Forecasting in Business and

Finance
Forecasting
Forecasting is a process of predicting or estimating future events based on
past and present data and most commonly by analysis of trends. A forecast,
must have logic to it. It must be defendable. It offer logic and so is
differentiated from lucky guess. After all, even a broken watch is right two
times a day.

For example, we could forecast average annual company income based on

data from the past 10 years. We could also use forecasting to predict how
many customer calls Phil, our salesman, is likely to receive in the next day. Or
how many product demos he'll lead over the next week. The data from
previous years is already available in our CRM, and it can help us accurately
predict and anticipate future sales and marketing events where Phil may be
needed.

2
More Example
 Pantai Hospital has to forecast how many patients will be
admitted during the Covid-19 pandemic to develop
departmental schedules for the period.

 Air Asia has to forecast how many aircrafts should be put

on rest in order to minimize its losses during the
pandemic era.

 Proton has to forecast its sales for the new car model X50
so that dealer stocks are of reasonable size for every
model.

3
4
Time Series Data and Modelling

5
6
Forecasting in General

7
 Forecasting Issues
How well can one forecast the future?

o The answer depends on the stability of the pattern of the profit,

sales and expenses for the product and services.

o The underlying pattern may be hard to find, but not impossible.

o When a pattern is found, the question remains, how long will it

persist? When will it change? Forecasters are willing to accept
the challenge.

8
 Theories Behind Business Forecasting at
Microeconomic (Firm) and Macroeconomic Level

1. Product Life Cycle Stages

 All products and services go through the following four
stages:
o Introduction to the market
o Growth of volume and share
o Maturation, where maturity is the phase of relative equilibrium
o Decline occurs, because of deteriorating sales; decline leads to
restaging or withdrawal

 The production system’s capabilities need to be adjusted with

changes or transitions between stages.
9
Product Life Cycle Stages (continued)
Life cycle stages provide a classification for understanding the
nature of evolving demand trends that will occur over time.

10
 Product Life Cycle Stages (continued)
 During the introductory phase, demand is led by the desire to “fill the pipeline.” This
means getting product into the stores or warehouses—wherever it must be to supply
the customers.

 When growth starts to occur, there is a trend line of increasing sales. The trick is to
estimate how fast demand will increase over time and for how long a period growth
will continue.

 When the new product or service stops growing, it is considered mature. This means
—its volume is stabilized at the saturation level for that brand. The competitors have
divided the market, and only extraordinary events, such as a strike at a competitor’s
plant, are able to shift shares and volumes.

 Finally, the product begins to lose share, volume drops, and, depending on the
strategy, the product is either restaged or terminated.
11
2. Macroeconomy Business Cycles
• Business Cycles
– Recurring cycles of recession and recovery
• Peak
– Transition from end of expansion to start of contraction
• Trough
– Transition point between recession and recovery
Quantitative Forecasting

13
 Forecasting with Data
 Mathematical equations are used for forecasting.
 It is then tested with data in a statistical model, often in time
series form to verify its forecasting ability.
 Statistical model do not make forecasts “the truth.”
 Good forecasting can be done without statistical model.
 With or without statistical model, no forecast is ever
guaranteed.
• The cardinal premise underlying all time series models is that
the historical pattern of the dependent variable can be used as
the basis for developing forecasts.
• In these models, historical data for the forecast variable are
analyzed in an attempt to discern any underlying pattern(s).
14
 Forecasting and Modeling
• Forecasting can be in univariate or
multivariate (multiple dependent variables)
modeling
• Modeling can be in economic-driven,
mathematic or even statistical based. The
formers are not always required to develop
accurate forecasts. Statistical modeling in
univariate basis is usually sufficient to provide
a good forecast.
• The principle of parsimony suggests that the
 Time Series and Extrapolation
 A time series is a stream of data (e.g., demand).

 Data are recorded at different time periods – monthly,

weekly, daily, etc.

 Forecasters predict (by extrapolation) the value(s) at a

future time.

 The pattern of the series is considered to be time-dependent.

External causes are not brought into the picture.

16
 Time Series Forecasting
• Time series or autoregressive forecasting
models will be most useful when economic
conditions can be expected to remain
relatively stable. So it is more suitable for
short run purposes.
• Reliance of time series models on analysis and
extrapolation of historical patterns carries
several important implications with respect to
technique selection:
1. Time series are best when applied to short-
term forecasts.
 Time Series (Random Variations)

o There are no specific assignable causes for

random variations.
o Values are a result of the economic environment
and the market place.

