1
REMITTANCES, GROWTH AND FINANCIAL SYSTEM DEVELOPMENT A comparative regression analysis
Frank van Roest S1344315
frankvanroest@gmail.com
University of Groningen Faculty of Economics
2
Abstract
The rates and levels of international migrants and corresponding remittances to their countries of origin have increased enormously since the last two decades. The aggregate impact of these remittances has only since recently been picked up by mainstream economic research. This study tries to find aggregate effects of remittances on economic growth in developing countries on the long run, as well as their effect through financial system development. In order to do so, a panel data set containing 30 developing countries (the highest relative recipients of remittances) over the period 1995-2009 is used. The results show that remittances have a positive, but only sometimes significant, effect on growth. Moreover, this study has shown that the impact of remittances through financial system development points to a positive conditional relationship, but remains to be ambiguous due to mixed and insignificant results.
Keywords
Remittances Growth
3
1. Introduction
Since the Second World War, a great deal of attention within the field of economics has been on development economics. This field mainly focuses on how to promote economic growth in developing countries. These countries often are recipients of foreign money flows. A large share of research considering economic growth in developing countries pays attention to either foreign direct investment (FDI) flows or foreign aid. With regard to developing countries, especially the last one has been under scrutiny. The effects of foreign aid on economic growth have been dealt with in many respects. According to Arndt, Jones and Tarp (2010) there is a positive and significant causal relationship between foreign aid and growth in the long run. A very influential paper by Craig Burnside and David Dollar (2000) basically states that growth by aid in developing countries is conditional on good policy, i.e. foreign aid leads to growth in countries with a good political and legal structure. Easterly (2003) and Easterly, Levine and Roodman (2004) counter this statement using the exact same statistical methods but with and extended data-set, leading to different results (no aggregate relationship between aid and growth). Nonetheless, there is no general consensus on this relationship. Meanwhile, a third flow of money from developing to developed countries has not received as much attention as did foreign aid. This third flow of money considers remittances; money transfers from migrants back to their country of origin.1 One could argue that the reason for this limited research is due to conventional wisdom. That is, the motivation of a remittance sending party is mostly altruistic. As a consequence, the use of remittances by the receiving party is mostly consumption-driven, as opposed to investment-driven money flows like FDI. Although it can be argued that an increase in national disposable income, and consequently an increase in consumption, can trigger growth from a demand perspective, it is also possible to have negative effects on growth, for example through inflation. It is likely that investment driven money flows like FDI yield better results in terms of long term growth than consumption-driven money flows like remittances, since investment (in growth channels like fixed capital and human capital) usually leads to higher productivity and subsequently to higher growth rates.
1
4
More recently the impact of remittances on economic growth has gained more attention, and with good reason. The growth of remittances in the last decade has been enormous. In 2004, total remittances to developing countries already amounted to $125 billion, exceeding foreign aid by 50%.2 According to the Migration and
Remittances Factbook 2011 of the World bank, 6 years later, worldwide remittances
in 2010 amounted to $440 billion.3 From this amount, a total of $325 billion was received by developing countries, sometimes exceeding FDI inflows (the actual amount of remittances is believed to be much higher because not all transactions are actually recorded). Compared to 2009, total remittances have grown by 6% in 2010. Despite the global financial crisis, in 2009 remittance flows only fell by 5,5%. By contrast, FDI flows fell by 40%. These figures show that the relative importance of remittances has increased enormously in the last one or two decades and that due to increasing migration of people around the globe these figures will only continue to rise.
Figure 1: Remittances compared to with other resource flows, 1991-2010
US$ billions
Source: Migration and Remittances Factbook 2011, p17 FDI = foreign direct investment; ODA = official development
2
5
Earlier studies on the impact of remittances are mainly micro-studies focusing on one country or a region. These micro studies show very mixed results. Some of these studies focus on poverty alleviation, whereas others focus on economic growth, varying from a positive impact to a negative or no impact on economic growth at all. However, there is only a small amount of studies focusing on the impact of remittances on growth on a macro level. Since remittance flows have recently increased very quickly, and are most likely to grow even more in the future, it is to be expected that their impact will increase as well. Therefore, the aggregate impact of remittances on economic growth deserves more attention, especially with respect to developing countries. However, those studies which do maintain a macro approach, show rather mixed results as well. These mixed results could be explained by the ambiguous nature of remittances; remittances are mostly consumption-driven, rather than investment-driven. However, when remittances are used for investment purposes, they could spur growth. Therefore, it is likely that remittances lead to higher growth rates in countries with a higher level of financial development. As a better developed financial system usually funnels money flows into growth channels through savings and investment.
