ISSN (Print) - 0012-9976 | ISSN (Online) - 2349-8846

A+| A| A-

Macroeconomic Determinants of Remittances

Remittances to India have been growing rapidly since 1991, making the country one of the largest recipients of remittances in the developing world. This paper analyses the determinants of remittances to India and finds that their growth over time can be explained by the increase in migration and total earnings of migrants. Remittances are also affected by the economic environment in source countries, and appear to be counter-cyclical - i e, higher during periods of low economic growth in India. None of the remaining economic or political variables considered in the paper, including political uncertainty, interest rates or exchange rate depreciation, are found to affect remittances significantly.

����������������

Macroeconomic Determinants of Remittances Evidence from India

Remittances to India have been growing rapidly since 1991, making the country one of the largest recipients of remittances in the developing world. This paper analyses the determinants of remittances to India and finds that their growth over time can be explained by the increase in migration and total earnings of migrants. Remittances are also affected by the economic environment in source countries, and appear to be counter-cyclical – i e, higher during periods of low economic growth in India. None of the remaining economic or political variables considered in the paper, including political uncertainty, interest rates or exchange rate depreciation, are found to affect remittances significantly.

POONAM GUPTA

IIIII
IntroductionIntroductionIntroductionIntroductionIntroduction

R
emittances from abroad have become a very important component of the balance of payments for developing countries in recent years. For some countries they have exceeded various types of capital flows. Global Development Finance [GDF 2003] shows that remittances to developing countries are higher than official aid flows and are also higher than most other types of private capital flows. Remittances have increased rapidly for India too in the past decade, making it one of the largest recipients of remittances in the world.1

Remittances to India have more than quadrupled between 1991 and 2003, and totalled about US $ 18 billion in 2003. The buoyancy of remittances has been instrumental in substantially reducing the current account deficit in the past few years. They have also been one of most stable flows in the balance of payments accounts of India. Remittances to India have increased at about 13 per cent a year since 1991, making India one of the largest recipients of remittances in the world. The movement of remittances around the trend has been low, and therefore remittances have been the most stable type of external flows in India.

This paper analyses the macroeconomic factors that might explain the dynamics of remittances to India. It finds that the structural factors that may help to explain the buoyancy of remittances during the 1990s, are the increase in the number of migrants to countries such as the US, Australia, and Canada. In particular the number of Indian migrants to the US doubled during the 1990s.2 Evidence also shows that the migration during this period consisted of more skilled people and professionals, and that it was accompanied by a sharp increase in the average earning of the migrants. It is also possible that the growth in measured remittances in the 1990s may have been partly due to informal channels of money transfer rendered less attractive by the exchange rate devaluations in the early 1990s, and the opening up of the capital account. Another possibility is that due to the reduction in duty on the import of gold, the illegal import of gold became less remunerative, resulting in remittances being channelled more through the official routes.

In order to explain the dynamics of remittances around the trend we analyse several economic and political variables, but do not find many risk-return type factors to be important in explaining the behaviour of remittances around the trend. Though there is only weak evidence that remittances are counter-cyclical for India (higher during the years of drought), we find remittances to be higher when economic conditions in the host country are benign.

The paper does not look at the macroeconomic impact of remittances flows or their welfare implications, but the general perception is that they have thus far been beneficial to India. Remittances have been crucial in improving the current account, and in the consequent build-up of foreign exchange in the last few years.3 They are probably also important from the social security point of view by providing a safety net to family members in a non-working-age group. The effect of remittances on output and employment generation would depend on the end-use of the transfers. The effect would be larger if remittances are geared more toward investment expenditure. If remittances are used for consumption, then the stimulation to production would come through the multiplier effect, especially if the economy is operating below capacity.4

The finding that remittances have increased at a trend rate in the last decade or so not only is in tandem with increased migration, that they have been mostly stable implies that in the years ahead the country’s policies in the external sector may need

Figure 1: IMF and RBI DataFigure 1: IMF and RBI DataFigure 1: IMF and RBI DataFigure 1: IMF and RBI DataFigure 1: IMF and RBI Data
Figure 2: RemittancesFigure 2: RemittancesFigure 2: RemittancesFigure 2: RemittancesFigure 2: Remittances

(In US $ million)

14000

6000 12000

5000

10000 8000 4000

6000 3000 4000

2000

2000 0

in US$ milli on

US $ in million

1000 0

1989-901990-911991-921992-93

-

1993-941994-951995-961996-97

-

1997-981998-99

1999-2000

1990:21991:21992:21993:21994:21995:21996:21997:21998:21999:22000:22001:22002:2

2003:2

gold imports

other transfers

transfers (RBI data-gold) '

worker’s remittances, IMF

' RBI

IMF’s BOP

to be adapted to absorb these sustained flows, especially to ensure that they do not create excessive liquidity in the economy or generate inflationary pressures. In particular it may be desirable to let imports ease up to a similar extent as the incremental annual flows of remittances. This would help alleviate inflationary pressures and would also be desirable from the liquidity management point of view. Such a policy may overall be welfare enhancing.

