Nonlinear growth effect of remittances in recipient countries: an econometric analysis of remittances-growth nexus in Bangladesh

The macroeconomic impacts of remittances flows on developing economies are not well understood. The paper is an attempt to understand the impact of inward remittances flows on per capita GDP growth in Bangladesh during 1974-2006. We find that the growth effect of remittances is negative at first but becomes positive at a later stage- a strong evidence of a non-linear relationship. This could be due to unproductive use of remittances in the beginning followed by more productive utilisation. Remittances positively affect per capita GDP growth in Bangladesh when the complementarity between remittances and financial development is incorporated into the analysis. Abstract : The macroeconomic impacts of remittances flows on developing economies are not well understood. The paper is an attempt to understand the impact of inward remittances flows on per capita GDP growth in Bangladesh during 1974-2006. We find that the growth effect of remittances is negative at first but becomes positive at a later stage- a strong evidence of a nonlinear relationship. This could be due to unproductive use of remittances in the beginning followed by more productive utilisation. Remittances positively affect per capita GDP growth in Bangladesh when the complementarity between remittances and financial development is incorporated into the analysis.


Introduction
The importance of the flow of workers" remittances in the economies of developing countries during the last few decades or so cannot be ignored at the face changing global order where most of the economies in the world are transforming themselves to the call of globalization and transmuting towards more open markets with freer flows of goods and factors across borders. Remittancesthe unrequited transfer of funds by the migrants to their families at homeare a source of foreign exchange which is much scarce in developing economies. It is a more stable and less volatile source of external finance when compared to the other forms of flows which include official development assistance and foreign direct investment (Ratha 2007). Given the surge in the flows of remittances world wide (IMF 2005, World Bank 2005, and Ratha 2007, especially in the developing countries where remittances are twice the size of official development assistance (ODA) and as large as foreign direct investment (FDI), it has become important to study the development impacts of remittances in those economies. Potentially remittances inflows can have strong development impacts in the economy. As Kapur (2004) notes: "Remittances finance consumption, land and housing purchases and philanthropy; they are an important source of social insurance in lower income countries; and they provide liquidity for small enterprises (in the absence of well functioning credit markets) as well as capital investmentsin equipment, land, wells and irrigation works and educationwith longer-term implications for economic development." However, remittances can also have counter effects in the source economy unlike ODA or FDI because it is an outcome of labour migration. Like there are externalities (positive and/or negative) associated with labour migration, the impact of remittances on the economy measured at the macroeconomic, household or community level can be either positive or negative in the country of origin. Table 1 compiles the major conclusions reached in the researches that fall under the heading of "development impact of remittances".  T. and Wets, J. (2006) The topic of "development impact of remittances" is a broad and complex one and the analysis is less straight forward. During the last decade or so there has been a surge in researches on remittances and consequently there has emerged a vast literature around the topic of "development impact remittances". This paper however is directed towards studying only the growth impacts of remittances in Bangladesh where "remittance and economic growth" is a topic in the "macroeconomic impact of remittances"a subsection of "development impact of remittances" literature. 1 This paper is organized as follows. Following this brief introduction the paper will discuss in section two the dynamics of remittances flows and other important macro economic variables in Bangladesh. Section three will provide a review of researches on remittances in Bangladesh followed by section four which reviews the literature on remittances and economic growth. In section five the paper will present an econometric model and provide econometric results on the impact of remittances on economic growth in Bangladesh during 1974Bangladesh during -2006 and conclude by outlining the main findings and limitations of the study and providing guidance for further research.

Remittances Flows in Bangladesh
Bangladesh has experienced continuous surge in inward remittances flows both in terms of volume as well as in terms of ratio to GDP over the last thirty years. According to Bangladesh Bank"s Quarterly 2 total inward remittances was USD 24 million in 1976, but the amount stood at a staggering USD 6584 million (just over 6.5 billion) in 2007.
According to IMF"s International Financial Statistics, October 2007, Bangladesh was the tenth largest remittances recipient economy among the developing countries measured by the average for the flows over the period 1990 -2005. It was on the top ten recipients of remittances list in 2006 by Ratha (2007) based on volume and ranked 14 th in 2005 (World Bank, Global Economic Prospect 2006). The remittances to GDP ratio reached to an impressive 9.4 percent in 2007 from a decent 3.5 percent in 1997.

