do financial reforms and financial development...

1

Do Financial Reforms and Financial Development Enhance Merchandise Trade Performance?

Evidence from Eastern and Southern African Countries

Ronald Mangani University of Malawi, Chancellor College, Zomba, Malawi

Presented at the trapca Trade Policy Research Forum 2014

Abstract This paper analyses how financial reforms, financial development and trade flows are related in seventeen Eastern and Southern African countries. Trade flows are measured in terms of the shares of exports and imports in world totals for each country, while financial development is proxied by domestic credit to the private sector as a proportion of GDP. Separate export and import models are estimated for each of the countries using a single equation framework. These include measures of financial development and financial reforms as regressors after controlling for income and exchange rate effects, and taking into account stationarity and potential cointegration of the variables. Although a preliminary trend analysis suggests that financial reforms have led to financial development in the region which, in turn has increased both imports and exports, the regression results show no strong causal relationships. Thus, there is no robust evidence that financial development is catalytic to trade, such that both exports and imports are largely exogenous to the selected regressors. This is a puzzle that further research must seek to address by interrogating the modelling procedure adopted.

1. Introduction This paper analyses how financial reforms and financial development are related to exports and imports in the Eastern and Southern Africa (ESA) region. Governments in this region (as well as in most of the rest of Africa) adopted financial reforms during the 1980s and the 1990s. The reforms were commonly part of broader unilateral Structural Adjustment Programmes (SAPs) supported by the International Monetary Fund (IMF) and the World Bank, which were aimed at achieving greater economic liberalisation. Financial reforms within SAPs have sequentially focused on, inter alia, the decontrol of interest rates and credit levels, removal of entry restrictions and credit rationing, current and capital account liberalisation, adoption of more market-oriented monetary and exchange rate policies, introduction of more market-based financial instruments, as well as strengthening of regulatory functions. At the multilateral level, financial services liberalisation has also drawn attention under the General Agreement on Trade in Services (GATS). At that level, financial services liberalisation (i.e., trade policy reform in financial services) focuses on the removal of discriminatory regulation, such as quantitative or qualitative regulations that discriminate against foreign and domestic financial services providers with respect to market entry or commercial presence. It entails opening up the domestic financial sector through all four modes of services supply, namely cross-border supply, consumption abroad, commercial presence and permission of entry of foreign natural persons. Policy coherence between SAPs and the multilateral efforts is central, such that the programmes of the IMF and World Bank emphasise the complementary relationship between trade policy reform and domestic reform in order to strengthen the domestic financial systems of member countries (see Jansen & Vennis, 2006).

2

There is no debate that the unilateral and (to a less extent) multilateral initiatives have resulted in more developed financial sectors among ESA countries. While the reforms and resultant financial development are largely justified as being catalytic to economic growth and development in general (see Francois & Schuknecht, 1999), their hypothesised effects on trade in goods is attracting growing attention among researchers and policy-makers. This paper analyses how financial reforms, financial development and trade flows are related in a selection of seventeen ESA countries. The underlying hypotheses are that (a) financial development increases with financial reforms; (b) trade flows increase with financial reforms (c) trade flows increase with financial development; and (d) the positive association between trade flows and financial development is accentuated by financial reforms. Relative to the existing literature, the paper provides new insights in two ways. First, we contend in this paper that financial development is potentially attainable without financial reforms (e.g., it is possible for private sector credit to exhibit an upward trend in real terms in the absence of major financial reforms), and that it is possible for reforms not to lead to financial development (e.g., when the banking sector becomes more risk-averse after reforms, preferring to lend more to the public sector than the private sector). Hence, we isolate the effects of the two – financial reforms and financial development - on trade flows. Second, most studies investigate the relationships in a panel data context in which few Africa countries are included among many others globally (see, for instance, Chen et al., 2012; Manova, 2005; Beck, 2002). Such studies do not explicitly bring out the specifics relating to African countries. To address this, we present the results from country-specific analyses. The paper does not succeed to provide firm evidence in support of financial reforms and financial development as catalysts for deepening trade in most of the countries included in the analysis. The paper proceeds as follows. The relevant literature is briefly explored in Section 2, while Section 3 presents a description of the methodology adopted. Section 4 presents some prima facie evidence of the associations spanning financial reforms, financial development and trade flows in the selected countries, while the results of estimating our trade flow models are presented and discussed in Section 5. Section 6 summarises the paper. 2. The Literature The key argument in support of financial development is that it is positively associated with growth and incomes (Mckinnon, 1973; Shaw, 1973). From standard Keynesian macroeconomic theory, this suggests that financial development should also increase exports and imports. Several authors build on this reasoning to provide further insights on how international trade flows are affected by financial development (see, for instance, Manova 2008; Berthou 2007; Chang et al., 2005; Hur et al., 2006; Beck 2002). Arguably, financial development can render specific differential benefits between firms that produce for the domestic market and those in the export sector. Exporting firms are assumed to be relatively more restricted by initial entry costs (and fixed costs of exporting) vis-à-vis firms that cater exclusively to the domestic markets. The roles of firm heterogeneity and fixed costs in exporting are increasingly being recognised by trade economists (Melitz 2003; Helpman et al., 2008). Ahn et al. (2011) demonstrate that domestic sales are not affected by the banks’ providing trade finance. Chen et al. (2012) observe that trade finance contractions alone could explain about one-third of the drop in exports during the 2008 global financial crisis. Manova (2008) identifies financial development as an added source of comparative advantage. Since every export also reflects an import, these arguments equally locate the significance of financial development on imports. The effect of financial development on a country’s trade balance is, therefore, unclear and varies according to each economy.

The theoretical expectation, therefore, is that both exports and imports should increase with financial development. However, most of the studies assume that financial development (generally measured as the size of the private financial sector relative to total output) is the direct result of financial reforms,

3

hence attribute the trade effects of financial development to such reforms. It is common in the literature, therefore, to consider increased credit to the private sector as a measure of both financial development (e.g., Manova, 2008), and financial reforms (e.g., Hanif et al., 2008).

