The Portable MBA in Finance and Accounting, 3rd Edition

192 Understanding the Numbers

extend linearly with a continuation of the same slope that was estimated in the
trend line fit through the data.
A potential problem with fitting a trend line through the data with re-
gression analysis is that each observation is treated the same way. That is, we
are not weighting the information contained in the latest set of observations
more heavily than those that occurred 30 quarters ago. Other statistical tech-
niques are available to address this concern. One of these is exponential
smoothing. Exhibit 6.8 presents the same quarterly sales data with a trend line
that has been exponentially smoothed.

EXHIBIT 6.6 Kellogg company’s quarterly sales (1990 –2000).

Index^10203040

Sales ($)

Quarters

1,900,000

1,800,000

1,700,000

1,600,000

1,500,000

1,400,000

EXHIBIT 6.7 Trend analysis for Kellogg company’s quarterly sales
(1990 –2000).

Actual

Fits

Forecasts

01020304050

Quarters

Salest = $1,475,002 + $8,357.73 ×t

MAPE:

MAD:

MSD:

Linear Trend Model

Sales ($)

1,900,000

1,800,000

1,700,000

1,600,000

1,500,000

1,400,000 67,504^4

7.65E+09