This article was downloaded by: [Northwestern University] On: 27 January 2015, At: 02:41 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

The European Journal of Development Research Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/fedr20

Remittances and economic growth in developing countries a

b

Gyan Pradhan , Mukti Upadhyay & Kamal Upadhyaya

b

a

ABEMIS Department , Westminster College , Fulton, USA

b

Department of Economics , Eastern Illinois University , Charleston, USA

c

Department of Economics and Finance , University of New Haven , West Haven, USA Published online: 04 Mar 2011.

To cite this article: Gyan Pradhan , Mukti Upadhyay & Kamal Upadhyaya (2008) Remittances and economic growth in developing countries, The European Journal of Development Research, 20:3, 497-506 To link to this article: http://dx.doi.org/10.1080/09578810802246285

PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions

The European Journal of Development Research Vol. 20, No. 3, September 2008, 497–506

Remittances and economic growth in developing countries Gyan Pradhana, Mukti Upadhyayb and Kamal Upadhyayac*

Downloaded by [Northwestern University] at 02:41 27 January 2015

a

ABEMIS Department, Westminster College, Fulton, USA; bDepartment of Economics, Eastern Illinois University, Charleston, USA; cDepartment of Economics and Finance, University of New Haven, West Haven, USA This paper examines the effect of workers’ remittances on economic growth in a sample of 39 developing countries using panel data from 1980– 2004 resulting in 195 observations. A standard growth model is estimated using both fixed-effects and random-effects approaches. The empirical results show a significant overall fit based on the fixed-effects method as the random-effects model is rejected in statistical tests. Remittances have a positive impact on growth. Since official estimates of remittances used in our analysis tend to understate actual numbers considerably, more accurate data on remittances is likely to reveal an even more pronounced effect of remittances on growth. Cet article examine l’impact des transferts mone´taires effectue´s par les travailleurs migrants sur la croissance e´conomique d’un e´chantillon de 39 pays en voie de de´veloppement, a` partir de donne´es de panel, constitue´ de 195 observations allant de 1980 a` 2004. Deux mode`les de croissance standard sont propose´s afin d’analyser ces donne´es empiriques, l’un base´ sur un mode`le a` effets fixes, et l’autre sur un mode`le a` effets ale´atoires. Les tests statistiques de´montrent que le premier mode`le est plus significatif que le second, ce qui sugge`re que les transferts mone´taires effectue´s par les travailleurs migrants ont bien un impact positif sur la croissance e´conomique du pays d’origine de ces derniers. Puisque les estimations officielles utilise´es dans notre e´tude ont tendance a` minimiser conside´rablement le volume de ces transferts, des donne´es plus pre´cises sont susceptibles de re´ve´ler un effet encore plus prononce´. Keywords: remittances; economic growth; panel data; fixed-effects estimation; randomeffects estimation

1. Introduction Remittances from migrant workers in rich countries are increasingly important to developing countries. Until recently, these payments received little attention from governments and financial markets because they were usually sent in small amounts. In recent years, however, remittances have become difficult to ignore as they are large in the aggregate and have become essential to many developing countries. The World Bank (2007a) estimates that total remittances to developing countries will amount to $240 billion in 2007, up from $31.2 billion in 1990. The actual number is believed to be much higher because in many instances these transactions are not recorded. Many source and recipient nations are just beginning to compile data from their banking systems. However, although transfers through banks are relatively easy to track, these transactions only account for a small percentage of the total flow. Most of the money is sent through wire-transfer agencies which are not well regulated or through unregulated channels that leave no trace. According to the World Bank (2006), total remittances could be as much as 50% higher than official estimates if those sent through informal channels are included.