Although a growing body of the literature shows that financial reforms and/or financial development have potential to positively impact on investment and growth in sub-Saharan Africa (SSA) (see, for instance, Fowowe, 2011; Odhiambo, 2011), the empirical evidence on the effects under study is scanty in the region. Odhiambo (2011) documents support for the argument that financial liberalisation has led to financial deepening in South African, Zambia, Tanzania and Lesotho, but he does not extend the analysis to cover the trade effects of financial development. In a panel data analysis that includes SSA countries, Manova (2008) finds that countries with better developed financial systems tend to export relatively more in highly external capital dependent industries, and that equity market liberalizations increase exports disproportionately more for sectors more reliant on outside funding or characterized by softer assets. Using comparable methods, Chen et al. (2012) confirm that financial reforms significantly affect exports of developing countries, including SSA countries. Effects on manufacturing exports and imports are also documented by Beck (2002) in a panel that includes developing countries. The evidence based on country-specific regressions using data from the ESA region is unavailable in the literature, and it quite likely that the results based on panels in which African countries are a small part of the dataset may generate misleading policy recommendations. We seek to address this gap. 3. Methodology 3.1 Scope This study is conducted on a selection of seventeen ESA countries for which distinct commencement dates for financial reforms can be identified, and adequate data are available on our variables of interest. The countries and commencement dates for the reforms are presented in Table 1. Although financial reforms are an ongoing activity and may have been implemented in various forms prior to the commencement dates chosen in this paper, we identify the starting point in such reforms as the point where distinctly major changes (such as relaxation of ownership restrictions and/or interest rate and credit regulation decontrols) occur. Liberalisation of credit market conditions was the first policy intervention implemented by most countries in the context of the SAPs.

Table 1: Countries in the sample and commencement years of financial reforms

Country Commencement Year Landmark Financial Reform Measure 1. Botswana 1987 Interest rate and entry liberalisation 2. Burundi 1989 Interest rate liberalisation 3. Congo, DR 2001 Interest rate and exchange rate liberalisation 4. Egypt 1991 Exchange rate and interest rate decontrols 5. Ethiopia 1992 Interest rate liberalisation; currency devaluation 6. Kenya 1991 Interest rate liberalisation 7. Lesotho 1993 Interest rate liberalisation 8. Madagascar 1994 Exchange rate liberalisation 9. Malawi 1989 Interest rate decontrol; deregulation of bank entry 10. Mauritius 1988 Full interest rate liberalisation 11. Rwanda 1995 Liberalization of the monetary and financial sectors 12. Seychelles 1993 Interest rate liberalisation 13. South Africa 1980 Interest and credit decontrols 14. Sudan 1997 Financial reform and liberalisation programme 15. Swaziland 1994 Interest rate and entry decontrols 16. Tanzania 1992 Interest rate decontrol, exchange rate liberalisation 17. Uganda 1991 Liberalisation of financial markets

4

3.2 Trade Flow Models Separate export and import functions are estimated in this paper for each of the countries in the sample. We augment the export and import functions typically used to study exchange rate effects by including financial development and financial reform variables. Building on Bahmani-Oskooee and Goswami (2004) and following Bahmani-Oskooee and Ratha (2008), as well as Mangani (2011), the following basic export value and import value models can be specified and augmented:

0 1 2 1t t t tEXPORT GDPF EXRATEα α α µ= + + + (1)

0 1 2 2t t t tIMPORT GDP EXRATEβ β β µ= + + + (2)

In equations (1) and (2), t denotes time, 𝐸𝑋𝑃𝑂𝑅𝑇! and 𝐼𝑀𝑃𝑂𝑅𝑇! are the values of exports and imports, while 1µ and 2µ are white noise error terms. In addition to the real exchange rate (

tEXRATE ), exports may also depend on the income level of a country’s trading partners ( tGDPF ),

while imports may depend on the level of the country’s income ( tGDP). The terms iα and jβ (

, 0,1,2i j = ) are parameters to be estimated. It is expected that both 1α and 1β should be positive,

indicating that exports increase with *tY while imports increase with tY , ceteris paribus. Moreover,

since the real exchange rate is defines as units of domestic currency per unit of foreign currency, it is expected that 2 0α > and that 2 0β < . This a priori expectation is a stylised long-run condition premised on the theoretical proposition that a higher external value of for domestic currency makes foreign goods price-attractive to locals, and makes local products unattractive to foreigners. (Marshall, 1923; Lerner, 1944). In the short-run, however, these signs may be reversed to depict the so-called J-curve effect, since imports and exports adjust to exchange rate movements with time lags. Such lags may be characterised as decision lag, recognition lag, production lag, replacement lag and delivery lag (see Magee, 1973; Junz & Rhomberg, 1973; Hsing, 1999). In line with our objectives, we augment equations (1) and (2) to allow for the effects of financial development ( tCREDIT ) and financial reforms ( tD ) by specifying the following basic models:

0 1 2 3 4 5 1t t t t t t t tEXPORT D GDPF EXRATE CREDIT DCREDITα α α α α α µ= + + + + + + (3)

0 1 2 3 4 5 1t t t t t t t tIMPORT D GDP EXRATE CREDIT DCREDITα α α α α α µ= + + + + + + (4)

In equations (3) and (4), the effects of financial development without regard for reforms are captured by the parameters 4α and 4β . While tCREDIT is a standard nominal scale variable, we capture the

effects of financial reforms by introducing the categorical variable tD which assumes a value of zero in the pre-reforms period and a value of unity from the year of reforms onwards. Therefore, the parameters 1α and 1β assess whether the intercept is different in the pre-reform and post-reform

periods, while the parameters 5α and 5β assess whether the impacts of financial development on exports and imports are different during the two periods. In particular, the pre-reform impacts of financial development remain 4α and 4β in the export and import functions respectively, while the

corresponding post-reform impacts become 4 5α α+ and 4 5β β+ . Guided by the exploratory trend analysis described in Section 4 below, an alternative modelling framework may include a squared financial development term. This is attempted as a robustness check for our specification, but it is found unrewarding as the terms yield generally insignificant coefficients. Moreover, our preference for the dummy variable approach is bolstered by the fact that it

