The Romanian Economic Journal

3

Remittances and GDP Dynamics in 11 Developing Countries: Evidence from Panel Cointegration and PMG Techniques Anupam Das 1 Murshed Chowdhury 2 Despite a plethora of research, the role of remittances on economic growth is yet to be understood. Is there any long run relationship between remittances and GDP? This paper contributes to the literature by answering this question for 11 top remittance-recipient developing countries. These countries are: Bangladesh, Dominican Republic, El Salvador, Gambia, Guatemala, Honduras, Jamaica, Lesotho, Philippines, Senegal and Sri Lanka. Using recently developed econometric techniques, i.e., panel cointegration and pooled mean group (PMG) approach; our results support a positive long run relationship between remittances and GDP. However, the magnitude of the remittance-GDP coefficient is rather quite small. We hypothesize that remittances may be used to increase consumption in these economies. Our results also imply that developing countries should formulate policies to divert this external resource into more productive sectors. Keywords: External resources, economic growth, investment, consumption, long run JEL Classifications: F24, F43, C33 1.Introduction With the onset of globalization, remittance flows have been proliferated significantly to developing countries. Even without counting the sizeable flows through informal channels, remittances in 1

Anupam Das, PhD. Assistant Professor, Department of Policy Studies, Mount Royal University, [email protected] 2 Murshed Chowdhury, PhD Candidate. Department of Economics, University of Manitoba, [email protected]

4

The Romanian Economic Journal

these countries constitute the second largest source of external finance after foreign direct investment (FDI) (Glytsos, 2005). Total workers’ remittances went up from US$1.5 billion in 1975 to US$325 billion in 2008 (World Bank, 2011a and 2011b). Despite a modest decline by US$18 billion in 2009, it has been estimated that remittances will be back to its 2008 level by the end of 2010 (World Bank, 2011a). Such a large volume of remittances has motivated researchers to find out if remittances could be an important tool for economic development (Chami et al., 2003). Indeed, concerns about the development effect of remittances derive partly from concerns about its ultimate effect on growth. Despite a plethora of literature, the role of remittances on enhancing economic growth is unclear. Recent empirical work has generally argued that remittances have a positive impact on economic growth in developing countries (Pradhan et al., 2008; Loxley and Sackey, 2008; Giuliano and Ruiz-Arranz, 2009; Ziesemer, 2006). This conventional wisdom, however, has been criticized by Barajas et al. (2009), Rajan and Subramaniam (2005) and Chami et al. (2003). Critics argue that the growth effect of remittances is either negative or, at best, zero. Therefore, existing literature confronts about the effectiveness of remittance on economic development of recipient countries. From a theoretical understanding, remittance can increase consumption and investment in recipient countries and hence increase output and aggregate demand. On the other hand, due to compensatory nature of remittance, such an increase in income may create idleness among recipients. As a result, remittances fail to increase savings or investment in the economy (Kapur and McHale, 2003). The positive consumption effect of remittances was advocated by Catrinescu et al. (2006), Faini (2001), Taylor (1992), Jongwanich (2007), IMF (2005), Adams (1998, 2002), Pradhan et al. (2008), Tansel and Yasar (2010), Stahl and Habib (1989) and Nishat and Bilgrami (1991). Among them Stahl and Habib (1989), and Taylor (1992) particularly showed that the large multiplier effect of remittances in developing countries could increase GDP by a significant amount. Pradhan et al. (2008) supported the positive impact of remittance on Year XIV, no. 42 December 2011