The purpose of this paper is to contribute to research on the aggregate effects of remittances on economic growth in developing countries in the long run, as well as their effect though financial system development. The contribution of this paper lies in its unique dataset as well as in its approach. The dataset contains the 30 highest relative recipients of remittances over the last 15 years, since the impact of remittances is expected to be the highest in this set of countries over this period. Moreover, this study does not solely endeavor to explore the impact of remittances, but also its impact through a country’s financial system development. To address the abovementioned issues, section 2 contains background information as well as a literature review on the relation between remittances, financial development, and economic growth. Section 3 elaborates on the methods and data. The results will be dealt with in section 4, and the last section concludes.
3
6
2. Background and Framework for Analysis
2.1 Background
Although research on remittances has increased in the last decade, results are rather mixed. Only the most recent studies use quantitative data and econometric techniques to measure the impact of remittances on economic growth, whereas the bulk of relevant literature focuses on poverty alleviation and income distribution (Glystos, 2002, p3). Most research on remittances are micro studies rather than macro studies or focus only on a particular set of countries. The different foci (and corresponding methods) on effects of remittances partly explain the mixed results. For example, Giuliano and Ruiz-Arranz (2009) and Catrinescu et al. (2008) show contradictory findings, but mainly claim that the success of remittances, in terms of growth, depends on a country’s political and economic policies and institutions. Glystos (2002), focusing on changes of remittances in Mediterranean countries, finds that the good done by rising remittances is not as great as the harm done by falling remittances. Rapoport and Docquier (2005) find an overall positive relationship between remittances and long-run economic performance. Finally, Chami, Fullenkamp and Jahjah (2005) find a negative correlation between remittances and GDP growth, because remittances, contrary to FDI, are not profit-driven. In general, one could argue that, on the positive side, remittances can contribute to the alleviation of poverty (Jongwanich, 2007) and are an important source for covering a country’s balance of payments (BOP) deficits. Furthermore, remittances can sometimes provide funds for families for savings (and investment). However, on the negative side, remittances can lead to inflation, or can be a disincentive for people to go to work and instead live of these remittances. Moreover, large inflows of money by remittances can lead to an appreciation of the real exchange rate, disadvantaging the tradable goods sector, leading to less exports (Dutch disease) and the migration of people (as a condition for remittances) can lead to a brain-drain.
7
8
often the case for people living in rural areas, where income can be very volatile due to seasonal differences. Some members of a community get send to another country to diversify the sources of income. This diversification then functions as a form of insurance. Nevertheless, the bulk of remittances have an altruistic cause (Chami et al. 2005, p59), and none of these motivations seem to be investment driven, also partly explaining the resilience of remittances during economic downturns. The majority of literature claims that remittances received are mainly consumption-driven, i.e. are spend on final goods. This implies that remittances are different by nature in comparison to FDI, which is investment-driven, and can have different impacts on economic growth.
2.2 Theoretical Framework for Remittances and Growth
9
supply. This is mainly emphasised by Chami et al. (2005). Most remittance transactions take place under asymmetric information, leading to moral hazard problems. The recipient of remittances is very likely to be reluctant in participating in the labour market. In their macro study, Chami et al. find a negative correlation between remittances and economic growth.
In addition, it is argued that remittances in order to enhance economic growth depend on some conditional factors. Catrinescu et al. (2008) conduct a macro study in which they investigate the hypothesis that the impact of remittances on long run growth is partly determined by the receiving country’s institutional quality. Their results show that remittances are more likely to enhance growth in countries which have higher developed institutions and policies. Furthermore, Giuliano and Ruiz-Arranz (2009) find that remittances boost growth in developing countries with a less developed financial system, hence, working complementary to the financial sector, especially when the financial sector does not meet credit needs. However, Aggarwala, Demirgüç-Kunt and Soledad Martínez Pería (2010) find a significant and positive link between remittances and financial system development, which in turn leads to economic growth. Finally, results of Mundaca (2009) show that improving the financial sector leads to better use of remittances. In general, although heavily debated among economists, there is evidence that financial system development leads to economic growth. Schumpeter (1912) already acknowledged the importance of the financial system for economic development. More recently, Levine (2004, p890, in: Handbook of economic growth, 2005) shows evidence for the impact of financial system development on growth. Levine claims in his work that “theoretical models
show that financial instruments, markets, and institutions may arise to mitigate the effects of information and transaction costs. In emerging to ameliorate market frictions, financial arrangements change the incentives and constraints facing economic agents. Thus, financial systems may influence saving rates, investment decisions, technological innovation, and hence long-run growth rates.” The same
author suggests that there are five main functions of the financial system in relation to growth:
10
• Facilitate the trading, diversification, and management of risk • Mobilize and pool savings
• Ease the exchange of goods and services
Although all of these five functions are important for economic growth, it is the fourth function, mobilizing and pooling savings, which is of particular interest for the impact of remittances on growth. When remittances are not used for direct consumption, they are most likely used for savings (for future consumption rather that investment). When a financial system is more effective at pooling and allocation these savings, as is argued by Levine, than economies of scale can be exploited. Accumulated savings can overcome investment indivisibilities, can improve resource allocation and can boost technological innovation. Moreover, access to multiple investors (by accumulated savings) can give rise to larger projects, for example infrastructure projects. All these factors can have positive influence on growth. It can therefore be argued that remittances have an increasing positive impact on growth when financial system development improves.