The rest of the paper is organised as follows. Section II discusses issues related to the measurement of remittances and discusses the trends in remittances to India. Section III discusses the literature and the possible determinants of remittances. Section IV discusses and interprets the empirical findings of the paper, and Section V concludes.

IIIIIIIIII
Measurement Issues and MagnitudeMeasurement Issues and MagnitudeMeasurement Issues and MagnitudeMeasurement Issues and MagnitudeMeasurement Issues and Magnitude
of Remittancesof Remittancesof Remittancesof Remittancesof Remittances

Definition and MeasurementDefinition and MeasurementDefinition and MeasurementDefinition and MeasurementDefinition and Measurement

The two main sources of data on private transfers to India are the RBI’s database – Handbook of Indian Economy and the RBI Bulletins, and the IMF’s Balance of Payments Statistics (BOPS). The RBI’s data on private transfers are available for the period 1990-2003, and the IMF’s data are available for 1975-2002.5

The IMF’s data are available separately for two components: worker’s remittances and other transfers. As per the IMF’s BOP manual, “worker’s remittances” include transfers by migrants employed in new economies and considered residents there (i e, they have stayed in the new economy for a year or more). “Other transfers” include charitable and religious contributions (including relief work) and gifts, etc. The two components exhibit very different dynamics overtime.

Disaggregated annual data are also available in RBI publications for the period 1990–2000. In order to get a better handle on different components of remittances, we try to reconcile the IMF data with the RBI data. As Table 1 and Figure 1 show, until 1999, “other transfers” in the IMF data seemed equivalent to the “import of gold” component of the RBI’s data, and “worker’s remittances” included all the other components of the RBI data. However, this practice seems to have changed in 1999, when there seems to be divergence in the respective data series of the IMF and the RBI. In particular, since the gold imports in the remittances data are shown to be zero, and other transfers of the IMF show a rapid increase, it implies that starting in 1999 some other type of remittances are included under this heading.

The lack of clarity and a possible break in the series makes time series analysis of the disaggregated data difficult. Therefore we focus only on total remittances in the paper. We use the IMF data for our analysis since, as Figure 2 shows, the total transfers data of the RBI and the IMF match quite well.

One limitation of the reported data for remittances is that it is probably underestimated because it does not include remittances sent through informal channels. Such transactions are often known as “hawala” [El-Qorchi 2002]. In some countries these

Table 1. Different Components of Remittances Data in RBI’s and IMF’s DatabasesTable 1. Different Components of Remittances Data in RBI’s and IMF’s DatabasesTable 1. Different Components of Remittances Data in RBI’s and IMF’s DatabasesTable 1. Different Components of Remittances Data in RBI’s and IMF’s DatabasesTable 1. Different Components of Remittances Data in RBI’s and IMF’s Databases

(In US $ millions)

Family Gifts and Repatriation Gold and Local Redemptions Other Total Worker’s Other Total
Maintenance Donations of Savings Silver of NRNR Transfers Remittances-Transfers- Remittances-
Deposits RBI Data IMF Data IMF IMF Data
1989-1990 720 405 1,161 0 0 11 2,297 2,291 0 2,291
1990-1991 626 417 1,027 0 0 14 2,084 2,092 0 2,092
1991-1992 702 344 2,738 0 0 13 3,797 3,792 0 3,792
1992-1993 730 445 1,604 1,076 0 9 3,864 3,049 1,181 4,230
1993-1994 514 838 2,241 1,670 0 24 5,287 3,617 1,670 5,287
1994-1995 1,727 587 3,665 2,100 0 33 8,112 6,013 2,100 8,113
1995-1996 1,003 1,359 4,198 1,943 0 37 8,540 6,588 1,948 8,536
1996-1997 2,518 726 1,935 2,718 3,427 1,111 12,435 9,719 2,716 12,435
1997-1998 5,232 526 2,699 3,418 11,875 9,178 2,699 11,878
1998-1999 7,661 650 171 1,859 10,341 10,012 331 10,343
1999-2000 7,423 734 13 4,120 12,290 10,151 2,237 12,388

Sources: RBI Bulletins and the IMF BOPS.

