Remittances, Institutions, and Economic Development Natalia Catrinescu (European Commission) Miguel Leon-Ledesma (University of Kent) Matloob Piracha (University of Kent) Bryce Quillin (World Bank). THE IS A DRAFT, PLEASE DO NOT CITE ABSTRACT There is considerable debate regarding the relative contribution of international migrants remittances to sustainable economic development. While the rates and levels of officially recorded remittances to developing countries has increased enormously over the last decade, academic and policy-oriented research has not come to a consensus over whether remittances contribute to longer-term growth by building human and financial capital or degrade growth by creating moral hazard and exerting a Dutch disease style impact on patterns of investment. This paper suggests that contradictory findings have emerged when looking at the remittances-growth link because of an omitted variable bias: specifically, remittances will be more likely to contribute to longer-term growth when the receiving countries political and economic policies and institutions create the incentives for financial and business investment and savings from remittances. Policies must favor savings and investment so that, at the margin, household income that exceeds the needs of basic subsistence can be saved or invested (including in human capital). I. INTRODUCTION Remittances resulting from international migrants transfers are becoming an increasingly important source of external finance for developing economies. For many low-income, net-emigration countries, remittances are the most important source of external financing. Theoretical and empirical investigations into remittances economic impact have produced highly mixed results. On the one hand, remittances can contribute to the alleviation of poverty and, in some instances, provide capital to fund households investments and savings. For a number of countries, international remittances have driven macroeconomic growth, mostly by increasing national disposable income. Yet, remittances can have a deleterious impact on national economic growth in the medium term. Remittances can fuel inflation, disadvantage the tradable sector, and reduce labor market participation rates as receiving households opt to live off of migrants transfers rather than by working. At the policy level, remittances contribution to growth might reduce the incentives for implementing sound macroeconomic policy or make needed structural reforms.
The key to increasing the longer-term development impact of remittances is to implement economic and governance policies that support a sound business investment environment, provide the prudential security of the financial sector as well as quality public services (e.g. education and health care). Policies must favor savings and investment so that, at the margin, household income that exceeds the needs of basic subsistence can be saved or invested (including investment in human capital). The policies that are required for income convergence with industrialized countries more generally are also the ones that increase the development impact of remittances. II. LITERATURE REVIEW 1. Effect of Remittances on Economic Growth The economic consequences of remittances are hard to disentangle they can affect growth through a variety of channels. Lucas (2005) divides the discussion of remittances into two aspects: the effects on poverty and inequality; and the influences upon investment, growth, and macro-economic stability. One could also add to this categorisation a third area of research, which looks at how remittances and, more importantly, the underlying migration affect human capital. When it comes to the overall impact of remittances on income inequality, Ratha (2003) finds that evidence is mixed. Some find that remittances sharpen inequality (Stark et al., 1986; Adams, 1991, 1998), while others claim that in the long run, the income distribution becomes more equal through the liquidity provided for capital accumulation, or through trickle down effects in the labor market (Taylor and Wyatt, 1996; Stark et al. (1986). The effect on poverty seems much less controversial, as remittances per se do not lower anyone s income. The effect of remittances on human capital accumulation as a factor of economic growth is also uncertain. Here the relationship between remittances and growth is very much tied to the nature of the underlying migration. Thus, the positive or negative effect on 1
human capital accumulation depends on skill distribution of the migrants, the country s size and geographical proximity to developed countries, and on the probability of migrant s return to the home country. Faini (2002) finds that the existent bias towards skilled migration exacerbates the brain drain in developing countries and restricts their ability to rely on unskilled migration as an engine for growth and convergence. Recently Deine et al. (2001) and Stark (2003) showed that the mere possibility to migrate increases the return to education, thereby fostering further investment in education and ultimately boosting growth. Faini (2002) posits that the welfare impact of the bran drain might be unwarranted if the negative effects of this phenomenon is offset by the positive impact of remittances, but in fact finds that skilled migrants have a lower propensity to remit. Lundborg and Segerstrom (1999, 2002) consider the effect of migration on technological change and show a positive outcome for economic development. In their North-South model in which growth is driven by improvements in product quality, migration has a net positive effect for capitalists, workers, and migrants in the sending country. The economic consequences that are most important for the purpose of this study, however, refer to the manner in which remittances affect savings within the framework of exogenous growth models, where the effect of an increase in the saving rate is to increase the level of per capita capital stock and therefore per capita output. There is a debate over the extent to which remittances actually boost the economy of the source country, since more of the income has been used for consumption purposes and not saved or invested (see Drinkwater et. al, 2002)). Recent strands of literature, however, indicate that remittances can lead to economic growth simply by increasing the migrant s household income, regardless of whether this additional income is spent on consumption or savings. For example, Ratha (2004) indicated that if remittances are invested, they contribute to output growth, and generate positive multiplier effect even if they are consumed. Lucas (forthcoming) confirms that the discussion on the extent to which remittances lead to additional investments has been confused. In his view, the issue of whether the cash 2