11
3. Model, Methods and Data
3.1 Model
In this section, two different growth models are constructed to test the hypothesis. Moreover, the second part of this section contains a motivation and explanation for the used methodology. The framework presented below is a simple neoclassical growth model allowing for remittances and financial development indicators to be included, based on Barro (1996). Similar neoclassical models are used by Jongwanich (2007), Catrinescu et al. (2008), Pradhan et al. (2008) and Giuliano and Ruiz-Arranz (2009).
git = β0 + β1Yi,t-1 + β2humit + β3gcgfit + β4remit + β5Xit + εit (1)
In this model, git represents annual percentage GDP per capita growth. The subscript i
denotes the country. Furthermore, the subscript t denotes the time in years. Yi,t-1
represents a lagged variable for GDP per capita at time t-1. This means that growth in one year is explained by growth in the previous year. The reason to include this variable is that growth theory predicts that developing countries, so countries starting with a relative low per capita income, tend to grow relatively fast compared to developed countries (convergence theory, also see Pradhan et al. 2008, p 501). Hence, the coefficient for this variable is expected to be negative, since the convergence theory predicts diminishing returns to capital. Next, the variable humit stands for
human capital, and is represented by the gross ratio of secondary school enrollment. Gross enrollment ratio is the ratio of total enrollment, regardless of age, to the population of the age group that officially corresponds to the level of education shown (secondary).4 Initially it was considered to include the literacy rate of adults (older than 15 years) as a proxy for human capital next to secondary school enrollment, but this variable is left out because of very poor data availability for the countries in this dataset. In addition, gcgfit is the gross capital formation as a share of GDP and works
as a proxy for investment. The coefficients of both human capital and gross capital formation are expected to have a positive sign. Furthermore, to incorporate
4
12
remittances in the equation, remit represents remittances over GDP.5 In this first
regression, the coefficient for remittances is expected to be ambiguous (and most likely insignificant). Alternatively, Chami et al. (2005) suggest that instead of using remittances over GDP, to use the percentage change in remittances (Δremit), since this
would capture the dynamic nature of capital flows. However, according to Catrinescu et al. (2008), this would not be an appropriate variable to explain growth, because remittances would need to increase every year to promote growth, eventually to a limit of 100%. Therefore, the level of remittances is chosen as explanatory variable, rather than the growth of remittances. Xit is a series of control variables. Finally, εit
captures the error term. The control variables used in both models are inflation,
government consumption, population growth and openness. Inflation refers to annual
percentage change in the consumer price index. According to Jongwanich (2007, p8), higher inflation rates are an indicator for macro-economic instability, which reflects symptoms of weakness. This instability can in turn lead to a hampering in savings and investment decisions. Therefore, it is expected that the coefficient for inflation is negative. Moreover, government consumption refers to general government final consumption as a percentage of GDP. Government consumption usually is non-productive. Consequently, it is expected that the coefficient for government spending is also negative. Furthermore, the annual population growth rate for year t is the exponential rate of growth of midyear population from year t-1 to t, expressed as a percentage. Population is based on the de facto definition of population, which counts all residents regardless of legal status or citizenship--except for refugees not permanently settled in the country of asylum, who are generally considered part of the population of the country of origin. There is no general consensus on the economic effects of population growth. However, it could be argued that population growth leads to an increase in the amount of economic agents. This in turn can lead to more consumption and more value creation and consequently to higher levels of growth. Moreover, it could be argued that higher population leads to higher levels of urbanization (the more people there are, the bigger the hubs will be). For example Henderson (2005) argues that in cities more innovation takes place due to knowledge sharing and spillovers, and consequently that higher rates of urbanization lead to more economic development (and higher economic growth rates).Therefore, the coefficient
5
13
for this variable is expected to be positive. Finally, openness refers to trade openness i.e. exports plus imports as a share of GDP. Since higher trade openness usually leads to more efficient allocation of resources, more competition and consequently more innovation and entrepreneurship, it is expected that openness leads to more economic growth (a positive coefficient).