5

allows us to explicitly determine how trade flows respond to the potentially changing patterns of financial development in the pre-reform and post-reform periods. The data used in this analysis are annual time series, generally for the period 1970 - 2012, but few data gaps exist for some countries as described in Table 2. Export and import flows are measured as percentages of world totals for each country. Following most of the literature (see Manova, 2008; Beck, 2002; Levine & Zervos, 1998; Levine et al., 2000), financial development is measured by domestic credit to the private sector. This has the advantage of measuring the part of savings that is channelled to private firms. Although private sector credit is not a direct measure of efficiency, it captures part of it, since it excludes credit to the private sector by the central bank. The fact that this measure includes credit extended to the non-exporting and non-importing private sector does not reduce its appropriateness, but rather it bolsters the inquisition since it answers the very question of interest: is private credit financing trade or not? To complete our conditioning set as described in equations (3) and (4), we use real gross domestic product per capita for each country and for the world to capture the aforesaid income effects. These are expressed in US dollars at constant prices and constant exchange rates, base 2005. Finally, we used the real exchange rate for each country against the US dollar. The data on the trade flows and the income measures are obtained from the UNCTADstat database. The data on credit to the private sector are from the e-library of the IMF, while those on the real exchange rates are obtained from the World Bank on-line database. Table 2: Data gaps Country Data Gap Botswana Data start in 1972 Ethiopia Data end in 2008 Lesotho Data start in 1973 Rwanda Data end in 2005 Uganda Regressions data start in 1980, trend analysis data start in 1970 Note: Save for the few exceptions included in this table, data were for the period 1970-2012 in all other cases The standard augmented Dickey - Fuller (ADF) test is applied to establish the order of integration for each variable, but the test results are compared with those generated by using the Dickey-Fuller test with generalised least squares detrending (DF-GLS) due to Elliot et al. (1996), the Phillips-Perron (PP) test and the KPSS test due to Kwiatkowski et al. (1992). Application of the ADF test uses an automatic lag length selection procedure for the autoregressive structure, based on the Schwarz information criterion for a maximum possible lag of 6. Guided by the graphical inspection of the time series, trend and intercept terms are included in most of the tests, but some few exceptions to this general rule are highlighted in the test results. Throughout, test statistics greater than critical values in absolute value terms establish stationarity. The ADF test results are not discernibly different from those derived using the DF-GLS test, the PP test, and the KPSS test. Hence, only the ADF test results are reported in this paper, while the rest are available upon request. Where variables are integrated of the same order and the order is greater than zero (typically unity), we apply the single-equation cointegration test proposed by Phillips and Ouliaris (1990) which involves testing for the unit root in the residuals of derived from a regression in the level of such variables. Unlike the alternative proposed by Engel and Granger (1987), the selected procedure does not require the inclusion of lagged variables, in line with our modelling framework. Moreover, equations (3) and (4) imply strict exogeneity of the right-hand variables, hence the Johansen type of tests are inappropriate. We also consider it inappropriate to introduce lags - as would typically be suggested by the autoregressive distributed lag (ARDL) procedure suggested by Pesaran et al. (2001) -since annual data are used. In effect, we consider the lag effects to be fully absorbed within the year, in line with the stylised fact that empirically tested distributed lag models that use monthly data hardly ever include as many as twelve lags. The actual application of the Phillips-Ouliaris test allows for a

6

linear trend in the cointegrating equation, while the long run variances are estimated using the Bartlett kernel with Newey-West fixed bandwidth. The ultimately chosen models are tested for serial correlation of up to order 4 using the Breusch-Godfrey test, and for autoregressive conditional heteroscastity of the same order. Moreover, we confirm that the models are correctly specified against Ramsey’s regression specification error test (RESET). All the tests as well as all the estimations reported in this paper are conducted using the EViews 8 software. 4. Some Prima Facie Evidence As a precursor to our formal regression analysis, Figure 1 presents time plots of aggregate trade flow and financial development variables for all the seventeen countries in the sample. Since the year of commencement of reforms is not standard across the countries (see Table 1), we consider the period 1970-1984 as the pre-reform period, and 1998-2012 as the post reform period in this investigation. The period 1985-1997 (between the two central vertical lines in the figure) is omitted to account for the differential commencement dates of the reforms. The key conclusions drawn from this analysis are as follows: First, financial reforms are associated with greater financial development. Private credit increased by 77.9 percent from an annual average of 15.9 percent of GDP per country in the pre-reform period, to an annual average of 28.2 percent. Moreover, the growth rate in private credit was very slow for all the countries during the pre-reform period (slope of 0.129 in the time trend) but very fast after the reforms (slope of 0.738). Second, trade flows are significantly lower during the post-reform period than before. The share of world exports by these countries has dropped by 43.7 percent from an average of 1.7 percent before reforms to only 0.9 percent after reforms. Import shares have also declined from 1.8 percent to 1.2 percent. Third, both the export and import shares exhibit an upward trend after reforms, compared with the downward trend witnessed during the pre-reform period. From slopes of -0.035 and -0.007 (i.e., negative trend slopes) in export and import shares of world trade respectively, the slopes are reversed after reforms to positive values of 0.016 and 0.040, respectively. It is this reversal that we attempted to capture by including a quadratic CREDIT term in our regressions, but without success. At this trend, the higher pre-reform average shares are bound to be surpassed over time. Figure 1: A trend analysis for all countries in the sample

Pre-reform means Export = 1.660 Import = 1.750 Credit = 15.865 Post-Reform means Export = 0.876 Import = 1.190 Credit = 28.222 Pre-reform slopes Export = -0.035 Import = -0.007 Credit = 0.129 Post-Reform slopes Export = 0.016 Import = 0.040 Credit = 0.738

0.000

5.000

10.000

15.000

20.000

25.000

30.000

35.000

40.000

0.000

0.500

1.000

1.500

2.000

2.500

1970

1973

1976

1979

1982

1985

1988

1991

1994

1997

2000

2003

2006

2009

2012

Private Cred

it (%

of G

DP, average)

Expo

rt, Impo

rt (%

of w

orld to

tals)

Export Import Credit

7

The picture described above is generally sustained when we exclude South Africa, which is by far the region’s most dominant economy (see Figure 2). More importantly, the aggregate trends are reflective of the trends depicted at the level of each country (see Appendix 1). An analysis of the country-specific trends shows that (a) private credit increases much faster after reforms than before in 76.4 percent of the countries studied; and (b) financial reforms are associated with a reversal or improvement in steadily declining export and import shares in 69.5 percent and 57.8 percent of the countries, respectively. The prima facie evidence, therefore, supports the assertion that financial development – achieved through financial reforms – is associated with an increase in exports and imports in most of the countries included in the study, but the association with imports is weaker than it is with exports. Although this evidence may be useful, it does not validate the assertion of causal relationships between financial development and trade flows, and may mislead policy interventions.

