Motives for Remittances Change During the Financial Crisis in Bosnia and Herzegovina

Introduction

Up to now, the investigation of motives for sending remittances has commonly been conducted within a framework that analyses the ‘crowding-out’ effect influencing the relation between social transfers and remittances, in which a negative relationship would suggest altruistic motives, whereas a positive relationship would be indicated by other motives such as insurance, inheritance, or the exchange of services.^[1] The ongoing global economic crisis offers a unique opportunity to investigate whether and how the motives to send remittances were changed by the crisis, which affected both their senders and their recipients. Analysis of two Household Budget Survey (HBS) datasets from Bosnia and Herzegovina (BiH), one from 2007 (before the crisis) and another from 2011 (during the crisis), allows the estimation of a model for the relationship between social transfer receipts and remittances to households. Comparison of findings from data collected in these two years yields insights into possible changes in the motives for sending remittances from 2007 to 2011, a period conspicuously marked by the onset of the 2008 economic crisis and its continuing effects. It would not be appropriate to conclude, on the basis of these results, that the change in motives was due to the crisis, since many other, unobservable factors might have influenced the motives for sending remittances; however, the comparison of motives between the two yearlong periods can provide useful insights.

During the war in BiH in the 1990s, about one quarter of the country’s total population fled the country.^[2] Even today, it is estimated that every third Bosnian lives abroad.^[3]As a result, the percentage of BiH’s GDP derived from remittances (around 13%) is among the highest in the world.^[4] Overall social transfers in BiH amount to about 14% of its GDP, the majority of which goes to pensions and health-care services.^[5] But in addition, a large proportion of these transfers is devoted to non-insurance-based benefits, among them veterans’ benefits, child-care allowances, and other forms of social assistance.

This situation, in which a nation has significant private and public transfers, motivates an analysis of the possible interaction between such transfers. The link between receipt of social and private transfers is known as the crowding-out effect.^[6] The sign of this effect, in addition, helps reveal possible motives for the sending of remittances. If the sign is negative, i.e. any increase in social transfers received by a household decreases the remittance income received; this suggests that remittances are driven by senders’ altruistic motives. On the contrary, if remittances are driven by the exchange motive, they will increase when social transfers increase.

Remittances are well known as a countercyclical international financial flow: their inflow to a country increases during economic downturns because migrants abroad send more money to their families at home if their need for support increases. The global economic crisis has changed this dynamic, because it has affected recipients and senders of remittances. Therefore, the cyclicality of these flows is an empirical question.

This study investigates the presence of the crowding-out effect in Bosnia and Herzegovina, and its change between 2007 and 2011, to answer the question: How do the motives for sending remittances change when senders face economic difficulties? The social transfers to be analysed here are restricted to non-contribution-based social transfers, such as veterans’ benefits and child-care allowances, because contribution-based benefits, such as unemployment benefits and pensions, cannot—unlike non-contribution-based transfers—be considered exogenous sources of income. As this study’s key objective is analysis of the crowding-out effect, then contribution-based benefits, which are received as alternatives to wages, should not be considered exogenous sources of income that increase their recipients’ overall incomes.

The paper is structured as follows. The next section reviews the literature on social transfers’ crowding-out effect on remittances. The third section briefly explains the characteristics of Bosnia-Herzegovina’s social security system and the inflows of remittances to the country. The fourth section provides the theoretical model for the non-monotonic crowding-out effect and presents new specifications for two empirical models: one that controls for such an effect, and another that does not. The method and the data used for these models are also described. Section five presents results of the estimation of different model specifications. Finally, section six explains the policy implications of this study’s results.

Literature on the Crowding Out Effect

Literature on social transfers’ crowding-out effect on remittances has been concerned with investigating its direction and the linearity of senders’ motives. Therefore, the theory about the effect has been empirically tested in cross-sectional settings. The ongoing global economic crisis offers a unique opportunity to investigate whether and how the motives for remittances were changed by the massive international downturn—a crisis that affected senders and recipient of remittances alike. This paper is the first contribution to the empirical literature that analyses the change in motives for the sending of remittances as a consequence of the recent global economic crisis.

