The Financial Crisis and Macroeconomic Policy:
Four Years On
John B. Taylor
Stanford University
MONFISPOL Conference
September 19, 2011
The Revival of Keynesian Discretionary
Fiscal Policy in the 2000s
• Economic Growth and Tax Relief Reconciliation Act of 2001
– Refund checks; first installment of 2001 tax rate cuts
• Economic Stimulus Act of 2008 (February)
– Rebate checks and credits
• American Recovery and Reinvestment Act of 2009 (February)
– One-time payments, withholding change, refunds
– More government spending too
• Miscellaneous interventions in 2009-10
– Cash for clunkers program
– First time home buyers program
• Tax Relief , Unemployment Insurance Reauthorization, and
Job Creation Act of 2010 (December))
– Temporary cut in payroll tax
• American Jobs Act of 2011?
Keynesian Discretionary Fiscal Policy First
Became Popular in the ‘60s & ‘70s
• First in academia in the 1950s and 1960s
– Arguments appeared in the major textbooks (Samuelson).
– Keynesian econometric models
• Then in practice: 1962 Economic Report of the President
– “The task of economic stabilization cannot be left entirely to
built-in stabilizers,” the report warned. “Discretionary budget
policy, e.g. changes in tax rates or expenditure programs, is
indispensible—sometimes to reinforce, sometimes to offset,
the effects of the stabilizers.”
– investment tax credit (1962), tax surcharge (1968)
– tax rebates (1975)
– countercyclical grants to states for infrastructure (1977-78)
• Keynesian discretionary policy continued to the late 1970s.
Keynesian Policy Fell Out of Favor in ‘80s & ‘90s
• Research raised doubts about discretionary policy
– Lucas and Sargent “After Keynesian Economics”
– Gramlich “the general idea of stimulating the economy
through state and local governments is probably not a very
good one”
• Soon automatic stabilizers dominated the budget cycle
• Bush 41: proposed tiny stimulus package in 1992
– Shift $10 billion in G from future to the present
– did not pass the Congress
• Clinton: proposed tiny stimulus in 1993
– $16 billion more G
– did not pass the Congress.
• Eichenbaum (1997) “there is now widespread agreement
that countercyclical discretionary fiscal policy is neither
desirable nor politically feasible.”
The Basic Model: A decline in I causes the
aggregate expenditure line to shift down
25_10
SPENDING
45-degree line
Original point
of spending
balance
Original E line
New E line
I falls by this
amount
New point of spending balance
Income or real GDP falls by this amount
(more than by amount I falls ).
New
income
level
Original
income
level
INCOME OR REAL GDP
Countercyclical discretionary fiscal policy:
Increase in G raises GDP depending on size of the
multiplier and amount of crowding out
25_10
SPENDING
45-degree line
G rises
INCOME OR REAL GDP
But Macro Models Differ Greatly
• Romer and Bernstein (Jan 2009) used estimated old
Keynesian models (without RE) to predict ARRA effect
– Large multipliers, around 1.5.
• Cogan, Cwik, Taylor and Wieland (Feb 2009) used
estimated New Keynesian model to predict ARRA effect
– Much smaller multipliers, around 0.5.
• What not use these existing macro models for the
evaluation of actual packages?
– Because they simply repeat the same prediction story over
again.
• So you learn virtually nothing
A Stylized Illustration
Consider two models relating stimulus S to output Y.
Model A is Y= αS +Z
Model B is Y = Z
where Z is a shock and α =1.5
Now, suppose that a stimulus is enacted: S = 2 and Y decreases by -1
According to Model A , Z = - 4
According to Model B, Z= -1
Now consider policy evaluation with counterfactual: S=0
Economists using Model A say:
Just as we predicted, the stimulus package worked.
Without it, Y would have fallen to -4 rather than -1. The decline in
output would have been 4 times as deep, a Great Depression 2.0.
Economists using Model B say
Just as we predicted the stimulus package did not work.
A Less Stylized Illustration
New Keynesian
Smets - ECB
6
Robert Barro
Harvard
6
5
5
With stimulus
With stimulus
4
4
3
3
2
2
If no
stimulus
1
If no
stimulus
1
0
0
-1
-1
2009
2010
2009
“The accumulation of hard data and real-life experience has
allowed more dispassionate analysts to reach a consensus that
the stimulus package, messy as it is, is working”
New York Times November 12, 2009
2010
Use a Direct Approach
• Micro data (used in 2001, 2008)
– Shapiro and Slemrod (2003, 2009),
– Johnson, Parker, Souleles (2006)
– Parker, Souleles, Johnson, and McClelland (2009).