received is actually spent on investments is not the point. Rather, one should examine whether families with incomes enhanced by remittances save more, recognizing that spending on education, housing, and land are forms of investment and that what may be an investment by one family may or may not be an investment for the country the question arises as to how the recipient of payments generated by remittances spends the income. Taylor (1999) also finds that many of migration s most important impact may not be found in the households that send migrants abroad and receive remittances. High levels of consumption (as opposed to investment) spending by remittance receiving households may result in positive impact of remittances on productive investment in migrant sending areas, provided that this consumption demand triggers investment by other households or by firms. There is significant empirical evidence that remittances lead to positive economic growth, be it through increased consumption, savings or investment. Lucas (forthcoming) cites several case studies that show signs that remittances may indeed have served to accelerate investment in Morocoo, Pakistan, and India. Glytsos (2002) models the direct and indirect effects of remittances on incomes and hence on investment in seven Mediterranean countries, and finds that investment rises with remittances in six out of the seven countries. Additionally, the results of the analysis conducted by Leon-Ledesma and Piracha (2004) for eleven transition economies of Eastern Europe during 1990-1999 show support for the view that remittances have a positive impact on productivity and employment both directly and indirectly through their effect on investment. A recent study by Roberts and Banaian (2004) on remittances in Armenia conclude that overall, empirical evidence suggests that the propensity to save out of remittance income is high (almost 40%) and remarkably consistent across studies. 3
There is also evidence of important multiplier effects from remittance spending, particularly from housing construction. 1 The multiplier effect can be rather high - Massey and Durrand (1996) find that every migradollar that enters a local economy generates as much as $4.00 in demand of goods and services. Moreover, Desai et al. (2003) indicate that additional consumption increases indirect tax receipts thus also increasing government consumption or savings. Thus, there is overwhelming evidence that remittances have enabled economic growth through greater rates of investment as well as through the important multiplier effects, raising income levels in the economy beyond the households of the recipients of remittances. There are, nevertheless, at least two points of reservation regarding the effects of remittances. One is the possibility that countries can face a situation similar to the Dutch disease in which the inflow of remittances causes a real appreciation, or postpones depreciation, of the exchange rate, restricting export performance and hence possibly limiting output and employment. More importantly, research by Chami et al (2003) shows that income from remittances may be plagued by a moral hazard problem, permitting the migrant s family members to reduce their work effort. This hypothesis can influence significantly our understanding of how remittances affect economic growth. We, therefore, use their model as a benchmark and extend it by using a slightly different (and longer) data set, but more importantly by adding institutional variables to determine the real effect of remittances. 2. Effect of Institutions on Economic Growth A pioneering paper by North (1990), who asserted that institutions are the underlying determinants of the long-run performance of economies, spurred a great deal of academic interest and burgeoning literature on the extent and manner in which institutions affect economic growth. Important empirical work carried out by Acemoglu et al (2001), Barro (1991), Easterly and Levine (2001), Rodrik et al (2002) and others has 1 See Roberts and Banaian (2004), Lucas (forthcoming), Glytsos (1993), Adelman and Taylor (1990), Zarate (2002)). 4
fostered a widespread agreement among economists studying economic growth that institutional quality is not only associated with positive economic growth, but also that this relationship is causal. Much less understood are the implications of the conclusion that quality of institutions matters. Rodrik (2004) points out that institutional causality is endogenous to income levels, making it difficult to disentangle the web of causality between prosperity and institutions. Most researchers, nevertheless, find that good institutions lead to economic growth mainly by increasing the efficiency of investment. Analysis by Aron (2000), drawing on a wide sample of literature in the field, shows that institutional framework affects growth because it is integral to the amount spent on both the costs of transactions and the costs of transformation (in the production process). Transaction costs are higher when property rights or the rule of law are not reliable. Transformation costs, too, can be raised substantially because unenforceable contracts mean using inexpensive technology and operating less efficiently and competitively on a short-term horizon. Better performing institutions may improve growth by increasing the volume of investment - for example, by eliminating bureaucratic red tape and rent-seeking costs and by improving the efficiency of investment for example, by enforcing well-defined property rights. Aron also finds evidence, albeit weak, for a direct relationship between institutions and growth. A number of empirical studies support this line of reasoning. Mauro (1995) considers the effect of corruption, efficiency of the judicial system, and other institutional indices on economic growth and finds that corruption lowers investment, thereby lowering economic growth. He accounts for endogeneity problems by using ethnolinguistic fractionalization as instrument, and finds that bureaucratic efficiency actually causes high investment and growth. 5
