Financial market structure and capital controls in Sub-Saharan Africa

Kofi Bondzie Afful (School of Economics and Finance, University of the Witwatersrand Johannesburg, Johannesburg, South Africa)
Tendai Gwatidzo (School of Economics and Finance, University of the Witwatersrand Johannesburg, Johannesburg, South Africa)
Mthokozisi Mlilo (School of Economics and Finance, University of the Witwatersrand Johannesburg, Johannesburg, South Africa)

Journal of Economic and Administrative Sciences

ISSN: 2054-6238

Article publication date: 25 November 2024

177

Abstract

Purpose

This study investigates the influence of capital controls on financial market structure in Sub-Saharan Africa (SSA). This is especially relevant as the former restrictions are relatively common on the sub-continent. At the same time, the sub-region’s financial markets are highly bank-based and focused on the short term, with stock markets being illiquid and stunted.

Design/methodology/approach

To achieve its research objectives, the study posits an original model and uses comparative statics to analyze the relation between the aforestated phenomena in a representative SSA economy. Key hypothesized conclusions derived therefrom are tested using panel econometrics.

Findings

The comparative static analysis illustrates that capital controls favor banks, making them monopolistic and inefficient. This is confirmed by the empirical investigation, as the said market restriction skews financial market structure towards a bank-dominated system.

Research limitations/implications

The study limits itself to capital controls and their effects on financial market structure. It does not particularly investigate the influence of different types of these restrictions. Specifically, it dichotomizes the influence of the examined controls on bank and stock markets.

Practical implications

The dissimilar influence of capital controls on banks relative to stock markets is critical for decision and policymakers. This paper highlights that capital controls may have unintended adverse effects on domestic financial markets. Also, they may not be the most appropriate policy to deepen markets and enhance domestic resource retention. There is, consequently, a need to determine fitting policies that attract rather than repel financial flows. Furthermore, capital controls may engender rather than address macroeconomic misalignment.

Social implications

As a social imperative, it is necessary to analyze SSA’s framework of capital restrictions to better understand how they distort market incentives and mechanisms. This would help identify adverse effects that retard social development.

Originality/value

This study extends existing literature by developing a novel analytical framework incorporating key characteristics of SSA economies. This helps to better understand the nature of the capital controls–financial market structure relation in imperfect market conditions.

Keywords

Citation

Bondzie Afful, K., Gwatidzo, T. and Mlilo, M. (2024), "Financial market structure and capital controls in Sub-Saharan Africa", Journal of Economic and Administrative Sciences, Vol. ahead-of-print No. ahead-of-print. https://doi.org/10.1108/JEAS-07-2023-0201

Publisher

:

Emerald Publishing Limited

Copyright © 2024, Kofi Bondzie Afful, Tendai Gwatidzo and Mthokozisi Mlilo

License

Published by Emerald Publishing Limited. This article is published under the Creative Commons Attribution (CC BY 4.0) licence. Anyone may reproduce, distribute, translate and create derivative works of this article (for both commercial and non-commercial purposes), subject to full attribution to the original publication and authors. The full terms of this licence may be seen at http://creativecommons.org/licences/by/4.0/legalcode


1. Introduction

A key tenet in international economics is freedom of movement of resources. Global market restrictions impair such flows. A common form of these limitations is capital controls. The latter is a relatively typical feature in Sub-Saharan Africa (SSA), despite most of these nations being comparatively small open economies. Between 1995 and 2020, about half of SSA’s 45 economies maintained a low-to-medium degree of capital restrictions (Sriram et al., 2020). About 10 of these may be described as highly restrictive (Ntembe, 2022). Approximately the same numbers, 10, are relatively financially open, namely: Botswana, Djibouti, Gambia, Kenya, Liberia, Mauritius, Sao Tome, Seychelles, Uganda and Zambia (Chinn and Ito, 2022; Fernández et al., 2021).

Juxtaposed with the above phenomenon is the sub-region’s anomalous financial market structure. It is characterized by bank-dominated systems focused on short-term financing (Nyantakyi and Sy, 2015; World Bank, 2022, 2023). Less than 25% of domestic financial resources have a maturity exceeding 10 years. Capital markets, especially stock exchanges, are incomplete, inefficient, illiquid and stunted (World Bank, 2022, 2023). As such, stock market financing is woefully inadequate.

The relation between capital controls and financial market structure is an area of study with scant literature. Concerning SSA, the literature is even more limited. Current studies on the mentioned topic do not particularly take into account the sub-continent’s anomalous financial market structure (Chokri, 2020a; Mutize et al., 2020; Pasricha et al., 2019; Rebucci and Ma, 2020). This study, consequently, answers the research question: What are the critical features of SSA’s capital control-financial market structure relation? Also, it investigates whether capital controls enhance or worsen the sub-region’s financial market structure aberration.

To answer the prior stated concerns, an original model is constructed and analyzed using comparative statics. The derived analytical framework is subsequently tested empirically. Regarding its organizational structure, the paper commences with an introduction followed by a review of relevant studies. The third section presents the novel framework. The data and research methodology employed are outlined in the fourth section, while the fifth segment presents and discusses the empirical results. The paper ends with a conclusion and summary.

2. Literature review

2.1 Capital controls–financial market structure relation

It is worth noting that the focus of this study is on capital controls and financial market structure rather than on banks and stock markets separately. Theories on the relation between the two may be categorized into interventionist and non-interventionist precepts. The non-interventionist schools of thought include the classical and parity frameworks. They theorize that markets are, generally, near perfect and will tend towards steady state equilibrium (Romer, 2019; Salvatore, 2013). Therefore, there is little or no need for capital controls (Aizenmann, 2019; Blanchard, 2018; Chokri, 2020b). Overall, they presume that banks and stock markets produce similar financial resources. These non-interventionist theories are silent on whether banks and stock markets compete or complement each other. According to them, therefore, financial market structure is irrelevant. These precepts agree that the said financial restrictions have immediate adverse effects, but such are almost instantly addressed. Consequently, capital controls do not affect financial market structure (Benigno et al., 2017; Graham et al., 2015; Schmitt-Grohe and Uribe, 2017).

The key interventionist precepts mainly consider capital controls and financial markets but not financial market structure. They include current account optimization, dilemma/trilemma dichotomy, Keynesian, macroeconomic stabilization, Mundell–Fleming, parity variants, portfolio balance and speculative attack theories (Graham et al., 2015; Iqbal, 2022; Magud et al., 2018; Montecino, 2015; Pasricha et al., 2019; Rebucci and Ma, 2020). The aforementioned precepts argue that financial systems are imperfect. One group, however, accepts that such flaws are short term, while the other cohort assumes they are long term (Krugman et al., 2014). Within the interventionist school of thought, some proponents find that the immediate effects of capital controls may be both positive and negative, resulting in either a J- or S-curve (Chokri, 2020a, b). Overall, interventionist precepts affirm that these restrictions affect the balance of payments, the international flow of goods and services as well as financial resources. These frameworks conclude that capital controls may temporarily minimize financial market swings and associated macroeconomic gyrations (Batini and Durand, 2020; Graham et al., 2015). Most of such theories see little or no difference between bank and stock market financing. Consequently, they theorize that capital controls have a uniform effect on both systems. Similar to their non-interventionist counterparts, they conclude that capital controls are irrelevant to financial market structure.