Figure 1. External Flows in Bangladesh
. In Figure 1 we can see that compared to other forms of external finances viz overseas development assistance (ODA) and foreign direct investment (FDI), remittances flows in The nature and size of remittances inflows warrant a systematic study of their macroeconomic impacts into the economy. According to Chami et al. (2008) there are three main features of remittances that provide impetus to study their macroeconomic impacts. These are: i) size of these flows relative to the size of the recipient economies; ii) the likelihood that these flows will continue unabated because of continued trend in globalization; and iii) These flows are quite distinct from those of official aid or private capital. (In contrast to these remittances are composed of numerous small transfers between private individuals which are tied to familial relationship in the home country).
Given the importance of studying the macroeconomic impact of remittances as outlined above, in my Ph.D. thesis, we endeavour to concentrate our study into four areas of the macroeconomy in Bangladesh where remittances can have potential impacts. These are the impact of remittances on macroeconomic volatility, financial development, real exchange rates and on economic growth. Our hypotheses are in Bangladesh remittances reduce macroeconomic volatility, promote financial development, and appreciate real exchange rate and through these channels affect economic growth. However, in this paper we will just concentrate on the impact of remittances on GDP growth in Bangladesh.
The macroeconomic effects of remittances in the economy can be decomposed into their effects on savings, investment, consumption and growth. As a basic macroeconomic identity we know that total disposable income equals consumption plus savings.
Remittances can be thought of augmenting total disposable incomes of the households. A part of which is consumed and rest is saved. In the portion of consumption expenditures some are actually investments as represented by expenditure incurred to buy durable household goods, finance education and health, purchase land or to provide capital for micro business start ups etc. So the impact of remittances on the receiving economy should be working through savings and investment as well as short run effects on aggregate demand and output through consumption. Some important graphs related to remittances and other macroeconomic variables in Bangladesh like investment, savings, consumption, financial deepening, and growth are provided through Figures 2 to 8. Figure 2 shows that remittances to GDP ratio and gross fixed investment to GDP ratio in Bangladesh seem to be positively connected over the period over the period 1990 -2006.

Figure 2. Remittances and Fixed Investment
Remittances can also go towards financing household consumption as we can see in Figure3.  Remittances can lead to increase in the demand for money and expand the supply of funds into the recipient countries" banking system. This could be affecting the broad measures of financial developmentcredit to private sector to GDP and M2 to GDPin Bangladesh.

Figure 5. Remittances and Financial Development
We can see from Figure 5, there seems to be a strong link between remittances to GDP ratios and the measures of financial deepening (M2/GDP and Domestic Credit/GDP) in Bangladesh. The graph shows a positive correlation between remittances and other two series leading to the conjecture that remittances inflow could contribute to the development of the financial sector in the Bangladesh economy.
In Figure 6 we will casually observe the dynamics of real effective exchange rate in  1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006  The overall trend is that REER is at the same level at the end of period compared to where it was in the beginning of the period. If we take a look at the data of REER for a bit longer time period, we can see instead that there is some sign of real depreciation compared to other South Asian economies. Figure 7 shows that over the longer time   It is difficult to say by looking at Figure 8 if there is any link between remittance and GDP growth in Bangladesh. However, though GDP growth per se is stochastic, trend in GDP growth in the graph is increasing over the period 1990 to 2006 during the time when remittances to GDP ratio are also steadily increasing. So there might be a relationship.
One can see another interesting observation in Figure 8. During the time when remittance was increasing steadily, especially from 1992 to 2001, the variability of GDP growth rate, measured as the deviation of GDP growth from the trend, was also considerably less.
This could mean that increased flows of remittances can reduce output volatility.

Review of Studies on Remittances in Bangladesh
Researches related to remittances in Bangladesh are scant and have not yet been directed in the area where this thesis aims to contribute which is the macroeconomic impact of remittances. The remittances literature related to Bangladesh tends to be both qualitative and quantitative in nature. Some earlier studies employ quantitative methods to analyse the empirical issues related to remittances. The later studies are primarily qualitative aimed towards understanding the nature and profiles of remitter and recipient households in a broader social context.

