## The Income Statement Other Components

You would then go down item by item on the income statement. Your next estimate would be for COGS. You have a whole range of options, including but not limited to a plain growth forecast (similar to what we used for sales).

A Plain Growth Forecast: You could repeat the sales exercise with COGS: A pure growth model would project that COGS' historical growth rate of (\$10, 754/\$10, 326)1/2 -1 « 2.05% will continue in 2002. If applied to the year 2001 COGS of \$10,754, your 2002 COGS forecast would thus be \$10, 754 ■ (1 + 2.05%) « \$10,975. A Pure Proportion of Sales Forecast: Forecast COGS not only relative to its own history, but also relative to your already projected sales of \$27,906 for 2002. You also know the historical relationship between COGS and sales, which you can use to predict a relationship between 2002 sales and 2002 COGS. For example, PepsiCo's COGS was \$10, 326/\$25,093 « 41.15% of sales in 1999,40.14% of sales in 2000, and 39.93% of sales in 2001. The simplest sales-based model might just project that COGS would be a slowly declining fraction of sales in 2002. In this case, your COGS forecast might be

Direct extrapolation of COGS is possible. But it can now also be projected in relation to (as a fraction of) sales.

An Economies-of-Scale Forecast: A more sophisticated model might pose economies of scale.

In this case, COGS would not go up proportionally with sales. Instead, COGS would have both a "fixed component," whose cost would not change with sales (e.g., the factories), and a "variable component," whose costs would increase with sales (e.g., the cola syrup) but less than one-to-one. You might try to plot COGS against sales for 1999-2001 and determine visually that a good line fit would be

This says that \$3.5 billion is unalterable factory costs, but for each extra dollar of sales, you have to purchase only 25 cents of syrup. Substituting in our estimated 2002 sales of \$27,906 million, you would project COGS for 2002 to be

E ( COGS2002 ) « \$3, 500 + 25% ■ (\$27,906) « \$10, 500 (24.4)

Or, you could use heavier statistical artillery and run a regression relating PepsiCo's COGS to sales over its most recent three years. (Don't worry if you do not know what this is.) Such a regression suggests that a better line fit would be

E ( COGS2002 ) « \$3, 506 + 26% -E ( Sales2002 ) (24.5)

so your prediction would change to

E ( COGS2002 ) « \$3, 506 + 26% ■ \$27,906 « \$10, 760 (24.6)

An Industry-Based Forecast: You could draw on information from other firms, such as Coca Cola. In 2001, Coca Cola had COGS of \$6,044 on sales of \$20,092, a ratio of 30%, which is much lower than PepsiCo's. This may not only suggest that Coca Cola's business is different, but also that PepsiCo may be able to lower its COGS in the future to meet "better practice" standards. Thus, you might want to lower PepsiCo's COGS estimate from \$10,760.

A Disaggregated Forecast: If you were even more sophisticated, you could recognize that COGS contains some depreciation. Thus, the history of PepsiCo's past capital expenditures could also influence your COGS estimate. You could throw past capital expenditures into your statistical regression, too, to come up with a better predictive formula.

The sky—your economic and econometric background knowledge—is your limit. For illustration's sake, let's adopt \$10,760 from Formula 24.6 as our predicted COGS in Table 24.3.

Other items in the You can repeat these forecasting processes to predict other income statement items. Again, table may followOdthesr you have many options. Like COGS, SG&A contains both fixed and variable expenses, as well as . depreciation that relates to past investments. SG&A might thus be modeled as a combination of a fixed component, plus a sales-variable component, plus a past capital expenditure-based component. There is also no need to remain consistent across different items—you could use one method to estimate COGS and another to estimate SG&A (or any other financial statement item, for that matter). For example, you could relate net interest income to how much debt PepsiCo currently has and what you know current interest rates are and what you believe future interest rates will be. But because no money (only scarce book space) is at stake, for the rest of the income statement, let's play it simple. The footnotes in Table 24.3 describe the method of projection for each item. Clearly, if your money was at stake, you would want to know as much about the business as possible and use this knowledge to come up with better models for the relationships between PepsiCo's financial variables. Again, the limit is only your knowledge— and for our PepsiCo example, it is obviously very limited, indeed.

### SIDE NOTE

In the appendix to this chapter, there are similar formulas for many pro forma components estimated with data from the universe of publicly traded companies. These can be used "in-a-pinch"—or even to help you gain some intuition about how important the fixed and variable components are in a particular data item. However, the formulas there are mechanistic and therefore definitely not particularly reliable in any individual case—so be careful.

statement model would rely on the income statement

0 0