Robert J. GordonRobert J. Gordon
Northwestern University and NBERNorthwestern University and NBER
CIRET Conference, New York CityCIRET Conference, New York City
October 14, 2010October 14, 2010
Controversial Issues About Controversial Issues About the Recession and the Recession and
RecoveryRecovery
The Plan: From Long-Run to Short-Run to
Policy The reasoning behind my pessimistic long-
term US growth forecast, recently summarized in Business Week
Graphs on dimensions of the weak labor market. The labor market is in worse condition than the product market.
New research on the “Demise of Okun’s Law” and the near-term division of output growth between productivity, hours, and employment growth– Why is the labor market in such bad shape compared
to the product market? The policy debate: what more can monetary
and fiscal policy do?
The Pessimistic Long-run Conclusion
Comparing 2007-2027 forecasts with 1987-2007 actual:
Output growth will slow from 2.9 to 2.4 Output per capita growth will slow
from 1.74 to 1.4 That is the slowest growth of income
per capita “since George Washington” Compare to 2.16 1929-2007 or 2.02
1891-2007
Growth in MFP vs. Ypc by Time Interval, 1891-2027
Growth in MFP and Real GDP per capita, selected intervals, 1891-2027
0
0.5
1
1.5
2
2.5
3
1891-1928 1928-1950 1950-1972 1972-1987 1987-2007 2007-2027
Pe
rce
nt
pe
r Y
ea
r
MFP
GDP/Pop
Components of Growth in Y/H, 1987-2007 vs. 2007-27
Components of Growth of Labor Productivity, Two Intervals
0
0.5
1
1.5
2
2.5
Output/Hour Capital Deepening Labor Quality MFP
Pe
rce
nt
pe
r Y
ea
r
1987-2007
2007-2027
From Y/H to Y/N, the Role of Falling LFPR
Components of Output Growth, Two Intervals
0
0.5
1
1.5
2
2.5
3
3.5
Output Output/Hour Output/Person Hours Population
Pe
rce
nt
pe
r Y
ea
r
1987-2007
2007-2027
Possible Further Room for Pessimism
These projections are based on the historical record of growth between years of “normal” utilization (1987, 2007)
No allowance here for long-run “tainting” effects of the current abysmal economy– Loss of skills and human capital– Years of low investment will increase the age of
the capital stock and reduce the growth of both capital quantity and capital quality
Policy Prescriptions for Long-Run Growth
Problem Slowdown reflects aging of population
and stagnation of educational attainment Solve the first by immigration,
particularly of high-skilled people Work on the second by better
government-run student loan programs and direct measures to address the rising relative price of college education (“higher education cost disease”)
Stimulate demand to avert long-run supply sclerosis
Next We’ll Look at Graphs of Raw Numbers
for Current US Labor Market
Now We’re Looking at– Magnitudes: How Severe Is This Episode?– Timing: Do Labor Market Indicators Change
at the Same Time as Output (Real GDP)?– Which Measures Are the Most Different from
1980-82? We Consider 1980-82 as a Single
Recession and also as two back-to-back recessions– (Jan-July 1980 and Jul 81 to Nov 82)
Output Gap vs. Gap in Output Gap vs. Gap in Aggregate Hours of Aggregate Hours of WorkWork
Conclusion to this point
Comparing the 9.6 level of U rate now to 10.8 in Nov & Dec 1982 is misleading– U rate in July 81 or even Jan 80 started higher– Overall increase in 2007-09 is greater– Much more incidence this time of long-term
unemployment and forced part-time The emergence of long-term unemployment:
is the US becoming more like Europe’s two decades 1985-2005?
Stylized fact: If the empl/pop ratio was the same today as in 2000, there would be 14 million more jobs (9 million from lower unemployment rate, 5 million from higher LFPR) “New Normal” for LFPR? For U Rate?
