JOHN KITCHEN'S WEBPAGE
The usual disclaimer applies: Any opinions or views expressed
on this page or in any of
the papers are those of the author and are not necessarily the views of any institution.
Real Time Forecasts for Real GDP Growth
Current Projection -- click here.
Previous discussion: The system now searches for best specification and best sample period by indicator.
Previously the system had a fixed 80-quarter historical sample. The program now searches
across a variety of specifications and across sample periods (from a minimum of 24 observations
to a maximum of 80 observations), choosing the best specification by indicator for a given
sample using a minimum Schwarz criterion, and searching for the best sample period for
minimizing the mean absolute error for a recent prediction period (at this time I am using
an 8-quarter period with an ending quarter 2 quarters prior). So, for 2007.Q1 for example,
the program searches for the best specification and sample by indicator for minimizing the
mean absolute error for that indicator’s prediction of real GDP growth over the 2004.Q4 to
2006.Q3 period. For the historical estimates, the 8-quarter evaluation period moves back
accordingly. The 2-quarter lag for the end of the 8-quarter period is to allow for proper
comparison for current quarter estimates when the prior quarter estimate is still not "final."
Also, the 8-quarter period represents a recent period for the quarter of interest, but one
for which, in real time, some quarters would have gone through an annual revision and other
more recent ones may not have.
It takes about 25 minutes for the program to run for any given
quarter’s estimate. Hence, to produce current vintage estimates for 1998 forward required
about 15 hours of computer time. The expanded search process means that the 25 minute run
time is the minimum time required to get an updated estimate for the current quarter as new
data come in; so perhaps it would be real time plus 25 minutes ... Looking at the historical
values, the composite, current vintage RTF-U indicator has the lowest mean absolute error and
lowest root mean square error as a predictor of real GDP growth compared to the individual
indicators for the full period examined from 1998 forward; RTF-U was third best for RMSE and
fifth best for MAE for the 6-quarter period leading into and during the 2001 recession -- from
2000.Q3 to 2001.Q4. Beyond the composite RTF-U indicator, the "best" individual indicators for
the full period were: real exports, industrial production, aggregate worker hours, shipments
of durables goods, and continuing UI claims. For the period around the recession (2000.Q3 to
2001.Q4), the "best" individual predictors were: industrial production, the work week,
shipments of durable goods, continuing UI claims, and ISM non-manufacturing index.
Description: Real time forecasts produce estimates for current quarter real GDP growth
based on a centered value from a set of forecasts from incoming indicators. The intent of the
system is to use data on key indicators as the data become available and generate a
purely data-based estimate of the current-quarter GDP growth estimate.
(A paper on it was published in Business Economics;
the intro description can be found
here.) The real time measures are based solely on available data and do not include
forecaster judgment or add factors. RTF-U (see table) -- the measure of underlying GDP
growth -- is based on current data for individual indicators and the estimated historical
relationship for each individual indicator and GDP growth; weights for combining the
individual indicator estimates are determined by the fits of the historical relationships.
RTF-Q (see chart) is a measure that attempts to capture the quarterly variation of GDP
growth, based on the historical relationship between current vintage actual GDP growth
and RTF-U estimates, and the prediction errors of the RTF-U estimate in the prior
quarter.
----
Paper on "Some Information on the Relative Valuations of Residential and Other Assets Using BEA Fixed Assets Data", January 2008.
Abstract: This paper presents information on the relative valuation of fixed assets, using data from the Bureau of Economic Analysis (BEA) coupled with observed longer-run historical relationshipsand "stylized facts" for macroeconomic relationships in the U.S. economy. Alternative scenarios are examined describing the likely extent of overvaluation of residential assets and the combination of real investment and price changes that would be required to resolve the overvaluation under varying assumptions. The scenarios presented illustrate how changes to the total nominal valuation of the asset stock typically are more easily accomplished through
changes to the price of the asset than from changes in real investment flows, and this is
particularly true for residential assets with their low depreciation rate. The approach provides a useful framework for the evaluation of alternative future paths and the scenarios presented
provide useful benchmarks against which alternative combinations of changes in real investment
and prices can be compared.
INTERNATIONAL DEFICITS, DEBT AND INCOME PAYMENTS:
Revised version of "Sharecroppers or Shrewd Capitalists?"
paper -- now published in Review of International Economics, November 2007.
Summary, less technical paper on "U.S. International Deficits,
Debt, and Income Payments: Key Relationships Affecting the Outlook" (November 2006) --
from presentation to NABE annual meetings in Boston, September 2006. Now published in
January 2007 Business
Economics.
