University of Neuchâtel, Institute of Economic Research (IRENE), Neuchâtel
Measurement error in historical data distorts descriptive analyses based on binary classifications. Modern replications of deficiencies in retrospective CPI estimates for the 19th century show that measurement issues cause misclassification of inflationary and deflationary episodes. We therefore underestimate the shortfall in real activity during deflation. Using various approaches to control for measurement error in 19th century US CPI data, a series of stylized facts emerge: (i) Real activity was on average substantially lower during deflations; (ii) CPI deflations were associated with at least as severe shortfalls in real activity as equity price declines and banking crises; (iii) Only severe deflations were associated with declines in real activity; (iv) Transitory and persistent deflations, as well as, monetary and nonmonetary deflations were equally associated with lower GDP growth.
Deflation real activity monetary history measurement error binary regressors misclassification bias bounds GMM