Slide Chap 5
-
Upload
izwan-yusof -
Category
Documents
-
view
220 -
download
0
Transcript of Slide Chap 5
-
8/12/2019 Slide Chap 5
1/35
Slide 1
Chapter 5
DR.NORAZNIN ABU BAKAR
-
8/12/2019 Slide Chap 5
2/35
Slide 2
Business and Economic Forecasting
Demand Forecasting is a criticalmanagerial activity which comes in two
forms:
Quantitative Forecasting+2.1047%Gives the precise amountor percentage
Qualitative Forecasting Gives the expected direction
Up, down, or about the same
-
8/12/2019 Slide Chap 5
3/35
Slide 3
What Went Wrong WithSUVs at Ford Motor Co?
Chrysler introduced the Minivan
in the 1980s
Ford expanded its capacity to produce theExplorer, its popular SUV
Explorers price was raised substantially in 1995
at same time competitors expanded theirofferings of SUVs.
Must consider response of rivals in pricing
decisions
-
8/12/2019 Slide Chap 5
4/35
Slide 4
Significance of Forecasting
Both public and private enterprises operate under
conditions of uncertainty.
Management wishes to limit this uncertainty by
predicting changes in cost, price, sales, and
interest rates.
Accurate forecasting can help develop strategies
to promote profitable trends and to avoid
unprofitable ones. A forecast is a prediction concerning the future.
Good forecasting will reduce, but not eliminate, the
uncertainty that all managers feel.
-
8/12/2019 Slide Chap 5
5/35
Slide 5
H ierarchy of Forecasts The selection of forecasting techniques depends in
part on the level of economic aggregation involved. The hierarchy of forecasting is:
National conomy(GDP, interest rates,inf lation, etc.)
sectors of the economy(durable goods)
industry forecasts(all automobile manufacturers)>firm forecasts(Ford Motor Company)
Product forecasts(The Ford Focus)
-
8/12/2019 Slide Chap 5
6/35
Slide 6
Forecasting Criter iaThe choice of a particular forecasting method
depends on several criteria:
1.costsof the forecasting method compared withits gains
2.complexityof the relationships amongvariables
3.timeper iod involved4.accuracyneeded in forecast5.lead timebetween receiving information and
the decision to be made
-
8/12/2019 Slide Chap 5
7/35Slide 7
Accuracy of Forecasting
The accuracyof a forecasting model is measuredby how close the actual variable, Y, ends up to the
forecasting variable, Y.
Forecast erroris the difference. (Y - Y)
Models differ in accuracy, which is often based on
the square root of the average squared forecast
error over a series of N forecasts and actual figures
Called a root mean square error, RMSE.
RMSE = { (Y - Y)2
/ N }
^
^
^
-
8/12/2019 Slide Chap 5
8/35Slide 8
Quantitative Forecasting
Deterministic TimeSeries Looks For Patterns
Ordered by Time
No Underlying Structure
Econometric Models
Explains relationships Supply & Demand
Regression Models
Like technical
security analysis
Like fundamentalsecurity analysis
-
8/12/2019 Slide Chap 5
9/35Slide 9
Time Series
Examine Patterns in the Past
TIMETo
X
X
X
Dependent Variable
Secular TrendCyclical Variation
Forecasted Amounts
The data may offer secular trends, cyclical variations, seasonal
variations, and random fluctuations.
-
8/12/2019 Slide Chap 5
10/35Slide 10
Elementary Time Series Models
for Economic Forecasting
1. Naive Forecast
Yt+1
= Yt
Method best whenthere is no trend, onlyrandom error
Graphs of sales over
time with and withouttrends
When trending down,the Nave predicts toohigh
NO Trend
Trend
^
time
time
-
8/12/2019 Slide Chap 5
11/35Slide 11
2. Nave forecast with adjustments
for secular trendsYt+1 = Yt + (Yt - Yt-1 )
This equation begins with last periods
forecast, Yt. Plus an adjustment for the change in the
amount between periods Ytand Yt-1.
When the forecast is trending up, thisadjustment works better than the purenave forecast method #1.