Knack and Keefer (1997) provide evidence on the relevance of informal institutions by using indices on interpersonal trust and civic norms among local economic agents to find a robust positive relationship between social capital (as measured by these indices) and both growth and investment. They use the same instrument as Mauro (1995) to find a causal effect between these institutional indexes and investment. Grigorian and Martinez (2000) find that quality of institutions has a very strong positive effect on the rate of industrial growth, not only by increasing the amount of investments made available in the economy, but also by improving the efficiency of resource allocation. They also posit that the marginal effect of institutional improvements on industrial growth might be stronger in transition economies, where traditions of market oriented business institutions were virtually absent for decades. 3. Institutions, Remittances, and Economic Growth: the Interplay The two strands of literature considered above seem to indicate that quality of institutions might play an important role in determining the exact effect of remittances on economic growth through the influence institutions exert on the volume and efficiency of investment. Nevertheless, there is surprisingly very little literature that would come even close to analyzing the interplay between institutions, remittance and growth. There is limited empirical work that suggests what the role of certain institutional arrangements in determining the effect that remittances exert on growth might be. Faini (2002) regressed income growth in source countries on a standard set of explanatory variables and on remittances. Faini found a positive impact of remittances on growth and interpreted the positive coefficient on the policy stance to indicate that in order for the full impact of remittances, which allow households to accumulate productive assets, to be realized, a sound policy environment is needed one that does not foster macroeconomic uncertainty, does not penalize agricultural activities and supports the build-up of social and productive infrastructures. Ratha (2003) finds that during 1996-2000, remittance receipts averaged 0.5 percent of GDP in countries with a higher-than-median level of 6
corruption compared to 1.9 percent in countries with lower-than-median level of corruption, giving an indication that corruption has an effect on the level of income generated from remittances. This could, as discussed above, potentially turn into higher rates of savings and investment. In his policy-oriented analysis, Sorensen (2004) mentions institutional development as one crucial factor in enhancing the effect of remittances. She concludes that direct policies aimed at valorizing remittances can be effective only alongside a wider engagement by the wealthier donor countries in programs promoting institutional capacity building and improving the market and financial environment. Moreover, there seems to be a plausible hypothesis that quality of institutions might also play a role as a determinant of migration. Holzmann and Munz (2004) posit that differences in political stability, the human rights situation, and the general rule of law may also affect migration with these factors serving as a proxy for the level of individually perceived insecurity. A forthcoming World Bank Study 2 hypothesizes that broad, quality of life considerations drive migration. Preliminary findings indicate that although the decision to migrate for more productive and lucrative jobs is certainly related to the search for a higher quality life, wage and unemployment differentials alone do not explain as much migration as these broad quality of life concerns. Risk-averse agents and households may be less motivated to exploit spreads in earnings across countries if their day-to-day lifestyle is comfortable and stable. Spreads in this security may motivate those who would otherwise stay at home to search for a better and more secure life. All of the above analysis and deliberation seem to warrant at the very least, a closer look at the nexus between institutions, remittances and economic growth. Considering the serious implications of the study by Chami et al (2003), which posits a moral hazard problem connected to remittances and finds that remittances lead to negative economic growth by reducing the migrant s household s work effort, the particular model in that 2 World Bank (2005), Labor Study, Chapter 7: Migration in Europe and Central Asia (forthcoming) 7
paper will be extended to analyze the role of institutions in the relationship between remittances and growth. III. DATA DESCRIPTION The data on remittances was collected from the World Bank s World Development Indicators (WDI) database. The WDI data on remittances represents current transfers by migrants who are employed or intend to remain employed for more than a year in another economy in which they are considered residents. The data is reported by countries in their balance of payments (BoP). There is a widespread consensus in the literature that the quality of data on remittances is extremely poor. Similarly, the dataset used in this paper has several important limitations. First of all, the panel is significantly unbalanced there are many missing values in the data set. In fact, if we look at the last 34 years (1970-2003), only three countries Columbia, the Dominican Republic, and Italy have reported observations for every year (Table 1). Only 40 out of the 126 countries have 20 or more observations in this entire period. The situation is improved only slightly when employee compensation is added, but the dataset is expanded to 137 countries, of which 55 countries have 20 or more observations in the 34 year period (see Table 2). Data for Europe and Central Asia is even scarcer as for most countries in this region observations exist only since 1995. This situation, however, is to be expected. As mentioned above, migration is a relatively new phenomenon in the region, and its causes diversified and magnitude increased after the Asian financial crisis of 1997. Nonetheless, continuous observations for ECA countries are rare even for the last five years. Another major problem with this data is that it only includes official flows, i.e. flows transmitted through official banking channels. Unofficial flows still represent a large 8
