Transcript Noneconomic Forecasts

Noneconometric Forecasts
Su, Chapter 11
Judgmental Forecasts
•
•
•
•
Based on subjective opinions
Quick and Painless
Most popular forecast used by business
Often used when many factors affect the
variable being forecast and little or no data
are available
Economic Indicators
• Much like naïve forecasts
• Examine many economic series over a long
period of time
– Determine which ones have turned down (up)
before the “total business climate” turns down
(up): Leading
– Turn down (up)at the same time: Coincident
– Turn down (up) after: Lagging
Harvard ABC Curves
• Early Example of economic indicators
– A: Index representing speculation [stock prices]
– B: Index representing business activity [dollar
volume of checks drawn on bank deposits]
– C: Index representing the money market
[interest rate on short term commercial loans]
• Collapsed during great depression
Constructing Indicators:
Determine Reference Cycles
• Reference Cycles: Reflect general business
cycle movements in the economy
• Indicators are all about timing - first
question is timing relative to what?
• “Total Business Climate” as reflected by
reference cycles
• Reference Cycles consist only of timing of
turning points
Reference Cycle Determination:
Recessions
The NBER does not define a recession in
terms of two consecutive quarters of decline
in real GNP. Rather, a recession is a
recurring period of decline in total output,
income, employment, and trade, usually
lasting from six months to a year, and
marked by widespread contractions in many
sectors of the economy.
Reference Cycle Determination:
Growth Recessions
A growth recession is a recurring period of slow
growth in total output, income, employment, and
trade, usually lasting a year or more. A growth
recession may encompass a recession, in which
case the slowdown usually begins before the
recession starts, but ends at about the same time.
Slowdowns also may occur without recession, in
which case the economy continues to grow, but at
a pace significantly below its long-run growth
Reference Cycle Determination:
Depressions
A depression is a recession that is major in
both scale and duration.
Reference Cycles and the NBER
• The determination that the last recession ended in
March 1991 is the most recent decision of the
Business Cycle Dating Committee of the National
Bureau of Economic Research.
• Announcement dates:
–
–
–
–
–
–
The March 1991 trough was announced December 22, 1992.
The July 1990 peak was announced April 25, 1991.
The November 1982 trough was announced July 8, 1983.
The July 1981 peak was announced January 6, 1982.
The July 1980 trough was announced July 8, 1981.
The January 1980 peak was announced June 3, 1980.
Reference Cycle Data
• Available from 1854 - present at the NBER
web site, along with other information
Constructing Indicators:
Study Cyclical Movements
• After determining the reference cycles, NBER
looked at many monthly, quarterly, and annual
time-series, removed trends, cyclical factors and
seasonal variation
• Cyclical movements were compared to the
Reference Cycle, series were separated into four
groups, based on movements
• In 1950: 80 Cyclical series, 30 Leading, 15
Coincident, 7 Lagging, 28 others
Current Cyclical Indicators
•
•
•
•
List is often revised
Currently consists of 250 series
Now maintained by The Conference Board
Web site Conference Board
(http://www.tcb-indicators.org)
Composite Indexes
• It’s hard to use 250 indicators, as each
reflects different parts of the economy
• Need aggregated indexes
• Composite Indexes: Weighted averages of a
group of indexes that are selected according
to a set of criteria. Summary Measures.
• Designed to signal changes in the direction
of aggregate activity
Index of Leading Indicators
•
•
•
•
•
•
•
•
•
•
Avg. wkly hrs, manufacturing
Avg wkly initial claims for unemp ins
Manufacturers' new orders, cons gds and mats
Vendor perf, slower deliveries diffusion index
Manufacturers' new orders, nondefense cap gds
Building permits, new private housing units
Stock prices, 500 common stocks
Money supply, M2
.308
Int rate spread, 10-year T bonds less fed funds
Index of consumer expectations
.181
.025
.049
.027
.013
.018
.032
.329
.018
Jan-01
Jan-99
Jan-97
Jan-95
Jan-93
Jan-91
Jan-89
Jan-87
Jan-85
Jan-83
Jan-81
Jan-79
Jan-77
Jan-75
116.6
Jan-73
Jan-71
108.5
Jan-69
Jan-67
Jan-65
Jan-63
Jan-61
Jan-59
Mar-01
Composite Index of Leading Indicators
106.9
115
110
105
100
95
90
85
80
75
70
Index of Coincident Indicators
•
•
•
•
Employees on nonagricultural payrolls
Personal income less transfer payments
Industrial production
Manufacturing and trade sales
.485
.274
.131
.110
Jan-99
Jan-97
Jan-95
Jan-93
Jan-91
Jan-89
Jan-87
