The relation between growth and inflation rates in Latin America.

Van Rensselaer, Kristen N. ; Copeland, Joe

ABSTRACT

Inflation has long been a problem for countries in Latin America. While some countries have made progress in addressing the problem, other countries have not been able to achieve sustained economic growth. Keynesian thought posits the notion that inflation and growth can be positively related, while other theories suggest that inflation is detrimental to long-run growth. Various empirical studies yield conflicting results. This study applies two Granger causality tests to seventeen Latin American countries in order to determine possible long-run relations. For most of the countries, no causality is found, while unidirectional causality and bi-directional causality is found in a limited number of countries.

INTRODUCTION

Historically, many Latin American countries have been plagued by inflation problems though the problem has not been universally severe. During the past few decades, some of the countries have made considerable progress toward controlling inflation while others still struggle in the battle against inflation. Some countries in the region have also encountered difficulties in maintaining respectable economic growth rates. The purpose of this paper is to explore the relations that appear to exist between inflation and economic growth in selected Latin American countries and to note some of the econometric problems that exist in analyses of this nature.

There is far from universal agreement regarding the relation between inflation and economic growth, and this relation has long held a central position in macroeconomics. The traditional Keynesian view would hold that inflation can act as a stimulus to economic growth. This view is typically expressed through a short run Phillips curve tradeoff (Paul, Kearney & Chowdhury, 1997) that arises due to sticky prices and wages. Thus, in the short run, faster real growth may be associated with inflation (Motley, 1998). However, the Phillips curve tradeoff tends to disappear as economic agents are able to anticipate price changes. These views were especially prominent in the 1960's and continued into the 1980's (Bruno & Easterly, 1996).

There are strong arguments that inflation is detrimental to economic growth, especially in the long run. Inflation makes it difficult for economic agents to make correct decisions since changes in relative prices become obscured (Harberger, 1998). Inflation imposes variable costs, especially menu costs. If inflation is high, the variability of inflation may also be high and this complicates the task of forecasting inflation. If inflation cannot be correctly anticipated, both savers and investors may be lead to make decisions that are detrimental to economic growth. This view is traditional and, in spite of the Keynesian views promulgated in the 1960's and thereafter, this view is incorporated into models of the new-growth literature (Bruno & Easterly, 1996).

The empirical evidence regarding the inflation-growth relation is mixed. Fischer (1993) found that real GDP growth is negatively related to inflation in cross sectional regressions and that low inflation is not necessary for growth. Using extreme bounds analysis, Levine and Zervos (1993) found that there is not a significant negative correlation between growth and inflation. They suggest that if inflation persists over a long time period, economic agents find mechanisms to adjust so that growth is not affected. However, cross-country regressions involving socio-economic variables often produce fragile results (Levine & Renelt, 1992).

Grier and Tullock (1989) found that the average level of inflation had a neutral effect on growth in the OECD countries but for the remainder of the world, inflation had a significant negative effect on growth. For Latin America, they found no statistical relation between inflation and growth. Barro (1997), in a world wide cross sectional analysis, found a negative relation between inflation and growth, though the effect could not be isolated for inflation rates below 20 percent. Other research by Barro (1996) indicates that the experiences of high inflation have adverse effects on investment and growth. Kocherlakota (1996) supports Barro's findings of a negative slope coefficient between growth and inflation and further shows that high growth tends to generate low inflation.

Until the 1970's, many studies found the negative effect of inflation on economic growth to be statistically insignificant, and in some instances, the effect was found to be positive (Sarel, 1996). The high and persistent inflation rates experienced by some countries during the 1970's and 1980's brought about a rethinking of the inflation-growth link.

The preceding review of cross-sectional studies, while not exhaustive, does not provide clear evidence of the growth-inflation relation. There are numerous problems associated with the use of cross-country regression analysis. Levine and Renalt (1992) and Temple (1999) review some of the methodological problems. Pooled cross-country data sets tend not to be informative regarding what happens in the lower ranges of inflation (Bruno & Easterly, 1998). The empirical relation between growth and inflation appears to be different for Africa and Latin America (Ericsson, Irons & Tryon, 2001).

