This paper analyzes economic developments in El Salvador during 1990–97. The paper assesses the prospects of the Salvadoran economy in facing the new challenges coming from its reinsertion into the global economy. The paper describes the evolution of trade competitiveness and evaluates the divergence from equilibrium of the real effective exchange rate. It concludes that the exchange rate behavior in the last decade appears to have followed roughly the path predicted by a long-term equilibrium exchange rate model consistent with a current account deficit of about 2 percent of GDP.


This paper analyzes economic developments in El Salvador during 1990–97. The paper assesses the prospects of the Salvadoran economy in facing the new challenges coming from its reinsertion into the global economy. The paper describes the evolution of trade competitiveness and evaluates the divergence from equilibrium of the real effective exchange rate. It concludes that the exchange rate behavior in the last decade appears to have followed roughly the path predicted by a long-term equilibrium exchange rate model consistent with a current account deficit of about 2 percent of GDP.

IV. Workers’ Remittances: Trends and Prospects 7

A. Introduction

40. Workers’ remittances have been an important source of foreign exchange for El Salvador since 1990, having contributed to raise the standard of living of an important section of the population. After a sharp increase in workers’ remittances in relation to GDP during 1990-95, the upward trend has slowed down and it is not expected to recover because of an anticipated slowdown in emigration and the negative impact resulting from the settlement and assimilation of earlier emigrants in the host country. In this context, it is important to identify the factors that explain the behavior of remittances in order to assess the prospects for these flows in the future. This is particularly relevant in the context of the impact of remittances on the long-run real equilibrium exchange rate.8 This section develops a model of workers’ remittances for El Salvador over the period 1976-96.

B. Determinants of Workers’ Remittances

41. The literature does not offer a comprehensive and systematic theory to explain the behavior of remittances. However, two main frameworks have been developed in the context of empirical research to model remittances. 9 One approach makes migration and remittance decisions endogenous to the family. Therefore, in this approach the decision process is sequential, determining first the overall level of savings of the migrant and then the migrant’s propensity to remit to the home country. This approach uses socio-demographic and income variables such as the composition of the family at home and abroad, the migrant income in the host country, family income in the country of origin, and the length of stay abroad.

42. The second approach seeks to explain workers’ remittances as a transfer of migrant savings from the host to the home country. Therefore, remittances would be the result of a portfolio allocation decision by the migrant worker, which will depend on incentives offered by the home country relative to the host country such as relative interest rates, relative prices, expectations and uncertainty.

43. The empirical evidence corroborates that demographic and income are significant determinants of workers’ remittances, but is less conclusive regarding the sensitivity of remittances to relative rates of return in the home and host country.

C. Workers’ Remittances to El Salvador in 1976-96

44. Emigration from El Salvador, mainly to the United States and mostly illegal, increased gradually in the 1970s and quickly intensified in the 1980s, pushed by a forced displacement of rural population and deteriorating economic conditions resulting from the civil war. According to demographic indicators of the Ministry of Planning (MIPLAN), in the 1970s about 250 thousand citizens left El Salvador compared with about 550 thousand in the 1980s and less than 100 thousand between 1990-96. However, it is only in the 1990s that recorded remittances reached high levels, increasing from US$26 million in 1976 to about a US$1 billion in 1996. They averaged US$780 million a year (10 percent of GDP or 60 percent of exports of goods and nonfactor services) over 1990-96, compared with US$125 million a year (3 percent of GDP or 13 percent of exports of goods and nonfactor services) in the 1980s (Table 1).

Table 1.

El Salvador: Workers’ Remittances, 1976-96

article image
Sources: Central Reserve Bank; and Fund staff estimates.

For an explanation on the estimates of these time series see Note I.

The nominal data were deflated using the U.S. CPI (1990=100).

45. One possible factor behind the sizable increase in remittances during 1990-96, 10 is the effect that the legalization of a large number of undocumented Salvadorans living in the United States had on securing job protection and in reestablishing links with the home country. Two important reforms to the immigration laws of the United States 11 allowed for estimated temporary or permanent regularization of the status of 340,000 Salvadorans in the United States and the ensuing job protection to those migrants. With the new acquired legal status, formerly undocumented Salvadorans were able to visit the home country and find investment opportunities in a country that had also ended the civil strife.12

Model specification and results

46. The model:



article image

The coefficients for SW, rear, and dum were expected to be positive, while the coefficient for t (proxy of length of the stay) was expected to be negative as the aging of the migrant’s population tends to weaken the ties with the home country.

