The World Bank Economic Review, 31(1), 2017, 1–23 doi: 10.1093/wber/lhv053 Article Remittances and Vulnerability in Developing Countries Giulia Bettin, Andrea F. Presbitero, and Nikola L. Spatafora Abstract This paper examines how international remittances are affected by structural characteristics, macroeconomic conditions, and adverse shocks in recipient economies. We exploit a novel, rich panel data set, covering bilat- eral remittances from 103 Italian provinces to seventy-nine developing countries over the period 2005–2011. We find that remittances are negatively correlated with the business cycle in recipient countries and in particu- lar increase in response to adverse exogenous shocks, such as large terms-of-trade declines. This effect is stron- ger where the migrant communities have a larger share of newly arrived migrants. Finally, we show that recipient-country financial development is negatively associated with remittances, suggesting that remittances help alleviate credit constraints. JEL classification: F33, F34, F35, O11 Key words: Remittances, Shocks, Business Cycles, Vulnerability This paper examines the drivers of remittances, with a focus on whether remittances should be viewed as a countercyclical shock absorber, helping smooth consumption during a downturn in recipient econo- mies, in contrast to the typically pro-cyclical private capital flows. This issue is particularly salient for two reasons. First, remittances to developing countries have grown steadily relative to capital flows, spurred by growing migration (figure 1). Remittances to developing countries are projected to reach USD 435 billion in 2014, more than three times the size of official development assistance, and USD 500 billion by 2017 (World Bank 2014). Second, remittances have proved very resilient since the onset of the global financial crisis (World Bank 2015).  Politecnica delle Marche (Italy) and MoFiR; her email Giulia Bettin is an assistant professor of economics at the Universita address is: g.bettin@univpm.it. Andrea F. Presbitero is an economist at the International Monetary Fund and MoFiR; his email address is: apresbitero@imf.org. Nikola Spatafora (corresponding author) is lead research economist for East Asia and the Pacific at the World Bank; his email address is: nspatafora@worldbank.org. We thank Giacomo Oddo, Roberto Tedeschi, and Simonetta Zappa at the Bank of Italy for clarifications on the remittances dataset. We also thank Andrew Foster, Caglar Ozden, Dilip Ratha, three anonymous referees, and participants at the 4th International Conference “Economics of Global Interactions: New Perspectives on Trade, Factor Mobility and Development,” the CSAE Conference 2014: “Economic Development in Africa” (Oxford), and seminars at the International Monetary Fund and the World Bank for useful comments. This research is part of a project on Macroeconomic Research in Low-Income Countries (project id: 60925), supported by the U.K. Department for International Development. The views expressed in this article are those of the authors and do not necessarily represent the views of the International Monetary Fund, the World Bank, or the policies of these institutions. Funding to pay the Open Access publication charges for this article was provided by the U.K. Department for International Development, project on Macroeconomic Research in Low-Income Countries (project id: 60925). C The Author 2015. Published by Oxford University Press on behalf of the International Bank for Reconstruction and Development / THE WORLD BANK. V This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs licence (http://creativecom- mons.org/licenses/by-nc-nd/4.0/), which permits non-commercial reproduction and distribution of the work, in any medium, provided the original work is not altered or transformed in any way, and that the work is properly cited. For commercial re-use, please contact journals.permissions@oup.com 2 Bettin, Presbitero, and Spatafora Figure 1. Capital Flows in Developing Countries Sources: World Development Indicators and International Debt Statistics (developing countries are defined as low and middle income countries), The World Bank. The existing empirical evidence on the role of remittances as a shock absorber is inconclusive. Some studies suggest that remittances are countercyclical with respect to output in the recipient country (Chami, Jahjah, and Fullenkamp 2005) because they are driven by altruism (Agarwal and Horowitz 2002; Osili 2007) or because household members migrate as part of a risk-diversification strategy aiming to insure against income shocks (Yang and Choi 2007). Other studies emphasize that remittances can be pro-cyclical, because migrants’ decision to remit is also driven by factors such as investment in physical and human capital (Yang 2008; Adams and Cuecuecha 2010; Cooray and Mallick 2013). This paper re-examines the question using a novel, rich panel dataset, covering bilateral remittances from 103 Italian provinces to seventy-nine developing countries over the period 2005–2011. In this data- set, remittances display significant variability, both over time and across source provinces and recipient countries. Italy can be considered a representative case study given that remittance outflows increased substantially in the last years, notwithstanding the global slowdown, and they flow to all world regions and are not limited to neighboring regions such as Eastern Europe, Central Asia, or Northern Africa. Specifically, the paper makes three main contributions to the literature. First, the availability of bilat- eral panel data for a large sample of recipient economies makes it possible to analyze systematically the correlation between remittances and the business cycle in recipient countries, controlling for numerous time-invariant factors at the level of the source-recipient pair. Relatedly, we consider separately the cycli- cal and trend components in GDP per capita. In addition, we control for specific factors of vulnerability in recipient countries, including in particular natural disasters, large declines in the terms of trade, and armed conflicts. In contrast, the existing literature focuses mainly either on bilateral remittances for a limited sample of recipient countries (Lueth and Ruiz-Arranz 2008; Frankel 2011; Docquier, Rapoport, and Salomone 2012) or on country pairs, such as the US-Mexico or Germany-Turkey corridors (Sayan 2004; Vargas-Silva 2008). Our results are also less likely to be biased by endogeneity, since remittances from Italian provinces (either individually or in the aggregate) are unlikely to affect significantly the busi- ness cycle in the recipient country. The World Bank Economic Review 3 Second, we analyze potential sources of heterogeneity in the response of remittances to economic con- ditions in recipient countries. In particular, we investigate whether remittances from source provinces with faster-growing, more recently arrived migrant communities respond more strongly to downturns in their countries of origin, consistent with more recent migrants being linked more closely by altruistic ties to their countries of origin. Third, the data on remittances cover the periods both before and after the 2007–08 global financial crisis. This is particularly relevant because the crisis affected both the migrants’ origin and host coun- tries, with an a priori ambiguous effect on remittances. On the one hand, the downturn in the country of origin might induce a positive change in remittances driven by altruism or insurance. On the other hand, the recession in the host country would reduce the income of migrants, including in particular temporary workers employed in the construction sector.1 Our key finding is that remittances from Italian provinces are negatively correlated with the business cycle in recipient countries and increase especially strongly in response to adverse exogenous shocks, such as declines in the terms of trade. This effect is stronger for communities with a larger share of newly arrived migrants. In addition, remittances are positively correlated with potential GDP in recipient coun- tries. These results are consistent with remittances being driven by both altruism and investment motives. Also, our results indicate that financial development in the recipient country is negatively associated with remittances, suggesting that remittances help alleviate credit constraints; further, international remittances and aid inflows are substitutes. The paper is structured as follows. Section 1 offers a detailed review of the existing literature on the macroeconomic determinants of remittances. Section 2 describes the data and the estimated model. Section 3 presents selected statistics about remittances outflows from Italian provinces to developing countries. Section 4 discusses the empirical results. 