You are on page 1of 60

Statistics for Managers

Using Microsoft® Excel
4th Edition
Chapter 15
Time-Series Forecasting and
Index Numbers
Statistics for Managers Using Microsoft Excel, 4e © 2004 Prentice-Hall, Inc.

Chap 15-1

Chapter Goals
After completing this chapter, you should be able
to:

Develop and implement basic forecasting models

Identify the components present in a time series

Use smoothing-based forecasting models, including
moving average and exponential smoothing

Apply trend-based forecasting models, including linear
trend and nonlinear trend

complete time-series forecasting of seasonal data

compute and interpret basic index numbers

The Importance of Forecasting

Governments forecast unemployment, interest
rates, and expected revenues from income taxes
for policy purposes

Marketing executives forecast demand, sales, and
consumer preferences for strategic planning

College administrators forecast enrollments to plan
for facilities and for faculty recruitment

Retail stores forecast demand to control inventory
levels, hire employees and provide training

Example: Year: 1999 2000 2001 2002 2003 Sales: 75. quarterly. etc.2 78.Time-Series Data    Numerical data obtained at regular time intervals The time intervals can be annually. hourly.7 80. daily.5 79.3 74.2 .

Time-Series Plot A time-series plot is a two-dimensional plot of time series data  the vertical axis measures the variable of interest  the horizontal axis corresponds to the time periods .

Time-Series Components Time Series Trend Component Seasonal Component Cyclical Component Irregular Component .

Trend Component  Long-run increase or decrease over time (overall upward or downward movement)  Data taken over a long period of time Sales nd e r t d r Up w a Time .

Trend Component (continued)   Trend can be upward or downward Trend can be linear or non-linear Sales Sales Time Downward linear trend Time Upward nonlinear trend .

Seasonal Component    Short-term regular wave-like patterns Observed within 1 year Often monthly or quarterly Sales Summer Winter Summer Spring Winter Spring Fall Time (Quarterly) Fall .

Cyclical Component    Long-term wave-like patterns Regularly occur but may vary in length Often measured peak to peak or trough to trough 1 Cycle Sales Year .

Irregular Component   Unpredictable. “residual” fluctuations Due to random variations of    Nature Accidents or unusual events “Noise” in the time series . random.

Multiplicative Time-Series Model for Annual Data   Used primarily for forecasting Observed value in time series is the product of components Yi  Ti  Ci  Ii where Ti = Trend value at year i Ci = Cyclical value at year i Ii = Irregular (random) value at year i .

Multiplicative Time-Series Model with a Seasonal Component   Used primarily for forecasting Allows consideration of seasonal variation Yi  Ti  Si  Ci  Ii where Ti = Trend value at time i Si = Seasonal value at time i Ci = Cyclical value at time i Ii = Irregular (random) value at time i .

Smoothing the Annual Time Series  Calculate moving averages to get an overall impression of the pattern of movement over time Moving Average: averages of consecutive time series values for a chosen period of length L .

. L = 7 Etc.Moving Averages     Used for smoothing A series of arithmetic means over time Result dependent upon choice of L (length of period for computing means) Examples:    For a 5 year moving average. L = 5 For a 7 year moving average.

Moving Averages (continued)  Example: Five-year moving average  First average: MA(5)   Second average: MA(5)   Y1  Y2  Y3  Y4  Y5 5 etc. Y2  Y3  Y4  Y5  Y6 5 .

Example: Annual Data
Year
1
2
3
4
5
6
7
8
9
10
11
etc…

Sales
23
40
25
27
32
48
33
37
37
50
40
etc…

Calculating Moving Averages
Average
Year

5-Year
Moving
Average

Year

Sales

1

23

3

29.4

2

40

4

34.4

3

25

5

33.0

4

27

6

35.4

5

32

7

37.4

6

48

8

41.0

7

33

9

39.4

8

37

9

37

10

50

11

40

etc…

3
29.4 

1 2  3  4  5
5
23  40  25  27  32
5

Each moving average is for a consecutive
block of 5 years

Annual vs. Moving Average

The 5-year
moving average
smoothes the
data and shows
the underlying
trend

Exponential Smoothing   A weighted moving average  Weights decline exponentially  Most recent observation weighted most Used for smoothing and short term forecasting (often one period into the future) .

larger W gives less smoothing The weight is:   Close to 0 for smoothing out unwanted cyclical and irregular components Close to 1 for forecasting .Exponential Smoothing (continued)  The weight (smoothing coefficient) is W     Subjectively chosen Range from 0 to 1 Smaller W gives more smoothing.

