Research in International Business and Finance 42 (2017) 1413–1427
Contents lists available at ScienceDirect
Research in International Business and Finance
journal homepage: www.elsevier.com/locate/ribaf
Full length Article
Remittances, banks and stock markets: Panel evidence from
T
developing countries
Haruna Issahakua,⁎, Joshua Yindenaba Aborb, Simon Kwadzogah Harveyc
a Department of Economics and Entrepreneurship Development, University for Development Studies, Box 520, Wa, UWR, Ghana
b Department of Finance, University of Ghana Business School, Legon, Ghana
c Bank of Ghana, One Thorpe Road, P.O. Box GP 2674, Accra, Ghana
A R T I C L E I N F O A B S T R A C T
JEL classification: The study investigates dynamic and causal linkages among international remittance inflows,
C33 banking sector development and stock market development in a large panel of developing
C58 countries. We use two stage least squares and impulse response functions to shed light on the
D53 remittance-bank-stock market nexus. We find that remittances promote banking sector devel-
F54 opment in low remittance receiving countries, but not in high remittance receiving economies.
G21
M21 We establish a bi-causal negative link between stock markets and remittances in countries with
developed banking systems. In low remittance recipient countries, remittances decrease stock
Keywords: market development; however, in remittance dependent countries, remittances promote stock
Banking sector development
Remittances market development. Again, stock market development promotes remittance inflows in re-
Developing countries mittance dependent countries, while obstructing it in low remittance recipient countries. We
Two stage least squares suspect lingering doubts about the quality of developing country stock markets to be behind this
Stock market development latter result, though the fact that most developing countries’ financial systems are bank based
could also play a role.
1. Introduction
Theoretical predictions on the financial development-remittance link are conflicting. The substitution and the complementarity
hypotheses are the leading theories in this regard. The ‘substitution hypothesis’ holds that, in a financially underdeveloped economy
where it is difficult to get credit from the formal financial sector, remittance serves as an alternative source of ‘credit’ for migrant
households. Thus, where there is low financial development, remittances provide liquidity and alleviate credit constraints. In con-
trast, where financial intermediaries function well, access to credit will be enhanced and remittances will subsidize consumption and
weaken work efforts (Bettin and Zazzaro, 2011). On the other hand, the ‘complementarity hypothesis’ posits that a developed
financial system will attract remittances in the formal credit system by lowering the costs of opening a deposit account and will allow
resources to be allocated efficiently. To date, the theory is also still vague on whether banks and stock markets are substitutes,
complements or independent. Boyd and Smith (1998) support the complementarity thesis by arguing that, through the provision of
liquidity and financial intermediation functions, equity markets support the banking sector in the mobilisation and allocation of
resources. In contrast, when firms issue stock to raise funds, their involvement in credit markets declines, implying that the two
segments of the financial system are substitutes (Arestis et al., 2001). More recently, Christophers (2015) has critiqued the con-
ventional classification of the financial system into market-based and bank-based systems. In the author’s view, ‘financial markets
⁎ Corresponding author.
E-mail addresses: iharuna@uds.edu.gh, harusconer@yahoo.com (H. Issahaku).
http://dx.doi.org/10.1016/j.ribaf.2017.07.080
Received 21 November 2016; Accepted 3 July 2017
Available online 16 July 2017
0275-5319/ © 2017 Elsevier B.V. All rights reserved.
H. Issahaku et al. Research in International Business and Finance 42 (2017) 1413–1427
are, in large part, banks and their interaction’(Christophers, 2015, p. 92).
Though there is a resurgence of literature regarding the positive effect of remittances on banking sector development and vice
versa, such studies do not simultaneously incorporate capital markets. In a sample of 94 non-OECD countries, Cooray (2012) shows
that remittances deepen the financial system in countries where government ownership of banks is low, but promote financial sector
efficiency in countries where government ownership of banks is high. Demirgüç-Kunt et al. (2011) further confirm that remittances
promote banking sector breadth and depth. Ezeoha (2013) and Adenutsi (2011) demonstrate that a well −developed financial sector
stimulates remittance inflows. These studies, however, do not incorporate stock market development in the analyses.
This omission presents a number of difficulties. By omitting stock market development, it is hard to tell whether (1) the link
between remittances and financial development still pertains when stock market development is controlled for, (2) capital markets
and banks impact remittance to the same degree and in the same direction, (3) banking sector and stock market development each
has an independent effect on remittances, or (4) that the financial system as a whole impacts on remittances but that the separate
contributions of banks and stock markets are indistinguishable.
Billmeier and Massa (2009) is the only study identified so far to have looked at the relationship between remittances and domestic
credit on the one hand and stock market development on the other hand. But their study faces a number of setbacks. Billmeier and
Massa (2009) use panel estimation with fixed effects without accounting for possible reverse causality between stock markets and
banking sector development, between stock market development and remittances, and between banking sector development and
remittances, since the theory predicts that remittances and financial development are interdependent. In addition, the use of stock
market capitalisation as a measure of stock market development is problematic, as it measures only one dimension of stock market
development − size. A stock market can be deep but is neither efficient nor stable. Also, a deep stock market does not necessarily
deliver access, since a stock market can be dominated by a few big firms. The authors do not examine variations in the impact of
credit and remittances on stock market development based on the size of remittances or the degree of banking sector development in
the economy. For instance, we will expect the effect of banking sector development and remittances on stock market development to
differ between remittance dependent countries and remittance ‘starved’ countries.
Levine (2005) laments that a major issue besetting the study of the finance − growth link is the use of weak proxies for financial
development. The same problem is manifest in the link between finance and remittances. Banking sector (financial) development is
often proxied by single variable measures such as the ratio of liquid liabilities to GDP, domestic credit provided by the banking sector,
deposits to GDP, credit provided by financial institutions to the private sector, the number of bank branches, and deposit accounts per
capita (e.g. Brown et al., 2013; Demirgüç-Kunt et al., 2011; Ezeoha, 2013; Nyamongo et al., 2012; Ramirez, 2013). Levine (2005) and
Bettin and Zazzaro (2011) argue that the financial development proxies listed above do not adequately mirror the key functions of the
financial system such as screening borrowers, spotting firms with great potential, solving information asymmetries, mobilising re-
sources and channelling same to profitable sources, monitoring managerial actions to prevent fraud and providing avenues for risk
sharing. Conclusions arising from such studies that use weak financial development proxies could be misleading and may not be
relevant for policy purposes.
Not that previous studies do not recognise the weaknesses embedded in the use of single variable proxies, they often deal with the
problem by using several of these single variable based measures to perform what they call ‘robustness checks’. This solution, though
laudable, is inadequate, as the alternative proxies themselves may be measuring different aspects or just one aspect of banking sector
development. It is now being recognised that financial development has at least four main dimensions −size, access, efficiency and
stability (WEF, 2013). Therefore, it is imperative that measures of banking and stock market development reflect the multi-
dimensional nature of financial markets. We address the gap in measuring financial development by constructing a composite index
each for banking sector development and stock market development after deriving the weights (factor loadings) from a panel
principal component analysis (PCA). Unlike previous proxies, our composite measures are able to capture the four key dimensions of
financial development− depth, access, efficiency and stability. By employing these multidimensional measures, we are able to assess
the link among remittances, banking sector and stock market development from a holistic standpoint.
This paper uses standard panel econometrics to assess the dynamic and causal links among banking sector development, stock
market development and remittances in a large panel of developing counties (DCs). In particular, we assess the impact each of
banking sector and stock market developments on remittances, the impact each of banking sector development and remittances on
stock market development, and the impact each of stock market development and remittances on banking sector development after
(1) controlling for potential endogeneity, (2) controlling for other variables known to influence the interaction between financial
development and remittances, and (3) performing a number of robustness checks by (i) using several econometric estimation
techniques to ensure rigour, (ii) splitting the sample based on the quantity of remittances received and the degree of banking sector
development to deal with potential non- linearities.
From the methodological perspective, we use an assortment of panel econometric estimation techniques. We first employ ordinary
least squares, fixed effects and random effect models to provide preliminary results. However, since these estimation techniques do
not adequately account for endogeneity, we resort to the use of two stage least squares and generalised impulse response functions
generated from panel vector autoregression.
