Revisiting the Effects of Remittances on Bank Credit: A Macro Perspective
| DOI | http://doi.org/10.1111/sjpe.12086 |
| Date | 01 November 2015 |
| Published date | 01 November 2015 |
REVISITING THE EFFECTS OF
REMITTANCES ON BANK CREDIT:
A MACRO PERSPECTIVE
Richard P.C. Brown*and Fabrizio Carmignani**
ABSTRACT
We investigate the effect of remittances on bank credit in developing countries.
Understanding this link is important in view of the growing relevance of remit-
tances as a source of external finance and of the beneficial impact that financial
intermediation is likely to have on economic growth. Our contribution is twofold.
First, we present a theoretical model of bank credit in a hypothetical remit-
tances-receiving country where: (1) the banking sector is imperfectly competi-
tive; and (2) bank rates change infrequently because of the presence of
adjustment costs. We show that in equilibrium, the relationship between remit-
tances and bank-credit is likely to be non-linear. Second, we look at the evidence
using a panel data set for a large group of developing and emerging economies
over the period 1970–2009. We find that at initially low levels of remittances, an
increase in remittances reduces the volume of credit extended by banks. How-
ever, at sufficiently high levels of remittances, the effect becomes positive. The
turning point of the relationship occurs at a level of remittances of about 2.5%
of GDP, which would imply that approximately 50% of our sample lies to each
side of this threshold.
II
NTRODUCTION
In capital scarce economies, individuals are often trapped in poverty not
because they lack profitable investment opportunities, but because they lack
the initial finance needed to undertake such investments. It is therefore not
surprising that the expansion of credit markets and banking services is now
regularly placed at the core of development and poverty reduction strategies.
1
Meanwhile, in recent years, remittances from abroad have come to be recog-
nized as an important source of external finance for households in many
developing countries. It is then important to understand how the inflow of
*The University of Queensland
**Griffith University
1
There is for instance a voluminous body of evidence suggesting that financial develop-
ment promotes economic growth. See Loayza and Ranciere (2006), Mishkin (2007), and ref-
erences therein.
Scottish Journal of Political Economy, DOI: 10.1111/sjpe.12086, Vol. 62, No. 5, November 2015
©2015 Scottish Economic Society.
454
remittances affects the volume of credit provided domestically. On the one
hand, remittances might substitute for domestic credit as individuals who
receive remittances would not need to borrow from the domestic banks (or
other financial institutions) to invest. On the other hand, remittances might
crowd-in domestic credit by increasing the volume of resources that banks can
potentially mobilize and therefore lend. Moreover, individuals who receive
remittances might become more financially literate and hence express a stron-
ger demand for banking services and financial intermediation. In this study,
we show that the balance between these two opposing effects depends on the
volume of remittances received by a country. More specifically, when a coun-
try receives only a small amount of remittances, their effect on credit is nega-
tive. However, when the flow of remittances grows beyond a certain
threshold, the effect becomes positive.
Our analysis is related to a burgeoning body of research that studies the
relationship between remittances and broad financial development. Findings
from this literature indicate that while remittances are often used to over-
come credit constraints in financially underdeveloped economies (Giuliano
and Ruiz-Arranz, 2009; Ramirez and Sharma, 2009; Arezki and Bruckner,
2011; Combes et al., 2011; Anzoategui et al., 2014), their aggregate effect on
financial deepening is on balance positive (Hunte, 2004; Martinez Peria
et al., 2008; Billmeier and Massa, 2009; Gupta et al., 2009; Aggarwal et al.,
2011; Demirguc-Kunt et al., 2011). However, there is also some (limited)
micro-level evidence suggesting that the likelihood of a household having a
bank account does not depend on the volume of remittances received
(Mogilevsky and Atamanov, 2009; Brown et al., 2013). This would in turn
indicate that remittances do not necessarily increase household’s demand for
banking services.