Despite the foregoing seeming uniformity within interventionist tenets, they differ with regard to the key transmission mechanism between capital controls and financial markets. In several of these precepts, the balance of payment, especially its current account, is the major medium connecting the former to the latter and, consequently, its structure. This is highlighted by the following concepts: current account optimization, dilemma/trilemma dichotomy and Keynesian theories (Graham et al., 2015; Magud et al., 2018; Montecino, 2015; Pasricha et al., 2019; Rebucci and Ma, 2020). An alternative transmission medium, espoused by parity and portfolio balance concepts, is the currency market. These tenets presume that capital controls affect asset markets, including financial resources. A third group, consisting of the Mundell–Fleming, parity variants and speculative attack concepts, combine both currency markets and current accounts as transmission mechanisms (Aizenmann, 2019; Alfaro et al., 2018; Chokri, 2020a, b; Magud et al., 2018). Despite the differences among interventionist proponents regarding the said transmission mechanisms, they are similar to non-interventionist theories, as they also do not particularly consider financial market structure.

2.2 Financial market structure in SSA

Financial market structure refers to the relative mix of banks versus stock markets in a macroeconomy. It influences the characteristics, transaction costs and type of predominant financial instruments (Bencivenga et al., 1996; Williamson and Tadelis, 2012). It is a critical feature that influences available tradeoffs and returns of financial assets. Specifically, it affects financial contracts, institutions and instruments (Cihak et al., 2012; Demirguc-Kunt and Levine, 2002, 2011).

Bank-dominated systems, one form of financial market structure, are characterized by debt financing, whether short or long term (Nyasha and Odhiambo, 2017). These institutions tend to have a more intimate relation with firms they lend to in order to minimize monitoring and other transaction costs (Boot and Thakor, 1997; Demirguc-Kunt and Maksimovic, 1996, 2011; Williamson and Tadelis, 2012). In this manner, they are able to offer more personalized assistance, products, resources and services. Market-based systems, on the other hand, differ as they provide more standardized contracts and instruments (Demirguc-Kunt and Levine, 2002). As a comparative advantage, they provide access to a greater array of market investors that are more impartial and removed (Ho and Iyke, 2017; Mutize et al., 2020).

SSA’s financial market structure is an aberration of orthodox theories, as it is bank-dominated and focused on the short end (Allen et al., 2011; Boot and Thakor, 1997; Demirguc-Kunt and Levine, 2012). Less than 6% of loans provided extend longer than 7–8 years. About 23% of financial advances and loans have a maturity between 3 and 6 years (Nyantakyi. and Sy, 2015; World Bank, 2022; World Bank, 2023). A little more than 70% of these resources cover 0–3 years. Long-term domestic financing exceeding 10 years is relatively scarce (Soumaré et al., 2021).

The second anomaly of SSA’s financial market structure is its severely underdeveloped capital markets, especially stock exchanges. Out of its 45 economies, there are only about 16 active functioning markets. Most of these have an average market capitalization to gross domestic product (GDP) ratio less than 30%. In comparison, in developed and emerging bank-dominated economies, their same ratios exceed 50%. Overall, capital financing provided by these systems is less than a third of domestic credit offered by the banking sector (World Bank, 2022, 2023). Additionally, while developed stock markets are relatively efficient and liquid, in SSA they are severely inefficient, illiquid and stunted.

The third facet of SSA’s deviant financial market structure is demonstrated by its incompleteness, especially at the long end. The sub-continent’s yield curve and supply of long-term financing are discontinuous (Mutize et al., 2020). Nigeria and South Africa have the main operational bond markets on the sub-continent. As well, the only active derivatives market is in South Africa (World Bank, 2022, 2023).

2.3 Research gap

From the literature reviewed, a research gap is identified. Firstly, there is a dearth of studies that critically examine the capital controls–financial market structure relation. Secondly, there is a scarcity of studies focused on SSA. This is highly relevant as the mentioned sub-region is characterized by moderate-to-high capital controls as well as anomalous financial market structures. This study, therefore, addresses the stated research gap.

3. Analytical model

3.1 Baseline model: no capital controls

To critically analyze the phenomena under study, a baseline framework without the mentioned restrictions is first developed. This is extended by subsequently including capital controls for further analysis. Herein, a financial resource synonymously refers to a contract or instrument. The representative macroeconomy is founded on disequilibrium theory, where, YitE, aggregate expenditure at time t exceeds national income, Yit, in the same time period (Schoder, 2019). The output gap is denoted as YitD+. Let Yi* symbolize domestic full employment output, while Yi,t1 is last year’s aggregate income. Then, YitD+ = Yi*Yi,t1 at the beginning of time t. In time t, a portion, αitY, of the said gap is financed by domestic, Γit, and foreign, Γjt, financing, respectively. Presume therefore that αitYYitD+ = YitEYit = Γit + Γjt.

If Γb,it and ΓM,it are domestic banking sector and stock market financing, respectively, then Γit = Γb,it + ΓM,it. As country i only produces short-term financial resources, then Γb,it and ΓM,it have the same limited maturity. Long-term finance is only available from country j, the characteristic advanced foreign economy. The form of external financing is ignored for ease of analysis. Suppose that Yj* and Yjt are income at full employment and time t, respectively, in economy j. Let Yj*Yjt denote the potential–actual output deficit of country j. As such, Yi*Yit are similarly defined for economy i. The direct exchange rate between economies i and j at time t is defined as follows:

(1)Yi*YitYj*Yjt=eij,t

According to Equation (1), eij,t is a function of the ratio of full employment-current output gap in country i to that of economy j. It suggests that a rise eij,t is due to an increase in the said deficit of country i or the vice versa condition in economy j. As both countries trade only between themselves, their excess demand is met by consuming more of the other economy’s output. This is in tandem with orthodox macroeconomic theory (Blanchard, 2018; Romer, 2019).

Assume imports, Zi, outstrip exports Xi in country i based on the precondition that both variables exceed zero. Γit+ is additional domestic supply of financial resources introduced into the domestic economy at time t. Presume that Γi,t1 is the previous year’s domestic finance, while Γit is domestic outflow to country j. Alternatively: Γit = Γit+ + Γi,t1 + Γit. Let δit be the proportion of Yt1 used to finance Γit+.