Jan-85
Jan-83
Jan-81
Jan-79
Jan-77
Jan-75
Jan-73
Jan-71
Jan-69
Jan-67
Jan-65
Jan-63
Jan-61
Jan-59
Composite Index of Coincident Indexes
120
110
100
90
80
70
60
50
40
Index of Lagging Indicators
•
•
•
•
•
•
•
Average duration of unemployment
Inventory/sales ratio, mfg & trade
Labor cost per unit of output, mfg
Average prime rate
Commercial and industrial loans
Cons. installment credit / pers inc. ratio
Consumer price index for services
.038
.122
.062
.243
.130
.218
.187
Jan-99
Jan-97
Jan-95
Jan-93
Jan-91
Jan-89
Jan-87
Jan-85
Jan-83
Jan-81
Jan-79
Jan-77
Jan-75
Jan-73
Jan-71
Jan-69
Jan-67
Jan-65
Jan-63
Jan-61
Jan-59
Composite Index of Lagging Indicators
115
110
105
100
95
90
85
80
Forecasting with the Leading
Composite Index
Diffusion Indexes
• Indicators - even those in the Composite Indexes do not consistently move together
• Hard to draw conclusions by watching the ups and
downs of many indexes
• Diffusion Indexes: Measure the relative strengths
of the momentum of expansions and contractions
in a group of indicators
• % of the indicators in any group that rises during a
given time span
How to Compute a Diffusion Index
• See the Conference Board web site for
details
Forecasting with Composite
Indexes: Pro’s and Con’s
• Advantages:
– Use is straightforward
– Reliable
– Free
• Shortcomings:
– Directional forecast only
– Short-run (10-12 months)
– “Measurement Without Theory”
Survey Data
• Based on the concept that you can find out
what will happen by going out and asking
people what they think will happen
• Two kinds:
– Public Opinion Surveys
– Expert Surveys
Public Opinion Surveys
• Typically based on groups of randomly
selected individuals from a specific
category: Consumers, Investors, Voters, etc.
• Collected by different methods
• Accuracy depends on size of sample and
how representative the sample is of the
underlying population
• Visible in media: Polls
Forward Data
• Public Opinion Polls used in economics
– Surveys of people’s opinions about the future
• Consumer Purchases: University of
Michigan Survey Research Center (Index of
Consumer Sentiment)
• Business Expenditures: BEA-SEC Survey,
part of the Annual Capital Expenditures
survey
Expert Surveys
• Survey expert opinions about the future
• Well-known surveys
– ASA-NBER/SPF: Now run by Philadelphia
FED, called Survey of Professional Forecasters
– NABE Consensus Forecast
– Blue Chip Economic Indicators
– Livingston Survey: Survey of Economists, also
run by Philadelphia FED. Biannual
Ja
n80
Ja
n81
Ja
n82
Ja
n83
Ja
n84
Ja
n85
Ja
n86
Ja
n87
Ja
n88
Ja
n89
Ja
n90
Ja
n91
Ja
n92
Ja
n93
Ja
n94
Ja
n95
Ja
n96
Ja
n97
Ja
n98
Survey of Consumer Confidence
130
120
110
100
90
80
70
60
Forecast Surveys: Advantages
• Easy to use, does not involve advanced
theory or econometric techniques
• Provide both direction and quantitative
results
Forecast Surveys: Disadvantages
• Accuracy depends on panel
• Judgmental
Exercise: Evaluating the
Composite Coincident Index
• P. 370: “The National Bureau defined that a
peak is reached if the next three months are
all downward; a trough is reached if the
next three months are all upward.”
• How reliable is this as a forecast of turning
points in the reference cycle?
• How often does the reference cycle peak
close to peaks in the Indicators?
Exercise: Step 1
• Copy the contents of coincident.txt into
Excel and convert it to cells.
• This contains dates, the Composite Index of
Coincident Indicators, and:
–
–
–
–
Peak = 1 at reference cycle peaks
Trough = 1at reference cycle troughs
Expan = 1 during expansions
Reces = 1 during recessions
Exercise: Step 2
• Insert a new column to the right of
“Coincident”
• Label “Delta Coincident”
• Write a formula calculating change in
coincident
Exercise: Step 3
• Insert a new column to the right of “Delta
CC”
• Label it “Delta CC 3”
• Write a formula using the logical functions
“If” and “And” so that this column = 1 if
“Delta CC” = 1 for 3 consecutive months
• Sum up “Delta CC 3” What does this tell
you?
Exercise: Step 4
• We now know how often the Composite
Index of Coincident Indicators turned down
for three consecutive months
• Want to find out how often a peak happened
near these events.
• Define near
Exercise: Step 5
• Let near be +/- three months.
• Construct a logical variable that is equal to
1 if Delta CC 3 equals 1 and Peak = 1 in
any of the three months before or after
Delta CC3
• Sum this up
Formula
• Column E: Delta CC 3
• Column F: Peak
• In Row 6:
=IF(AND(E6=1,OR(F3=1,F4=1,F5=1,F6
=1,F7=1,F8=1,F9=1)),1,0)
Exercise: Evaluation
• How many times was the formula =1?
• How many peaks in the sample?
• How many times did CC turn down for
three consecutive months?
– Is this an “overestimate”?
– Correction?