(and for the most part unknown) share of the total remittance flows. In the context of ECA, studies indicate that unofficial flows might be as much as two times larger than officially-recorded flows 3. Nevertheless, in conducting any type of panel estimation with remittances data, one should also keep in mind the fact that better technology, decreased transfer transaction cost, and efforts to crack down on money laundering have generated a decrease in the unrecorded portion of remittance, which might create difficulties in determining whether a higher recorded amount represents remittance growth or improved reporting. Yet another issue is the different standards employed by different countries in measuring and reporting worker remittances. For example, the WDI definition indicates that some developing countries classify workers' remittances as a factor income receipt (and thus as a component of GNI). Including employee compensation data can help deal with this classification problem. Despite all the problems mentioned above, remittance flows tend to be the best measured aspect of the migration experience 4. This dataset, which includes observation for 137 countries over 34 years is, to our knowledge, the best available data on remittances. We also collected data on corruption indicators (Transparency International) and the UN human development indicator. The TI Corruption Perceptions Index (CPI) 5 ranks countries in terms of the degree to which corruption is perceived to exist among public officials and politicians. It is a composite index, drawing on corruption-related data in expert surveys and reflects the views of business people and analysts from around the world, including experts who are locals in the countries evaluated. Although the CPI 3 See for example, Moldovan Microfinance Alliance Report (2003) Abroad Work, Remittances, and their Utilization Patterns ; Roberts and Banaian (2004), Remittances in Armenia: Size, Impacts, and Measures to Enhance Their Contribution to Development, (mimeo) 4 See Adams and Page (2003), International Migration, Remittances, and Poverty in Developing Countries, World Bank Policy Research Working Paper 3179, December 2003 5 Transparency International, http://www.transparency.org/pressreleases_archive/2003/2003.10.07.cpi.en.html, retrieved April 3, 2005 9
index now spans 155 countries, it is available only starting 1995 and as few as 36 countries have continuous observations during 1995-2003. The human development index (UNHDI) 6 focuses on three measurable dimensions of human development: living a long and healthy life, being educated and having a decent standard of living. Thus it combines measures of life expectancy, school enrolment, literacy and income to allow a broader view of a country s development than does income alone. The UNHDI spans 180 countries and some 100 of them have continuous observations for 28 years. However, it is not exactly a measure of quality of institutions and any coefficients on this variable should be treated with care. Additionally, we collected governance research indicators developed by Kaufmann, Kraay, and Mastruzzi (2003) 7, for six dimensions of governance: Voice and Accountability (measures the extent to which citizens of a country are able to participate in the selection of governments). Political Stability and Absence of Violence (captures the idea that the quality of governance in a country is compromised by the likelihood of wrenching changes in government). Government Effectiveness (focuses on inputs required for the government to be able to produce and implement good policies and deliver public goods). Regulatory Quality (measures the incidence of market-unfriendly policies such as price controls or inadequate bank supervision, as well as perceptions of the burdens imposed by excessive regulation in areas such as foreign trade and business development). Rule of Law (measures the success of a society in developing an environment in which fair and predictable rules form the basis for economic and social interactions, and importantly, the extent to which property rights are protected). 6 United Nations Human Development Reports, http://hdr.undp.org/statistics/, retrieved April 3, 2005 7 Kaufmann, Kraay, Mastruzzi (2003) Governance Matters III: Governance Indicators for 1996-2002, World Bank Policy Research Working Paper 3106, http://www.worldbank.org/wbi/governance/pdf/govmatters3_wber.pdf, retrieved April 3, 2005 10
Control of Corruption (measures perceptions of corruption conventionally defined as the exercise of public power for private gain). The use subjective institutional measures instead of objective institutional measures in growth empirics are quite consistently verified and are considered to be a promising research avenue (see Moers, 1999). However, Rodrik (2004) points out that the most commonly-used institutional quality measure are based on surveys of domestic and foreign investors, thus capturing perceptions rather than any of the formal aspects of the institutional setting. This in his view creates two important problems perceptions are shaped not just by institutional environment, but also by many other aspects of the economic environment, creating endogeneity and reverse causality issues, and even when causality is established, the results do not indicate the specific institutional design that led to the measured outcome. Despite these obvious shortcomings of the data, we employ the indices described above merely to find an indication that the institutional environment might have an impact on the relationship between remittances and economic growth, leaving it for further research to consider in more detail on how significant this impact is and how exactly does the relationship play out. IV. METHODOLOGY OF ANALYSIS As mentioned above, our model extends the work of Chami, Fullenkamp, and Jahjah 8 (hereafter CFJ), which posits that since the sending of remittances takes place under asymmetric information and uncertainty, remittances are burdened with a moral hazard problem that limits their ability to contribute to positive business and human capital investment in developing economies, thus leading to negative economic growth. We briefly outline CFA model before considering any extensions. The model uses panel data on remittances, per capita GDP, gross capital formation (formerly categorized as gross domestic investment), and net private capital flows, all reported over the 1970-1998 8 Chami, Ralph, Connel Fullenkamp, and Samir Jahjah s (2003), Are Immigrant Remittance Flows a Source of Capital for Development?, IMF Working Paper WP/03/189 11
