HARMONIC-ANALYSIS, TIME-SERIES VARIATIONS AND THE DISTRIBUTIONAL PROPERTIES OF FINANCIAL RATIOS

This paper models the behaviour of financial ratios using the techniqu es of continuous time stochastic calculus. Previous work in the area h as been restricted to models based on first order stochastic different ial equations. However, in the present paper we model a ratio's displa cement from its long term mean in terms of a second order stochastic d ifferential equation. In this way we show that higher order equations may be used to provide more flexible modelling procedures than those p reviously studied. The paper begins by describing a ratio's evolution in terms of a particular form of second order stochastic differential equation. A solution is then obtained for this equation. We then show how the parameters of the model may be estimated from a given set of e mpirical data. The model is then applied to a data set of four ratios for 118 UK companies covering a period of 37 years. For three of the f our ratios, clear evidence of a period emerges. Previous work suggests that these ratios are well described by mean reversion processes. Our empirical analysis, however, suggests there is a significant tendency for adjustments to 'overshoot' the targeted long term mean. Taken tog ether with prior work in the area, the paper begins to provide a broad picture of the way in which financial ratios evolve over time.