18
 Time Series and Extrapolation continued

The data in time series may consist of several different kinds

of variations.

Important among them are:

o random variations
o increasing or decreasing trend
o seasonal variations

19
Time Series Forecasting

20
 Naïve Model (Random Walk)
• Uses recent past as the best indicator of the
future. ˆ
Yt 1 Yt

e Y  Yˆ
• The error associated
t t witht this model is
computed as:
Example for the Naïve Model
Week Sales (in Forecast
$1,000)
1 9 -
2 8 9
3 9 8
5 12 9
6 9 12
7 12 9
8 11 12
9 ? 11
 Averaging Models

• The basic premise of these models is that a

weighted-average of past observations can
be used to smooth the fluctuations in the
data in the short term.
 Simple Average Model
• Similar to the naïve model, this model uses
part of the historical data to make a forecast.
n
 Yt
ˆ
Yt 1  t 1
n
 Moving Average Model
• Recent observations play an important role in
the forecast.
• As new observations become available, a new
average is computed.
• The choice of using a smaller or larger
number of observations has implications for
the forecast.
 Moving Average
A n-month moving average is the sum of the observed
values during the past n months divided by n.
Moving Average Method n =3

Month Sales Forecast Formula

1 100
2 80 MA(3)
3 90
4 110 90.00 =(100+80+90)/3
5 100 93.33 =(80+90+110)/3
6 110 100.00 =(90+110+100)/3
7 95 106.67 =(110+100+110)/3
8 115 101.67 =(100+110+95)/3
9 120 106.67 =(110+95+115)/3
10 90 110.00 =(95+115+120)/3
11 105 108.33 =(115+120+90)/3
12 110 105.00 =(120+90+105)/3 26
 Weighted Moving Average
 The weighted moving average (WMA) makes
forecasts more responsive to the most recent actual
occurrences (e.g., demand).

 The most recent n periods are used in forecasting.

 Each period is assigned a weight between 0 and 1.

 The total of all weights adds up to one (1).

27
Weighted Moving Average
(using monthly demands)

Example:
Forecast (4) = 0.2*(Demand 1) + 0.3*(Demand 2) + 0.5*(Demand 3)
Weight
Weighted Moving Average Method n=3
s
0.2
0.3
Month Sales Forecast Calculation 0.5
1 100
2 80 WMA(3)
3 90
4 110 89.00 = 0.2*100 + 0.3*80 + 0.5*90
5 100 98.00 = 0.2*80 + 0.3*90 + 0.5*110
6 110 101.00 = 0.2*90 + 0.3*110 + 0.5*100
7 95 107.00 = 0.2*110 + 0.3*100 + 0.5*110
8 115 100.50 = 0.2*100 + 0.3*110 + 0.5*95
9 120 108.00 = 0.2*110 + 0.3*95 + 0.5*115
10 90 113.50 = 0.2*95 + 0.3*115 + 0.5*120
11 105 104.00 = 0.2*115 + 0.3*120 + 0.5*90
12 110 103.50 = 0.2*120 + 0.3*90 + 0.5*105 28
 Exponential Smoothing Model

• The model relies on the assumption that the

data are stationary.

• Most recent observations play a more

important role than the distant past.
2 3
Ŷt 1 Yt   (1   )Yt  1   (1   ) Yt  2   (1   ) Yt  3
 Exponential Smoothing Model
• The model depends on three pieces of data:
– Most recent actual
– Most recent forecast
– Smoothing constant.

• The value of alpha assigned as a smoothing

constant is critical to the forecast.
• The best alpha should be chosen on the basis
of minimal sum of error squared.
 Exponential Smoothing Model
• Several approaches are followed in selecting
the smoothing constant.
– If a great amount of smoothing is desired, a small
alpha should be chosen.
– The choice of alpha is affected by the
characteristics of the time series. If sharp ups and
downs are noticed in the data, the best
smoothing constant is 0.1. That is alpha chosen
should equal 0.1.
– If the data show that the past is very different
from the present, then alpha of 0.9 is
appropriate.
 Exponential Smoothing
 The Exponential Smoothing (ES) method forecasts the
demand for a given period t by combining the forecast of the
previous period (t-1) and the actual demand of the previous
period (t-1).
 The actual demand for the previous period is given a weight of
α and the forecast of the prior period is given a weight of (1 -
α).

 α is a smoothing constant whose value lies between 0 and 1 (0

≤ α ≤ 1).