5

The Romanian Economic Journal

economic growth. Also since the official estimates of remittance flow underestimate the actual flow, the use of more accurate data was likely to produce more pronounced relationship between remittance and economic growth (Pradhan et al., 2008). Tansel and Yasar (2010) recently estimated a Keynesian-type simultaneous and dynamic macroeconometric model to observe the relationship between remittance flows and economic growth in Turkey. Their findings also suggested a positive impact of remittances on growth. Negative relationship between remittance and economic growth was predicted by Chami et al. (2003). Following Becker’s (1974) economics of family model, they argued remittance flow was not profit driven, instead compensatory. Using the panel dataset for 113 developing countries, Chami et al. (2003) argued that the high flow of remittances created a dependency among recipients which in turns induced them to reduce their labor market participation. In such a situation, output might fall in home countries. Rajan and Subramanian (2005) used a growth equation with valid instrumental variables; nevertheless, they couldn’t find any robust and positive impact of remittances on longterm growth. The other set of literature argued that remittances were not necessarily productive in terms of the overall economy. Because remittances were mainly used for daily consumption purpose, this resource would fail to create sufficient savings required for desired growth (Chami, 2003; Sofranko and Idris, 1999). This set of literature concludes that remittances’ effect of growth is negative. However, a recent paper by Das (2012) argues that remittances can bring a favorable outcome even if it is used for consumption. Using various specifications of investment, consumption and growth equations for four developing countries over 1975 to 2006, Das (2012) showed that remittance can have a positive impact on economic growth either through consumption or investment. The size of the coefficient will be large if the effect comes from investment, while any effect through consumption will produce a smaller coefficient. Year XIV, no. 42

December 2011

6

The Romanian Economic Journal

It is clear from the above discussion that the findings are far from unanimous. Ambiguous results may be due to several factors, including different types of countries in the same panel and misspecified econometric models. Inconclusive results on remittancegrowth dynamics warrant for further examination. This paper contributes to this debate by focusing on a number of important issues. First, we examine a remittance-output relationship over the period 1985 to 2009 for a panel of 11 remittance-recipient developing countries, namely Bangladesh, Dominican Republic, El Salvador, Gambia, Guatemala, Honduras, Jamaica, Lesotho, Philippines, Senegal and Sri Lanka. The selection of these countries is guided by the fact that all these countries belong to a group of top remittance recipient countries in 2009 (World Bank, 2011a). Such a selection of countries increases the chance of producing statistically unambiguous results. Second, we conduct both panel as well as country-specific disaggregated unit root tests. Third, having found that all variables in our dataset have the same order of integration, we employ the panel cointegration test propagated by Pedroni (1997). Finally, test results from the panel cointegration motivate us to estimate a long run relationship between remittance and GDP by using Pooled Mean Group (PMG) approach suggested by Pesaran et al. (1999). This approach allows for both short-run adjustments and heterogeneity across countries. Moreover, PMG imposes cross-country homogeneity restrictions on the long run and identifies common long run coefficients. Indeed, this model performs better than dynamic fixed effect models since convergence speeds toward steady state can vary across countries- an assumption consistent with neoclassical growth model (Bassanini and Scarpetta, 2002). Same speed of adjustment can be assumed only when the growth rates of technology and population are same across the countries in our dataset. The rest of the paper is structured as follows. In section II, we discuss the methodology that is used to examine remittance-growth relationships. Section III presents the results and finally, section IV concludes the paper. Year XIV, no. 42 December 2011

7

The Romanian Economic Journal

2.Methodology 2.1 The Behavioral Equation

The growth equation to be estimated in this paper is drawn from the basic neoclassical growth model of Solow (1956). Given the humancapital data is not widely available for developing countries; this model has become the standard procedure in neoclassical growth empirics. We assume that GDP (ln gdp ) has a long run relationship with investment as a share of GDP (ln invy ) and employment (ln emp ) . Time series data on labor force is not readily available for most of the developing countries. Thus, the economically-active population is used as a proxy of employment. Economically active population is defined as the number of people that belongs to the age group from 15 years to 64 years. Finally, to examine the long run relationship between remittances and growth, remittance as a share of GDP (ln remy ) is incorporated in our modified neoclassical model. First differences of all these variables enter into the short run equation. The growth equation discussed above is estimated using a large panel of 11 countries over the period 1985 to 2009 (See Table 1 for the list of countries). Data are collected and calculated from the World Development Indicators (WDI) and the UNData published by the World Bank and the United Nations respectively.