(2a)YtE=αtYYi*+(1αtY)Yit=(XitZit)+(1+δit)Yit+Γi,t1
(2b)Yit=(αtY+δit)1[(αtYYi*(XitZit)Γi,t1]

If there is no foreign financing, then YtEYt = αtY (Yi*Yi,t1) = Γit. In this instance, capital controls have little or no effect on the financial system. Equation (2a) conjectures that domestic spending is affected by the current account, previous income and present financing. Equation (2b) postulates that current macro output is inversely related to the sum of the current year’s fraction of the output gap financed, αtY, as well as the proportion of last year’s output injected as additional domestic financial resources at time t.

As country j is close to steady state, let a change in Yj*Yjt denoted as Δ (Yj*Yjt) be relatively small but greater than zero. According to Equation (3) below, domestic financing of excess expenditure may be approximated by the sum of local currency value of country j’s output gap and domestic expenditure less than the totality of full employment output of country i and foreign financing of local excess demand. Equation (3) is underpinned by two preconditions: Γb,it + ΓM,it > 0 and Yi* > YitE. As such, eij,t (Yj*Yjt) > Yi* + Γjt. This implies that the domestic currency value of the output gap of country j exceeds the combined value of local full employment output and external financing for country i.

(3)Γb,it+ΓM,iteij,t(Yj*Yjt)+YitE[Yi*+Γjt]

In the absence of capital controls, financing for country i’s excess expenditure does not affect its financial market structure. Rather, it enables it to source the needed funding externally, as local resources are inadequate.

3.2 Capital controls

This segment investigates the effects of capital controls on financial market structure. Capital restrictions are imposed at the latter end of time t. Its effects occur with a year’s lag at the end of t+1. Presume that the said capital controls are quota-type restrictions, a form of market ceiling. Γ̅i is the fixed maximum stipulated capital outflow. It raises the cost of international flows between countries, distorting financial market incentives (Schmitt-Grohe and Uribe, 2017).

P1.

This study proffers a financial instruments–capital controls hypothesis that extends the tenets of asset specificity and uncertainty within transaction costs theory (Williamson, 1998; Williamson and Tadelis, 2012). Let νixb be the value of bank financial instrument x in country i, while νixM is the analogous securitized non-collateral stock market contract. The value of each financial asset is comprised of two parts. The first portion, the commodity or tangible portion, is symbolized by an additional subscript k. For bank and market instruments, these are represented by νixb,k and νixM,k, respectively. The second part, the value of financial rights, is the intangible facet symbolized by subscript l. The non-physical value of bank and stock market resources are denoted by νixb,l and νixM,l accordingly (Bencivenga et al., 1996; Williamson, 1998). Therefore:

(4a)νixb=νixb,k+νixb,l
(4b)νixM=νixM,k+νixM,l

Presuppose that νixb,l and νixM,k are asymptotic to but never equal to 0. Thus, for instruments issued by banks, their tangible value significantly exceeds the alternative. Assume the opposite for stock market resources.

Let wib,l and wiM,l be the degree or weight of intangibility for each respective bank and market instrument, such that wib,l = 1 − wib,k and wiM,l = 1 − wiM,k. Neither wib,l nor wiM,l equal zero. However, wib,k = νixb,k/νixb and wiM,k = νixM,k/νixM. These assumptions presume that there exists a tradeoff between the commodity and financial rights values of an instrument.

It is hypothesized that under capital controls cixb, the cost of estimating νixb remains the same as before the imposition of the said restriction. On the other hand, cixM, the analogous cost for a stock market resource is a direct function of the intensity of the imposed restrictions, f(Γ̅i), as well as ζΓ̅i, the lost financial flows occurring as result of the instituted capital controls. This is depicted in Equation (5). Both f(Γ̅i) and ζΓ̅i are implicit positive functions of wiM,l. Given the condition that cixM > 0, then:

(5)cixM=f(Γ̅i)+ζΓ̅i

As such, vixbc and vixMc, the respective net values of bank and stock market instruments are as follows:

(6a)vixbc=νixb
(6b)vixMc=νixMcixM
Combining Equations (4a), (4b), (5), (6a) and (b) implies that when Γ̅i > 0, then vixbc > vixMc. This concludes that capital controls adversely affect the value of intangible financial instruments, including those issued by stock markets.

Γb,it+1 is bank financing at time t+1 after imposition of the examined market restrictions. ΓM,it+1 and Γj,t+1 are similarly defined for stock market and foreign financing. ϕbi is sensitivity of Γb,it+1 to Γ̅i computed as ϕbi = (νixb,kνixb,l), as depicted in Formulation (7a). ΓM,it+1 and Γj,t+1 are computed in a similar manner, as illustrated in Equations (7b) and (7c).

(7a)Γb,it+1=Γb,it(1+ϕbiΓ̅i);0<ϕbi<1
(7b)ΓM,it+1=ΓM,it(1+ϕMiΓ̅i);0<ϕMi1
(7c)Γj,t+1=ΓM,it(1+ϕjiΓ̅i);0<ϕji1

Although Equations (7a)–(7c) are derived from preceding axioms, Formulation (7a) postulates that capital controls increases both the value and quantity of bank instruments. The key reason is because financial assets with greater commodity value are a more certain store of value in uncertain environments engendered by capital controls. The opposite relation holds for the two alternative instruments, namely, stock market and foreign financing. A consequence of this is a monopolistic banking system characterized by higher returns ■

For ease of analysis, let ϕbi + ϕMi + ϕji = 1 such that | ϕbi | = |ϕMi + ϕji |. This implies that changes in foreign and stock market financing are offset by bank funding. As ϕMi + ϕji < 0, then ϕMi + ΔΓM,it+1 = − (ΔΓM,it+1). This based on the assumption that Γj,t+1Γjt < 0 and Γi,t+1Γit > 0. This implies that after the institution of capital controls, the increase in local financing makes up for the decrease in stock and foreign financing. Formally, this may be stated as |(Γb,it+1Γb,it)| = |(ΓM,it+1ΓM,it) + (Γj,t+1Γjt)|, whereas (ΓM,it+1ΓM,it) + (Γj,t+1Γjt) < 0. The reason is because the imposed restrictions discourage potential foreign resource inflows, as external investors cannot freely repatriate their domestic earnings due to a limitation on capital outflows.

Formulation (8) suggests that a change in country i’s excess demand at time t+1, the left-hand side of the equation, is a function of capital controls as well as the sensitivity of the three sources of funding to the stated market restriction.

(8)Δ[Yi,t+1EYi,t+1]=Γ̅i(ϕbiΓb,itϕMiΓM,itϕjiΓM,it)

From prior formulations, the reaction function, i.e. Equation (9), is derived. It argues that foreign financing after the imposition of capital controls is a function of four components. The first is the domestic economy’s past year’s excess demand YitEYit. The second is the ratio of ex-post excess demand after capital controls to the sum of one and the product of capital control and its effect on local bank instruments. The third determinant is the value of the previous year’s domestic market financing. The final elements are the magnitude of capital controls and the sensitivity of foreign funding to the examined market intervention.