The second equation allows for the level of development of a country’s financial system to interact with remittances and is presented below.
git = β0 + β1Yi,t-1 + β2humit + β3gcfit + β4remit + β5findevit (2)
+β6(remit ∙ findevit) + β7Xit + εit
The former model does not capture any interaction effects between a country’s financial system development and remittances, whereas the latter one does. The variable findevit represents a country’s financial development. There is no general
consensus on how financial development can be defined, although three proxies are widely used throughout a lot of studies as financial development indicators and are directly used in this paper (Levine, 2004, p890, in: Handbook of economic growth, 2005): DEPTH, BANK and PRIVY. DEPTH, which measures the size of financial intermediaries, equals currency plus demand and interest-bearing liabilities of banks and nonbank financial intermediaries divided by GDP (also referred to as M2/GDP). Furthermore, BANK, which measures the relative degree to which the central bank and commercial banks allocate credit, equals the ratio of bank credit divided by bank credit plus central bank domestic assets. Finally, PRIVY equals domestic credit to private sector divided by GDP (the more credit is allocated to private firms, the more financial intermediaries are engaged in researching firms, risk management, mobilizing savings, etc.). Similar models are also used in Giuliano and Ruiz-Arranz (2009) and Mundaca (2009). The coefficient of findevit is expected to be positive,
since a better developed financial system usually leads to more savings and investment and, subsequently, to economic growth. Of particular interest for this study is the value of β6, since this captures the conditional effects of remittances and
financial system development. The coefficient of this interaction term (remit ∙ findevit )
14 3.2 Data on Remittances
This paper examines the aggregate effects of remittances on growth. Thus, the level of analysis is on a global level. Therefore, the panel data of this study contains aggregate country information. In total 30 countries are included in the dataset over a period of 15 years (1995-2009). Table 3 in appendix III provides a list of the countries used in this research in alphabetical order. These countries are the highest relative recipients of remittances (over GDP), for 2009 and are all low- or middle-income countries. These countries are deliberately chosen, for it is to be expected that those countries experience the highest impact of remittances. Moreover, only in the last one or two decades the amount of (officially recorded) remittance flows have increased significantly. Moreover, there is a higher data availability and higher data accuracy for later years. Therefore, a period of 15 years is chosen, for single year periods as well as 3 and 5 year averages. Including data from the 1970s, for example, would most likely result in a very unbalanced panel due to missing values and also captures a period in which remittances flows were relatively low.
Data on remittances are derived from the World Development Indicators (WDI). 6 Since remittances is a collective term for different kinds and origins of money flows, it is important to give a comprehensive definition of remittances, also used by the IMF’s Balance of Payments (BoP) Manual (2010): ‘Remittances represent household
income from foreign economies arising mainly from the temporary or permanent movement of people to those economies. Remittances include cash and noncash items that flow through formal channels, such as via electronic wire, or through informal channels, such as money or goods carried across borders.’ Remittances arguably
constitute of two main components. These are: compensation of employees and
worker’s remittances.7 Both these components are recorded in the current account. Compensation of employees refers to income earned by short-term workers in an economy where they are non-residents (for example seasonal workers) for less than a
6
The online version of the WDI can be found on http://data.worldbank.org/data-catalog/world-development-indicators/wdi-2011, the databank can be found on
http://databank.worldbank.org/ddp/home.do?Step=12&id=4&CNO=2
7
15
year. If migrants live in a host country for more than a year, regardless of their immigration status, they are considered residents. Transfers from residents back to their country of origin are considered workers’ remittances.8 Although remittances conceptually also include cash and other items sent through informal channels, these are not recorded in the data (for example Catrinescu et al. (2009) claim that actual data on remittances is very poor, since large quantities of international remittances are sent through informal channels such as ‘hawala’ service providers).9 Additionally, the aforementioned two components which constitute remittances could be extended with money flows resulting from social benefits (social security funds and pension funds) or transfers to NPISHs to account for total remittances, but this would also be inconsistent with the data used in this paper.10 Consequently, the definition of remittances used in this paper is the one given by the IMF’s BoP Manual consisting of compensation of employees and worker’s remittances and the one used in the WDI.
