. 81
( 208 .)


Business Analysis and 2
Valuation Tools

cate their views of the ¬rm™s prospects to investors; bankers and debt market participants
need forecasts to assess the likelihood of loan repayment. Moreover, there are a variety
of contexts (including but not limited to security analysis) where the forecast is usefully
summarized in the form of an estimate of the ¬rm™s value”an estimate that, after all,
can be viewed as the best attempt to re¬‚ect in a single summary statistic the manager™s
or analyst™s view of the ¬rm™s prospects.
Prospective analysis includes two tasks”forecasting and valuation”that together
represent approaches to explicitly summarizing the analyst™s forward-looking views. In
this chapter, we focus on forecasting. Valuation is the topic of the following two chap-
ters. The key concepts discussed in this chapter are again illustrated using analysts™ fore-
casts for Nordstrom.

Forecasting is not so much a separate analysis as it is a way of summarizing what has
been learned through business strategy analysis, accounting analysis, and ¬nancial anal-
ysis. For example, a projection of the future performance of Nordstrom as of early 1999
must be grounded ultimately in an understanding of questions such as:
• From business strategy analysis: What will Nordstrom™s recent focus on restruc-
turing to enhance shareholder value mean for future margins and sales volume?
What will it imply about the need for working capital and capital expenditures?
• From accounting analysis: Are there any aspects of Nordstrom™s accounting that
suggest past earnings and assets are overstated, or expenses or liabilities are over-
stated? If so, what are the implications for future accounting statements?
• From financial analysis: What are the sources of the improvement in Nordstrom™s
margin in 1998? Is the improvement sustainable? Has Nordstrom™s shift in business
strategy translated into improvements in asset utilization in 1998? Can any such im-
provements in efficiency be sustained or enhanced? Will Nordstrom change its debt


376 Prospective Analysis: Forecasting

Prospective Analysis: Forecasting

The upshot is that a forecast can be no better than the business strategy analysis, ac-
counting analysis, and ¬nancial analysis underlying it. However, there are certain tech-
niques and knowledge that can help a manager or analyst to structure the best possible
forecast, conditional on what has been learned in the previous steps. Below, we summa-
rize an approach to structuring the forecast, some information useful in getting started,
and some detailed steps used to forecast earnings, balance sheet data, and cash ¬‚ows.


The Overall Structure of the Forecast
The best way to forecast future performance is to do it comprehensively”producing not
only an earnings forecast, but a forecast of cash ¬‚ows and the balance sheet as well. A
comprehensive approach is useful, even in cases where one might be interested primarily
in a single facet of performance, because it guards against unrealistic implicit assump-
tions. For example, if an analyst forecasts growth in sales and earnings for several years
without explicit consideration of the required increases in working capital and plant as-
sets and the associated ¬nancing, the forecast might possibly imbed unreasonable as-
sumptions about asset turnover, leverage, or equity capital infusions.
A comprehensive approach involves many forecasts, but in most cases they are all
linked to the behavior of a few key “drivers.” The drivers vary according to the type of busi-
ness involved, but for businesses outside the ¬nancial services sector, the sales forecast is
nearly always one of the key drivers; pro¬t margin is another. When asset turnover is ex-
pected to remain stable”as is often realistic”working capital accounts and investment in
plant should track the growth in sales closely. Most major expenses also track sales, subject
to expected shifts in pro¬t margins. By linking forecasts of such amounts to the sales fore-
cast, one can avoid internal inconsistencies and unrealistic implicit assumptions.
In some contexts, the manager or analyst is interested ultimately in a forecast of cash
¬‚ows, not earnings per se. Nevertheless, even forecasts of cash ¬‚ows tend to be grounded
in practice on forecasts of accounting numbers, including sales and earnings. Of course, it
would be possible in principle to move directly to forecasts of cash ¬‚ows”in¬‚ows from
customers, out¬‚ows to suppliers and laborers, and so forth”and in some businesses, this
is a convenient way to proceed. In most cases, however, the growth prospects and pro¬t-
ability of the ¬rm are more readily framed in terms of accrual-based sales and operating
earnings. These amounts can then be converted to cash ¬‚ow measures by adjusting for the
effects of noncash expenses and expenditures for working capital and plant.

