Journal Issue
Share
Article

El Salvador: Selected Issues

Author(s):
International Monetary Fund. Western Hemisphere Dept.
Published Date:
January 2015
Share
  • ShareShare
Show Summary Details

Investment Drivers in Central America: an Application to El Salvador1

1. Low domestic investment and FDI in El Salvador. Average public and private investment in El Salvador during 2008-13 was only 2.4 and 11.7 percent of GDP, respectively, compared to 22.7 percent of GDP in CAPDR. Meanwhile, foreign direct investment averaged only 1.8 percent of GDP in this period, compared to the regional average of 4.8 percent of GDP.

2. Growth diagnostic of crime and low productivity in tradables. The US Partnership for Growth, in their 2011 constraints analysis of El Salvador (based on the Hausman, Rodrick and Velasco (2004) methodology) noted that nearly 11 percent of GDP is “spent or foregone due to crime” in El Salvador, nearly double the figure for Costa Rica. The report also cites low productivity in the tradables sector as a key impediment to private investment and growth. However, while crime is very high in El Salvador, it is more widespread in Honduras but does not appear to have a significant negative effect on investment in that country.

3. Other potential drivers of private investment in El Salvador. A panel regression is used to examine potential effects of variables such as inflation, public debt levels, human capital levels (proportion of the labor force with secondary school education), characteristics of exports (their level as percent of GDP and complexity) and institutional variables on investment in Central America. Among the latter, the regression includes a policy uncertainty variable constructed from a 5-year moving standard deviation of the government’s primary balance, a variable for political uncertainty proxied by the frequency of elections, competitiveness scores, and economic institution quality indicators. Figure 1 provides a regional comparison for some of the variables. Annual data from 1995–2012 including all Central American countries (El Salvador, Costa Rica, Dominican Republic, Guatemala, Honduras, Nicaragua, and Panama) from the IMF, the World Bank World Development Indicators (WDI), MIT observatory of economic complexity, World Economic Forum (WEF) and International Country Risk Guide (ICRG) were used for the regression specification below:

Figure 1.El Salvador: Elections, Human Capital and Exports

Source: ICRG, MIT Observatory of Economic Complexity, World Bank WDI and Fund staff estimates.

where β1 denotes the coefficient on exports as percent of GDP, β2 the coefficient on the exchange rate regime, β3 is the effect of education, β4 the level of inflation, β5 is the effect from the country’s level of export complexity, β6 captures the effect of the volatility of the primary balance, β7 and β8 effects from the level of debt (which is allowed to be a binomial in order to capture possible nonlinear “threshold” effects like those identified by Reinhart and Rogoff (2010, 2011)), β9 and β10 denote the effects from the duration of the electoral cycle (also allowed to be a binomial for possible non-linear effects), and finally, β11 and β12 are interactive coefficients between changes in openness and complexity index and changes in debt and complexity index, respectively.

4. Estimates and goodness of fit. The text table below shows the estimated coefficients for each specification: (1) a baseline which controls for global conditions such as deviations of the U.S. GDP from trend and dotcom crisis and the global financial crisis using dummies in 2001 and 2009, (2) one which includes country fixed effects, (3) one which adds survey competitiveness scores and (4) one which replaces competitiveness scores with ICRG’s measure for quality of institutions.2 The time-series fit for El Salvador is relatively good up to 2006–07 when the country experienced a jump in investment driven by Bancolombia’s purchase of Banco Agricola in December of that year (amounting to 4.5 percent of GDP). The model also over-predicts both the slump and the recovery related to global financial crisis, partly driven by the protracted global slowdown post-crisis. Estimated coefficients are qualitatively similar across the first two specifications (with and without country fixed effects); one notable change is the fact that complexity becomes positive and significant on its own if country fixed effects are included. Both competitiveness and institutional quality measures have positive and significant effects but their inclusion (or the inclusion of fixed effects) undoes the significance of the education variable and some interactions of the complexity variable. This is likely because survey measures already incorporate certain aspects of these other variables included in the regression (e.g. competitiveness measures partially capture the quality of labor force).

Private Investment, data and model fit for El Salvador

(percent of GDP)

Source: Fund staff estimates.

5. Model implications. Lowering the political uncertainty and increasing the educational level, competitiveness scores and economic institutions to regional levels could increase investment between 1 to 6 percent of GDP. The regression estimates suggest that, El Salvador’s private investment level as a percent of GDP would be 3.6–5 percent higher if its electoral cycle was increased from the current average of 20 months to 47 months (the case of Guatemala, Honduras and Costa Rica). If the quality of institutions and competitiveness scores reached the levels in Costa Rica (best institutions in the region) or Panama (highest competitiveness score) then investment as proportion of GDP would increase by 1.2 percent of GDP and 5.8 percent of GDP, respectively. It is, however, noteworthy that—given competitiveness levels in the region are relatively low—a comparison with South Korea’s competitiveness score illustrates that investment in El Salvador would increase by 16 percent of GDP. Finally, an increase in education to the levels observed in Costa Rica (the highest in the region) would generate an increase in investment of about 2.3 percent of GDP.