period. It first examines the relationship between worker remittances and per capita GDP growth using standard population-averaged cross-section estimation. Δ = β i 0 + β1 y0i + β 2wri + β 3gcfi + β 4 y npcf + u i i where y is the log of real GDP per capita, y 0 is the initial value of y, wr is the log of worker remittances to GDP ratio, gcf is the log of gross capital formation to GDP ratio, and npcf is the log of net private capital flows to GDP ratio. CFA then considers an alternative to this estimation using change in the log workers remittances to GDP ratio as independent variable: Δ = β i 0 + β1 y0i + β 2Δwri + β 3gcf i + β 4 y npcf + u They point out that this latter estimation is preferred as it captures the dynamic nature of private transfers. They then used a panel estimation to provide heterogeneity in the estimated coefficients and to capture dynamic effects, estimating one-way and two-way fixed effects, i i One way fixed effects: Two-way fixed effects: Δ it = β 0i + β1δwrit + β 2 gcf it + β 3 y npcf + u Δ it = β 0i + β1t + β 2Δwrit + β 3gcfit + β 4 y npcf + u it it it it Lastly, lagged income gap and interest gap between the sending country and the U.S. (as a proxy for receiving countries) is introduced as an instrument to account for the endogeneity problem. The two-stage regressions are estimated as follows: First stage: Δwr = β + β ( r r ) + ε it 0 1( yi yus ) t 1 + β 2 i us t it where wr is the log of worker remittances to GDP ratio, y is per capita GDP, and r is the money marker interest rate. Second stage: Δ it = β 0 + β Δdwr ˆ 1 + β 2 gcfit + β 3 y npcf + U it it 12
where ŵr are fitted growth rates of remittances. Their results show a negative correlation between remittance growth and per capita GDP which, according to the authors, indicates that remittances fluctuate countercyclically. Furthermore, the coefficients on results of the two-stage regressions are also negative and robust. The authors therefore conclude that remittances have a causal negative impact on GDP growth. They also interpret the contrast between the negative effect of remittances and the positive effect of foreign direct investment (as part of net private capital flows) as evidence that remittances should not be considered equivalent to capital flows. CFA model, however, has important drawbacks. Lucas (forthcoming) notes that crosssection studies of the type utilized by CFA are notoriously sensitive to the precise list of controls introduced, and the range permitted in this study is relatively low. The inclusion, on the other hand, of a control on investment in the home country means that any growth generated by remittances through increased investment goes unrecognized. Moreover, when it comes to the results of the study, it can be exceedingly difficult to disentangle the reverse causality and determine whether remittances cause slower growth, or whether slower growth causes more migration and possible more remittances per migrant. The instrument used in the model does not seem to be effective in eliminating the bias, indicated particularly by the insignificance of the interest rate gap differential in the first stage. We therefore extend CFA as follows. Firstly, we prefer to use remittance/gdp ratio instead of growth of remittance/gdp ratio, as we believe that remittances should enter the model like, for instance, investment does to capture its effects on growth. We therefore think that the first equation and its associated extensions are more appropriate for this kind of analysis. However, for comparison purposes we conduct the panel estimations using both approaches. 13
Secondly, and more importantly, we augment CFA model with institutional variables to get an indication of whether institutions has a role to play in the analysis of remittances effect on growth. If they do, one would expect the negative coefficient on remittances to become insignificant or even positive, and the coefficient on investment to increase. V. EMPIRICAL ESTIMATES The results of the analysis conducted in accordance with the model are indicated in Tables 4 to14. The main result of our analysis confirms the hypothesis and expectation laid out above most of the coefficient estimates on worker remittances generally are statistically insignificant, and in some specification where they are statistically significant, they are positive. The cross section analysis conducted as the average over the 1970-2003 period (Tables 4 and 5) yields insignificant results. However, including the Transparency International (TI) corruption index and the UN human development indicator (UNDHI) in the quadratic equation makes the positive coefficient highly significant (Table 5). It suggests that there is a robust positive correlation between increase in remittances and GDP growth if institutional quality is accounted for. The same cross section estimation run over the 1991-2003 period (Tables 6 to 9) yields less significant, but still positive coefficients, thus refuting the CFA s proposition that remittances are countercyclical in nature. The panel estimation (Tables 9 to 12) gives similar results the coefficients are insignificant, with the exception of the estimations that control for the TI corruption index and UNHDI. The coefficient remains the same whether random effects or fixed effects are used, and both for the 1970-2003 and the 1991-2003 time periods. The results provide support for the hypothesis that quality of institutions is an important factor in determining how remittances affect growth. The replication of the two-stage regression yields exclusively insignificant results, regardless of whether only lagged income gap, or both income gap and interest rate are 14