*Email: [email protected] ISSN 0957-8811 print/ISSN 1743-9728 online q 2008 European Association of Development Research and Training Institutes DOI: 10.1080/09578810802246285 http://www.informaworld.com

Downloaded by [Northwestern University] at 02:41 27 January 2015

498

G. Pradhan et al.

While India, China and Mexico account for a third of remittance flows to the developing world, many other developing countries also rely heavily on these flows. For instance, these payments provide more than 20% of GDP for Tonga, Lesotho and Jordan; more than 15% of GDP for Albania, Nicaragua, Yemen and Moldova; more than 10% of GDP for Lebanon, El Salvador, Cape Verde and Jamaica; and more than 5% of GDP for many countries, including Morocco, Dominican Republic, Vanuatu, Philippines, Honduras, Uganda, Ecuador and Sri Lanka (Ratha 2004). In 36 countries, remittances exceed all other imports of public and private capital. Remittances are more than twice the size of net official flows and are second only to foreign direct investment as a source of external finance for developing countries. The benefits from workers’ remittances to recipient countries are numerous. As indicated, remittances make up a significant proportion of GDP of many developing countries that understandably have become dependent on these payments. Remittances can finance the much needed investment in developing countries and contribute to increased productivity and economic growth. They are believed to reduce poverty since it is usually the poor in developing countries who migrate and send back remittances. By raising incomes of households, remittances can also lead to higher consumption that can have a multiplier effect on aggregate demand and output. While development loans come with a liability and obligation to pay interest, remittances do not. Moreover, they cannot be wasted by governments since they are sent directly to the individuals for whom they are intended. It has also been argued that they are a more stable source of funding than foreign direct or foreign portfolio investments which tend to be particularly volatile in developing countries. The international migration of unskilled labor can also be beneficial to the country of emigration in other ways. Unskilled workers are more plentiful, so their loss is likely to be less significant to the sending economy (Perkins, Radelot, Snodgrass, Gillis and Romer 2001). Another potential benefit is training. Unskilled workers who go abroad and return to their native countries after a few years may bring back usable skills acquired abroad. Despite their benefits, there are economic costs associated with reliance on remittances. They may promote idleness among those who benefit. Moreover, it has been argued that remittances may have minimal impact on saving and investment since they are used mainly for daily consumption (Kapur and McHale 2005). As such, the notion that remittances help to promote economic growth may be mistaken. Another potential negative impact of remittances is that such inflows may cause an appreciation in the value of a country’s currency, lower net exports, and adversely affect economic growth. A distinction between unskilled labor and skilled or educated labor may be pertinent. This differentiation may not be critical from a global perspective. It has been argued that according to the basic principles that govern the benefits from international trade, world output is maximized when everyone works where the compensation and the productivity are the greatest. However, the distinction is important from the viewpoint of developing countries because skilled and unskilled labor have different opportunity costs (Kapur and McHale 2005). The migration of educated, highly skilled workers is harmful to most developing countries and has been denounced as ‘brain drain’ for two reasons. The first is that these qualified people represent one of the scarcest resources of developing countries. This loss of human capital lowers productivity and economic growth. The second is that, in most cases, the state has invested a great deal of time, effort and money on their education. When these workers leave to work in foreign countries, not only are their services lost but a high cost must be incurred to train replacements. Although it raises world output, international migration of skilled workers is likely to worsen the distribution of income between rich and poor countries. To offset this effect, some have proposed a tax on the brain drain, to be collected by the governments of developed countries to which professional and technical personnel from developing countries have migrated.