Economic and Political Weekly June 30, 2006

Figure 3a: Annual Remittances to IndiaFigure 3a: Annual Remittances to IndiaFigure 3a: Annual Remittances to IndiaFigure 3a: Annual Remittances to IndiaFigure 3a: Annual Remittances to India
Figure 4: Remittances to Selected CountriesFigure 4: Remittances to Selected CountriesFigure 4: Remittances to Selected CountriesFigure 4: Remittances to Selected CountriesFigure 4: Remittances to Selected Countries

ri gh t scale In per cent 4 3 2 1 0 197519781981198419871990199319961999 private transfers in US $ bn Source: IMF’s BOPS and author’s calculations. 22222 Figure 3b: RemittancesFigure 3b: RemittancesFigure 3b: RemittancesFigure 3b: RemittancesFigure 3b: Remittances (In current, constant US $ million) 5000 4000 3000 2000 1000 0 16000 14000 12000 10000 8000 6000 4000 2000 0 US $ in mln China Mexico Korea India Pakistan 1986 1991 1996 2001 2002 Source: IMF’s BOPS, series on total private current transfers. -----(-5 ) -Figure 5: Remittances and Current Account BalanceFigure 5: Remittances and Current Account BalanceFigure 5: Remittances and Current Account BalanceFigure 5: Remittances and Current Account BalanceFigure 5: Remittances and Current Account Balance 20 10 0 –10 –20 (–7.5) (4.6)(1.4)(–2.9) (–5.9) 2.4 -9.9 13.18.1 -13.9 12.3 14.7 -16.0 -10.9 -10.1 as per cent of GDP

16000 12000 8000 4000 0

US $ in mln

1990

1995

2000

2001

2002

Current account balance except for remittances

Remittances

Source: IMF’s IFS database.

reflection of increased globalisation and the associated move

1990:21991:21992:21993:21994:21995:21996:21997:21998:21999:22000:22001:22002:22003:2

in current US $

in constant US $

ment of people. However, the increase in remittances to India

Source: IMF’s BOPS and author’s calculations.

flows are estimated to be very high. For India estimates have put the remittances through hawala at about US $ 6 billion a year [Reddy 1997].

Finally, if the remuneration to bring money through hawala changes over time, this would result in discrete jumps in the reported figures on remittances, making the time series data less comparable over time. This is likely to happen for instance, if the official exchange rate is aligned more closely with the shadow exchange rate; or if the rules on declaration of remittances become more stringent, thus inducing the agents to remit funds through the official channels. This probably happened in India in the 1990s and in the post-September 11, 2001 period (we include exchange rate depreciation and a dummy for the period after the September 11, 2001 to account for these events in the regressions later).

Magnitude and Trend of RemittancesMagnitude and Trend of RemittancesMagnitude and Trend of RemittancesMagnitude and Trend of RemittancesMagnitude and Trend of Remittances

Private transfers have become an important component of the balance of payments in India since 1991. Figure 3 shows that the transfers increased steadily during the 1970s, remained more or less flat in the 1980s and picked up sharply in the 1990s. The sharpest increase in transfers took place during 1991-97. Remittances to India more than quadrupled between 1991 and 2003, and totalled about US $ 18 billion in 2003, making India one of the largest recipients of remittances. In terms of percentage of GDP, remittances equalled about 3 per cent in 2003.

Remittances have increased in tandem with the increase in remittances to other developing and emerging countries, a has been somewhat sharper than that to many other countries (Figure 4). Another country where the remittances have increased at a similar pace in the past decade is China. India accounted for about 10 per cent of total remittances to developing countries, and about 25 per cent of total remittances to Asian countries in 2002.

Remittances have been an important component of India’s current account, accounting for about half of the receipts on invisibles and 20 per cent of the total receipts in the current account in 2002. Increased remittances, coupled with an improved trade balance, have been instrumental in the recent improvement of the current account of India.

One issue which often figures in the discussion of external flows is the sustainability or volatility of these flows. An important feature of remittances is that they have proven to be one of the most stable forms of external flows to India, on the current as well as on capital account. Jadhav (2003) shows that the volatility of remittances is lower than that of NRI deposits or portfolio flows. In addition, we find that remittance receipts are also substantially less volatile than exports of goods and services.

IIIIIIIIIIIIIII
Literature and Determinants of RemittancesLiterature and Determinants of RemittancesLiterature and Determinants of RemittancesLiterature and Determinants of RemittancesLiterature and Determinants of Remittances

In the literature several issues related to international migration, and more specifically issues related to remittances have been studied, such as estimating the impact of migration on the domestic economy (in terms of lost human capital or tax revenue); analysing the incentives behind remittances (for support of family, or investment purposes); and assessing the effects of the remittances on the native country (effects on the balance of payments and growth).