The empirical literature analysing motives for sending remittances primarily uses the crowding-out framework, where the sign of the relationship between receipt of remittances and change in income, usually as a result of a change in the receipt of social transfers, is used to reveal possible motives for the sending of remittances. Remittances can be motivated by either altruism or exchange.^[7]Becker’s conception of altrustic motives for transfers is based on the idea of interdependent preferences. According to this idea, people have preferences about what others consume (e.g. parents have preferences about what their children consume) and consequently, the utility of transfers depends not only on their own consumption but also on that of others. Thus a migrant will increase the amount of remittance payments sent to family members left behind once their income has been negatively affected by adverse conditions in the migrants’ home country. So if these transfers are motivated by altruism, a decrease in a recipient’s income (as a result of crisis or reduced social transfers) will increase private transfers. This effect is interpreted as the crowding out of private transfers by public transfers.^[8] This crowding-out effect implies that social transfers’ positive effects can be neutralised by remittances, as the giving of support to vulnerable groups—the intended outcome of social transfers—will, at least partially, be transferred to senders of remittances.^[9]

The opposite effect is expected if the sending of remittances is motivated by exchange, resulting in remittances being ‘crowded in’ by public transfers.^[10] If private transfers are predominantly motivated by exchange, meaning that such transfers are used as payment for the provision of certain services by a recipient to a sender, then the sign of the relationship between them is positive, because the rise in income of the provider of services through the receipt of social transfers increases the ‘price’ of such services. Other subtypes of the ‘exchange’ motive can be explained in the same way. For example, if remittances are motivated by the sender’s self-interested intention to increase inheritance claims, then the recipient’s increase in income also increases the expected gain of inheritance and, consequently, remittance transfers can be seen as an ‘investment’. During a global economic crisis, however, the interplay of changes in income and social transfer among remittance senders and their recipients complicates the expected crowding-out effect. Moreover, such changes may affect people’s motivation to send remittances, which need to be tested empirically.

Up to now, the majority of empirical studies have failed to find a strong crowding-out effect between social transfers and remittances.^[11] One explanation might be that most of these studies did not account for the possibility of a non-monotonic relationship between public and private transfers.^[12] An increase in income may cause the motives for transfers to change, thus causing the sign of the relationship between public and private transfers to be different at different levels of recipients’ income. Such linear models are not correctly specified and therefore cannot capture the true crowding-out effect. In addition, empirical evidence from developed countries that possess a long history of public transfers that may already have replaced private transfers might be misleading. Therefore, recent studies have focused on the collection of evidence from developing countries, allowing for a non-monotonic relationship between public and private transfers. Cox et al. investigated this possibility by a threshold model and estimated the transfer derivatives to be - 0.4 for the poorest households and almost zero for richer households in the Philippines.^[13] In a study of the relationship between public pensions for the elderly and private transfers in South Africa, Jensen estimates that for each rand increase in public pension income, transfers made by people’s children were reduced by 0.25-0.30 Rand.yAlthough recent studies have addressed the issue of change in motives across income distribution, the present study is the first to also address the change in motives over time.

Remittances and Social Protection in Bosnia and Herzegovina. Background

Modelling the Crowding-Out Effect

This paper analyses the possible crowding-out effect of social transfers on remittances in order to find evidence about the dominant motives for sending remittances, as well as how these motives have changed during the recent global economic crisis. Its findings reveal the possible predominant motives for remitting by Bosnians and their dynamics over time, which can indirectly be interpreted as effects of the global economic crisis on these motives.^[18]

Two different model specifications are outlined here. In the first specification, the motives of remittances are assumed to be monotonic. In the second, assuming that the motives for remittances are non-monotonic, the impact of social transfers on remittances was estimated based on the theoretical description of the relationship provided by Cox et al.^[19] Cox et al. hypothesised that the relationship between social transfers and remittances should be negative at low levels of income, then begin to increase at some threshold level (K) such as the poverty line, as exchange motives supplant altruistic motives for the sending of remittances. But as income increases, the presence of exchange motives causes remittances first to increase (up to the recipient’s income level, depicted by I’ in figure 1) and then to decrease and eventually cease (at the income level depicted by I’’), making the relationship between social transfer and remittances negative in the first part and having an inverse U-shape in the second. Cox et al. suggest that the appropriate specification of the empirical model needs to be nonlinear to capture such a relationship and to account for non-monotonic motives by specifying a point where the break occurs. Moreover, the break where the receiving of remittances ceases needs to be identified. In terms of model specifications with alternative dependent variables, it should be noted that Cox et al. suggest the relationship outlined above for the amount of remittances received, but not necessarily for the likelihood that remittances will be received, which means that the theoretical discussion presented above of the non-monotonic and non-linear effect of social transfers on remittances should be controlled for in models only where the dependent variable is the amount of remittances.