Data for all other variables are derived from the WDI as well, apart from the variable BANK, used as one of the indicators for financial development. Data for the BANK variable (this measures the relative degree to which the central bank and commercial banks allocate credit) is derived from the dataset compiled by Beck, Demirgüç-Kunt and Levine (2010).11
3.3 Endogeneity and Conditionality
A problem widely addressed in studies with respect to remittances related to economic growth, is the problem of endogeneity i.e. two-way causality between remittance flows and the level of economic growth. A possible reason for this is that low economic growth (mainly in developing countries) leads to an increasing amount of migrants. These migrants then in turn remit money to their country of origin.
8
For a complete and even more comprehensive definition, see Appendix 5 of the IMF’s Balance of Payments Manual, 2010, page 272-277. The definition for remittances used in this paper is referred to as personal remittances in the BoP Manual.
9
For more information about ‘hawala’ as an alternative remittance system see for example: http://www.interpol.int/public/financialcrime/moneylaundering/hawala/default.asp
10
NPISH stands for non-profit institutions serving households, these include for example charity organizations, religious organizations, member clubs and trade unions
11
The dataset provided by Beck, Demirgüç-Kunt and Levine can be found on
16
17
endogeneity problem. Consequently, in this study there is opted for a conditionality approach to minimize the effects of endogeneity.
3.4 Methods
18
19
4. Empirical Results
This section contains the results of the comparative regressions. In total 3 sets of regressions are ran. First, all the used tests will be discussed, justifying the used prediction methods. Secondly, the results of the regressions will be discussed in section 4.2 and are shown in appendix IV.
4.1 Tests
The first set of regressions, one with and one without the interaction variables, uses single year data. The second set of regressions uses 3 year averages to reduce possible autocorrelation due to business cycles and the third uses 5 year averages. The summary statistics of the 3 sets of regressions are shown in tables 6, 7 and 8 in appendix IV. A correlogram of all the explanatory variables for these sets is shown in appendix VI. For all of the in total 6 regressions, a Hausman test has been conducted. Except for the first regression for single year data, the Hausman test showed out that the random effects model is inconsistent, with respective chi-squared values (and p-values) of 12.20(0.1425), 34.06(0.0007), 12.40 (0.0882), 33.69 (0.0008), 51.31(0.0000) and 31.31(0.0030). This implies that the fixed effects model outperforms the random effects model, as was expected. Running heteroscedasticity tests (Breusch-Pagan and White) for all datasets, shows out that for both datasets, H0
is rejected (for both tests and both datasets, H0 is rejected at a 5% significance level).
20
do not vary greatly per used dataset. Finally, it should be mentioned that the performed tests on heteroscedasticity and normality are ran on ordinary least squares regressions, because these test on a fixed-effect model are too complex for this study.
4.2 Results
The values of all coefficients of the single year regressions, 3 year average regressions and 5 year average regressions are presented in tables 3, 4 and 5, respectively, in appendix VI. Of particular interest for this study are the coefficients for remittances (β4) and the coefficients for the conditional effects of remittances interacted with the
financial development indicators (β6) of equation (1) and (2).
21
school enrollment most likely pays of years or even decades later. Secondly, data on school enrollment in developing countries might be poor. Finally, the quality of secondary school in developing countries can be very low.
With regard to the coefficients for the conditional effects of remittances interacted with the financial development indicators (rem_M2, rem_DomCred and rem_Bank), it is seen that for the first two indicators DEPTH (rem_M2) and PRIVY (rem_DomCred) the coefficients are adjacent to nil. The third indicator BANK (rem_Bank) has a positive, but insignificant, coefficient. Although in total these indicators point to a positive conditional relationship, these results are rather ambiguous as well. And, therefore, it is hard to draw any valuable conclusions based on these results. It was expected that there would be a positive relationship for all three indicators, but the regressions show out differently. One good reason for the ambiguous outcomes of the conditionality effects is the potentially false assumption that the used indicators for financial development (DEPTH, PRIVY and BANK) lead to economic growth. The results show this is most certainly not true for every case. More often than not the coefficient for these indicators is negative, rather than positive. The negative (but mostly insignificant) values of these indicators imply that financial system development does not necessarily lead to economic growth in the first place, partly explaining the ambiguous results of the interaction terms. Additionally, there might be another good explanation for these outcomes. Remittances can be successful conditional on the financial sector, but the discussion considering this success revolves around remittances to be either supplementary or complimentary to the financial system. For example, it can be argued that remittances can be useful as a substitute to a poorly operating financial system for providing funds for investment opportunities. This would mean that the impact of remittances becomes bigger, the less a country’s financial system is developed i.e. the coefficient measuring conditionality (β6) in this case is expected to be negative (this is for
22
are large inequalities. Not just between social classes, but also between regions. This could for example mean that in large economic hubs, like cities, with better access to the financial system, remittances can work complementary to the financial system. Meanwhile, in poorer and usually more rural areas, people have less access to the possibilities a proper financial system provides. In this case it is very likely that remittances work supplementary to the financial system. This illustrates that both phenomena can operate in the same time in one country, automatically blurring the results.