Remittances and Savings
The earliest paper on remittances in Bangladesh is a micro level study by Mahmud and Osmani (1980) who quantitatively modelled expenditure and saving behaviour of the migrant and the household receiving remittances. They found that migrants remit more than eighty percent of their income from abroad, and that there is a significant difference in the saving rate between remittance receiving and non-receiving household. They also showed that savings ratio monotonically increases with income and for the highest remittance receiving income group saving ratio was found to be almost three fourth of income. One central finding of their study is that remittances go into unproductive use.
An early quantitative study at the macroeconomic level of remittances in Bangladesh in comparison to other Asian countries is by Qubria (1986) who showed that the steady flows of remittances have eased the foreign exchange constraints, improved the balance of payments and has augmented national savings.

Remittances, Consumption and Welfare
There is a general tendency among the early authors to conclude that remittances go into consumption. However this may not be necessarily bad. Using computable general equilibrium (CGE) modelling framework, Stahl and Habib (1989) showed that even if only small proportions of remittances income go to direct investment while the majority go to serve the purpose of consumption needs, remittances could still be developmental.
They set up CGE model and quantitatively explored whether expansion in consumption increase aggregate demand. The study found evidence that in Bangladesh remittances tend to be spent within those sectors which have relatively strong linkage with the rest of the economy. Thus many sectors not directly benefiting from the remittance expenditure would nonetheless experience an increase in demand for their output inducing investment in their sectors and expanding employment. In a theoretical paper not exactly related to Bangladesh but in general applicable to labour exporting Asian labour-surplus economies, Qubria (1997) simultaneously combined migration and remittance in a model to derive and prove a series of propositions. The main theoretical conclusions were that if remittances per capita exceed the source-country wage rate then aggregate GNP of the labour-exporting country will increase and also that if migration is accompanied by sufficient remittances then it will enhance welfare in the source country irrespective of the criteria that is used to measure welfare.

Remittances Utilisation, Micro Finance Institutions (MFI) and Socio-Economic Context
The later studies of remittances flows in Bangladesh were mainly qualitative except some very recent ones. The objectives of these studies had been to identify the nature and dynamics of remittances, recipient households and the remitter in a broader social context. Murshid et al (2002) is one such study. It contained a micro-level survey data methodology to identify the socio economic profiles of migrant family and discussed channels both official and unofficial through which remittances come and various types of remittances. It also discussed the prospects and challenges of channelling remittances via the official way through the national and multinational banks. Murshid et al (2002) also provided a Keynsian type analysis of macroeconomic effects of remittances on the economy. Using a short-run Keynsian structural equation type model this study estimated a remittances multiplier in the economy which equalled 3.33 indicating a one million taka (Bangladeshi local currency) increase in remittances would increase national income by 3.33 million taka. Note that this is just a level effect not a growth effect. This type structural equation type estimation is not without problems because of the assumption of fixed price and interest, as the model ignores the monetary and the real sector.
Nonetheless this is so far the only study in Bangladesh which tried to econometrically estimate some macroeconomic effects of remittances. A major study by Siddiqui and Abrar (2003) used survey data and tried to qualitatively investigate nature and dynamics of people linked through remittances and the linkage between workers" remittances and micro-finance institutions in channelling and transferring remittances in Bangladesh. By surveying one hundred households, this study provided detailed characteristics of remittance receiving households, their socio economic profiles and the different ways remittances are used by them. For example, the study found that more than half the remitters, three quarter of whom are less than thirty five years of age, were married although the highest number of remittances recipient members were the parents who are typically more than fifty years of age, half of whom are illiterate. The study reported that some portion of remittances were invested in land and did not go towards savings while a substantial part of remittances were used to finance migration of other family members.
Finally the study discussed how the extended network of the multi finance institutions (MFI) can be used to channel inward remittances and direct them to productive usage. In a similar paper, Azad (2004)  The study later gives recommendations on how to coordinate all these factors to make remittances transfers efficient. One major finding of this study is that the general perception that remittances are put to unproductive use has no empirical validity anymore. The study further notes that such observation may have been valid in late 1970 or early 1980s. However in 1990s the migrants families have tried to effectively utilise the remittances they received by investing in nutritious food for the family members, health, education, land purchase and financing migration of other family members. De Bruyn and Kuddus (2005) studied the various impacts that remittances have on households and on the broader community level in Bangladesh and discussed the various means through which government or NGOs can enhance the impact of remittances. In the end they discussed major hindrance and opportunities to efficiently utilize remittances for productive purpose which have an impact on the society as a whole.

