Our cyclical models assume that corporate performance will be affected by the business or an industry cycle, and that corporate performance as measured by returns on capital and rates of growth and reinvestment follow cyclical patterns. Models of this sort have the advantage that they inherently embed a margin of safety at times of peak or optimal performance, and on the other hand also provide for a more optimistic or realistic perspective in the event of atypically depressed performance in the near term.
Cycle duration: We currently use cycles with a duration of four to six years (four to six reporting periods from a trough to the previous trough). This appears consistent with more recent business cycle durations, for example in Europe and the United States.
Cycle peaks and troughs: We use two main alternatives:
- Cycles ranging from historical minima to historical maxima
- Cycles with a range around the historical average
Cycle form: We use the following alternatives:
- Seesaw or zig-zag patterns – essentially a recurring linear progression from minima to maxima
- Trigonometric functions
- Asymmetric patterns – fast improvements from cycle troughs to average historical performance with decreasing improvements to cycle peaks, followed by sudden drops to cycle troughs
We prefer asymmetric patterns since these appear to resemble actual historical performance patterns very closely. These are somewhat more demanding to implement, but we use solutions that fit historical performance patterns very well and translate into realistic valuation perspectives.