5. Results 5.1 Unit roots and cointegration Appendix 2 shows the results for the ADF tests in levels and first differences of the variables. The results show that most of the variables are I(1) processes, save for a few exceptions which are I(0), viz: IMPORT in Seychelles and South Africa; EXRATE in Seychelles; and GDP in Seychelles. Based on these results, cointegration could be suspected in all the models except for the export and import functions in Seychelles, as well as the import function in South Africa. The results of applying the Phillips-Ouliaris cointegration test are reported in Table 3.Cointegration may not be rejected in the export models for Burundi, Congo DR, Lesotho, and Malawi, as well as in

Note: Exports and imports are expressed as percentages of world totals, and are summed across the countries in the sample in order to obtain the aggregates. Private credit is expressed as a percentage of each country’s GDP, and averaged across the countries in the sample to obtain the aggregates.

Figure 2: A trend analysis for all countries in the sample except South Africa

Pre-reform means Export = 0.667 Import = 0.891 Credit = 12.768 Post-Reform means Export = 0.388 Import = 0.633 Credit = 21.143 Pre-reform slopes Export = -0.033 Import = 0.009 Credit = 0.184 Post-Reform slopes Export = 0.013 Import = 0.024 Credit = 0.687

Note: Exports and imports are expressed as percentages of world totals, and are summed across the countries in the sample in order to obtain the aggregates. Private credit is expressed as a percentage of each country’s GDP, and averaged across the countries in the sample to obtain the aggregates.

0.000

5.000

10.000

15.000

20.000

25.000

30.000

0.000

0.200

0.400

0.600

0.800

1.000

1.200

1.400

1970

1973

1976

1979

1982

1985

1988

1991

1994

1997

2000

2003

2006

2009

2012

Private Cred

it (%

of G

DP, average)

Expo

rt, Impo

rt (%

of w

orld to

tals)


8

the import functions for Sudan and Tanzania. Therefore, error correction terms are included in the resulting six regression models for these countries. Appendix 3 reveals that the models ultimately chosen in this analysis pass the Breusch-Godfrey test for serial correlation of up to order 4, and are correctly specified against Ramsey’s RESET, showing no need for the inclusion of quadratic or cubic terms. Although most of them also pass the ARCH test, this is not the case for five of them (i.e., the export models for Botswana, Rwanda and South Africa, as well as the import models for Ethiopia and Rwanda). To address the ARCH effects, we generate White’s heteroskedasticity-consistent standard errors and covariances in such models. The RESET results provide added justification for not including a quadratic term in our models, notwithstanding that it may be suggested in the graphical illustrations in Appendix 1.

Table 3: Phillips-Ouliaris cointegration test results Model

Country EXPORT IMPORT Botswana 3.460 (0.561) 2.520 (0.912) Burundi 6.069 (0.007)* 3.189 (0.685) Congo, DR 6.104 (0.007)* 3.664 (0.462) Egypt 4.269 (0.220) 2.887 (0.666) Ethiopia 3.571 (0.510) 3.158 (0.701) Kenya 4.517 (0.151) 4.211 (0.238) Lesotho 5.685 (0.018)* 4.185 (0.253) Madagascar 4.976 (0.068) 4.586 (0.135) Malawi 5.208 (0.044)* 4.042 (0.299) Mauritius 3.002 (0.763) 2.946 (0.785) Rwanda 4.813 (0.103) 4.414 (0.191) Seychelles NA NA South Africa 3.753 (0.421) NA Sudan 3.004 (0.763) 6.313 (0.004)* Swaziland 3.203 (0.678) 2.629 (0.886) Tanzania 3.299 (0.634) 5.242 (0.041)* Uganda 2.853 (0.822) 4.433 (0.192) Note: Entries are test statistics under the null hypothesis of no cointegration, as well as their corresponding p-values (the latter in parentheses). * indicates evidence of cointegration. NA implies that the test is not applicable, based on the unit root test results in Appendix 2. 5.2 Estimation results: the export model The estimation results for the export model are presented in Table 4. The results are quite uninspiring in several ways. First, financial development impacts on exports in only two of the 17 countries (Burundi and Mauritius) prior to reforms, and the impact is negative in both countries. Moreover, financial development only increases exports after reforms relative to the pre-reform period in Mauritius and, in fact, it lowers them after reforms in Swaziland. Reforms also increase the exogenous component of exports (independently of financial development) in Congo DR, Madagascar, Malawi and Tanzania, but lower these exogenous exports in Mauritius. Therefore, the results do not render firm support for the assertion that financial development increases exports, nor that the effect of financial development on exports has increased after financial reforms in the ESA region. A second major observation is that hardly much else actually explains exports in these countries. Apart from the long-run effects depicted by the error correction terms in the four country equations where they are applicable, the only other causal effects on exports arise from foreign income and the exchange rate in Malawi and Swaziland, and the exchange rate in Mauritius and South African. The foreign income effect in Malawi and Swaziland is, in fact, negative. This is in conflict with the prior expectation of a positive relationship, and depicts the countries’ poor export performance. The exchange rate effects are mixed, depicting the positive Marshall-Lerner condition in Mauritius, and

9

the negative J-curve effect in Malawi, South Africa and Swaziland. The scarcity of causal effects could not be improved by replacing the GDPF variable with the country’s own output, nor by including quadratic and cubic terms, nor by adding autoregressive and distributed lag terms.

