Scraped data and prices in macroeconomics

This dissertation consists of three chapters on the microeconomic behavior of prices and its implications for macroeconomic models. It uses Scraped Data collected on a daily basis from online retailers to provide empirical insights on the behavior of individual prices in a much larger set of countries and economic contexts that has been previously possible in the micro-price literature. The first chapter presents stylized empirical facts on price stickiness in four emerging economies. It shows that the distribution in the size of price changes is bimodal--with few changes close to zero percent--the aggregate hazard functions are upward sloping or hump-shaped, and there is synchronization of price changes for competing brands. These facts challenge commonly-held views in the price-stickiness literature that have greatly influenced theoretical work in the past. The second chapter, co-authored with Roberto Rigobon, formally tests one of these facts--the bimodality of the size of changes--in a larger sample of 37 supermarkets in 23 countries. It uses two statistical tests--Hartigan's Dip and Silverman's Bandwidth--and proposes a new method--the Proportional Mass Test--to measure the degree of unimodality around zero and the largest mode. The evidence rejects unimodality at zero percent, but finds support for the existence of large modes away from zero. The third chapter provides alternative price indexes in Argentina, where official statistics have become unreliable in recent years. It shows that annual inflation is consistently two to three times higher than officially reported. The paper serves as an introduction to scraped-price indexes. which can be computed automatically every day and can serve as early-warning indicators for inflation in countries with volatile macroeconomic settings.