1. Determinants of Remittances: Existing Literature There exists a large literature on the determinants of migrants’ remittances.2 However, the empirical evi- dence remains inconclusive as to how remittances react to business cycles in the migrants’ home country and whether they help mitigate economic hardship. At the microeconomic level, some studies find that remittances increase to compensate relatives for negative shocks to their income—the altruism motive (Agarwal and Horowitz 2002). Other studies find a positive correlation between remittances and the economic conditions of families back home, suggesting that remittances are driven by self-interest motives such as investment or inheritance.3 In any case, positive shocks to migrants’ income in host countries are likely to translate into larger remittances (Bettin, Lucchetti, and Zazzaro 2012).4 Likewise, some macroeconomic studies find that remittances are negatively correlated with income levels in the recipient country (El-Sakka and McNabb 1999; Chami, Jahjah, and Fullenkamp 2005; Frankel 2011; Singh, Haacker, Lee, and Le Goff 2011), mitigate the adverse effect of food-price shocks on the level and instability of household consumption in vulnerable countries (Combes, Ebeke, Etoundi, 1 According to the Italian National Institute of Statistics, the unemployment rate for foreign-born workers increased from 10.2 percent in 2005 to 12.1 percent in 2011. 2 Rapoport and Docquier (2006) provide an exhaustive review of modern theoretical and empirical literature on remit- tances. See also Salomone (2009) and Stark (2009). 3 Lucas and Stark (1985) and Osili (2007) both show that remittances are positively correlated with the income of recipient households. Analogously, de la Briere, Sadoulet, de Janvry, and Lambert (2002) and Hoddinott (1994) show that remit- tances are positively correlated with household wealth. 4 Macroeconomic studies have considered a wide range of potential determinants, including exchange rates (Faini 1994), interest rate differentials (El-Sakka and McNabb 1999), the size of the diaspora abroad and transaction costs (Freund and Spatafora 2008), the skill and gender composition of migrant stocks (Faini 2007; Adams Jr. 2009; Niimi, Caglar, and Schiff 2010), and the interaction with immigration policies (Docquier, Rapoport, and Salomone 2012). 4 Bettin, Presbitero, and Spatafora and Yogo 2014), reduce output growth volatility in developing economies (Bugamelli and Paterno  2011; Chami, Hakura, and Montiel 2012), and react positively to natural disasters (Yang 2008; Mohapatra, Joseph, and Ratha 2012; Ebeke and Combes 2013). In contrast, other studies find that remittances are procyclical with respect to the recipient countries, consistent with an investment motive (Sayan 2004; Sayan 2006; Lueth and Ruiz-Arranz 2008; Giuliano and Ruiz-Arranz 2009; Durdu and Sayan 2010; Cooray and Mallick 2013). And some works do not find any significant correlation either with the busi- ness cycle in migrants’ home countries (Akkoyunlu and Kholodilin 2008) or with specific shocks such as armed conflicts (Naude ´ and Bezuidenhout 2012). Ruiz and Vargas-Silva (2014) argue that the cyclicality of remittances with respect to the recipient economy is country- or corridor-specific and unlikely to be stable over time. In particular, the degree of cyclicality may depend on other country-level characteristics.5 Relatedly, microeconomic evidence sug- gests that remittances decline over time, as migrants’ commitment and attachment to their relatives and their home country weakens—the “remittance decay” hypothesis.6 Recent studies have also investigated whether remittances represent an important channel in propa- gating global shocks (Barajas, Chami, Fullenkamp, and Garg 2010). Barajas, Chami, Ebeke, and Tapsoba (2012), in particular, show that remittances may prove destabilizing since they are more effec- tive in channeling economic downturns than booms from the source to the recipient countries. One important limitation of many cross-country analyses of the macroeconomic determinants of remittances is their reliance on aggregate remittance flows to developing countries, disregarding the het- erogeneity across source economies. To overcome this limitation, some studies use bilateral data on remittances to control for host countries’ characteristics, such as output fluctuations. In most cases, how- ever, the geographical coverage is limited to a single remittance corridor.7 In other cases, the geographi- cal perspective is wider but still limited. For instance, Lueth and Ruiz-Arranz (2008) use a panel dataset on bilateral remittances for eleven European and Asian recipient countries during the period 1980– 2004. The same dataset has been used in Barajas, Chami, Ebeke, and Tapsoba (2012), while Frankel (2011) merges their data with other bilateral data on remittances from the Inter-American Development Bank and the European Commission (Jimenez-Martin, Jorgensen, and Labeaga 2007). Docquier, Rapoport, and Salomone (2012) merge the sources used by Frankel (2011) with a database from the European Central Bank and a Romanian database; the resulting dataset include eighty-nine sending countries but is still limited to forty-six recipient countries, both developing and developed. Many existing analyses are also subject to endogeneity concerns. In many recipient countries, remit- tances represent a nonnegligible share of GDP; results might therefore be biased by reverse causality from remittances to GDP, with the exception of the few studies that focus on exogenous income shocks (see, for instance, Yang 2008). In cross-country aggregate analyses, this issue is addressed through GMM techniques (Cooray and Mallick 2013) or instrumental variables.8 Among studies using bilateral 5 For instance, Arezki and Bru ¨ ckner (2012) show that the impact of rainfall-driven income shocks on remittance inflows decreases with the level of financial development in the country. 6 Amuedo-Dorantes and Pozo (2006), de la Briere, Sadoulet, de Janvry, and Lambert (2002), Echazarra (2011), and Makina and Masenge (2015), among others, show the existence of an inverted-U shaped pattern of remittances over time, which is consistent with the “remittance decay” hypothesis (Poirine 1997). Remittances first grow due to an increase in migrants’ earning power and to the initial strong commitment to the relatives in the home country. Then, as the attachment becomes weaker and temporary migration often translates into permanent settlement, they tend to decrease over time. 7 Sayan (2004) and Akkoyunlu and Kholodilin (2008) focus on the Germany-Turkey remittance corridor, while Vargas-Silva (2008) and Ruiz and Vargas-Silva (2014) look at U.S.-Mexico remittances. Durdu and Sayan (2010) consider both corridors. 8 For instance, Abdih, Chami, Dagher, and Montiel (2012) analyze the impact of remittances on institutional quality and use geographical variables (coastal area, distance to the closest sending country) to instrument remittance inflows. Barajas, Chami, Fullenkam, Gapen, and Montiel (2009) introduce the ratio of remittances to GDP of all other recipient countries, which is likely to proxy for global reductions in transactions costs, as an instrument for remittances. The World Bank Economic Review 5 data, Frankel (2011) addresses endogeneity issues concerning the size of migrant stocks but disregards the possible bias related to recipient countries’ GDP. Lueth and Ruiz-Arranz (2008) acknowledge this problem but maintain that GMM estimates based on lagged recipient-country growth confirm their results. However, it remains unclear whether such estimates address the issue satisfactorily, given rising concerns about the performance of GMM in a context of weak instruments and over-fitting of endoge- nous variables (Roodman 2009; Bazzi and Clemens 2013). 2. Empirical Strategy and Data The Empirical Model To identify the effect of business cycle fluctuations and financial development on remittances we esti- mate a model in which remittances are a function of a set of independent variables constructed by exploiting information on migrants’ origin countries and bilateral information at the province-country level. In the baseline specification, total bilateral remittances between the source province i and the recip- ient country j at time t (REMi;j;t ) are a function of the logarithm of actual GDP per capita over potential GDP per capita in the recipient country (CYCLEj;t ), the log of trend GDP per capita (TRENDj;t ), the log of 1 þ the bilateral stocks of migrants (MIGRANTSi;j;t ), the log of population level (POPj;t ), the log of official aid per capita (AIDj;t ), and the log of the share of credit to the private sector over GDP (FINDEVj;t ): REMi;j;t ¼ a1 CYCLEj;t þ a2 TRENDj;t þ b1 MIGRANTSi;j;t (1) þ b2 POPj;t þ b3 AIDj;t þ b4 FINDEVj;t þ li;j þ ei;j;t where ei;j;t is the standard error term. To control for any time invariant bilateral unobservables, we include country-province pair fixed effects (li;j ) in equation 1.9 The key coefficient of interest is the response of remittances to the business cycle in the recipient coun- try (a1 Þ. If remittances are countercyclical with respect to output fluctuations in the recipient country (a1 < 0), this suggests an altruistic motivation behind transfers. A positive response of remittances to the long-run output trend in the recipient country (a2 > 0) instead offers evidence in favor of an invest- ment motive for remittances: investment-driven remittances may be particularly sensitive to long-term prospects in the migrants’ origin country, as proxied by trend output. Remittances to country j are expected to be positively associated with the stock of migrants from country j in a given province i (b1 > 0). Remittances might also be correlated with the total population of the recipient country (b2 > 0): for instance, this might be associated