Another innovation we bring to the literature is splitting the sample not just based on the level of remittances received, but also
based on the degree of banking sector development. In doing so, we allow the estimated coefficients to vary based on the state of
banking sector development in the country and based on the relative size of remittances in the economy. This allows us to capture
potential non-linearities and give policy recommendations while recognising that all developing countries neither receive the same
amount of remittances nor have they attained the same level of banking sector development. We are yet to encounter a study on
remittances, banks and stock markets which performs this two-way sample split based on banking sector development and
1414
H. Issahaku et al. Research in International Business and Finance 42 (2017) 1413–1427
remittances. Most of the existing studies split the sample based on financial development only (for example Giuliano and Ruiz-Arranz,
2009), based on natural resource endowment only (for example Billmeier and Massa, 2009), or based on the type of bank ownership
(for example Cooray, 2012).
To summarise, the contributions of this paper are as follows: (1) we overcome the use of inadequate proxies for financial de-
velopment by constructing multidimensional indices of banking sector and stock market development based on a PCA, (2) compared
to previous studies, we use an array of econometric techniques which enables us to overcome endogeneity, model selection bias and
to contribute to panel estimation procedure, (3) unlike previous studies, we study the relationship among (not between) stock market
development, banking sector development and remittances from varied directions (causality from: banks & stock markets to re-
mittances, banks & remittances to stock markets, and stock markets & remittances to banks), and (4) last but not the least, we capture
potential non-linearities in the relationships among remittances, banks and stock markets that might arise from differences in the
state of banking sector development across countries and from differences in the weight of remittances across developing economies.
This paper finds that the relationship among remittances, banks and stock markets is quite complex and nuanced. We find that the
relationship between banking sector development and remittances is of a nonlinear nature (inverse u-shape). Remittances promote
banking sector development in low remittance receiving countries, but not in high remittance receiving economies. The paper further
finds that the relationship between stock market development and remittances is mutually negative in countries with advanced
banking systems and in remittance ‘starved’ countries. We suspect the much publicised low capitalisation, weak stock market in-
frastructure, poor stock market supervision and regulation, and the illiquid nature of developing country stock markets to be behind
this undesirable outcome. The finding that the relationship between banking sector development and stock market development is
mutually positive is, however, quite re-assuring.
The rest of the paper is organised as follows. Section 2 presents the data and discusses the construction of our composite measures
of financial development. The empirical models and econometric estimation procedure are discussed in Section 3. Section 4 presents
the results and Section 5 concludes.
2. Data and index construction
2.1. Data
We analyse the causal links among stock market development, banking sector development and remittances using an unbalanced
panel of 61 developing countries (see Appendix B) over the period 1999–2013. A description of the data and their sources is reported
in Table 1. We are confronted with one of the issues that have ‘plagued’ the study of the finance-remittance nexus − how to measure
financial development to capture the functions of the financial system, such as easing information and transaction cost, mobilising
and channelling resources to productive sectors, and facilitating financial exchanges. It is by carrying out these functions that the
financial system is able to promote growth and deliver prosperity (Levine, 2005). Yet so far, the studies on the link between banking
sector development and remittances do not use measures that address this multifaceted nature of the functions of the financial sector.
Demirgüç-Kunt et al. (2011) use banking sector breadth (number of branches, or deposits accounts per capita) and banking sector
depth (deposits to GDP, or loans to GDP) to proxy for banking sector development. The problem with these measures is that a deep
and broad banking sector is not necessarily efficient or stable. A bank can reach out to as many people as possible and yet unable to
insulate itself from insolvency. Also, the loans and credit given could be concentrated in the hands of a few wealthy people leading to
undesirable economic outcomes. Other studies have resorted to the use of other quantity based proxies of banking sector develop-
ment, such as broad money supply to GDP and domestic credit provided by the banking sector (Chowdhury, 2011; Giuliano and Ruiz-
Arranz, 2009), and bank deposits to GDP (Aggarwal et al., 2011). These quantity based measures of banking sector development do
not adequately reflect banks’ key function of selecting entrepreneurs and channelling savings towards viable projects (Bettin and
Zazzaro, 2011).
The only study identified so far to have investigated the link between stock market development and remittances, Billmeier and
Massa (2009), use stock market capitalisation as a share of GDP to proxy stock market development. There are two main problems
Table 1
Description of variables.
Variable Notation Description Data Source
Remittances REMIT Personal remittances as a percentage of GDP WDI
Economic Openness TRADE Total exports plus total imports as a percentage of GDP WDI
Inflation Rate CPI Consumer price index (CPI) WDI
Economic Growth INCOME Growth in per capita GDP WDI
Financial Openness FDI Foreign direct investment as a percentage of GDP WDI
Investment INV Gross fixed capital formation as a percentage of GDP WDI
Institutional Quality INST Institutional quality proxied by the economic freedom index of the Heritage Foundation HF
Baking Sector Development BSD A composite measure of banking sector advancement encompassing size, access, efficiency and GFDD
stability dimension of banking sector development
Stock Market Development SMD A composite indicator of stock market advancement encompassing size, access, efficiency and stability GFDD
dimension of stock market development
Note: IFS is International Financial Statistics, WDI is World Development Indicators, HF is Heritage Foundation, GFDD is Global Financial Development Database.
1415
H. Issahaku et al. Research in International Business and Finance 42 (2017) 1413–1427
with this measure. First, the mere listing of firms does not guarantee the efficient allocation of resources (Beck and Levine, 2004).
Second, this measure only captures stock market depth but does not tell us how stable or liquid stock markets are.
We address the measurement shortcomings by recognising that financial development is multifaceted. There is the depth (size)
dimension, the access (breadth) dimension, the efficiency dimension, and the stability dimension. Therefore, any credible measure of
financial development must incorporate these dimensions. We develop measures for banking sector and stock market development
which capture these four dimensions. Our composite measure of banking sector development (BSD) envelopes credit provided by
deposit money banks as a share of GDP − a measure of the depth of financial intermediation, number of bank branches per thousand
adults as a measure of access, net interest margin as a measure of efficiency, and the bank Z-score as a measure of banking sector
stability. Similarly, our composite index of stock market development (SMD) consists of four measures − stock market capitalization
to GDP as a proxy for stock market size (depth), stock market value traded as a measure of access, turnover ratio (turnover/
capitalisation) as a measure of stock market efficiency, and volatility of stock price as a stability measure. The procedure used in
constructing the indices is discussed next.
2.2. Computations of indices and associated weights
Indices are computed for SMD and BSD. A major problem in the construction of indices is the choice of weights. In order not to
arbitrarily assign weights, Principal Components Analysis (PCA) is used to generate the weights. The composite index for SMD is
computed as follows:
4
SMDit = ∑ wr Indicatorrit
r=1
Where wr are the weights (factor loadings) derived from PCA; Indicatorrit is indicator r for country i at time t; r denotes the 4
subcomponents of SMD − stock market capitalization to GDP as a measure of stock market size (depth), stock market value traded as
a measure of access, turnover ratio (turnover/capitalisation) as a measure of stock market efficiency, and volatility of stock price
index as a stability measure.
The analogous index for BSD is:
4
BSDit = ∑ wr Indicatorrit
r=1
r denotes the four measures of banking sector development − banking sector credit to GDP, accounts per thousand adults, net
interest margin, and bank Z-score.