Figure 1

Non-monotonic relationship between social transfers and remittances. Source: Cox / Hansen / Jimenez, How Responsive Are Private Transfers to Income?, 2199.

Compared to previous studies, this paper uses a new approach to the problems of non-monotonic motives for sending remittances.^[20] In the second model specification, the poverty line was chosen as a threshold point where the motives begin to change. We can assume that remittances to lower-income recipients may be more altruistically motivated, so they decrease after an increase in social transfers and their poverty-reduction goal may not necessarily be reached. Remittances to higher-income recipients may be motivated more by exchange and would therefore change in the same direction as social transfers. In such relations, social transfers would increase poverty and inequality among households, particularly when the receipt of remittances by the non-poor is matched with ineffective social transfer policies. There is some evidence to support these ideas, but most studies have been based on the analysis of cross-sectional data, which does not assure an appropriate capture of the dynamic effects of changes in transfers. Therefore, it is necessary to test these ideas by using datasets from different time periods.

Results

The results of the three models from specification (1) are presented in Table 2. These results reveal the average influence of social transfers on remittances, or the ‘average’ (monotonic) motive for sending remittances, regardless of poverty status.

To a large extent, these results are in line with those of previous empirical studies. They provide evidence regarding the expected influence of the key demographic characteristics on receipt of remittances, particularly place of residence, age, gender, and marital status of a household’s head. Married household heads receive more remittance income. Female-headed households also receive more in remittances, which is in line with most previous studies.^[23]Households with an older head receive less remittance income, all else being equal. We also found that several variables used in most of the previous studies, such as a household’s size and number of children living in it, were statistically insignificant, so they were excluded in the model’s final specification.

The main variable of interest in this study, the variable for social transfers, has a statistically significant and positive coefficient in the model estimated with both 2007 and 2011 data, as well as with 2007 data only. This result suggests that the receipt of social transfers generally increases the amount received through remittances; that is, remittances are primarily driven by the exchange motive. However, this direction changes in 2011, which would indicate that in a situation of worsened economic conditions among recipients, altruism becomes the main motive for sending remittances. This also suggests that remittances are still countercyclical, even when remittance senders are also experiencing the effects of an economic downturn.

Table 3 presents results of the estimation of three models that control for possible non-monotonic motives for remittances.

In model 3, the benchmark category is that of non-poor households. Consequently, the original coefficient for social transfers measures their effect on receipt of remittances by non-poor households. Its squared term tests the possibility of the nonlinearity of this effect among non-poor households. The coefficient for the interaction term between social transfers and the dummy variable for poor households measures the difference in the effect of social transfers on remittance receipts between the poor and the non-poor. As can be seen from Table 3, the coefficient in front of the tst variable was positive in 2007 and became negative in 2011, which suggests that the main motive for sending remittances to non-poor households changed from exchange in 2007 to altruism in 2011. The squared term was statistically significant in 2007, but was no longer so in 2011, which shows that the exchange motive was nonlinear, while the altruism motive was not. However, the coefficient in front of the interaction term between social transfers and a household’s poverty status (tstpoor) changed its sign from 2007 to 2011, which suggests that the main motive for sending remittances to poor households changed from altruism to exchange.

Conclusions

The empirical analysis presented above is a first study on the dynamics of motives for sending remittances during a situation of global economic crisis when both senders and recipients of remittances are experiencing a change in income. The results of the analysis should provide useful insight into the existence of a relationship between these two types of transfers and its change over time, possibly as a consequence of the global economic crisis.

The results suggest that the predominant motive for sending remittances to Bosnia-Herzegovina in 2007 was exchange, which changed to altruism in 2011. Remittance receipts during the crisis decreased on average as a result of an increase in social transfers, revealing that remittance payments were influenced by a crowding-out effect caused by social transfers. Controlling for a possible non-monotonic pattern in the motives for sending remittances, the evidence supports the hypothesis that different motives compelled the sending of remittances to poor versus non-poor people. Moreover, the change in motives over time was observed for both poor and non-poor groups: remittances paid to poor people were primarily altruistically motivated in 2007, but this changed to an exchange motive in 2011, whereas for non-poor households the motives and their change over time showed exactly the opposite pattern.

From a policy perspective, these results suggest that the effect of large, category-based, and inefficient social transfers is made additionally inefficient by the presence of the crowding-out effect on remittances. Consequently, inequality between the recipients and non-recipients of either private or public transfers widens further. Finally, inflows of remittances cannot be considered to be a remedy for inefficient forms of social transfer, but on the contrary raise the importance of the proper targeting of social transfers.