various control variables are included, the CBI proxy is often not significant. Following the approach
suggested by Temple (1998), we also conclude that in those regressions in which the CBI proxy is
significant, the coefficient of the TOR becomes significant only after high inflation countries are
added to the sample.
The remainder of the paper is organised as follows. Section 2 presents the model and our data.
Section 3 offers our results. The final section contains some concluding comments.
2. The model and the data
Following Cukierman et al. (1992) we have used the transformed inflation rate D as dependent
variable in order to reduce heteroscedasticity of the error in the regressions. D is defined as the
inflation rate (p) divided by one plus the inflation rate:
D = p/(1 + p)
(1)
So the transformed inflation rate (if positive) takes a value from 0 to 1. When inflation is 100% (p=1),
D is 0.5. D has been calculated for each year, and subsequently the averages for the various estimation
periods have been calculated as the simple means of these annual observations. The inflation rate is
taken from the World Bank’s 2000 World Development Indicators CD-rom.
One serious drawback of many studies on the relationship between inflation and CBI is that
control variables are often lacking. Following Campillo and Miron (1997), we therefore also estimate
multivariate models. We include apart from indicators for CBI and openness (export and import as share
of GDP), also political instability (proxied by the number of government transfers), the log of GDP per
capita, a dummy for the exchange rate regime (one in case of a more or less fixed exchange rate) and the
debt-to-GDP ratio as control variables.
Our indicator for openness (OPEN, defined as sum of export and import in relation to GDP)
and the log of the level of GDP per capita (GDPCAP) are from the World Bank’s 2000 World
Development Indicators CD-rom. Our proxy for political instability (PI, defined as the total number of
government transfers) is from the new World Bank data set of political indicators.3 An exchange rate
dummy - which is one if the country had at the beginning and the end of the period a more or less
fixed exchange rate regime - is used to examine the impact of the exchange rate regime. This variable
is denoted as XRATE. It is constructed using information reported in the IMF's Annual Report. The
external debt-to-GDP ratio at the beginning of the estimation period (DEBT) is also from the World
Bank. So the estimated cross-section model is:4
D = c0 + c1CBI + c2OPEN + c3GDPCAP + c4PI + c5XRATE + c6DEBT
(2)