(9)Γj,t+1=[(YitEYit)(Yi,t+1EYi,t+1)(1+ϕbiΓ̅)2ΓM,it](1+ϕjΓ̅)2

The foregoing analysis suggests that capital controls do not affect banks and stock markets equally. Rather, they adversely impact foreign financial and domestic stock market instruments. In the case of SSA, it appears that these restrictions favor bank issued financing. The latter may be used as a substitute for lost foreign long-term inflows. Consequently, capital controls magnify bank dominance. Currency markets may not be immune, as the said financial restrictions could alter YijdYjid.

4. Data and empirical methodology

4.1 Data

The empirical analysis uses annual data between 1995 and 2021. Secondary data on all variables excluding the capital control indicators were obtained from World Bank (2022) and World Bank (2023). The study selected these years as a result of missing data and the fact that most SSA stock markets were established at the latter end of the 1990s.

The dataset is originally an unbalanced panel. This is because of the unsystematic nature of unavailable data for some of the sampled countries. Using data imputation processes, the compiled data are enhanced to a balanced panel (Austin et al., 2021; Eddings and Machenko, 2012; De Silva, De Livera, Lee and Moreno-Betancur, 2020; Nakai and Weiming, 2011).

To evaluate potential biases with data imputation, the study computes a set of rigorous pre-regression diagnostic tests. The pre-regression analysis enables the study to determine if said data enhancement introduced potential biases or weaknesses that are critical to the empirical analysis and findings. The results of the diagnostic tests are subsequently presented and discussed. The study uses STATA 18 for its empirical analysis.

Data on capital controls are compiled from Chinn and Ito (2022). Missing data preclude sampling all 45 SSA economies (World Bank, 2022, 2023). Although Fernández et al. (2021) develop an alternative capital controls dataset, it is not used in this study because it has more missing data for the sampled economies. The Chinn-Ito capital controls index ranges between 2.31 and −2.31. A value of 2.31 implies almost perfect capital mobility. At the other extreme, −2.31 denotes complete capital restriction.

The sampled countries are listed in Table 1. Of the analyzed economies, Botswana, Gambia, Kenya, Mauritius, Uganda and Zambia have relatively low capital restrictions, as their average Chinn-Ito indices are greater than zero. For the remaining 29, their average capital controls indicator is less than 0, highlighting prevalence of these restrictions on the sub-continent. Of these, only four of have a mean indicator between 0 and −1, namely, Cape Verde, Madagascar, Nigeria and Rwanda. Majority of the outstanding 25 have a mean index between −1 and −1.70.

From Table 1, only nine of the sampled economies have a standard deviation greater than 0.5. Even so, these do not markedly exceed 1. This implies that the intensity of capital controls in SSA does not significantly change over time. This is corroborated by the low coefficient of variation and trend indicators.

In 29 of the examined countries, about 83% of the sampled SSA countries, capital controls had a negative trend, meaning they became more restrictive over time. Such restrictions may have been implemented to address macroeconomic and structural misalignments arising from comparatively weak currencies, high net imports and significant foreign currency components of public debt (Batini and Durand, 2020; Sriram et al., 2020). These restrictions may be intended to minimize the adverse effects of currency mismatch, debt unsustainability and high indebtedness (Batini and Durand, 2020; Blanchard, 2018; Montecino, 2015).

Financial market structure, ϑbMi, is computed as ΓbiΓMiΓMi+Γbi. The literature notes that there are three key dimensions of financial market structure, namely, activity, efficiency and size. In more developed countries with available data, a composite measure integrating all three components may be computed as the aggregate financial market structure variable. In the case of SSA, however, data on the first two aspects, activity and efficiency, are highly limited. As a consequence, this study uses data on the size dimension of financial market structure.

Following the work of Demirguc-Kunt and Levine (2002) and Levine (2002), Γb,it and ΓM,it are proxied by the bank credit-to-GDP ratio and market capitalization ratio, respectively. Using data obtained from the World Bank (2022) and the World Bank (2023) covering 1995–2021, the study finds that all the examined countries in SSA, excluding South Africa, are bank-based. This includes those with active stock markets. For countries without a stock market, ϑbMi = 1, denoting complete bank dominance (World Bank, 2022, 2023).

Several studies have considered the influence of other variables on financial market structure, including governance and institutional factors as well as macroeconomic variables. These include Allen et al. (2015), Demirguc-Kunt and Levine (2012), Ho and Iyke (2017) and Nyasha and Odhiambo (2017). As the study concentrates mainly on the effects of capital controls on the regressand, it subsumes the influence of such potential regressors in the residual error term.

4.2 Empirical methodology

4.2.1 Hypothesis tests

A key conclusion derived from the comparative statics analysis is that capital controls encourage banks but discourage stock markets. As a first test of validity, a Pearson correlation statistic, r, is computed between capital controls and financial market structure. r2 is the square of r. The computed parameter is −0.21. Formulation (10) is used to convert the correlation to a t-statistic, where n is the sample size. The t-statistic for the correlation coefficient is −6.23. The remaining hypothesis test results are presented in Table 2. As a preliminary result, it confirms that capital controls stimulates excess bank dominance.

(10)tstatistic=rn21r2

To test whether capital controls raise bank returns, the study uses data on return on assets (ROA) and return on equity (ROE) for objectivity. It tests the null hypothesis that sampled SSA countries with negative (restrictive) capital control indices have greater bank returns than the opposite cohort – economies with little or no capital controls, signified by economies with a zero or positive capital controls index. Sample 1 is the group with negative capital controls indices, while Sample 2 is the set with a zero or positive index. The mean ROA and ROE indicators for the two groups are computed. These are transformed to a t-statistic with associated pooled standard deviations using Equations (11a) and (11b) accordingly. The mean ROA or ROE for Samples 1 and 2 are individually represented by τ̅1 and τ̅2. The related degrees of freedom are derived as n1+n22 where n1 and n2 are the respective sizes of Sample 1 and Sample 2. s12 and s22 are the corresponding variance for Samples 1 and 2, while sp2 is the pooled sample variance of the two groups.

(11a)t=τ̅1τ2̅((sp2((1/n1)+(1/n2)))1/2
(11b)sp=((s12*(n11)+(s22*(n21))n1+n22

The hypothesis test results in Table 2 illustrate that when ROA is used as the return parameter, the resulting test statistic, 1.14, does not reject the null hypothesis that capital controls do not alter bank returns. On the other hand, using ROE, the associated test statistic of 5.58 has a p-value of 0.00. This suggests that in economies with capital controls, banks have higher returns. The contrasting results from the ROA and ROE parameters are inconclusive.