10

Table 4: Estimation results: export function Variable

Const. D GDPF EXRATE CREDIT D* CREDIT ECT

Botswana 0.003 (1.410)

-0.003 -(1.207)

0.000 (0.092)

-0.002 (-1.123)

-0.001 (-1.193)

0.000 (-0.108)

Burundi 0.000

(-0.717) 0.001

(1.399) 0.000

(-0.756) 0.000

(-0.015) 0.000

(-2.139)* 0.000

(0.644) 0.843

(4.727)*

Congo DR -0.011 (-3.576)*

0.011 (2.377)*

0.000 (1.553)

0.000 (-0.266)

-0.006 (-1.511)

0.006 (1.107)

-0.285 (-2.744)*

Egypt -0.006 0.012 0.000 -0.010 -0.001 -0.001

(-0.685) (1.329) (-0.450) (-1.353) (-0.632) (-0.228)

Ethiopia -0.002 0.001 0.000 0.000 0.000 0.000

(-2.691)* (1.386) (1.359) (0.376) (-1.055) (0.812)

Kenya -0.002 0.002 0.000 0.000 -0.001 0.001

(-0.724) (0.799) (-0.332) (-0.308) (-1.470) (0.956)

Lesotho 0.000 0.000 0.000 0.000 0.000 0.000 -0.689

(-0.122) (1.550) (-0.430) (-0.382) (-0.250) (-0.509) (-4.657)*

Madagascar -0.001 0.002 0.000 0.000 0.000 0.000

(-2.025)* (1.966)* (-0.740) (0.565) (0.502) (-0.196)

Malawi 0.000 0.001 0.000 0.000 0.000 0.000 0.397

(-0.723) (2.292)* (-2.268)* (-2.435)* (-1.275) (1.119) (3.305)*

Mauritius -0.008 -0.004 0.000 0.000 -0.001 0.001

(-2.410)* (-3.087)* (-1.672) (2.962)* (-5.194)* (3.974)*

Rwanda 0.000 0.000 0.000 0.000 0.000 0.000

(-1.183) (0.352) (0.539) (-0.610) (1.248) (-0.035)

Seychelles -0.003 -0.003 0.000 0.001 0.000 0.000

(-0.410) (-1.107) (0.749) (0.563) (-0.128) (0.090)

South Africa -0.001 -0.020 0.000 -0.027 0.000 0.000

(-0.035) (-0.651) (0.395) (-2.004)* (0.004) (-0.014)

Sudan -0.005 0.004 0.000 0.000 0.000 -0.002

(-1.852) (1.098) (0.867) (0.066) (0.133) (-0.428)

Swaziland 0.001 0.000 0.000 -0.002 0.000 0.000

(1.532) (0.351) (-3.428)* (-4.692)* (1.071) (-2.060)*

Tanzania -0.003 0.005 0.000 0.000 0.000 0.000

(-2.953)* (3.789)* (-1.012) (0.419) (1.451) (-0.837)

Uganda 0.000 0.002 0.000 0.000 0.000 0.000

(-0.361) (1.566) (-1.639) (-1.786) (-0.108) (0.150) Note:

Entries are OLS parameter estimates and their corresponding t-statistics (in parentheses). ECT are error correction terms distilled from the corresponding model in levels, arising from evidence of cointegration. * indicates statistical significance at 5%. 5.3 Estimation results: the import model Table 5 reports the results of estimating the import model across the 17 countries. The results are also quite uninspiring, but slightly better than those observed for the export model. First, financial development impacts on exports in only four of the 17 countries (Egypt, Rwanda, Swaziland, Uganda)

11

prior to reforms, and the impact is positive in the first three countries but negative in Uganda. Moreover, financial development only increases imports after reforms relative to the pre-reform period in Kenya and Uganda, but it actually lowers them in Egypt and Swaziland. Reforms also increase the exogenous component of imports in Kenya and Seychelles independently of financial development, but they lower the exogenous imports in South Africa. Therefore, the results do not render firm support for the assertion that financial development increases imports, nor that the effect of financial development on imports has increased after financial reforms in the ESA region. Table 5: Estimation results: import function

Variables

Const. D GDP EXRATE CREDIT D* CREDIT ECT

Botswana

0.000 -0.003 0.000 -0.002 0.000 0.001 (0.343) (-1.529) (2.128)* (-1.888) (-0.585) (0.934)

Burundi

0.000 0.000 0.000 0.000 0.000 0.000 (-0.230) (-0.240) (1.289) (-2.233)* (-1.440) (1.954)*

Congo DR

0.001 0.001 0.000 0.000 -0.001 0.001 (0.325) (0.144) (1.694) (0.208) (-0.390) (0.296)

Egypt 0.006 0.018 -0.001 -0.081 0.013 -0.016

(0.264) (0.738) (-0.834) (-4.137)* (2.423)* (-2.228)*

Ethiopia -0.002 0.003 0.000 -0.002 0.000 0.000

(-1.266) (1.569) (2.011)* (-1.095) (0.141) (0.116)

Kenya -0.006 0.007 0.000 0.000 0.001 -0.001

(-2.620)* (2.053)* (2.367)* (-0.498) (1.182) (-1.061)

Lesotho 0.000 -0.001 0.000 -0.001 0.000 0.000

(0.863) (-1.148) (-0.189) (-1.829) (-0.017) (0.034)

Madagascar 0.000 0.000 0.000 0.000 0.001 0.000

(0.034) (0.321) (4.207)* (-0.567) (0.713) (-0.147)

Malawi -0.001 0.001 0.000 0.000 0.000 0.000

(-1.268) (1.047) (0.675) (-1.628) (0.264) (0.353)

Mauritius 0.000 -0.002 0.000 0.000 0.000 0.000

(0.479) (-1.547) (1.645) (-1.557) (-1.113) (0.152)

Rwanda 0.000 0.000 0.000 0.000 0.000 0.000

(-0.385) (-0.190) (-0.121) (-0.743) (2.514)* (-0.070)

Seychelles 0.004 0.001 0.000 0.000 0.000 0.000

(20.994)* (3.329)* (0.728) (1.194) (0.374) (-0.023)

South Africa 0.842 -0.238 0.000 0.002 0.051 -0.052

(12.563)* (-3.224)* (0.976) (0.049) (1.532) (-1.542)

Sudan -0.002 -0.002 0.000 0.002 0.000 0.004 0.308

(-1.135) (-0.622) (2.083)* (0.516) (-0.275) (1.869) (2.785)*

Swaziland 0.000 -0.001 0.000 -0.002 0.000 -0.001

(0.429) (-0.732) (-0.146) (-3.836)* (2.394)* (-2.431)*

Tanzania -0.002 0.003 0.000 0.000 0.001 0.000 0.306

(-1.358) (0.933) (-0.084) (-1.218) (1.828) (0.254) (2.773)*

Uganda -0.001 0.002 0.000 0.000 -0.009 0.008

(-0.429) (0.961) (-0.082) (-0.948) (-3.645)* (3.201)* Note: Entries are as described in Table 4 above.

