OPENNESS AND INFLATION: A NEW ASSESSMENT*

Maria Cristina T. Terra PUC-Rio

Abstract In models where inflation is generated due to lack of precommitment in monetary policy, an important prediction is that more open economies should have lower inflation rates. Romer [1993] provides a clever test that appears to support this prediction. However, as the present paper shows, Romer’s regression results may have been driven by indebted countries going through the debt crisis of the 1980’s. An alternative explanation of the data is offered.

I. Introduction Romer [1993] uses a Barro-Gordon type model1 to argue that openness puts a check on the government’s incentive to engage in unanticipated inflation, because of induced exchange rate depreciation.2 Therefore, in absence of precommitment in monetary policy, more open economies will tend to have lower inflation rates. Romer finds that empirically, over a large sample of countries and over a broad time frame, “average inflation rates are lower in smaller, more open economies. This relationship is significant, quantitatively large and

*

I thank Kenneth Rogoff, Olivier Blanchard, Peter Kenen, Laurence Ball, Afonso Bevilaqua, an anonymous

referee, and participants in the Department of Economics seminar at PUC-Rio and in the Modeling Inflation session at the Econometric Society 7th World Congress for helpful comments and suggestions. All remaining errors are my own. I am grateful to CNPq, Brazil, for financial assistance.

1

Kydland and Prescott [1977] and Barro and Gordon [1983] are classic references in this literature.

2

Rogoff [1985] presents the argument formally.

1

robust.” Moreover, “these findings suggest that models in which the absence of precommitment in monetary policy leads to inefficiently high average levels of inflation are essential to understanding inflation in most of the world.” This comment suggests that the negative link between inflation and openness Romer found may be largely driven by the responses of severely indebted countries (SICs) to the debt crisis of the 1980s. Figures 1 and 2 show the scatterplots of the logarithms of average inflation and openness for the subperiods 1973 to 1981 (pre-debt crisis) and 1982 to 1990 (the debt crisis period). Different groups of countries are separated according to their indebtedness level.3 The solid lines in the figures represent the predicted value of inflation, obtained from a regression of openness on inflation for the countries in each group. As the scatterplots show, there is indeed a strong negative relation between openness and inflation among SICs during the debt crisis period. However, this correlation is not observed for SICs during the pre-crisis period, and it is not observed among other groups of countries for either sub-period. The empirical results are analyzed in section 2, section 3 presents a theoretical argument for the empirical findings, and section 4 concludes. II. Empirical Evidence As in Romer [1993], we use the ratio between imports and GDP as the proxy for openness.4 Inflation is measured as the annual change in the logarithm of the GDP deflator. The

3

The classification used was the one in the World Bank Development Report, 1993, p.328-329, using standard

World Bank definitions of degree of indebtedness averaged over three years (1989-1991). High income countries, which did not have a debt problem, were not classified according to indebtedness level, and the group “all other countries” was created for them. Appendix A shows the list of countries in each group. 4

All regressions were also run using the ratio between imports plus exports to GDP as a measure of openness.

The results were qualitatively the same as the ones presented here.

2

national accounts data are taken from the International Financial Statistics (published by the International Monetary Fund). Our first approach follows Romer [1993] by pooling the full sample of 114 countries and by averaging data over the full 1973 to 1990 time period. The results, given in column (1) of Table 1, replicate Romer’s main findings. A regression of openness on inflation yields a significant negative coefficient for openness. Furthermore, openness explains over 10 percent of the variation in inflation. The sample was then divided into four groups of countries according indebtedness level5, and the same regression was run for each group of countries separately. The results are presented in columns (2), (5), (8) and (11). The null hypotheses that the coefficient on openness is equal to zero cannot be rejected at the 5% of significance level for any group of countries, except for the SICs (column (2)). For the SICs, the coefficient of openness is at least two times larger, in absolute value, than any other, and openness alone accounts for over 22% of inflation variation among those countries. It is important to note that the time period under study incorporates two distinct economic environments in the world. During the 1970’s, developing countries were contracting their debts; after the debt crisis began in 1982, they came under great pressure to make significant net repayments. Thus it makes sense to consider treating these two time periods separately. The next set of regressions uses the average of inflation rates and openness for the pre-debt crisis period (1973-1981) and for the debt crisis period (19821990), again taking each group of countries separately.6 These regressions are the ones used to calculate predicted inflation in figures I and II.

5

See footnote 3.

6

This sample of countries is slightly different from the ones in Romer [1993]. Countries for which data was

not available in the International Financial Statistics were excluded. Some countries for which data was available, but Romer [1993] did not use, were included. The complete list of countries is in the appendix.

