The Real Effects of Monetary Shocks: Evidence from Micro Pricing Moments

Abstract

This paper evaluates the informativeness of eight micro pricing moments for monetary non-neutrality. Frequency of price changes is the only robustly informative moment. The ratio of kurtosis over frequency is significant only because of frequency, and insignificant when non-pricing moments are included. Non-pricing moments are additionally informative about monetary non-neutrality, indicating potential omitted variable bias and the inability of pricing moments to serve as sufficient statistics. In contrast to existing theoretical work, this ratio has an ambiguous relationship with monetary non-neutrality in a quantitative menu cost model. We show which modeling ingredients explain this discrepancy, providing guidance on modeling choices.

Multi-Product Pricing: Theory and Evidence from Large Retailers

“…Second, improved datasets can balance representation of the economy by incorporating small-and medium-size stores, and by including durables, services, and other goods. Third, advancements in identification of monetary policy shocks and longer data can provide corroborating evidence on the role of price synchronization in monetary transmission (Hong et al, 2020). Finally, adding information about retailers' production costs (Dedola, Kristoffersenz, and Züllig, 2019), inventories and stockouts (Kryvtsov and Midrigan, 2013), relationships with suppliers (Anderson et al, 2017), and the costs of price adjustments (Levy et al, 1997) is crucial for further understanding the elements of the price-decision problem faced by retailers.…”

Empirical Investigation of a Sufficient Statistic for Monetary Shocks

“…Evidence is scarcer on kurtosis. For US PPI data, Hong et al (2020) report and average kurtosis of 4.9. With consumer price data, Cavallo (2018) report a median kurtosis of 4.8 in a large sample of countries based on "scraped" data.…”

“…Their estimates show that the ratio of kurtosis over frequency correlates negatively with CIR P as predicted by the sufficient statistic theory and that both frequency and kurtosis taken separately correlate equally and significantly with CIR P and with the expected sign. 5 A related analysis is developed by Hong et al (2020), who inspect the correlation between the response of sectoral producer price indices in the United States and several cross-sectional moments of the distribution of price changes. They document a positive correlation between the price response to monetary shocks and the ratio of kurtosis over frequency as well as frequency alone, but provide mixed evidence on the role of kurtosis, which is entered as the sole regressor in several specifications.…”

“…See e.g. columns 9 to 13 in Table 2 in Hong et al (2020), or the cross sectional regressions of Table 16 where both kurtosis and frequency are statistically significant and with the expected signs (especially so when fixed effects are not used).…”