• Macro data
– Special BEA satellite account
– “Personal Income and Output” (monthly to mid ’09)
– “Effect of the ARRA on Selected Federal Government
Sector Transactions” (quarterly)
Monthly Data on Rebate Payments in 2001 and 2008
($ billions, annual rates)
April
May
June
July
August
September
October
2001
0
0
0
95.1
223.1
144.9
2.5
2008
23.3
577.1
334.4
164.1
12.4
0
0
Temporary stimulus meets
permanent income hypothesis
Temporary stimulus meets
permanent income hypothesis again
Cash for clunkers: incentives really matter
Based on
Mian and Sufi (2010)
Quarterly Data
Billions of dollars
320
2008
280
240
200
160
2001
2009
120
80
40
0
01
02
03
04
05
06
07
08
09
10
Effects of Three Stimulus Packages on Disposable Personal Income
Billions of dollars
12,000
11,600
11,200
Disposable personal income
with stimulus
10,800
and without stimulus
10,400
10,000
Personal consumption expenditures
9,600
07Q1
07Q3
08Q1
08Q3
09Q1
09Q3
10Q1
10Q3
Quaterly disposable personal income, with and
without stimulus, and personal consumption
11Q1
Quarterly PCE Regressions With and Without Stimulus Payments
(1)
(2)
(3)
Disposable Personal
Income
.817
(40.9)
----
-----
Disposable Personal
Income--Without Stimulus
----
.857
(73.0)
.851
(60.4)
Stimulus Payments
-----
-----
0.128
(0.81)
Oil Price ($/bbl lagged 2 quarters)
-2.41
(-4.71)
-2.55
(-4.14)
-2.55
(-4.61)
Net Worth (lagged 2 quarters)
.021
(8.53)
.017
(7.32)
.018
(7.97)
Standard error of regression
76.9
65.8
66.3
Billions of dollars
(annual rates)
280
240
200
160
120
- Temporary transfers and
tax credits to persons
80
- Grants to state and
local governments
40
- Federal government consumption
- Federal government investment
0
09Q1 09Q2 09Q3 09Q4 10Q1 10Q2 10Q3 10Q4 11Q1
Major Federal Budget Categories of ARRA
Billions of dollars
1,300
With ARRA
1,200
Without ARRA
1,100
1,000
Federal Government Purchases
900
800
700
600
500
2000
2002
2004
2006
2008
Effect of ARRA on Federal Government Purchases
2010
Billions of dollars
2,400
With ARRA
grants
2,200
Total Receipts of
State and Local
Governments
2,000
1,800
Without ARRA
grants
1,600
1,400
1,200
2000
2002
2004
2006
2008
2010
Effect of ARRA on Receipts of State and Local Governments
Billions of dollars
1,900
1,800
1,700
1,600
1,500
1,400
1,300
1,200
1,100
2000
2002
2004
2006
2008
2010
State and Local Government Purchases: 2000.1 - 2011.1
Billions of dollars
500
450
400
350
300
250
200
00
01
02
03
04
05
06
07
08
09
State and Local Government Expenditures
Other Than for Purchases of Good and Services
10
Billions of dollars
180
160
140
120
100
80
60
40
20
2000
2002
2004
2006
2008
Net Borrowing By State and Local Governments
2010
Billions of dollars
(annual rates)
150
ARRA Grants
100
Other expenditures
50
Government purchases
0
-50
Borrowing (net)
-100
-150
2009Q1
2009Q3
2010Q1
2010Q3
2011Q1
ARRA Grants and State and Local Budgets
(change from 2008.4 when ARRA grants were zero)
State and local government budget constraint
Gt + Et + Lt = Rt + At
where
G = Government purchases of goods and services
E = Expenditures other than for the purchases
L = Lending or borrowing (-), net
A= ARRA grants (exogenous)
R = Revenues excluding ARRA grants (exogenous)
Estimated 3-equation system (1969Q1- 2011Q1)
Gt = 3.86 + 0.864Gt-1 + 0.124Rt - 0.114At
Et = -3.83+ 0.818Et-1 + 0.0398Rt + 0.113At
Lt = .0321 - 0.864Gt-1 - 0.818Et-1 + .836Rt + 1.001At
Billions of dollars
140
120
ARRA grants
100
80
60
40
20
Counterfactual
0
07Q1
07Q3
08Q1
08Q3
09Q1
09Q3
10Q1
10Q3
Actual and Counterfactual ARRA grants
to State and Local Governments
11Q1
Billions of dollars
240
Counterfactual
simulation
200
160
120
Dynamic
simulation
80
Historical
data
40
0
07Q1
07Q3
08Q1
08Q3
09Q1
09Q3
10Q1
10Q3
11Q1