Figure 6: Trend Growth of RemittancesFigure 6: Trend Growth of RemittancesFigure 6: Trend Growth of RemittancesFigure 6: Trend Growth of RemittancesFigure 6: Trend Growth of Remittances

9.0

8.5

8.0

7.5

7.0

6.5

6.0

5.5

5.0

= y = 0.03 T + 6.75 R2 = 0.80

1990:11991:11992:11993:11994:11995:11996:11997:11998:11999:12000:12001:12002:12003:1

in log US $

The literature broadly distinguishes between an altruistic motive to remit earnings to the migrant’s native country (mostly for consumption by the family), and remittances sent to either invest in the native country or to repay previously borrowed funds.6

In order to analyse the dynamics of remittances one can think of an optimising framework whereby a migrant maximises his utility by choosing the optimal level of his own consumption, remittances to family in his native country for their consumption needs, and investment in various available instruments in the native country as well as in the host country. Remittances to support family members at home would depend on the income of the migrants, and on the needs and income of the beneficiaries.7 Remittances for investment (in deposits, property, stocks, etc) would be influenced by risk-return considerations. Determinants of remittances such a framework imply are: income of the migrant, economic conditions in the native country (migrants are likely to remit more during the periods when their family’s income is low); return factors including domestic interest rates, interest rates abroad, and return in the stock market or return on property; and the risk of default, which could be proxied by domestic political uncertainty, geopolitical conditions, or rating downgrades.

The main results established in the literature are: remittances are motivated more by an altruistic motive than by an investment motive; remittances are counter-cyclical, i e, higher under adverse economic outcomes in the native country; they are used more for consumption than for investment; and they do not respond much to relative rates of return on investments in the home country.8

Explanatory VariablesExplanatory VariablesExplanatory VariablesExplanatory VariablesExplanatory Variables

In the econometric exercise below we include movements in US employment (non-agricultural employment), LIBOR, or oil prices, as proxies for the economic environment in the host countries. For economic conditions in India we consider variables such as industrial growth, a dummy for drought years (defined as a year when agricultural growth is negative), or return on the Bombay Stock Exchange (BSE). We also include agricultural or GDP growth rates, though quarterly data are available only since 1997.

For risk factors we include dummy variables for rating downgrades by leading credit rating agencies, for governments

Table 3. Quarterly Average of Remittances duringTable 3. Quarterly Average of Remittances duringTable 3. Quarterly Average of Remittances duringTable 3. Quarterly Average of Remittances duringTable 3. Quarterly Average of Remittances during
Specific PeriodsSpecific PeriodsSpecific PeriodsSpecific PeriodsSpecific Periods

Period and Duration Remittances (HP Filtered)

All -6.89

Drought 1991:2-1992:1, 1995:2-1996:1,

1997:2-1998:1, 2000:2-2001:1,

2002:2-2003:1 65.3 Rating 1991:2, 1997:1, 1997:4, 1998:1,2,4;

2000:4, 2001:3,4 -35.7 Asia 1998:1-1998:4 -315.8** Geo-Polt 2002:1 42.6 S11 2001:4, 2002:3 48.3

Notes: Calculated for HP residuals. ** indicates significance at 5 per cent levels respectively.

Table 4: Results of Unit Root TestsTable 4: Results of Unit Root TestsTable 4: Results of Unit Root TestsTable 4: Results of Unit Root TestsTable 4: Results of Unit Root Tests

D-F Statistic P Value

Remittances (in constant US$) -2.38 0.39 HP filtered remittances -4.2 .00 Earnings of migrants (in constant US$) -2.69 0.24 US non-agricultural employment -2.34 0.40 US non-agricultural employment

(in per cent change) -1.96 0.29 Oil prices (in constant US$) -2.84 0.18 Oil prices (in per cent change) -4.03 0.00 LIBOR (quarterly change) -2.48 0.12 Nasdaq (in per cent change) -3.03 0.03

TableTableTableTableTable
2: Correlation Coefficients between Dependent and Independent Variables2: Correlation Coefficients between Dependent and Independent Variables2: Correlation Coefficients between Dependent and Independent Variables2: Correlation Coefficients between Dependent and Independent Variables2: Correlation Coefficients between Dependent and Independent Variables
11111

Remit-US Employ-Oil Nasdaq: FII LIBOR: Exchange Drought BSEC: RIB S11 Asia Rate tances: HP ment: Price: Per Change Rate: Per Cent Filtered Per Cent Per Cent Cent Lagged Change Change Change Change Per Cent Change