used as instruments, and regardless of whether the estimations are conducted over the 1970-2003, or the 1991-2003 time periods. Including the TI corruption index, UNHDI, or governance indicators in the analysis does not change these results. It is interesting to note that when region dummy variables are included in the crosssectional analysis, the resulting coefficients on both Central Europe and Europe and Central Asia are negative and statistically significant. Further research could include interactive terms and a more focused analysis of how results differ for these two subregions. A robustness check including data on compensation of employees yields inconclusive results and thus is not reported here. Moreover, the results are also very sensitive to changes in the considered periods and countries included. Further analytical consideration and research is necessary to determine the most sensible option to be employed within the framework of this particular model. Note that lack of reliable data, lack of significance of the control variables, issues of conceptual endogeneity as described above, etc. affect the scope of our analysis and therefore the results obtained here cannot be conclusive and should be considered with caution. The important point, however, is that the few specifications that do give robust results always establish a positive relationship between remittances and growth, and because of this we can say, with some level of confidence, that the CFA conclusion that remittances and growth has a negative relationship does not hold. VI. CONCLUSION AND POLICY RECOMMENDATIONS Both the conceptual and empirical analyses seem to point to the fact that institutions can play a role in how remittances affect economic growth. A sound institutional environment has been found to affect the volume and efficiency of investment; hence in the presence of good institutions, remittances could be invested in a greater amount and 15
more efficiently, ultimately leading to higher output. At the very least, this conclusion warrants further research in this area. If further evidence supports the proposition laid out above, this could also have significant policy implications. A number of researchers have expressed skepticism regarding the ability of governments to affect the manner in which remittances are used. For example, Kapur (2004) points out that active government attempts to encourage or require remittances to be invested are unlikely to have significant economic benefit. If, however, it is found that institutions matter for the manner in which remittances are used, then the best way for recipient country governments to ensure that remittances contribute to positive economic growth is to foster better quality of institutions, thus ensuring that a greater proportion of remittances is utilized for productive investment. Based on the analysis and conclusions above, recommendations could be divided into suggestions for further research and available policy options: 1) Recommendations for Further Research Further research should test whether the proposition is robust to the use of other types of growth regressions and models. Noting the difficult endogeneity relationship between remittances, institutions, and growth on one hand, and remittances, institutions and investment on the other hand, it might be helpful to focus first on how institutions influence the effect of remittances on investment first, and then determine what the general implications for economic growth are. However, even if a separate equation for investment is introduced, the direct and indirect effects of remittances and institutions on growth will most likely be difficult to disentangle and as such will have little value in devising policy recommendations. Therefore qualitative case studies, and even a collection of anecdotal evidence might be more helpful for determining the exact manner in which institutions influence remittance 16
use and what policies might help governments ensure that remittances foster economic growth. 2) Policy Recommendations Even in the absence of more tangible evidence on the specific role of institutions, the World Bank and other development organizations could implement policies aimed at achieving a better understanding of remittances and encourage better transparency in their use. This requires, above all, better data collection practices. There are various ways in which data collection could be improved. For example, Kapur (2004) put forward the proposition that remittance data should become part of the IMF s Special Data Dissemination Standards (SDDS) to both address the severe problem of consistency and timeliness of remittance data. This, however, will not resolve the issue of recording unofficial flows. One potential way to determine the amount of both official and unofficial flows would be by including consistent, uniform, and comprehensive questions regarding migration and remittance flows in household surveys. Currently, only some household surveys include questions on migration. The results of household surveys could then be extrapolated and compiled in a central location that would comprise data on all the countries. Furthermore, unofficial remittances could be brought out into the more transparent official stream by policies aimed at reducing transfer costs and banking the un-banked. Finally, and just as importantly, development organizations should keep in mind that institutions per se are associated with better growth. Thus, even in the absence of solid evidence that would establish the link between institutions and remittances, policies aimed at improving the quality of institutions would be both desirable and necessary. 17