To overcome the aforementioned problem, the study tests the next hypothesis. It infers that banks with capital controls tend to operate in a monopolistic environment. Consequently, they will be less efficient, having higher cost-income ratios. The associated test statistic, 13.29, has a p-value of 0.00. This avers that capital controls foster higher cost-income ratios. This confirms that capital controls encourage banks at the expense of stock markets. Missing data preclude completing similar tests for the latter financial systems in SSA.

4.2.2 Regression analysis

The study employs parametric panel econometrics for the remaining portion of its empirical analysis because of its underlying research objectives as well as the features of the data employed. The methodology comprises of pre-regression diagnostic tests, panel regression and post-regression analysis (Baltagi, 2013; Biorn, 2017; Wooldridge, 2010). The following pre-regression diagnostic test statistics are computed: cointegration, multi-collinearity and stationarity. The panel regression employs a generalized method of moments (GMM) methodology because of its rigor and robustness when there are likely endogenous regressors in the model (Greene, 2017; Kivet et al., 2016; Pesaran, 2015). The regressor coefficients, the beta coefficient, are: β1, β2, β3, β4, β5 , β6. They are the respective coefficients of the examined regressors included in the regression model.

(12)ϑbMit=β1Γit+β2eijt+β3πt+β4Yitθ+β5ψit+β6ZitXit+ɛit

The econometric regression is estimated from Equation (12). In the said formulation, the dependent variable, ϑbMit, is posited as a function of capital controls, Γ̅it; dollar exchange rate, eijt; inflation rate, πt; real GDP per capita, Yitθ; size of the informal sector, ψit and terms of trade, ZitXit. The latter is estimated as the ratio of imports to exports. ɛit is the associated residual error term. Table 3 below summarizes the variables employed in the regression analysis.

5. Analysis, findings and results

5.1 Pre-regression diagnostic tests

5.1.1 Multi-collinearity tests

The multi-collinearity test statistics are reported in Table 4. The key indicators are condition index, eigenvalue, r-squared, tolerance factor, variance inflation factor (VIF) and square root of the VIF (Baltagi, 2013; Biorn, 2017; Pesaran, 2015; Wooldridge, 2010). The VIF parameters are all less than 2, far below the critical value of 10. The remaining multi-collinearity indicators are also relatively low, highlighting no multi-collinearity.

5.1.2 Panel stationarity tests

To investigate stationarity, the study computes the Levin et al. (2002) and Fisher-type unit-root test statistics of the examined variables (Baltagi, 2013; Biorn, 2017; Wooldridge, 2010). Level data of the dollar exchange rate, GDP per capita, inflation and terms of trade are non-stationary. Using the first-differenced form of the data, consequently, the first-differenced version of the data is tested. The results in Table 5 conclude that the latter dataset is stationary. It is rather used for the regression.

5.1.3 Panel cointegration test

The computed Kao (1999), Pedroni (2004) and Westerlund (2007) cointegration test statistics are presented in Table 6. They are similarly based on the first differences, as the level variables were not stationary. Underlying the test is the null hypothesis that the panel is not cointegrated, while the alternative asserts the opposite. The test results presented in Table 6 confirms that the model has a cointegrating relationship.

5.2 Panel regression

Table 7 presents the GMM regression results. In confirmation of the aforestated hypothesis, the paper finds that capital controls skew economies towards bank-based systems. It confirms the supposition that such restrictions distort financial flows from foreign and stock market sources. They may foster a monopolistic environment for local banks. This may be because capital controls add to existing uncertainties, making it difficult to evaluate the worth of financial instruments with relatively high intangible values. This finding is supported by Soumaré et al. (2021).

A high dollar exchange rate is also found to tilt SSA economies towards bank dominance. Parity relations between currency and associated asset markets are possibly disturbed in such conditions. The key effects of the noted exchange rate are twofold. It adversely affects goods and financial markets (Chokri, 2020b). In the product economy, output is restricted, as the supply of needed capital is limited (Romer, 2019). Although a weak local currency encourages exports, it worsens its trade balance in SSA, as the sub-region is a net importer. According to Alfaro et al. (2018) and Salvatore (2013), depreciated domestic currency alters parity relations in the financial sector. Additionally, it raises local interest rates. Foreign financing becomes relatively expensive. This empirical finding adds another dimension to the relation between currency markets and financial systems, as espoused by Cavusoglu et al. (2019), Mundell (2000) and Montecino (2015).

Inflation, on the other hand, was found to encourage a market-based economy in SSA. Alternatively, it discourages bank-dominated systems. Romer (2019) and Salvatore (2013) aver that this is because of the fixed income characteristics of bank financing. They do not compensate for inflation. On the other hand, stock market instruments may be able to pass on price increases better than banks, as highlighted by Mutize et al. (2020). For one thing, firms, listed or unlisted, may increase their product or service price margins as general prices rise. Shares and related financial instruments, therefore, easily adjust to extenuating market conditions (Soumaré et al., 2021). These results are in tandem with the work of Ho and Iyke (2017). Overall, empirical evidence on the relation between inflation and stock markets is mixed.

From the results, rising GDP per capita encourages bank dominance. In SSA, the former may increase disposable income and spending. This would increase bank deposits by customers. Banks then use such deposits to finance their expanded lending. Added to this, rising incomes may raise bank borrowings. On the other hand, stock market investments remain limited on the sub-continent, as share trading remains an almost elitist activity (Nyasha and Odhiambo, 2017; World Bank, 2023). This result disagrees with mainstream theory (Allen et al., 2015; Cihak et al., 2012; Demirguc-Kunt and Levine, 2012; Graham et al., 2015).

The following independent variables had no effect on the regressand, size of the informal sector and the terms of trade. From the Wald statistic, it may be inferred that the regression model is appropriately specified and fits the data. The Sargan test results highlight that the utilized instruments are valid. There were no first- or second-order auto-correlations.

6. Conclusion and summary

This study investigated the influence of capital controls on financial market structure in SSA. Taking into account the unique characteristics of both variables on the sub-continent, an original analytical framework was developed. Premised on its underlying assumptions, it finds that capital controls encourage banks but deter securitized and foreign financial instruments. The first part of the empirical analysis confirms the supposition that capital controls engender banks, making them more inefficient. The regression results highlight that the said restrictions skew financial markets towards bank-based systems.

As a policy implication, this study, therefore, corroborates the recommendation of Batini and Durand (2020) that capital controls are better utilized as a temporary policy instrument to minimize exchange rate shocks and related anomalies. When implemented permanently, they tend to have significantly adverse consequences by distorting market mechanisms. Most of SSA has maintained these controls for more than a decade, making them an almost perpetual phenomenon. The sub-region, as a result, must critically evaluate the success of these restrictions thus far. This can be done through research that investigates different facets of the complex capital controls–financial markets relation. Furthermore, their immediate, medium and long-term effects should be considered as demonstrated by the J- and S-curve phenomena. More studies on this topic can assist decision and policymakers.