Downloaded by [Northwestern University] at 02:41 27 January 2015

The European Journal of Development Research

499

The proceeds of the tax would be transferred to the poor countries of origin as compensation for their loss of talent (Perkins et al. 2001). The preceding discussion indicates that, in theory, the effects of workers’ remittances on economic growth, poverty and income distribution may be ambiguous. What about the empirical evidence? There is a growing body of literature in recent years that has examined the economic effects of remittances (Ozden and Schiff 2005). These studies serve to underscore the increasing importance of remittances provided by migrant workers from developing countries working in other countries. For instance, Ratha (2003) emphasizes the growing importance of remittances as a source of external funds for developing countries. His study finds that remittances are the second largest source of external funding (after foreign direct investment) for these countries. Edwards and Ureta (2003) examine the effects of remittances on education in El Salvador and report that remittances have an important effect on school retention. Stahl and Arnold (1986) show that remittances spent on consumption have positive multiplier effects on aggregate demand. Adelman and Taylor (1990) provide evidence from Mexico to show that remittances offset some of the output losses from emigration of highly skilled workers. They find that for every dollar of remittance Mexico received, its output increased by about three dollars. Similarly, Desai, Kapoor and McHale (2001) show that remittances from emigration of Indian workers more than offset the loss of tax revenue. They estimate the net financial loss from Indian emigration to the United States at 0.3% to 0.6% of GDP but remittances amounted to more than 2%. Adams (1998, 2002) uses household surveys in Pakistan to suggest that remittances resulted in higher savings and investment. The empirical evidence on the effect of remittances on economic growth, poverty and income inequality is more mixed. Catrinescu, Leon-Ledesma, Piracha and Quillin (2005) find a weak positive effect of remittances on long-term growth, and Faini (2001), Taylor (1992) and Stark and Lucas (1988) find a positive association between remittances and economic growth. Jongwanich (2007) examines the effects of remittances on growth and poverty in selected Asian and Pacific countries using panel data for 1993– 2000. He finds that remittances have a small positive impact on growth but a significant favorable impact on poverty reduction. Similarly, a recent IMF (2005) study of 101 developing countries finds a positive and significant impact of remittances on poverty reduction but no impact on economic growth. On the other hand, Chami, Fullenkamp and Jahjah (2003) find a negative impact of remittances on economic growth for 113 countries. Adams and Page (2003) use data for 74 developing countries to examine the effects of international migration and remittances on poverty. They conclude that remittances serve to reduce poverty. Taylor (1999) and Adelman and Taylor (1990) suggest that remittances have reduced income inequality in Mexico. On the other hand, McCormick and Wahba (2002) find that remittances have increased income inequality in Egypt. The large remittance flows to developing countries may be expected to have important implications for economic growth in these countries. Yet very few studies have examined the relationship between remittances and economic growth in a broad cross-section of developing countries. Moreover, as discussed above, the available empirical evidence on the relationship is quite ambiguous. In light of these considerations, this paper makes an attempt to contribute to the existing literature by examining the impact of remittances on economic growth using panel data for 39 developing countries. Section 2 discusses remittance flows across the world by region. Section 3 discusses the theoretical background and methodology. The empirical results and analysis are presented in section 4. Finally, section 5 summarizes and concludes the paper. 2. Remittance flows by region Earlier we discussed the significance of remittances for developing countries. In this section we discuss the trends in the global flow of remittances over the past 15 years by region.

500

Downloaded by [Northwestern University] at 02:41 27 January 2015

Table 1.

G. Pradhan et al. Remittance receipts by region (US$ bn).

Eastern Europe and Central Asia East Asia and Pacific Region South Asia Latin America and Caribbean Middle East and North Africa Sub-Saharan Africa High Income OECD countries High Income Non-OECD countries World

1990

1995

2000

2005

Annual growth* (%, 1990– 2005)

3.2 3.3 5.6 5.8 11.4 1.9 36.5 1 68.6

8.1 9.7 10 13.4 13.6 3.2 43.4 1.1 102.3

13.4 16.7 17.2 20.1 13.2 4.6 45.8 0.7 131.8

30.8 43.9 34.9 47.6 23.5 7.4 68.1 1.2 257.5

15.1 17.3 12.2 14.0 4.8 9.1 4.1 1.2 8.8

*Authors’ calculations. Source: Global Economic Prospect 2006, and World Development Indicators, 2007, Washington DC: World Bank.