Table 1: Worker Remittances (percent of GDP) (1970 2003) Country Mean Min Max St. Dev. Freq. Lebanon 0.3313325 0.0514004 0.6403061 0.2237674 8 Samoa 0.2671451 0.1794927 0.3825151 0.0762469 16 Tonga 0.200987 0.1180229 0.3710318 0.0757166 16 Yemen, Rep. 0.190534 0.1172514 0.3103261 0.0576831 14 Jordan 0.1759377 0.0262784 0.2491776 0.0552101 32 Bosnia and Herzegovina 0.1705852 0.124956 0.2396933 0.0446899 6 Cape Verde 0.1579204 0.1107981 0.21187 0.0308565 18 Albania 0.1551125 0.1035235 0.2237601 0.0360223 11 Serbia and Montenegro 0.1079907 0.0430375 0.1466817 0.0364781 8 Dominica 0.1039417 0.0376735 0.1964047 0.0412868 14 Tajikistan 0.0852514 0.0464666 0.1435598 0.0514067 3 Egypt, Arab Rep. 0.0818332 0.0286842 0.145833 0.03637 27 Kiribati 0.0797618 0.0578232 0.104811 0.0156607 10 Vincent and the Grenadines 0.0793358 0.0701286 0.0911643 0.00923 5 El Salvador 0.0722603 0.0030442 0.1404709 0.0478137 27 Nicaragua 0.067138 0.0055779 0.1355964 0.0393693 12 Morocco 0.0659386 0.0532942 0.0961891 0.0108356 29 Haiti 0.0596714 0.0310368 0.0862064 0.0169979 19 Portugal 0.0585341 0.018894 0.1015454 0.0258514 29 Uganda 0.0569736 0.0389879 0.0855895 0.0219973 4 St. Kitts and Nevis 0.0553245 0.0136277 0.119545 0.044245 11 Jamaica 0.0552941 0.0100207 0.1633866 0.0467051 28 Grenada 0.0541876 0.0391983 0.0662208 0.0117513 5 Burkina Faso 0.0533151 0.0097772 0.0910156 0.0216757 25 Pakistan 0.0514183 0.0169951 0.1024763 0.0254865 28 St. Lucia 0.0500953 0.0344082 0.0738494 0.0175485 6 Sri Lanka 0.0496337 0.0022576 0.0776882 0.0211605 29 Vanuatu 0.0464327 0.0265397 0.0847256 0.0162008 18 Dominican Republic 0.04636 0.0077793 0.1340372 0.0305094 34 Somalia 0.0444819 0.0102252 0.0948517 0.0338949 5 Benin 0.0421589 0.013164 0.0803244 0.0137736 30 Honduras 0.0404315 0.0067429 0.1146757 0.0342295 17 Tunisia 0.0399647 0.030552 0.0509198 0.0055461 28 Belize 0.0382963 0.0148312 0.098733 0.023714 19 Comoros 0.0365306 0.0106921 0.0820036 0.0189217 16 Mali 0.0358847 0.0188923 0.0585387 0.008854 28 Ecuador 0.0339959 0.0048281 0.0825963 0.0266265 14 Maldives 0.0330303 0.0228303 0.0432303 0.014425 2 Georgia 0.0300705 0.0136827 0.0532945 0.0130214 7 Bangladesh 0.0297449 0.0018656 0.0671791 0.0140259 28 Sudan 0.0286494 0.0039788 0.0684734 0.0206582 27 Antigua and Barbuda 0.0285369 0.0240059 0.0341156 0.0039418 5 Guatemala 0.026678 0.0054453 0.0829538 0.0203443 16 Guyana 0.0264828 0.0023842 0.0705921 0.0202267 9 Croatia 0.0261588 0.0195366 0.0303593 0.0033606 11 18
Mongolia 0.0254653 0.0056577 0.0526499 0.0211187 6 Barbados 0.0230312 0.0121836 0.0395324 0.0082282 16 Turkey 0.0226323 0.0097532 0.0381915 0.0067885 30 Greece 0.0209556 0 0.0290321 0.0063121 27 Macedonia, FYR 0.0202112 0.0102416 0.0314111 0.0063371 8 Senegal 0.0197927 0.005495 0.0272945 0.0045345 26 Nepal 0.0181095 0.0097665 0.031209 0.0079358 11 Azerbaijan 0.0177587 0.0108356 0.0261878 0.0064924 5 Malta 0.0163035 0.0000564 0.0567549 0.0167493 32 Cambodia 0.013984 0.0029079 0.0299641 0.0123237 11 Togo 0.0136661 0.0000203 0.0409495 0.009289 30 India 0.0133781 0.0044824 0.0251337 0.0054805 29 Seychelles 0.0131127 0.0020627 0.023914 0.0092343 6 Sao Tome and Principe 0.012739 0.0002843 0.0672994 0.0150288 18 Mozambique 0.0119379 0.0119379 0.0119379 1 Cyprus 0.0116614 0.0001134 0.0184845 0.0061419 11 Nigeria 0.0113486 0.0000862 0.053003 0.0178136 23 Peru 0.0103004 0.0033087 0.0140422 0.0034551 14 Kyrgyz Republic 0.0097996 0.0005961 0.0313574 0.0128697 11 Mexico 0.0095711 0 0.0211882 0.0044092 25 Costa Rica 0.0093724 0.0064066 0.0125841 0.0020237 9 Colombia 0.0089267 0.0008335 0.0386452 0.0091555 34 Algeria 0.0087227 0.0050965 0.0166503 0.0036647 15 Paraguay 0.0084205 0.0000357 0.0204427 0.0090298 19 Guinea-Bissau 0.0067382 0.004099 0.0091209 0.0019238 5 New Zealand 0.0060614 0.001548 0.0093944 0.0021317 29 Mauritania 0.0057485 0.000568 0.0421088 0.0084564 24 Armenia 0.0057135 0.0033337 0.008411 0.001731 9 Spain 0.0056427 0.0032668 0.0089379 0.0016271 29 Philippines 0.0056287 0.0013393 0.0128365 0.0026717 27 Oman 0.0051378 0.0032236 0.0105652 0.0018876 18 Poland 0.0050935 0.0038367 0.0057969 0.0006616 9 Ethiopia 0.0047577 0.0014446 0.0091594 0.0028304 8 Niger 0.0038901 0.0014586 0.0086432 0.0021353 22 Indonesia 0.003445 0.0001171 0.0100476 0.0030589 21 Belgium 0.0032861 0.0002967 0.0052784 0.0017814 27 Bolivia 0.0032414 0.0000961 0.0133583 0.004844 23 Latvia 0.0032288 0.0001609 0.0124749 0.0050985 7 Kazakhstan 0.0032026 0.0012881 0.0043581 0.00133 4 Uruguay 0.0030343 0.0029079 0.0031606 0.0001786 2 Ukraine 0.0030273 0.00221 0.0037346 0.0007682 3 Djibouti 0.0028664 0.0020575 0.0031875 0.0005409 4 Panama 0.0026492 0.0003884 0.0068804 0.001871 23 Ghana 0.0023323 0.0000888 0.0086454 0.0024831 25 Austria 0.0022527 0.0008535 0.0035377 0.0007314 28 Italy 0.0022248 0.0001965 0.0052271 0.0017124 34 Suriname 0.0021176 0.0001078 0.0135489 0.0046437 8 Trinidad and Tobago 0.0021109 0.0000277 0.0079562 0.0025634 27 Guinea 0.00209 0.0000321 0.0112021 0.0031814 13 Namibia 0.0020378 0.0006966 0.0029902 0.0007742 14 19
Slovenia 0.0018233 0.0005117 0.0038608 0.0012281 12 Belarus 0.0017414 0.0000246 0.0044282 0.0014975 7 Madagascar 0.0016492 0.0000324 0.0039593 0.0013829 19 Cameroon 0.0015722 0.0002908 0.0034696 0.0008526 17 Brazil 0.0015351 0.0000222 0.0044013 0.0014857 25 Aruba 0.0014399 0.0004049 0.0029148 0.0008225 11 Rwanda 0.0014317 0.0002496 0.0042063 0.0010943 21 Thailand 0.0013997 0.0013997 0.0013997 1 Zimbabwe 0.0012763 0.0000919 0.0026255 0.0009836 6 Lesotho 0.0011858 0.0001613 0.0025974 0.0007584 9 Korea, Rep. 0.0010597 0.0000472 0.0027143 0.0008374 23 China 0.0010501 0.000222 0.0049241 0.0010883 22 Russian Federation 0.0008486 0.0006704 0.001183 0.0002898 3 Moldova 0.0007853 0.000388 0.001817 0.0004417 8 France 0.0006658 0.0004341 0.0011473 0.0002039 28 Switzerland 0.0006376 0.0005067 0.0008472 0.0001068 21 Hungary 0.0005736 0.0001404 0.0011458 0.0003445 8 Lithuania 0.0005132 4.76E-06 0.0025049 0.0008043 11 Ireland 0.0005078 0.0004564 0.0005843 0.0000591 4 Congo, Rep. 0.0005055 0.0003899 0.0007923 0.0001466 9 Chad 0.0004938 0.0000323 0.001164 0.0004602 8 Botswana 0.0004641 0.0000106 0.0017343 0.0008473 4 Sweden 0.0004392 0.000092 0.0008419 0.0002457 16 Malawi 0.0004237 0.0002662 0.0006489 0.0001064 10 Slovak Republic 0.0002714 0.0001976 0.0003452 0.0001044 2 Estonia 0.0002673 4.39E-06 0.0010941 0.000411 10 Gabon 0.0002207 0.0000111 0.0007314 0.0002822 16 Norway 0.0001538 0.0000888 0.0002356 0.0000399 15 Argentina 0.0001504 0.000103 0.0001802 0.0000257 7 Romania 0.0001221 0.000054 0.000283 0.0000644 10 Japan 0.0001201 0.0000484 0.0002484 0.0000834 8 20