Another area of research interest is the exploration of the effects of different forms of capital restrictions on selected economic variables. One question that may be considered is whether the gradual but selected removal of specific capital controls may be more optimal compared to their immediate elimination. Alternatively, instead of a general application, perhaps if these restrictions are targeted at specific financial flows, they might be more potent at addressing particular macroeconomic misalignment with minimal adverse effects.

One of the underlying justifications of capital controls in SSA was to stem capital flight. Yet, these same governments do not appear to have provided adequate and appropriate incentives to enhance retention of domestic resources. Possibly, strategies of the latter nature may be more effective at discouraging capital outflows. There is, consequently, a need to determine fitting policies that attract more than repel financial flows. This may require a more holistic and integrated approach, as outlined by Sriram et al. (2020), where prudent macroeconomic management and increased political stability may be more effective. Other studies argue for policies directed at more specific financial flows and stakeholders.

For policymakers, a more critical analysis of SSA’s framework of economic policies is required. This is because although a policy might originally be intended to promote economic activity, it may actually distort market mechanisms. Examples include regulations targeted at attracting foreign direct investment but simultaneously restricting outward financial flows. This uncoordinated ad-hoc approach to addressing deep and multi-dimensional macroeconomic concerns may aggravate existing anomalous conditions. Further research could help identify such policies. Additional studies to explore such concerns can help untangle the current mix of regulatory frameworks with contrasting objectives.

Insights gleaned from the results herein may be incorporated into extended research to design a more robust financial market structure. This can be combined with an array of scenario analysis to critically evaluate the potential effects of hypothetical as well as actual market interventions. The disparate effects of capital controls on banks and stock markets highlight the need for further analysis on market controls and financial instruments.

As noted by an anonymous reviewer, this study focused particularly on SSA, such that its findings cannot be completely generalized. However, it highlights possible concerns in the sub-region about the interrelation between different financial instruments and economic policies. It therefore highlights research opportunities about the complex nature of these variables.

Due to unavailable data, this study was unable to explore the “trilemma” concern related to capital controls. This is another potential research area where a richer and robust dataset could aid in critical investigation of the effects of balance of payments, capital controls, exchange rates and monetary policy acting together. Trilemma research in SSA and its effects on financial markets are, therefore, open to further exploration.

Furthermore, other areas for further research raised by an anonymous reviewer include the introduction of other independent variables and other types of financial markets. These can be considered in future research. They are critical opportunities to provide additional insight and understanding about the relation between capital controls and financial markets.

Several SSA countries are members of different economic groupings. Restricting asset flows from non-member states may be a stipulated condition. The effects of these on domestic financial systems need to be explored in the literature. It is hoped that this study will stimulate further studies in this area of scholarly inquiry.

Descriptive statistics – capital controls in SSA

CountryMeanMedianSt. DevTrendCV
1. Angola−1.63−1.930.35−1.21−0.21
2. Benin−1.19−1.230.21−1.08−0.18
3. Botswana1.622.311.110.190.69
4. Burkina Faso−1.19−1.230.21−1.08−0.18
5. Burundi−1.66−1.930.34−1.22−0.21
6. Cape Verde−0.83−1.231.01−1.84−1.21
7. Central African Republic−1.19−1.230.21−1.08−0.18
8. Chad−1.19−1.230.21−1.08−0.18
9. Comorros−1.23−1.230.00−1.230.00
10. Congo (Democratic Republic)−1.19−1.230.10−1.09−0.10
11. Guinea−1.69−1.930.34−1.25−0.20
12. Guinea-Bissau−1.24−1.230.19−1.28−0.16
13. Eriterea−1.46−1.930.58−0.70−0.40
14. Eswatini−1.19−1.230.21−1.08−0.18
15. Ethiopia−1.23−1.230.00−1.230.00
16. Gabon−1.19−1.230.21−1.08−0.18
17. Gambia2.312.310.002.320.00
18. Ghana−1.43−1.230.42−0.87−0.30
19. Kenya0.931.040.580.610.63
20. Lesotho−1.22−1.230.07−1.18−0.06
21. Madagascar−0.53−0.160.49−0.21−0.93
22. Malawi−1.38−1.230.39−1.20−0.29
23. Mali−1.19−1.230.21−1.08−0.18
24. Mauritania−1.19−1.230.21−1.09−0.18
25. Mauritius1.441.200.681.490.47
26. Mozambique−1.23−1.230.00−1.230.00
27. Nigeria−0.85−0.660.39−1.29−0.46
28. Rwanda−0.34−1.231.13−1.83−3.35
29. Senegal−1.19−1.230.21−1.08−0.18
30. Sierra Leone−1.49−1.580.53−1.41−0.36
31. South Africa−1.19−1.230.21−1.08−0.18
32. Tanzania−1.21−1.230.10−1.16−0.08
33. Togo−1.23−1.230.00−1.230.00
34. Uganda1.982.310.781.160.39
35. Zambia2.002.310.891.200.45

Note(s): ST.DEV and CV denote standard deviation and coefficient of variation

Source(s): Authors' own computation from Chinn and Ito (2022)

Hypothesis test statistics and test results

Hypothesis test (null)Test statisticp-value
1. H0: Correlation between capital controls and financial market structure is zero−6.230.00
2. H0: ROA for countries with negative capital control indices is less than the same indicator for economies with positive indicators1.140.56
3. H0: ROE for countries with negative capital control indices is less than the same indicator for economies with positive indicators5.580.00
4. H0: Cost-income ratio for countries with negative capital control indices is less than the same indicator for economies with positive indicators13.290.00

Variables in empirical analysis

VariableType of variable
Financial market structureDependent variable
Capital control indexControl/Independent variable
Dollar exchange rateControl/independent variable
Inflation rateControl/independent variable
Real GDP per capitaControl/independent variable
Size of informal sectorControl/independent variable
Terms of tradeControl/independent variable

Multi-collinearity test results

VariableVIFVIF SQTFR-SQ.E.VC.I
GDP per capita1.401.180.60.500.631.33
Financial market structure1.261.120.740.240.281.28
Size of informal sector1.241.110.760.290.511.22
Inflation1.211.100.790.200.531.53
Terms of trade1.201.100.80.270.621.61
Dollar exchange rate1.161.080.840.520.291.29

Note(s): VIF SQ.; TF; R.SQ; E.V and C.I denote VIF square root, tolerance factor, R-squared, eigenvalue and condition index, respectively

Source(s): Authors' own computation

Panel stationarity tests

VariableLLC statisticp-valueFisher-type statisticp-value
Kaopen−8.48690.00−11.92670.00
GDP per capita44.00060.0027.390.00
Financial market structure−3.210.00−4.120.00
Size of informal sector28.020.00−4.320.00
Inflation8.41870.006.330.00
Terms of trade−27.710.00−29.160.00
Dollar exchange rate−14.02490.00−4.81380.00

