Short and Variable Lags
Gergely Buda, Vasco M. Carvalho, Giancarlo Corsetti, João B. Duarte, Stephen Hansen, Álvaro Ortiz, Tomasa Rodrigo, José V. Rodríguez Mora – 2023
Abstract
We study the transmission of monetary policy shocks using daily consumption, corporate sales and employment series. We find that the economy responds at both short and long lags that are variable in economically significant ways. Consumption reacts in one week, reaches a local trough in one quarter, recovers, and declines again after three quarters. Sales follow a similar pattern, but the initial drop, while delayed (one month), is deeper. In contrast, employment falls monotonically for five quarters albeit with a smaller impact reaction. We show that these short lags are masked by time aggregation at lower —quarterly— frequencies.
Summary
TK
Demonstrates cool way to calculate level impulse responses from Y/Y growth IRFs.
There’s a lot of smoothing/filtering here – they use a 90-day moving average on the daily data, and then in addition use y/y growth rates in the local projections. So in effect you have extremely smoothed data, which might be necessary given the inherent noisiness of daily data. Something to potentially consider for my purposes as an alternative to LOESS.