12

As with the export model, a further observation is that hardly much else actually explains imports in these countries. Nonetheless, domestic output has significant positive effects on imports in five countries (Botswana, Ethiopia, Kenya, Madagascar and Sudan). This agrees with the theoretical proposition of a positive relationship. The exchange rate effects are also trace but, where prevalent, they are consistent with the theory: domestic currency depreciation reduces imports in Burundi, Egypt and Swaziland. The two error correction terms are also significant All in all, the weak endogeneity of imports is as puzzling as it is for exports. 6. Conclusion This paper analyses how financial reforms, financial development and trade flows are related in seventeen countries drawn from the Eastern and Southern Africa (ESA) region. The paper measures trade flows in terms of the shares of exports and imports in world totals for each country, while financial development is proxied by domestic credit to the private sector as a proportion of GDP. Separate export and import functions are estimated for each country, where the effects of financial reforms and financial development are controlled for income and exchange rate effects. Although time plots of the series suggest that declining trends in exports and imports have been reversed after financial reforms in the ESA region, the regression results fail to establish firm causal effects. Of the seventeen countries, financial development impacts on exports only in Burundi and Mauritius prior to reforms, and the impact is negative. Moreover, financial development only increases exports after reforms relative to the pre-reform period in Mauritius and lowers them after reforms in Swaziland. Reforms also increase exogenous exports in Congo DR, Madagascar, Malawi and Tanzania, but lower them in Mauritius. In terms of imports, financial development impacts on exports only in Egypt, Rwanda, Swaziland, Uganda prior to reforms, and the impact is positive in the first three countries but negative in Uganda. Moreover, financial development only increases imports after reforms relative to the pre-reform period in Kenya and Uganda, but it actually lowers them in Egypt and Swaziland. Reforms also increase the exogenous component of imports in Kenya and Seychelles independently of financial development, but they lower the exogenous imports in South Africa. All in all, the results do not render firm support for the assertion that financial development increases trade flows, nor that the effect of financial development on trade flows has increased after financial reforms in the ESA region. Moreover, the results largely show that exports and imports are not strongly influenced by changes domestic or foreign income levels and exchange rate movements in most countries in the region. It is quite possible that ESA countries ought to address some structural factors within their economies (such as productive efficiency and competitiveness) before they can reap the trade benefits of financial reforms and financial development. This possibility holds even for the relatively more advanced countries in the region in which the stylised supply-side constraints may not be as adverse as in the weaker economies. References Ahn, J., M. Amiti, and D. Weinstein, 2011. Trade Finance and the Great Trade Collapse, American Economic Review: Papers & Proceedings, 101:3, 298–302. Bahmani-Oskooee, M., and Goswami, G., 2004. Exchange rate sensitivity of Japan’s bilateral trade flows, Japan and the World Economy 16, 1-15. Bahmani-Oskooee, M., and Ratha, A, 2008. Exchange rate sensitivity of US bilateral trade flows. Economic Systems 32, 129-141.

13

Beck, T., 2002. Financial development and international trade: Is there a link? Journal of International Economics 57, 107–131.

Berthou, A., 2007. Credit Constraints and Zero Trade Flows: the Role of Financial Development, Mimeo, University of Paris. Chang, Y., M. Hung, and C. Lu, 2005. Trade, R&D Spending and Financial Development, Applied Financial Economics 15 (11): 809–819. Chen, X., A. Munasib, and D. Roy, 2012. Financial Reforms and International Trade, IFPRI Discussion Paper 01182, International Food Policy Research Institute. Elliott, G., T. Rothenberg and J. Stock, 1996. Efficient Tests for an Autoregressive Unit Root, Econometrica 64, 813-836. Engel, R. and C. Granger. 1987. Cointegration and Error Correction: Representation, Estimation and Testing, Econometrica 66: 251-76. Fowowe, B., 2011. Financial Sector Reforms And Private Investment, In Sub-Saharan African Countries, Journal of Economic Development 36(3):79-97. Francois, J., and L. Schuknecht, 1999. Trade in Financial Services: Procompetitive Effects and Growth Performance. Hanif, M., F. Husain and S. Jafri, 2008. Financial Sector Reforms and International Trade Competitiveness: A Case Study of Pakistan, State Bank of Pakistan, Karachi, Pakistan.

Helpman, E., M. Melitz, and Y. Rubinstein, 2008. Estimating Trade Flows: Trading Partners and Trading Volumes, Quarterly Journal of Economics 123 (2): 441–487. Hsing, H., 1999 in Kamoto, E., 2006. The J-Curve Effect on the Trade Balance in Malawi and South Africa. MA Thesis, The University of Texas at Arlington. Hur, J., M. Raj, and Y. Riyanto, 2006. Finance and Trade: A Cross-country Empirical Analysis on the Impact of Financial Development and Asset Tangibility on International Trade, World Development 34 (10): 1728–1741. Jansen, M. and Y. Vennis, 2006, Liberalising Financial Services Trade in Africa: Going Regional and Multilateral, WTO Staff Working Paper ERSD-2006-03. Junz, H.B., and Romberg, R., 1973. Price competitiveness in export trade among industrial countries. American Economic Review 3, 314-327. Kwiatkowski, D., P. Phillips, P Schmidt, and Y. Shin. 1992. Testing the Null Hypothesis of Stationary against the Alternative of a Unit Root: How Sure Are We That Economic Time Series Have a Unit Root?, Journal of Econometrics, 54: 159-78. Lerner, A.P., 1944. The Economics of Control: Principles of welfare Economics. MacMillan: New York. Levine, R., Loayza, N., Beck, T., 2000. Financial intermediation and economic growth: Causality and Causes, Journal of Monetary Economics 46, 31–77.