Table 1.Estimated Regression Coefficients
Dependent variable:

Private investment (% of GDP)
Baseline

(1)
Fixed-Effects

(2)
Competitiveness

(3)
Institutions

(4)
Open_t0.145***0.102***0.925***0.145***
(0.0300)(0.0235)(0.295)(0.0188)
ExRate_t0.4470.2300.535**−0.329
(0.336)(0.591)(0.239)(0.238)
Educ_t0.0660**−0.02160.04200.0365
(0.0333)(0.0779)(0.0335)(0.0386)
Debt_t−0.125**−0.242***−0.121**−0.130***
(0.0521)(0.0704)(0.0600)(0.0448)
debt2_t0.000539***0.00106***0.000586**0.000590***
(0.000183)(0.000371)(0.000238)(0.000205)
Inflation_t−0.109**−0.101***−0.114**−0.0964**
(0.0512)(0.0288)(0.0522)(0.0385)
MoNoElec_t1.180***1.501***0.691**1.429***
(0.319)(0.515)(0.327)(0.265)
MoNoElec2_t−0.0170***−0.0212***−0.00985**−0.0192***
(0.00504)(0.00788)(0.00456)(0.00368)
Complex_t-1−0.4284.097**0.6401.596
(1.596)(1.753)(2.092)(1.466)
Open_t x Complex_t-10.0290**−0.01340.01840.0140
(0.0130)(0.0174)(0.0132)(0.0114)
Debt_t x Complex_t-1−0.0584*−0.0477**−0.0622−0.0500*
(0.0331)(0.0185)(0.0389)(0.0277)
Financial_Crisis Dummy−0.264−0.509−0.139−0.207
(0.654)(0.677)(0.669)(0.579)
Dotcom_Bubble Dummy−1.375*−1.417−1.337−1.425
(0.706)(1.037)(0.818)(0.907)
PolicyVolatility_t0.2350.3350.6040.570
(0.508)(0.494)(0.566)(0.623)
Competitiveness Score0.150**
(0.0710)
Quality of Institutions0.230***
(0.0766)
Constant−11.69*−64.80***12.61
(6.164)(23.76)(11.35)
Robust standard errors in parentheses (*** p<0.01, ** p<0.05, * p<0.1)
Robust standard errors in parentheses (*** p<0.01, ** p<0.05, * p<0.1)

6. Economic fundamentals matter. Higher openness and lower debt have positive effects on private investment which generate further positive effects when combined with a higher level of exports complexity. Exports of goods and services in El Salvador are only 24 percent of GDP. The model predicts that investment would increase between 5.1 to 7.3 percent of GDP if El Salvador’s exports to GDP increased to the level of Honduras (33 percent of GDP), a country with an export structure similar to that of El Salvador which is heavy on textile maquila. If El Salvador’s debt (the highest in the region) were to increase to 75 percent of GDP, then the model predicts a drop in investment of about 2.1 percent of GDP in all specifications. However, if debt were to decrease to 40 percent of GDP—which is the lower end of the estimated range for debt sustainability for El Salvador—then the increase in investment predicted by the model would be of 1.1 – 2.1 percent of GDP. The effects from increased openness and lower debt on investment interact positively with greater exports complexity. For most specifications, increasing complexity alone did not yield a large or significant effect on investment. This may be because a country’s move to a high value-added production model is more credible if it already has a relatively large and well-developed export base together with a certain expected level of macroeconomic stability. Thus, if this increase in complexity (e.g. to Panama-like levels) is accompanied by a higher openness level or a lower debt level, then investment would increase between 5.1 to 8.4 percent of GDP in the former case and between 1.8 and 2.6 percent in the latter. Finally, inflation is estimated to have a negative effect on investment. El Salvador has the lowest inflation rate in the region; if it were to increase to around 7 percent (the level of Nicaragua, the highest in the region) then the model predicts that investment would fall by 0.6 percent of GDP.

7. Main conclusions. A simple panel regression for the drivers of investment in Central America, shows that variables such as education, openness and low levels of inflation have a positive effect on private investment. High levels of export complexity alone do not appear to correlate with private investment. Nevertheless, when an increase in complexity is coupled with either higher levels of openness or lower levels of debt, investment rises. Interestingly, the model also finds that lower political uncertainty has a positive impact on private investment. Finally, survey measures such as competitiveness also have a strong effect on investment.

References

Prepared by Joyce Wong and Heba Hany.

A specification which included both the competitiveness score and the ICRG measure for quality of institutions was problematic due to the high correlation between those two measures.

Other Resources Citing This Publication