32
 Exponential Smoothing (continued)
The equation for the forecast for period t is:
Forecast (t)
= α*Actual Demand (t-1) + (1- α )*Forecast (t-1).
= α*Actual Demand (t-1) + [1*Forecast (t-1)]-[α *Forecast (t-1)].
=Forecast (t-1) + α*[Actual Demand (t-1) -Forecast (t-1)]

The equation can also be written as:

Forecast (t) = Forecast (t-1) + α*{Actual Demand (t-1)–Forecast (t-1)}

33
 Exponential Smoothing (continued)
Example: F(3) = F(2) + α*{(A(2) – F(2)} = 100 + 0.2*(80 – 100) = 96.

Exponential Smoothing Method (sales are actual sales indicated by A in equation)

alpha = 0.2
Month Sales Forecast Comment and Calculation

Forecast for period 1 should be available before

1 100 100 starting the calculations. If it is not given then set
it equal to the sales of period 1.

2 80 100.00 =(100 + 0.2(100 -100))

3 90 96.00 =(100 + 0.2(80 -100))
4 110 94.80 =(96 + 0.2(90 -96))
5 100 97.84 =(94.8 + 0.2(110 -94.8))
6 110 98.27 =(97.84 + 0.2(100 -97.84))
7 95 100.62 =(98.27 + 0.2(110 -98.27))
8 115 99.50 =(100.62 + 0.2(95 -100.62))
9 120 102.60 =(99.5 + 0.2(115 -99.5))
10 90 106.08 =(102.6 + 0.2(120 -102.6))
11 105 102.86 =(106.08 + 0.2(90 -106.08))
12 110 103.29 =(102.86 + 0.2(105 -102.86))

34
 Time Series (Random Variations and Increasing Trend)

There is a constant rate of change (increasing values) as

time goes by.

35
 Time Series (Random Variations and
Decreasing Trend)
There is a constant rate of change (decreasing values) as
time goes by.

36
 Time Series Analysis – Trend Line
 If the time series exhibits an increasing or decreasing trend
then a trend analysis is more appropriate.
 A trend line defines the relationship between demand forecast
and the time period by the following equation.
Y = a + bX, where, Y is the demand forecast and X is
the time period.
 X is the independent variable and Y is the dependent variable
since the demand depends on the time period.

37
 Time Series Analysis – Trend Line
(continued)

 In the equation, Y = a + bX, a is the intercept on the

Y-axis. a gives the value of demand (variable Y) when
X = 0.
 The slope of the line is b which gives the change in
the value of demand (variable Y) for a unit change in
the value of X.
 The “Intercept” and “Slope” functions in Excel are
used to calculate a and b respectively.

38
 Time Series Analysis – Trend Line
(continued)

Example: Consider the demand data given in the table below.

Y= 2.73 + 8.65 X
oThe Excel functions give b = 8.65 and a = 2.73.
oUse them in equation, Y = a + bX, to make a forecast.
oFor example, for period 11 (X = 11),
Forecast = 2.73 + 11*8.65 = 97.87.
oSimilarly, for period 12,
Forecast = 2.73 + 12*8.65 = 106.52.

39
 Time Series Analysis – Trend Line
(continued)

The forecasts (values on straight line) and the actual

demand (values on zigzag line) have been plotted in
the following figure.

40
 Time Series Analysis – Trend Line
(continued)

 For any given time period, the difference between the

forecast (values on straight line) and the actual
demand (values on zigzag line) gives the error in that
period.

 The trend analysis method minimizes the sum of the

squares of these errors in calculating the values of a
and b.

41
 Time Series (Random Variations and
Seasonal Variations)
o Seasonal (cyclical) variations may also be present.
o Examples: demand for resort hotels & home heating
oil.

42
 Seasonal Forecast Step 1

Quarterly demand for last four years is given in the table below.
We use a 5-step process to forecast.
Step 1: Find average quarterly demand for each quarter.

Demand

Quarter Year 1 Year 2 Year 3 Year 4

Fall 2530 2690 2790 2860

Winter 2300 2420 2410 2600

Spring 1900 2000 2105 2175

Summer 1510 1775 1875 1945

Average 2060 2221 2295 2395

=(1510 + 1900 + 2300 + =(1775 + 2000 + 2420 =(1875 + 2105 + 2410 =(1945 + 2175 + 2600 +
Formula
2530)/4) + 2690)/4) + 2790)/4) 2860)/4)

43
 Seasonal Forecast Step 2

Step 2: Compute Seasonal Index (SI) for each quarter

for each year.