2
The existing work is almost exclusively empirical and focuses on the estima-
tion of linear relationships. In our study, instead, we combine theory and
empirics to show that the relationship between remittances and credit pro-
vided by banks is likely to be non-linear. In the first part of our study, we
present a theoretical model of banking and remittances in a hypothetical
developing economy where the credit market is imperfectly competitive. We
prove that the existence of positive adjustment costs (a likely occurrence in
many economies) may be sufficient to generate a non-linear effect of remit-
tances on bank credit. The intuition for this result is as follows. A representa-
tive bank sets deposit and lending rates to maximize its profit. Starting from
an initial position where the rates are set at the profit maximizing level, an
increase in remittances reduces the demand for loans. This has two conse-
quences: (1) the volume of credit provided by the bank declines (as less credit
is demanded) and (2) the bank is no longer maximizing profits. To return to a
profit-maximizing equilibrium, the bank should reset its rates. However,
positive adjustment costs imply that the bank will do that only if the loss of
2
A related branch of this literature looks at the effect of financial development and finan-
cial transaction costs on remittances (Freund and Spatafora, 2008; Adenutsi, 2011; Mokerjee
and Roberts, 2011).
REVISITING THE EFFECTS OF REMITTANCES ON BANK CREDIT 455
Scottish Journal of Political Economy
©2015 Scottish Economic Society
profits from not adjusting is sufficiently large to make the adjustment of rates
worthwhile. This loss of profits in turn is proportional to the inflow of remit-
tances. So, for low levels of remittances, the loss is small, the bank does not
adjust the rates, and credit declines. But for sufficiently high levels of remit-
tances, the avoided loss is large enough to offset the adjustment cost and
hence the bank resets its rates to stimulate greater demand for loans. As a
consequence, more credit is provided in the new equilibrium.
Is there any evidence that the relationship between remittances and bank
credit is non-linear? In the second part of our study, we look at the data.
A simple scatter plot with a non-parametric fit suggests that the relation-
ship is possibly U-shaped, meaning that the positive effect of remittances
on credit occurs only at sufficiently high levels of remittances. Accordingly,
we estimate a regression model of credit using a quadratic specification,
with linear and squared remittances as explanatory variables. This estima-
tion exercise is conducted using a panel data set for a large group of coun-
tries over the period 1970–2009. In the baseline regression, we control for
several other determinants of credit and we deal with unobserved country
heterogeneity using standard fixed effects estimators. We also present
dynamic specifications that include a lagged dependent variable in addition
to other controls. These dynamic specifications are estimated using the dif-
ference-GMM estimator. Two key findings emerge. First, when remittances
are entered linearly in the credit equation, their estimated coefficient is
positive, but generally not different from zero at usual confidence levels.
Second, in a quadratic specification, the coefficient of the linear term on
remittances is negative, the coefficient of the quadratic term is positive, and
both coefficients are strongly significant. This is evidence that the effect of
remittances on credit is U-shaped rather than linear. The choice of a para-
metric (quadratic) specification might, however, be restrictive. So, we test
the robustness of our findings by estimating threshold effect regressions as
in Hansen (1999), Masten et al. (2008) and Combes et al. (2011). These
further estimates substantially confirm the results obtained from the qua-
dratic specification.
The rest of the study is organized as follows. Section II presents the theo-
retical argument. Section III is devoted to the data analysis. Section IV con-
cludes. Some mathematical proofs, additional econometric results, and data
description and sources are given in the Appendix.
II THEORETICAL FORMALIZATION
We start our analysis by presenting a stylized model of the credit market of a
developing or emerging economy that receives remittances. The focus is not
on the channels through which remittances are transferred from abroad, but
rather on the impact that the inflow of remittances has on the domestic supply
and demand of credit.
Our formalization of the credit market hinges on two assumptions. First,
the market is imperfectly competitive. More specifically, in line with the
456 RICHARD P. C. BROWN AND FABRIZIO CARMIGNANI
Scottish Journal of Political Economy
©2015 Scottish Economic Society
To continue reading
Request your trial