with lower costs of transferring remittances, or proxy for greater investment opportunities. The sign of the relationship between remittances and aid (b3 ) is a priori ambiguous. Remittances and aid could be positively correlated, as found by (Kpodar and Le Goff 2012), reflecting a positive impact of remittances on absorptive capacity in recipient countries or the capacity of the diaspora to influence foreign-aid policy in the host country (Milner and Tingley 2010). Conversely, remittances could be nega- tively correlated with aid (Amuedo-Dorantes, Pozo, and Vargas-Silva 2007) because they reduce the need for aid and hence donors’ willingness to provide it. 9 In the working paper version of the paper (Bettin, Presbitero, and Spatafora 2014), we also control for a set of province- specific control variables, including time-variant measures of the trend and cyclical component of GDP, population, and financial development at the provincial level. The main results about the counter-cyclical behavior of remittances with re- spect to economic conditions in the recipient countries are confirmed. When we control explicitly for provincial GDP, we find that remittances are influenced by economic conditions in the migrants’ host province, consistent with the view that developing countries that receive sizable remittance inflows are vulnerable to external shocks (Barajas, Chami, Ebeke, and Tapsoba 2012). 6 Bettin, Presbitero, and Spatafora As regards financial development in the recipient country, countries with more developed credit mar- kets could attract greater remittances (b4 > 0), as a result of either lower transaction costs (Freund and Spatafora 2008) or the capacity of an efficient banking system to channel profit-driven remittances towards growth-enhancing projects (Bettin and Zazzaro 2012). On the other hand, remittances and financial development may be substitutes (b4 < 0): migrants whose relatives have limited access to financial resources at home may transfer resources to relax liquidity constraints and fund either con- sumption or investments in physical and human capital (Giuliano and Ruiz-Arranz 2009). Since we are interested in assessing the potential role of remittances as shock absorbers in recipient countries, we augment equation 1 by including three specific factors of vulnerability for developing countries, which can plausibly be treated as exogenous: (1) an indicator equal to unity if country j expe- rienced natural disasters in year t (DISASTERj;t ); (2) an indicator equal to unity if armed conflicts occurred in country j at time t (WARj;t ); and (3) the log of the terms-of-trade index (TTj;t ). Adverse shocks in these exogenous variables, controlling for the cyclical component of output per capita, may be particularly likely to evoke a sympathetic (or, alternatively, insurance-type) response among migrants. A novel feature of our panel dataset is the bilateral source province-recipient country dimension. This makes it possible to assess whether the reaction of remittances to the cyclical component of GDP in the recipient country depends on the structure of the migrant community at the provincial level. In particu- lar, our hypothesis is that the strength of (altruistic) ties to the country of origin diminishes with time. Further, faster growing migrant communities will on average be more recently settled. We therefore aug- ment the baseline specification (equation 1) by interacting the CYCLEj variable with the growth rate of the bilateral stock of migrants, DMIGRANTSi;j;½t;tÀnŠ : REMi;j;t ¼ a1 CYCLEj;t þ a2 TRENDj;t þ b1 MIGRANTSi;j;t þ b2 POPj;t þ b3 AIDj;t þ b4 FINDEVj;t þ q1 DMIGRANTSi;j;½t;tÀnŠ (2) þ q2 DMIGRANTSi;j;½t;tÀnŠ à CYCLEj;t þ li;j þ ei;j;t The growth rate of the migrant stock is computed over time horizons n, ranging from two to four years, and is winsorized at the 99th percentile to minimize the effect of extremely high growth rates stemming from low initial migrant stocks. For robustness, we also estimate an alternative specification employing the growth rate of the migrant stock lagged by two years (that is, DMIGRANTSi;j;½tÀ2;tÀ4Š ) in both the level and interac- tion terms. Under the hypothesis that altruistic feelings decrease over time, a larger share of recently settled migrants will lead to a more negative response of remittances to the cyclical component of GDP (q2 < 0). The Estimator Since the dependent variable REMi;j;t has a significant share of nonrandomly distributed zeros (that is, many empty country-province cells), equations 1 and 2 are estimated using the Fixed Effects Poisson estimator. Despite deriving originally from the analysis of count data, the Poisson estimator can also be applied to non- negative continuous variables (Wooldridge 2010). Poisson regression estimates are consistent in the presence of heteroskedasticity and more efficient than standard OLS estimates, especially when considering large sam- ples. Thanks to its multiplicative form, the Poisson specification also provides a natural way to deal with zero observations in the dependent variable instead of either transforming or excluding them from the sample (Silva and Tenreyro 2006; Burger, van Oort, and Linders 2009). Additionally, in the Poisson model coeffi- cients on variables expressed in logarithms may be interpreted as elasticities (Cameron and Trivedi 2005). In the baseline estimates, we include country-province pair to control for unobservable heterogeneity; we also include time fixed effects and additionally control for the potential correlation of errors at the bilateral level by clustering standard errors by country-province pairs. In the robustness section we control for unobservables by including separately country and province fixed effects, instead of country-province pair fixed effects, so as to identify the effect of specific time- The World Bank Economic Review 7 Table 1. Variables: Definition, Sources, and Summary Statistics Variable Definition Source Mean St. Dev. REMi;j;t Total official remittances at constant prices from prov- Bank of Italy 0.888 8.431 ince i to country j in year t CYCLEj;t Logarithm of actual GDP over potential GDP in coun- WEO 0.002 0.019 try j in year t; potential GDP is calculated by apply- ing the Hodrick–Prescott filter (with the smoothing parameter set at 6.25) to the GDP series at constant prices TRENDj;t Logarithm of potential GDP in country j in year t, cal- WEO 10.772 2.445 culated by applying the Hodrick–Prescott filter (with the smoothing parameter set at 6.25) to the GDP ser- ies at constant prices MIGRANTSi;j;t Logarithm of 1 þ the stock of migrants living in prov- ISTAT 4.913 1.604 ince i and coming from country j in year t WARj;t Indicator ¼ 1 if country j experienced armed conflicts in UCDP/PRIO Armed 0.196 0.397 year t; both interstate and intrastate conflicts are con- Conflict Dataset sidered, in which the government of country j repre- sents one of the warring parties DISASTERj;t Indicator ¼ 1 if country j experienced natural disasters EM-DAT, CRED 0.768 0.422 in year t TTj;t Logarithm of the terms-of-trade index of country j WEO 4.651 0.235 POPj;t Logarithm of population in country j in year t WDI 16.910 1.537 FINDEVj;t Logarithm of the ratio of domestic credit to the private WDI 3.350 0.632 sector over GDP in country j in year t AIDj;t Logarithm of official aid per capita received in country j WDI 2.982 1.388 in year t FISCALBALANCEj;t Fiscal balance (þ surplus/ - deficit) as a share of GDP in WDI À0.017 0.040 country j in year t EXTERNAL DEBTj;t External debt stocks as a share of GDP in country j in WDI 0.384 0.237 year t EXECUTIVE CONSTj;t Constraint on the executives’ index in country j in year Polity IV - Center for 5.236 1.709 t (1 ¼ unlimited authority; 7 ¼ Executive parity or Systemic Peace subordination) DMIGRANTSi;j;tÀn Percentage growth rate of the migrant stock MIGRAN ISTAT 0.377 0.670 TSi;j;t calculated as ðMIGi;j;t À MIGi;j;tÀn Þ=MIGi;j;tÀn . with n ¼ 2; 3; 4 The variable is winsorized at the 99th percentile. Statistics are calculated for n ¼ 2. DISTANCEi;j Logarithm of the kilometric distance between province i Built-in STATA routine 8.415 0.764 and country j Notes: WDI: World Development Indicators (The World Bank); WEO: World Economic Outlook (International Monetary Fund); ISTAT: Italian National Institute of Statistic; CRED: Centre for Research on the Epidemiology of Disasters; PRIO: Peace Research Institute Oslo. Sources: Authors’ analysis based on data described in the text. invariant variables such as bilateral distance to the migrants’ country of origin. We also control for the potential correlation of errors at the bilateral level by clustering standard errors by country-province pairs. Data, Sample, and Sources The variables used in equations 1 and 2 are constructed using data collected from many sources. Here we provide an overview; a precise definition of each variable and of its sources is in table 1. The main data source is a detailed panel dataset on bilateral outward remittances from 103 Italian provinces to seventy-nine developing countries, compiled by the Bank of Italy (see table 2 for a list of 8 Bettin, Presbitero, and Spatafora Table 2. List of Countries Remittances from Italy / Remittances Remittances from Total remittance from Italy / top province outflows from Italy (%) GDP (%) in Italy / GDP (%) EAST ASIA AND THE PACIFIC Cambodia 0.0001 