Remittances: HP filtered 1 US employment 0.24 1 Oil price 0.05 0.15 1 Nasdaq 0.09 0.21 -0.20 1 FII 0.22 0.70 0.37 0.11 1 LIBOR 0.27 0.32 0.06 0.08 0.25 1 EXCC(-1) 0.01 -0.11 -0.01 -0.07 -0.22 -0.30 1 Drought 0.18 -0.17 -0.02 -0.16 -0.23 -0.09 0.23 1 BSEC 0.15 -0.07 0.20 0.23 0.08 0.13 -0.06 -0.05 1 RIB -0.14 0.16 -0.09 0.01 -0.03 -0.32 0.04 -0.19 -0.17 1 S11 0.08 -0.44 0.02 -0.27 -0.16 -0.20 -0.07 0.07 -0.08 -0.07 1 Asia -0.18 0.19 -0.23 0.01 -0.03 -0.33 0.14 -0.08 -0.18 0.86 -0.08 1 1 Rate -0.01 -0.12 -0.32 -0.11 -0.21 -0.29 0.07 0.06 -0.14 0.32 0.06 0.44 0.??

Note: 1A coefficient higher than .23 is significantly different from zero at a 10 per cent level. Source: Author’s calculations.

Economic and Political Weekly June 30, 2006

Figure 7: Total Migration to US and RemittancesFigure 7: Total Migration to US and RemittancesFigure 7: Total Migration to US and RemittancesFigure 7: Total Migration to US and RemittancesFigure 7: Total Migration to US and Remittances
Figure 8: Total Earnings of Indians in US and RemittancesFigure 8: Total Earnings of Indians in US and RemittancesFigure 8: Total Earnings of Indians in US and RemittancesFigure 8: Total Earnings of Indians in US and RemittancesFigure 8: Total Earnings of Indians in US and Remittances

16

50

16

1000

40

800

1212 30

600

88

= Earnings = 4.26 t + 7.54 R2 = 0.91

Migration = 72798t + 328036

400

US $ in billion

US $ in billion

US $ in billion

In thousands

20 4R2 = 0.89

=

104

200

0

0

0

0

1994 1995 1996 1997 1998 1999 2000 2001

19941995199619971998199920002001

-----
-----
-----
-----
Total earnings of Indians aged 18-64 in the US

-

Indians aged 18-65 in the US Remittances

Source: Desai et al (2001).

resigning mid-term, and for periods of geopolitical tensions on the border with Pakistan.9 We also include a dummy for the Asian crisis period. This period also coincided with the issuance of the Resurgent Indian Bond (RIB) yielding an attractive interest to Indians abroad.

Since remittances are likely to be higher during the periods of festivals or marriages, we also test for the robustness of our results by including a separate dummy for the October-December and April-June periods as these coincide with either the period of major festivals or with the auspicious months of the wedding season (we also tested for seasonality of the quarterly remittances data but found that the data do not exhibit a seasonal pattern).

In addition, we control for a dummy variable for the post-September 11, 2001 period in the regressions to reflect the effect of the strengthening of regulations and a clampdown on hawala transactions after September 11.

Since a depreciation of the currency would render remittances more profitable, align the official exchange rate closely with the black market exchange rate, or even raise expectations of an appreciation in the future, it would probably increase remittances sent through the official channels. However, since there is a potential endogeniety in the depreciation and remittances variables (more remittances would imply stronger rupee), we include lagged values of the exchange rate depreciation. Further details

Table 5: Regression Results for the Level of RemittancesTable 5: Regression Results for the Level of RemittancesTable 5: Regression Results for the Level of RemittancesTable 5: Regression Results for the Level of RemittancesTable 5: Regression Results for the Level of Remittances

I II III IV

C 2.25*** 3.16 2.17 -1.5

(3.07) (2.06) .96 (-1.06) Trend .008*** .001 -.00 -.004

(3.01) (.33) -.64 (-.82)

Lagged dependent variable .67*** .13 .03

(6.38) (.64) (.16) Earnings of migrants in the US .58*** .65 -.69

(2.74) (-1.08) (-1.2) Lagged earnings of migrants 1.25** 1.29** in the US (1.99) (2.10) Oil prices .39 (.36) Oil prices lagged .57

(.50) Number of observations 53 29 28 29 R2, Adj. R2 .84,.83 0.50; 0.46 .60;.54 0.55; 0.50

Note: Dependent variable is log of remittances (in constant US$); log of total estimated earnings of the Indian migrants in the US (in constant US$). Estimates obtained from OLS, with heterosckedasticity consistent standard errors. Since the variables are I(1) their lagged values are included in the regression.

