Working Paper: NBER ID: w17978
Authors: Mario J. Crucini; Christopher I. Telmer
Abstract: We provide three sets of variance decompositions on microeconomic international relative price data. The first shows that the overall distribution of absolute deviations from the Law of One Price (LOP) is dominated by cross-sectional variation in long-term averages, not by time-series variation around the long-term averages. The second shows that time-series variation in changes in LOP deviations is dominated by idiosyncratic, goods-specific variation, not by aggregate variation such as that arising from nominal exchange rates. The third shows that time-series and cross-sectional variance are connected across goods. Goods that exhibit high cross-sectional variance also exhibit high time-series variance. Moreover, when this connection is made conditional on the tradeability of a goods, a two-factor structure for the goods-specific cross-section is revealed. We argue that this factor structure, in addition to our other variance decompositions, is informative for the construction of models that can synthesize the micro and macroeconomic behavior of relative prices.
Keywords: real exchange rates; law of one price; variance decomposition
JEL Codes: E3; F3; F4
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Cross-sectional variance in long-term absolute deviations from the LOP (C21) | Time series variance in long-term absolute deviations from the LOP (C22) |
| Changes in LOP deviations (C29) | Goods-specific factors (L68) |
| Cross-sectional variance in long-term absolute deviations from the LOP (C21) | Cross-sectional variance in time series (C22) |
| Cross-sectional variance in long-term absolute deviations from the LOP (C21) | Tradeability of goods (F10) |
| Positive relationship between cross-sectional variance and time series variance (C32) | High variability in long-term prices (E39) |