Table 2: Worker Remittances and Compensation of Employees (percent of GDP) (1970 2003) Country Mean Min Max St. Dev. Freq. Lebanon 0.3313325 0.0514004 0.6403061 0.2237674 8 Samoa 0.2671451 0.1794927 0.3825151 0.0762469 16 Tonga 0.200987 0.1180229 0.3710318 0.0757166 16 Yemen, Rep. 0.190534 0.1172514 0.3103261 0.0576831 14 Jordan 0.1759377 0.0262784 0.2491776 0.0552101 32 Bosnia and Herzegovina 0.1705852 0.124956 0.2396933 0.0446899 6 Albania 0.1612581 0.0180641 0.2703361 0.0594579 12 Cape Verde 0.1579204 0.1107981 0.21187 0.0308565 18 Serbia and Montenegro 0.1079907 0.0430375 0.1466817 0.0364781 8 Dominica 0.1039417 0.0376735 0.1964047 0.0412868 14 Moldova 0.0880821 0.0005819 0.1614215 0.048071 9 Tajikistan 0.0852932 0.0464666 0.1435598 0.051383 3 Georgia 0.0827444 0.0566204 0.1287059 0.0265658 7 Egypt, Arab Rep. 0.0818332 0.0286842 0.145833 0.03637 27 Kiribati 0.0797618 0.0578232 0.104811 0.0156607 10 Vincent and the Grenadines 0.0793358 0.0701286 0.0911643 0.00923 5 El Salvador 0.073878 0.0045527 0.1415823 0.0454409 28 Jamaica 0.0703278 0.0232644 0.1774717 0.0449959 28 Morocco 0.0659386 0.0532942 0.0961891 0.0108356 29 Haiti 0.0596714 0.0310368 0.0862064 0.0169979 19 Portugal 0.0585341 0.018894 0.1015454 0.0258514 29 Uganda 0.0569736 0.0389879 0.0855895 0.0219973 4 St. Kitts and Nevis 0.0553245 0.0136277 0.119545 0.044245 11 Grenada 0.0541876 0.0391983 0.0662208 0.0117513 5 Nicaragua 0.054147 0.0017512 0.1355964 0.0440585 15 Armenia 0.053755 0.0445067 0.0830379 0.0119599 9 Burkina Faso 0.0533151 0.0097772 0.0910156 0.0216757 25 Pakistan 0.0514188 0.0169951 0.1024763 0.0254866 28 St. Lucia 0.0500953 0.0344082 0.0738494 0.0175485 6 Sri Lanka 0.0496337 0.0022576 0.0776882 0.0211605 29 Dominican Republic 0.0485298 0.0077793 0.1506816 0.0340254 34 Philippines 0.04815 0.0171957 0.0978365 0.0270235 27 Vanuatu 0.0464327 0.0265397 0.0847256 0.0162008 18 Somalia 0.0444819 0.0102252 0.0948517 0.0338949 5 Benin 0.0421589 0.013164 0.0803244 0.0137736 30 Tunisia 0.0399647 0.030552 0.0509198 0.0055461 28 Belize 0.0382963 0.0148312 0.098733 0.023714 19 Comoros 0.0365306 0.0106921 0.0820036 0.0189217 16 Mali 0.0358847 0.0188923 0.0585387 0.008854 28 Croatia 0.0304254 0.0210958 0.0368536 0.0050106 11 Bangladesh 0.0297449 0.0018656 0.0671791 0.0140259 28 Syrian Arab Republic 0.0290164 0.0067672 0.0907783 0.0198678 27 Sudan 0.028732 0.0039788 0.0687679 0.0207641 27 Antigua and Barbuda 0.0285369 0.0240059 0.0341156 0.0039418 5 Guyana 0.0264828 0.0023842 0.0705921 0.0202267 9 21
Mongolia 0.0254653 0.0056577 0.0526499 0.0211187 6 Ecuador 0.0252689 0.0000528 0.0829463 0.027399 19 Honduras 0.0252078 0.0005166 0.1156789 0.0327771 30 Barbados 0.0230312 0.0121836 0.0395324 0.0082282 16 Macedonia, FYR 0.0229841 0.0153274 0.0373395 0.006858 8 Greece 0.0229113 0.0022014 0.0302576 0.0058175 27 Turkey 0.0226323 0.0097532 0.0381915 0.0067885 30 Senegal 0.0197927 0.005495 0.0272945 0.0045345 26 Kuwait 0.0196475 0.003579 0.0548776 0.0112978 27 Mozambique 0.0192664 0.006559 0.0342029 0.008374 24 Nepal 0.0181095 0.0097665 0.031209 0.0079358 11 Guatemala 0.0173379 0.0000141 0.0845897 0.0203209 27 Paraguay 0.0164477 0.0030036 0.0397614 0.0128681 29 Malta 0.0163035 0.0000564 0.0567549 0.0167493 32 Latvia 0.0138708 0.0073094 0.0268342 0.0071072 8 Togo 0.0136661 0.0000203 0.0409495 0.009289 30 India 0.0133781 0.0044824 0.0251337 0.0054805 29 Sao Tome and Principe 0.0133733 0.001265 0.0672994 0.0147765 18 Cambodia 0.0133598 0.0007443 0.0307132 0.0125758 12 Seychelles 0.0131127 0.0020627 0.023914 0.0092343 6 Azerbaijan 0.0130644 0.0009828 0.0261878 0.0097036 7 Mexico 0.0117212 0.0013156 0.0233113 0.0045344 25 Cyprus 0.0116614 0.0001134 0.0184845 0.0061419 11 Nigeria 0.0113486 0.0000862 0.053003 0.0178136 23 Slovenia 0.0108833 0.0030583 0.018967 0.0043485 12 Peru 0.0103004 0.0033087 0.0140422 0.0034551 14 Kyrgyz Republic 0.0097996 0.0005961 0.0313574 0.0128697 11 Colombia 0.0096174 0.001756 0.0388502 0.0088278 34 Algeria 0.0087227 0.0050965 0.0166503 0.0036647 15 Maldives 0.0082627 0.0029171 0.0432303 0.0092213 20 New Zealand 0.0060614 0.001548 0.0093944 0.0021317 29 Mauritania 0.0057485 0.000568 0.0421088 0.0084564 24 Spain 0.0056427 0.0032668 0.0089379 0.0016271 29 Poland 0.005384 0.0017322 0.0071768 0.001439 10 Guinea-Bissau 0.0053176 0.001143 0.0091209 0.0029125 7 Oman 0.0051378 0.0032236 0.0105652 0.0018876 18 Ethiopia 0.0047577 0.0014446 0.0091594 0.0028304 8 Costa Rica 0.0044513 0.000651 0.0135576 0.004366 27 Bulgaria 0.0042728 0.0032828 0.0051874 0.0006893 8 Iceland 0.0042543 0.0002009 0.0089346 0.0037669 27 Niger 0.0038901 0.0014586 0.0086432 0.0021353 22 Bolivia 0.003863 0.0003326 0.0168606 0.0057258 28 Indonesia 0.003445 0.0001171 0.0100476 0.0030589 21 Belgium 0.0032861 0.0002967 0.0052784 0.0017814 27 Israel 0.0029665 0.0010059 0.0061694 0.0014588 33 Djibouti 0.0028664 0.0020575 0.0031875 0.0005409 4 Panama 0.0026492 0.0003884 0.0068804 0.001871 23 Ghana 0.002384 0.0001048 0.0086454 0.0024412 25 Belarus 0.0023504 0.0000246 0.0063225 0.0020158 11 Austria 0.0022527 0.0008535 0.0035377 0.0007314 28 22
Uruguay 0.0022292 0.0005865 0.0031606 0.0014268 3 Italy 0.0022248 0.0001965 0.0052271 0.0017124 34 Ukraine 0.0021892 0.0001347 0.0066617 0.0025367 8 Suriname 0.0021424 0.0000769 0.0158596 0.003773 16 Guinea 0.0021195 0.0000321 0.0112021 0.0031686 13 Trinidad and Tobago 0.0021109 0.0000277 0.0079562 0.0025634 27 Namibia 0.0020378 0.0006966 0.0029902 0.0007742 14 Turkmenistan 0.0018577 0.0018577 0.0018577 1 Kazakhstan 0.0017589 0.0000541 0.0045156 0.001931 8 Madagascar 0.0016492 0.0000324 0.0039593 0.0013829 19 Romania 0.0015992 0.0002537 0.0031257 0.0011701 10 Cameroon 0.0015722 0.0002908 0.0034696 0.0008526 17 Russian Federation 0.0015224 0.0002629 0.0032172 0.0011261 10 Brazil 0.0014943 0.0000847 0.0044474 0.0015776 29 Aruba 0.0014399 0.0004049 0.0029148 0.0008225 11 Rwanda 0.0014317 0.0002496 0.0042063 0.0010943 21 Thailand 0.0013997 0.0013997 0.0013997 1 Zimbabwe 0.0012763 0.0000919 0.0026255 0.0009836 6 China 0.0012509 0.000222 0.0051089 0.0011846 22 Lesotho 0.0011858 0.0001613 0.0025974 0.0007584 9 Korea, Rep. 0.0010597 0.0000472 0.0027143 0.0008374 23 Angola 0.0006832 0.0006832 0.0006832 1 Lao PDR 0.0006673 0.0002268 0.0019582 0.0006823 10 France 0.0006658 0.0004341 0.0011473 0.0002039 28 Switzerland 0.0006376 0.0005067 0.0008472 0.0001068 21 Hungary 0.0005736 0.0001404 0.0011458 0.0003445 8 Lithuania 0.0005132 4.76E-06 0.0025049 0.0008043 11 Ireland 0.0005078 0.0004564 0.0005843 0.0000591 4 Congo, Rep. 0.0005055 0.0003899 0.0007923 0.0001466 9 Chad 0.0004938 0.0000323 0.001164 0.0004602 8 Botswana 0.0004641 0.0000106 0.0017343 0.0008473 4 Sweden 0.0004392 0.000092 0.0008419 0.0002457 16 Malawi 0.0004237 0.0002662 0.0006489 0.0001064 10 Argentina 0.0002929 0.0000673 0.0007276 0.0001642 22 Slovak Republic 0.0002714 0.0001976 0.0003452 0.0001044 2 Estonia 0.0002673 4.39E-06 0.0010941 0.000411 10 Gabon 0.0002207 0.0000111 0.0007314 0.0002822 16 United States 0.0001877 0.0000192 0.0002983 0.0001079 27 Norway 0.0001538 0.0000888 0.0002356 0.0000399 15 Japan 0.0001201 0.0000484 0.0002484 0.0000834 8 Venezuela, RB 0.000102 0.0000161 0.0002477 0.0000813 15 Chile 0.0000822 4.14E-06 0.0001841 0.0000727 12 23