The econometric estimation was based in the following two equations: the first was an unrestricted equation and the second equation was reformulated as an error-correction model.


Where, ln and D ln are the logarithm and the differences of logarithms for a particular variable.

47. These equations were estimated by least ordinary squares using annual data for the 1976-96 period.13 The results are shown in Table 2, the t statistics are shown between parentheses under each coefficient.

Table 2.

Parameters Estimates for the Workers’ Remittances Model

article image

48. In general, the specified equations shown in Table 2 provided good results. The coefficients of all the variables had the expected signs and were statistically significant, with higher long-term elasticities as expected. Both coefficients of variables representing demographic factors, the stock of workers, and the proxy for length of the stay were statistically significant. The estimated coefficient for stock of work was positive and for the length of the stay negative as expected. However, with a more significant impact of a change in the stock of workers than length of the stay on a change in workers remittances. The economic variable, real income in the host country also had the expected sign, with the value of the short run of real income in the host country similar to the values obtained by other studies on remittances. For example Swamy (1981) reported income elasticities in the range of 1.4-1.3. Finally, the dummy variable had a positive and significant impact on remittances for the period specified.

D. Conclusions

49. The findings in this section corroborate the results obtained in other studies on remittances, in which the demographic variables have a significant impact on the behavior of workers’ remittances. In particular, the positive and significant effect of the stock of migrant labor and the declining propensity to remit associated to the duration of the migration. The estimates presented also support the hypothesis that economic activity in the host country has an important effect on the level of remittances, and that changes in the U.S. immigration laws might had a significant impact on remittances over 1990-96.

50. The results suggest also that prospects for a continued growth of these flows at the rates observed over 1990-95 are low. It is more plausible to expect that the level of remittances will stabilize in nominal terms and fall in real terms because the growth of the stock of migrant workers is bound by the deceleration of migration from El Salvador in the early 1990s, and by the effect that more restrictive U.S. immigration policies will have on future migration flows. Also, the legalization of the status of migrant Salvadorans in the United States is a once-for-all effect on the level of remittances, which will tend to weaken after these migrants become permanent residents of the United States. With a stagnant stock of workers, the effect of a declining propensity to remit as the duration of migration increases will tend to be the dominant factor.

Note I. Data Sources on Workers’ Remittances

Official data

1. Official data on workers’ remittances are prepared by the central reserve bank (BCR). The sources and methods of calculation have changed over the last three decades. In general, the objective has been to quantify money transfers and consequently, the series does not include remittances in kind (in the form of consumer durables and other goods).

2. In the 1970s, the source of information was the foreign exchange records of commercial banks. However, this source lost reliability in the late 1970s due to exchange controls and episodes of multiple exchange rates, with transactions shifting to the parallel market.

3. Consequently, the BCR changed the quantification method in the 1980s because it could no longer rely on the foreign exchange records of commercial banks. For the period 1980-86, the BCR used the 1979 number as a benchmark multiplying it by the average rate of growth of workers’ remittances in 1977-79. After 1987, the BCR developed a new methodology based on migration and economic factors that were applied to the 1979 benchmark, and revised the series back to 1980. To estimate the growth rates for migration, the BCR used informed guesses on the stock of Salvadorans living in the United States (based on estimates of the U.S. Immigration and Naturalization Service for 1979 and the U.S. embassy in El Salvador for 1985 and 1987) and assumed decreasing rates of emigration according to the intensity of the internal conflict, which resulted in higher emigration rates for the early 1980’s. In addition, the BCR assumed that three economic variables had a direct effect on remittances: the level of interest rates on saving accounts, the change in the parallel market exchange rate, and the rate of inflation. The effect of the first two variables was assumed to be positive and negative for the latter variable. To estimate the effect of these economic variables on the growth of remittances, the three rates were added up and the combined rate used to adjust the emigration rate estimated previously to arrive to a net increase in remittances.

4. Starting in 1991, the BCR resumed the use of information on foreign exchange transactions as reported by commercial banks and nonbank financial institutions. Additionally, it began to use information on workers’ remittances reported by the exchange houses, which were authorized to start operations in April 1990.