^
-
8/12/2019 Slide Chap 5
12/35Slide 12
3. Linear & 4. Constant rate of growth
Used when trend has a
constant AMOUNT ofchange
Yt = a + bT, where
Yt are the actual
observations and
Tis a numerical time
variable
Used when trend is a
constantPERCENTAGE rate
Log Yt = a + bT,
whereb is the
continuouslycompounded growth
rate
Linear Trend Growth Uses a Semi-log Regression
-
8/12/2019 Slide Chap 5
13/35Slide 13
More on Constant Rate of Growth Model
a proof
Suppose: Yt= Y0( 1 + G)twhere g is the
annual growth rate
Take the natural log of both sides: Ln Yt= Ln Y0 + t Ln (1 + G)
but Ln ( 1 + G ) g, the equivalentcontinuously compounded growth rate
SO: Ln Yt= Ln Y0 + t g
Ln Yt= a + b twherebis the growth rate
^
^
-
8/12/2019 Slide Chap 5
14/35Slide 14
Numerical Examples: 6 observations
MTB > Print c1-c3.Sales Time Ln-sales
100.0 1 4.60517109.8 2 4.69866
121.6 3 4.80074
133.7 4 4.89560
146.2 5 4.98498
164.3 6 5.10169
Using this sales
data, estimate
sales in period 7
using a linear anda semi-log
functional
form
-
8/12/2019 Slide Chap 5
15/35Slide 15
The regression equation is
Sales = 85.0 + 12.7 Time
Predictor Coef Stdev t-ratio p
Constant 84.987 2.417 35.16 0.000
Time 12.6514 0.6207 20.38 0.000
s = 2.596 R-sq = 99.0% R-sq(adj) = 98.8%
The regression equation is
Ln-sales = 4.50 + 0.0982 Time
Predictor Coef Stdev t-ratio pConstant 4.50416 0.00642 701.35 0.000
Time 0.098183 0.001649 59.54 0.000
s = 0.006899 R-sq = 99.9% R-sq(adj) = 99.9%
-
8/12/2019 Slide Chap 5
16/35
Slide 16
Forecasted Sales @ Time = 7 Linear Model Sales = 85.0 + 12.7 Time
Sales = 85.0 + 12.7( 7)
Sales = 173.9
Semi-Log Model Ln-sales = 4.50 + 0.0982
Time
Ln-sales = 4.50 +
0.0982( 7 )
Ln-sales = 5.1874
To anti-log:
e5.1874 = 179.0
linear
-
8/12/2019 Slide Chap 5
17/35
Slide 17
Sales Time Ln-sales
100.0 1 4.60517
109.8 2 4.69866
121.6 3 4.80074133.7 4 4.89560
146.2 5 4.98498
164.3 6 5.10169
179.0 7 semi-log
173.9 7 linear
Which prediction
do you prefer?
Semi-logis
exponential
7
-
8/12/2019 Slide Chap 5
18/35
Slide 18
5. Declining Rate of Growth Trend
A number of marketing penetration modelsuse a slight modification of the constant rate
of growth model In this form, the inverse of time is used
Ln Yt = b1b2( 1/t ) This form is good for patterns
like the one to the right
It grows, but at continuouslya declining rate time
Y
-
8/12/2019 Slide Chap 5
19/35
Slide 19
6. Seasonal Adjustments: The Ratio to Trend Method
Take ratios of the actual to
the forecasted values for past
years.
Find the average ratio. Thisis the seasonal adjustment
Adjust by this percentage by
multiply your forecast by the
seasonal adjustment
If average ratio is 1.02, adjust
forecast upward 2%
12 quarters of data
I II III IV I II III IV I II III IV
Quarters designated with roman numerals.
-
8/12/2019 Slide Chap 5
20/35
Slide 20
Let D = 1, if 4th quarter and 0otherwise Run a new regression:
Yt= a + bT + cD the c coefficient gives the amount of the adjustment for the
fourth quarter. It is an I ntercept Shi f ter.
With 4 quarters, there can be as many as three dummy variables;
with 12 months, there can be as many as 11 dummy variables
EXAMPLE: Sales = 300 + 10T + 18D12 Observationsfrom the first quarter of 2002 to 2004-IV.
Forecast all of 2005.
Sales(2005-I) = 430; Sales(2005-II) = 440; Sales(2005-III) = 450;
Sales(2005-IV) = 478
7.Seasonal Adjustments: Dummy Variables
-
8/12/2019 Slide Chap 5
21/35
Slide 21
Soothing Techniques
8. Moving Averages
A smoothing forecast
method for data thatjumps around
Best when there is no
trend
3-Period Moving Ave.
Yt+1 = [Yt+ Yt-1+ Yt-2]/3
*
*
*
*
*Forecast
Line
TIME
Dependent Variable
-
8/12/2019 Slide Chap 5
22/35
Slide 22
Smoothing Techniques
9. First-Order Exponential Smoothing
A hybrid of the Naive
and Moving Average
methods
Yt+1 = wYt+(1-w)Yt A weighted average of
past actual and pastforecast.
Each forecast is a function
of all past observations
Can show that forecast is
based on geometricallydeclining weights.
Yt+1 = w.Yt+(1-w)wYt-1+(1-w)2wYt-1 + Find lowest RMSE to pick
the best alpha.
^ ^
^
^
-
8/12/2019 Slide Chap 5
23/35
Slide 23
First-Order Exponential
Smoothing Example for w= .50Actual Sales Forecast
100 100 initial seed required
120 .5(100) + .5(100) = 100
115
130
?
1
2
3
4
5
-
8/12/2019 Slide Chap 5
24/35
Slide 24
First-Order Exponential
Smoothing Example for w= .50Actual Sales Forecast
100 100 initial seed required
120 .5(100) + .5(100) = 100
115 .5(120) + .5(100) = 110
130
?