Macroeconomic Determinants of Remittances
One recent study on remittances in Bangladesh has been involved with the current issues in remittances literature -determinants of workers" remittances. A macroeconomic paper by Barua, Majumder, Akhteruzzaman (2007) used a balanced panel dataset of bilateral remittances flows from ten major destination countries for Bangladeshi migrants" over the period 1993 to 2005 and identified the major determinants of the inflow of workers" remittances in Bangladesh. Income differential was found as one significant determinant of remittances. High inflation in home country vis-a-vis the host countries exerts negative influence on inward remittances whereas depreciation (or devaluation) of currency leads to increased flow of remittances.

Remittances and Poverty
Khan (2008)  However, the effect of remittances on long term economic growth could also be negative.
For instance inward remittances are by products of outflows of work force-often skilled or semi skilled-that should be negatively affecting growth (World Bank 2005). Chami, Fullenkamp, and Jahjah (2003) argue that a substantial amount of remittances are motivated by altruism and arrive in the economy to compensate for any income losses incurred by the migrants" family members due to bad economic condition at home. This creates a moral hazard problem as the migrant would not know whether income loss is due to genuine bad economic condition or not. As a result this may encourage the recipient households who participate in local labour market to choose more leisure over work and cause an aggregate reduction in labour supply which may negatively affect economic growth. Another way remittances can negatively affect growth is by diminishing technological capacity of the economy through appreciating the real exchange rate (Chami et al 2008). Usually an economy"s technical capacity will largely depend on the size of its tradable goods sector. Production in some component of the traded goods sector such as non-traditional manufactures intended for export can spur technological diffusion and increase the technological capacity for other firms in the economy. Arrival of remittances inflow can appreciate the economy"s real effective exchange rate and render the tradable sector uncompetitive. This "Dutch disease" effect will shrink tradable goods sector and allow resources to shift to the non-tradable goods sector and thus diminish the growth of technological capacity of the whole economy and subsequently reducing economic growth.
Relatively recently there has emerged a substantial empirical literature to measure the impact of remittances on GDP growth. These papers are discussed below.
Chami, Fullenkamp, and Jahjah (2003) Chami, Fullenkamp, and Jahjah (2003) (henceforth CFJ) use aggregate remittance data for a sample of 83 countries over the 1970 -1998 period to examine the relationship between workers remittances and per capita GDP growth. The study used panel regressions of growth of real GDP per capita on workers" remittances to GDP conditional on initial per capita income, investment to GDP ratio, inflation rate, regional dummy and the ratio of net capital flows to GDP. Later the study replaced workers" remittances to GDP variable with a change in that ratio as a regressor to incorporate the dynamic nature of private transfers. Overall, Chami et al (2003) found that, the investment to GDP and net private capital flows to GDP ratio were positively affecting growth but workers" remittances to GDP ratio either was not significant or negatively related to growth. In order to test whether the relationship could be non-linear, CFJ used a squared term of the remittances to GDP ratio and still found the same result. However, when change in the ratio was added to replace the variable in levels, it was found that annual change in remittances to GDP ratio negatively and significantly affects growth. To account for possible endogeneity problem such that the main causes of remittances are also affected by it, CFJ used some instruments for remittances and also conducted a two stage instrumental variable technique to estimate the relationship. In the first stage each workers" remittances to GDP ratio was estimated as a function of each country"s income gap and real interest gap relative to the United States. Having estimated this equation, the fitted values of remittances to GDP are obtained. In the second-stage per capita GDP growth regression is estimated using fitted values of remittances to GDP ratio from stage one as regressor conditional on the same variables as before. Here too the changes in remittances to GDP ratio are found to be negatively related to growth.