3

Columns (3), (6), (9) and (12) of table I present the results of the regressions of openness on inflation during the pre-debt crisis period, and columns (4), (7), (10) and (13) show the regressions for the debt crisis period. The coefficient of openness is negative and significant at conventional significance levels only for the SICs, and during the debt crisis its value is almost two times larger than during the pre-debt crisis period (-2.495 versus 1.253, respectively). For the moderately and less indebted countries, the coefficients for openness are also larger during the debt crisis, compared to the pre-debt crisis period, although they are not significant. For the less indebted countries, and for the group of countries not classified by indebtedness level (countries with no debt problems), there is virtually no link between openness and inflation: the t-statistics for the coefficient of openness is lower than 1 and the R2 is lower than 2% in regressions (9), (10), (12) and (13). Overall, the results in table I indicate that the negative relation between openness and inflation identified in column (1) for the whole sample of 114 countries, for averages of the data from 1973 to 1990, is actually caused by the severely indebted countries, over the debt crisis period. III. Inflation and Debt Crisis The empirical result that the effect of openness on inflation is only significant for severely indebted countries during the debt crisis can be explained with an argument that relates to the literature that recognizes the dual resource transfer necessary for a country’s repayment of a large external debt.7 On the one hand, trade surpluses have to be generated to make debt repayments, in order to effect the required external transfer. On the other hand, due to the fact that in many developing countries the foreign debt is largely a public sector

7

See Cohen [1987], Rodrik [1990], and Bevilaqua [1993], for references in this literature.

4

liability, the public sector must raise resources by taxing the private sector, thereby effecting an internal transfer. Take two countries with same debt burden, and therefore needing the same trade surplus to make the external transfer. Assuming identical price elasticities, the less open economy will need a larger exchange rate devaluation to generate the trade surplus. The devaluation of the exchange rate, in turn, tightens the internal constraint by raising the value of external liabilities in domestic currency: more resources will then have to be transferred from the private to the public sector. When inflation tax is the major mechanism for this transfer, a higher inflation rate will result. Hence, the less open a country is, the higher its inflation will be during a debt crisis. Terra [1995] presents a formal model for this argument. This is a topic for further research. IV. Concluding Remarks One interpretation of the evidence is that countries that “over-borrowed” -- the SICs -- are precisely the ones with less precommitment in monetary policy. Therefore the negative link between inflation and openness is stronger among them. This explanation would explain why openness is empirically important as a determinant of inflation only for the severely indebted countries. This interpretation, however, does not account for the fact that the relation is notably stronger during the debt crisis period. It is important to note that the alternative explanation of the data offered here is not contradictory with the precommitment argument. Both absence of precommitment in monetary policy and debt crisis may be an important determinants of inflation. The former generates a negative link between openness and inflation among SICs, and the latter strengthens this link. Therefore, they may be viewed as complementary and, as the data indicates, important explanations for the behavior of inflation across countries.

5

5. References Barro, Robert and David Gordon, “Rules, Discretion and Reputation in a Model of Monetary Policy”, Journal of Monetary Economics, XII (July 1983), 101-121. Bevilaqua, Afonso, “Public External Debt and Dual Resource Transfers”, Chapter 3 of Ph.D. dissertation, University of California, Berkeley (1993). Cohen, D., “Domestic and External Debt Constraints of LDCs”, in R. Bryant and R. Portes, eds., Global Macroeconomics (Macmillan, 1987). Kydland, Finn and Edward Prescott, “Rules Rather than Discretion: the Inconsistency of Optimal Plans”, Journal of Political Economy, LXXXV (June 1977), 473-492. Rodrik, D., “The Transfer Problem in Small Open Economies: Exchange Rate and Fiscal Policies for Debt Service”, Ricerche Economiche, XLIV (April-September 1990), 231-250. Rogoff, Kenneth, “Can International Monetary Policy Cooperation Be Counterproductive?”, Journal of International Economics, XVIII (May 1985), 199217. Romer, David, “Openness and Inflation: Theory and Evidence”, The Quarterly Journal of Economics, CVIII, (November 1993) , 869-903. Terra, Cristina, “Openness and Inflation: A New Assessment”, Chapter 2 of Ph.D. Dissertation, Princeton University, Princeton (1995). World Bank, World Bank Development Report, (Oxford: Oxford University Press, 1993)