Time series analysis, which does have the advantage of unmasking relations for individual countries, has yet to provide evidence of a universal relation between growth and inflation. Paul, Kearney, and Chowdhury (1997) performed Granger causality tests on 70 countries for the 1960-1989 time period. For 40 percent of the countries, no causality was found; for approximately one-third of the countries, unidirectional causality was found; and for approximately one-fifth of the countries, bidirectional causality was found. In 19 of the countries, the direction of causality ran from growth to inflation, while the direction ran from inflation to growth for seven countries.

In this analysis, the correlation between the growth and inflation rates for each of the 17 Latin American countries is examined. Then the data are subjected to Granger causality tests in order to determine if long run relations exist and to determine the direction of causality. The data cover the 40-year period from 1961 through 2000.

DATA

Annual data used for the 17 Latin American countries were extracted from the World Bank's World Development Indicators: 2002 (cd-rom version). Annual inflation rates reflect annual percent changes in the CPI. Changes in the CPI are used in place of the GDP deflator since, by construction, the GDP deflators are negatively correlated with growth rates (Sarel, 1996). For this study, economic growth rates are defined as annual percent changes in real GDP. The countries included in the study are those Latin American countries for which 40 years of annual CPI and GDP growth rates could be obtained.

As a first step in analyzing the growth-inflation relation, the simple correlation coefficients between the annual percentage increases in the real GDP growth rate and the CPI were examined for the each country, and those correlation coefficients appear in Table 1. Only two countries, Paraguay and Uruguay, have positive coefficients, and in both cases, the resulting p-values are extremely large. Only two of the countries, Costa Rica and Mexico, had reasonably high coefficients, and in both instances, the level of significance is quite high.

Table 2 shows the mean growth and inflation rates for the 17 countries. These rates are the simple arithmetic average rates for the 40-year time period. Only three of the 17 countries had average growth rates below two percent, and seven countries had average growth rates in excess of four percent.

Three of the countries (Argentina, Bolivia, and Peru) had triple-digit average inflation rates. Argentina experienced triple-digit inflation in 15 of the 40 study years. Bolivia only had that experience for three of the 40 years, but in one year the inflation rate was in the 5-digit range. Panama had the lowest average inflation rate, but its economy has been dollarized for quite some time.

The correlation coefficients indicate that, for 15 of the 17 countries, a possible negative relation exists between growth and inflation, while a positive relation could exist for two of the countries. However, the preceding correlation analysis indicates nothing about the direction of causality. Furthermore, well-known difficulties with time series analysis may be present. If the data are nonstationary, then the results found may be spurious (Murray, 1994).

METHODOLOGY

In order to gain insight into the direction of causality between inflation and growth, two causality techniques will be used. While no statistical test can indicate true causality, inference can be made as to which variable may precede another variable.

The first technique is the test for Granger causality (Granger, 1969). With Granger causality, to determine if growth (Y) "Granger causes" inflation (P), the following two regression equations are estimated:

[P.sub.t] = [alpha][m.summation over (i=1)][[beta].sub.i][P.sub.t-i] + [e.sub.t] (1)

[P.sub.t] = [alpha][m.summation over (i=1)][[empty set].sub.i][P.sub.t-i] + [n.summation over (i=1)][[delta].sub.i][Y.sub.t-i] + [u.sub.t] (2)

The first equation is the current inflation values regressed on lagged values of inflation. The second equation adds lagged values of growth to determine if past values of growth assist in the prediction of current values of inflation. An F-test is then used to determine whether the coefficients of the lagged growth rates in the second equation may be considered to be zero. To test for causation from inflation to growth, the test is repeated where growth is regressed on past values of growth and then compared to the regression of current values of growth regressed on past values of growth and inflation. The direction of causation between growth and inflation can be one of four possibilities: inflation to growth, growth to inflation, bidirectional between growth and inflation, or no relationship between growth and inflation.

