Following on from the work of Birchenhall, Jessen, Osborn & Simpson (1999) on predicting US business cycle regimes we apply the same methodology to construct a one period ahead model of classical business cycle regimes in the UK. Birchenhall et al generated the regime data from the NBER dating of peaks and troughs. In the UK there is no comparable dating committee and our first task is to date the UK peaks and troughs. Application of a simple mechanical rule based on changes in GDP produces a set of acceptable turning points, with one exception that is attributable to the 3-day working week in 1974. We are left with 3 peaks at 1973 Q3, 1979 Q2 and 1990 Q2 together with troughs at 1975 Q3, 1981 Q1 and 1992 Q2. Starting with a number of leading indicators several parsimonious one-period-ahead models are selected largely on the basis of the SIC criterion. A number of interesting results have emerged from this investigation. One model that performs well is based on "real" stock market prices and "real" term structure, together with differences in "real" M4. The first two terms are interpreted as measures of expected return on investment and the third term is interpreted as a measure of unexpected growth in liquidity. Both stock prices and M4 anticipate the 73-75 recession, the 79-81 recession is anticipated by the term structure while the 90-92 recession is anticipated by M4.
