Deck 17: Time Series Forecasting and Index Numbers

Question 1

Removing the seasonal effect by dividing the actual time series observation by the estimated seasonal factor associated with the time series observation is called deseasonalization.

Answer 1

True

Question 2

Exponential smoothing is a forecasting method that applies equal weights to the time series observations.

Answer 2

False

Question 3

The smoothing constant is a number that determines how much weight is attached to each observation.

Answer 3

True

Question 4

Holt-Winters double exponential smoothing would be an appropriate method to use to forecast a time series that exhibits a linear trend with no seasonal or cyclical patterns.

Question 5

A Paasche index more accurately provides a year-to-year comparison of the annual cost of selected products in the market basket than a Laspeyres index.

Question 6

The forecaster who uses MSD (mean squared deviations) to measure the effectiveness of forecasting methods would prefer method 1, which results in several smaller forecast errors, to method 2, which results in one large forecast error equal to the sum of the absolute values of several small forecast errors given by method 1.

Question 7

A positive autocorrelation implies that negative error terms will be followed by negative error terms.

Question 8

Cyclical variation exists when the magnitude of the seasonal swing does not depend on the level of a time series.

Question 9

When using moving averages to estimate the seasonal factors, we need to compute the centered moving average if there is an odd number of seasons.

Question 10

Dummy variables are used to model increasing seasonal variation.

Question 11

Simple exponential smoothing is an appropriate method for prediction purposes when there is a significant trend present in a time series.

Question 12

While a simple index is calculated by using the values of one time series, an aggregate index is computed based on the accumulated values of more than one time series.

Question 13

When deseasonalizing a time series observation, the actual time series observation is divided by its seasonal factor.

Question 14

A time series decomposition method would not be used to forecast seasonal data.

Question 15

A simple exponential forecasting method would not be used to forecast seasonal data.

Question 16

A univariate time series model is used to predict future values of a time series based only upon past values of a time series.

Question 17

Dummy variable regression would be an appropriate method to use to forecast a time series that exhibits a linear trend with no seasonal or cyclical patterns.

Question 18

Trend refers to a long-run upward or downward movement of a time series over a period of time.

Question 19

The simple moving average method is primarily useful in determining the impact of trend on a time series.

Question 20

Forecasters using a multiplicative decomposition model or time series regression model, assume that the time series components are changing over time.

Question 21

Simple exponential smoothing is a forecasting method that applies equal weights to the time series observations.

Question 22

When a forecaster uses the ________ method, she or he assumes that the time series components are changing slowly over time.

A) time series regression
B) exponential smoothing
C) index number
D) multiplicative decomposition

Question 23

Seasonal variations are periodic patterns in a time series that must last at least one year.

Question 24

In the Durbin-Watson test, if the calculated d-statistic is greater than the upper value of the d-statistic, then

A) we do not reject H₀, which says the error terms are not autocorrelated.
B) we do reject H₀, which says the error terms are not autocorrelated.
C) the test is inconclusive.
D) we do reject H₀, which says the error terms are positively or negatively autocorrelated.

Question 25

Exponential smoothing is designed to forecast time series described by regular and seasonal components that are always changing over time.

Question 26

The no-trend time series model is given by

A) TR_t = β₀ + β_1t.
B) TR_t = β₀.
C) TR_t = β₀ + β_1t + β_2t².
D) TR_t = β₀ + β_ln(t).

Question 27

All of the following are forecasting methods except

A) Holt-Winters double exponential smoothing.
B) simple exponential smoothing.
C) time series regression.
D) MAD autocorrelation.

Question 28

Causal variables can be used in forecasting models.

Question 29

Box-Jenkins methodology is a more sophisticated approach to forecasting a time series with components that might be changing over time.

Question 30

If the errors produced by a forecasting method for 3 observations are −1, −2, and −6, then what is the mean squared error or deviation?

A) 9
B) −9
C) 3
D) 13.67

Question 31

If the errors produced by a forecasting method for 3 observations are +3, +3, and −3, then what is the mean squared error?

A) 9
B) 0
C) 3
D) −3
E) 2

Question 32

If the errors produced by a forecasting method for 3 observations are +3, +3, and −3, then what is the mean absolute deviation?

A) 9
B) 0
C) 3
D) −3

Question 33

Random shock is a value that is assumed to have been randomly selected that is the same for each and every time period.

Question 34

When a forecaster uses the ________ method, she or he assumes that the time series components are changing quickly over time.

A) time series regression
B) simple exponential smoothing
C) Box-Jenkins
D) multiplicative decomposition

Question 35

Three criteria used to compare two forecasting methods are the mean absolute deviation, the mean squared deviation, and the mean absolute percentage error.