44
 Seasonal Forecast Step 3

Step 3: Calculate the average SI for each

quarter.

45
 Seasonal Forecast Step 4

Step 4: Calculate the average quarterly demand for next year.

First, the yearly demand has to be estimated or calculated for next year using
one of the forecasting techniques.

Suppose the estimated demand is 2,800.

Therefore, the average quarterly demand = 2,800/4 = 700. The calculations are
shown below.

46
 Seasonal Forecast Step 5

Step 5: Forecast demand for the four quarters of next year.

Multiply the average demand by the SI for each quarter.

For example, forecast for Spring quarter = 638 = 700*0.912.

47
Exponential Smoothing With Seasonal Effect and Time Trend

48
 Forecast with Regression Factor
 Regression is the more advance structural forecast with more explanatory
insight from the exogenous explanatory variables (time series is purely
endogenous model)

 Regression establishes a relationship between two sets (at least) of numbers that
are time series.

 For example, when a series of Y numbers (such as the monthly sales of cameras
over a period of years) is causally connected with the series of X numbers (the
monthly advertising budget), then it is beneficial to establish a relationship
between X and Y in order to forecast Y.

 In regression analysis X is the independent variable and Y is the dependent

variable.
49
 Regression Analysis
 The regression analysis gives the relationship between X and Y
by the following equation.
Y = a + bX,
where, a is the intercept on the Y-axis
(value of the variable Y when X = 0); and b is the slope of the line which
gives the change in the value of variable Y for a unit change in the value of
X.
 The “Intercept” function in Excel calculates a and the “Slope”
function in Excel is used to find the value of b.

50
 Regression Analysis (continued)
Example: Use the data given in the following table for ten-pairs of X and Y.
o The Excel functions give b = 50.23 and a = 62.44.
o Use them in equation, Y = a + bX, to forecast.
o Y = 62.44 + 50.23*15
o Suppose X = 15, then
Forecast Y= 62.44 + 50.23*15 = 815.84.
Observation 1 2 3 4 5 6 7 8 9 10
Number
Independent 10 12 11 9 10 12 10 13 14 12
Variable (x)
Dependent 400 600 700 500 800 700 500 700 800 600
Variable (y)

51
 Regression Analysis continued

The forecasts (values on straight line) and the actual

demand data points have been plotted in the following
figure.

52
 Regression Analysis (continued)
 For any given time period, the difference between the forecast
values and the actual demand gives the error in that period.

 The regression analysis minimizes the sum of the squares of

these errors in calculating the values of a and b.

 An assumption that is generally made in regression analysis is

that the relationship between the correlate pairs is linear.

 However, if nonlinear relations are hypothesized, there are

strong, but more complex methods for doing nonlinear
regression analyses.

53
 Coefficient of Determination (continued)
The coefficient of determination (r2), where r is the value of the coefficient of
correlation, is a measure of the variability that is accounted for by the
regression line for the dependent variable.

The coefficient of determination always falls between 0 and 1.

 For example, if r = 0.8, the coefficient of determination is r2 = 0.64 meaning

that 64% of the variation in Y is due to variation in X.

 The remaining 36% variation in the value of Y is due to other variables.

 If the coefficient of determination is low, multiple regression analysis may

be used to account for all variables affecting the independent variable Y.

54
 Performance Forecast
Business performance model explains the drivers for the business performance:
Y = a + b 1 X1 + b 2 X2 + b 3 X3 + b 4 X4 + b 5 X5

Performance = a + b1SIZE + b2CAPEX + b3GROWTH + b4RISK + b5CG

Performance:
ROA, ROE, TOBINQ, STOCK RETURNS, etc…
Explanatory Forces:
Company size such as market capitalization, or total assets.
CAPEX is Capital expenditure proxy for future investment
GROWHT proxy for potential growth from sales growth
RISK can be refers to many dimension such as leverage, stock price volatility, etc.
CG is corporate governance representing effectiveness internal management and
control 55
 Forecast Error
The forecasting errors are computed as,
Deviation/Error (t) = {Demand (t) – Forecast (t)}^2.

Underestimate:
Demand is greater than the forecast. Error term is positive.

Overestimate:
Demand is smaller than the forecast. Error term is negative.

56
 Error Analysis (continued)
The most commonly used method to measure errors is Mean
Absolute Deviation (MAD).

To calculate MAD, take the sum of the absolute measures of the

errors and divide that sum by the number of observations. MAD
treats all errors linearly.