Each indicator is normalised on a scale of 1–7 to allow for easy comparison of country performance across indicators. Where 1 and
7, respectively, are the worst and best possible outcomes. The formula for conversion is given by:
[country score − sample minimum]
6 + 1
[sample maximum − sample minimum]
Where the sample minimum and sample maximum are, respectively, the lowest and highest country scores in the sample of countries
covered. In the case of variables for which higher values indicate a worse outcome (eg. stock price volatility) we use the following
conversion formula both to rescale from 1 to 7 and to ensure that 1 and 7 still depict worst and best outcomes:
[country score − sample minimum]
− 6 + 7
[sample maximum − sample minimum]
3. Empirical models and econometric estimation procedure
3.1. Empirical models
We model the relationship among remittances, stock market and banking sector development as follows:
REMITit = β1SMDit + β2BSDit + Xi′t α + ut + ηi + eit (1)
SMDit = α1REMITit + α2BSDit + Xi′t β + ut + ηi + eit (2)
BSDit = b1REMITit + b2SMDit + Xi′t b + ut + ηi + eit (3)
Where SMD is stock market development, REMIT is remittances as a percentage of GDP, BSD is banking sector development, Xit is a
vector of control variables, namely inflation (CPI), investment (INV), institutional quality (INST), trade openness (TRADE), per capita
GDP growth (INCOME), and financial openness (FDI). ut is time effect, ηi is country specific effect, and eit is the stochastic error term
assumed to be independently and identically distributed.
Stock market development (SMD) refers to a composite measure of stock market development which captures the size, efficiency,
1416
H. Issahaku et al. Research in International Business and Finance 42 (2017) 1413–1427
access and stability dimension of stock market development. The components of the index are stock market capitalization to GDP as a
proxy for stock market size (depth), stock market value traded as a measure of access, turnover ratio (turnover/capitalisation) as a
measure of stock market efficiency, and volatility of stock price index as a stability measure.
Banking sector development (BSD) is a composite measure of banking sector development which captures four key dimensions of
banking sector development − depth, access, efficiency, and stability. Banking sector credit to GDP is used as a measure of financial
depth, number of bank branches per thousand adults is used as a measure of access, net interest margin as a measure of efficiency,
and bank Z-score is a proxy for banking sector stability.
Remittances (REMIT) are measured as migrant remittances and as a percentage of GDP. For a number of developing countries,
remittances receipt is on the ascendency and in some cases has outstripped FDI. Though remittance’s impact on financial develop-
ment is mixed, there is now an emerging consensus that remittances stimulate financial development (Billmeier and Massa, 2009). By
increasing disposable income, bank deposits, and credit volume, remittances increase the amount of capital available for investment
in stocks, bonds and other assets. Also, remittance recipients sometimes save part of the remittance receipts for future use. Banks also
follow up on remittance recipients to provide extra financial services beyond just the receipt of remittances. Due to the above reasons,
it is envisaged that remittances will advance both banking sector and stock market development.
Economic growth is proxied by growth in per capita GDP in constant 2000 U.S dollars. Economic growth has been found to anchor
financial development. The higher the economic growth, holding all other factors constant, the higher the demand for both banking
and nonbanking financial services.
Institutions are proxied by the Heritage Foundation’s Index of Economic Freedom. The index is made up of 10 components with
equal weights− fiscal burden, banking and finance,1 trade policy, government intervention, black market, monetary policy, property
rights, capital flows and foreign investment, wages and prices, and regulation. Countries are given scores between 0 and 100 with
higher values associated with higher institutional quality. This index has been chosen over other institutional quality measures due to
the following reasons. First, unlike other indicators, data on the Index of Economic Freedom is available for a larger number of
developing countries on annual basis and for a relatively longer timespan. Secondly, the measure is comprehensive and covers a wide
dimension of institutional development. The index is receiving wide usage in the extant literature.
Inflation is measured as the consumer price index. It has been generally found that inflation interferes with the decision making
processes of economic agents pertaining to nominal quantities. This discourages financial intermediation while encouraging in-
vestment in real assets. Thus, inflation is expected to be negatively correlated with financial market development and remittances.
Investment is critical for both stock market capitalisation and financial intermediation by banks, since banks and stock markets
constitute alternative ways of intermediating savings to investment projects. Investment is measured as gross fixed capital formation/
GDP.
Trade openness and financial openness are deemed to promote the development of financial markets and the flow of funds across
countries, including remittances. The ratio of trade to GDP is used to proxy for trade openness, while FDI as a percentage of GDP is
used as a proxy for financial openness.
Ruiz and Vargas-Silva (2010) argue that the effect of remittances on economic magnitudes could vary based on the size of such
flows relative to GDP. If this proposition is true, then it will have implication for Eqs. (1)–(3) above. It will mean that the impact of
remittances on stock market and banking sector development and vice versa will vary with the amount of remittances received. Also,
the lack of consensus on the link between remittances and financial market development suggests the possible existence of non-
linearities or threshold effects. While some studies predict a positive link (see Aggarwal et al., 2011; Cooray, 2012; Ezeoha, 2013),
others find a negative link (e.g. Brown et al., 2013). And yet, some studies view remittances and financial development as substitutes
(e.g. Giuliano and Ruiz-Arranz, 2009; Ramirez, 2013). Moreover, Chowdhury (2011) finds the impact of financial development on
remittances to be neutral. Our suggested threshold models are of the following form. The split sample model for remittances is
specified as:
REMITit = β1SMDit + β2BSDit + Xi′t α + ut + ηi + eit, if REMITit ≤ γ (4)
REMITit = δ1SMDit + δ2BSDit + Xi′t δ + ut + ηi + eit, if REMITit > γ (5)
The model for stock market development is given by:
SMDit = α1REMITit + α2BSDit + Xi′t β + ut + ηi + eit, if REMITit ≤ γ (6)
SMDit = ϕ1 REMITit + ϕ2 BSDit + Xi′t ϕ + ut + ηi + eit, if REMITit > γ (7)
That of the banking sector is of the form:
BSDit = b1REMITit + b2SMDit + Xi′t b + ut + ηi + eit, if REMITit ≤ γ (8)
BSDit = ∝1 REMITit + ∝2 SMDit + Xi′t ∝ +ut + ηi + eit, if REMITit > γ (9)
REMIT is the threshold variable used in splitting the sample. Where γ is an exogenously determined threshold defined by the
1 The inclusion of banking and finance in the institutional quality measure might pose linearity issues with BSD and SMD. However, a correlation test shows that this
problem is not serious. The correlation between institutional quality and BSD and SMD are respectively 0.21 and 0.04.
1417
H. Issahaku et al. Research in International Business and Finance 42 (2017) 1413–1427
median level of remittances received. All other variables are as defined earlier. The sample split allows the regression coefficients to
vary based on the amount of remittances received. We use an exogenously determined threshold as opposed to one endogenously
determined because most endogenously threshold models are based on OLS which is plagued with endogeneity. Further robustness
checks are examined later by splitting the sample based on the level of banking sector development.
3.2. Econometric estimation procedure −Two stage least squares
The 2SLS approach is employed in this study due to the potential endogeneity between remittances and financial development
variables. For instance, remittance has been found to promote banking sector development (Demirgüç-Kunt et al., 2011) and stock
market development (Billmeier and Massa, 2009). At the same time, a well-developed financial system is found to promote re-
mittance inflow (Cooray, 2012). These illustrations imply a potential endogeneity between remittances and financial development.
Other potential sources of endogeneity include a reverse causality between remittances and economic growth, and the inter-
dependent nature of banking institutions and stock markets.
We note that the selection of instruments is a nontrivial matter in 2SLS estimations. It is always difficult to get ‘good’ instruments.
In this regard we use the lags of the variables (remittances, banking sector development) as instruments. We include household
consumption growth as an additional instrument for remittances given the abundance of literature to the effect that remittances
trigger consumption. Specification tests relating to instrument validity and endogeneity are conducted. The Woolridge’s over-iden-
tifying restrictions test is employed to assess instrument validity. The null hypothesis is that the instruments are valid. With regards
endogeneity we employ two tests: Woolridge’s robust score test for exogeneity and the robust regression exogeneity test. The null
hypothesis in both cases is that the variables are exogenous. In models where we fail to reject the null hypothesis, alternative robust
panel least squares estimates are provided (in Appendix A) since OLS is said to be less biased in the absence of endogeneity.
4. Empirical results
4.1. Descriptive statistics
Table 2a presents a characterisation of remittances inflows in developing countries based on the levels of stock markets, in-
stitutional and banking sector advancement. We determine the level of financial development by splitting our sample based on the
median values. All observations less than or equal to the median level define a low level of development of the variable in question,
while those above define a high level of development of the variable concerned. For instance, all observations below and up to the
median level of stock market development are deemed to describe a low level of stock market development, while all observations
above the median level of stock market development are deemed to describe a high level of stock market advancement. The choice of
the median is based on the fact that compared to the mean, the median is less susceptible to the influence of outliers.
