Trends, cycles, and structural breaks in cliometrics

Abstract

The calculation of trends and their growth rates, along with the related calculation of cycles, is an important area of cliometrics. The methods traditionally employed to estimate trend were either the estimation of regressions containing simple functions of time, typically in conjunction with a method to deal with regime shifts or structural breaks, or simple unweighted moving averages. In both cases the cycle was determined by residual and, because the trend was, possibly locally, deterministic, the cyclical component took up most of the fluctuations in the observed series. The last 25 years or so, however, have seen major developments in both macroeconomics and time series econometrics and statistics on the modelling of trends and cycles that allow all components to be stochastic and perhaps determined by the statistical properties of the observed time series. This chapter provides a survey of these developments.