Source(s): Authors' own computation

Panel cointegration tests

Cointegration testTest statisticp-value
Westerlund cointegration test−6.400.00
Kao cointegration test−45.810.00
Pedroni cointegration test−17.920.00

Source(s): Authors' own computation

GMM panel regression

VariablesCoefficientStandard errorZ-statisticp-value
Capital controls index (lag)−0.03250.0158−2.060.04
Dollar exchange rate−0.00010.0001−7.170.00
Inflation0.00020.00012.270.02
Real GDP per capita−0.00050.0000−3.490.00
Size of informal sector−0.00130.0012−0.980.33
Terms of trade−0.00310.006−0.560.58
GMM post-regression diagnostic test statisticValue
Wald statistic80.88
p-value0.00
Sargan test (H0: over-identifying restrictions are valid)3.571
p-value0.69
Arellano–Bond first order test (H0: no autocorrelation)−1.29
p-value0.23
Arellano–Bond second order test (H0: no autocorrelation)−1.49
p-value0.14
Number of observations858

Source(s): Authors' own computation

Conflict of interest/Funding statement: This statement is to indicate that the authors have no conflict of interest to report. As well, no funding or sponsorship was provided by any external entity for this research paper.

References

Aizenmann, J. (2019), “A modern reincarnation of Mundell-Fleming's trilemma”, Economic Modelling, Vol. 81, pp. 444-454, doi: 10.1016/j.econmod.2018.03.008.

Alfaro, L., Chari, A. and Kanczuk, F. (2018), “The real effects of capital controls: firm-level evidence from a policy experiment”, Journal of International Economics, Vol. 108, pp. 191-210, doi: 10.1016/j.jinteco.2017.06.004.

Allen, F., Otchere, I. and Senbet, L.W. (2011), “African financial systems: a review”, Review of Development Finance, Vol. 30 No. 3, pp. 1-35, doi: 10.1016/j.rdf.2011.03.003.

Allen, F., Carletti, E., Cull, R., Qian, J., Senbet, L. and Valenzuela, P. (2015), “The African financial development and financial inclusion gaps”, Journal of African Economies, Vol. 23 No. 5, pp. 614-642, doi: 10.1093/jae/eju015.

Austin, P.C., White, I.R., Lee, D.S. and Buuren, S.V. (2021), “Missing data in clinical research: a tutorial on multiple imputation”, Canadian Journal of Cardiology, Vol. 37 No. 9, pp. 1322-1331, doi: 10.1016/j.cjca.2020.11.010.

Baltagi, B.H. (2013), Econometric Analysis of Panel Data, John Wiley and Sons, Chichester.

Batini, N. and Durand, L. (2020), Capital Account Developments: Flows, Restrictions and Policy Toolkits, IEO Background Paper No. BP/20-02/03 for IEO evaluation of “IMF Advice on Capital Flows.”, International Monetary Fund, WA.

Bencivenga, V., Smith, B.D. and Starr, R.M. (1996), “Equity markets, transactions costs and capital accumulation: an illustration”, The World Bank Economic Review, Vol. 10 No. 2, pp. 241-265, doi: 10.1093/wber/10.2.241.

Benigno, G., Chen, H., Otrok, C., Rebucci, A. and Young, E.R. (2017), “Optimal capital controls and real exchange rate policies: a pecuniary externality perspective”, Journal of Monetary Economics, Vol. 84, pp. 147-165, doi: 10.1016/j.jmoneco.2016.10.004.

Biorn, E. (2017), Econometrics of Panel Data: Methods and Applications, Oxford University Press, Oxford, London.

Blanchard, O. (2018), “On the future of macroeconomic models”, Oxford Review of Economic Policy, Vol. 34 Nos 1-2, pp. 43-54, doi: 10.1093/oxrep/grx045.

Boot, A.W.A. and Thakor, A.V. (1997), “Financial system architecture”, Review of Financial Studies, Vol. 10 No. 3, pp. 693-733, doi: 10.1093/rfs/10.3.693.

Cavusoglu, N., Goldberg, M.D. and Stillwagon, J. (2019), “New evidence on the portfolio balance approach to currency returns”, Working Paper Series, Institute for New Economic Thinking, Vol. 89, pp. 1-40.

Chinn, M.D. and Ito, H. (2022), “The Chinn-Ito index: a de jure measure of financial openness”, Updated from: Chinn, M. D. and Ito, H. (2006). “What matters for financial development? Capital controls, institutions, and interactions”, Journal of Development Economics, Vol. 81 No. 1, pp. 163-192.

Chokri, Z. (2020a), “Restrictive policy impacts in emerging economies”, Cogent Economics and Finance, Vol. 8 No. 1, pp. 1-18.

Chokri, Z. (2020b), “Capital controls to manage foreign exchange reserves and foreign debts”, Zbornik Radova Ekonomskog Fakulteta U. Rijeci, Vol. 38 No. 1, pp. 271-294.

Cihak, M., Demirguc-Kunt, A., Feyen, E. and Levine, R. (2012), “Benchmarking financial systems around the world”, Policy Research. #6175, World Bank.

De Silva, A.P., De Livera, A., Lee, K.J., Moreno-Betancur, M. and Simpson, J.A. (2020), “Multiple imputation methods for handling missing values in longitudinal studies with sampling weights: comparison of methods implemented in Stata”, Biometrical Journal, Vol. 63 No. 2, pp. 354-371, doi: 10.1002/bimj.201900360.

Demirguc-Kunt, A. and Levine, R. (2002), “Bank-based and market-based financial systems: cross-country comparisons”, Journal of Financial Intermediation, Vol. 11 No. 4, pp. 398-428, doi: 10.1006/jfin.2002.0341.

Demirguc-Kunt, A. and Levine, R. (2011), Optimal Financial Structures and Development: the Evolving Importance of Banks and Markets, Mimeo, World Bank.

Demirguc-Kunt, A. and Levine, R. (2012), “The evolving importance of banks and securities markets”, The World Bank Economic Review, Vol. 23 No. 5, pp. 23-54.

Demirguc-Kunt, A. and Maksimovic, V. (1996), “Stock market development and financing choices of firms”, The World Bank Economic Review, Vol. 10 No. 2, pp. 341-370, doi: 10.1093/wber/10.2.341.

Eddings, W. and Machenko, Y. (2012), “Diagnostics for multiple imputation in Stata”, STATA Journal, Vol. 12 No. 3, pp. 353-367.

Fernández, A., Klein, M., Rebucci, A., Schindler, M. and Uribe, M. (2021), Capital Control Measures: A New Dataset, International Monetary Fund, Washington, DC.