… where: Ei = exponentially smoothed value for period i Ei-1 = exponentially smoothed value already computed for period i .1 Yi = observed value in period i W = weight (smoothing coefficient). 3.Exponential Smoothing Model  Exponential smoothing model E1  Y1 Ei  WYi  (1  W )Ei1 For i = 2. 4. 0 < W < 1 .

12 (.8)(31.437)=31.2)(40)+(.872 (.8)(26.296 27.8)(31.840 (.8)(33.437 (.12)=26. Suppose we use weight W = .4 26.8)(23)=26. E1 = Y1 since no prior information exists Ei  WYi  (1  W )Ei1 .2)(32)+(.549 31.2)(48)+(.8)(32.2)(33)+(.2)(48)+(.437 31.840)=32.8)(26.958 etc.296)=27.697 etc.872)=33. Exponentially Smoothed Value for this period (Ei) 23 (.697 (.4 (.697)=36.4)=26.12 26.840 32.872 33. Forecast from prior period (Ei-1) -23 26.Exponential Smoothing Example  Time Period (i) 1 2 3 4 5 6 7 8 9 10 etc.8)(26.296 (.2)(50)+(.549 (.2 Sales (Yi) 23 40 25 27 32 48 33 37 37 50 etc.549)=31.2)(37)+(.2)(25)+(.2)(27)+(.8)(27.

since the trend is upward sloping and the weighting factor is only . Smoothed Sales  Fluctuations have been smoothed  NOTE: the smoothed value in this case is generally a little low.Sales vs.2 .

Forecasting Time Period i + 1 The smoothed value in the current period (i) is used as the forecast value for next period (i + 1) :  Yˆi1  Ei .

Exponential Smoothing in Excel  Use tools / data analysis / exponential smoothing  The “damping factor” is (1 .W) .

Trend-Based Forecasting  Estimate a trend line using regression analysis  Year Time Period (X) Sales (Y) 1999 2000 2001 2002 2003 2004 0 1 2 3 4 5 20 40 30 50 70 65 Use time (X) as the independent variable: Yˆ  b0  b1X .

905  9.Trend-Based Forecasting (continued)  Year Time Period (X) Sales (Y) 1999 2000 2001 2002 2003 2004 0 1 2 3 4 5 20 40 30 50 70 65 The linear trend forecasting equation is: Yˆi  21.5714 Xi .

Trend-Based Forecasting (continued)  Year Time Period (X) Sales (y) 1999 2000 2001 2002 2003 2004 2005 0 1 2 3 4 5 6 20 40 30 50 70 65 ?? Forecast for time period 6: Yˆ  21.33 .5714 (6)  79.905  9.

r2 and standard error to that of linear model to see if this is an improvement  Can try other functional forms to get best fit .Nonlinear Trend Forecasting  A nonlinear regression model can be used when the time series exhibits a nonlinear trend  Quadratic form is one type of a nonlinear model: Yi  0  1Xi  2 X  i 2 i  Compare adj.

Exponential Trend Model  Another nonlinear trend model: Xi 0 1 Yi  β β  εi Transform to linear form: log(Yi )  log(β0 )  Xi log(β1 )  log( ε i ) .

Exponential Trend Model (continued)  Exponential trend forecasting equation: log(Yˆi )  b0  b1Xi where b0 = estimate of log(β0) b1 = estimate of log(β1) Interpretation: (βˆ1  1)  100% is the estimated annual compound growth rate (in %) .

Model Selection Using Differences  Use a linear trend model if the first differences are approximately constant (Y2  Y1 )  ( Y3  Y2 )    ( Yn  Yn-1 )  Use a quadratic trend model if the second differences are approximately constant [(Y3  Y2 )  ( Y2  Y1 )]  [(Y4  Y3 )  ( Y3  Y2 )]    [(Yn  Yn-1 )  ( Yn-1  Yn-2 )] .