Estimations of workers’ remittances for the 1979-90 period

5. Remittances started to grow slowly in the mid-1970s and much more rapidly in the 1980s (see Table 1). However, based on more comprehensive migration data and the evolution in the flows after 1990—when a more reliable source of information was available—the official data on workers’ remittances for 1979-90 appear to be underestimated. For this reason, the series for that period has been adjusted using the following data and assumptions: (a) net emigration information from the official migration statistics; (b) the stock of migrants for 1979 was assumed to be the cumulative yearly flow starting from 1950, which takes care of a possible underestimation of the official migration data; (c) the distribution by age given by the statistics on migration was used to estimate the workers among the migrant population, corrected by the migrant population of previous years that were projected to become inactive or active on account of age; and (d) using the value of remittances for 1978 and the stock of migrant workers estimated for 1978 as explained in (b) above, an average remittance by migrant worker was estimated for 1978. To calculate the new series of workers’ remittances, the average remittance per migrant worker estimated for 1978 (adjusted for the variation of the GDP deflator of the United States) was multiplied by the stock of migrant workers estimated for a given year. The new series gives a higher level of remittances for the period 1980-90.

Data sources

The data sources used for the estimated model are the following:

Remittances (re): for the real value of remittances, the source was the BCR data for the period 1976-78 and 1992-96. For the period 1979-90, remittances were recalculated as described above. These flows were deflated by the U.S. CPI (1990=100).

Migrants’ income in the host country (rear): measured as a weighted average of the weekly earnings of the construction and service sectors in the United States deflated by the U.S. CPI.

Stock of workers abroad (SW): estimated on the base of emigration data published by MIPLAN.

Length of stay (t): a proxy consisting of a simple linear trend was used.

Dummy variable (dum): a dummy variable with a value of one for the period 1991-96, and zero elsewhere.


  • Banco Central de Reserva de EI Salvador, 1993, Recopilación de Cifras de las Remesas Familiares en El Salvador, (mimeo).

  • Elbadawi, Ibrahim A. and Rocha , Roberto de Rezende, 1992, Determinants of Expatriate Workers’ Remittances in North Africa and Europe, World Bank Policy Research Series WPS 1038, November 1992.

    • Search Google Scholar
    • Export Citation
  • Fundacion Salvadorefia para el Desarrollo Económico y Social (FUSADES), 1995, ElManejo de las Remesas en El Salvador, Boletín Economico y Social No. 111.

    • Search Google Scholar
    • Export Citation
  • Funkhouser, Edward , 1992, Mass Emigration, Remittances, and Economic Adjustment: The Case of El Salvador in the 1980s, Immigration and the Work Force: Economic Cosequences for the United States and Source Areas. Eds. George J. Borjas and Richard B. Freeman. Chicago: University of Chicago Press.

    • Search Google Scholar
    • Export Citation
  • Hagan, Jacqueline and Baker, Susan Gonzalez, 1993, Implementing the U.S. Legislation Program: The Influence of Immigrants Communities and Local Agencies on Immigration Policy Reform, International Migration Review, Vol. 27, No. 3: 51336.

    • Search Google Scholar
    • Export Citation
  • Lianos, Theodore, 1997, Factors Determining Migrant Remittances: The Case of Greece, International Migration Review, Vol 31, No. 1: 7287.

    • Search Google Scholar
    • Export Citation
  • Russell, Sharon Stanton , 1986, Remittances from International Migration: A Review in Perspective, World Development 14:6; June: 677696.

    • Search Google Scholar
    • Export Citation
  • Swamy, Gurushri, 1991, International Migrant Workers’ Remittances: Issues and Prospects, World Bank Staff Working Papers, No. 481.

  • Straubhaar, Thomas, 1986, Determinants of Workers’ Remittances: The case of Turkey, Welwirtschaftliches Archiv; 122, No. 4: 72840.

  • Torres, Julio Roberto, 1997, Metodología de Regislro de las Remesas Familiares, Boletín Económico, BCR, No. 108, Junio: 1419.


Prepared by Florencia Frantischek.


For an analysis of this effect see Section III.


For a review of the literature on international workers’ remittances see: Elbadwi and Rocha (1992).


The increase is less abrupt when the official series on remittances for the period 1979-90 is adjusted as described in the Note at the end of the section.


The Immigration Reform and Control Act (IRCA) of November 1986 and the Immigration Act of 1990 (IMMACT90).


For a field study on the effect of the legalization process of undocumented immigrants from Guatemala see Hagan and Baker (1993).


Statistical tests were run to rule out the possibility of spurious regressions of the model. Even though the logarithm of the stock of workers was found to be stationary using the augmented Dickey-Fuller (ADF) test, a spurious correlation was considered unlikely as the ADF test may not capture the presence of a unit root when the sample period extends beyond the period of sustained growth of migration. The logs of remittances and real earnings were found to be nonstationary using the same ADF test. In addition, the Johansen cointegration test accepted two cointegrating equations for the logs of remittances, real earnings and the stock of workers.