14

Levine, R., Zervos, S., 1998. Stock markets, banks, and economic growth, American Economic Review 88, 537–558. Magee, S., 1973. Currency contracts, pass-through, and devaluation. Brookings Papers of Economic Activity 1, 303-323. Mangani, R., 2011. The Exchange Rate Sensitivity of Foreign Trade: Evidence from Malawi, Trade Policy Review 4:5-23. Manova, K, 2008b. Credit Constraints, Heterogeneous Firms, and International Trade, NBER Working Papers 14531. Cambridge, MA, US: National Bureau of Economic Research. Marshall, A., 1923. Money, Credit and Currency. MacMillan: London.

McKinnon R., 1973. Money and Capital in Economic Development. The Brookings Institution, Washington, D.C. Melitz, M., 2003. The Impact of Trade on Intra-industry Reallocations and Aggregate Industry Productivity, Econometrica 71 (6): 1695–1725. Odhiambo, N. 2011. The Impact of Financial Liberalisation in Developing Countries: Experiences from Four SADC Countries. Organisation for Social Science Research in Eastern and Southern Africa, Addis Ababa. Pesaran, H.M., Shin, Y., and Smith, R.J., 2001. Bounds testing approach to the analysis of level relationships. Journal of Applied Economics 16, 289-326. Shaw, E.S., 1973. Financial Deepening in Economic Development. New York: Oxford University Press.

15

Appendix 1: Plots of Exports, Imports and Private Credit Note: Export and Import are exports and imports as percentages of world totals; private credit is domestic credit to the private sector as percentage of GDP. The commencement of financial reforms is identified by the vertical line cutting through the plots.

0 5 10 15 20 25 30 35

0.000

0.020

0.040

0.060

0.080

1970

1975

1980

1985

1990

1995

2000

2005

2010

Private Cred

it (%

of G

DP)

Expo

rt, Impo

rt (%

of w

orld to

tals)

Botswana


0.00

5.00

10.00

15.00

20.00

25.00

0.000 0.002 0.004 0.006 0.008 0.010 0.012

1970

1975

1980

1985

1990

1995

2000

2005

2010

Private Cred

it (%

of G

DP)

Expo

rt, Impo

rt (%

of w

orld to

tals)

Burundi


0.00

2.00

4.00

6.00

8.00

0.000 0.050 0.100 0.150 0.200 0.250 0.300

1970

1975

1980

1985

1990

1995

2000

2005

2010

Private Cred

it (%

of G

DP)

Expo

rt, Impo

rt (%

of w

orld to

tals)

DR Congo


0.00

10.00

20.00

30.00

40.00

50.00

60.00

0.000

0.200

0.400

0.600

0.800

1.000 1970

1975

1980

1985

1990

1995

2000

2005

2010

Private Cred

it (%

of G

DP)

Expo

rt, Impo

rt (%

of w

orld to

tals)

Egypt


0.00

5.00

10.00

15.00

20.00

25.00

30.00

0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080

1970

1975

1980

1985

1990

1995

2000

2005

2010

Private Cred

it (%

of G

DP)

Expo

rt, Impo

rt (%

of w

orld to

tals)

Ethiopia


0.00

10.00

20.00

30.00

40.00

50.00

0.000

0.050

0.100

0.150

0.200

1970

1975

1980

1985

1990

1995

2000

2005

2010

Private Cred

it (%

of G

DP)

Expo

rt, Impo

rt (%

of w

orld to

tals)

Kenya


16

0

5

10

15

20

25

30

0.000

0.005

0.010

0.015

0.020

0.025

0.030

1970

1975

1980

1985

1990

1995

2000

2005

2010

Private Cred

it (%

of G

DP)

Expo

rt, Impo

rt (%

of w

orld to

tals)

Lesotho


0.00

5.00

10.00

15.00

20.00

25.00

0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070

1970

1975

1980

1985

1990

1995

2000

2005

2010

Private Cred

it (%

of G

DP)

Expo

rt, Impo

rt (%

of w

orld to

tals)

Madagascar


0.00

5.00

10.00

15.00

20.00

25.00

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

1970

1975

1980

1985

1990

1995

2000

2005

2010

Private Cred

it (%

of G

DP)

Expo

rt, Impo

rt (%

of w

orld to

tals)

Malawi


0.00

20.00

40.00

60.00

80.00

100.00

120.00

0.000

0.010

0.020

0.030

0.040

0.050

1970

1976

1982

1988

1994

2000

2006

2012 Pr

ivate Cred

it (%

of G

DP)

Expo

rt, Impo

rt (%

of w

orld to

tals)

MauriQus


0.00

2.00

4.00

6.00

8.00

10.00

12.00

0.000

0.005

0.010

0.015

0.020

1970

1975

1980

1985

1990

1995

2000

2005

2010 Private Cred

it (%

of G

DP)

Expo

rt, Impo

rt (%

of w

orld to

tals)

Rwanda


0.00

5.00

10.00

15.00

20.00

25.00

30.00

35.00

0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080

1970

1975

1980

1985

1990

1995

2000

2005

2010 Private Cred

it (%

of G

DP)

Expo

rt, Impo

rt (%

of w

orld to

tals)

Seychelles


17

0.00

50.00

100.00

150.00

200.00

0.000

0.200

0.400

0.600

0.800

1.000

1.200

1.400

1970

1975

1980

1985

1990

1995

2000

2005

2010 Private Cred

it (%

of G

DP)

Expo

rt, Impo

rt (%

of w

orld to

tals)

South Africa


0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00

0.000

0.020

0.040

0.060

0.080

0.100

0.120

1970

1975

1980

1985

1990

1995

2000

2005

2010 Private Cred

it (%

of G

DP)

Expo

rt, Impo

rt (%

of w

orld to

tals)

Sudan


0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00

0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035

1970

1975

1980

1985

1990

1995

2000

2005

2010 Private Cred

it (%

of G

DP)

Expo

rt, Impo

rt (%

of w

orld to

tals)