The second technique used to investigate the direction of causality is the method used by Paul, Kearney, and Chowdhury (1997). This variation of the Granger causality model allows for "instantaneous causality" by allowing current values of inflation to play a role in determining current values of growth and vice versa. The authors consider this to be advantageous for annual data due to the speed with which information is transmitted through an economy. The model used in this paper differs slightly from the one suggested in Paul, Kearney, and Chowdhury (1997). Paul, et. al., includes the growth rate of money as a possible determinant of inflation. This paper excludes this variable since the focus is on the relationship between inflation and growth only. The following models will be used to further test for causality. For these models, the significance of the coefficients will be used to determine the direction of causality.

[P.sub.t] = [alpha][l.summation over (i=1)][[empty set].sub.i][P.sub.t-i] + [p.summation over (i=0)][gamma][Y.sub.t-i] + [v.sub.t] (3)

[Y.sub.t] = [alpha][q.summation over (i=1)][[lambda].sub.i][Y.sub.t-i] + [r.summation over (i=0)][[eta].sub.i] [P.sub.t-i] + [z.sub.t] (4)

RESULTS

Before proceeding with the causality tests, it is important to determine the order of integration of the data. Estimation of the equations requires the use of stationary data. If both growth and inflation rates are nonstationary, cointegration tests are a more appropriate technique to discern long-run relationships and Granger causality. Dickey-Fuller (1979, 1981) and Phillips-Perron (1988) Unit Root tests are two commonly used techniques to determine whether or not time series is stationary. More than one test is typically used due to the low power of the tests. If the tests confirm one another, greater confidence can be placed in the results (Enders, 1995).

Table 3 reports the results of the unit root tests. The lag length was selected using Akaike Information Criterion (AIC). After selecting the lag length, the selection of appropriate model is addressed. The choice for the models includes a time trend and constant, a constant, or no time trend or constant (none). The power of unit root tests is very sensitive to the deterministic regressors selected. The models were selected using the technique outlined by Doldado, Jenkinson, and Sosvilla-Rivero (1995). The null hypothesis of the Augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) test is the time series contains a unit root or integrated of order one. Failure to reject the null means the series is integrated of order one, I(1), or higher while rejecting the null means the series is stationary or integrated of order zero, I(0).

The five-percent MacKinnon Critical Values for both the ADF and PP test statistic for a lag of zero and no constant or time trend is -1.96. Since most of the models fit this description the reader is referred to MacKinnon (1991) for other critical values.

The ADF and PP test gave conflicting results for the growth rate for Trinidad & Tobago. In this case, the order of integration was determined by the PP test. The reason for favoring the PP test is due to the assumptions required for both tests. The ADF test assumes the error terms are independent with a constant variance. The PP test assumes the error terms are weakly dependent and heterogeneously distributed. Due to the nature of the data, the assumptions for the PP test are more reasonable than the ADF test.

For the growth rate variables, sixteen are stationary and one is nonstationary. For the inflation variables, three are stationary and fourteen are nonstationary. When the null hypothesis of a unit root is not rejected, this means the time series could be integrated of order one or higher. For the growth and inflation rates that were found to be nonstationary, these variables were checked for higher orders of integration. While the results are not reported here, it can be concluded that the nonstationary variables are integrated of order one. For the variables having unit roots, the first difference of the variable is used in the estimation of the causality regressions.

Table 4 reports the results of the Granger Causality Tests as discussed in the methodology section. A lag length of two was chosen for the tests. These results should be interpreted with caution due to the nonnormality and/or heteroscedasticity of the error terms. At the ten percent level of significance, it appears the growth precedes inflation in Columbia, Honduras, and Venezuela. Inflation precedes growth in Haiti. There is no detection of bidirectional causality between inflation and growth for any of the countries using this technique. The statistical evidence indicates that for the majority of the countries (Argentina, Bolivia, Costa Rica, Dominican Republic, El Salvador, Guatemala, Jamaica, Mexico, Panama, Paraguay, Peru, Trinidad & Tobago, and Uruguay) there is no causal relationship between growth and inflation.