Interestingly, countries with substandard stock markets receive a higher median amount of remittances compared to countries
with advanced stock markets. In the interim, this suggests a negative relationship between stock markets and remittances, but only in
a non-causal manner. On the contrary, countries with a more developed banking system receive a higher median amount of re-
mittances compared to countries with a relatively underdeveloped banking sector. Remittances as a share of GDP is 2.4184% in
countries with high banking sector development compared to 1.6350% for countries with a less developed banking sector. For
institutional quality, the results are quite mixed. While countries with low institutional quality have a higher median amount of
remittances received, countries with better institutional quality, however, have a higher mean level of remittances received.
We test the statistical significance of these differences in the medians using a series of nonparametric tests namely, Wilcoxon/
Mann-Whitney, Wilcoxon/Mann-Whitney (tie-adj.), Med. Chi-square, Adj. Med. Chi-square, Kruskal-Wallis, Kruskal-Wallis (tie-adj.),
van der Waerden. For stock market and banking sector development, the observed differences are significant at 1% level for all the
test statistics. Despite the statistical significance, we must reiterate that the findings here are merely descriptive; causal relationships
Table 2a
Average remittances based on levels of banking and stock market development.
Mean Median Max. Min. Std. Dev. Obs. Mean differencea (p = 0.05)
Low banking sector development 3.8527 1.6350 31.5237 0.0145 5.3276 452
High banking sector development 4.6489 2.4184 40.6220 0.0489 5.9821 457 Significant
Low stock market development 4.8632 2.5818 40.6220 0.0155 6.1343 454
High stock market development 3.6441 1.5812 31.5237 0.0145 5.1152 455 Significant
Low institutional quality 4.2860 2.4469 40.6220 0.0145 5.4961 443
High institutional quality 4.3056 1.7859 31.5237 0.0227 5.9743 441 Not significant
All 4.2530 1.9974 40.6220 0.0145 5.6769 909
a tests are based on the following nonparametric tests: Wilcoxon/Mann-Whitney, Wilcoxon/Mann-Whitney (tie-adj.), Med. Chi-square, Adj. Med. Chi-square,
Kruskal-Wallis, Kruskal-Wallis (tie-adj.), van der Waerden.
1418
H. Issahaku et al. Research in International Business and Finance 42 (2017) 1413–1427
Table 2b
Average level of banking sector development based on the degree of stock market development.
Mean Median Max Min. Std. Dev. Obs. Mean differencea (p = 0.01)
Low stock market development 2.6537 2.4806 5.9844 1.2084 1.0335 457
High stock market development 3.2288 2.8718 6.9999 1.3867 1.2471 458 significant
All 2.9415 2.7052 6.9999 1.2084 1.1804 915
a tests are based on the following nonparametric tests: Wilcoxon/Mann-Whitney, Wilcoxon/Mann-Whitney (tie-adj.), Med. Chi-square, Adj. Med. Chi-square,
Kruskal-Wallis, Kruskal-Wallis (tie-adj.), van der Waerden.
Table 3
Remittances, banks and stock markets eOLS, Fixed Effects and Random Effects estimates.
(1) (2) (3) (4) (5) (6) (7) (8) (9)
SMD BSD REMIT
Regressors OLS FE RE OLS FE RE OLS FE RE
SMD 0.454*** 0.166*** 0.186*** −0.554*** −0.351* −0.346*
(0.0261) (0.0580) (0.0499) (0.184) (0.198) (0.179)
BSD 0.656*** 0.299** 0.331*** 0.454** 0.366 0.339
(0.0428) (0.125) (0.106) (0.229) (0.565) (0.534)
REMIT −0.0292*** −0.0304 −0.0280* 0.0166** 0.0176 0.0153
(0.00954) (0.0200) (0.0153) (0.00792) (0.0265) (0.0239)
FDI −0.000250 0.0332** 0.0326** −0.0107 0.00882 0.00612 −0.197*** −0.000341 −0.00132
(0.0138) (0.0155) (0.0151) (0.0109) (0.00922) (0.00913) (0.0671) (0.0562) (0.0555)
INV −0.0379 0.182*** 0.164*** 0.246*** 0.0696 0.0843 −0.207 0.0990 0.0803
(0.0427) (0.0591) (0.0554) (0.0359) (0.0855) (0.0794) (0.219) (0.271) (0.271)
TRADE −0.239*** −0.0313 −0.0652 0.279*** 0.0966 0.114* 0.260* 0.473 0.462
(0.0376) (0.0811) (0.0746) (0.0300) (0.0725) (0.0669) (0.156) (0.305) (0.291)
INST −0.198* −0.569** −0.508*** 0.292*** 1.017*** 0.946*** 0.333 −0.515 −0.383
(0.110) (0.227) (0.193) (0.0888) (0.171) (0.164) (0.598) (1.095) (1.036)
CPI −0.114** −0.0698 −0.0747 0.278*** 0.228*** 0.228*** 0.461** 0.210 0.220
(0.0456) (0.0705) (0.0662) (0.0356) (0.0407) (0.0402) (0.193) (0.185) (0.185)
INCOME 0.0266** 0.0201** 0.0210*** −0.0279** −0.0143* −0.0148** 0.0456 0.0638* 0.0632*
(0.0121) (0.00758) (0.00726) (0.0130) (0.00737) (0.00731) (0.0644) (0.0372) (0.0372)
Constant 2.397*** 2.473** 2.408*** −3.596*** −4.851*** −4.700*** −3.198 −0.778 −1.286
(0.494) (1.024) (0.906) (0.374) (0.834) (0.798) (2.632) (4.903) (4.671)
Observations 663 663 663 663 663 663 663 663 663
R-squared 0.329 0.156 0.499 0.501 0.051 0.081
No. of countries 57 57 57 57 57 57 57 57 57
Table 3 presents OLS, Fixed effects (FE), Random (RE) estimates with stock market development (SMD), Banking sector development (BSD) and remittances (REMIT)
as dependent variables. The control variables are financial openness (FDI), trade openness (TRADE), investment (INV), institutional quality (INST), inflation (CPI) and
per capita GDP growth (INCOME). All variables are defined in Table 1. Robust standard errors in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1.
will be established later. For institutional quality, none of the tests is significant at the 5% level.
Based on the standard deviations, remittances are more variable under low levels of stock market development and under high
levels of institutional and banking sector advancement. The median amount of remittances, as a share of GDP for our sample
countries, is 1.9974% while the mean is 4.2530%. In some developing countries, remittances as a percentage of GDP exceed 60%. The
results confirm that remittances play a crucial role in developing countries.
Table 2b describes the degree of banking sector advancement based on the level of stock market development. Countries with
more developed stock markets also have more developed banks and vice versa, based on both the mean and median. This points to a
possible complementarity between the two segments of the financial system. The median difference is statistically significant at 1%
level.
4.2. Preliminary results
Table 3 reports ordinary least squares (OLS), fixed effects (FE) and random effects (RE) estimates2 of the relationship among
remittances, banks and capital markets. All standard errors are corrected to ensure homoscedastic errors. The control variables in all
the regressions are financial openness (FDI), trade openness (TRADE), inflation (CPI), investment (INV), institutional quality (INST),
2 The Hausman test shows that the fixed effect model is valid for models where BSD and SMD are the dependent variables while the random effect is valid in models
where remittance is the dependent variable.
1419
H. Issahaku et al. Research in International Business and Finance 42 (2017) 1413–1427
and per capita GDP growth (INCOME). The main regressors for the stock market models are banking sector development and re-
mittances. The main regressors in the banking sector models are banking sector development and remittances, while banking and
stock market development are the main regressors in the remittance models.
In the models where stock market development is the regressor, banking sector development positively and significantly affects
stock market quality in all the three estimation methods, confirming complementarity between the two segments of the financial
system. Interestingly, the coefficient is large. The least coefficient suggests that a 10% increase in banking sector development will
induce a 2.99% improvement in stock market development in DCs. This finding is plausible because banks and stock markets are
interdependent. The market provides an avenue for banks to raise finance for investment. This finance obtained from the market
enables banks to operate efficiently, increase their value and, for that matter, appreciate the value of their listed stocks. These
findings are consistent with Christophers (2015) and Cheng (2012).