Swaziland


0.00

5.00

10.00

15.00

20.00

0.000

0.020

0.040

0.060

0.080

0.100

0.120

1970

1975

1980

1985

1990

1995

2000

2005

2010 Private Cred

it (%

of G

DP)

Expo

rt, Impo

rt (%

of w

orld to

tals)

Tanzania


0.00

5.00

10.00

15.00

20.00

0.000

0.020

0.040

0.060

0.080

0.100

1970

1975

1980

1985

1990

1995

2000

2005

2010

Private Cred

it (%

of G

DP)

Expo

rt, Impo

rt (%

of w

orld to

tals)

Uganda


18

Appendix 2: Unit Root Tests Note: Entries are absolute test statistics under the null hypothesis of a unit root. Statistics for the test in level are entered against “L”, while those for the test in first differences are entered against “D”. The 5% absolute critical value is 3.521. The L and D test statistics for GDPF are 1.715 and 1.549*. Rejection of the null hypothesis is marked with *. NA implies that the test is not applicable. Variables Country CREDIT EXPORT IMPORT EXRATE GDP Botswana L: 0.592

D: 4.594* L: 1.611 D: 7.301*

L: 2.010 D: 5.867*

L: 2.143 D: 5.414*

L: 2.197 D: 5.630*

Burundi L: 2.585 D: 7.050*

L: 1.443 D: 9.547*

L: 1.026 D: 3.775*

L: 2.710 D: 3.732*

L: 1.467 D: 5.366*

Congo, DR L: 0.961 D: 5.153*

L: 2.420 D: 3.720*

L: 2.067 D: 6.513*

L: 1.265 D: 5.639*

L: 1.228 D: 4.363*

Egypt L: 3.241 D: 4.748*

L: 2.206 D: 8.250*

L: 2.825 D: 5.844*

L: 3.083 D: 4.567*

L: 0.202 D: 3.605*

Ethiopia L: 2.844 D: 5.232*

L: 0.692 D: 6.934*

L: 1.225 D: 8.396*

L: 2.902 D: 4.532*

L: 2.123 D: 4.833*

Kenya L: 2.445 D: 7.953*

L: 2.383 D: 7.104*

L: 1.197 D: 8.890*

L: 1.038 D: 7.112*

L: 2.082 D: 4.129*

Lesotho L: 1.919 D: 6.916*

L: 2.966 D: 8.027*

L: 3.077 D: 5.589*

L: 2.242 D: 5.608*

L: 1.707 D: 5.632*

Madagascar L: 2.271 D: 4.726*

L: 2.556 D: 8.825*

L: 2.434 D: 7.299*

L: 1.711 D: 7.178*

L: 1.713 D: 7.418*

Malawi L: 0.875 D: 6.099*

L: 1.843 D: 8.302*

L: 2.473 D: 7.825*

L: 2.647 D: 5.409*

L: 1.500 D: 7.450*

Mauritius L: 2.360 D: 6.314*

L: 1.886 D: 6.812*

L: 3.357 D: 5.211*

L: 1.689 D: 5.373*

L: 0.644 D: 6.137*

Rwanda L: 1.157 D: 7.167*

L: 2.563 D: 8.471*

L: 2.889 D: 7.319*

L: 2.462 D: 6.689*

L: 1.676 D: 6.257*

Seychelles L: 1.383 D: 6.167*

L: 1.611 D: 7.301*

L: 3.980* NA

L: 3.956 NA

L: 3.851 NA

South Africa L: 0.756 D: 4.234*

L: 1.090 D: 5.575*

L: 3.768* NA

L: 3.405 D: 5.567*

L: 0.465 D: 4.414*

Sudan L: 0.563 D: 4.683*

L: 2.625 D: 4.051*

L: 1.539 D: 7.254*

L: 1.742 D: 6.264*

L: 1.211 D: 4.642*

Swaziland L: 2.197 D: 8.875*

L: 2.640 D: 5.601*

L: 2.305 D: 6.186*

L: 3.070 D: 5.630*

L: 3.025 D: 4.956*

Tanzania L: 2.095 D: 5.995*

L: 1.701 D: 7.235*

L: 0.863 D: 8.738*

L: 1.621 D: 3.402*

L: 1.544 D:3.660*

Uganda L: 0.412 D: 4.944*

L: 2.008 D: 5.091*

L: 3.094 D: 8.004*

L: 1.792 D: 4.430*

L: 2.577 D: 4.359*

19

Appendix 3: Diagnostic Tests Table 3: Phillips-Ouliaris Cointegration Test Results

Country Export Model Import Model BG(4) ARCH(4) RESET(3) BG(4) ARCH(4) RESET(3) Botswana 0.062 0.031* 0.543 0.328 0.270 0.546 Burundi 0.713 0.113 0.667 0.879 0.547 0.466 Congo, DR 0.335 0.324 0.897 0.467 0.223 0.992 Egypt 0.522 0.899 0.443 0.992 0.055 0.765 Ethiopia 0.661 0.765 0.785 0.578 0.048* 0.862 Kenya 0.091 0.777 0.895 0.897 0.066 0.765 Lesotho 0.067 0.252 0.291 0.892 0.567 0.889 Madagascar 0.231 0.452 0.234 0.894 0.097 0.263 Malawi 0.111 0.888 0.786 0.784 0.432 0.221 Mauritius 0.109 0.280 0.345 0.409 0.765 0.234 Rwanda 0.413 0.022* 0.770 0.203 0.023* 0.675 Seychelles 0.333 0.359 0.254 0.906 0.886 0.887 South Africa 0.761 0.011* 0.675 0.456 0.657 0.356 Sudan 0.342 0.268 0.998 0.752 0.768 0.876 Swaziland 0.422 0.222 0.279 0.502 0.855 0.345 Tanzania 0.120 0.245 0.357 0.231 0.432 0.773 Uganda 0.444 0.342 0.336 0.398 0.887 0.236 Note: Entries are p-values for accepting the null hypothesis of no0. 4th order serial correlation (BG(4)), not 4th order ARCH effects (ARCH(4)) and no quadratic and cubic terms (RESET(3). * denotes a rejection of the corresponding null hypothesis at 5% significance level.

do financial reforms and financial development...

Documents