Getting Started: Points of Departure
Every forecast has, at least implicitly, an initial “benchmark” or point of departure”
some notion of how a particular amount, such as sales or earnings, would be expected to
Prospective Analysis: Forecasting

10-3 Part 2 Business Analysis and Valuation Tools

behave in the absence of detailed information. For example, in beginning to contemplate
1999 pro¬tability for Nordstrom, one must start somewhere. A possibility is to begin
with the 1998 performance. Another starting point might be 1998 performance adjusted
for recent trends. A third possibility that might seem reasonable”but one that generally
turns out not to be very useful”is the average performance over several prior years.
By the time one has completed a business strategy analysis, an accounting analysis,
and a detailed ¬nancial analysis, the resulting forecast might differ signi¬cantly from the
original point of departure. Nevertheless, simply for purposes of having a starting point
that can help anchor the detailed analysis, it is useful to know how certain key ¬nancial
statistics behave “on average.”
In the case of some key statistics, such as earnings, a point of departure or benchmark
based only on prior behavior of the number is more powerful than one might expect. Re-
search demonstrates that some such benchmarks for earnings are not much less accurate
than the forecasts of professional security analysts, who have access to a rich informa-
tion set. (We return to this point in more detail below.) Thus, the benchmark is often not
only a good starting point, but also close to the amount forecast after detailed analysis.
Large departures from the benchmark could be justi¬ed only in cases where the ¬rm™s
situation is demonstrably unusual.
Reasonable points of departure for forecasts of key accounting numbers can be based
on the evidence summarized below. Such evidence may also be useful for checking the
reasonableness of a completed forecast.

THE BEHAVIOR OF SALES GROWTH. Sales growth rates tend to be “mean-revert-
ing”: ¬rms with above-average or below-average rates of sales growth tend to revert over
time to a “normal” level (historically in the range of 7 to 9 percent for U.S. ¬rms) within
no more than three to ten years. Figure 10-1 documents this effect for U.S. ¬rms for
1979“1998. All ¬rms are ranked in terms of their sales growth in 1979 (year 1) and
formed into ¬ve portfolios based on the relative ranking of their sales growth in that year.
Firms in portfolio 1 have the top twenty percent of rankings in terms of their sales
growth in 1979, and those in portfolio 2 fall into the next twenty percent; those in port-
folio 5 have the bottom twenty percent sales growth ranks. The sales growth rates of each
of the ¬ve portfolios plotted in Figure 10-1 in year +1 to year +10 are averaged across
three experiments. The sales growth rates of ¬rms in each of these ¬ve portfolios are
traced from 1979 through the subsequent nine years (years 2 to 10). The same experi-
ment is repeated with 1984 and then 1989 as the base year (year 1).
The ¬gure shows that the group of ¬rms with the highest growth initially”sales
growth rates of more than 50 percent”experience a decline to about 6 percent growth
rate within three years and are never above 13 percent in the next seven years. Those
with the lowest initial growth experience an increase to about 8 percent growth rate by
year 5, and never fall below 5 percent after that. All ¬ve portfolios, irrespective of their
starting growth levels, revert to “normal” levels of sales growth of between 7 and 9 per-
cent within ¬ve years.
378 Prospective Analysis: Forecasting

Prospective Analysis: Forecasting

Figure 10-1 Behavior of Sales Growth over Time for U.S. Companies
for 1979“1998
50% Top Fifth
Sales Growth

Second Fifth
Third Fifth
10% Fourth Fifth
Bottom Fifth
1 2 3 4 5 6 7 8 9 10

One explanation for the pattern of sales growth seen in Figure 10-1 is that as industries
and companies mature, their growth rate slows down due to demand saturation and intra-
industry competition. Therefore, even when a ¬rm is growing rapidly at present, it is
generally unrealistic to extrapolate the current high growth inde¬nitely. Of course, how
quickly a ¬rm™s growth rate reverts to the average depends on the characteristics of its
industry and its own competitive position within an industry.