Another way of measuring forecast error is Mean Square Error

(MSE), which penalizes a forecast much more for extreme
deviations than it does for small ones.

57
 Example for the Naïve Model
Week Sales (in Random Error Absolute Squared
$1,000) Walk Error Error
Forecast

1 9
2 8 9 −1 1 1
3 9 8 1 1 1
5 12 9 3 3 9
6 9 12 −3 3 9
7 12 9 3 3 9
8 11 12 −1 1 1
Sum 2 12 30
Mean 0.33 2 5
 Error Analysis (continued)
Example: Consider the demand and forecast given for 10
periods in the table below.

The sum of the absolute errors for 10 periods is 171.

Therefore, MAD = 171/10 = 17.10.
Absolute
Period Demand Forecast Error
Error
1 212 206.0 6.0 6.0
2 224 207.0 17.0 17.0
3 220 210.0 10.0 10.0
4 211 212.0 -1.0 1.0
5 198 205.0 -7.0 7.0
6 236 209.0 27.0 27.0
7 219 224.0 -5.0 5.0
8 296 238.0 58.0 58.0
9 280 249.0 31.0 31.0
10 252 261.0 -9.0 9.0
Total 127.0 171.0

59
 Analyzing Forecast Error
 To select a forecasting method, say exponential smoothing,
calculate the values of MAD/MSE choosing different values
of α. The value of α that minimizes MAD/MSE will be
selected.

 Similarly, for moving average, find MAD/MSE for different

values of n. The n that minimizes MAD/MSE will be selected.

 A similar procedure can be used with the weighted moving

average method to find the best combination of weights.

60
 Analyzing Forecast Error
• Whenever a manager evaluates alternative
forecasting techniques in terms of their
accuracy, it is necessary to go beyond the
computation of error.
• Managers are generally concerned with two
forms of accuracy:
– Accuracy of the technique in predicting the
underlying patterns or relationship of past data.
– Accuracy of the changes in the pattern. That is,
how fast forecasting procedure can respond to
that basic change. (We will discuss this in later
chapters.)
62
Qualitative Forecasting

63
 The Delphi Method

Delphi is a forecasting method that relies on expert estimation of future

events.

The experts submit their opinions to a single individual (leader of the group)
who maintains anonymity of responses.

The leader combines the opinions into a report which is disseminated to all
participants. We hope the report is fair and balanced.

The participants are asked whether they wish to reevaluate and alter their
previous opinions in the face of the body of opinion of their colleagues.

64
 The Delphi Method
Gradually, the group is supposed to move toward consensus. If it does not, at
the least, a set of different possibilities can be presented to management.

The Delphi method is meant to put all participants on an equal footing with
respect to getting their ideas heard.

There is no evidence that the Delphi method provides forecasts (and/or

predictions) with smaller errors than other techniques.

It is apparent that managers gain greater perspective about forces that should
be considered when they are contemplating possible outcomes. That is a
positive benefit of Delphi.

65
 Business Cycles: Leading Indicators
• Economic series that tend to rise or fall in advance of rest of the
economy
•Average weekly hours of production workers
•Initial claims for unemployment insurance
•Manufacturers’ new orders
•Institute of Supply Management Index of New Orders
•New orders for nondefense capital goods
•New private housing units authorized by local building permits
•Yield curve: spread between 10-year T-bond yield and federal funds rate
•Stock prices, 500 common stocks
•Money supply (M2) growth rate
•Index of consumer expectations for business conditions
 Business Cycles: Coincident Indicators
• Economic series that tend to rise or fall in tandem with the rest of
the economy
•Employees on nonagricultural payrolls
•Personal income less transfer payments
•Industrial production
•Manufacturing and trade sales
 Business Cycles: Lagging Indicators
• Economic series that tend to rise or fall behind the rest of the
economy
•Average duration of unemployment
•Ratio of trade inventories to sales
•Change in index of labor cost per unit of output
•Average prime rate charged by banks
•Commercial and industrial loans outstanding
•Ratio of consumer installment credit outstanding to personal income
•Change in consumer price index for services
 Tutorial
The table below shows the movement of the price of a commodity
over 12 months.
Month 1 2 3 4 5 6 7 8 9 10 11 12
Price 25 30 32 33 32 31 30 29 28 28 29 31
 Calculate a 6 month moving average for each month. What is
the forecast for month 13?
 Apply exponential smoothing with smoothing constants of 0.7
and 0.8 to derive forecasts for month 13.
 Which of the two forecasts based on exponential smoothing for
month 13 do you prefer and why?

69