Year XIV, no. 42

December 2011

8

The Romanian Economic Journal Table 1 List of Countries in the Dataset Bangladesh

Dominican Republic

El Salvador

Gambia

Guatemala

Honduras

Jamaica

Lesotho

Philippines

Senegal

Sri Lanka

2.2 Tests for Unit Roots

Conventionally, long-run parameters exhibit cointegrating relationships among a set of I (1) variables (Asteriou, 2009). Given our dataset includes time period T which is fairly long (25 years), it is highly likely that macroeconomic variables will be characterized by unit-root process (Nelson and Plosser, 1982). It is necessary to check the variables for the order of integration before examining any long run relationship. Hence, unit root tests for all variables in our dataset are imperative. We employ two different types of panel unit root tests to determine the level of integration for all variables in our dataset. The first one is known as Hadri’s Lagrange Multiplier (LM) test, propagated by Hadri (2000). This test statistics is defined as the simple average of the individual LM-statistics in the panel. 1 N ∑ i =1φi N

ϕ =

Year XIV, no. 42

December 2011

9

The Romanian Economic Journal

The null hypothesis of this test is that the panel is stationary. Hadri (2000) argued that the standardized panel LM test is asymptotically normally distributed: θ=

N (φ − µv )

σν

⇒ N (0,1)

where, µv = E [∫ V 2 ] and σ 2 v = VAR[∫ V 2 ]. V is given by a model specific functional of Brownian motions defined in Hadri (2000). The second test is by Im et al. (2003). The null hypothesis of this test is that the series contains a unit root. The limiting distribution of this test is given as: θ=

(

N t ADF − µ ADF 2 σ ADF

) ⇒ N (0,1)

where, ADF stands for augmented Dickey Fuller test. In this 2 distribution, µ ADF and σ ADF are from Monte Carlo simulations. Average estimated ADF t-statistics is represented by t ADF . It should be noted that Hadri test has a very strong null hypothesis that the variable is stationary for all panels. Thus, to verify our results we also perform a modified Dickey-Fuller t test for a unit root for each individual panel in which the series is transformed by a generalized least square regression. Results are presented in Table 2.

Year XIV, no. 42

December 2011

10

The Romanian Economic Journal Table 2 Stationarity Tests for Relevant Variables

Variable lngdp

Hadri Test

Im et al. Test

11.28***

7.28

∆lngdp

1.69

-8.56***

lninvy

9.96***

-0.83

-0.83

-11.03***

11.49***

-1.48

∆lninvy lnemp ∆lnemp

7.04***

-7.53***

lnremy

4.19***

-3.77***

∆lnremy

-0.44

-10.70***

DickeyFuller Test

Determination

11

Mostly NonStationary

11

Mostly NonStationary

8

Mostly NonStationary

11

Mostly NonStationary

Notes: 1) *** Indicates significance at the 1% level, 2) The null hypothesis of Hadri Test is stationarity, 3) The null hypothesis of Im et al. Test is non-stationarity, 4) The null hypothesis of Dickey Fuller Test is non-stationarity

Results from both Hadri (2000) and Im et al. (2003) tests show that two out of four variables (i.e., ln gdp and ln invy ) are clearly I (1) . For these two variables, the null of stationarity is rejected at the level, but not rejected at the first difference by Hadri test. On the other hand, the null hypothesis of non-stationarity is not rejected at the level while Year XIV, no. 42

December 2011

11

The Romanian Economic Journal

it is rejected at the first difference by Im et al. At least one test supports that the other two variables, i.e., ln remy and ln emp are I (1) . Results from Dickey-Fuller tests indicate that the null hypothesis of non-stationarity is rejected for all panels for ln gdp , ln invy and ln remy . For the other variable (i.e., ln emp ), the null hypothesis is not rejected for 8 out of 11 countries. These results suggest that variables in our dataset are mostly non-stationary. Based on these results, a panel cointegration approach will be the most appropriate one. 2.3 Panel Cointegration Technique

Since the results from unit root tests show cointegrating relationships in our dataset and because the primary focus of this paper is to find out a long run relationship between remittances and GDP, we employ a panel cointegration technique suggested by Pedroni (1999). This approach works better than other traditional pooled techniques since it controls for the country size, and heterogeneity. Panel cointegration technique is based on an examination of the residual of a spurious regression performed using I (1) variables. The residual should be I (0) if the variables are cointegrated. The residual of the hypothesized cointegrating regression can be established from the following equation:

(ln gdp)i ,t = α i + β i1 (ln invy )i ,t + β i2 (ln remy )i ,t + β i3 (ln emp )i ,t + φi t + ε i ,t where i=1,…,N; t=1,…,T, and N is the number of countries in the panel and T is the number of observations over time. The estimated residual becomes: )

) )

)

ε it = ρ i ε it −1 + υ it

With the null hypothesis of no cointegration, the residual is I (1) and ) ρ i = 1 . There are two alternative hypotheses. 1) the homogeneous alternative (also known as the within-dimension test), ρ) i = ρ < 1 for all

(

Year XIV, no. 42

)

December 2011

12

The Romanian Economic Journal

i, and 2) heterogeneous alternative (also known as the betweendimension or group statistics) ρ) i < 1 for all i. Pedroni (1997) discusses seven panel cointegration statistics. Four of those are based on the within-dimension test while the rest three are based on the between-dimension or group statistics approach. Under an appropriate standardization (i.e., using appropriate mean and variance), the asymptotic distribution of these statistics follows a normal distibuton. K=

K NT − µ N v

⇒ N (0,1)

where K NT represents the corresponding form of test statistics with respect to N and T. µ and v are the moments of the Brownian function that are given in Pedroni (1999). Numerical values of µ and v depend upon the presence of a constant, time trend, and the number of regressors in the cointegration test regression. 2.4 Pooled Mean Group

After finding out cointegrating relationships from both unit root and panel cointegration tests, we employ a recently developed econometric approach: the Pooled Mean Group (PMG). A detailed derivation of this approach can be found in Asteriou (2009). The PMG estimator has number of advantages. First, this approach provides common long run coefficients for all p panels while allowing for differences in short run and error correction coefficients. Since growth in technology and population are not same across countries in our dataset, it is legitimate to argue that there will be differences in the speed of error correction across panels (Bassanini and Scarpetta, 2002). Indeed, this characteristic makes PMG approach less restrictive than dynamic fixed effect models which assume common short-run and long-run coefficients as well as a common speed of error correction across panels. Year XIV, no. 42

December 2011

13

The Romanian Economic Journal

Using the PMG approach, our growth equation can be written in the following manner: ∆ ln gdpit = α i (ln gdpi ,t −1 −β 'i xi ,t −1 ) + ∑ δ 'ij ∆ ln gdpi ,t − j + ∑ φ 'ij x i ,t − j + ∑ ϕ 'ij z i ,t − j + µi + ∈it m −1

n−1

q −1

j =1

j =1

j =1

where, xi , j is the vector of non-stationary variables; zi , j is the vector of stationary variables; µ i represents the fixed-effect;∈it represents the vector of standard errors. α i is the error correction coefficient. β 'i represents the long run parameters, and finally, δ 'ij , φ 'ij and ϕ 'ij represent country specific short-run coefficient vectors. 2. Results and Discussions Results from the panel cointegration tests are presented in Table 3. All seven panel-cointegration test statistics are reported in the table. Among those, three out of four within within-dimension-based, panel ρ, panel PP and panel ADF statistics demonstrate the rejection of the null hypothesis of no cointegration. All three between dimensionbased statistics, i.e., group ρ, group PP, and group ADF reject the null hypothesis of no cointegration. It has been argued by Gutierrez (2003) that the group ρ statistic has the best power among the test statistics of Pedroni (1999). Thus, it can be concluded that the variables in our dataset exhibit long run cointegrated relationships.

Year XIV, no. 42

December 2011

14

The Romanian Economic Journal Table 3 Results from the Panel Cointegration Estimates

Statistic

Probabilities

Panel υ-statistic

-0.701

0.311

Panel ρ-statistic

1.904

0.065

Panel PP statistic

1.860

0.070

Panel ADF statistic

2.423

0.021

Alternative Hypothesis: Individual AR coefficient Group ρ-statistic

2.687

0.010

Group PP statistic

2.280

0.029

Group ADF statistic

2.838

0.007

Notes: The null hypothesis of Pedroni’s (1997) panel cointegration procedure is no coingretaion. One shortcoming of Pedroni (1997) panel approach is that it does not provide estimation for long run relationships or the speed of adjustment towards the long run (Murthy, 2007). Therefore, we continue with PMG estimations to find out the long run relationship between remittances and GDP and also the short run speed of adjustment. The following table (Table 4) presents results from the PMG estimates. Overall results suggest that the long run coefficient for remittance to GDP ratio is 0.04. This coefficient is significant only at Year XIV, no. 42