Remittances negatively impact stock market development in all three models. This finding is significant at 5% respectively in the
OLS and RE models. This result, though undesirable, is not so surprising. Most DCs are known to have poorly capitalised stock
markets. Also, DC stock markets are highly illiquid and are thus an unattractive investment alternative for remittance recipients.
Moreover, emerging country stock markets are observed to exhibit higher price volatility relative to advanced country stock markets.
Such high volatilities scare away risk averse investors (Singh, 1997). Additionally, to invest in the stock market, the investor requires
some minimum level of financial literacy, which is often lacking amongst most households in developing countries. Remittances are
often used to finance child schooling, for household consumption, to finance businesses, to acquire household assets, and for the
construction of houses. The coefficients are, however, relatively small, thereby discounting the economic significance of the reduction
in stock market development following an increase in remittance inflows.
Both remittances and stock market development enter our banking sector regressions significantly with a positive sign. The
remittance estimate is significant in only one of the models. The finding that remittances promote banking sector development is not
surprising. The banking system is an important channel for the transmission of remittances. The receipt of remittances help boost
bank credit and deposits (Aggarwal et al., 2011) and also makes banks more efficient, especially when government ownership of
banks is high (Cooray, 2012). Remittances further accelerate banking sector development by improving banking sector breadth and
depth (Demirgüç-Kunt et al., 2011).
The estimates for the stock market variable in the banking sector regressions are highly significant in all three models. Stock
markets provide important avenues for banks to raise money (by issuing stocks and bonds) to finance their projects. The finding of a
positively reinforcing bi-causality between banking sector and capital market development is refreshing. It implies that DCs can
simultaneously develop the banking sector and the market without having to worry about any backlash. As we have in advanced
countries, strong banks and strong markets can exist together in DCs. Strong banks and strong markets will anchor accelerated
growth.
Improvement in banking quality is found to positively stimulate remittance inflows. The result is only significant in the OLS
regression. The empirical literature suggests that a properly functioning banking system will reduce the cost of remitting which could
increase remittance flows (Giuliano and Ruiz-Arranz, 2009). Stock market development negatively enters all the remittances re-
gressions and is significant in all three models. The result, that stock market development does not promote remittances inflow, is less
surprising since migrants hardly invest in stocks in the home country and low financial literacy also limits the participation of
remittance recipients in stock markets.
4.3. Two stage least squares estimates (2SLS) estimates −full sample
Table 4 presents 2SLS estimates for our sample. The results generally confirm the findings from the OLS, FE and RE models. These
results are not susceptible to endogeneity bias. The Woolridge’s over-identifying restrictions tests show that we have used valid
instruments. Furthermore, Woolridge’s robust score test for exogeneity and/or the robust regression exogeneity test reject the null
hypothesis of exogeneity for all the models, model 12 being the only exception. For model 12, we present the corresponding robust
OLS estimates in Appendix A. The Wald test confirms that our regressor variables jointly affect the regressands at 1% level. These test
combined, show that our models are adequate.
Banking sector development positively and significantly promotes capital market advancement, confirming the earlier results of a
complementary relationship between the two financial segments. Interestingly, the coefficient is large. It depicts that a 10% increase
in banking sector development will lead to a 6.80% increase in stock market development in developing countries, which is an
improvement over the previous estimates. Again, remittances negatively impact stock market development. Consistent with the
preliminary results, both remittances and stock market development enter our banking sector regressions with significantly positive
signs. After accounting for endogeneity, stock market advancement still has a negative effect on remittances while banking sector
quality has a positive effect.
4.4. Does the quantity of remittance received affect the relationship among banks, stock markets and remittances?
A problem with our results so far is that we have assumed that the established relationships hold, irrespective of the size of
remittances in the overall economy. For instance, we might expect that the impact of remittances and banks on stock markets might
be different in countries where remittances are a large share of GDP compared to countries where remittances only form a minute
proportion of GDP. To address this problem, we partition our sample into two, based on the median amount of remittances received.
For the purpose of this study, all countries below the median level of remittances are regarded as low remittance recipient countries,
1420
H. Issahaku et al. Research in International Business and Finance 42 (2017) 1413–1427
Table 4
Relationship among banks, stock markets and remittances −2SLS estimates (full sample).
(10) (11) (12)
Regressors SMD BSD REMIT
SMD 0.486*** −0.612***
BSD 0.680*** 0.572**
REMIT −0.0209* 0.0162*
(0.0109) (0.00867)
(0.0480) (0.262)
FDI 0.000861 −0.00219 −0.243***
(0.0153) (0.0116) (0.0726)
INV −0.0392 0.261*** −0.361
(0.0476) (0.0368) (0.239)
TRADE −0.219*** 0.270*** 0.198
(0.0404) (0.0325) (0.172)
INST −0.288** 0.299*** 0.320
(0.124) (0.0980) (0.669)
CPI −0.111* 0.205*** 0.296
(0.0572) (0.0417) (0.227)
INCOME 0.0261* −0.0339** 0.0294
(0.0137) (0.0160) (0.0740)
(0.0292) (0.211)
Constant 2.654*** −3.337*** −1.759
(0.544) (0.403) (2.834)
Diagnostics
Observations 561 561 561
R-squared 0.322 0.489 0.042
No. of countries 57 57 57
Wald Chi2 (stat) 245.57*** 784.82*** 29.46***
Woolridge’s OIR (p-value) 0.7243 (0.3947) 0.0622 (0.8031) 0.6541 (0.4187)
Woolridge Rob. 0.0208 8.5459** 5.2782*
Rob. Reg. 0.0103 5.3364** 2.05662
Table 4 presents 2SLS results for the entire sample. Stock market development (SMD), Banking sector development (BSD) and remittances (REMIT) are the dependent
variables. The control variables are financial openness (FDI), trade openness (TRADE), investment (INV), institutional quality (INST), inflation (CPI) and per capita
GDP growth (INCOME). All variables are defined in Table 1. The diagnostic tests reported include: (1) number of observations, (2) R squared test of model explanatory
power, (3) number of countries, (4) Wald Chi-square for joint significance of coefficients (5) the Woolridge’s overidentifying restrictions test of which the null
hypothesis is that the instruments are valid, (6) Woolridge’s robust score test for exogeneity of which the null hypothesis is that remittances is exogenous, (7) Robust
regression exogeneity test of which the null hypothesis is that remittance is exogenous. Robust standard errors are in parentheses. *, **, *** respectively denote
significance at 10%, 5%, 1% level of significance.
while those above the median are characterised as high remittance recipient countries. The results of the 2SLS estimates for the split
sample are reported in Table 5.
These results largely reinforce our earlier finding. The main exception is that we now gain a new insight into the relationship
between banking sector development and remittances. The results depicting remittances as the dependent variable (model 17–18)
show that, below the median remittances received, banking sector progress is positively and significantly linked to the inflow of
remittances but negatively (but not significantly) related to remittances above the median. This shows a non-linear association
between baking sector development and remittances. The reasoning is that, at low levels of remittances, an improvement in the
banking sector, involving an expansion of banking coverage to include remittance recipients, a decrease in the cost of remitting and
an increase in banking efficiency will lead to a corresponding rise in remittance inflows. However, as the level of banking sector
development reaches a certain magnitude, migrants no longer increase their remittances sent. Above the median, remittances respond
positively to favourable changes in trade openness, improvements in institutional quality and rising inflation.
The other insight gained by splitting the sample is that the effect of remittances on banking sector development is positive (and
significant) at low levels of remittances, but negative (and insignificant) at high levels of remittances. The implication is that remittances
improve banking sector development in low remittance recipient countries, but may not impact banking sector development in high
remittance recipient countries. Thus, there is a limit to which developing countries can rely on remittances to develop their banking sector.