Furthermore, taking a look at the rest of the results, we see that the coefficient for the lagged variable (lag1_GDPpercapita) is practically zero (and insignificant), implying that the results are inconsistent with the neo-classical convergence theory. Moreover, inflation as well a general government consumption (Infl and GovCons) have a slightly negative effect (or no effect), mostly significant in the regressions for 3 and 5 year averages. This is as expected. In contrast to what was expected, it can be seen that the coefficient for population growth (PopGrowth) is negative with large coefficients. It was predicted from theory that higher growth rates for population leads to higher economic growth rates. This might be the case for more developed countries with high quality institutions, but it is obvious from the results in this study that this is not the case for developing countries. Finally, trade openness (Trade) has a positive, sometimes significant, effect on growth, as was expected.
23
5. Discussion, limitations and Conclusions
This section contains a discussion and limitations of the used methods and results and ends with the conclusions.
5.1 Discussion and Limitations
24
remittances through institutional quality, although this impact is very likely to exist (see for example Catrinescu et al., 2009). Moreover, the indicators used to reflect financial system development were believed to spur economic growth, which is not always the case according to the results of this study. Thirdly, it could be argued that different estimation techniques can be used, resulting in different outcomes. Authors like Giuliano and Ruiz-Arranz and Catrinescu et al. use more sophisticated methods like the Generalized Method of Moments (GMM)12 regressions to estimate the impact of remittances. However, using these more sophisticated and complex ways of estimation is beyond the scope of this study. Comparing the (in essence rather weak) results of this study with other studies, it can be seen that the implied positive (or at least not negative) relation between remittances and growth is in line with most studies (Guiliano and Arranz, 2009, Pradhan et al., 2008, Jongwanich, 2007). However, Chami et al. (2005) found a negative relation. This difference can be explained by the fact that Chami et al. use a different dataset, variables and model. Their dataset contains 113 countries for the period 1970-1998. Remittances were much lower in this in this period than for example in the last decade. Moreover, using a dataset of 113 countries, total remittances tend to be much lower as well, since countries receiving relatively low amounts of remittances are included as well. Furthermore, Chami et al. explain economic growth by using remittance growth, rather than remittances over GDP. Most importantly, they use a model to test whether remittances flows act like usual capital flows (like FDI), taking into account that remittances can compensate for bad economic outcomes. This leads to a negative relationship between remittances and growth. Using this different approach evidently leads to different outcomes. If the results in terms of the conditional effects of this study are compared to relevant other studies, in particular by Mundaca (2009) and Guiliano and Arranz (2009), it appears that these are in line with those of Mundaca, who also finds a positive relation between remittances and growth through financial system development, but opposite to Guiliano and Arranz. Possible reasons for this have been discussed earlier in section 4.2. Next to these reasons, the differences can be explained by using a different dataset, containing over 100 countries over the
12
25
period 1975-2002. As argued before for the (quite similar) dataset used by Chami et al., this can lead to different results.
