726  Cyclical Companies
DCF values (the values are indexed for comparability). It shows that the DCF 
value is far less volatile than the underlying cash flow, because no single year’s 
performance has a significant impact on the value of the company.
In the real world, the share prices of cyclical companies are less stable than 
the example in Exhibit 37.1. Exhibit 37.2 shows the earnings per share (EPS) 
EXHIBIT 37.2  Share Prices and Earnings per Share: 15 Cyclical Companies
Trough + 2
Trough + 1
Trough
Peak + 2
Peak + 1
Peak
Peak – 1
Peak – 2
Peak – 3
–1.5
–1.0
–0.5
0
0.5
1.0
1.5
2.0
2.5
Share price
EPS
Index
EXHIBIT 37.1  The Long-Term View: Free Cash Flow and DCF Volatility
Free cash flow pattern, Company A, $ million
 
0 
1 
2 
3 
4 
5 
6 
7 
8 
9 
10
After-tax operating profit 
10 
9 
6 
3 
– 
(2) 
3 
18 
7 
6 
10
Net investment 
(3) 
(3) 
(2) 
(2) 
(1) 
(3) 
(5) 
(3) 
(3) 
(3) 
(3)
Free cash flow 
7 
6 
4 
1 
(1) 
(5) 
(3) 
15 
4 
3 
7
DCF value 
34 
33 
27 
28 
30 
35 
40 
33 
33 
34 
31
Free cash flow and DCF value patterns
Index
DCF value
Free cash flow
250
200
150
100
50
0
–50
–100
1
2
3
5
7
8
9
10
Period, years
3
1
Cash flows valued from any 1 year forward
2
4
6
Period, years

Share Price Behavior  727
and share prices, both indexed, for 15 companies with a four-year cycle. The 
share prices are more volatile than the DCF approach would predict, which 
suggests that market prices exhibit the bias of anchoring on current earnings.
How can this apparent anomaly be explained? We examined equity analysts’ 
consensus earnings forecasts for cyclical companies to look for clues to these com-
panies’ volatile stock prices. Consensus earnings forecasts for cyclical companies 
appeared to ignore cyclicality entirely. The forecasts invariably showed an upward-
sloping trend, whether the companies were at the peak or trough of the cycle.
What became apparent was not that the DCF model was inconsistent with 
the facts, but that the market’s projections of earnings and cash flow (assuming 
the market followed the analysts’ consensus) were to blame. This conclusion was 
based on an analysis of 36 U.S. cyclical companies during 1985 to 1997. We di-
vided them into groups with similar cycles (e.g., three, four, or five years from 
peak to trough) and calculated scaled average earnings and earnings forecasts. We 
then compared actual earnings with consensus earnings forecasts over the cycle.1
Exhibit 37.3 plots the actual earnings and consensus earnings forecasts for 
the set of 15 companies with four-year cycles in primary metals and manu-
facturing transportation equipment. The consensus forecasts do not predict 
the earnings cycle at all. In fact, except for the next-year forecasts in the years 
following the trough, the earnings per share are forecast to follow an upward-
sloping path with no future variation.2
EXHIBIT 37.3  Actual EPS and Consensus EPS Forecasts: 15 Cyclical Companies
Trough + 6
Trough + 5
Trough + 4
Trough + 3
Trough + 2
Peak + 2
Peak + 1
Peak 
Peak – 1
Peak – 2
Peak – 3
–0.6
–0.3
0
0.3
0.6
0.9
1.2
1.5
Actual EPS
Forecast EPS
Peak – 4
Trough + 1
Trough
1 Note that we have already adjusted downward the normal positive bias of analyst forecasts to focus 
on just the cyclicality issue. V. K. Chopra, “Why So Much Error in Analysts’ Earnings Forecasts?” Finan-
cial Analysts Journal (November/December 1998): 35–42.
2 Similar results were found for companies with three- and five-year cycles.

728  Cyclical Companies
One explanation could be that equity analysts have incentives to avoid 
predicting the earnings cycle, particularly the down part. Academic research 
has shown that earnings forecasts have a positive bias that is sometimes attrib-
uted to the incentives facing equity analysts at investment banks.3 Pessimistic 
earnings forecasts may damage relations between an analyst’s employer—an 
investment bank—and a particular company. In addition, companies that are 
the target of negative commentary might cut off an analyst’s access to man-
agement. From this evidence, we could conclude that analysts as a group are 
unable or unwilling to predict the cycles for these companies. If the market 
followed analyst forecasts, that behavior could account for the high volatility 
of cyclical companies’ share prices.
We know that it is difficult to predict cycles, particularly their inflection 
points. So it is unsurprising that the market does not get it exactly right. How-
ever, we would be surprised if the stock market entirely missed the cycle, 
as the analysis of consensus forecasts suggests. To address this issue, we re-
turned to the question of how the market should behave. Should it be able to 
predict the cycle and therefore exhibit little share price volatility? That would 
probably be asking too much. At any point, the company or industry could 
break out of its cycle and move to one that is higher or lower, as illustrated in 
Exhibit 37.4.
3 The following articles discuss this hypothesis: M. R. Clayman and R. A. Schwartz, “Falling in Love 
Again: Analysts’ Estimates and Reality,” Financial Analysts Journal (September/October 1994): 66–68; J. 
Francis and D. Philbrick, “Analysts’ Decisions as Products of a Multi-Task Environment,” Journal of Ac-
counting Research 31, no. 2 (Autumn 1993): 216–230; K. Schipper, “Commentary on Analysts’ Forecasts,” 
Accounting Horizons (December 1991): 105–121; B. Trueman, “On the Incentives for Security Analysts to 
Revise Their Earnings Forecasts,” Contemporary Accounting Research 7, no. 1 (1990): 203–222.
EXHIBIT 37.4  When the Cycle Changes
New trend
established
New trend
Cycle returns
Old trend
Today
Earnings