The paper presents an econometric framework for the construction of a consistent panel of purchasing power parities (PPPs) which makes it possible to combine all the PPP benchmark data from various phases of the International Comparison Program with the data on national price movements in the form of implicit deflators from national accounts. The method improves upon the current practice used in the construction of the Penn World Tables (PWT), and similar tables produced by the World Bank which tend to be anchored on a selected benchmark. The econometric formulation is based on a regression model for the national price levels where the disturbances are assumed to be heteroskedastic and spatially correlated across countries. The regression model along with data on country specific price movements are combined using a state-space formulation and optimum predictions of PPPs are obtained. As a property of the method presented in the paper, we show that the resulting PPP predictions are weighted averages of extrapolations of PPPs from different benchmarks-thus the method provides a formal approach which has a simple intuitive interpretation. The smoothed PPP predictions (and standard errors) obtained through the state-space are produced for both ICP-participating and non-participating countries and non-benchmark years. A complete tableau of PPPs for 141 countries spanning the period 1970 to 2005 is compiled using the method. Results for some selected countries are presented and the new series are compared and contrasted with the currently available PWT series. Extrapolated series for the remaining countries are available from the authors upon request.
The harmonic model has important properties which enable OLS (Ordinary Least Squares) to be used to estimate efficiently a stable seasonal pattern. These properties are considered, and a procedure is suggested to facilitate a trade-off between simplicity and precision in the specification of the model. Two empirical examples illustrating this procedure are presented. where" and (Ut) is a stationary series with u=[:J [0S A, sin Al ... COS Aj X = C~S 2A, sin 2AI ... cos 2A6 (2.4) cos nAI sin nAI .. . cos nA (2.3) 6 (2.1) Yt = L {(Xk cos Akt +{jk sin Akt } + Ut, k=l (2.2)erties and suggest a procedure that enables them to be used to advantage.
The Harmonic Model and Its PropertiesIn what follows it will be assumed that interest lies in estimating the seasonal component of the monthly series (Yt) and that any trend in the data has been removed by a suitable linear filter. This filter will affect the estimate of the seasonal component, but this estimate (from the de trended data) can always be adjusted by multiplication by a factor" derived from the frequency response function of the filter.Hannan's model [13, p. 32] is adopted, i.e.,It is also assumed that observations have been taken over m complete years, so that t= 1, 2, . .. , nand n = 12m. Defining vectors Y, 0, and u, and the (n X 11) matrix X as:S.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.