of government size x Dummy log(inflation) x Dummy Growth rate of trade openness x Dummy Growth rate of external debt/GDP ratio x Dummy Governance x Dummy

Panel B

0.124** (0.063)

3.951* (2.422) -0.139 (0.122) 0.332*** (0.114)

0.080(0.122)

0.025** (0.013) -0.135 (0.125) -0.417** (0.238)

-0.125 (0.088)

-0.080(0.119)

-0.090*** (0.035)

0.192*** (0.006)

3.093*** (1.102)

Interaction of Chinn-Ito's index with

[1]Initial GDP per capita [2]Schooling

[3]Government size [4]Inflation

[5]Trade openness [6]Foreign debt [7]Governance

log(initial GDP per capita) x Chinn-Ito log(initial secondary enrolment) x Chinn-Ito Growth rate of government size x Chinn-Ito log(inflation) x Chinn-Ito Growth rate of trade openness x Chinn-Ito Growth rate of external debt/GDP ratio x Chinn-Ito Governance Chinn-Ito

1.230*** (0.536) 1.009** (0.591) 1.231*** (0.503)

0.268*** (0.055)

0.063 (0.054)

1.236*** (0.546) 1.208*** (0.526)

1.230*** (0.551) 1.238*** (0.550)

-0.033 (0.027)

-0.072* (0.043)

-0.019*** (0.005)

1.313*** (0.490)

The dependent variable is the growth rate of real GDP per capita. Numbers in parentheses are the corresponding robust standard errors.

* Denotes significant at 10% level.

** Denotes significant at 5% level.

*** Denotes significant at 1% level.

Table 5

Robustness check to estimation method.3

Panel A Interaction of dummy financial liberalization with

[1]Schooling [2]Government size [3]Inflation [4]Trade openness [5]Foreign debt [6]Governance

log(initial secondary enrolment) x Dummy Growth rate of government size x Dummy log(inflation) x Dummy Growth rate of trade openness x Dummy Growth rate of external debt/GDP ratio x Dummy Governance x Dummy 0.454** (0.247) 0.124 (0.10) -0.282** (0.129) -0.125** (0.071) -0.118*** (0.045) 0.589*** (0.260)

Panel B Interaction of Chinn-Ito's index with

[1]Schooling [2]Government size [3]Inflation [4]Trade openness [5]Foreign debt [6]Governance

log(initial secondary enrolment) x Chinn-Ito Growth rate of government sizex Chinn-Ito log(inflation) x Chinn-Ito Growth rate of trade openness x Chinn-Ito Growth rate of external debt/GDP ratio x Chinn-Ito Governance x Chinn-Ito 0.335*** (0.189) 0.023 (0.072) -0.385** (0.197) 0.001 (0.038) -0.031** (0.018) 0.363*** (0.112)

a Coefficients are estimated for the period 1970/1974-2005/2010. Dependent variable: Growth rate of real GDP per capita. Numbers in parentheses are the corresponding robust standard errors. ** Denotes significant at 5% level. ** Denotes significant at 1% level.

ARTICLE IN PRESS

10 O. Njikam / Review of Development Finance xxx (2017) xxx—xxx

Appendix.

Table A

Dates indicating the switch toward a more liberal financial domestic regime.

Country Year Country Year Country Year

1. Angola 1990 16. Ethiopia 1990 31. Niger 1989

2. Benin 1989 17. Gabon 1990 32. Nigeria 1987

3. Botswana 1991 18. Gambia, The 1986 33. Rwanda 1990

4. Burkina Faso 1989 19. Ghana 1988 34. Sao Tome & Principe 1990

5. Burundi 1986 20. Guinea-Bissau 1989 35. Senegal 1989

6. Cameroon 1990 21. Guinea 1989 36. Seychelles 1990

7. Cape Verde 1990 22. Kenya 1991 37. Sierra Leone 1991

8. Central African Republic 1990 23. Lesotho 1980 38. South Africa 1980

9. Chad 1990 24. Liberia 1990 39. Sudan 1990

10. Comoros 1990 25. Madagascar 1994 40. Swaziland 1990

11. Congo, Republic of 1990 26. Malawi 1988 41. Tanzania 1991

12. Congo, Democratic Republic of 1990 27. Mali 1989 42. Togo 1989

13. Cote-d'Ivoire 1989 28. Mauritania 1990 43. Uganda 1988

14. Djibouti 1990 29. Mauritius 1993 44. Zambia 1992

15. Equatorial Guinea 1990 30. Mozambique 1994 45. Zimbabwe 1991

Source: Reinhart and Tokatlidis (2003).

Table B

Sample of SSA countries.