IMF (2005)
The IMF (2005)  openness and financial development. The study also used two instruments for remittances. These include distance between the migrants" home and main destination country, and a dummy measuring whether the home and the main destination country shared a common language. Since these instruments are time invariant, a panel estimation was not undertaken rather a cross-sectional estimation over the averages of 1970 -2003 was done. Using this cross-country growth regression framework, no significant link was found between real par capita GDP growth and remittances.

Faini (2006)
Faini (2006)  International Country Risk Guide index as an indicator of institutional quality. The estimated coefficient on the remittances to GDP ratio was found to be positive and significant in the OLS regressions of the study. Later foreign aid to GDP ratio was included as additional control variable and remittance was still found to be positively affecting growth. In order to encounter the endogeneity problem, the study also used an instrumental variable estimation procedure. Remittances to GDP ratio were instrumented by distance from the home country to the main destination country. The result was still in support of a positive impact of remittances on per capita GDP growth but the estimated coefficient in remittances to GDP became insignificant. Finally, in order to see how the impact of remittances interacts in deferential policy environment, the study added two additional regressors: black market exchange premium and inflation. These variables measure the overall stability of the macroeconomic system. Higher values of these variables would imply weak macro policy environment. In such a environment the incentive to use remittance income for investment purpose will be relatively weak. To test this hypothesis, the study included two interacting terms as regressors. Such as remittances to GDP interacted with black market exchange premium and with inflation rate. The estimated coefficient on the interacting terms was found to be negative and significant meaning that the positive impact of remittances on growth gets diluted when the macroeconomic policy environment is unstable and unsound.

Catrinescu, Leon-Ledesma, Piracha and Quillin (2006)
The study by Catrinescu, Leon-Ledesma, Piracha and Quillin (2006)  The study also found evidence of a positive relationship between growth and remittances though the magnitude of this relationship was a little weak.

Guiliano and Ruiz-Arranz (2006)
The study by Guiliano and Ruiz-Arranz (2006)  World Bank (2006) The World Bank (2006) study conducts a panel data estimates on the impact of remittances and growth using a sample of 67 countries measured over the period of 1991 -2005. The control variables used were (all in logs) initial GDP per capita, secondary school enrolment ratio, the private domestic credit to GDP ratio, ICRG political risk index, real export and import to GDP ratio, inflation rate, real exchange rate overvaluation, government consumption, and time period dummies. A SGMM estimation procedure was performed with the use of external as well as time varying instruments to control for the endogeneity of remittances. These instruments were formed by multiplying the share of a country"s migrants to its top five OECD destinations and a measure of the respective OECD country"s economic performance, such as GDP per capita, GDP growth rate or unemployment rate. The other instruments were average output per capita of the top country destinations across the world weighted by the inverse of the distance between remittance sender and recipient country. Instrumented this way, the regression specifications found a positive and significant relationship between remittances to GDP ratio and per capita GDP growth. The magnitude of the estimated effect of remittances on growth was found to relatively small in economic terms. Increase in remittance to GDP ratio from 0.7 percent to 2.3 percent is estimated to have led to an increase in 0.27 percent increase in GDP growth. However, when the investment to GDP ratio is included along with the rest of control variables, the estimated coefficient on remittances to GDP looses significance. This led the study to conclude that one of the main channels through which remittances work is through increasing domestic investment.
World Bank (2008) According to this World Bank (2008) study what impact remittances have on growth will depend on the policies being implemented by the home country. This study found out a number of areas that are complements to remittances in the matter of enhancing economic growth. This study included interaction terms with for remittances and education, remittances and financial depth, remittances and institutional qualities, and remittances and macroeconomic policy distortions. This study used a GMM instrumental variable procedure to estimate four growth regressions for each of the complementary sector outlined above. The instruments used were the time varying one in World Bank (2006). All regressions used the same control variables as in World Bank (2006) but each included a remittances and the respective complementary terms as additional control. The argument is remittances augments growth in the presentence of complementary policies that encourage education, increase financial depths, improve institutional qualities and reduce macroeconomic policy distortions. The study found a negative and significant coefficient on the total remittances to GDP ratio, but positive and significant coefficients on each of the interaction terms. This, the argued, implied a net positive impact of total remittances on GDP.