5.1 Conclusions
26
References
Adams R.H. Jr., ‘International Remittances and the Household: Analysis and Review of Global Evidence’, Journal of African Economies, Vol.15, 2006, pp.396-425
Aggarwal R, A. Demirgüç-Kunt and M. Soledad Martínez Pería, “Do remittances promote financial development?”, Journal of Development Economics, 2010
Arndt C., S. Jones and F. Tarp, “Aid, Growth, and Development”, World Institute for
Development Economics Research, Working Paper No. 2010/96, September 2010
Barajas A., R. Chami, C. Fullenkamp, M. Gapen and P. Montiel, “Do Workers’ Remittances Promote Economic Growth?”, IMF Working Paper 09/153, July 2009
Barro R.J., “Determinants of Economic Growth: a Cross Country Empirical Study”,
National Bureau of Economic Research (NBER) Working Paper 5698, August 1996
Beck T., A. Demirgüç-Kunta and R. Levine "Financial Institutions and Markets Across Countries and over Time: Data and Analysis", World Bank Policy Research
Working Paper No. 4943, May 2009, pp. 597-605
Brown, R.P.C., “Migrant’s Remittances, Savings and Investment in the Pacific”,
International Labour Review, Vol.133, 1994, pp. 1–19
Burnside C. and D. Dollar, “Aid, Policies, and Growth.”, American Economic Review, September 2000, 90(4), pp. 847-868
Catrinescu N. M. Leon-Ledesma, M. Piracha and B. Quillin, “Remittances, Institutions, and Economic Growth”, World Development, 37(1), 2009, pp. 81-92
27
Clarke K.A., “The Phantom Menace: Omitted Variable Bias in Econometric Research Conflict Management and Peace Science”, Conflict Management and Peace Science, Vol. 22, 2005, pp. 341-352
Easterly W., R. Levine and D. Roodman, “Aid Policies, and Growth: Comment”,
American Economic Review, June 2004, 94(3), pp. 774-780
Easterly W., “Can Foreign Aid Buy Growth?”, Journal of Economic Perspectives, Summer 2003, 17(3), pp. 23-48
Funkhauser, E., “Remittances from international migration: a comparison of El Salvador and Nicaragua”, Review of Economics and Statistics, 77(1), 1995, pp. 137-146
Giuliano P. and M. Ruiz-Arranz, “Remittances, financial development, and growth”,
Journal of Development Economics, Vol. 90, 2009, pp. 144-152
Glytos, N.P., “Dynamic Effects of Migrant Remittances on Growth: An Econometric Model with an Application to Mediterranean Countries”, Centre of Planning and
Economic Research, No.74, April 2002, pp. 1-41
Hansen H. and F. Tarp, “Aid and growth regressions”, Journal of Development
Economics, 2001, Vol. 64, pp. 547-570
Henderson J.V., “Urbanization and Growth”, in: P. Aghion and S.N. Durlauf, “Handbook of Economic Growth 1A”, North Holland, 2005, pp. 1543-1591
IMF Balance of Payments and International Investment Position Manual,
International Monetary Fund, Sixth Edition, 2010
Jonwanich J., “Workers’ Remittances, Economic Growth and Poverty in Developing Asia and the Pacific Countries”, United Nations Economic and Social Commission for
28
Levine, R. “Finance and Growth: Theory and Evidence”, in: P. Aghion and S.N. Durlauf, “Handbook of Economic Growth 1A”, North Holland, 2005, pp. 865-934
Lucas R.E.B. and O. Stark, “Motivations to remit: Evidence from Botswana”, Journal
of Political Economy, Vol. 93, 1985, pp. 901-918
Mundaca B.G., “Remittances, Financial Market Development, and Economic Growth: The Case of Latin America and the Caribbean”, Review of Development Economics,
13(2), 2009, pp. 288-303
Pradhan G., M. Upadhyay and K. Upadhyay, “Remittances and economic growth in developing countries”, The European Journal of Development Research, Vol. 20, No.3 September 2008, pp. 497-506
Rapoport H. and F. Docquier, “Economics of Migrants’ Remittances”, Institute for
the Study of Labor (IZA), Discussion paper No. 1531, March 2005
Ratha, D., ‘Worker’s Remittances: An Important and Stable Source of External Development Finance’, Global Development Finance, World Bank, 2003
“Theorie der Wirtschaftlichen Entwicklung. Leipzig: Dunker &
29
APPENDIX I: Top Remittance-Receiving Countries
Figure 2: Top Remittance-Receiving Countries, 2010, US$ billions
30
Figure 3: Top Remittance-Receiving Countries, 2009, percentage of GDP
31
APPENDIX II: Migration figures