Question 36

The multiplicative Winters method is used to forecast time series when there are no seasonal factors that are part of the model.

Question 37

The Box-Jenkins methodology can be used to identify what is called an autoregressive-moving average model.

Question 38

Which of the following is not a component of time series?

A) trend
B) seasonal
C) cyclical
D) irregular
E) smoothing constant

Question 39

The multiplicative Winters method used to forecast time series applies a seasonal factor SN_T to the forecasting model.

Question 40

Multiplicative decompositions assume that time series components remain essentially constant over time.

Question 41

The ________ component of a time series refers to the erratic time series movements that follow no recognizable or regular pattern.

A) trend
B) seasonal
C) cyclical
D) irregular

Question 42

The ________ component of a time series measures the fluctuations in a time series due to economic conditions of prosperity and recession with a duration of approximately 2 years or longer.

A) trend
B) seasonal
C) cyclical
D) irregular

Question 43

Seasonal variations are periodic patterns in a time series that are repeated over time. For which one of the following time series variables is it not possible to recognize seasonal variations?

A) quarters of the year
B) months of the year
C) days of the week
D) hours of the day
E) years

Question 44

A major drawback of the aggregate price index is that

A) it does not take into account the fact that some items in the market basket are purchased more frequently than others.
B) it is difficult to compute.
C) it is computed by using the values from a single time series or based on a single product.
D) percentage comparisons cannot be made to the base year.

Question 45

Suppose that the unadjusted seasonal factor for the month of April is 1.10. The sum of the 12 months' unadjusted seasonal factor values is 12.18. The normalized (adjusted) seasonal factor value for April

A) is larger than 1.1.
B) is smaller than 1.1.
C) is equal to 1.1.
D) cannot be determined with the information provided.

Question 46

The ________ component of a time series consists of erratic and unsystematic fluctuations in the time series data.

A) trend
B) seasonal
C) cyclical
D) irregular

Question 47

A sustained long-term change in the level of the variable that is being forecasted per unit of time is

A) a trend.
B) a time series.
C) seasonality.
D) a change due to business cycles.

Question 48

Assume that the current date is February 1, 2003. The linear regression model was applied to a monthly time series based on the last 24 months' sales (from January 2000 through December 2002). The following partial computer output summarizes the results. Determine the predicted sales for this month.

A) 45.9
B) 42.7
C) 44.3
D) 109.1
E) 113.4

Question 49

A sequence of values of some variable or composite of variables taken at successive, uninterrupted time periods is called a

A) least squares (linear) trend line.
B) moving average.
C) cyclical component.
D) time series.
E) seasonal factor.

Question 50

When the moving average method is used to estimate the seasonal factors with quarterly sales data, a(n) ________ period moving average is used.

A) 2
B) 3
C) 4
D) 5
E) 8

Question 51

Which of the following time series forecasting methods would not be used to forecast a time series that exhibits a linear trend with no seasonal or cyclical patterns?

A) dummy variable regression
B) linear trend regression
C) Holt-Winters double exponential smoothing
D) multiplicative Winters method
E) both dummy variable regression and multiplicative Winters method

Question 52

Assume that the current date is February 1, 2003. The linear regression model was applied to a monthly time series based on the last 24 months' sales (from January 2000 through December 2002). The following partial computer output summarizes the results. At a significance level of .05, what is the value of the rejection point in testing the slope for significance?

A) 1.717
B) 1.96
C) 2.074
D) 1.645
E) 2.064

Question 53

In general, the number of dummy variables used to model constant seasonal variation is equal to the number of

A) seasons.
B) seasons minus 1.
C) seasons plus 1.
D) seasons minus 2.
E) seasons divided by two.

Question 54

The ________ component of a time series reflects the long-run decline or growth in a time series.

A) trend
B) seasonal
C) cyclical
D) irregular

Question 55

When the magnitude of the seasonal swing does not depend on the level of a time series, we call this ________ variation.

A) increasing seasonal
B) cyclical seasonal
C) constant seasonal
D) decreasing seasonal
E) no seasonal

Question 56

Since a(n) ________ index employs the base-period quantities in all succeeding periods, it allows for ready comparisons for identical quantities of goods purchased between the base period and all succeeding periods.

A) simple
B) aggregate
C) Laspeyres
D) Paasche
E) quantity

Question 57

Which of the following time series forecasting methods would not be used to forecast seasonal data?

A) dummy variable regression
B) simple exponential smoothing
C) time series decomposition
D) multiplicative Winters method

Question 58

Those fluctuations that are associated with climate, holidays, and related activities are referred to as ________ variations.

A) trend
B) seasonal
C) cyclical
D) irregular

Question 59

A restaurant has been experiencing higher sales during the weekends, compared to the weekdays. Daily restaurant sales patterns for this restaurant over a week are an example of a(n) ________ component of a time series.