——— transfers

Source: Desai et al (2001).

on data sources and on construction of variables are provided in the Appendix.

Bivariate Association between Transfers andBivariate Association between Transfers andBivariate Association between Transfers andBivariate Association between Transfers andBivariate Association between Transfers and
Other VariablesOther VariablesOther VariablesOther VariablesOther Variables

Remittances transfers are not found to be correlated significantly with most of the variables, except US non-agricultural employment (Table 2). (The coefficients are also found to be insignificant for most of the variables in multivariate regressions.) We also do not find transfers during some of the events – such as geopolitical tensions, or the aftermath of September 11, to be significantly different than during the rest of the period (Table 3). Remittances were lower during the Asian crisis; however since this period also coincides with other events, including the issuance of the RIB, the effect of the Asian crisis cannot be isolated.

IVIVIVIVIV
Econometric Specification and EmpiricalEconometric Specification and EmpiricalEconometric Specification and EmpiricalEconometric Specification and EmpiricalEconometric Specification and Empirical
ResultsResultsResultsResultsResults

Time Series PropertiesTime Series PropertiesTime Series PropertiesTime Series PropertiesTime Series Properties

We first test the time series properties of the variables, which in turn would determine the regression specification. The dependent variable was found to be an I(1) process (test for unit root was conducted for the specification including a constant and a trend). Similarly, the variables related to migration were found

Table 6: Regression Results for Cyclical ComponentTable 6: Regression Results for Cyclical ComponentTable 6: Regression Results for Cyclical ComponentTable 6: Regression Results for Cyclical ComponentTable 6: Regression Results for Cyclical Component
of Remittancesof Remittancesof Remittancesof Remittancesof Remittances

I II

C -.04 (-1.22) -.076** (-1.98)
Per cent change in non-agricultural
employment in the US .13** (2.04)
Per cent change in oil prices -.001 (-.59) -.001 (-.44)
Drought .082* (1.89) .057 (1.15)
Change in LIBOR .10* (1.74)
Asian crisis -.11** -(2.12) -.16* (-1.67)
Rho .17 (1.16)
Numbers of observations 53 54
R2, Adj. R2 .13;.06 .16;.07

Note: Dependent variable is HP filtered series of log of remittances (in constant US $); in column I above an OLS specification (with heteroskedasticity consistent errors were used); in column II the OLS specification indicated presence of autocorrelation, therefore an AR1 specification is used.

to be an I(1) process. However, the test ruled out the presence of a cointegrating relationship between these two variables. Other variables measured in levels, such as oil prices, US employment were also found to be I(1) processes, but the tests ruled out cointegrating relationships between these variables and remittances. However, the variables which were measured in first differences, as percentage changes, or as deviation from a HP trend were found to be I(0) processes (Table 4).

Econometric SpecificationEconometric SpecificationEconometric SpecificationEconometric SpecificationEconometric Specification

We estimate the following two specifications of the regression equation:

Transt = c + a trendt +∑ iXit + £t, £t ≈ N(0, a2), t = 1,2,..,T (1)i

Res1 = c +∑ iXit + £t, £ t ≈ N(0, a2), t = 1,2,..,T(2)

i

Our dependent variable is measured in constant US dollars. In the first specification (equation 1) we regress remittances on a linear trend and a set of explanatory variables in order to explain the trend in the series. In the second specification (equation 2) we estimate the de-trended series for transfers (the HP filtered series). The estimates are obtained either through ordinary least squares, with heteroskedasticity consistent standard errors; or if there is autocorrelation through an AR1 specification.10

Linear TrendLinear TrendLinear TrendLinear TrendLinear Trend

Remittances exhibit a strong linear trend (Figure 6). The factors that may explain the buoyancy of remittances over time include increase in the pool of Indians settled abroad, and the fact that the migration consists increasingly of more skilled people, and therefore people with higher average earnings. As mentioned earlier, the spurt in the early 1990s may also be due to the informal channels of money transfers being rendered less attractive with the realignment of the exchange rate in the early 1990s, the opening up of the capital account; and the reduction in duty on import of gold, which probably made illegal import of gold less remunerative.

The results for specification in equation 1 are reported in Table 5. The first column shows that the series for remittances exhibits a very strong linear a trend. In order to explain the trend we include either the number of migrants (we use the data on migrants in the US, and interpolate the quarterly series from annual data, measured in log) or their average earnings abroad (proxied by the average earnings in the US, quarterly data interpolated from annual data). Only few observations are available for these variables, nevertheless, results (column 2 Table 3) show that the migration/earnings of migrants explain the trend in remittances. We also include oil prices, however its coefficient is insignificant. Since the dependent and independent variables in levels are all I(1) processes, with no conintegrating vector, we include lagged values of all the variables in the regressions.