In the banking sector development model, above the median level of remittances received, trade openness, investment and stock market
development enter the model positively and are all highly significant. This means that remittance dependent countries cannot rely on
remittances to develop their banking sector. To advance their banking sector will require more than remittances; they will need to boost
and stabilise capital investment, upgrade capital markets and fortify trade competitiveness. The literature supports the fact that ‘too much’
remittances can lead to undesirable economic outcomes including rising inflation, a loss of international competitiveness due to domestic
currency appreciation, a decline in the supply of labour, an increase in the consumer demand for non-tradable goods and a consequent
reduction in the production of tradable goods (Acosta et al., 2009). Remittance deposits with banks are mostly transitory funds and may
not be kept in banks long enough for the banks to generate long term loans from them. Thus in the long run remittances may not
significantly improve bank lending. A two-way scatterplot and quadratic regression curve with 95% confidence interval of banking sector
development and remittances confirms an inverse u-shape relationship (see Fig. 1).
1421
H. Issahaku et al. Research in International Business and Finance 42 (2017) 1413–1427
Table 5
Relationship among banks, stocks and remittances based on the median level of remittances −2SLS estimates.
(13) (14) (15) (16) (17) (18)
VARIABLES SMD SMD BSD BSD REMIT REMIT
Below median Above median Below median Above median Below median Above median
(≤1.9974) (> 1.9974) (≤1.9974) (> 1.9974) (≤1.9974) (> 1.9974)
SMD 0.560*** 0.383*** −0.418*** 0.286**
(0.0392) (0.0402) (0.154) (0.130)
BSD 0.738*** 0.563*** 0.646*** −0.189
(0.0504) (0.0891) (0.183) (0.184)
REMIT −0.0377* 0.0855** 0.0546*** −0.0292
(0.0217) (0.0334) (0.0170) (0.0287)
FDI −0.0629** 0.0725*** 0.0403* −0.0328** −0.156* −0.156***
(0.0279) (0.0180) (0.0216) (0.0148) (0.0804) (0.0454)
INV −0.00621 0.0107 0.179** 0.336*** −0.0566 0.0596
(0.0801) (0.0640) (0.0762) (0.0376) (0.247) (0.164)
TRADE −0.141*** −0.357*** 0.198*** 0.360*** −0.0397 0.471***
(0.0437) (0.0682) (0.0428) (0.0421) (0.130) (0.128)
INST −0.130 −0.627*** 0.343** 0.0388 1.871*** 0.932**
(0.201) (0.174) (0.164) (0.125) (0.604) (0.395)
CPI −0.0546 −0.231*** 0.219*** 0.166*** −0.461** 0.447***
(0.0811) (0.0744) (0.0631) (0.0550) (0.187) (0.129)
INCOME 0.0240 0.00825 −0.0138 −0.0652*** 0.0555 0.0138
(0.0182) (0.0221) (0.0189) (0.0167) (0.0903) (0.0483)
Constant 1.334* 4.870*** −3.105*** −2.526*** −6.267** −6.038***
(0.753) (0.816) (0.666) (0.562) (2.584) (1.771)
Diagnostics
Observations 280 281 280 281 280 281
R-squared 0.422 0.282 0.557 0.481 0.100 0.119
Wald Chi2 (stat) 269.95*** 87.56*** 550.80*** 361.65*** 29.83*** 36.85***
Woolridge’s OIR (p- 0.4004 (0.5269) 0.7978 (0.3717) 0.5086 (0.4757) 0.1155 (0.7340) 2.2626 (0.1325) 1.9955 (0.1578)
value)
Woolridge Rob. 0.7986 1.1951 6.0353** 4.4703 9.1494** 0.2118
Rob. Reg. 0.3488 0.7032 6.8095** 2.5634* 5.2015** 0.1004
Table 5 presents 2SLS results for split samples based on the amount of remittances received. Stock market development (SMD), Banking sector development (BSD) and
remittances (REMIT) are the dependent variables. The control variables are financial openness (FDI), trade openness (TRADE), investment (INV), institutional quality
(INST), inflation (CPI) and per capita GDP growth (INCOME). All variables are defined in Table 1. The diagnostic tests reported include: (1) number of observations,
(2) R squared test of explanatory power, (3) Wald Chi-square for joint significance of coefficients, (4) the Woolridge’s overidentifying restrictions test of which the null
hypothesis is that the instruments are valid, (5) Woolridge’s robust score test for exogeneity of which the null hypothesis is that remittances is exogenous, (6) Robust
regression exogeneity test of which the null hypothesis is that remittance is exogenous. Robust standard errors are in parentheses. *, **, *** respectively denote
significance at 10%, 5%, 1% level of significance.
Another key revelation is that in low remittance receiving countries, remittances decrease stock market development while in
high remittance receiving countries, remittances promote stock market development. The intuition behind this finding is that, when
migrant households receive remittances in excess of their consumption and welfare needs, they invest the ‘excess’ cash in capital
markets.
To summarise, our robustness checks of splitting the sample based on the size of remittances in the economy corroborates our
earlier finding that financial markets are interdependent even in the presence of remittances, and that stock markets do not promote
remittance inflow. The new insights are that banking sector development promotes remittances inflow only in countries that are less
remittance dependent; remittances promote banking sector development in less remittance-dependent economies; remittances pro-
mote stock market development in remittance dependent countries but lowers it in nondependent countries.
4.5. What about the degree of banking sector development?
In this subsection, we present further robustness checks by discussing the impact of the degree of banking sector progress on the
relationships among remittances, stock markets and banks. The reasoning is that a well-developed banking sector may enhance stock
market advancement more than a shallow one. A more efficient banking sector might even have a better capacity to channel re-
mittances to productive sectors of the economy. The results of 2SLS estimates of our split sample based on the median level of banking
quality are shown in Table 6. The results confirm our findings in the previous section but with some interesting qualifications.
Though banking sector progress positively influences stock market advancement, the impact is quantitatively greater at a high level
of banking sector development. Stock market development promotes banking development irrespective of the degree of banking
sector advancement. The impact of remittances on banking sector progress is not responsive to the degree of banking sector de-
velopment. This finding, combined with the earlier findings, suggests that the impact of remittances on the banking sector does not
1422
H. Issahaku et al. Research in International Business and Finance 42 (2017) 1413–1427
Fig. 1. Scatter plot of banking sector development (BSD) and remittances.
depend on the degree of banking sector advancement but instead on the volume of remittances relative to the size of the economy.
Banking sector development stimulates remittance inflows only at low levels of banking sector development. This finding is
plausible because when domestic financial markets are shallow, migrant relatives rely on remittances as an alternative source of
capital (Issahaku et al., 2016). Lastly, stock market development does not promote remittance inflows especially at high levels of
banking sector development.
4.6. Simulations
We perform simulation exercises to see the responses of banking and stock market advancement and remittances to own shocks
and shocks from other system variables. This shock analysis exercise can guide policy makers and practitioners on the probable
behaviour of the three policy variables in the advent of positive or negative shocks to domestic and global financial markets and in
the advent of positive or negative technological shocks. We present generalised impulse responses generated from a three variable
(BSD, SMD, REMIT) panel vector autoregression (PVAR) in Fig. 3. Each simulation exercise involves response to one unit positive
shock of the relevant variable. Error bands (2 standard deviations) are generated using Monte Carlo simulations with 5000 repetitions
to enhance precision. Impulse responses are valid only when the underlying equations are stable. Fig. 2 shows that the inverse roots of
the autoregressive characteristic polynomial fall within the unit cycle. This implies that the PVAR is stable enough to produce reliable
impulse response functions.
Rows 1, 2 and 3 respectively depicts the responses of stock market development, banking sector development and remittances to
own shocks and shocks from other system variables. Positive innovations in stock markets lead to a steady and persistent rise in stock
market development. This means that the introduction of new technologies, significant infrastructure upgrading or any stock market
related significant positive event in DC stock markets would massively raise stock market depth, breadth, stability and efficiency. In
the advent of a positive shock to banking sector advancement (stock market advancement), stock market advancement (banking
sector advancement) improves gradually over the long run confirming the positive mutuality between banks and stock markets
established earlier. A shock to remittances however, leads to a lacklustre rise in stock market development and an insignificant
change in banking sector development, especially in the initial years of the shock. Banking sector development shocks have a steady
and persistent impact on banking progress. Remittances are largely responsive to own shocks. Positive shocks from stock markets
only lower remittance inflows after 5 years following the shock. This means that with the low level of financial literacy in developing
countries even sudden improvements in stock markets will not motivate migrants to send money home to capitalise on the booming
stock markets. Shocks to banking sector have insignificant effects on remittances. For all three variables, own shocks are the most
important. For stock markets, own shocks are most important followed respectively by banking sector and remittances shocks.