Country Study period Country Study period Country Study period

1. Angola 1985-2010 16. Ethiopia 1980-2010 31. Niger 1970-2010

2. Benin 1970-2010 17. Gabon 1970.2010 32. Nigeria 1970-2010

3. Botswana 1970-2010 18. Gambia, The 1970-2010 33. Rwanda 1970-2010

4. Burkina Faso 1970-2010 19. Ghana 1970-2010 34. Sao Tome & Principe 1985-2010

5. Burundi 1970-2010 20. Guinea-Bissau 1990-2010 35. Senegal 1970-2010

6. Cameroon 1970-2010 21. Guinea 1990-2010 36. Seychelles 1970-2010

7. Cape Verde 1970-2010 22. Kenya 1970-2010 37. Sierra Leone 1970-2010

8. Central African Republic 1970-2010 23. Lesotho 1975-2010 38. South Africa 1970-2010

9. Chad 1970-2010 24. Liberia 1970-2010 39. Sudan 1970-2010

10. Comoros 1985-2010 25. Madagascar 1970-2010 40. Swaziland 1970-2010

11. Congo, Republic of 1970-2010 26. Malawi 1980-2010 41. Tanzania 1970-2010

12. Congo, Democratic Republic of 1970-2010 27. Mali 1990.2010 42. Togo 1970-2010

13. Cote-d'Ivoire 1970-2010 28. Mauritania 1985-2010 43. Uganda 1970-2010

14. Djibouti 1990-2010 29. Mauritius 1970-2010 44. Zambia 1970-2010

15. Equatorial Guinea 1985-2010 30. Mozambique 1980-2010 45. Zimbabwe 1970-2010

References

Abed, G.T., 2003. Unfulfilled promise. Finance Dev. 40 (1), 1-11.

Ahmed, A.D., 2013. Effects of financial liberalization on financial market development and economic performance of the SSA region: an empirical assessment. Econ. Model. 30, 261-273.

Arellano, M., Bover, O., 1995. Another look at the instrumental variable estimation of error-component models. J. Econom. 68 (1), 29-51.

Azam, J.-P., Fosu, A., Ndung'u, N., 2002. Explaining slow growth in Africa. Afr. Dev. Rev. 14 (2), 177-220.

Blundell, R., Bond, S., 1998. Initial conditions and moment restrictions in dynamic panel data models. J. Econom. 87 (1), 115-143.

Bruno, G.S.F., 2005. Approximating the bias of theLSDV estimator for dynamic unbalanced panel data models. Econ. Lett. 87, 361-366.

Chang, R., Kaltani, L., Loayza, N.V., 2009. Openness can be good for growth: the role of plocy complementarities. J. Dev. Econ. 90, 33-49.

Chang, C.C., Mendy, M., 2012. Economic growth and openness in Africa: what is the empirical relationship? Appl. Econ. Lett. 19, 1903-1907.

Chinn, M.D., Ito, H., 2008. A new measure of financial openness. J. Comp. Policy Anal. 10 (3), 309-322.

Collier, P., Gunning, J.W., 1999. Why has Africa grown slowly? J. Econ. Per-spect. 13 (3), 3-22.

Edison, H., Levine, R., Ricci, L., Slok, T., 2002. International financial integration and economic growth. J. Int. Money Finance 21 (6), 749-776.

Gibson, H.D., Tsakalotos, E., 1994. The scope and limits of financial liberalization in developing countries: a critical survey. J. Dev. Stud. 30 (3), 578-628.

Helliwell, J.F., 1994. Empirical linkages between democracy and economic growth. Br. J. Polit. Sci. 24, 225-258.

Levine, R., 2001. International financial liberalization and economic growth. Rev. Int. Econ. 9, 668-702.

Levine, R., 2006. Finance and Growth: Theory and Evidence. In: Aghion, P., Durlauf, S. (Eds.), Handbook of Economic Growth. Elsevier.

Mehran, H., Ugolini, P., Briffaux, J.P., Iden, G., Lybeck, T., Swaray, S., Hayward, P., 1998. Financial sector development in sub-Saharan Africa Countries. In: IMF Occasional Paper., pp. 169.

^ ARTICLE IN PRESS

O. Njikam/ Review of Development Finance xxx (2017) xxx—xxx

Özdemir, D., 2014. Economic growth and financial liberalization in the EU accession countries. Appl. Econ. Lett. 21 (15), 1036-1044.

Reinhart, C.M., Tokatlidis, I., 2003. Financial liberalization: the African experience. J. Afr. Econ. 12 AERC Supplement 2, ii53-ii88.

Rodrik, D., 1999. The New Global Economy and Developing Countries: Making Openness Work. Overseas Development Council, Policy Essay no. 24, Washington, D.C.

Rodrik, D., Subramanian, A., 2009. Why did financial globalization disappoint? IMF Staff Paper 56 (1), 112-138.

Savvides, A., 1995. Economic growth in Africa. World Dev. 23 (3), 449-458. Stiglitz, J.E., 2000. Capital market liberalization, economic growth, and instability. World Dev. 28 (6), 1075-1086. Williamson, J., 1993. Democracy and the 'Washington Consensus'. World Dev. 21 (8), 1329-1336.

Yalta, A.Y., Yalta, A.T., 2012. Does financial liberalization decrease capital flight? A panel causality analysis. .Int. Rev. Econ. Finance 22, 92-100.