December 2011

15

The Romanian Economic Journal

the 10% level. In other words, remittances increase GDP in the long run by only 4%. Needless to say, the size of this coefficient is small and thus, differs from the existing literature which suggests a large, positive impact of remittances on GDP growth. Table 4 PMG Estimates of the Growth Equation Dependent Variable: lnGDP Explanatory Variables

Long-Run Coefficients

Lninvy

0.098*** (4.12)

Lnemp

2.276*** (29.01)

Lnremy

0.041* (1.84) Short-Run Coefficients

Error correction coefficient

-0.120** (-2.03)

∆ lninvy

0.079*** (3.65)

∆ lninvy (First lagged)

0.035** (2.23)

∆ lnemp

1.727(0.98)

∆ lnremy

-0.034**(-1.99)

∆ lnremy (First lagged) Constant Year XIV, no. 42

-0.012 (-1.23) -2.706** (-2.14) December 2011

16 Number of Observations

The Romanian Economic Journal 275

Notes: 1) *** Indicates significance at the 1% level. 2) Figures in brackets are z-statistics. 3) First lag is included in the short run equation whenever the variable (without lag) is found to be significant. Two other variables in the long run equation, i.e., ln invy and ln emp are found to be very significant (at better than the 1% level) and positive. These results support the assumption of neoclassical growth model. Not surprisingly, most of the variables in the short run are found to be insignificant. As expected the error correction term in the short run equation is significant at the 5% level. The coefficient of this term is negative 0.12, suggesting that approximately 12% of the deviations from the long run equilibrium are corrected in the first year. Results from estimated panel cointegration and PMG procedure suggest positive and long run relationship between remittance and GDP in Bangladesh. However, remittances impact on GDP is very weakly significant (statistically) and quite small in magnitude. Although our results suggest that there are long run relationships between variables in our dataset, these results should be interpreted with caution. First, we argue that remittances in these countries may be used for consumption purposes. Das (2012) showed that the effect of remittances on growth can be positive even if it is used for consumption; however, the magnitude would be small. We can expect a large magnitude of the coefficient only when remittances are used to increase investment in the recipient countries. Second, we hypothesize that the small magnitude of the coefficient is due to a large transaction of remittances through unofficial channels, which are not captured in our estimations. The estimates of unofficial remittances are still Year XIV, no. 42

December 2011

17

The Romanian Economic Journal

judgmental. In fact, El-Qorchi et al. (2003) estimated that unofficial transfers of remittances to the developing world are approximately $10 billion per year. Page and Plaza (2006) pointed out a number of countries and regions where the amount of unrecorded remittances is higher than other countries. Most of these countries are also highremittance recipient countries and thus, included in our dataset. Therefore, we argue that our estimation results would have been different if we could include unrecorded remittance flows into our estimation. These results have important policy implications for developing countries. One important result derived from our estimation is that the impact of remittances on GDP is quite small (in size) in the long run. Countries in our dataset already attract a significant flow of remittances. Policy-makers simply need to formulate policies so that this large size of external resource is diverted towards more productive investment sector. New policies may be required to improve the financial sector to encourage remittance flows through the formal sector. In recent years, developing countries have taken number of initiatives to encourage migrants to send money through the formal sector. These initiatives include setting up new branches in host countries and remittance transfer through mobile phones (Das, 2012). 4.Conclusion Remittance flows have increasingly become one of the most important sources of external financing in many developing countries. Sharp increase in remittance in recent decades claims the investigation of effectiveness of such flow on economic growth of the recipient countries. Bangladesh, Dominican Republic, El Salvador, Gambia, Guatemala, Honduras, Jamaica, Lesotho, Philippines, Senegal and Sri Year XIV, no. 42