Similarly, for banks, own shocks are prime, mostly followed by shocks from stock markets. Remittances respond mainly to own
shocks and shocks to stock markets.
5. Conclusion
What dynamic and causal relationships characterise remittances, stock market development, and banking sector development? Do these
1423
H. Issahaku et al. Research in International Business and Finance 42 (2017) 1413–1427
Table 6
Relationship among banks, stock markets and remittances based on the degree of banking sector development −2SLS estimates.
(19) (20) (21) (22) (23) (24)
VARIABLES SMD SMD BSD BSD REMIT REMIT
Below median Above median Below median Above median Below median Above median
(≤2.7052) (> 2.7052) (≤2.7052) (> 2.7052) (≤2.7052) (> 2.7052)
SMD 0.268*** 0.269*** −0.154 −0.960***
(0.0377) (0.0228) (0.305) (0.261)
BSD 0.562*** 0.973*** 1.171** −0.0605
(0.102) (0.101) (0.593) (0.464)
REMIT 0.0186 −0.0530*** 0.0116 −0.00167
(0.0146) (0.0154) (0.0105) (0.00753)
FDI 0.00910 −0.0252 −0.0240** 0.0116 −0.226*** −0.304***
(0.0209) (0.0232) (0.0111) (0.00926) (0.0815) (0.109)
INV −0.0853 0.0431 0.144*** 0.209*** −1.138*** 0.574*
(0.0593) (0.0873) (0.0394) (0.0460) (0.303) (0.345)
TRADE −0.170** −0.234*** 0.0878** 0.176*** −0.0541 0.578***
(0.0689) (0.0449) (0.0408) (0.0272) (0.282) (0.181)
INST −0.102 −0.415** 0.0367 0.129 2.747*** −1.936***
(0.175) (0.189) (0.106) (0.0999) (0.935) (0.743)
CPI −0.155** −0.0686 0.0646* 0.136*** 0.190 0.216
(0.0668) (0.0892) (0.0363) (0.0465) (0.277) (0.328)
INCOME 0.0273* 0.00288 −0.0115 −0.0269* 0.0853 −0.183
(0.0142) (0.0294) (0.0111) (0.0154) (0.0751) (0.124)
Constant 2.115*** 2.450** −0.633 −1.436*** −8.449** 4.616
(0.656) (0.989) (0.398) (0.505) (3.856) (3.672)
Diagnostics
Observations 267 294 267 294 267 294
R-squared 0.187 0.348 0.186 0.457 0.111 0.127
Wald Chi2 (stat) 57.66*** 226.06*** 72.92*** 334.52*** 37.33*** 46.69***
Woolridge’s OIR (p- 1.4885 (0.2225) 0.6018 (0.4379) 2.2634 (0.1325) 0.4158 (0.5190) 1.6168 (0.2035) 0.2504 (0.6168)
value)
Woolridge Rob. 1.2661 0.1433 6.2194** 2.7475 6.1265** 2.0585
Rob. Reg. 0.6551 0.0759 3.3064** 1.2027 2.6135* 0.9732
Table 6 presents 2SLS results for split samples based on the degree of banking sector development. Stock market development (SMD), banking sector development
(BSD) and remittances (REMIT) are the dependent variables. The control variables are financial openness (FDI), trade openness (TRADE), investment (INV), in-
stitutional quality (INST), inflation (CPI) and per capita GDP growth (INCOME). All variables are defined in Table 1 The diagnostic tests reported include: (1) number
of observations, (2) R squared test of explanatory power, (3) Wald Chi-square for joint significance of coefficients, (4) the Woolridge’s overidentifying restrictions test
of which the null hypothesis is that the instruments are valid, (5) Woolridge’s robust score test for exogeneity of which the null hypothesis is that remittances is
exogenous, (6) Robust regression exogeneity test of which the null hypothesis is that remittance is exogenous. Robust standard errors are in parentheses. *, **, ***
respectively denote significance at 10%, 5%, 1% level of significance.
characterised relationships, if at all they exist, vary according to the amount of remittances received and to the state of banking sector
development? To shed some light on these crucial questions, in this paper we assessed the interactive linkages among remittances, banking
sector development and stock market quality using data for a large number of developing countries. The data cover the period 1999–2013.
We also constructed a multidimensional index each for banking and stock market development, which unlike the financial development
proxies in the extant literature, captures the size, access, efficiency and stability dimensions of banking sector and stock market development.
We find that remittances promote banking sector development in less remittance dependent countries, but not in remittance dependent
countries. Another interesting finding is the significant and positively reinforcing mutual bond between banking sector development and
stock market advancement. Thus, it is fairly conclusive that banks and stock markets in developing countries are mutually dependent. The
association between stock market development and remittances is quite complex. We find a bi-causal negative connection between stock
markets and remittances in countries with advanced banking systems. In low remittance recipient countries, remittances decrease stock
market development; however, in remittance dependent countries, remittances promote stock market development. Again, stock market
development obstructs remittance inflows in low remittance recipient countries; however, stock market development promotes remittance
inflows in remittance dependent countries. These findings hold after (i) controlling for endogeneity, (ii) incorporating other key explanatory
variables, (iii) employing alternative estimation procedures such as OLS, FE, RE and 2SLS.
The findings have some implications for policy. The instances of negative relationships between stock market development and
remittances confirm lingering doubts about the quality of stock markets in developing countries. Developing country stock markets
are mostly poorly capitalised, illiquid, volatile, and are often anchored by poor infrastructure, compared to their advanced country
counterparts. A number of measures can be undertaken to reverse this arrangement. First, other developing countries can follow the
steps of India, China and Ethiopia by taking advantage of their citizens abroad to issue diaspora bonds. Second, a stock market
centred educational campaign on financial literacy by the government and regulatory authorities can enhance citizen and migrant
participation in stock markets. Lastly, massive investments in stock market infrastructure by the state and development partners, and
1424
H. Issahaku et al. Research in International Business and Finance 42 (2017) 1413–1427
Fig. 2. Stability of PVAR.
Fig. 3. Impulse responses.
an easing of stock market listing conditions by regulatory authorities will also help boost stock market development.
The non-linear link between banking sector development and remittances is noteworthy, especially the fact that remittance does not
promote banking sector quality in high remittance receiving countries. The implication is that remittance dependent countries cannot rely on
remittance inflows to advance banking sector development. High remittance receiving countries can promote banking sector development by
upgrading capital markets, securing trade competitiveness and by building a pro-banking sector domestic investment framework.
Overall, we present perhaps the first evidence of a bidirectional causality among remittances, banking sector development and
stock market quality in a single study. We also construct unique composite measures for banking sector development and stock
market development. While these measures may not be the best yet, they certainly advance the methodology for measuring financial
market development. We encourage future studies to further explore the construction of multidimensional indices for banking sector
and stock market development, and to use them to re-assess the ‘fragile’ bond between finance, remittances and economic growth.
1425
H. Issahaku et al. Research in International Business and Finance 42 (2017) 1413–1427
Declaration of interest
Authors declare that there are no conflict of interests of any kind, either actual or potential.
Appendix A
See Table A1.
Table A1
Panel least squares estimates for models that failed the endogeneity test.