Table 5 reports the results of the Paul, Kearney, and Chowdhury (1997) alternative variation of the Granger causality test. This model allows for instantaneous causality between growth and inflation. The previous model only allowed for a lagged relationship. To expedite the reporting of the results, only the countries where a statistically significant relationship is evidenced will be given in Table 5. The table reports the regression coefficients and the p-values. As stated before, these results should be interpreted with caution due to the violation of assumptions concerning the error term. Where applicable, the regression results have been corrected for heteroscedasitcity.

Using this alternative model, growth "causes" inflation in Honduras and Mexico. Inflation "causes" growth in Haiti. There is a bidirectional relationship for Argentina, Costa Rica, and Venezuela. There was no relationship between growth and inflation for Bolivia, Columbia, Dominican Republic, El Salvador, Guatemala, Jamaica, Panama, Paraguay, Peru, Trinidad & Tobago, and Uruguay. Table 6 provides a summary of the results of both causality tests. As might be expected, more relationships are detected between inflation and growth rates when instantaneous causality is included in the model.

CONCLUSION

This study has explored the possible relationship between real GDP growth and the rate of inflation, as measured through the GDP deflator, in Latin America over the period of 1961 to 2000. In addition to correlation analysis, Granger causality is used to investigate the nature of the relationship between growth and inflation rates.

The correlation analysis revealed mixed results. For countries where the correlation coefficient is statistically significant, the relationship between growth and the rate of inflation is negative. However there are two possible problems with this analysis. First, the correlation analysis does not indicate the causal flow of the relationship. Is growth having a negative impact on inflation or vice versa? Second, this analysis may yield spurious results if the data is nonstationary.

Unit root analysis is conducted prior to the estimating the Granger causality models. In the cases where the variable is determined to be nonstationary, the variable is first differenced prior to the estimation of the Granger causality models. Two forms of Granger causality tests are conducted. The interpretation of these results should be treated with caution since, due to the nature of the data, the regression models tend to experience violation of the assumptions about the error terms.

For the seventeen countries under study, there was no statistical evidence of a relationship between growth and inflation in eleven countries (Bolivia, Columbia, Dominican Republic, El Salvador, Guatemala, Jamaica, Panama, Paraguay, Peru, Trinidad & Tobago, and Uruguay) using the Paul, Kearney and Chowdhury (1997) variation of the Granger causality test. In these countries, economic growth and inflation are not statistically linked. In Haiti where inflation tends to "cause" growth, the preponderance of the evidence indicate an adverse relationship between inflation and growth. For this country, increases in inflation may lead to decreased economic growth. In Honduras and Mexico where growth "causes" inflation, the negative relationship is more beneficial for the economy. For these countries, as the economy experiences positive growth, this tends to decrease inflation. Argentina, Costa Rica, and Venezuela have more complicated relationships between inflation and growth. Growth and inflation appear to have bidirectional causality.

For the Latin American countries used in Paul, Kearney, and Chowdhury (1997), fifteen (Bolivia, Columbia, Costa Rica, Dominican Republic, El Salvador, Guatemala, Haiti, Honduras, Jamaica, Panama, Paraguay, Peru, Trinidad & Tobago, Uruguay, and Venezuela) of the countries are also used in this study. For these fifteen countries, the same general conclusion was reached about causality for Costa Rica (Y [left and right arrow] P) and Honduras (Y [right arrow] P). Also, both studies conclude that there is no statistical evidence of causality between growth and inflation for El Salvador, Guatemala, Jamaica, Paraguay, Trinidad & Tobago, and Uruguay. While this study differs in the model (see endnote 1) and the time period used, the studies reach similar conclusions for eight of the fifteen countries. However, the fragility of the results of statistical analysis indicates a need for continual analysis and model development.

REFERENCES

Barro, R.J. (1997). Determinants of Economic Growth. Cambridge, MA: The MIT Press.