1
2
3
4
5
-
8/12/2019 Slide Chap 5
25/35
Slide 25
First-Order Exponential
Smoothing Example for w= .50Actual Sales Forecast
100 100 initial seed required
120 .5(100) + .5(100) = 100
115 .5(120) + .5(100) = 110
130 .5(115) + .5(110) = 112.50
? .5(130) + .5(112.50) = 121.25
Period 5
ForecastMSE = {(120-100)2+ (110-115)2+ (130-112.5)2}/3 = 243.75
RMSE = 243.75 = 15.61
1
2
3
4
5
-
8/12/2019 Slide Chap 5
26/35
Slide 26
Direction of sales can be indicated by other variables.
TIME
Index of Capital Goods
peakPEAK Motor Control Sales
4 Months
Example: Index of Capital Goods is a leading indicator
There are also lagging indicators and coincident indicators
Qualitative Forecasting
10. Barometric Techniques
-
8/12/2019 Slide Chap 5
27/35
Slide 27
LEADING INDICATORS*
M2 money supply (-12.4) S&P 500 stock prices (-11.1)
Building permits (-14.4)
Initial unemployment claims
(-12.9) Contracts and orders for
plant and equipment (-7.4)
COINCIDENT
INDICATORSNonagricultural payrolls
(+.8)
Index of industrial
production (-1.1)
Personal income less
transfer payment (-.4)
LAGGING INDICATORS Prime rate (+2.0)
Change in labor cost per unit
of output (+6.4)*Survey of Current Business, 1994See pages 206-207 in textbook
Time given in month s from change
-
8/12/2019 Slide Chap 5
28/35
Slide 28
Handling Multiple Indicators
Diffusion Index: Wall Street With LouisRuykeyserhas eleven analysts. If 4 are
negative about stocks and 7 are positive, the
Diffusion Index is 7/11, or 63.3%.above 50% is a positive diffusion indexomposite Index: One indicator rises
4% and another rises 6%. Therefore, theComposite Index is a 5%increase.
used for quantitative forecasting
-
8/12/2019 Slide Chap 5
29/35
Slide 29
Qualitative Forecasting
11.Surveys and Opinion Polling Techniques
Sample bias-- telephone, magazine
Biased questions--
advocacy surveys
Ambiguous questions
Respondents may lie on
questionnaires
New Products have no
historical data -- Surveys
can assess interest in new
ideas.
Survey Research Center
of U. of Mich. does repeat
surveys of households on
Big Ticket items (Autos)
Common Survey Problems
-
8/12/2019 Slide Chap 5
30/35
Slide 30
Qualitative Forecasting
12. Expert OpinionThe average forecast from several experts
is a Consensus Forecast.
Mean Median
Mode
Truncated Mean Proportion positive or negative
-
8/12/2019 Slide Chap 5
31/35
Slide 31
EXAMPLES:
IBES, First Call, and Zacks Investment --
earnings forecasts of stock analysts of companies
Conference Board macroeconomic predictions
Livingston Surveys--macroeconomic forecasts
of 50-60 economists
Individual economists tend to be lessaccurate over time than the consensus
forecast.
-
8/12/2019 Slide Chap 5
32/35
Slide 32
13. Econometric Models Specify the variables in the model
Estimate the parameters
single equation or perhaps several stage methods
Qd = a + bP + cI + dPs+ ePc But forecasts require estimates for future prices,
future income, etc.
Often combine econometric models with time seriesestimates of the independent variable.
Garbage in Garbage out
-
8/12/2019 Slide Chap 5
33/35
Slide 33
example
Qd = 400 - .5P + 2Y + .2Ps
anticipate pricing the good at P = $20
Income (Y) is growing over time, the estimate
is: Ln Yt= 2.4 + .03T, and next period is T
= 17. Y = e2.910= 18.357
The prices of substitutes are likely to be
P = $18. Find Qdby substituting in predictions for P, Y,
and Ps
Hence Qd = 430.31
-
8/12/2019 Slide Chap 5
34/35
Slide 34
14. Stochastic Time Series A little more advanced methods incorporate into time
series the fact that economic data tends to drift
yt= a+ byt-1+ et In this series, if ais zero and bis 1, this is essentially
the nave model. When ais zero, the pattern is called arandom walk.
When ais positive, the data drift. The Durbin-Watson
statistic will generally show the presence ofautocorrelation, or AR(1), integrated of order one.
One solution to variables that drift, is to use first
differences.
-
8/12/2019 Slide Chap 5
35/35
Cointegrated Time Series
Some econometric work includes several stochastic
variable, each which exhibits random walk with drift
Suppose price data (P) has positive drift
Suppose GDP data (Y) has positive drift
Suppose the sales is a function of P & Y
Salest= a + bPt+ cYt
It is likely that P and Y are cointegrated in that they
exhibit comovement with one another. They are notindependent.
The simplest solution is to change the form into first
differences as in: DSalest= a + bPt+ cYt