Working Paper: NBER ID: w10614
Authors: Chiyoung Choi; Nelson C. Mark; Donggyu Sul
Abstract: Three potential sources of bias present complications for estimating the half-life of purchasing power parity deviations from panel data. They are the bias associated with inapproiate aggregation across heterogeneous coefficients, time aggregation of commodity prices, and downward bias in estimation of dynamic lag coefficients. Each of these biases have been addressed individually in the literature. In this paper, we address all three biases in arriving at our estimates. Analyzing an annual panel data set of real exchange rates for 21 OECD countries from 1948 to 2002, our point estimate of the half-life is 5.5 years.
Keywords: Purchasing Power Parity; Panel Data; Halflife; Bias Estimation
JEL Codes: C32; F31; F47
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| inappropriate pooling across heterogeneous coefficients (C20) | biased estimates of the halflife (C41) |
| time aggregation of commodity prices (Q02) | biased estimates of the halflife (C41) |
| downward bias due to inclusion of a constant in dynamic regression (C22) | biased estimates of the halflife (C41) |
| controlling for biases (C90) | more accurate estimate of the halflife (C41) |
| controlling for biases (C90) | halflife of purchasing power parity (PPP) convergence (F31) |