Barro, R. J. (1996). Inflation and growth. Review (Federal Reserve Bank of St. Louis), 78(3). Retrieved September 30, 2002, from http://www.stls.frb.org/docs/publications/review/96/05rb2.pdf.

Bruno, M. & W. Easterly (1996). Inflation and growth: In search of a stable relationship. Review (Federal Reserve Bank of St. Louis) 78(3). Retrieved September 30, 2002, from http://www.stls.frb.org/docs/publications/review/96/05/9505mb.pdf.

Bruno, M. & W. Easterly (1998). Inflation crises and long-run growth. Journal of Monetary Economics, 41(1), 3-26.

Dickey, D.A. & W.A. Fuller (1979). Distribution of the estimates for autoregressive time series with a unit root. Journal of the American Statistical Association, 78(366), 427-437.

Dickey, D.A. & W.A. Fuller (1981). Likelihood ratio statistics for autoregressive time series with a unit root. Econometrica, 49(4), 1057-1072.

Doldado, J.J., T. Jenkinson, & S. Sosvilla-Rivero (1990). Cointegration and unit roots. Journal of Economic Surveys, 4, 249-73.

Enders, W. (1995). Applied Econometric Time Series, New York: Wiley Publishing.

Ericsson, N.R., J.S. Irons & R. Tryon (2001). Output and inflation in the long run. Journal of Applied Econometrics, 16(3), 241-253.

Fischer, S. (1993). The role of macroeconomic factors in growth. Journal of Monetary Economics, 32(3), 485-512.

Granger, C.W.J. (1969). Investigating causal relations by econometric models and cross-spectral methods. Econometrica, 37(3) 424-438.

Grier, K.B. & G. Tullock (1989). An empirical analysis of cross-national economic growth. 1951-80, Journal of Monetary Economics, 24(2), 259-276.

Harberger, A.C. (1998). A vision of the growth process. American Economic Review, 88(1), 1-32.

Kockherlokota, N. (1996). Commentary. Review (Federal Reserve Bank of St. Louis) 78(3). Retrieved September 30, 2002, from http://www.stls.frb.org/docs/publications/review/96/05/9605nk.pdf.

Levine, R. & D. Renelt (1992). A sensitivity analysis of cross-country growth regressions. American Economic Review 82(4), 942-963.

Levine, R. & S.J. Zervos (1993). What have we learned about policy and growth from cross-country regressions? American Economic Review 83(2), 426-430.

Motley, B. (1998). Growth and inflation: A cross-country study. Economic Review (Federal Reserve Bank of San Francisco) 1,15-28.

MacKinnon, J.G. (1991). Critical values for cointegration tests. Long-Run Economic Relationships: Readings in Cointegration. Oxford: Oxford University Press.

Murray, M.P. (1994). A drunk and her dog: An illustration of cointegration and error correction. American Statistician 48(1), 37-39.

Paul, S., C. Kearney & K. Chowdhury (1997). Inflation and economic growth: A multi-country empirical analysis. Applied Economics 29(10), 1387-1401.

Phillips, P.C.B. & P. Perron (1988). Testing for a unit root in time series regression. Biometrica, 75(2), 1361-1401.

Sarel, M. (1996). Nonlinear effects of inflation on economic growth. IMF Staff Papers, 43(1), 199-215.

Temple, J. (1999). The new growth evidence. Journal of Economic Literature 37(1), 112-156.

Kristen N. Van Rensselaer, University of North Alabama

Joe Copeland, University of North Alabama

Table 1: Growth-Inflation Correlations, By Country, 1961-2000

Country Correlation Coefficient p-value

Argentina -0.40 0.01
Bolivia -0.21 0.19
Columbia -0.18 0.27
Costa Rica -0.69 0.00
Dominican Republic -0.26 0.11
El Salvador -0.36 0.02
Guatemala -0.21 0.20
Haiti -0.18 0.26
Honduras -0.30 0.06
Jamaica -0.19 0.24
Mexico -0.60 0.00
Panama -0.08 0.63
Paraguay 0.13 0.41
Peru -0.45 0.00
Trinidad & Tobago -0.00 0.98
Uruguay 0.15 0.35
Venezuela -0.26 0.10