Graham, M., Peltomaki, J. and Sturludóttir, H. (2015), “Do capital controls affect stock market efficiency? Lessons from Iceland”, International Review of Financial Analysis, Vol. 41, pp. 82-88, doi: 10.1016/j.irfa.2015.05.009.

Greene, W. (2017), Econometric Analysis, 8th ed., Pearson Incorporated, Cambridge, MA.

Ho, S.Y. and Iyke, B.N. (2017), “Determinants of stock market development: a review of the literature”, Studies in Economics and Finance, Vol. 34 No. 1, pp. 143-164, doi: 10.1108/sef-05-2016-0111.

Iqbal, A.S. (2022), “The Mundell-Fleming model and the impossible trinity”, in Foreign Exchange, Palgrave Macmillan.

Kao, C. (1999), “Spurious regression and residual-based tests for cointegration in panel data”, Journal of Econometrics, Vol. 90 No. 1, pp. 1-44, doi: 10.1016/S0304-4076(98)00023-2.

Kivet, J.F., Pleus, M. and Poldermans, R.W. (2016), “Accuracy and efficiency of various GMM inference techniques in dynamic micro panel models”, Econometrics, Vol. 5, pp. 2-54.

Krugman, P.R., Melitz, M.J. and Obstfeld, M. (2014), International Economics: Theory and Policy, 10th ed., Pearson Canada.

Levin, A., Lin, C.F. and Chu, C.S.J. (2002), “Unit root tests in panel data: asymptotic and finite-sample properties”, Journal of Econometrics, Vol. 108 No. 1, pp. 1-24, doi: 10.1016/S0304-4076(01)00098-7.

Levine, R. (2002), “Bank-based or market-based financial systems: which is better?”, Journal of Financial Intermediation, Vol. 11 No. 4, pp. 398-428, doi: 10.1006/jfin.2002.0341.

Magud, N.E., Reinhart, C.M. and Rogoff, K.S. (2018), “Capital controls: myth and reality - a portfolio balance approach”, Annals of Economics and Finance, Vol. 19 No. 1, pp. 1-47.

Montecino, J.A. (2015), “Capital controls and the real exchange rate: do controls promote disequilibria?”, Journal of International Economics, Vol. 114, pp. 80-95, doi: 10.1016/j.jinteco.2018.05.005.

Mundell, R. (2000), “Currency areas, exchange rate systems and international monetary reform”, Journal of Applied Economics, Vol. 3 No. 2, pp. 217-256, doi: 10.1080/15140326.2000.12040550.

Mutize, M., Tefera, E. and McBride, N. (2020), “The development of financial markets in Africa: trends, challenges and prospects”, Journal of Economics and Behavioral Studies, Vol. 12 No. 1, pp. 46-54.

Nakai, M. and Weiming, K. (2011), “Review of the methods for handling missing data in longitudinal data analysis”, International Journal of Mathematical Analysis, Vol. 5 No. 1, pp. 1-13.

Ntembe, A. (2022), “A single currency for Africa: challenges and possibilities”, in Amin, A.A., Tawah, R.N. and Ntembe, A. (Eds), Monetary and Financial Systems in Africa, Palgrave Macmillan, Cham.

Nyantakyi, E.B. and Sy, M. (2015), “The banking system in Africa: main facts and challenges”, Africa Economic Brief, Vol. 6 No. 6, pp. 1-16.

Nyasha, S. and Odhiambo, M.N. (2017), “Are banks and stock markets compliments or substitutes? Empirical evidence from three countries?”, Managing Global Transitions: International Research Journal, Vol. 15 No. 1, pp. 81-102.

Pasricha, G., Falagiarda, M., Bijsterbosch, M. and Aizenman, J. (2019), “Domestic and multilateral effects of capital controls in emerging markets”, Journal of International Economics, Vol. 115, pp. 48-58, doi: 10.1016/j.jinteco.2018.08.005.

Pedroni, P. (2004), “Panel cointegration: asymptotic and finite sample properties of pooled time series tests with an application to the PPP hypothesis”, Econometric Theory, Vol. 20 No. 3, pp. 597-625, doi: 10.1017/S0266466604203073.

Pesaran, M.H. (2015), Time Series and Panel Data Econometrics, Oxford University Press, Oxford.

Rebucci, A. and Ma, C. (2020), “Capital controls: a survey of the new literature”, in Dixit, A., Edwards, S., Judd, K. and Kletzer, K. (Eds), Oxford Research Encyclopedia of Economics and Finance, Oxford University Press.

Romer, D. (2019), Advanced Macroeconomics, 5th ed., McGraw-Hill Irwin Publications, New York.

Salvatore, D. (2013), International Economics, 11th ed., John Wiley & Sons Incorporated.

Schmitt-Grohe, S. and Uribe, M. (2017), “Is optimal capital-control policy countercyclical in open-economy models with collateral constraints?”, IMF Economic Review, Vol. 65 No. 3, pp. 498-527, doi: 10.1057/s41308-017-0032-6.

Schoder, C. (2019), “A Keynesian dynamic stochastic disequilibrium model for business cycle analysis”, Economic Modelling, Vol. 86, pp. 117-132, doi: 10.1016/j.econmod.2019.06.006.

Soumaré, S., Kanga, D., Tyson, J. and Raga, S. (2021), “Capital market development in Sub-Saharan Africa: progress, challenges and innovations”, Working Paper 2, A Joint FSD Africa–ODI research programme for financial sector development in Africa, Overseas Development Institute, London.

Sriram, B., El Aynaoui, K., Loungani, P., Ocampo, J.A. and Pedraglio, R. (2020), “IMF advice on capital flows to Africa and the Middle East”, Background Paper. BP/20-02/11, Independent Evaluation Office, International Monetary Fund.

Westerlund, J. (2007), “Testing for error correction in panel data”, Oxford Bulletin of Economics and Statistics, Vol. 69 No. 6, pp. 709-748, doi: 10.1111/j.1468-0084.2007.00477.x.

Williamson, O.E. (1998), “Transaction cost economics: how it works; where it is headed”, De Economist, Vol. 146 No. 1, pp. 23-58, doi: 10.1023/a:1003263908567.

Williamson, O.E. and Tadelis, S. (2012), “Transaction cost economics”, in Gibbons, R. and Roberts, J. (Eds), Prepared for the Handbook of Organizational Economics, Princeton University Press.

Wooldridge, J.M. (2010), Econometric Analysis of Cross Section and Panel Data, Massachusetts Institute of Technology Press, Cambridge, MA.

World Bank (2022), Global Financial Development Data 2022, World Bank, Washington, DC.

World Bank (2023), World Development Indicators 2023, World Bank, Washington, DC.

Acknowledgements

The authors would like to thank two anonymous reviewers for their comments and suggestions that have helped to enhance the paper.

Corresponding author

Kofi Bondzie Afful can be contacted at: kba.gimpa.edu@gmail.com

Related articles