0.0001 0.0000 Indonesia 0.0011 0.0000 0.0000 Malaysia 0.0001 0.0000 0.0000 Philippines 0.1151 0.0054 0.0031 Thailand 0.0019 0.0001 0.0000 Vietnam 0.0003 0.0000 0.0000 EUROPE AND CENTRAL ASIA Albania 0.0256 0.0166 0.0019 Armenia 0.0001 0.0001 0.0000 Azerbaijan 0.0000 0.0000 0.0000 Belarus 0.0007 0.0001 0.0000 Bosnia and Herzegovina 0.0007 0.0003 0.0001 Georgia 0.0052 0.0037 0.0011 Kazakhstan 0.0003 0.0000 0.0000 Kyrgyz Republic 0.0005 0.0009 0.0003 Moldova 0.0113 0.0169 0.0018 Turkey 0.0042 0.0000 0.0000 Ukraine 0.0208 0.0012 0.0001 LATIN AMERICA AND THE CARIBBEAN Argentina 0.0056 0.0001 0.0000 Bolivia 0.0058 0.0028 0.0009 Brazil 0.0279 0.0001 0.0000 Colombia 0.0203 0.0007 0.0001 Costa Rica 0.0004 0.0001 0.0000 Dominica 0.0000 0.0001 0.0000 Dominican Republic 0.0171 0.0028 0.0005 Ecuador 0.0235 0.0031 0.0010 El Salvador 0.0025 0.0009 0.0006 Guatemala 0.0004 0.0001 0.0000 Haiti 0.0001 0.0001 0.0000 Honduras 0.0012 0.0006 0.0002 Jamaica 0.0002 0.0001 0.0000 Mexico 0.0010 0.0000 0.0000 Nicaragua 0.0003 0.0003 0.0001 Panama 0.0004 0.0001 0.0000 Paraguay 0.0011 0.0005 0.0001 Peru 0.0258 0.0016 0.0006 Venezuela, RB 0.0008 0.0000 0.0000 MIDDLE EAST AND NORTH AFRICA Algeria 0.0005 0.0000 0.0000 Egypt, Arab Rep. 0.0028 0.0001 0.0000 Iran, Islamic Rep. 0.0001 0.0000 0.0000 Jordan 0.0003 0.0001 0.0000 Lebanon 0.0003 0.0001 0.0000 Libya 0.0010 0.0001 0.0000 Morocco 0.0554 0.0050 0.0006 Tunisia 0.0148 0.0027 0.0002 The World Bank Economic Review 9 Table 2. (continued) Remittances from Italy / Remittances Remittances from Total remittance from Italy / top province outflows from Italy (%) GDP (%) in Italy / GDP (%) SOUTH ASIA Bangladesh 0.0281 0.0025 0.0007 India 0.0223 0.0001 0.0000 Pakistan 0.0090 0.0004 0.0001 Sri Lanka 0.0088 0.0017 0.0004 SUB-SAHARAN AFRICA Benin 0.0010 0.0012 0.0001 Burkina Faso 0.0023 0.0022 0.0004 Burundi 0.0001 0.0004 0.0000 Cameroon 0.0025 0.0009 0.0001 Cape Verde 0.0006 0.0031 0.0011 Chad 0.0001 0.0001 0.0000 Congo, Dem. Rep. 0.0011 0.0008 0.0002 Coˆ te d’Ivoire 0.0043 0.0015 0.0001 Ethiopia 0.0006 0.0002 0.0000 Gabon 0.0001 0.0001 0.0000 Gambia, The 0.0004 0.0031 0.0008 Ghana 0.0045 0.0012 0.0001 Guinea 0.0003 0.0005 0.0001 Guinea-Bissau 0.0002 0.0016 0.0002 Kenya 0.0013 0.0003 0.0001 Madagascar 0.0005 0.0005 0.0001 Mali 0.0011 0.0010 0.0001 Mauritania 0.0001 0.0002 0.0000 Mauritius 0.0004 0.0003 0.0001 Mozambique 0.0001 0.0001 0.0000 Niger 0.0002 0.0003 0.0000 Nigeria 0.0092 0.0004 0.0000 Senegal 0.0420 0.0261 0.0027 Seychelles 0.0000 0.0001 0.0000 Sierra Leone 0.0001 0.0005 0.0001 South Africa 0.0002 0.0000 0.0000 Sudan 0.0002 0.0000 0.0000 Tanzania 0.0007 0.0003 0.0001 Togo 0.0012 0.0032 0.0004 Uganda 0.0004 0.0002 0.0000 Zambia 0.0001 0.0001 0.0000 Note: Data computed as averages over the sample period (2005–2011). Sources: Authors’ analysis based on Bank of Italy data. recipient countries included in the sample).10 Our sample includes developing countries where data on the control variables for the extended model specification are available; this choice preserves a large cov- erage in terms of geographical and income distribution. To minimize the noise in the data and the effect of outliers, the estimation sample is limited to countries with at least 200 observations; in addition, we consider province-country pairs for which the bilateral stock of migrants is greater than fifteen 10 Data on remittance flows to 204 destination countries are collected as part of a monthly survey carried out by the Bank of Italy on a provincial basis since 2005. The dataset is publicly available at: www.bancaditalia.it/statistiche/rapp_ estero. 10 Bettin, Presbitero, and Spatafora individuals.11 The dataset provides data on remittances at constant prices and yearly frequency for the period 2005–11. Data consider only remittances sent through formal channels and predominantly reflect transfers carried out through money-transfer operators and the postal system. The banking system has been included in the survey only since 2010, and accounts for 5 to 10 percent of total remittances. All formal transactions are reported, regardless of the amount. As a caveat, the dataset does not include remittances sent through informal channels. Bilateral data on migrant stocks for the period 2005–11, collected by the Italian National Institute of Statistics (ISTAT) from the population registers of Italian municipalities, represent the stock of the for- eign population resident in each Italian province, by citizenship, at the beginning of each year. Data on the age structure of the foreign resident population in each province are unavailable. Instead, we use the growth rate of the number of migrants over a two-, three-, or four-year window in each province as a measure of how recently established a migrant community is. The data refer to official foreign residents and do not account for undocumented migrants residing in Italy. For each recipient country, GDP at constant prices for the period 1950–2012 is drawn from the IMF World Economic Outlook database. The cyclical and trend components are extracted using the Hodrick-Prescott filter. Data on total population for the period 2005–11, as well as the level of financial development, proxied by domestic credit to private sector as a share of GDP, are drawn from the World Development Indicators database. The annual frequency of natural disasters is drawn from the International Emergency Disasters data- base (EM-DAT) built by the Centre for Research on the Epidemiology of Disasters.12 Data on armed conflicts are drawn from the UCDP/PRIO Armed Conflict Dataset (Themne ´ r and Wallensteen 2013).13 The terms of trade are drawn from the IMF World Economic Outlook database. 3. Remittances from Italy to Developing Countries Total remittances from Italy to developing countries doubled between 2005 and 2011, reaching almost f 7 billion, in line with the growth in the stock of foreign residents in Italy (figure 2). After 2007, how- ever, the growth rate of remittances slowed down significantly, reflecting the impact of the global finan- cial crisis and of the euro area crisis on Italian output and unemployment. Indeed, remittances declined in 2010, although 2011 saw a rapid recovery, consistent with the global pattern of international remit- tances (figure 1). The geographic distribution of remittances from Italy largely mimics the global distribution (figure 3). The East Asia and Pacific region is the main recipient of both Italian and global remittances to devel- oping countries. The region’s share of remittances from Italy increased by 10 percentage points between 2005 and 2011. Europe and Central Asia’s share of remittances from Italy is twice as high as its share of global remittances, reflecting the relatively large number of migrants from Eastern Europe in Italy. South Asia accounts for a rising share of remittances from both Italy and the world. In contrast, Sub-Saharan Africa accounts for a limited share of remittances. 11 Our main results are robust to upward and downward changes in those two thresholds. 12 The data are accessible at www.cred.be/emdat/. A disaster is defined as a “situation or event, which overwhelms local capacity, necessitating a request to national or international level for external assistance.” Formally, an event is classi- fied as a disaster whenever it fulfills at least one out of four selection criteria: ten or more people killed; one hundred or more people affected, injured, or homeless following the disaster; declaration of a state of emergency; or calls for inter- national assistance. See www.emdat.be/criteria-and-definition. 13 The most recent version (4-2013) is available at www.pcr.uu.se/research/ucdp/datasets/ucdp_prio_armed _conflict_da- taset/. The World Bank Economic Review 11 Figure 2. Remittance Outflows to Developing Countries and Foreign Residents in Italy Sources: Bank of Italy and ISTAT. Focusing on individual countries, China, Romania, and the Philippines were the major recipients of remittances from Italy in both 2005 and 2011.14 Transfers to Bangladesh, Sri Lanka, and Georgia increased dramatically between 2005 and 2011. Colombia is the only country listed that registered a decrease in remittances from Italy over this period. The stock of resident migrants by country of origin is positively correlated with remittances to the relevant recipient country in 2011.15 The distribution of foreign residents and their recent growth are highly heterogeneous across Italian provinces. The share of migrants in the total population is much larger in the North than in the South, reflecting the economic divide between northern and southern regions (figure 4, panel a): over the period 2005–2010, foreign residents accounted for about 10% of the population in Prato, Brescia, and Reggio Emilia and 8.5% in Milan. In contrast, in some provinces in the South and the islands (for instance, Oristano, Taranto, and Cagliari), the share of migrants in the total population is around or just above 1%. Even in Naples and Palermo, the largest cities in the South, this share is about 1.5%, while in Rome foreign residents represent 6.6% of the population. Analyzing recent migration patterns, the growth in foreign residents has varied widely across provinces (fig- ure 4, panel b). Between 2005 and 2010, the overall stock of migrants almost doubled (figure 2), the median province experienced a 94% increase in foreign residents, and the number of migrants increased in all provin- ces. However, a quarter of all provinces experienced migrant growth rates below 77%, while the number of foreign residents increased by more than 138% in one-tenth of all provinces. The correlation between the growth rate of migrant stocks and the initial share of migrants in the population is negative, that is, provinces with relatively fewer foreign residents attracted more migrants. For instance, in Oristano, where foreign 14 The Italy-China remittance corridor was the single most important at the EU level in 2010. The Italy-Romania and Italy-Philippines corridors were among the ten biggest corridors from Europe. See http://epp.eurostat.ec.europa.eu/ statistics_explained/index.php/Migrant_remittance_and_cross-border_or_seasonal_compensation_transfer_statistics. 