Working Paper: NBER ID: w21490
Authors: Alberto Cavallo
Abstract: This paper introduces Scraped Data as a new source of micro-price information to measure price stickiness. Scraped data, collected from online retailers, have no time averaging or imputed prices that can affect pricing statistics in traditional sources of micro-price data. Using daily prices of 80 thousand products collected in five countries with varying degrees of inflation, including the US, I find that relative to previous findings in the literature, scraped online prices tend to be stickier, with fewer price changes close to zero percent, and with hump-shaped hazard functions that initially increase over time. I show that the sampling characteristics of the data, which minimize measurement biases, explain most of the differences with the literature. Using the cross-section of countries, I also show that only the relative frequency of price increases over decreases correlates with inflation.
Keywords: Scraped Data; Price Stickiness; Measurement Bias
JEL Codes: E30; E60
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Measurement biases (C83) | distorted frequency and size of price changes (E39) |
| time averaging (C22) | misrepresentation of price dynamics (E30) |
| relative frequency of price increases (E30) | higher inflation (E31) |