Table 2: Mean Growth And Inflation Rates

Country Mean Growth Rate Mean Inflation Rate

Argentina 2.59 242.90
Bolivia 2.74 353.78
Columbia 4.23 19.25
Costa Rica 4.87 14.18
Dominican Republic 5.43 11.95
El Salvador 3.10 9.78
Guatemala 4.05 9.22
Haiti .95 10.32
Honduras 3.99 8.57
Jamaica 1.95 16.16
Mexico 4.72 26.83
Panama 4.54 2.85
Paraguay 4.52 13.09
Peru 3.14 331.25
Trinidad & Tobago 3.75 8.18
Uruguay 1.93 53.12
Venezuela 2.72 17.01

Table 3: Unit Root Test Results For Growth And Inflation Rates

Country Var. Lag Model

Argentina Y 0 None
 P 1 None

Bolivia Y 0 None
 P 0 None

Columbia Y 0 Constant
 P 0 Constant

Costa Rica Y 0 Constant
 P 0 Constant

Dominican Y 0 Constant
Republic P 0 None

El Salvador Y 1 None
 P 0 None

Guatemala Y 0 None
 P 0 None

Haiti Y 1 None
 P 0 None

Honduras Y 0 Constant
 P 0 None

Jamaica Y 1 None
 P 0 None

Mexico Y 0 Constant
 P 2 None

Panama Y 1 Constant
 P 0 None

Paraguay Y 0 None
 P 0 None

Peru Y 1 None
 P 0 None

Trinidad & Tobago Y 1 None
 P 0 None

Uruguay Y 2 Constant
 P 0 None

Venezuela Y 0 Constant
 P 4 None

Country ADF PP Conclusion

Argentina -4.83 -4.83 I(0)
 -3.69 -3.29 I(0)

Bolivia -3.34 -3.34 I(0)
 -5.39 -5.39 I(0)

Columbia -3.65 -3.65 I(0)
 -2.75 -2.75 I(1)

Costa Rica -4.50 -4.50 I(0)
 -1.75 -1.75 I(1)

Dominican -6.14 -6.14 I(0)
Republic -1.26 -1.26 I(1)

El Salvador -2.49 -2.14 I(0)
 -1.20 -1.20 I(1)

Guatemala -1.63 -1.63 I(1)
 -2.14 -2.14 I(0)

Haiti -3.40 -6.36 I(0)
 -1.76 -1.76 I(1)

Honduras -4.75 -4.75 I(0)
 -1.14 -1.14 I(1)

Jamaica -2.43 -4.01 I(0)
 -0.95 -0.95 I(1)

Mexico -4.12 -4.12 I(0)
 -1.08 -1.58 I(1)

Panama -3.95 -4.13 I(0)
 -1.05 -1.05 I(1)

Paraguay -2.12 -2.12 I(0)
 -1.63 -1.63 I(1)

Peru -3.43 -3.54 I(0)
 -1.22 -1.22 I(1)

Trinidad & Tobago -1.82 -4.06 I(0)
 -1.13 -1.13 I(1)

Uruguay -4.73 -3.96 I(0)
 -1.31 -1.31 I(1)

Venezuela -5.53 -5.53 I(0)
 -1.41 -1.82 I(1)

Table 4: Results Of Granger-Causality Test

 Ho: Inflation does not Cause Inflation
Country
 F-Value p-Value

Argentina 2.39 0.11
Bolivia 0.69 0.51
Columbia 0.52 0.60
Costa Rica 1.01 0.38
Dominican Republic 0.04 0.96
El Salvador 0.95 0.40
Guatemala 0.44 0.65
Haiti 2.69 0.08
Honduras 1.28 0.29
Jamaica 1.12 0.34
Mexico 0.34 0.71
Panama 1.30 0.29
Paraguay 1.40 0.26
Peru 0.63 0.54
Trinidad & Tobago 2.24 0.12
Uruguay 1.15 0.33
Venezuela 1.27 0.29