A) trend
B) seasonal
C) cyclical
D) irregular

Question 60

In the multiplicative decomposition method, the centered moving averages provide an estimate of

A) trend × seasonal.
B) trend × cycle.
C) seasonal × cycle.
D) trend × irregular.
E) seasonal × irregular.

Question 61

When there is ________ seasonal variation, the magnitude of the seasonal swing does not depend on the level of the time series.

A) cyclical
B) constant
C) irregular
D) increasing

Question 62

A forecasting method that weights recent observations more heavily is called ________.

A) time series analysis
B) first-order autocorrelation
C) multiplicative decomposition
D) exponential smoothing

Question 63

The demand for a product for the last six years has been 15, 15, 17, 18, 20, and 19. The manager wants to predict the demand for this time series using the following simple linear trend equation: tr_t = 12 + 2t. Use this equation to forecast the demand for this product, and then calculate the MAD.

A) MAD = 1.333
B) MAD = 1.6
C) MAD = 2.0
D) MAD = 2.333
E) MAD = 2.5

Question 64

XYZ Company, Annual Data Based on the information given in the table above, we can conclude that, in general,

A) the forecasting method is underestimating demand.
B) the forecasting method is overestimating demand.
C) we cannot determine whether the predictions are underestimating or overestimating demand.

Question 65

In a given week, the NYSE (New York Stock Exchange) is generally open from Monday through Friday. If we wanted to use the multiple regression method with dummy variables to study the impact of the day of the week on stock market performance, we would need ________ dummy variables.

A) 5
B) 52
C) 4
D) 3
E) 365

Question 66

Weighting in exponential smoothing is accomplished by using ________.

A) first-order autocorrelation
B) smoothing constants
C) the Durbin-Watson method
D) multiplicative decomposing

Question 67

XYZ Company, Annual Data Based on the information given in the table above, what is the average forecast error?

A) -1.3333
B) 1.6667
C) -2.5
D) -3.3333
E) 4.5

Question 68

The demand for a product for the last six years has been 15, 15, 17, 18, 20, and 19. The manager wants to predict the demand for this time series using the following simple linear trend equation: tr_t = 12 + 2t. What are the forecast errors for the 5th and 6th years?

A) 0, −3
B) 0, +3
C) +2, +5
D) −2, −5
E) −1, −4

Question 69

The purpose behind moving averages and centered moving averages is to eliminate ________.

A) constant variation
B) cyclical variation
C) seasonal variation
D) regular variation

Question 70

The upward or downward movement that characterizes a time series over a period of time is referred to as ________.

A) seasonal variation
B) cyclical variation
C) a trend
D) irregular variation

Question 71

When using simple exponential smoothing, the value of the smoothing constant α cannot be

A) negative.
B) greater than zero.
C) greater than 1.
D) .99.
E) negative or greater than 1.

Question 72

The ________ test is a test for first-order positive autocorrelation.

A) Durbin-Watson
B) MSD
C) MAD
D) multiplicative Winters

Question 73

When there is first-order autocorrelation, the error term in period t is related to the error term in period ________.

A) t
B) t + 1
C) t − 1
D) t − 2

Question 74

XYZ Company, Annual Data Based on the information given in the table above, what is the MSD?

A) 1.3333
B) 1.6667
C) 2.5
D) 3.3333
E) 4.5

Question 75

Periodic patterns in time series that repeat themselves within a calendar year or less are referred to as ________.

A) constant variations
B) cyclical variations
C) seasonal variations
D) regular variations

Question 76

The Holt-Winters double exponential smoothing method is used to forecast time series data with ________.

A) autocorrelation
B) a linear trend
C) cyclical patterns
D) moving averages

Question 77

The demand for a product for the last six years has been 15, 15, 17, 18, 20, and 19. The manager wants to predict the demand for this time series using the following simple linear trend equation: tr_t = 12 + 2t. Use this equation to forecast the demand for this product, and then calculate the MSD.

A) MSD = 6
B) MSD = 3.3333
C) MSD = 7.0
D) MSD = 2
E) MSD = 2.4

Question 78

The recurring up-and-down movement of a time series around trend levels that last more than one calendar year is called ________.

A) constant variation
B) cyclical variation
C) seasonal variation
D) irregular variation

Question 79

XYZ Company, Annual Data Based on the information given in the table above, what is the MAD?

A) 1.3333
B) 1.6667
C) 2.5
D) 3.3333
E) 4.5

Question 80

The Durbin-Watson statistic is used to detect ________.

A) first-order autocorrelation
B) exponential smoothing
C) multiplicative decomposing
D) irregular variation