Introduction

Abstract

Modelling trends and cycles in time series has a long history in empirical economics, stretching back to the latter part of the nineteenth century. Until then, few economists recognised the existence of regular cycles in economic activity nor the presence of longer-term, secular movements. Rather than cycles, they tended to think in terms of ‘crises’, used to mean either a financial panic or a period of deep depression. The early studies of business cycles, notably the Sunspot and Venus theories of Jevons and Moore and the rather more conventional credit cycle theory of Jugler, are discussed in detail in Morgan (1990). The analysis of secular movements was even rarer, a notable example being Poynting (1884), who was the first to introduce moving averages. Although such movements are nowadays typically referred to as trends, the term ‘trend’ was coined only in 1901 by Hooker (1901) when analysing British import and export data. The early attempts to take into account trend movements, typically by detrending using simple moving averages or graphical interpolation, are analysed by Klein (1997).