15 There are some outliers, notably China, whose share of total remittances significantly exceeds its share of total migrants. This may reflect an incorrect classification of some trade payments to China as remittances. Owing to this, and to the difficulties in estimating the cyclical component of Chinese GDP, we exclude China from the sample. 12 Bettin, Presbitero, and Spatafora Figure 3. Remittances by Region of Destination Sources: Bank of Italy and World Bank Migration & Remittances Factbook 2011. resident accounted for only 0.4% of the population in 2005, migrant stocks increased by 150%; in Prato, where foreign residents accounted for 7.8% of the initial population, migrant stocks increased only by 61%. Migrants’ remittances, measured as a share of provincial value added, show a great degree of variabil- ity across Italian provinces (figure 4, panel c). Larger remittance outflows reflect primarily the presence of larger migrant communities, although the North-South divide in this case is slightly less marked.16 Between 2005 and 2010, remittance outflows in the median province on average equaled 0.2% of pro- vincial GDP, ranging from less than 0.1% of GDP in ten Italian provinces (mainly located in the South) 16 The correlation across Italian provinces between the share of remittances in value added and the share of migrants in the total population equals 0.28. The World Bank Economic Review 13 Figure 4. Remittances and Migrants across Italian Provinces Sources: Bank of Italy and ISTAT. See table 1 for definitions. Data on the stock of foreign residents refer to the end of the year. Data on provincial value added are available only until 2010. to 0.3% of GDP in Genoa, 0.4% in Milan, 0.6% in Florence, 0.9% in Rome, and 3.9% (the maximum value) in Prato, all provinces with a very large share of foreign residents. In sum, the richness of the dataset in terms of destination countries and the heterogeneity across source provinces make it possible to identify how remittances react to economic shocks in destination countries. In addition, the variability in the localization patterns of migrants across provinces allows us to assess if the response of remittances is affected by the structure of migrant community in the province. 4. Results The Counter-Cyclicality of Remittances In estimating equation 1, we start with a parsimonious specification, controlling for the trend and cyclical components of GDP, country size, and migrant stock (table 3, column 1). We then estimate the baseline model (column 2). We then add the three variables capturing exogenous shocks, first one at the time and then jointly (columns 3 through 6). For comparability purposes, these results are based on a common sample driven by the specification with the largest set of controls (column 6). Finally, we show results for the pre- ferred specification, as reported in column 3 but based on the largest available sample (column 7). 14 Bettin, Presbitero, and Spatafora Table 3. Baseline and Extended Specification (1) (2) (3) (4) (5) (6) (7) MIGRANTS i;j;t 0.775*** 0.702*** 0.699*** 0.702*** 0.702*** 0.699*** 0.527*** [0.162] [0.111] [0.104] [0.111] [0.112] [0.104] [0.105] POP j;t 3.711*** 0.169 0.435 0.171 0.166 0.442 À0.243 [1.281] [0.827] [0.841] [0.827] [0.818] [0.835] [0.612] CYCLE j;t À3.065*** À2.068*** À1.799*** À2.051*** À2.071*** À1.817*** À1.438*** [0.426] [0.455] [0.489] [0.505] [0.460] [0.536] [0.406] TREND j;t 1.603*** 1.499*** 1.096** 1.496*** 1.498*** 1.099** 1.337*** [0.513] [0.466] [0.497] [0.465] [0.464] [0.497] [0.358] FINDEV j;t À0.685*** À0.672*** À0.686*** À0.686*** À0.668*** À0.686*** [0.134] [0.126] [0.137] [0.134] [0.128] [0.111] AID j;t À0.103*** À0.090*** À0.103*** À0.103*** À0.090*** À0.089*** [0.013] [0.011] [0.013] [0.013] [0.011] [0.014] TT j;t À0.486*** À0.490*** À0.325** [0.180] [0.179] [0.134] DISASTERS j;t À0.004 0.007 [0.020] [0.020] WAR j;t 0.003 À0.006 [0.040] [0.043] Observations 19,130 19,130 19,130 19,130 19,130 19,130 23,419 Number of pair 3,467 3,467 3,467 3,467 3,467 3,467 3,750 Notes: The table reports regression coefficients and (in brackets) the associated robust standard errors clustered by country-province pairs. *significant at 10%; **significant at 5%; ***significant at 1%. Estimations are carried out by using the Poisson Fixed Effects estimator. The dependent variable is the value of total official remittances at constant prices from province i to country j in year t (REMi;j;t ). A constant and a set of province-country pairs (i; j) and year (t) dummies are included. Sources: Authors’ analysis based on data described in the text and table 1. The results consistently show that remittances increase in response to cyclical output declines in the recipient country, CYCLEj . The reaction of remittances to growth slowdowns is statistically significant and economically relevant across all specifications. The elasticity equals À3 in the most parsimonious specification. It declines (in absolute value) after including the other country-specific controls; in particu- lar, the elasticity equals À1.8 in the specification that includes an indicator for a negative terms-of-trade shock. The elasticity is somewhat smaller in the large sample (column 7) but still statistically significant and well above unity (in absolute value). The absolute (as opposed to proportional) impact on remittan- ces of the recipient country’s economic cycle depends on the initial level of remittances, which is in turn a function of the migrant stock residing in a specific province. On average over the sample period, the impact on remittances of a 1 percent cyclical decline in the recipient country’s GDP ranges from 13 mil- lion US $ for remittances to the Philippines to 263 US $ for remittances to Dominica. Even controlling for cyclical variations in output per capita, remittances increase significantly in response to a terms-of-trade deterioration, which represents one of the major factors of vulnerability in recipient countries. This result is consistent with a particularly altruistic response to major and/or clearly exogenous shocks. Remittances are approximately 0.5 percent larger when recipient countries experience a one percent decrease in their terms of trade (column 3), reinforcing the response of remittances to cyclical output down- turns. Again, this elasticity shrinks slightly when the model is estimated using the large sample (column 7). We do not find strong evidence of a positive response of remittances to natural disasters, in line with Lueth and Ruiz-Arranz (2008) but in contrast to other studies (Yang 2008; Mohapatra, Joseph, and Ratha 2012; Ebeke and Combes 2013). Again, remittances are not significantly affected by the outbreak of armed conflicts, in line with Naude ´ and Bezuidenhout (2012).17 When controlling jointly for these 17 These effects remain insignificant when disasters are expressed as log (1þannual frequency), as in Beine and Parsons (2015) and when armed conflicts are expressed in terms of their annual frequency. The World Bank Economic Review 15 three measures of exogenous shocks, remittances are again negatively associated with the terms of trade, but not with armed conflicts or natural disasters (column 6). The response of remittances to cyclical fluctuations is stronger in migrant communities that on aver- age have settled more recently, as indicated by the negative and highly significant coefficient on the inter- action between the cycle and the (two-year) growth rate of the bilateral stock of migrants, DMIGRANTSi;j;ðt;tÀ2Þ (table 4, columns 2 and 5). Likewise, the response of remittances to terms-of-trade shocks is stronger in more recently established migrant communities, as indicated by the negative coeffi- cient on the interaction between the terms of trade and the growth rate of the migrant stock (table 4, col- umns 4 and 5). All this is consistent with the notion that the counter-cyclical behavior of remittances can be attributed to altruistic feelings and that the strength of such feelings diminishes over time. Moreover, these interaction effects are quantitatively large. The elasticity of bilateral remittances with respect to cyclical GDP equals À1.1 where the migrant community is relatively old (DMIGRANTSi;j;½t:tÀ2Š ¼ 0) but more than doubles to À2.5 where the migrant community is relatively young (DMIGRANTSi;j;½t;tÀ2Š ¼ 1, that is, the migrant stock doubled in size in the previous two years). Similarly, a one percent terms-of-trade deterioration is associated with a 0.18 percent increase in remit- tances from relatively