Does Remittances Affect Economic Growth in Bangladesh? Econometric Results
In this section we are going to provide an econometric analysis of the growth impact of remittances in Bangladesh. Firstly we will discuss the basic model. The basic model has three alternate specifications. Each of these specifications will be estimated using an OLS, Instrumental Variable-Two Stage Least Square (IV-2SLS), and IV-GMM estimators. In the end we will discuss some limitations of our results.

Econometric Model
Our basic model for the analysing whether remittances affect growth in Bangladesh is: where y is the per capita GDP growth, our variable of interest is Rem which is log of remittances 3 to GDP ratio and Z is a vector of control variables. The control variables were those which were frequently included in the "growth -remittances" regressions in the literature. Here they are (all in logs) gross capital formation to GDP ratio (lgcf), populations growth (lpop), government consumption to GDP ratio (lgov), M2 to GDP ratio (lm2), inflation rate (linf). The last three controls are included as a proxy to the capture effects of government size, financial development and relative macroeconomic stability.
Next we recognise the fact that the "growth -remittances" relation in Bangladesh may be non-linear. As a result we provide an alternative specification of [1] by including the squared term of our variable of interest Rem: all the variables here are same as [1], only exception is 2 Re t m which is squared term of the variable Rem (lrem2).
Finally, to capture the interactive effect, if there is any, between remittances and some other control variable included in Z, we will modify the specification [2] further. As it is found in the literature, remittances could be affecting economic growth via its interacting effect with the level of financial development. To capture this interaction between remittances and financial development we will include an interacting term in [2] and provide our third and final specification: is the log of remittances to GDP times log of M2 to GDP ratio (lm2rem).

Econometric Results
We present the econometric results of our models outlined in equations [1] through [3].
We report an OLS estimation followed by an IV-2SLS and a GMM-IV estimators. The results are presented at  . This time the estimated coefficient on the remittances variable is negative and significant, and the estimated coefficient on the squared remittances term is positive and highly significant. In this specification the R 2 increases from the previous one implying a better fit. Lastly we assume that remittances could be affecting growth through its interacting effect on financial development. This is why we include a interaction term between remittances and M2 to GDP ratio, and provide our estimation result of Equation [3]. We see that this remittance and growth relationship has a positive sign though insignificant. The coefficient on the squared remittances term is positive and highly significant. The R 2 also increases slightly.
In a similar way, we estimate equations [1] to [3] using an alternative estimator IV-2SLS.
Her we take account for the fact that remittances could be endogenous, hence we use an instruments for this variable. Since the major portions of remittances that are sent by the Bangladeshi migrants usually come from Middle Eastern Asian Countries, we chose the GDP per capita of Saudi Arabia as an instrument for remittances. The result is almost similar to the OLS estimates. Without the squared remittances and interactive term, remittances" impact on growth is negative but not significant. When squared remittances variable is added both coefficients of remittances and squared remittances are significant while the former has a negative sign and the latter has a positive. After the interaction term is added onto the regression, remittances seem to positively affect growth though the estimate is statistically not significant.
Finally we provide estimations of equation [1] to [3] using GMM-IV estimator. The instrument used for remittances is same as beforeper capita GDP of Saudi Arabia. The estimated coefficients are identical to those of IV-2SLS however there seem to have been improvements in terms of reductions in standard errors of the estimates. The t-ratios improved for some estimates. Hence some estimates which were insignificant previously are now significant. Remittances negatively affect growth when neither squared remittances term nor the interaction terms are added. When the squared remittances term is added, remittance still negatively and significantly affects growth but the estimated coefficient on the squared remittances term is positive and significant as usual. When the interaction term is added, remittances positively affect growth but the estimated coefficient is not significant.
The main findings are a) There is strong evidence that there is a non-linear relationship between flows of inward remittances and economic growth in Bangladesh. The estimated coefficient on the squared remittances variable is positive and significant in all six specifications where it is included. This implies that inflows of remittances during 1974 -2006 in Bangladesh reduce per capita GDP growth rates in the initial phase but enhance growth rates at a later phase. This could be due to the fact that in the early periods remittances were put to unproductive use (Mahmud and Osmani 1980) whereas in the later periods remittances were utilised for more productive purpose (Siddidui 2004 preferences from consuming out of remittances incomes to investing in health, education and financing micro enterprises. The consumption to investment shift could also stem from overall macroeconomic situation. In the early part of the 1974 -2006 periods, especially from early 70s to late 80s, the economy was plagued with distortions with a relatively closed economy coupled with high tariffs and a repressed financial sector.
Through the initiation structural adjustment programmes and reforms in the real and financial sectors of the economy in the early 1990s these constraints have gradually relaxed. The consequences of such changes in the economic policies mean that the overall macroeconomic situation was conducive to investment and private sector growth in the latter part of the time frame whereas in the earlier part it was not so. Hence it can be argued that the lack of activities of MFIs and NGOs accompanied by an unfavourable investment climate during the earlier phase led the households to put their remittances into unproductive use or into conspicuous consumption. But at the later stage this trend was reversed because of increased MFI and NGO activities along with a favourable macroeconomic environment conducive to investment motivated households towards productive use of their remittances receipts.
There could also be a productivity based explanation for the non-linear effect. Inflow of remittances might have led to the real effective exchange rate appreciation i.e. the Dutch disease effect by squeezing the tradable goods sector and reducing technological capacity in the overall economy and thus reducing growth in the early phase. In the later phase the situation should have been exacerbated due to diminishing returns combined with Dutch disease effect. However, due to favourable investment climate, relative openness of the economy and proliferations of MFIs and NGOs in channelling remittances into productive investments, the Dutch disease and diminishing return effects were perhaps outweighed thru overall productivity gains in the economy at the later phase which contributed to growth. Thus remittances may have a non-linear effect on per capita GDP growth in Bangladesh with growth falling first and then rising later on. b) Out of the nine estimates, remittances" effects on growth had been negative six times.
Of these six estimates, four had been statistically significant. A firm conclusion that remittances negatively affect growth cannot be reached from here because three of these four times, the negative effect remittances had been neutralised by positive effect of the squared remittances variable added in the regression. That is in these four significant estimates, three times the squared remittances variable was included in the speciation whose estimated coefficient was positive and significant. So considering equation [ However these estimates are not significant.