Borrowing (net) by State and Local Governments:
Historical and Counterfactual without ARRA
Billions of dollars
2,350
2,300
Counterfactual simulation
Dynamic simulation
Historical data
2,250
2,200
2,150
2,100
2,050
2,000
07Q1
07Q3
08Q1
08Q3
09Q1
09Q3
10Q1
10Q3
11Q1
Total Expenditures by State and Local Governments:
Historical and Counterfactual without ARRA
Billions of dollars
500
480
460
Dynamic
simulation
440
Historical
data
420
Counterfactual
simulation
400
380
360
07Q1
07Q3
08Q1
08Q3
09Q1
09Q3
10Q1
10Q3
11Q1
Other Expenditures by State and Local Governments:
Historical and Counterfactual without ARRA
Billions of dollars
1,880
Counterfactual
simulation
1,840
1,800
Dynamic simulation
Historical data
1,760
1,720
1,680
1,640
07Q1
07Q3
08Q1
08Q3
09Q1
09Q3
10Q1
10Q3
11Q1
Purchases of Goods and Services by State and Local Governments:
Historical and Counterfactual without ARRA
The Plausibility of the Counterfactual
• Weren’t many states borrowing constrained after the
crisis?
• Not clear but in any case Fed’s Flow of Funds data
show that that net borrowing would have increased
even with such borrowing constraints.
• Net borrowing =
net increase in liabilities - net acquisition of financial assets
• Net borrowing = - $118 billion from 2008 to 2010.
Net increase in liabilities = $53 billion
Net acquisition of financial assets = $171 billion.
• Thus state and local government added significantly to
their financial assets as ARRA grants came in.
• With no ARRA they would not have done so.
Why the Negative Effect on Purchases?
• “Other expenditures” consist largely of Medicaid,
TANF, and other transfer programs
• ARRA conditioned states’ receipt of additional
Medicaid grants on their not reducing benefits or
restricting eligibility rules
• In some states, this meant undoing benefit
reductions or eligibility restrictions that were
implemented in the previous 6 months
– July 1, 2008 is the date in Section 5001 of ARRA
• This “hold-harmless” provision, may have
forced states to reallocate funds that would
have been used for purchases
Test By Splitting ARRA Grants into Medicaid and Other
Billions of dollars
140
120
Total ARRA grants
100
80
Other
60
40
Medicaid
20
0
07Q1
07Q3
08Q1
08Q3
09Q1
09Q3
10Q1
10Q3
11Q1
Regressions with ARRA grants split into Medicaid (M) and other (N)
Constant
G
3.356
(3.5)
Dependent Variables
E
L
-2.471
2.691
(-2.1)
(1.3)
G(-1)
0.882
(17.8)
-------
-0.877
(-12.8)
E(-1)
------
0.875
(20.2)
-0.734
(-11.0)
R
0.108
(2.6)
0.028
(3.2)
0.829
(13.8)
M
-0.318
(-2.3)
0.129
(2.6)
1.200
(6.7)
N
-0.002
(-0.03)
0.076
(1.6)
0.851
(13.0)
Net Effect on Federal, State& Local
Government Purchases
• If government purchases have a greater impact
on GDP than temporary transfers—which the
permanent income theory predicts—then ARRA
could have had a negative effect
• According to the simulations the cumulative
negative effect on state and local government
purchases was $85 billion (341/4). Larger than
the $30 billion (119/4) cumulative positive effect
of ARRA on federal government purchases.
Cross check on GDP growth and G-contribution
Percent, annual rate
6
Growth rate of real GDP
4
2
0
Contribution of
government purchases
-2
-4
-6
-8
07Q1
07Q3
08Q1
08Q3
09Q1
09Q3
10Q1
10Q3
Parrallel Developments
in Monetary Policy
Toward Discretion: ‘60s & ‘70s
• Fed did not follow Milton Friedman’s rules message “of
setting itself a steady course and sticking to it.” (AEA 1968)
• 1965-1970s saw a series of boom-bust cycles in monetary
policy with inflation rising steadily higher at each cycle.
• The wage and price controls of the 1970s
– Epitome of interventionist policy
– Defended by Fed Chair Burns:
“wage rates and prices no longer respond as they once did to
the play of market forces.”