 Ho: Growth does not Cause Inflation
Country
 F-Value p-Value

Argentina 0.17 0.85
Bolivia 2.11 0.14
Columbia 4.53 0.02
Costa Rica 0.96 0.39
Dominican Republic 0.99 0.38
El Salvador 0.06 0.94
Guatemala 0.73 0.49
Haiti 0.10 0.91
Honduras 2.49 0.10
Jamaica 0.52 0.60
Mexico 0.63 0.54
Panama 0.85 0.44
Paraguay 0.54 0.59
Peru 2.09 0.14
Trinidad & Tobago 0.32 0.73
Uruguay 2.26 0.12
Venezuela 4.80 0.02

Table 5: Alternative Granger Causality Test

 Inflation Model (Equation 3)

Country [alpha] [P.sub.t-1] [P.sub.t-2]

Argentina 343.61 0.546 -0.197
 (0.01) (0.00) (0.26)

Costa Rica 2.716 -0.243 -0.086
 (0.73) (0.18) (0.68)

Honduras -3.044 0.045 -0.112
 (0.27) (0.83) (0.59)

Mexico 1.099 0.083 -0.323
 (0.88) (0.67) (0.34)

Venezuela -6.702 0.412 0.476
 (0.20) (0.12) (0.07)

Country [Y.sub.t] [Y.sub.t-1]

Argentina -42.307 -7.579
 (0.01) (0.64)

Costa Rica -2.578 0.967
 (0.10) (0.14)

Honduras -0.611 0.748
 (0.15) (0.09)

Mexico -2.317 0.892
 (0.00) (0.45)

Venezuela -2.543 3.438
 (0.00) (0.00)

Country [Y.sub.t-2] Adj [R.sup.2]

Argentina -16.883 0.41
 (0.30)

Costa Rica 1.086 0.33
 (0.22)

Honduras 0.49 0.12
 (0.28)

Mexico 1.236 0.26
 (0.20)

Venezuela 0.317 0.33
 (0.76)

 Growth Model (Equation 4)

Country [alpha] [Y.sub.t-1] [Y.sub.t-2]

Argentina 3.926 -0.049 -0.256
 (0.00) (0.77) (0.12)

Costa Rica 2.361 0.410 0.111
 (0.03) (0.04) (0.54)

Haiti 0.786 0.093 0.130
 (0.29) (0.57) (0.41)

Venezuela 0.894 0.409 0.072
 (0.36) (0.05) (0.71)

Country [P.sub.t] [P.sub.t-1]

Argentina -0.004 0.000
 (0.01) (0.89)

Costa Rica -0.125 -0.052
 (0.00) (0.19)

Haiti -0.034 0.025

 (0.68) (0.78)

Venezuela -0.086 0.081
 (0.01) (0.10)

Country [P.sub.t-2] Adj [R.sup.2]

Argentina 0.002 0.26
 (0.20)

Costa Rica 0.021 0.39
 (0.55)

Haiti -0.178 0.05
 (0.05)

Venezuela 0.090 0.17
 (0.06)

Note: Parameter estimates with p-values in parenthesis.

Table 6: Summary Of Results

 Granger Causality Test

Y [right P [right Y [right and No
arrow] P arrow] Y arrow] P Causality

Columbia Haiti None Argentina
Honduras Bolivia
Venezuela Costa Rica
 Dominican Republic
 El Salvador
 Guatemala
 Jamaica
 Mexico
 Panama
 Paraguay
 Peru
 Trinidad & Tobago
 Uruguay

 Alternative Test for Granger Causality

Y [right P [right Y [right and No
arrow] P arrow] Y arrow] P Causality

Honduras Haiti Argentina Bolivia
Mexico Costa Rica Columbia
 Venezuela Dominican Republic
 El Salvador
 Guatemala
 Jamaica
 Panama
 Paraguay
 Peru
 Trinidad & Tobago
 Uruguay