Table 1: Estimated number of international migrants at mid-year for 1990-2010
Year Estimated number of international migrants at mid-year 1990 155 518 065 1995 165 968 778 2000 178 498 563 2005 195 245 404 2010 213 943 812
Table 2: International migrants as a percentage of the population for 1990-2010
Year International migrants as a percentage of the population 1990 2.9 1995 2.9 2000 2.9 2005 3.0 2010 3.1
Source: United Nations, Department of Economic and Social Affairs, Population Division (2009). Trends in International Migrant Stock: The 2008 Revision (United Nations database,
32
APPENDIX III: Alphabetical list of countries
Table 3: Alphabetical list of countries used in the regressions (30 in total)
Albania Armenia Bangladesh
33
APPENDIX IV: Results
Table 3: Single year regression results
(1) (2)
VARIABLES GDPGrowth GDPGrowth
lag1_GDPpercapita -0.000357 -5.85e-05 (0.000441) (0.000283) GFCF 0.0837* 0.0906 (0.0471) (0.0604) SchoolEnr -0.0705 -0.0855 (0.0598) (0.0788) Rem 0.0490 0.0769 (0.0529) (0.216) Infl 0.00377 0.00784 (0.0226) (0.0262) GovCons -0.0845 -0.00376 (0.0736) (0.102) PopGrowth -2.305** -2.495** (1.118) (0.955) Trade 0.0283 0.0486 (0.0281) (0.0324) M2 0.158 (0.109) DomCred -0.267* (0.150) Bank -0.868 (5.629) rem_M2 -0.0108 (0.00647) rem_DomCred 0.0105 (0.00730) rem_Bank 0.150 (0.240) Constant 8.764** 8.363 (3.559) (4.990) Observations 248 241 R-squared 0.082 0.130 Number of countrynmbr 28 27
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Y1-1 represents a lagged variable for GDP per capita at time t-1; GFCG represent gross fixed capital
34
Table 4: 3 year average regression results
(1) (2)
VARIABLES GDPGrowth GDPGrowth
lag1_GDPpercapita -0.000148 -0.000264 (0.000553) (0.000527) GFCF 0.00261 -0.0189 (0.0570) (0.0799) SchoolEnr -0.0678 -0.105 (0.0496) (0.0665) Rem 0.143* 0.147 (0.0826) (0.210) Infl -0.00599*** -0.00606*** (0.00195) (0.00217) GovCons -0.102 0.00227 (0.0997) (0.119) PopGrowth -2.296 -1.937 (1.418) (1.199) Trade 0.0465* 0.0545* (0.0227) (0.0303) M2 0.161 (0.117) DomCred -0.0795 (0.164) Bank -8.323 (4.935) rem_M2 -0.00536 (0.00611) rem_DomCred 0.00339 (0.00845) rem_Bank 0.201 (0.228) Constant 7.219* 9.277* (3.991) (5.085) Observations 101 99 R-squared 0.237 0.310 Number of countrynmbr 28 27
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Y1-1 represents a lagged variable for GDP per capita at time t-1; GFCG represent gross fixed capital
35
Table 5: 5 year average regression results
(1) (2)
VARIABLES GDPGrowth GDPGrowth
lag1_GDPpercapita 0.000120 -3.17e-05 (0.000526) (0.000516) GFCF 0.0287 -0.0387 (0.0818) (0.141) SchoolEnr -0.0656** -0.0391 (0.0288) (0.0282) Rem 0.0580 0.285 (0.0893) (0.286) Infl -0.00943*** -0.00996*** (0.00293) (0.00278) GovCons -0.223* -0.321 (0.130) (0.228) PopGrowth -3.462* -2.292 (1.829) (1.920) Trade 0.00483 0.0136 (0.0209) (0.0288) M2 -0.0375 (0.105) DomCred 0.207 (0.216) Bank -8.263 (6.302) rem_M2 0.00140 (0.00405) rem_DomCred -0.00941 (0.00955) rem_Bank 0.0571 (0.452) Constant 14.06*** 14.98*** (3.836) (5.256) Observations 69 68 R-squared 0.294 0.384 Number of countrynmbr 28 27
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Y1-1 represents a lagged variable for GDP per capita at time t-1; GFCG represent gross fixed capital
36
APPENDIX V: Summary Statistics
Table 6: Single year summary statistics
rem_Bank 436 10.22302 7.696522 .0363559 39.22383 rem_DomCred 446 435.3378 452.237 .3909897 2878.988 rem_M2 444 682.2211 780.5515 .8972667 5298.37 M2 460 47.15482 35.24031 6.06028 228.407 Trade 454 87.89538 36.08971 22.86587 222.882 PopGrowth 480 1.253617 1.223909 -6.034338 3.889371 GovCons 441 14.42317 7.036395 4.36433 50.36216 Infl 433 19.46292 213.8278 -100 4447.866 Rem 452 12.83637 9.216827 .0581861 50.41634 SchoolEnr 308 66.43439 24.8291 14.52556 114.0797 GFCF 418 23.15758 8.732744 -11.67995 76.69301 lag1_GDPpe~a 476 1567.155 1434.801 139.408 8175.142 GDPGrowth 479 4.036098 6.389053 -30.9 88.95766 Variable Obs Mean Std. Dev. Min Max
Table 7: 3 year average summary statistics
37
Table 8: 5 year average summary statistics
38
APPENDIX VI: Correlograms of the explanatory variables