December 2011

18

The Romanian Economic Journal

Lanka have placed themselves in the 2009 list of top remittance recipient countries in the world. As a result, the question whether remittances could influence the long run growth in these countries is interesting and important. We apply a number of econometric procedures to answer this question. Our dataset covers the period 1985 to 2009. Although remittance-GDP relationship is one of the most controversial in the empirical literature, a number of studies found a significant relation between these variables (Paul et al., 2011). We too find a similar result for top remittance recipient countries. Results from unit root tests and panel cointegration suggest that there exist long run cointegrating relationships among variables in our dataset. Additionally, we use a recently developed econometric technique: Pooled Mean Group (PMG) to find out the size and the sign of remittance coefficient in the growth equation. This technique allowed for short-run adjustments and heterogeneity across countries. Results from PMG estimation suggest rather a small impact of remittances on economic growth in top remittance recipient countries. Only 4% growth in these countries is due to remittance flows during the period 1985 to 2009. If remittances are used for to increase consumption in the recipient countries, the long run remittancegrowth coefficient can be small in size. Otherwise, the large transaction of unofficial remittances could also be responsible for small coefficient size. These hypotheses can be proved with microeconomic investigations with household data, which is beyond the scope of this present research. Finally, this research raises some additional questions. First, does our panel result support individual country-specific time series analysis? Second, does the growth effect of remittances come through Year XIV, no. 42

December 2011

19

The Romanian Economic Journal

consumption or investment? These questions are important and hence, left for future research.

References Adams Jr., R.H. (2002), “Precautionary Saving from Different Sources of Income: Evidence from Rural Pakistan”, Policy Research Working Paper, 2761, Washington, DC: The World Bank. Adams Jr., R.H. (1998), “Remittances, Investment, and Rural Asset Accumulation in Pakistan”, Economic Development and Cultural Change, 47: 155–173. Asteriou, D. (2009), “Foreign Aid and Economic Growth: New Evidence from a Panel Data Approach for Five South Asian Countries”, Journal of Policy Modeling, 31: 155-161. Barajas, A., Chami, R., Connel, F., Gapen, M. and Montiel, P. (2009), “Do Workers’ Remittances Promote Economic Growth?” IMF Working Paper, WP/09/153, Washington, DC: IMF. Bassanini, A. and Scarpetta, S. (2002), “Does Human Capital Matter for Growth in OECD Countries? A Pooled Mean-Group Approach”, Economics Letters, 74: 399-405. Becker, G. (1974), “A Theory of Social Interaction”, Journal of Political Economy, 82: 1063-1093. Catrinescu, N., León-Ledesma, M., Piracha, M., and Quillin, B. (2006), “Remittances, Institutions, and Economic Growth”, IZA Discussion Paper, 2139. Year XIV, no. 42

December 2011

20

The Romanian Economic Journal

Chami, R., Fullenkamp, C. and Jahjah, S. (2003), “Are Immigrant Remittance Flows a Source of Capital for Development?” IMF Working Paper, WP/03/189, Washington, DC: IMF. Das, A. (2012), “Remittance Behavior of Migrants and its Macroeconomic Effects in Four Developing Countries”, International Journal of Applied Behavioral Economics, 1(1), Forthcoming. El- Qorchi, M., Maimbo, S.M. and Wilson, J. (2003), “Informal Funds Transfer Systems: An Analysis of the Informal Hawala System”, IMF Occasional Paper, 222, Washington, DC: International Monetary Fund. Faini, R. (2001), “Development, Trade, and Migration”, Washington, DC: International Monetary Fund. Giuliano, P. and Ruiz-Arranz, M. (2009), “Remittances, Financial Development and Growth”, Journal of Development Economics, 90(1): 144-152. Glytsos, N.P. (2005), “The Contribution of Remittances to Growth”, The Journal of Economic Studies, 32: 468-496. Gutierrez, L. (2003), “On the Power of Panel Cointegration Test: A Monte Carlo Comparison”, Economic Letters, 80: 105-111. Hadri, K. (2000), “Testing for Stationarity in Heterogeneous Panel Data”, The Econometrics Journal, 3: 148-161. Im, K.S., Pesaran, M.H. and Shin, Y. (2003), “Testing for Unit Roots in Heterogeneous Panels”, Journal of Econometrics, 115: 53-74.

