Saving out of Remittances: Evidence from Ethiopia and Kenya
| Author | Seife Dendir |
| DOI | http://doi.org/10.1111/imig.12343 |
| Published date | 01 August 2017 |
| Date | 01 August 2017 |
Saving out of Remittances: Evidence from
Ethiopia and Kenya
Seife Dendir*
ABSTRACT
This article examines the saving behaviour of remittance recipients in Ethiopia and Kenya.
The few existing estimates of savings from remittances, often obtained indirectly using expen-
diture analysis, vary widely. The analysis presented here relies on ordinal saving categories
reported in response to a direct survey question. The results reveal that the savings rate is
higher in Kenya than Ethiopia, in the raw data and the multivariate model, although in both
countries it rises with receipt size. Interestingly, gender is a robust predictor of differential sav-
ings, albeit with a contrasting pattern: women save more than men in Ethiopia, while the
reverse is true in Kenya. In other results, a pre-migration agreement about remittances has a
strong positive effect on savings. Savings rates are also positively associated with financial
inclusion in Ethiopia and formal channel flows in Kenya. Many of these results are confirmed
by analysis of broader forms of saving/investment.
1. INTRODUCTION
In today’s global economy, remittances constitute a major part of capital flows to developing coun-
tries. According to estimates by the World Bank, remittances sent to these countries through official
channels totalled more than $440 billion in 2015. Official development assistance, in comparison,
amounts to about a third of remittance flows annually. Importantly, since rebounding in 2010 after
the financial crisis, remittance flows have grown steadily despite the continued weakness in the
world economy, further strengthening their role as a crucial source of financing for the developing
world (World Bank, 2014; World Bank, 2015).
A large body of evidence documents how households in developing countries deploy remittance
receipts for various purposes. On a basic level, poor households use remittances to augment con-
sumption and help alleviate the poverty burden (see Adams and Page, 2005; Yang and Martinez,
2006; Lokshin et al., 2010, among others). Households also use remittances to finance investment
in human capital of children. Remittance receipts are associated with reduced infant morbidity and
mortality, and improved rates of literacy and school attendance (Hildebrandt and McKenzie, 2005;
Lopez-Cordova, 2005). Moreover, remittances serve as insurance, helping smooth household con-
sumption in the advent of adverse shocks (Yang and Choi, 2007). Remittance receipts are also used
to fund entrepreneurial activities (Woodruff and Zenteno, 2007) and asset –typically land and
housing –purchases (Osili, 2004). In contrast, remittance flows could also engender perhaps unin-
tended, negative consequences such as appreciation of the local currency and reductions in labour
supply (Amuedo-Dorantes and Pozo, 2004; Kim, 2007).
* Radford University, Radford, Virgina
doi: 10.1111/imig.12343
©2017 The Author
International Migration ©2017 IOM
International Migration Vol. 55 (4) 2017
ISS N 00 20- 7985 Published by John Wiley & Sons Ltd.
An important issue regarding remittances is the extent to which households use them to finance
basic consumption vis-
a-vis investment. This is because the allocation of receipts between the two
types of use partly determines the long-term impact of remittances on economic development. As
noted, remittances are often used to compensate shortfalls in consumption due to low and uncertain
incomes, which is a vital contribution per se. However, it is fair to argue that remittances are likely
to have a more direct and enduring impact on household, local and wider economies if they finance
investments in education, health, and capital for entrepreneurial activities. The related literature has
therefore focused on examining household expenditure patterns and determining whether remit-
tances are mostly spent on so-called “consumption”or “investment”items (de Brauw and Rozelle,
2008; Adams and Cuecuecha, 2010; Zhu et al., 2012, among others).
Very few studies have directly investigated the saving behaviour of households relative to remit-
tance income. Arguably this is mostly due to the broader difficulty of measuring any kind of saving
with a reasonable degree of precision. Brown (1994) perhaps provides one of the earliest attempts
to directly examine the extent to which remittances are saved vis-
a-vis consumed. Using data from
the South Pacific (Western Samoa and Tonga), the study reports that contrary to the commonly
held view that remittances are used exclusively for consumption purposes, most recipients in fact
save and/or invest a considerable part of their receipts. Specifically, while close to 60 per cent of
households report financial savings, remittance-receiving households were more than 50 per cent
more likely to hold such savings than non-recipients. The average savings rate exclusively among
recipients was also positively related to remittance income.
More recently, Adams (2002) and Zhu et al. (2012) have estimated the marginal propensity to
save (MPS) out of remittance receipts explicitly. Using data from rural Pakistani households,
Adams (2002) measures saving as net real and financial saving (i.e. expenditure on land and hous-
ing, farm animals, education and financial savings) and estimates the MPS out of external remit-
tances to be 0.71, the second highest among seven different income sources. The study notes that
this is consistent with the precautionary model of savings, which says more variable (less pre-
dictable) income sources are likely to be saved at a much higher rate. Zhu et al. (2012) examine
the MPS from remittances of rural-urban migrant households in China. They define savings in two
ways: narrowly, as a residual –the difference between a household’s total income and living
expenses; and broadly, by adding expenditures on durables and education to the narrow measure.
The MPS estimated in their preferred specification is not statistically different from zero, however,
and at 0.20 for broad savings, is less than half the MPS from other income sources. The result
prompted them to conclude that migrant remittances are unlikely to increase household savings and
finance productive investment in rural China.
1
Other studies have made indirect inferences about the saving rate out of remittances by analyzing
income and expenditure data of recipient households. Often, the stated purpose of these studies is
to compare the propensity to consume at the margin of different goods from remittance income. If
the expenditure data cover a broad range possible consumption goods, the exercise therefore allows
them to roughly gauge the saving rate as a residual. However, the implied savings rate out of
remittance income from these studies also varies widely, depending on subject country, data,
methodology and type of remittance. Zhu et al. (2014), for instance, examine consumption patterns
of Chinese households and report that remittances are spent on non-housing expenditures ‘virtually
dollar-for-dollar’, for an implied MPS of zero.
2
Davies et al. (2009), in contrast, find very low mar-
ginal propensity to consume (MPC) out of remittances by Malawian households, with estimates of
0.07 and 0.20 for male and female headed families, respectively. Since they consider total non-dur-
able consumption, the implied MPS must thus be very high.
3
Adams and Cuecuecha (2010) and
(2013) examine expenditure patterns of remittance receiving households in Guatemala and Ghana,
respectively. Although they do not estimate the MPC per se, they find that the marginal budget
share spent on ‘investment goods’(education, housing and health) rises with remittance receipt –
vis-
a-vis the counterfactual of non-receipt –while that on food consumption declines. A similar
Saving out of Remittances: Evidence from Ethiopia and Kenya 119
©2017 The Author. International Migration ©2017 IOM
To continue reading
Request your trial