Table 1 presents the flow of remittances by region for 1990, 1995, 2000 and 2005. The last column of the table reports the average yearly growth rate of remittances by region between 1990 and 2005. As reported in the table, with the exception of the Middle East and North Africa, and Sub-Saharan Africa, remittance flows have been growing at double digits in the rest of the developing world. In Eastern Europe and Central Asia, the amount of remittances has increased almost ten-fold between 1990 and 2005. During the same period, remittance flows increased more than 13 times in East Asia and the Pacific region. In Latin America and the Caribbean, South Asia, and Sub-Saharan Africa, remittance flows increased approximately eight, six and four times. The growth of remittances in the Middle East and North Africa, High Income OECD and High Income Non-OECD countries is not that pronounced, however. In the 1990s, countries in Eastern Europe and Central Asia made the transition from centrally planned to free market economies which permitted labor migration in search of better paying jobs in the oil-rich Middle East and industrialized Western Europe. This migration resulted in a significant growth of remittances to Eastern Europe and Central Asia. With the expansion in economic activity, Japan, South Korea, Hong Kong, Taiwan and Singapore began to experience labor shortages. In order to address this shortage, these countries adopted more liberal policies with regard to migrant workers in the 1990s. As a consequence, there was an influx of migrant labor in these countries, particularly from countries such as Thailand, the Philippines and Indonesia, which resulted in a significant growth in remittance flows for these countries. For religious reasons, Pakistani and Bangladeshi migrant workers have always been preferred in Saudi Arabia and other Islamic Middle Eastern countries. With the implementation of the economic liberalization program in India, the government lifted many regulations with regard to foreign exchange and travel abroad for work. As might be expected, a significant number of migrant workers began to travel to foreign countries, particularly the Middle East. Moreover, the explosion in the information technology industry in the United States and other industrialized countries attracted a large number of South Asians, particularly Indians, to migrate to these countries. The overall consequence was a huge increase in the flow of remittances to South Asia. Countries in Latin America and the Caribbean have been the main sources of migrant workers for the United States and Canada. There is also an active movement of migrant workers within the Latin American and Caribbean region itself. For instance, Mexico receives a large number of migrant workers from its southern neighbors. Moreover, the implementation of NAFTA and the economic boom of the 1990s increased the sharply demand for workers in the United States and Canada.

Downloaded by [Northwestern University] at 02:41 27 January 2015

The European Journal of Development Research

501

This huge arrival of migrant workers from Mexico and other countries in Latin America as well as the Caribbean has led to a considerable growth in remittances to this region. Until the late 1980s, countries such as Tanzania, Zambia and Mozambique restricted the mobility of people, often even from one region of the country to another. As pressure mounted from donor agencies and industrialized countries, countries in Sub-Saharan Africa began to adopt more liberal economic policies and more democratic political systems that guaranteed basic human rights such as mobility of labor both within and across national borders. In addition, the end of apartheid policies in South Africa opened the door for migrant workers from the Sub-Saharan region. The result was that remittances became one of the main sources of foreign exchange earnings for these countries. It is also apparent from Table 1 that the growth of remittances has not been very high in the Middle East and North Africa, and in OECD countries. However, the remittance flows to these regions in absolute terms have been much larger than in other regions, mainly due to the relatively free movement of labor within the Middle East and within OECD countries. 3.

Theoretical background and methodology

To estimate how remittances affect growth of per capita income, our model reflects several considerations. As noted above, remittances received can add to domestic consumption and savings. To the extent that remittances help families in the source country to maintain a minimum standard of living, remittances can raise the family members’ productivity. A part of remittances can also raise saving and investment in the source country. This is the variable of our main interest and we include it directly in the model. Growth theory predicts that because of diminishing returns to capital, countries that start out with a low per capita income tend to grow relatively fast which allows them to catch up with countries that were already at higher stages of development. We thus control for initial income while studying the influence of other factors and expect a negative sign for the coefficient of initial income. One of the important factors that determine growth is the rate at which a country saves and invests. Most past research attributes a significant portion of per capita income growth to the share of investment in GDP. This ratio measures changes in the capital stock in the economy and represents the growth effects of factor accumulation. Among the factors of production commonly recognized as crucial to growth, as suggested by Mankiw, Romer and Weil (1992) is human capital. Empirically, however, this variable does not perform very well. An important reason is that human capital that is supposed to reflect the status of education, training, and health among the working population has eluded a clear definition or measurement. The part of growth not directly attributable to factor accumulation is total factor productivity which is determined by many factors – economic, social and political. In this study, we use only two of them to avoid high correlation among the independent variables and for reasons of data availability. The first is the degree of openness of a country captured by total trade-to-GDP ratio or simply export – GDP ratio. Although openness is an outcome-based rather than a policy-based measure, the use of this variable as an indicator of an economy’s external orientation has been abundantly justified on theoretical and practical grounds (Warner 2003). The second is a ‘polity’ variable that provides ‘annual information on political regime and authority characteristics for all independent states’ and is constructed under the Polity IV project maintained by the University of Maryland (Polity IV Project website). The polity variable is the difference between two scores, one for democracy and one for autocracy. A country with a high

