Quantitative Techniques

Forecasting

Forecasting
Introduction
Forecasting is the establishment of future expectations by the analysis of past data, or the formation of opinions. Forecasting is an essential element of capital budgeting.

Capital budgeting requires the commitment of significant funds today in the hope of long term benefits. The role of forecasting is the estimation of these benefits. 2

Forecasting Techniques and Routes

Techniques Routes
Top-down route Bottom-up route

Quantitative

Qualitative

Simple regressions Multiple regressions Time trends Moving averages

Delphi method Nominal group technique Jury of executive opinion Scenario projection
3

Quantitative Forecasting
Quantitative: Regression with related variable Data set of Sales as related to both time and the number of households.
HISTORICAL DATA YEAR 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

HOUSEHOLDS
815 927 1020 987 1213 1149 1027 1324 1400 1295 1348 1422

SALES
2109 2530 2287 3194 3785 3372 3698 3908 3725 4129 4532 4487

Quantitative Forecasting
Quantitative: Sales plotted related to households.
SalesUnits Related to Number of Households
5000

Sales Units

4000 3000 Sales 2000 1000 0 0 500 1000 1500 Numbe r of House holds
5

Quantitative Forecasting
Quantitative: Sales regressed on households.
SUMMARY OUTPUT SALES REGRESSED AS A FUNCTION Edited output OF HOUSEHOLDS from the Excel regression. Regression Statistics Multiple R 0.824389811 R Square 0.67961856 Adjusted R Square 0.644020623 <== "Strength" of the regression Standard Error 429.2094572 Observations 11

Coefficients Standard Error Y Axis Intercept -348.218 913.798 Number of Households 3.316 0.759

t Stat P-value -0.381 0.712 4.369 0.002

6

Quantitative Forecasting
Quantitative: Sales regressed on households.

Predicting with the regression output.

Regression equation is: Sales(for year) = -348.218 + ( 3.316 x households).

Assuming that a separate data set forecasts the number of households at 1795 for the year 2006, then:

Sales(year 2006) = -348.218 + ( 3.316 x 1795)

= 5,604 units.
7

Quantitative Forecasting
Quantitative: Multiple Regression Sales as a function of both time and the number of households.
HISTORICAL DATA YEAR 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

HOUSEHOLDS
815 927 1020 987 1213 1149 1027 1324 1400 1295 1348 1422

SALES
2109 2530 2287 3194 3785 3372 3698 3908 3725 4129 4532 4487

Quantitative Forecasting:
Multiple Regression Line Information From the Excel spreadsheet.
SUMMARY OUTPUT MULTIPLE REGRESSION: SALES ON YEARS and HOUSEHOLDS Regression Statistics Multiple R 0.9216 R Square 0.8494 Adjusted R Square 0.8118 <== "Strength" of regression. Standard Error 312.1217 Observations 11

Y Axis Intercept Calendar Year Households

Coefficients Standard Error -382643.9164 127299.584 193.3326 64.376 0.1368 1.194

t Stat P-value Lower 95% -3.006 0.017 -676197.474 3.003 0.017 44.880 0.115 0.912 -2.616
9

Quantitative Forecasting: Using Multiple Regression

Multiple regression equation is:
Sales in year = -382643.91 +(193.33 x Year) + (0.1368 x Households)

Forecast of sales for the year 2005 is:

Sales in year 2005 = -382643.91 + (193.33 x 2005) + (0.1368 x 1586) = 5200 Units

(Note: the sales forecast relies upon a separate forecast of the number of households, given as 1 586, for 2005.) 10

Quantitative Forecasting
Quantitative: Time Series Regression Sales plotted as a function of time.
Plot of Past Sales Units By Year
5000

Sales Units

4000 3000 2000 1000 0 1990 1995 Year

11

Sales

2000

2005

Quantitative Forecasting: Fitted Regression Line

Sales Regression: Line Fit Plot
5000 4000

Sales

3000 2000 1000 0 1990

Actual

Predicted

1995 Year

2000

2005
12

Quantitative Forecasting: Regression Line Information

From the Excel spreadsheet.
EDITED SUMMARY OUTPUT REGRESSION OF SALES ON YEARS Regression Statistics Multiple R 0.9215 R Square 0.8492 Adjusted R Square 0.8324 <== "Strength" of regression. Standard Error 294.5125 Observations 11

Y axis intercept Slope of line

Coefficients Standard Error -395541.56 56077.1544 199.87 28.0807

t Stat -7.0535 7.1178

P-value 0.0001 0.0001

13

Quantitative Forecasting: Regression Line Use

Equation for the regression line is:
Sales in year = -395541.56 + (199.87 x Year)

Forecast of sales for the year 2005 is:

Sales in 2005 = -395541.56 + (199.87 x 2005)

= 5198 Units
(Note: the large negative Y axis intercept results from using the actual calendar years as the X axis scale.)
14

Quantitative Forecasting:
Regression: Auto Forecast by Excel.
Sales by Year, With Automatic Three Year Prediction
6000 5000 4000 SALES

Sales

3000 2000 1000 0 1990

Simple Linear Regression, Forecast Out to Year 2005 1995 2000 2005 2010

Year
15

Quantitative Forecasting:
Moving Average- Auto Plot
Sales Units Per Year With Fitted Two Year Moving Average
5000

Sales Units

4000 3000 2000 1000 0 1990 1995 Years 2000 2005

16

SALES 2 per. Mov. Avg. (SALES)

Quantitative Forecasting:
Notes on Excel Auto Plot.
Excel will plot, and automatically forecast, a data series which has a functional relationship. For example, a regression trend line.
The auto plot is driven through the Chart menu as Add Trendline. A particular forecast is specified via the dialog box. Future point data values cannot be read from the automated trendline. Non-functional relationships, such as a moving average, can be plotted, but cannot be automatically forecast.
17

Forecasting Routes
Top-Down
where international and national events affect the future behaviour of local variables.

18

Forecasting Routes

Where local events affect the future behaviour of local variables.

Bottom-Up
19

Forecasting: Summary
Sophisticated forecasting is essential for

capital budgeting decisions Quantitative forecasting uses historical data to establish relationships and trends which can be projected into the future Qualitative forecasting uses experience and judgment to establish future behaviours Forecasts can be made by either thetop down or bottom up routes.
20