old migrant communities but a 0.4 percent increase from relatively young migrant communities. For the median value of DMIGRANTSi;j;½t;tÀ2Š (0.18), the elasticity with respect to cyclical GDP equals À1.3, and a one percent terms-of-trade deterioration is associated with a 0.2 percent increase in remittances. Robustness checks broadly confirm these results. In particular, when the growth rate of the migrant stock is computed over a three- or four-year period,18 remittances from more recently established com- munities remain significantly more responsive to cyclical fluctuations in recipient output, although the interaction with the terms of trade loses statistical significance (table 4, columns 6 and 7). The same applies when the growth rate of the migrant stock is lagged by two years (table 4, column 8). Overall, these findings suggest that remittances can indeed play a significant role in stabilizing output during downturns, smoothing consumption, and mitigating the effects of macroeconomic fluctuations in developing countries. Other Results The elasticity of remittances with respect to trend GDP per capita in recipient countries ranges between 1.1 and 1.6 across different model specifications. The positive association between remittances and trend GDP, a proxy for expected GDP, supports the hypothesis that remittances are driven not only by altru- istic motives but also by investment motives.19 Remittances are negatively correlated with financial development in recipient countries: on average, a 1 percent reduction in the ratio of domestic credit to the private sector over GDP translates into a 0.7 percent increase in migrants’ transfers. This suggests that remittances act as a substitute for financial development, helping to overcome the financing constraints of households living in countries with less efficient financial institutions (Giuliano and Ruiz-Arranz 2009). Remittances and foreign aid are also substitutes, in line with Amuedo-Dorantes, Pozo, and Vargas-Silva (2007), suggesting that they do not reinforce each other in mitigating business-cycle fluctuations in recipient countries. The effect is statisti- cally significant across all specifications, even if the magnitude is relatively modest: on average, a 1 per- cent increase in aid per capita is associated with a 0.1 percent reduction in remittances. Bilateral remittances are strongly correlated with the size of the migrant community in the source province. This 18 Data availability does not allow us to calculate the change in migrant stocks over longer time horizons. Using a four- year horizon already implies a significant reduction in sample size compared to the baseline regressions (see table 3). 19 Evidence that remittances might be profit-driven is also provided in Lueth and Ruiz-Arranz (2008), where remittances are shown to be positively correlated with the recipient country’s growth rate of GDP per capita. Table 4. Interactions 16 (1) (2) (3) (4) (5) (6) (7) (8) MIGRANTS i;j;t 0.398*** 0.419*** 0.392*** 0.394*** 0.418*** 0.410*** 0.350*** 0.341*** [0.136] [0.139] [0.134] [0.136] [0.139] [0.110] [0.081] [0.083] DMIGRANTSi;j;ðt;tÀ2Þ À0.065 À0.074 À0.06 0.932 1.116** [0.061] [0.060] [0.060] [0.571] [0.555] DMIGRANTSi;j;ðt;tÀ3Þ 0.558 [0.370] DMIGRANTSi;j;ðt;tÀ4Þ À0.216 [0.248] DMIGRANTSi;j;ðtÀ2;tÀ4Þ 0.004 [0.309] POP j;t À0.902* À0.883* À0.69 À0.698 À0.669 À0.421 0.221 À0.238 [0.518] [0.521] [0.518] [0.510] [0.511] [0.511] [0.517] [0.498] CYCLE j;t À1.544*** À1.107*** À1.375*** À1.394*** À0.895*** À0.636** À0.568* À0.890*** [0.241] [0.289] [0.251] [0.252] [0.299] [0.297] [0.291] [0.288] TREND j;t 1.688*** 1.668*** 1.552*** 1.535*** 1.502*** 1.599*** 1.826*** 1.722*** [0.229] [0.227] [0.227] [0.230] [0.228] [0.214] [0.223] [0.224] AID j;t À0.059*** À0.059*** À0.055*** À0.054*** À0.054*** À0.056*** À0.057*** À0.055*** [0.013] [0.013] [0.014] [0.013] [0.013] [0.013] [0.013] [0.013] FINDEV j;t À0.460*** À0.457*** À0.475*** À0.476*** À0.474*** À0.486*** À0.496*** À0.489*** [0.073] [0.073] [0.073] [0.073] [0.072] [0.071] [0.072] [0.073] TT j;t À0.219*** À0.177*** À0.178*** À0.197*** À0.264*** À0.238*** [0.055] [0.062] [0.061] [0.062] [0.065] [0.059] CYCLE j;t *DMIGRANTSi;j;ðt;tÀ2Þ À1.373** À1.556** [0.588] [0.621] CYCLE j;t *DMIGRANTSi;j;ðt;tÀ3Þ À1.252*** [0.297] CYCLE j;t *DMIGRANTSi;j;ðt;tÀ4Þ À0.500*** [0.104] CYCLE j;t *DMIGRANTSi;j;ðtÀ2;tÀ4Þ À0.501** [0.197] TT j;t *DMIGRANTSi;j;ðt;tÀ2Þ À0.214* À0.256** [0.125] [0.120] TT j;t *DMIGRANTSi;j;ðt;tÀ3Þ À0.131 [0.080] TT j;t *DMIGRANTSi;j;ðt;tÀ4Þ 0.044 Bettin, Presbitero, and Spatafora Table 4. (continued) (1) (2) (3) (4) (5) (6) (7) (8) [0.053] TT j;t *DMIGRANTSi;j;ðtÀ2;tÀ4Þ À0.000 [0.066] Observations 16,670 16,670 16,670 16,670 16,670 16,670 16,670 16,670 Number of pair 3,621 3,621 3,621 3,621 3,621 3,621 3,621 3,621 Notes: The table reports regression coefficients and (in brackets) the associated robust standard errors clustered by country-province pairs. *significant at 10% level; **significant at 5% level; ***significant at 1% level. Estimations are carried out using the Poisson Fixed Effects estimator. The dependent variable is the value of total official remittances at constant prices from province i to country j in year t (REMi;j;t ). A constant and a set of province-country pairs The World Bank Economic Review (i; j) and year (t) indicators are included. Sources: Authors’ analysis based on data described in the text and table 1. 17 18 Bettin, Presbitero, and Spatafora Table 5. Robustness: Sample Definition (1) (2) (3) (4) Low-income Middle-income Large migrant No large countries countries communities recipients MIGRANTS i;j;t 0.740*** 0.610*** 0.879*** 0.735*** [0.149] [0.115] [0.153] [0.130] POP j;t À18.750*** 0.421 0.455 À0.437 [2.238] [0.922] [0.969] [0.887] CYCLE j;t À5.720*** À2.287*** À2.249*** À1.889*** [1.401] [0.457] [0.472] [0.454] TREND j;t À2.684*** 1.018** 1.489*** 1.918*** [0.753] [0.487] [0.522] [0.575] AID j;t À0.234*** À0.099*** À0.106*** À0.099*** [0.056] [0.015] [0.012] [0.015] FINDEV j;t À0.008 À0.678*** À0.714*** À0.799*** [0.200] [0.145] [0.137] [0.151] Observations 2,554 16,576 9,266 17,112 Number of pair 503 2,964 1,684 3,153 Notes: The table reports regression coefficients and (in brackets) the associated robust standard errors clustered by country-province pairs. *significant at 10% level; **significant at 5% level; ***significant at 1% level. Columns 1 and 2 report estimates based on, respectively, low- and middle-income countries, as defined by the World Bank. Column 3 excludes all country-province-year observations for which the stock of migrants is less than one hundred individuals. Column 4 excludes large worldwide remittances recipients (China, India, Indonesia, Brazil, Nigeria, Pakistan, Bangladesh). Estimations are carried out using the Poisson Fixed Effects estimator. The dependent variable is the value of total official remittances at constant prices from province i to country j in year t (REMi;j;t ). A constant and a set of province-country pairs (i; j) and year (t) indicators are included. Sources: Authors’ analysis based on data described in the text and table 1. finding is consistent with a large body of literature (Lueth and Ruiz-Arranz 2008; Frankel 2011; Docquier, Rapoport, and Salomone 2012). The elasticity is generally stable at around 0.7 across alterna- tive specifications, similar to the value of 0.8 estimated by Docquier, Rapoport, and Salomone (2012) using a Poisson regression. The populations of the origin country, POPj , is positively associated with remittance flows. This may reflect either lower costs of sending remittances to larger countries, or simply greater investment opportunities. However, the coefficient is statistically significant only in the most par- simonious specification (column 1) and is less precisely estimated once we control for aid flows and the level of financial development. Robustness Checks This section tests the robustness of the findings. We first investigate the impact of changes in the sample composition (table 5). We then allow for additional covariates (table 6). Finally, we employ a different estimation method (table 7). Different samples We test the validity of the finding about the cyclical response of remittances across a number of alternative subsamples. First, we split the sample between low-income and middle- income countries (table 5, column 1 and 2, respectively) to analyze whether the countercyclical behavior of remittances depends on the recipient country’s income level. The coefficient on CYCLEj remains neg- ative in both cases, although the countercyclical effect is much stronger in low-income countries (a1 ¼ À5:7) than in middle-income countries (a1 ¼ À2:3). In low-income countries, the coefficient on trend GDP is negative, suggesting a more limited investment motive for remittances. In these countries, the negative correlation between remittances and aid is also stronger (perhaps because foreign aid is larger and more variable, so that its effects can be more precisely estimated), suggesting that remittances could help offset limited aid flows. The World Bank Economic Review 19 Table 6. Robustness: Additional Covariates (1) (2) (3) (4) MIGRANTS i;j;t 0.528*** 0.542*** 0.396*** 0.401*** [0.139] [0.141] [0.092] [0.092] POP j;t À1.603 À1.413 À7.723*** À8.604*** [1.247] [1.096] [1.371] [1.568] CYCLE j;t