502

G. Pradhan et al.

index for democracy tends to have a low index for autocracy; yet there are many instances in the data where a low score for one is not associated with a low score for the other for the same country. For this reason, we use the difference between the two as reflecting the nature of political regime that influences economic growth. Since the durability of a regime should matter more for growth, we estimate our model with a one-period lag (of five years) of this variable as well as its contemporaneous value. The estimated model can be summarized as follows:

Downloaded by [Northwestern University] at 02:41 27 January 2015

Dyi;t ¼ a0 þ byi;t21 þ g1 remiti;t þ g2 invi;t þ g3 openi;t þ g4 politi;t þ ai þ 1i;t

ð1Þ

where Dyi,t is the annual percentage growth of real per capita GDP in country i in the five-year period t, yi,t21 is the logarithm of per capita income in country i in time t 2 1 measured in international purchasing power parity dollars. Thus the dependent variable, after we multiply log differences by 100, gives the percentage growth of per capita income. Remiti,t is the logarithm of per capita remittance received measured in constant US dollars.1 The variable invi,t is the fraction of income invested, openi,t is openness measured as either total trade in goods and services as a percentage of GDP or exports as a percentage of GDP. Our political variable that indicates the strength of political institutions is politi,t is ‘polity’ measured as the difference between the scores for democracy and autocracy. As indicated earlier, an alternative specification of the model replaces politi,t with politi,t21. The values of both democracy and autocracy range between 0 (lowest) and 10 (highest). Finally, ai represents the country-specific effects that must be controlled for before we estimate the effects of remittances and other variables on growth in the model. This composite variable allows us to control explicitly for such factors as culture, work ethic, and others that influence the growth rate of an economy and hence prevents a misspecification of the model. All the variables on the right-hand side of Equation (1) are averages for five-year periods. These periods are: 1980 –84, 1985 –89, 1990 –94, 1995– 99, and 2000 –04. The data were compiled mainly from World Development Indicators CD-ROM 2007, and a few missing data points were filled in from the IMF and country sources. The sample consists of 39 low- and middle-income countries for which data on remittances and other variables were available. Thus, we have a panel data of 39 £ 5 ¼ 195 observations. The estimation of Equation (1) requires some consideration of possible country-specific factors that affect growth but are not easily measured. When the unobservable country-specific variables are correlated with the included right-hand-side variables, the model can generate misleading results. To address this problem, we could potentially use either fixed-effect or random-effect models. We prefer the fixed-effects approach since the random-effects estimation requires that the omitted variables be uncorrelated with the included right-hand-side variables for the same country, which seems unrealistic in the context of the growth model we consider. We do, however, check if the random-effects version outperforms the fixed-effects version of the model, using the Hausman test. The country-varying variables differ across countries, but are constant within a country over time. To avoid correlation of the unobserved country-specific variables with the included right-hand-side variables, the fixed-effects method adjusts all the model variables by subtracting the mean of each over time. The constant country-specific variables and the intercept then drop out of the estimated equation and therefore the estimated parameters reflect more accurately the effects of the included independent variables. On the other hand, the random-effects model assumes that the country-specific coefficients do not necessarily assume a fixed value but are instead drawn from a random distribution with a constant mean and variance. Further, the country effects on growth are not correlated with the