THE BEHAVIOR OF EARNINGS. Earnings have been shown, on average, to follow a
process that can be approximated by a “random walk” or “random walk with drift”; thus,
the prior year™s earnings is a good starting point in considering future earnings potential.
As will be explained in more detail later in the chapter, it is reasonable to adjust this sim-
ple benchmark for the earnings changes of the most recent quarter (that is, changes ver-
sus the comparable quarter of the prior year after controlling for the long-run trend in
the series). Even a simple random walk forecast”one that predicts next year™s earnings
will be equal to last year™s earnings”is surprisingly useful. One study documents that
professional analysts™ year-ahead forecasts are only 22 percent more accurate (on aver-
age) than a simple random walk forecast.1 Thus, a ¬nal earnings forecast will usually not
differ dramatically from a random walk benchmark.
The implication of the evidence is that, in beginning to contemplate future earnings
possibilities, a useful number to start with is last year™s earnings; the average level of
earnings over several prior years is not. Long-term trends in earnings tend to be sus-
tained on average, and so they are also worthy of consideration. If quarterly data are also
considered, then some consideration should usually be given to any departures from the
long-run trend that occurred in the most recent quarter. For most ¬rms, these most recent
changes tend to be partially repeated in subsequent quarters.2
Prospective Analysis: Forecasting

10-5 Part 2 Business Analysis and Valuation Tools

THE BEHAVIOR OF RETURNS ON EQUITY. Given that prior earnings serves as a
useful benchmark for future earnings, one might expect the same to be true of rates of
return on investment, like ROE. That, however, is not the case, for two reasons. First,
even though the average ¬rm tends to sustain the current earnings level, this is not true
of ¬rms with unusual levels of ROE. Firms with abnormally high (low) ROE tend to ex-
perience earnings declines (increases).3
Second, ¬rms with higher ROEs tend to expand their investment bases more quickly
than others, which causes the denominator of the ROE to increase. Of course, if ¬rms
could earn returns on the new investments that match the returns on the old ones, then
the level of ROE would be maintained. However, ¬rms have dif¬culty pulling that off.
Firms with higher ROEs tend to ¬nd that, as time goes by, their earnings growth does not
keep pace with growth in their investment base, and ROE ultimately falls.
The resulting behavior of ROE and other measures of return on investment is charac-
terized as “mean-reverting”: ¬rms with above-average or below-average rates of return
tend to revert over time to a “normal” level (for ROE, historically in the range of 10 to
15 percent for U.S. ¬rms) within no more than ten years.4 Figure 10-2 documents this
effect for U.S. ¬rms for 1979“1998. All ¬rms are ranked in terms of their ROE in 1979
(year 1) and formed into ¬ve portfolios. Firms in portfolio 1 have the top twenty percent
ROE rankings in 1979, those in portfolio 2 fall into the next twenty percent, and those in
portfolio 5 have the bottom twenty percent sales growth ranks. The average ROE of ¬rms
in each of these ¬ve portfolios is then traced through nine subsequent years (years 2 to
10). The same experiment is repeated with 1984 and 1989 as the base year (year 1), and
the subsequent years as years +2 to +10. Figure 10-2 plots the average ROE of each of
the ¬ve portfolios in years 1 to 10 averaged across these three experiments.
The most pro¬table group of ¬rms initially”with average ROEs of 27 percent”
experience a decline to 17 percent within three years. By year 10, this group of ¬rms has
an ROE of 14 percent. Those with the lowest initial ROEs (“33 percent) experience an
increase in ROE until they reach a level of 13 percent in year 10. Three of the ¬ve port-
folios record an average ROE in the range of 13 to 15 percent by year 10, even though
they start out in year 1 with a wide range of average ROEs.
The pattern in Figure 10-2 is not a coincidence; it is exactly what the economics of
competition would predict. The tendency of high ROEs to fall is a re¬‚ection of high prof-
itability attracting competition; the tendency of low ROEs to rise re¬‚ects the mobility of
capital away from unproductive ventures toward more pro¬table ones.
Despite the general tendencies documented in Figure 10-2, there are some ¬rms
whose ROEs may remain above or below normal levels for long periods of time. In some
cases, the phenomenon re¬‚ects the strength of a sustainable competitive advantage (e.g.,
Wal-Mart), but in other cases, it is purely an artifact of conservative accounting methods.
A good example of the latter phenomenon in the U.S. is pharmaceutical ¬rms, whose
major economic asset (the intangible value of research and development) is not recorded
on the balance sheet and is therefore excluded from the denominator of ROE. For those
¬rms, one could reasonably expect high ROEs”in excess of 20 percent”over the long
run, even in the face of strong competitive forces.
380 Prospective Analysis: Forecasting

Prospective Analysis: Forecasting

Figure 10-2 Behavior of ROE over Time for U.S. Companies


. 81
( 208 .)