Paper on "Sharecroppers or Shrewd Capitalists?"
(February/March 2006) NOTE: This is an older version ... newer version at top of page ...
For slides from September 2006 presentation at the National
Association for Business Economics
annual meetings session on Capital
Inflows and
the Trade Deficit: Measurement and Implications of Global Imbalances click
HERE.
The slides show updated projections as of the end of August.
Abstract: Large and increasing U.S. international deficits and debt have led to an
apparent conventional wisdom that the United States will pay an increasing share
of total U.S. output over time to service the growing international debt. This
paper presents a detailed framework and analysis of the issues surrounding the
question of whether the U.S. is, in fact, on track to be a society of "sharecroppers"
or rather is actually more consistent with being a society of "shrewd capitalists."
The base scenario projects that the U.S. likely will experience continued growth in
its net international debt position, but with a relatively minor cost of servicing
that debt in terms of the associated net international income flows. Alternative
scenarios based on other analysts’ projections also are presented to illustrate the
reliability of the modeling framework and to show how alternative future paths for
key variables affect the outcomes. The detailed analysis provides insights into
how the underlying relationships affect the final result. In particular, valuation
changes -- and notably valuation changes beyond those resulting from exchange rate
changes -- have played, and likely will continue to play, a large role in the
determination of the U.S. net international investment position. In general, the
results indicate that there is a higher likelihood for the U.S. international financial
position to be "sustainable" and manageable -- even if we were to observe persisting
trade deficits -- than is typically considered to be the case. [Original paper written in
February; revised in March. Updated revised version in process.]
OTHER:
Projections of GDP-Dependent Long-Run Equity Returns [Revised May 2006]
Abstract: This paper provides an analysis of long-run
equity returns based on simulations that account for stochastic real GDP growth projections,
alternative relative equity valuations compared to GDP, and estimated relationships between
adjusted dividend returns and relative equity valuations. Data for total corporate equities
and nonfarm nonfinancial corporate equities were drawn from the Flow of Funds Accounts and
from the National Income and Product Accounts. Data for the S&P500 index also were used.
Simulation results point to projected long-run centered broad-based real equity returns
over the next 45 years in the range of about 5-1/2 to 6-1/4 percent. Outcomes based on a
projected distribution of real GDP growth centered on the Social Security Trustees
intermediate real GDP growth assumption -- real GDP growth averaging 2.1 percent over
the next 45 years -- point to projected long-run, broad-based real equity returns over
the next 45 years centered at about 5-1/2 percent. An assumption of higher real GDP growth
such as that made by Robert Gordon or by public and private consensus forecasts yields a
centered real equity return in roughly the 6 percent to 61/4 percent range.
Payroll jobs outlook:
Payroll Jobs Outlook Discussion Archive
Return to top
"Some Information on the Relative Valuations of Residential and Other Assets Using BEA Fixed Assets Data", January 2008.
"Sharecroppers or Shrewd Capitalists?" February 2006.
Projections of GDP-Dependent Long-Run Equity Returns,
manuscript, September 2005. See discussion above.
"A Top-Down View of Personal Income Tax Receipts: The Role of the
Stock Market and the Business Cycle," manuscript, March 2004 -- revised June 2004. This
paper illustrates how the technical relationship between personal income tax receipts and
total taxable personal income is related to key economic variables reflecting the performance
of the business cycle and the relative valuation of the stock market.
"A Note on the Observed Downward Bias in Real-Time Estimates
of Payroll Jobs Growth in Early Expansions," manuscript, August 2003. This is the paper
cited in the Wall Street Journal article, "THE OUTLOOK: Job Gauge Skips Key Group:
Start-Ups," September 22, 2003.
-------> Click here for an excel file with the
data used in the paper.
"When Real Time Data Fail." An examination of 1992 media
reports on payroll jobs announcements and how they were later shown to be fundamentally
wrong.
"Real Time Forecasting in Practice" as published (with R. Monaco).
Original manuscript, August 2003 (Note: This is the paper that won the National Association for
Business Economics Mennis Award for contributed papers at the 2003 NABE annual meetings in
Atlanta. Published in Business
Economics, October, 2003). Cited in article in The New York Times, Sunday
Febuary 1, 2004, "Growth Forecasts, Without the Wait" by Daniel Gross.
"A Note on Interest Rates and Structural Federal Budget
Deficits," manuscript, October 2002.
"Observed Relationships Between Economic And Technical
Receipts Revisions In Federal Budget Projections,"
National Tax Journal, Vol. LVI, No.2, June 2003, pp. 337-353.
Return to top
|