The European Journal of Development Research

503

Downloaded by [Northwestern University] at 02:41 27 January 2015

right-hand-side variables included in the model. Estimated with a maximum likelihood method, the random-effects model, if true, provides an efficient set of estimates whereas the fixed-effects estimation yields more consistent parameters. 4. Empirical results The fixed-effects estimation of our panel data model yields the results as shown in Table 2. For all four specifications shown, the Hausman test verifies the superiority of the fixed-effects model since the random-effects model is found to be inconsistent. The regressions show a significant overall fit. The R2 for our panel model where cross-sectional units dominate ranges between 0.34 and 0.38 and the F ratio is significant at the 1% level. Between broader (total trade-to-GDP) and narrower (exports-to-GDP) measures of openness, we find better results with the export share of GDP though Table 2 reports the results for both. Further, the political regime variable ( polity) contributes significantly to growth, both in terms of size and statistical significance, with a lag, which seems reasonable. Our main variable of interest however, is international remittances. Columns (2) and (4) of Table 2 exhibit a positive and significant impact of remittances on per capita growth. Holding other variables constant, a one percentage point increase in per capita growth requires on average a 25 times increase in remittances from the sample average of 47.6 international PPP (purchasing power parity) dollars per capita to about 1200 dollars. Since remittances are just about 1.5% of average per capita income in the sample countries, we do not postulate a large contribution of remittances to overall growth. It is, however, evident that not all remittances are spent in consumption but a fraction is saved and invested leading to some impact on long-term growth. As pointed out earlier, it is possible that measured remittances are an underestimate of actual remittances and the underestimates may vary across years. If this is true, the precision with which the remittance coefficient is estimated could be lowered substantially. Table 2.

Fixed-effects estimates (Dependent variable: annual percentage growth of per capita real GDP).

constant yt21 remitt invt opent

(2) 34.8561*** (5.00)*** 2 6.2235*** (2 6.91)*** 0.3935 (1.95)** 2.4252 (2.57)*** 1.5521 (2.05)**

(3) 34.5165*** (5.05)*** 2 6.0094 (2 6.72)*** 0.2779 (1.35) 2.6309 (2.82)*** 1.1437 (1.51)

expgdpt polityt

No. observations No. countries adj. R 2 F

2.0680 (3.21)*** 20.0055 (20.15)

0.0080 (0.22)

polityt21 195 39 0.337 15.35***

(4) 40.4083*** (5.61)*** 26.9027 (27.49)*** 0.4067 (2.13)** 2.2426 (2.56)***

0.0749 (2.09)** 195 39 0.355 16.65***

195 39 0.362 17.14***

(5) 39.753*** (5.58)*** 2 6.6765 (2 7.27)*** 0.2886 (1.45) 2.3379 (2.69)*** 1.7233 (2.7)*** 0.0644 (1.83)* 195 39 0.376 18.19***

Notes: (1) Each specification was also run using random effects. The Hausman test, however, rejects random effects estimation in each case since P(chi-sq . .05) far exceeds the critical chi-sq value. (2) *, **, *** indicate significance at the 1%, 5%, and 10% levels, respectively.

