Price Duration Using Daily Online Data: Time- or State-Dependent?

Using daily retail prices gathered through web scraping in Mexico, we analyze if price changes can be characterized by time-dependent features, like the duration of the price spell, and/or by variables associated with the state of the economy. Through the lens of a duration model, we find evidence of both time- and state-dependency behavior. Favoring time-dependency, on the one hand, estimates indicate that price spells exhibit greater risk of ending every seven days relative to other days in between. Advocating for state-dependency, the probability of price changes seems to be affected by variations in the USD/MXN exchange rate, variations in real point of sales expenditures and the COVID-19 pandemic. Finally, leveraging data gathered via direct visits to brick-and-mortar stores, we also find time-dependency and state-dependency in the duration of price spells.