Conclusion and Further Research
This paper consists of a part of my Ph.D. thesis which is macroeconomic impact of remittances on the Bangladesh economy. The "growth -remittances" relationship is a part of the macroeconomic impact of remittances. However the results of this paper have some severe limitations. Firstly, this study is severely constrained by limitation of data.
Only 29 observations based on annual data were available to perform the econometric study. In order for the econometric exercise to be meaningful at least more than 30 observations are required. In later part on my Ph.D. thesis I may have to include other similar economies of South Asia and perform a panel study using country specific dummies to make analysis more informative and meaningful. Secondly, due to unavailability of human capital formation data (gross enrolment ratio) for the time period 1974 -2006, this important conditioning variable was not included in our regression.
Thus we have not adequately controlled for the other factors which are important in the growth regressions. These also include macroeconomic volatility, exchange rate misalignment, and institutional quality. In later part on my thesis I will estimate the macroeconomic volatility and exchange rate misalignment in Bangladesh. There is evidence in the literature that remittances reduce macroeconomic volatility and induce real exchange rate appreciation. Once these channels are included in the growth analysis along with the financial development measure, the true nature and effects of remittances on growth can be unveiled. Based on whatever information, data and model we have now, we can conclude that the effect of remittances on per capita GDP growth of Bangladesh is at best inconclusive. Nonetheless there is strong evidence that this "growth -remittances" effect is non-linear. At early stages of the economy, remittances reduce growth rates but increase growth rates at a later stage. This could be due to unproductive use of remittances in the beginning followed by productive utilization later on. Any negative impact that remittances may have on growth might be potentially mitigated by this non-linearity. On the other hand, it may be possible that remittances positively affect per capita GDP growth in Bangladesh when the complementarity between remittances and financial development is incorporated into the analysis. But firm conclusions can only be made when other important control variables are added onto the analysis and also when the current data set can be expanded either with quarterly frequency or by including panel of other similar South Asian economies.