Downloaded by [Northwestern University] at 02:41 27 January 2015

504

G. Pradhan et al.

The initial per capita income is highly significant. The magnitude and sign of the estimated coefficient suggests that the convergence of per capita income, conditional on the determinants of the steady state included in the regression, has been occurring during the sample period at a fairly rapid pace of over 6% a year. While this seems to be much higher than the usual predictions of the neoclassical or endogenous growth theory, it is not too out of line from the usual prediction of panel data models in the literature (see Barro and Sala-i-Martin 2004). Furthermore, reducing observations to five-year averages could to some extent capture the effect of relatively short-term factors that would disappear in the long-run convergence rates. There is one reason, however, why the convergence estimate may look fairly sensible. Our parameter reflects conditional convergence only within the sample of low- and middle-income countries. The exclusion of highincome countries by design may have contributed to a faster convergence rate as well. The investment rate influences output growth positively, as expected. Its coefficient is statistically significant at the 1% level. The size of this coefficient indicates that a 10% increase in the investment rate, such as from the overall mean of 20.10% to 22.11% of output, leads to a 24.3% rise in the change in per capita output. Around the mean value, the output growth would rise from 1.01% to 1.25% per year. This supports the hypothesis of a relatively large contribution of investment to the growth in output and per capita income. In all regressions, including those not reported in Table 2, the investment –output ratio remains highly significant. The openness variable measured by trade –GDP ratio is statistically significant when included along with the contemporaneous political variable, whereas its p-value rises to .133 when openness is clustered with polity lagged one period. The inclusion of lagged polity also lowers the size and significance of the coefficient of remittances. Focusing back on openness, a 10% increase in the trade –GDP ratio (from the average 54.3% to about 60%) can raise average growth by 0.16 percentage points. On the other hand, an equal percentage increase in export share of GDP leads to a slightly higher percentage growth in real GDP per capita. Our last variable in Table 2 is polity, which measures the effect of an increase in the score for democracy relative to autocracy. It fails to provide a significant impact on contemporaneous growth though it matters much more for growth after a period. This result supports the assertion frequently found in the empirical literature that political stability and stronger political institutions provide a conducive environment for sustained growth in the long run. Finally, we note that the way polity enters the empirical model affects the coefficient of remittances. The remittance coefficient in column (3) falls in size from its value in column (2) but more importantly its p-value drops from .04 to .13 causing the coefficient to lose statistical significance even at the 10% level.2 It is possible that once we control for investment as a percentage of GDP and the relevant polity variable, remittances have a greater impact on short-run consumption than on long-run growth. 5. Summary and conclusion This paper has examined the effect of workers’ remittances on economic growth in a sample of 39 developing countries. A standard growth model is estimated in which per capita growth of real GDP depends on remittances per capita, investment, openness, and ‘polity’ which measures the difference in scores between democracy and autocracy. All our variables are measured in five-year periods and we have five such periods for each country from 1980 to 2004 resulting in a panel data of 195 observations. The model is estimated using both fixed-effects and random-effects approaches. The empirical results show a significant overall fit. Remittances have a positive impact on growth although the impact is not very large in size and the coefficient of the remittances variable is significant in two of four specifications. The results are reasonably good although

The European Journal of Development Research

505

Downloaded by [Northwestern University] at 02:41 27 January 2015

we recognize that the official estimates of remittances used in our analysis tend to understate actual numbers. All other variables performed quite well, with the exception of the polity variable for which we get mixed results. However, a higher score on polity as countries get more democratic or less autocratic does raise the rate of economic growth after a period. A continued increase in efforts to measure remittances accurately will probably increase the precision and efficiency with which the long-term effect of remittances on growth can be determined. For individual countries, framing an appropriate manpower policy towards foreign employment of its residents will partly depend on these improvements in data. Notes 1. 2.

To avoid taking logarithms of 0 for some years for some countries, a constant 5 was added to all per capita remittance numbers. The countries that only reported $0 per capita for all or most periods were dropped from the sample. Remittances are not much correlated with either polity or polity lagged one period. The correlation coefficient of remittances with polityt is 20.01 and with polityt21 is þ0.04.