Model Selection Using Differences  (continued) Use an exponential trend model if the percentage differences are approximately constant (Y2  Y1 ) (Y3  Y2 ) (Yn  Yn-1 )  100%   100%     100% Y1 Y2 Yn-1 .

correlation between values 2 periods apart pth order Autoregressive models: Yi  A 0  A 1Yi-1  A 2 Yi-2     A p Yi-p  δi Random Error .correlation between consecutive values 2nd order .Autoregressive Modeling   Used for forecasting Takes advantage of autocorrelation    1st order .

has acquired a number of office units (in thousands of square feet) over the last eight years. Develop the second order Autoregressive model. Year Units 97 98 99 00 01 02 03 04 4 3 2 3 2 2 4 6 .Autoregressive Model: Example The Office Concept Corp.

8125 X Variable 2 -0.Autoregressive Model: Example Solution  Develop the 2nd order table  Use Excel to estimate a regression model Excel Output Coefficients Intercept 3.9375Yi2 .9375 Year Yi Yi-1 Yi-2 97 98 99 00 01 02 03 04 4 3 2 3 2 2 4 6 -4 3 2 3 2 2 4 --4 3 2 3 2 2 Yˆi  3.8125Yi1  0.5  0.5 X Variable 1 0.

5  0.9375Yi2 Yˆ2005  3.5  0.5  0.8125Yi1  0.625 .8125(6 )  0.Autoregressive Model Example: Forecasting Use the second-order equation to forecast number of units for 2005: Yˆi  3.8125(Y2004 )  0.9375(4 )  4.9375(Y2003 )  3.

… . Form a series of “lagged predictor” variables Yi-1 . decrease p by 1 and repeat . Yi-2 .Autoregressive Modeling Steps 1.Yi-p 3. Choose p (note that df = n – 2p – 1) 2. Use Excel to run regression model using all p variables 4. this model is selected If null hypothesis not rejected. Test significance of Ap   If null hypothesis rejected.

Selecting A Forecasting Model  Perform a residual analysis     Look for pattern or direction Measure magnitude of residual error using squared differences Measure residual error using MAD Use simplest model  Principle of parsimony .

Residual Analysis e e 0 0 Random errors T T Cyclical effects not accounted for e e 0 0 T Trend not accounted for T Seasonal effects not accounted for .

Measuring Errors   Choose the model that gives the smallest measuring errors Sum of squared errors (SSE) n SSE   (Yi  Yˆi )  Mean Absolute Deviation (MAD) n 2 MAD  i1  Sensitive to outliers   Y  Yˆ i1 i i n Not sensitive to extreme observations .

Principal of Parsimony   Suppose two or more models provide a good fit for the data Select the simplest model  Simplest model types:     Least-squares linear Least-squares quadratic 1st order autoregressive More complex types:   2nd and 3rd order autoregressive Least-squares exponential .

0 otherwise Q3 = 1 if third quarter. 0 otherwise Q2 = 1 if second quarter. 0 otherwise (Quarter 4 is the default if Q1 = Q2 = Q3 = 0) .Forecasting With Seasonal Data  Recall the classical time series model with seasonal variation: Yi  Ti  Si  Ci  Ii  Suppose the seasonality is quarterly  Define three new dummy variables for quarters: Q1 = 1 if first quarter.

3.Exponential Model with Quarterly Data Xi 0 1 Q1 Q2 Q3 Yi  β β β 2 β3 β 4 ε i βi provides the multiplier for the ith quarter relative to the 4th quarter (i = 2. 4)  Transform to linear form: log(Yi )  log(β0 )  Xilog(β1 )  Q1log(β2 )  Q 2log(β3 )  Q3log(β 4 )  log( ε i ) .

Estimating the Quarterly Model  Exponential forecasting equation: log(Yˆi )  b0  b1Xi  b 2Q1  b3Q 2  b 4Q3 where b0 10  βˆ0 b0 = estimate of log(β0). so b1 ˆ 10  β 1 b = estimate of log(β ). to fourth quarter βˆ 4 = estimated multiplier for third quarter relative to fourth quarter . so 1 Interpretation: 1 etc… ˆ  1)  100% = estimated quarterly compound growth rate (in %) (β 1 βˆ = estimated multiplier for first quarter relative to fourth quarter 2 βˆ3 = estimated multiplier for second quarter rel.