Regression Results for HP Filtered SeriesRegression Results for HP Filtered SeriesRegression Results for HP Filtered SeriesRegression Results for HP Filtered SeriesRegression Results for HP Filtered Series

Next we estimate the equation for HP filtered series for remittances. In the regressions (Table 6), the variables found to be significant are either variables which measure the state of the US economy (non-agricultural employment, changes in labour, oil prices, and the return on Nasdaq are used as proxies for the source country’s business cycle conditions), or that of the Indian economy. Remittances are found to be higher when economic conditions in the source country are benign. Results also show that remittances are higher during drought years, though this result is somewhat weaker, as it is not obtained in all the specifications.

The only other variable we find to be important in the regressions is the dummy for the period around the Asian crisis, which has a negative coefficient and is significantly different from zero. This may be due to the uncertainty around the Asian crisis, or

AppendixAppendixAppendixAppendixAppendix
Data Sources and DefinitionsData Sources and DefinitionsData Sources and DefinitionsData Sources and DefinitionsData Sources and Definitions

Variable Name Definition/Construction of Variable Source

Remittances Private transfers on current account, expressed in constant US$ IMF, BOPS Drought Dummy takes a value 1 if it is a drought year – i e, agricultural growth Constructed using the data on

is negative agriculture growth LIBOR, change in LIBOR 12 month LIBOR in US$, change in LIBOR IFS Return on Nasdaq Quarterly return on Nasdaq, per cent change in Nasdaq Constructed using the data from IFS Asian crisis Dummy takes a value 1 for the quarter in which crisis occurred in Asia ConstructedusingIFSexchangerate data Per cent change in oil prices Per cent change in oil prices in constant US$ Calculations using data from IFS S11 dummy Dummy variables take a value one for four quarters after September 2001 Constructed Return on BSE M-o-m percentage return in BSE index in dollar terms Handbook of Statistics on the Indian

Economy, RBI Exchange rate change Quarterly percentage change in exchange rate with respect to US dollar IFS Political uncertainty Dummy equals 1 in the quarters during which the central government Dow Jones Newswire

resigned mid-term Geopolitical tensions Dummy takes a value 1 for the quarters of Kargil war, nuclear tests, and Dow Jones Newswire border stand-off in summer 2002. Rating changes Dummy equals one for the quarters in which the rating/outlook is revised Constructed using information from down, and the following month S and P, Moody’s Issuance of RIB, IMD bonds Dummy equals one for the quarters in which RIB, IMD were issued Constructed

Economic and Political Weekly June 30, 2006

due to expectations of a depreciation of the rupee. Around the same time the RBI had floated a high yielding RIB to attract investments from non-resident Indians. The negative effect on remittances during the Asian crisis may also be due to the diversion of remittances into these bonds. However, this explanation is probably not valid because we do not find the interest rate on NRI deposits, or changes in policies regarding NRI deposits (especially since 2002) to be correlated significantly with the behaviour of remittances.

Remittances are also not found to be affected by the movement of oil prices. This is despite the fact that a non-trivial percentage of remittances to India originate from the Gulf countries. This could possibly be due to the fact that oil price changes have offsetting effects – higher oil prices may result in higher income and more remittances from the oil producing countries in the Gulf, but may negatively impact the economies in other source countries, and lower remittances originating from there.

Several other economic and political variables were included in the regressions but their coefficients were not found to be important, including rating changes, return on the domestic stock market, and exchange rate changes. Finally, we did not find exchange rate variables to be significant, and neither did the period since September 11, 2001 witness any unusual pattern in remittances.

VVVVV
ConclusionConclusionConclusionConclusionConclusion

This paper analysed the recent behaviour of remittances to India. It finds that commensurate with the increase in the number of migrants from India and the migration of high skilled workers over time, private transfers to India on the current account have been very robust in the past decade. The paper also finds that private transfers have been a stable source of funds and have not been affected by risk-return considerations to the same extent that flows on capital account have been, such as portfolio investment or even NRI deposits. Thus they have proved to be a source of strength in the balance of payments in India.

The econometric analysis shows that not many macroeconomic factors are important in explaining the behaviour of remittances around the trend over time. Among the variables that are found to be significantly associated with the movements in remittances include indicators of economic activity in the source countries. Remittances are higher when economic conditions abroad are benign, and remittances are also found to be somewhat countercyclical, i e, higher during the periods of negative agriculture growth.

