Discussion of KahnRich on Tracking the New Economy - PowerPoint PPT Presentation

1 / 15

About This Presentation

Title:

Discussion of KahnRich on Tracking the New Economy

Description:

Number of Views:88

Avg rating:3.0/5.0

Slides: 16

Provided by: robert733

Category:

more less

Transcript and Presenter's Notes

Title: Discussion of KahnRich on Tracking the New Economy

1
Discussion of Kahn-Rich on Tracking the New
Economy

2
Productivity Growth is the Hot Topic in
Macroeconomics

Look at the numbers, 6.8 in 2003Q2, 8.1 in
2003Q3
How much of this is an unusual cyclical event?
How much of this is an acceleration of trend?
Advertisement Read another paper, Exploding
Productivity Growth Historical Context,
Possible Causes, Future Implications BPEA
20032, forthcoming
Brookings disallows NBER Yellow-covered Papers
Will be on my web site by Monday morning. Search
Google (not that Scottish University).
97 double-spaced pages

3
Two New Ideas in Kahn-Rich Paper

Idea 1. Productivity growth shifts between
regimes. High and Low productivity growth.
Idea 2. We can do better in estimating the
trend of productivity growth by using outside
information going beyond productivity growth
itself
Info A Real compensation per hour
Info B Real consumption per hour

4
Does their Approach Signal Increase in late 90s
Trend Earlier than Other Methods?

Their claim
One could not decisively conclude that there was
a return to a higher growth regime on the basis
of productivity alone.
Only the corroborating evidence from other
cointegrated series can swing the balance
strongly in favor of a regime switch.
Does Their Series Detect Post-1995 Acceleration
Faster than Alternatives?

5
Motivations for the Kahn-Rich Reevaluation of
Growth Trends

One can be skeptical about both ideas
We treat that trend as a stochastic process
whose mean growth rate has two regimes.
Why two?
Why not high medium low?
Why not four regimes? Ten regimes?
Use two additional series to provide additional
information
Skeptic Why do these two series provide
independent information? Why do the authors not
take us through the arithmetic of labors income
share?

6
Additional justifications for their approach

Criterion for use of two additional variables,
consumption and real wage
1 We show that aggregate productivity data
alone do not provide as clear . . . a signal of
changes in trend growth as does the joint signal
from the series we examine.
What aspect of time-series dynamics provide an
additional contribution from those two series?
2 We do not have to choose break dates
Nor do H-P filter nor Kalman time-varying
coefficient
3 How long regimes last (contingent on only two
regimes?)

7
Arithmetic of Labors Share

Does their real compensation variable provide
additional information?
S (WH/PY)
Change in log labors income share
?s ?w ?p (?y ?h)
Usual cyclical behavior, labors share rises in
recessions, shrinks in recoveries (like now).
In using compensation per hour as a proxy for the
productivity trend, they are making a statement
about the cyclical behavior of labors share.
But they have no model.

8
Consumption/Hour

How is Consumption/Hour related to productivity?
C/H (C/Y) (Y/H)
So now we need a model of the share of
consumption in GDP. Lots of models Keynesian,
RBC, but not in this paper.

9
Alternative Productivity Trends to Compare to
Kahn-Rich

1 Hodrick-Prescott (H-P) filter
Parameter value meaning
Everyone uses 1600. This is the square of
5/(1/8) 40. When detrending GDP, a 5 percent
GDP gap causes the trend to decelerate by 1/8
percent per quarter.
Great Depression 25 percent GDP gap, trend
decelerates at 5/8 percent per quarter, 2.5
percent per year, 10 percent after four years.
Starting at 3 percent per year in 1929, trend by
1933 is growing at 3 25 percent or -22 percent
per year.
H-P parameter 6400 is much better than 1600

10
2 Kalman Filter with Time-Varying Coefficients

The Kalman filter explains the change in
productivity growth (?pt) by a time-varying
constant and any set of other explanatory
variables (ßXt)
(5) ?y - ?h(t) a(t) ßX(t) w(t)
The next step is to specify a time-series process
for the time-varying productivity trend, and the
most straightforward is a random walk
(6) a(t) a(t-1) v(t)

11
Implementation of Kalman Filter Productivity
Trend with Time-Varying Coefficient

Replace ßX(t) by change in output deviation from
trend (or change in unemployment rate).
Could also use supply-shock variables, e.g.,
change in oil prices, that cause temporary
changes in productivity
Kalman filter can use more info than H-P
Compare K-filter with H-P filter
Without output variable they are identical, same
smoothness coefficient

12
Kalman with Output vs. H-P
13
Kahn-Rich Compared
14
Summary Problems with their Productivity Growth
Trend

Problem 1. Why is it so Jagged?
Problem 2. Why must there be only two regimes?
Problem 3. Why does the behavior of the real
wage tell us something about the productivity
trend?
What if super-normal productivity growth goes
into profits? How long does it take for real
wage growth to catch up?
Problem 4. Why does consumption tell us
anything about the productivity trend. What is
the theory of the behavior of the C/Y ratio?

15
Thanks to the Authors

For bringing the Productivity Growth Trend to the
Attention of this Audience one day after the
announcement of 8.1
For helping to focus attention on the strengths
and weaknesses of Hodrick-Prescott and Kalman
And for reminding us that there is a lot of
macroeconomics devoted to explaining the change
in labors income share and the C/Y ratio

Write a Comment

User Comments (0)