Year XIV, no. 42

December 2011

21

The Romanian Economic Journal

IMF (2005), “World Economic Outlook”, Washington, DC: International Monetary Fund. Jongwanich, J. (2007), “Workers’ Remittances, Economic Growth and Poverty in Developing Asia and the Pacific Countries”, UNESCAP Working Papers, WP/07/01. Kapur, D. and McHale, J. (2003), “Migration’s New Payoff”, Foreign Policy, 139: 48-57. Loxley, J. and Sackey, H. (2008), “Aid Effectiveness in Africa. African Development Review”, 20: 163-199. Murthy, N.R.V. (2007), “Panel Cointegration Analysis: An Empirical Example, in Cointegration for the Applied Economist”, In Rao, B. (ed.) Cointegration for the Applied Economics, (Second Edition): 222256. Nelson, C. and Plosser, C. (1982), “Trends and Random Walks in Macroeconomic Time Series: Some Evidence and Implications”, Journal of Monetary Economics, 10: 130-162. Nishat, N. and Bilgrami, N. (1991), “The Impact of Migrant Workers’ Remittances on Pakistan Economy”, Pakistan Economic and Social Review, 29: 21-41. Page, J. and Plaza, S. (2006), “Migration Remittances and Development: A Review of Global Evidence”, Journal of African Economies, 15: 245-336. Paul, B.P., Uddin, M.G.S. and Noman, A.M. (2011), “Remittances and Output in Bangladesh: An ARDL Bounds Testing Approach to Cointegration”, International Review of Economics, 58: 229-242. Year XIV, no. 42

December 2011

22

The Romanian Economic Journal

Pedroni, P. (1999), “Critical Values for Cointegration Tests in Heterogeneous Panels with Multiple Regressors”, Oxford Bulletin of Economics and Statistics, 61: 653-670. Pedroni, P. (1997), “Panel Cointegration; Asymptotic and Finite Sample Properties of Pooled Time Series Tests, with an Application to the PPP Hypothesis: New Results”, Working Paper, Indiana University, April. Pesaran, M.H., Shin, Y. and Smith, R. (1999), “Pooled Mean Group Estimation of Dynamic Heterogeneous Panels”, Journal of the American Statistical Association, 94: 621-634. Pradhan, G., Upadhaya, M. and Upadhaya, K. (2008), “Remittances and Economic Growth in Developing Countries”, The European Journal of Development Research, 20: 497-506. Rajan, R. G. and Subramaniam, A. (2005), “What Undermines Aid’s Impact on Growth?” NBER Working Paper, 11657. Stahl, C.W. and Habib, A. (1989), “The Impact of Overseas Workers’ Remittances on Indigenous Industries: Evidence from Bangladesh”, The Developing Economies, 27: 269–285. Sofranko, A.J. and Idris, K. (2009), “Use of Overseas Migrants’ Remittances to the Extended Family for Business Investment: A Research Note”, Rural Sociology, 64: 464-481. Solow, R. M., (1956), “A Contribution to the Theory of Economic Growth”, The Quarterly Journal of Economics, 70: 65-94.

Year XIV, no. 42

December 2011

23

The Romanian Economic Journal

Tansel, A. and Yasar, P. (2010), ‘Macroeconomic Impact of Remittances on Output Growth: Evidence from Turkey”, Migration Letters, 7: 132-143. Taylor, J.E. (1992), “Remittances and Inequality Reconsidered: Direct, Indirect and Intertemporal Effects”, Journal of Policy Modelling, 14: 187-208. World Bank (2011a), “Migration and Remittances Factbook 2011”, Second Edition, Washington, DC: The World Bank. Available at: http://siteresources.worldbank.org/INTLAC/Resources/Factbook20 11-Ebook.pdf World Bank (2011b), “World Development Indicators”, Washington, DC: The World Bank. Ziesemer, T. (2006), “Worker Remittances and Growth: The Physical and Human Capital Channel”, UNU-Merit Working Paper Series, 020, United Nations University. Available at: http://www.merit.unu.edu/publications/wppdf/2006/wp2006020.pdf

Year XIV, no. 42

December 2011