The paper does not look at the contribution of remittances to the economic development or their welfare implications. For this issue to be addressed fully, it may be useful to examine more disaggregated data.

m

Email: pgupta@imf.org

NotesNotesNotesNotesNotes

[This paper was written when the author was at the resident representative office of the IMF in New Delhi, India. I would like to thank Ashoka Mody, Sudip Mohapatra, Helene Poirson, Catriona Purfield and Jerry Schiff for their helpful comments.]

1 Remittances (also known as current transfers) include worker’s remittances and other private transfers on the current account.

2 See Desai et al (2001).

3 Since some of the remittances are likely to filter out of the economy through higher imports, the net impact on current account balance is perhaps smaller than the total flow of remittances.

4 Though we are unaware of any study on India that looks at the end-use of remittances, evidence from other countries shows that remittances are mostly used for consumption and for investment in land and property.

5 We have used the data for gross transfers (i e, we did not net out remittances paid) in the analysis, however, since the transfers paid are very small, the net and gross transfers do not differ much in India.

6 See Chami et al (2003) for a detailed discussion of these issues.

7 Researchers have used either the household level data [e g, Lucas and Stark 1985] or the aggregated macro level data [e g, Chami et al 2003; Straubhaar 1986] to analyse the possible determinants and effects of remittances.

8 Evidence on the contrary is found in Straubhaar (1986), who finds remittances to Turkey to be sensitive to temporary domestic political instability. He also finds that remittances do not respond strongly to the incentives offered to migrants to remit.

9 Similar variables have been used by Gupta and Gordon (2003, 2004) to analyse the determinants of NRI deposits and portfolio flows.

10 Jadhav (2003) analyses the determinants of workers remittances to India. Using a log linear regression specification, he includes oil prices, US GDP, an interest rate variable (difference between NRI interest rate and LIBOR) and exchange rate depreciation as the explanatory variables. He finds remittances to be associated positively with the oil prices and an exchange rate depreciation. The analysis in this paper differs from Jadhav (2003) in many important aspects. First, we use either stationary variables or include lagged values of the I(1) variables in the regressions in order to eliminate the problem of spurious regressions. Second, we use a more complete specification by including a trend and/or the variables on the RHS, which may explain the trend behaviour in remittances. Finally, we include a somewhat more comprehensive set of explanatory variables.

ReferencesReferencesReferencesReferencesReferences

Chami, R, C Fullenkamp and S Jahjah (2003): ‘Are Immigrant Remittance Flows a Source of Capital for Development?’, IMF Working Paper 03/ 189, International Monetary Fund, Washington DC.

Desai, M A, D Kapur and J McHale (2001): ‘The Fiscal Impact of the Brain Drain: Indian Emigration to the US’, draft.

El-Qorchi, M (2002): ‘Hawala’, Finance and Development, Vo 39, No 4, December, International Monetary Fund, Washington DC.

GDF (2003): ‘Worker’s Remittances: An Important and Stable Source of External Development Finance’, Chapter 7, pp 157-75, World Bank, Washington DC.

Gordon, James P, Poonam Gupta (2003): ‘Portfolio Flows to India: Do Domestic Fundamentals Matter?’, IMF Working Paper 02/42, International Monetary Fund, Washington DC.

– (2004): ‘Non-Resident Deposits in India: In Search of Return?’, Economic and Political Weekly, Vol 39, No 37, September 19, pp 4165.

Jadhav, Narendra (2003): ‘Maximising Developmental Benefits of Migrant Remittances: The Indian Experience’, paper presented at the joint conference of DFID-World Bank, London.

Lucas, R E B and O Stark (1985): ‘Motivations to Remit: Evidence from Botswana’, Journal of Political Economy, Vol 93, No 5. Reddy, Y V (1997): ‘Capital Flight: Myths and Realities’, address at Centre for Economic and Social Studies, Hyderabad, June 21. Straubhaar, T (1986): ‘The Determinants of Worker’s Remittances: The Case of Turkey’, Weltwirtschaftsliches Archive, 122 (4), pp 728-40.

Economic and Political WeeklyEconomic and Political WeeklyEconomic and Political WeeklyEconomic and Political WeeklyEconomic and Political Weekly

available at

Alter Media BookshopAlter Media BookshopAlter Media BookshopAlter Media BookshopAlter Media Bookshop

Brahmaswom Madom Bldg, M G Road, Trissur 680 001, Kerala Email: info@altermediaindia.com

To read the full text Login

Get instant access

New 3 Month Subscription
to Digital Archives at

₹826for India

$50for overseas users

Comments

(-) Hide

EPW looks forward to your comments. Please note that comments are moderated as per our comments policy. They may take some time to appear. A comment, if suitable, may be selected for publication in the Letters pages of EPW.

Back to Top