• Insider reports showed little strategy or systematic thinking
• Econometric studies show that Fed’s responses to inflation
were unstable over time in the 1970s
– Rising implicit inflation target
• Not rule-like as it would become in the 1980s and 1990s.
U.S. Inflation and Expected Inflation 1965-1980
14
CPI Inflation
Livingston Survey
12
10
8
6
4
2
0
56
58
60
62
64
66
68
70
72
74
76
78
80
82
84
1965-1980: monetary policy not
well described by a rules-based
price stability objective
1965-79
From “Has the Fed Gotten
Tougher on Inflation?” The
FRBSF Weekly Letter, March
31, 1995, by John P Judd and
Bharat Trehan of the San
Francisco Fed
Toward Rules in the ‘80s –’90s
• Dramatic shift of policy under Volcker in 1979-87.
– Volcker (1983) “We have…gone a long way toward
changing the trends of the past decade and more.”
• Policy continued in ‘80s and ‘90s under Greenspan.
• Additional evidence of more rules-based policy
– more predictable and transparent decision-making process
– focus on expectations of future policy actions.
– announcing interest rate decisions when making them.
• Transcripts of the FOMC in the 1990s, many references
to rules.
• Actual monetary policy closer to simple policy rule
Monetary policy gets more
predictable, inflation targets,
rules-based
1965-79
From “Has the Fed Gotten
Tougher on Inflation?” The
FRBSF Weekly Letter, March
31, 1995, by John P Judd and
Bharat Trehan of the San
Francisco Fed
1987-92
1993-94
Illustrative monetary policy chart from St Louis Fed
February 2007, Bill Poole
The Swing Away from Rules in Recent Years
• Interest rates too low for too long in 2003-05
• On-again, off-again bailouts financed by central
bank’s balance sheet
– on for BSC creditors’ bailout, off for Lehman
creditors’ bailout, on for AIG creditors’ bailout, off for
TARP role out
• Government regulators and supervisors deviated
from sound regulatory rules, especially at large
banks
Illustrative monetary policy chart from St Louis Fed
February 2007, Bill Poole
Illustrative monetary policy chart from St Louis Fed
February 2007, Bill Poole
Chart from Kansas City Fed, 2009, Tom Hoenig
(2000-2009)
Percent
10
Real GDP growth
2009Q3 - 2011Q1
1983Q1 - 1984Q3
8
6
4
2
0
1
2
3
5
4
Quarter since recession ended
The Recovery in Historical Context
6
7
Percent
11
10
9
8
7
6
5
4
3
2
50
55
60
65
70
75
The Unemployment Rate
80
85
90
95
00
05
10
More
intervention,
Percent
11
10
9
8
7
6
5
4
3
2
50
55
60
65
70
75
The Unemployment Rate
80
85
90
95
00
05
10
More
intervention,
more
unemployment
Percent
11
10
9
8
7
6
5
4
3
2
50
55
60
65
70
75
The Unemployment Rate
80
85
90
95
00
05
10
More
intervention,
more
unemployment
Percent
11
10
Less
Intervention,
9
8
7
6
5
4
3
2
50
55
60
65
70
75
The Unemployment Rate
80
85
90
95
00
05
10
More
intervention,
more
unemployment
Percent
11
10
Less
intervention,
less
unemployment
9
8
7
6
5
4
3
2
50
55
60
65
70
75
The Unemployment Rate
80
85
90
95
00
05
10
More
intervention,
more
unemployment
Percent
11
10
More
intervention,
Less
intervention,
less
unemployment
9
8
7
6
5
4
3
2
50
55
60
65
70
75
The Unemployment Rate
80
85
90
95
00
05
10
More
intervention,
more
unemployment
Percent
11
10
More
intervention,
more
unemployment
Less
intervention,
less
unemployment
9
8
7
6
5
4
3
2
50
55
60
65
70
75
80
85
90
95
00
05
10
The Unemployment Rate and Unintended Consequences
Conclusion
• Revival of discretionary policy has been ineffective
• Fiscal stimulus packages
– People largely saved the transfers and tax rebates
– Federal government increased purchases by a tiny
amount
– State and local governments used stimulus grants to
reduce borrowing rather than increase expenditures,
and they shifted expenditures away from purchases
– The results do not support the view that things would
have been worse
• Provide evidence against “Model A”
• Plus counterfactual simulations
• Parallel developments for monetary policy
• Results are consistent with consensus prior to the
revival, based on experience of 30 years ago