6

Appendix Severely Indebted Countries

Moderately Indebted Countries

Less Indebted Countries

Algeria Bangladesh* Bahrain Argentina Benin Barbados Bolivia Cameroon Botswana Brazil Central African Rep.* Burkina Faso Burundi* Chile Cape Verde** Congo Colombia El Salvador Ecuador Costa Rica Fiji Egypt Dominican Republic Grenada** Ethiopia Gabon Iran Ghana Gambia* Korea Guyana Greece Lesotho Honduras Guatemala Malaysia Ivory Coast* Haiti Malta Jamaica* Hungary** Mauritius Jordan India Oman Kenya Indonesia Paraguay Liberia Malawi Portugal Madagascar Nepal Saudi Arabia* Mauritania Pakistan Solomon Islands** Mexico Papua New Guinea South Africa* Morocco Philippines Surinam Myanmar** Rwanda Swaziland Nicaragua Senegal Thailand Niger Sri Lanka Trinidad & Tobago Nigeria Togo Zimbabwe Panama Tunisia Peru Turkey Sierra Leone Uruguay Somalia* Venezuela Sudan* Yemen** Syrian Arab Republic Tanzania Uganda* Zaire Zambia * Countries included only in regressions (1), (2), (5), (8), and (11) of table I. ** Countries included only in the other regressions of table I.

7

All Other Countries Australia Austria Belgium Burma* Canada Cyprus Denmark Finland France Germany Hong Kong Iceland Ireland Israel Italy Japan Kuwait Luxembourg Netherlands New Zealand Norway Singapore Spain Sweden Switzerland Taiwan United Arab Emirates United Kingdom United States

Table I

Whole Sample Severely Indebted Countries (1) Openness

(2)

(3)

-1.007 -1.020 -2.149 (-

(-

(-

Moderately

Less Indebted

Indebted Countries

Countries

All Other Countries

(4)

(5)

(6)

(7)

(8)

(9)

(10)

-2.055

-1.041

-0.698

-0.651

-0.874

-0.237

0.672

(-2.362)

(-0.548)

(0.883)

(-2.875) (-1.409) (-1.185) (2.042)

3.787) 3.639) 3.056) Log Income

-0.049

0.305

0.448

-0.342

-0.002

(-

(1.759)

(4.297)

(-0.316)

(-0.008)

0.802) Constant

-1.729 -1.374 -1.039 (-

(-

14.792) 2.918 R-Square

Sample

(-

-3.184

-1.731

-5.056

(-2.526) (-7.463) (-6.311)

3.962)

-1.954

-1.635

-2.378

-2.663

(-

(-1.900)

(-

(-1.601)

10.792)

11.636)

0.114

0.129

0.226

0.317

0.069

0.483

0.173

0.256

0.013

0.095

114

114

34

34

29

29

22

22

29

29

Size t-statistics are in parentheses. The regressions that iclude income, also include dummy variables for alternative data sources (see Romer [1993]).

8

Table II

Severely Indebted

Moderately

Less Indebted

All Other

Countries

Indebted Countries

Countries

Countries

Pre-

Debt

Pre-

Debt

Pre-

Debt

Pre-

Debt

Debt

Crisis

Debt

Crisis

Debt

Crisis

Debt

Crisis

Crisis Openness

-1.253 (-

Crisis -2.495

-1.289

(-2.175) (-1.404)

Crisis -2.154

-0.067

(-1.900)

(-

-1.237 (-

-0.893

-1.355

Sample

0.207

(-0.531) (0.351)

0.231 (0.227)

(-2.105) (-4.819)

-1.408

-1.672

-2.128

-2.056

-2.624

(-4.084)

(-

(-6.971)

(-

(-6.248)

6.664) R-Square

-0.260

0.316)

2.677) Constant

Crisis

13.109)

8.828)

0.234

0.145

0.076

0.131

0.005

0.013

0.005

0.002

30

30

26

26

23

23

25

24

Size t-statistics are in parentheses.

9

Figure 3

All Other Countries

Less Indebted Countries

1000 1000

100 100

10

10

Inflation (% - log scale)

1

1 0

0.5

1

1.5

0

Moderately Indebted Countries

0.5

1

1.5

Severely Indebted Countries 1000

1000

100 100

10 10

1

1 0

0.5

1

1.5

0

Openness

Pre-Debt Crisis

10

0.5

1

1.5

Figure 4

All Other Countries

Less Indebted Countries

1000 1000

100 100

10

10

Inflation (% - log scale)

1

1 0

0.5

1

1.5

0

0.5

Moderately Indebted Countries

1

1.5

Severely Indebted Countries

1000

1000

100

100

10

10

1

1 0

0.5

1

0

1.5

Openness

Debt Crisis

11

0.5

1

1.5