Working Paper: NBER ID: w10304
Authors: Shiusheng Chen; Charles Engel
Abstract: Recently, Imbs et. al. (2002) have claimed that much of the purchasing power parity puzzle can be explained by aggregation bias'. This paper re-examines aggregation bias. First, it clarifies the meaning of aggregation bias and its applicability to the PPP puzzle. Second, the size of the bias' is shown to be much smaller than the simulations in Imbs et. al. (2002) suggest, if we rule out explosive roots in the simulations. Third, we show that the presence of non-persistent measurement error especially in the Imbs et. al. (2002) data can make price series appear less persistent than they really are. Finally, it is now standard to recognize that small-sample bias plagues estimates of speeds of convergence of PPP. After correcting small sample bias by methods proposed by Kilian (1998) and by So and Shin (1999), the half-life estimates indicate that heterogeneity and aggregation bias do not help to solve the PPP puzzle.
Keywords: No keywords provided
JEL Codes: F31; C22; C23
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| nonpersistent measurement error (C83) | persistence of price series (E30) |
| small sample bias (C83) | halflife of deviations from PPP (F31) |
| correcting for small sample bias (C83) | halflife estimates (C41) |
| aggregation bias (C43) | perception of the PPP puzzle (F31) |
| aggregation bias (C43) | speed of price adjustment (D41) |