References Adams, R. (1998), ‘Remittances, Investment, and Rural Asset Accumulation in Pakistan,’ Economic Development and Cultural Change, October, 155– 173. ———. (2002), ‘Precautionary Saving from Different Sources of Income: Evidence from Rural Pakistan,’ Policy Research Working Paper 2761. Washington, DC: World Bank. Adams, R., and Page, J. (2003), ‘The Impact of International Migration and Remittances on Poverty,’ paper prepared for DFID/World Bank Conference on Migrant Remittances. London, 9 – 10 October. Adelman, I., and Taylor, J.E. (1990), ‘Is Structural Adjustment with a Human Face Possible? The Case of Mexico,’ Journal of Development Studies, 26, 387– 407. Barro, R., and Sala-i-Martin, X. (2004), Economic Growth, Cambridge, MA: MIT Press. Catrinescu, N., Leon-Ledesma, M., Piracha, M., and Quillin, B. (2005), IZA Discussion Paper No. 2139. http://ssrn.com/abstract ¼ 902342 Chami, R., Fullenkamp, C., and Jahjah, S. (2003), ‘Are Immigrant Remittance Flows a Source of Capital for Development?,’ IMF Working Paper 01/189, Washington, DC: IMF. Desai, M., Kapoor, D., and McHale, J. (2001), ‘The Fiscal Impact of the Brain Drain: Indian Emigration to the U.S.,’ Weekly Political Economy Discussion Paper, Harvard University. www.wcfia.harvard.edu/ seminars/pegroup. Edwards, A., and Ureta, M. (2003), ‘International Migration, Remittances, and Schooling: Evidence from El Salvador,’ NBER Working Paper No. w9766, June. Faini, R. (2001), Development, Trade, and Migration, Washington, DC: IMF. International Monetary Fund (IMF) (various years), International Financial Statistics Yearbook, Washington, DC: IMF. International Monetary Fund (IMF) (2005), World Economic Outlook, Washington, DC: IMF. Jongwanich, J. (2007), ‘Workers’ Remittances, Economic Growth and Poverty in Developing Asia and the Pacific Countries,’ UNESCAP Working Paper WP/07/01, Bangkok: ESCAP. Kapur, D., and McHale, J. (2005), Give Us Your Best and Brightest: The Global Hunt for Talent and Its Impact on the Developing World, Washington, DC: Center for Global Development. Mankiw, N.G., Romer, D., and Weil, D. (1992), ‘A Contribution to the Empirics of Economic Growth,’ Quarterly Journal of Economics, 102, 407– 437. McCormick, B., and Wahba, J. (2002), ‘Return Migration and Geographical Inequality: The Case of Egypt,’ unpublished paper, University of Southampton. Ozden C., and Shciff M. (eds.) (2005), International Migration, Remittances, and the Brain Drain, Washington, DC: World Bank Publications. Perkins, D., Radelet, S., Snodgrass, D., Gillis, M., and Romer, M. (2001), Economics of Development (5th ed.), New York: W.W. Norton & Company. Ratha, D. (2003), ‘Workers’ Remittances: An Important and Stable Source of External Development Finance,’ Global Development Finance, World Bank, 157– 175 ———. (2004), ‘Why Should We Care About Workers’ Remittances? And What Should We Do?,’ Expert Meeting of the Global Commission on International Migration. Geneva, 20– 21 January.

Downloaded by [Northwestern University] at 02:41 27 January 2015

506

G. Pradhan et al.

Stahl, C., and Arnold, F. (1986), ‘Overseas Workers’ Remittances in Asian Development,’ International Migration Review, 20, 4, 899– 925. Stark, O., and Levhari, D. (1982), ‘On Migration, Risk in LDCs,’ Economic Development and Cultural Change, 31, 1, 191– 196. Stark, O., and Lucas, R. (1988), ‘Migration, Remittances and the Family,’ Economic Development and Cultural Change, 36, 3, 465– 481. Taylor, J.E. (1992), ‘Remittances and Inequality Reconsidered: Direct, Indirect and Intertemporal Effects,’ Journal of Policy Modeling, 14, 2, 187– 208. ———. (1999), ‘The New Economics of Labor Migration and the Role of Remittances,’ International Migration, 37, 1, 63 – 86. University of Maryland, Polity IV. www.cidcm.umd.edu/inscr/polity/, accessed February 2005. Warner, A. (2003), ‘Once More into the Breach: Economic Growth and Integration,’ Center for Global Development Working Paper No. 34. World Bank (2006), Global Economic Prospects, Washington, DC: World Bank. World Bank (2007a), Global Economic Prospects, Washington, DC: World Bank. World Bank (2007b), World Development Indicators CD-ROM, Washington, DC: World Bank.