À3.194*** À2.432*** À1.382*** À2.257*** [0.534] [0.558] [0.411] [0.852] TREND j;t 2.092*** 2.076*** 2.168*** 2.246*** [0.559] [0.567] [0.540] [0.522] FINDEV j;t À0.837*** À0.840*** À0.664*** À0.691*** [0.130] [0.127] [0.161] [0.154] AID j;t À0.092*** À0.096*** À0.135*** À0.139*** [0.013] [0.012] [0.044] [0.044] FISCAL BALANCE j;t 1.856*** 1.374* [0.525] [0.800] EXTERNAL DEBT j;t 0.014 À0.19 [0.199] [0.165] EXECUTIVE CONST j;t 0.234*** 0.237*** [0.032] [0.033] Observations 16,129 16,079 8,463 8,463 Number of pair 2,772 2,760 1,389 1,389 Notes: The table reports regression coefficients and (in brackets) the associated robust standard errors clustered by country-province pairs. *significant at 10% level; **significant at 5% level; ***significant at 1% level. Estimations are carried out using the Poisson Fixed Effects estimator. The dependent variable is the value of total official remittances at constant prices from province i to country j in year t (REMi;j;t ). A constant and a set of province-country pairs (i; j) and year (t) indica- tors are included. Sources: Authors’ analysis based on data described in the text and table 1. Next, we drop observations that may add noise and lead to small-sample bias. In particular, we are concerned about province-country pairs that are characterized by a limited number of resident migrants. Here, remittances may be driven by idiosyncratic factors, which could be largely unrelated to macroeco- nomic conditions in the recipient country as a whole. To avoid this possibility, we exclude all observa- tions where the migrant community numbers less than one hundred migrants (MIGRijt < 100), rather than using the threshold as in the baseline model. Although this threshold implies dropping fifteen coun- tries (all provinces are still included in the sample) and reduces the original sample by about one-half, the results from our baseline model remain valid (column 3).20 A related concern is that, in large recipi- ent countries, macroeconomic conditions could be highly heterogeneous within the country. Further, migrant remittances may be largely driven by conditions within some region of the country, rather than in the recipient country as a whole. Hence, we drop from the sample the recipient countries with the largest population (Bangladesh, Brazil, China, India, Indonesia, Nigeria, and Pakistan).21 Again, the ear- lier findings are confirmed, even controlling for other explanatory variables (column 4). Additional covariates We next augment the baseline model with additional regressors and find a positive impact on remittances of macroeconomic stability and institutional quality in the recipient coun- try (table 6). In particular, remittances are larger to countries with higher fiscal balances (as a ratio of GDP) or with stronger institutions (as measured by the constraints on the executive; cf. Singh, Haacker, Lee, and Le Goff 2011). In contrast, external debt does not have a significant impact. The negative effect on remittances of the cyclical component of output in recipient countries holds even when controlling jointly for these covariates (column 4). 20 Results are robust to alternative specifications of the threshold up to MIGRijt < 500. 21 We drop countries with a total population above the 95th percentile of the sample distribution. 20 Bettin, Presbitero, and Spatafora Table 7. Robustness: PPML Estimator—Baseline and Extended Specification (1) (2) (3) (4) (5) (6) (7) MIGRANTS i;j;t 0.948*** 0.948*** 0.948*** 0.948*** 0.948*** 0.948*** 0.928*** [0.039] [0.038] [0.038] [0.038] [0.038] [0.038] [0.035] POP j;t 3.869*** 0.399 0.65 0.407 0.383 0.646 0.359 [1.149] [0.796] [0.813] [0.796] [0.787] [0.806] [0.622] DISTANCE i;j 0.443*** 0.445*** 0.444*** 0.445*** 0.445*** 0.444*** 0.403*** [0.160] [0.159] [0.159] [0.159] [0.160] [0.159] [0.148] CYCLE j;t À2.981*** À1.945*** À1.694*** À1.899*** À1.957*** À1.690*** À1.440*** [0.485] [0.471] [0.498] [0.515] [0.475] [0.540] [0.409] TREND j;t 1.595*** 1.466*** 1.093** 1.457*** 1.464*** 1.092** 1.294*** [0.499] [0.461] [0.486] [0.460] [0.459] [0.485] [0.357] FINDEV j;t À0.686*** À0.673*** À0.689*** À0.690*** À0.675*** À0.669*** [0.131] [0.122] [0.133] [0.131] [0.124] [0.105] AID j;t À0.107*** À0.093*** À0.106*** À0.106*** À0.093*** À0.092*** [0.012] [0.011] [0.012] [0.013] [0.011] [0.012] TT j;t À0.454*** À0.452*** À0.291** [0.173] [0.171] [0.125] DISASTERS j;t À0.011 À0.002 [0.021] [0.020] WAR j;t 0.011 0.003 [0.042] [0.045] Observations 19,130 19,130 19,130 19,130 19,130 19,130 23,419 R-squared 0.901 0.914 0.915 0.914 0.914 0.915 0.897 Notes: The table reports regression coefficients and (in brackets) the associated robust standard errors clustered by country-province pairs. *significant at 10% level; **significant at 5% level; ***significant at 1% level. Estimations are carried out using the Poisson Pseudo Maximum Likelihood (PPML) estimator (Silva and Tenreyro 2006). The dependent variable is the value of total official remittances at constant prices from province i to country j in year t (REMi;j;t ). A constant and a set of province (i), country (j), and year (t) indicators are included. Sources: Authors’ analysis based on data described in the text and table 1. Different fixed effects The last robustness exercise relates to the fixed effects in our estimation. Here, we drop country-province pair fixed effects and control separately for country and province fixed effects. Our main results from the baseline and the augmented specification, and in particular the significant negative coefficient on the cyclical component of recipient GDP, are largely confirmed (tables 7). The magnitude of the elasticities is also quite similar, lending support to the validity of our main findings. The distance to migrants’ home country, DISTANCEi;j , is positively correlated with remittances. A priori, we would instead expect distance to be positively correlated with remittance transfer costs and therefore negatively correlated with remittances.22 The result may arise because remittance data only takes into account official transactions. Migrants from nearby regions, such as Eastern Europe or the Mediterranean, may send remittances informally, for instance bringing them in person when they travel back home. In contrast, migrants from distant countries are relatively more likely to use formal, if expen- sive, remittance channels. 5. Conclusions We analyze how remittances are affected by structural characteristics, macroeconomic conditions, and adverse shocks in both source and recipient economies using a novel, rich panel dataset on bilateral 22 Lueth and Ruiz-Arranz (2008), Frankel (2011), and Docquier, Rapoport, and Salomone (2012) indeed find a negative, significant correlation between remittances and bilateral distance. The World Bank Economic Review 21 remittances from 103 Italian provinces to seventy-nine developing countries over the period 2005–2011. Remittances are negatively correlated with the business cycle in recipient countries and increase espe- cially strongly in response to adverse exogenous shocks, such as large declines in the terms of trade. The counter-cyclical behavior of remittances is stronger in provinces where the migrants’ community has a larger share of newly arrived migrants. These results are consistent with international remittances being driven at least partly by altruism motives. We also find that remittances are positively correlated with potential GDP in recipient countries, that remittances and foreign aid are substitutes, and that financial development in the recipient country is negatively associated with remittance transfers, suggesting that remittances help alleviate credit con- straints. All these results hold even controlling for unobserved country-province pairs fixed effects, which capture time-invariant institutional and geographical factors, which may drive remittance flows. We conclude that remittances may indeed contribute significantly to macroeconomic stability in recipient countries. This effect should be considered together with their positive impact on poverty alle- viation and growth, emphasized in the existing literature. From a policy perspective, our results suggest that reducing barriers to remittances to developing countries, by reducing these countries’ reliance on more volatile capital inflows, may also increase their resilience to exogenous shocks. One way to achieve this is to reduce barriers to international migration (Clemens 2011). In addition, it remains important to promote migrants’ access to financial services in host countries. Overall, these findings corroborate the efforts being carried out by international financial institutions and the private sector to reduce the costs of migrant remittances and to foster freer remittance flows. References Abdih, Y., R. Chami, J. Dagher, and P. Montiel. 2012. “Remittances and Institutions: Are Remittances a Curse?” World Development 40 (4): 657–66. Adams, R. H., Jr. 2009. “The Determinants of International Remittances in Developing Countries.” World Development 37 (1): 93–103. Adams, R. H., Jr, and A. Cuecuecha. 2010. “Remittances, Household Expenditure and Investment in Guatemala.” World Development 38 (11): 1626–41. Agarwal, R., and A. W. Horowitz. 2002. “Are International Remittances Altruism or Insurance? Evidence from Guyana Using Multiple-Migrant Households.” World Development 30 (11): 2033–44. Akkoyunlu, S., and K. A. Kholodilin. 