December, 1998, the last month of comprehensive data available for this study.
Comparisons of price instability in different periods are commonly based on estimation
of the coefficients of variation of detrended domestic prices in different time periods
(Hazell et al. (1990), Knudsen and Nash (1990), Mendoza (1998)). This approach is used
in this study. Detrended real domestic market prices (P*Dk) for each commodity in each
country were obtained by deflating nominal monthly prices by domestic CPI’s, and then
detrending to yield a residual stochastic variable, εtk, via equation (1):
11
PDtk = β0k + β1kT + Σ γikDm + εtk (1)
m=1
where the Dm’s are dummy variables accounting for monthly seasonality, T represents a
time trend variable, and subscripts t and k denote monthly observations and the two
policy periods, respectively. Once the detrended, deseasonalized real price residuals were
obtained, coefficients of variation were estimated and then compared across the two
policy periods, and their equality tested using the following form of the t-test (Greene,
1997):
(PD 1 - PD2 )-m
(2)
s 1 + s2_
A similar process was followed using international prices for the same commodities and
time periods in order to compare trends in domestic price instability with those in
international markets (these results are only briefly cited here but are discussed in detail
in Villoria, 2000).
Sources of Price Instability