(25) (26) (27) (28) (29) (30) (31) (32)
Regressors SMD SMD SMD REMIT SMD SMD BSD REMIT
Full Sample Below median Above median Above median Below median Above median Above median Above median
remittance remittance remittance BSD BSD BSD BSD
SMD 0.187 0.265*** −0.916***
(0.115) (0.0202) (0.224)
BSD 0.656*** 0.714*** 0.585*** −0.303** 0.579*** 0.979*** −0.190
(0.0428) (0.0460) (0.0804) (0.149) (0.0797) (0.0904) (0.432)
REMIT −0.0292*** −0.0315 0.0453 0.0106 −0.0573*** −0.00321
(0.00954) (0.0197) (0.0297) (0.0121) (0.0143) (0.00740)
FDI −0.000250 −0.0448* 0.0548*** −0.120*** 0.0149 −0.0340 0.0129 −0.260**
(0.0138) (0.0250) (0.0168) (0.0442) (0.0182) (0.0220) (0.00881) (0.103)
INV −0.0379 −0.0444 −0.0124 0.197 −0.103** 0.0490 0.201*** 0.411
(0.0427) (0.0697) (0.0557) (0.148) (0.0510) (0.0788) (0.0421) (0.319)
TRADE −0.239*** −0.140*** −0.381*** 0.465*** −0.178*** −0.259*** 0.161*** 0.595***
(0.0376) (0.0428) (0.0615) (0.114) (0.0642) (0.0417) (0.0249) (0.168)
INST −0.198* −0.192 −0.403*** 0.668* −0.168 −0.226 0.182** −2.065***
(0.110) (0.182) (0.151) (0.352) (0.158) (0.162) (0.0858) (0.668)
CPI −0.114** −0.0567 −0.205*** 0.417*** −0.141** −0.0768 0.166*** 0.241
(0.0456) (0.0599) (0.0658) (0.115) (0.0549) (0.0752) (0.0410) (0.289)
INCOME 0.0266** 0.0235 0.0125 8.98e-05 0.0218* 0.0114 −0.0197 −0.159
(0.0121) (0.0165) (0.0190) (0.0415) (0.0127) (0.0240) (0.0125) (0.1000)
Constant 2.397*** 1.723** 4.055*** −5.057*** 2.394*** 1.780** −1.704*** 5.607*
(0.494) (0.693) (0.746) (1.595) (0.617) (0.871) (0.445) (3.358)
Obs. 663 318 345 345 319 344 344 344
R-squared 0.329 0.420 0.273 0.096 0.196 0.376 0.453 0.147
F-statistic 36.94*** 41.02*** 10.54*** 4.88*** 10.12*** 36.25*** 44.88*** 8.45***
Robust standard errors in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1.
Appendix B
See Table A2.
Table A2
List of countries included in the study.
1. Argentina 16. Egypt 31. Latvia 46. Peru 61. Vietnam
2. Armenia 17. El Salvador 32. Macedonia 47. Philippines
3. Bangladesh 18. Fiji 33. Malawi 48. Poland
4. Barbados 19. Georgia 34. Malaysia 49. Romania
5. Bolivia 20. Ghana 35. Mexico 50. Russia
6. Bosnia and Herzegovina 21. Guyana 36. Mongolia 51. South Africa
7. Botswana 22. Hungary 37. Morocco 52. Sri Lanka
8. Brazil 23. India 38. Namibia 53. Tanzania
9. Bulgaria 24. Indonesia 39. Nepal 54. Thailand
10. Burkina Faso 25. Iran 40. Nigeria 55. Trinidad and Tobago
11. China 26. Jamaica 41. Oman 56. Tunisia
12. Colombia 27. Jordan 42. Pakistan 57. Thailand
13. Costa Rica 28. Kazakhstan 43. Panama 58. Turkey
14. Cote d'Ivoire 29. Kenya 44. Papua New Guinea 59. Uganda
15. Croatia 30. Kyrgyz Republic 45. Paraguay 60. Ukraine
1426
H. Issahaku et al. Research in International Business and Finance 42 (2017) 1413–1427
References
Acosta, P.A., Lartey, E.K.K., Mandelman, F.S., 2009. Remittances and the Dutch disease. J. Int. Econ. 79 (1), 102–116. http://dx.doi.org/10.1016/j.jinteco.2009.06.
007.
Adenutsi, D.E., 2011. Financial development, international migrant remittances and endogenous growth in Ghana. Stud. Econ. Finance 28 (1), 68–89. http://dx.doi.
org/10.1108/10867371111110561.
Aggarwal, R., Demirgüç-Kunt, A., Pería, M.S.M., 2011. Do remittances promote financial development? J. Dev. Econ. 96 (2), 255–264. http://dx.doi.org/10.1016/j.
jdeveco.2010.10.005.
Arestis, P., Demetriades, P.O., Luintel, K.B., 2001. Financial development and economic growth: the role of stock markets. J. Money Credit Banking 33 (1), 16–41.
Beck, T., Levine, R., 2004. Stock markets, banks, and growth: panel evidence. J. Banking Finance 28 (3), 423–442.
Bettin, G., Zazzaro, A., 2011. Remittances and financial development: substitutes or complements in economic growth? Bull. Econ. Res.. http://dx.doi.org/10.1111/j.
1467-8586.2011.00398.x. (no-no).
Billmeier, A., Massa, I., 2009. What drives stock market development in emerging marketsinstitutions, remittances, or natural resources? Emerging Markets Rev. 10
(1), 23–35. http://dx.doi.org/10.1016/j.ememar.2008.10.005.
Boyd, J.H., Smith, B.D., 1998. The evolution of debt and equity markets in economic development. Econ. Theory 12 (3), 519–560.
Brown, R.P.C., Carmignani, F., Fayad, G., 2013. Migrants’ remittances and financial development: macro- and micro-level evidence of a perverse relationship. World
Econ. 36 (5), 636–660. http://dx.doi.org/10.1111/twec.12016.
Cheng, S.-Y., 2012. Substitution or complementary effects between banking and stock markets: evidence from financial openness in Taiwan. J. Int. Financial Markets
Institutions Money 22 (3), 508–520. http://dx.doi.org/10.1016/j.intfin.2012.01.007.
Chowdhury, M.B., 2011. Remittances flow and financial development in Bangladesh. Econ. Modell. 28 (6), 2600–2608. http://dx.doi.org/10.1016/j.econmod.2011.
07.013.
Christophers, B., 2015. Against (the idea of) financial markets. Geoforum 66, 85–93.
Cooray, A., 2012. Migrant remittances, financial sector development and the government ownership of banks: evidence from a group of non-OECD economies. J. Int.
Financial Markets Institutions Money 22 (4), 936–957.
Demirgüç-Kunt, A., Córdova, E.L., Pería, M.S.N., Woodruff, C., 2011. Remittances and banking sector breadth and depth: evidence from Mexico. J. Dev. Econ. 95 (2),
229–241.
Ezeoha, A.E., 2013. Financial determinants of international remittance flows to the sub-Saharan african region. Int. Migration 51, e84–e97. http://dx.doi.org/10.
1111/imig.12061.
Giuliano, P., Ruiz-Arranz, M., 2009. Remittances, financial development, and growth. J. Dev. Econ. 90 (1), 144–152. http://dx.doi.org/10.1016/j.jdeveco.2008.10.
005.
Issahaku, H., Harvey, S.K., Abor, J.Y., 2016. Does development finance pose an additional risk to monetary policy? Rev. Dev. Finance 6 (1), 91–104.
Levine, R., 2005. Finance and growth: theory and evidence. Handbook of Economic Growth, vol. 1. pp. 865–934.
Nyamongo, E.M., Misati, R.N., Kipyegon, L., Ndirangu, L., 2012. Remittances, financial development and economic growth in Africa. J. Econ. Bus. 64 (3), 240–260.
http://dx.doi.org/10.1016/j.jeconbus.2012.01.001.
Ramirez, M.D., 2013. Do financial and institutional variables enhance the impact of remittances on economic growth in Latin America and the Caribbean? A Panel
Cointegration Analysis. Int. Adv. Econ. Res. 19, 3-273-288. http://dx.doi.org/10.1007/s11294-013-9407-2.
Ruiz, I., Vargas-Silva, C., 2010. Monetary policy and international remittances. J. Developing Areas 43 (2), 173-186.
Singh, A., 1997. Financial liberalisation, stockmarkets and economic development. Econ. J. 107 (442), 771–782.
WEF, 2013. The Financial Development Report 2012. In: Reuttner, I. (Ed.), The World Economic Forum, Geneva, Switzerland.
1427