2008. “A Link between Workers’ Remittances and Business Cycles in Germany and Turkey.” Emerging Markets Finance and Trade 44 (5): 23–40. Amuedo-Dorantes, C., and S. Pozo. 2006. “The Time Pattern of Remittances: Evidence from Mexican Migrants.” Well-being and Social Policy 2 (2): 49–66. Amuedo-Dorantes, C., S. Pozo, and C. Vargas-Silva. 2007. “Remittances and the Macroeconomy: The Case of Small Island Developing States.” Working Paper Series UNU-WIDER Research Paper, World Institute for Development Economic Research (UNU-WIDER). Arezki, R., and M. Bru¨ ckner. 2012. “Rainfall, Financial Development, and Remittances: Evidence from Sub-Saharan Africa.” Journal of International Economics 87 (2): 377–85. Barajas, A., R. Chami, C. Ebeke, and S. J.-A Tapsoba. 2012. “Workers’ Remittances: An Overlooked Channel of International Business Cycle Transmission?” IMF Working Papers 12/251, International Monetary Fund. Barajas, A., R. Chami, C. Fullenkam, M. Gapen, and P. Montiel. 2009. “Do Workers’ Remittances Promote Economic Growth?” IMF Working Papers 09/153, International Monetary Fund. Barajas, A., R. Chami, C. Fullenkamp, and A. Garg. 2010. “The Global Financial Crisis and Workers’ Remittances to Africa: What’s the Damage?” Journal of African Development 12 (1): 73–96. Bazzi, S., and M. Clemens. 2013. “Blunt Instruments: Avoiding Common Pitfalls in Identifying the Causes of Economic Growth.” American Economic Journal: Macroeconomics 5 (2): 152–86. Beine, M., and C. Parsons. 2015. “Climatic Factors as Determinants of International Migration.” The Scandinavian Journal of Economics 117 (2): 723–67. 22 Bettin, Presbitero, and Spatafora Bettin, G., R. Lucchetti, and A. Zazzaro. 2012. “Endogeneity and Sample Selection in a Model for Remittances.” Journal of Development Economics 99 (2): 370–84. Bettin, G., A. F. Presbitero, and N. Spatafora. 2014. “Remittances and Vulnerability in Developing Countries.” Policy Research Working Paper 6812, The World Bank. Bettin, G., and A. Zazzaro. 2012. “Remittances and Financial Development: Substitutes or Complements in Economic Growth?” Bulletin of Economic Research 64 (4): 509–36. Bugamelli, M., and F. Paterno . 2011. “Output Growth Volatility and Remittances.” Economica 78 (311): 480–500. Burger, M., F. van Oort, and G.-J. Linders. 2009. “On the Specification of the Gravity Model of Trade: Zeros, Excess Zeros and Zero–Inflated Estimation.” Spatial Economic Analysis 4 (2): 167–90. Cameron, A. C., and P. K. Trivedi. 2005. Microeconometrics. No. 9780521848053 in Cambridge Books, Cambridge University Press. Chami, R., D. S. Hakura, and P. J. Montiel. 2012. “Do Worker Remittances Reduce Output Volatility in Developing Countries?” Journal of Globalization and Development 3 (1): 1–25. Chami, R., S. Jahjah, and C. Fullenkamp. 2005. “Are Immigrant Remittance Flows a Source of Capital for Development?” IMF Staff Papers 1 (52): 55–81. Clemens, M. A. 2011. “Economics and Emigration: Trillion-Dollar Bills on the Sidewalk?” Journal of Economic Perspectives 25 (3): 83–106. Combes, J.-L., C. Ebeke, M. N. Etoundi, and T. Yogo. 2014. “Are Foreign Aid and Remittance Inflows a Hedge against Food Price Shocks in Developing Countries?” World Development 54 (1): 81–98. Cooray, A., and D. Mallick. 2013. “International Business Cycles and Remittance Flows.” The B.E. Journal of Macroeconomics 13 (1): 1–33. de la Briere, B., E. Sadoulet, A. de Janvry, and S. Lambert. 2002. “The Roles of Destination, Gender, and Household Composition in Explaining Remittances: An Analysis for the Dominican Sierra.” Journal of Development Economics 68, 309–28. Docquier, F., H. Rapoport, and S. Salomone. 2012. “Remittances, Migrants’ Education and Immigration Policy: Theory and Evidence from Bilateral Data.” Regional Science and Urban Economics 42 (5): 817–28. Durdu, C. B., and S. Saya. 2010. “Emerging Market Business Cycles with Remittance Fluctuations.” IMF Staff Papers 57 (2): 303–25. Ebeke, C., and J.-L. Combes. 2013. “Do Remittances Dampen the Effect of Natural Disasters on Output Growth Volatility in Developing Countries?” Applied Economics 45, 2241–54. Echazarra, A. 2011. “Accounting for the Time Pattern of Remittances in the Spanish context.” Working Paper 5- 2010, Centre for Census and Survey Research, University of Manchester. El-Sakka, M. I. T., and R. McNabb. 1999. “The Macroeconomic Determinants of Emigrant Remittances.” World Development 27 (8): 1493–502. Faini, R. 1994. “Workers Remittances and the Real Exchange Rate: A Quantitative Framework.” Journal of Population Economics 7 (2): 235–45. ———. 2007. “Remittances and the Brain Drain: Do more Skilled Migrants Remit More?” World Bank Economic Review 21 (2): 177–91. Frankel, J. 2011. “Are Bilateral Remittances Countercyclical?” Open Economies Review 22 (1): 1–16. Freund, C., and N. Spatafora. 2008. “Remittances, Transaction Costs, and Informality.” Journal of Development Economics 86 (2): 356–66. Giuliano, P., and M. Ruiz-Arranz. 2009. “Remittances, Financial Development, and Growth.” Journal of Development Economics 90 (1): 144–52. Hoddinott, J. 1994. “A Model of Migration and Remittances Applied to Western Kenya.” Oxford Economic Papers 46, 459–476. Jimenez-Martin, S., N. Jorgensen, and J. Labeaga. 2007. The Volume and Geography of Remittances from the EU. Tech. Rep., Report to the European Commission. Kpodar, K., and M. Le Goff. 2012. “Do Remittances Reduce Aid Dependency?” Working Papers 34, FERDI. Lucas, R. E., and O. Stark. 1985. “Motivations to Remit: Evidence from Botswana.” Journal of Political Economy 93 (5): 901–18. Lueth, E., and M. Ruiz-Arranz. 2008. “Determinants of Bilateral Remittance Flows.” The B.E. Journal of Macroeconomics 8 (1 (Topics)), Article 26. The World Bank Economic Review 23 Makina, D., and A. Masenge. 2015. “The Time Pattern of Remittances and the Decay Hypothesis: Evidence from Migrants in South Africa.” Migration Letters 12 (1). Milner, H. V., and D. H. Tingley. 2010. “The Political Economy of US Foreign Aid: American Legislators and the Domestic Politics of Aid.” Economics & Politics 22 (2): 200–32. Mohapatra, S., G. Joseph, and D. Ratha. 2012. “Remittances and Natural Disasters: Ex-post Response and Contribution to Ex-ante Preparedness.” Environment, Development and Sustainability 14 (3): 365–87. Naude ´ , W., and H. Bezuidenhout. 2012. “Remittances Provide Resilience against Disasters in Africa.” UNU-MERIT Working Paper Series 026, United Nations University, Maastricht Economic and social Research and training cen- tre on Innovation and Technology. Niimi, Y., O. Caglar, and M. Schiff. 2010. “Remittances and the Brain Drain: Skilled Migrants do Remit Less.” Annales d’Economie et de Statistique (97–98), 123–41. Osili, U. O. 2007. “Remittances and Savings from International Migration: Theory and Evidence Using a Matched Sample.” Journal of Development Economics 83 (2): 446–65. Poirine, B. 1997. “A Theory of Remittances as an Implicit Family Loan Arrangement.” World Development 25 (4): 589–611. Rapoport, H., and F. Docquier. 2006. “The Economics of Migrants’ Remittances.” In S. Kolm, and J. Mercier Ythier, eds., Handbook on the Economics of Giving, Altruism and Reciprocity, vol. 2, Elsevier, 1135–98. Roodman, D. 2009. “A Note on the Theme of too Many Instruments.” Oxford Bulletin of Economics and Statistics 71 (1): 135–58. Ruiz, I., and C. Vargas-Silva. 2014. “Remittances and the Business Cycle: A Reliable Relationship?” Journal of Ethnic and Migration Studies 40 (3): 456–74. Salomone, S. 2009. “Anthology on Migrants Remittance and Development: Research Perspectives.” In R. J. Adams, H. de Haas, and U. Osili, eds., Remittances. Overview of the Existing Literature, New York: Social Science Research Council. Sayan, S. 2004. “Guest workers’ Remittances and Output Fluctuations in Host and Home Countries: The Case of Remittances from Turkish Workers in Germany.” Emerging Markets Finance and Trade 40 (6): 68–81. Sayan, S. 2006. “Business Cycles and Workers’ Remittances: How do Migrant Workers Respond to Cyclical Movements of GDP at Home?” IMF Working Papers 06/52, International Monetary Fund. Silva, J. M. C. S., and S. Tenreyro. 2006. “The Log of Gravity.” The Review of Economics and Statistics 88 (4): 641–58. Singh, R. J., M. Haacker, K.-w. Lee, and M. Le Goff. 2011. “Determinants and Macroeconomic Impact of Remittances in Sub-Saharan Africa.” Journal of African Economies 20 (2): 312–40. Stark, O. 2009. “Reasons for Remitting.” World Economics 10 (3): 147–58. Themne ´ r, L., and P. Wallensteen. 2013. “Armed conflict, 1946–2012.” Journal of Peace Research 50 (4). Vargas-Silva, C. 2008. “Are Remittances Manna from Heaven? A Look at the Business Cycle Properties of Remittances.” The North American Journal of Economics and Finance 19 (3): 290–303. Wooldridge, J. M. 2010. Econometric Analysis of Cross Section and Panel Data. The MIT Press, 2nd edn. World Bank. 2014. Migration and Development Brief. Tech. Rep. 23, World Bank, Washington, DC. ———. 2015. Global Economic Prospects, January 2015: Having Fiscal Space and Using It. World Bank, Washington, DC. Yang, D. 2008. “Coping with Disaster: The Impact of Hurricanes on International Financial Flows, 1970–2002.” The B.E. Journal of Economic Analysis & Policy 8 (1 (Advances)), Article 13. Yang, D., and H. Choi. 2007. “Are Remittances Insurance? Evidence from Rainfall Shocks in the Philippines.” The World Bank Economic Review 21 (2): 219–48.