43.017X i  .082Q1  . so 10b1  βˆ1  1. so 10b2  βˆ 2  0.073Q 2  .845 10b 4  βˆ 4  1.022. so 10b0  βˆ0  2691.827 b3 = -.022Q 3 b0 = 3. so 10b3  βˆ3  0.53 b1 = .Quarterly Model Example  Suppose the forecasting equation is: log(Yˆi )  3.017.073.082.040 b2 = -. so b4 = .052 .43  .

sales in Q4 are 105.7% of average 4th quarter sales. after adjusting for the 4% quarterly growth rate .53 Unadjusted trend value for first quarter of first year βˆ1  1.Quarterly Model Example (continued) Value: Interpretation: βˆ0  2691.827 Ave.040 4.0% = estimated quarterly compound growth rate βˆ 2  0.052 Ave.845 Ave. sales in Q3 are 84. after adjusting for the 4% quarterly growth rate after adjusting for the 4% quarterly growth rate βˆ 4  1. βˆ3  0.2% of average 4th quarter sales.5% of average 4th quarter sales. sales in Q2 are 82.

Index Numbers  Index numbers allow relative comparisons over time  Index numbers are reported relative to a Base Period Index  Base period index = 100 by definition  Used for an individual item or measurement .

Simple Price Index  Simple Price Index: Pi Ii   100 Pbase where Ii = index number for year i Pi = price for year i Pbase = price for the base year .

Index Numbers: Example  Airplane ticket prices from 1995 to 2003: Index Year Price (base year = 2000) 1995 272 85.6 2000 320 100.8 2002 366 114.0 2001 348 108.0 1997 295 92.4 2003 384 120.0 I1996 P1996 288   100  (100 )  90 P2000 320 Base Year: P2000 320 I2000   100  (100 )  100 P2000 320 I2003 P2003 384   100  (100 )  120 P2000 320 .2 1999 322 100.0 1996 288 90.2 1998 311 97.

Index Numbers: Interpretation P1996 288   100  (100 )  90 P2000 320  Prices in 1996 were 90% of base year prices I2000 P2000 320   100  (100 )  100 P2000 320  Prices in 2000 were 100% of base year prices (by definition. since 2000 is the base year)  I2003 P 384  2003  100  (100 )  120 P2000 320 Prices in 2003 were 120% of base year prices I1996 .

Aggregate Price Indexes  An aggregate index is used to measure the rate of change from a base period for a group of items Aggregate Price Indexes Unweighted aggregate price index Weighted aggregate price indexes Paasche Index Laspeyres Index .

Unweighted Aggregate Price Index  Unweighted aggregate price index formula: n IU( t )  (t) P  i i1 n P i1 P i1 n (t) i (0) P  i i1  100 t = time period n = total number of items = unweighted price index at time t IU( t ) n (0) i i = item = sum of the prices for the group of items at time t = sum of the prices for the group of items in time period 0 .

8 I2004  P   P 410  100  (100)  118.Unweighted Aggregate Price Index: Example Automobile Expenses: Monthly Amounts ($): Index Year Lease payment Fuel Repair Total (2001=100) 2001 260 45 40 345 100.1 2003 305 55 45 405 117.8 345 2001 2004 Unweighted total expenses were 18.4 2004 310 50 50 410 118.0 2002 280 60 40 380 110.8% higher in 2004 than in 2001 .

Weighted Aggregate Price Indexes  Laspeyres index n I  (t) L P i1 n (t) i P i1 Q (0) i (0) i Q  Paasche index n  100 (0) i I  (t) P P i 1 n P i1 Q(i 0 ) : weights based on period 0 quantities (t) i Q (0) i (t) i Q  100 (t) i Q(i t ) : weights based on current period quantities Pi( t ) = price in time period t Pi( 0 ) = price in period 0 .

Common Price Indexes  Consumer Price Index (CPI)  Producer Price Index (PPI)  Stock Market Indexes  Dow Jones Industrial Average  S&P 500 Index  NASDAQ Index .

. business experiences. etc. and habits. changing technologies.Pitfalls in Time-Series Analysis   Assuming the mechanism that governs the time series behavior in the past will still hold in the future Using mechanical extrapolation of the trend to forecast the future without considering personal judgments.

quadratic and exponential models .Chapter Summary    Discussed the importance of forecasting Addressed component factors of the time-series model Performed smoothing of data series    Moving averages Exponential smoothing Described least square trend fitting and forecasting  Linear.

Chapter Summary (continued)     Addressed autoregressive models Described procedure for choosing appropriate models Addressed time series forecasting of monthly or quarterly data (use of dummy variables) Discussed pitfalls concerning time-series analysis .