Documenting and Explaining the Change in
Cyclical Labor-market Behavior
Documenting– A new approach to disentangling trends and
cycles Use of “outside information” from inflation
equation to determine the unemployment rate gap
– A new approach to data Total Economy not NFPB Sector Conventional vs. Unconventional Measures
– Initial finding as reported before: hours gap > output gap in 2008-09, the reverse of 1980-82
The Output Identity: Simple The Output Identity: Simple Version and Conventional Version and Conventional
VersionVersion
NN
L
L
E
E
H
H
YY
NN
L
L
E
E
E
E
H
H
YY
H
H
P
P
P
P
PP
Conventional Compared to Conventional Compared to Unconventional IdentityUnconventional Identity
NN
L
L
E
E
E
E
H
H
YY
H
H
P
P
P
P
PP
NN
L
L
E
E
H
H
YY
H
H
H
H
II
Kalman Trends, Conv vs. Unconv Output
Aggregate Hours
Total Economy Labor Productivity (Y/H)
Output Gap vs. Gap in Output Gap vs. Gap in Aggregate Hours of WorkAggregate Hours of Work
Close-up for Period since 1986
Regression Analysis
Changes in hours gap regressed on– Own lags– Current and lagged changes in output gap– Error-correction term (lagged level of
hours gap)– “End of expansion” dummy variables
(capture overhiring end of expansion, underhiring beginning of recovery)“Early recovery productivity bubble”
Long-run Coefficients: The “Demise of Okun’s Law”
Explanations Offered in My Research
The “Disposable Worker” Hypothesis Similar sources as rising US inequality Increased market power of managers and
highly paid professionals– Increased share of executive incomes coming
from stock options Reduced market power of workers due to:
– Declining unions, declining real minimum wage, low-skilled immigration, and imports
Implications for the Unemployment Rate
Translate forecast growth in hours into the hours gap
Regression coefficients imply roughly 60% of an improvement in hours gap flows into employment rate gap
Optimistic path, output gap shrinks 0.8 points per year (3.3 vs. 2.5)
Pessimistic path, output gap stays where it is forever (2.5 percent growth forever)
Implied Unemployment Paths
Unemployment Rate: Actual and ProjectedOptimistic and Pessimistic Outlooks
0.00%
2.00%
4.00%
6.00%
8.00%
10.00%
12.00%
1986 1992 1998 2004 2010 2016
Year
Optimistic Projection
Pessimistic Projection
Actual
Reasons This May BeToo Optimistic
Why does pessimistic U Rate decline even though output gap is fixed?– Error-correction term implies mean reversion both of
negative hours gap and positive productivity gap Division of predict real GDP growth between
productivity and hours: mean reversion may be delayed– Much publicized “reluctance to hire”– Matched with “reluctance to invest”– Firms are not expecting the 3.3 growth scenario
If the 3.3 growth rate actually happened, short-term outcome would be more productivity growth and less hours and employment growth
But will that output scenario happen?
Reasons To Be Skeptical of Optimistic
Scenario Reinhart-Rogoff. Recoveries after
financial crises are slower than after garden-variety recessions
Ineffectiveness of monetary policy
Political paralysis just as Obama stimulus is about to be withdrawn
The Fed is Out of Ammunition
Textbook IS-LM model still taught in intermediate undergrad macro
Monetary policy is ineffective if:– Horizontal LM curve– Vertical IS curve
The Fed now is plagued by both Why should QE2 work since QE1
didn’t?
The Fed Can’t Control the Cost of Business Borrowing
Fundamental Causes of Weak Recovery (Vertical
IS) Consumption
– Collapse of Household Net Worth– Record-high indebtedness
Residential Construction– Foreclosures and Under-water Mortgages– People walk away from under-water– Their credit is tainted for years– Their houses add to supply but not to demand– My mortgage broker’s story, 3 vs. 80
Consumption Problem:Household Balance
SheetThe Twin Peaks of Household Net Worth
400
450
500
550
600
650
700
750
800
Per
cen
tage
of
Dis
po
sab
le P
erso
nal
Inco
me
Total Assets
Total Liabilities
Net Worth
Sources : Federal Reserve Board Flow of Funds Accounts and Bureau of Economic Analysis NIPA Tables . Details in Appendix C-4.
-150
-100
-50
0
50
100
1970 1975 1980 1985 1990 1995 2000 2005 2010
Housing Starts Used to be a Leading Indicator,
but Not Any MoreQuarterly Housing Starts, 1970-2010
0
500
1000
1500
2000
2500
1970 1975 1980 1985 1990 1995 2000 2005 2010
Qu
arte
rly
Ho
usi
ng
Star
ts (
tho
usa
nd
s)
U.S. Census Bureau Manufacturing, Mining and Construction Statistics
Where is Fiscal Policy?
Krugman NYT Monday: it wasn’t tried.
Look at government employment excluding census workers
The Obama stimulus was too small and too much was wasted on tax cuts and capital-intensive infrastructure spending
What Ended the Great Depression?Chart Extends 1929-41 Quarterly
0
5
10
15
20
25
30
1929 1931 1933 1935 1937 1939 1941
Spen
ding/
GDP R
atio
State and Local Spending/GDP
Transfer Payments/GDP
Federal Spending/GDP
How Does the Obama Stimulus Measure Up?
0
5
10
15
20
25
30
1980 1985 1990 1995 2000 2005 2010
Spen
ding/
GDP R
atio
Federal Spending/GDP
State and Local Spending/GDP
Transfer Payments/GDP
The Current Debate,What About Inside vs.
Outside Govt Debt? Monetary Policy is parralized but still
can support fiscal expansion A fiscal stimulus does not raise outside
debt (“held by the public”) if the Fed buys the bonds
This is the classic Milton Friedman “heliocopter drop” of money
In today’s terminology, QE2 supports a fiscal stimulus
But where is the political will?
Conclusions
A Real GDP path of 3.3 brings the U rate down to 6 percent by early 2014
Reasons this may be too optimistic Pessimistic path leaves U rate at 8 percent in
2016, Europe déjà vu 1985-2005 Model has mean-reversion built in, underhiring
in 2009-10 will be reversed, productivity bubble will be reversed.
Basic problems: Reinhart-Rogoff, ineffective monetary policy, lack of political will on fiscal policy, about to become much worse after the elections