Download From Instability to Deflation

From Instability to Deflation
Steve Keen
University of Western Sydney
www.debtdeflation.com/blogs
The permanent crisis?
• Longest downturn since WWII
Index Start of Recession = 100
Rise in Unemployment during Recession
240
230
220
210
200
190
180
170
160
150
140
130
120
110
100
90
80
70
1949
1954
1958
1960
1970
1974
1980
1991
2001
2008
100
0
1
2
3
4
5
www.debtdeflation.com/blogs
6
7
8
The permanent crisis?
• The Great Depression and the Lesser Depression:
Unemployment: Great Depression and Today
30
U-6 Unemployment since 2008
Unemployment 1930-1942
28
26
Percent of workforce
24
22
20
18
16
14
12
10
8
6
4
2
0
0
1
2
3
4
5
6
7
8
www.debtdeflation.com/blogs
9
10
11
12
13
The Financial Instability Hypothesis
•
•
•
•
•
•
•
•
Economy in historical time
Debt-induced recession in recent past
Firms and banks conservative re debt/equity, assets
Only conservative projects are funded
– Recovery means most projects succeed
Firms and banks revise risk premiums
– Accepted debt/equity ratio rises
– Assets revalued upwards…
“Stability is destabilising”
– Period of tranquility causes expectations to rise…
Self-fulfilling expectations
– Decline in risk aversion causes increase in investment
– Investment expansion causes economy to grow faster
Rising expectations leads to “The Euphoric Economy”…
The Financial Instability Hypothesis
• Asset prices rise: speculation on assets profitable
• Increased willingness to lend increases money supply
– Money supply endogenous, not controlled by CB
• Riskier investments enabled, asset speculation rises
• The emergence of “Ponzi” financiers
– Cash flow less than debt servicing costs
– Profit by selling assets on rising market
– Interest-rate insensitive demand for finance
• Rising debt levels & interest rates lead to crisis
– Rising rates make conservative projects speculative
– Non-Ponzi investors sell assets to service debts
– Entry of new sellers floods asset markets
– Rising trend of asset prices falters or reverses
The Financial Instability Hypothesis
• Boom turns to bust
• Ponzi financiers first to go bankrupt
– Can no longer sell assets for a profit
– Debt servicing on assets far exceeds cash flows
• Asset prices collapse, increasing debt/equity ratios
• Endogenous expansion of money supply reverses
• Investment evaporates; economic growth slows
• Economy enters a debt-induced recession
– Back where we started...
• Process repeats once debt levels fall
– But starts from higher debt to GDP level
• Final crisis where debt burden overwhelms economy
– Modeling Minsky…
Keen 1995 Model Foundations: Nonlinear dynamics
• Growth Cycle model (Goodwin 1967, Blatt 1983)
• Capital K determines output Y via the accelerator:
K
1/3
Accelerator
K
1/3
Y
Y
Goodwin's cyclical growth model
Accelerator
1.50
Employment
Wages
• Y determines employment L via productivity a:
l
r
1
a
Y Productivity
l
Labour
/
1
a
r
1
N
LabourPopulation
Productivity
.96
"NAIRU"
+
10
L
WageResponse
/
L
1.25
L
l
r
/
l
1.00
• L determines employment rate l via population N:
100
N
1
Population
Initial Wage
.75
*
l
r
PhillipsCurve
/
dw/dt
l
.50
w
L
+
1/S
+
Integrator
*
0
2
4
6
Time (Years)
8
10
W
• l determines rate of change of wages w via Phillips Curve
+
.96
"NAIRU" 10
WageResponse
*
Pi
PhillipsCurve
I
dK/dt
dw/dt
1.2
1.1
Wages
+
- Y
Goodwin's cyclical growth model
1.3
1.0
• Integral of w determines W (given initial value)
3
Initial Capital
1
Initial Wage
dw/dt
+
1/S
+
Integrator
+
.9
w
L
1/S
+
Integrator
*
W
.8
.7
.9
.95
1
Employment
1.05
• Y-W determines profits P and thus Investment I…
Y
W
+
-
Pi
I
dK/dt
• Closes the loop:
1
Initial Capital
dK/dt
+
1/S
+
Integrator
Modelling Depressions as “Black Holes”
• Goodwin model: No role for debt
• Debt essential element to introduce Minsky
• For debt, essential that capitalists wish to invest more than they earn
– “Debt seems to be the residual variable in financing decisions.
Investment increases debt, and higher earnings tend to reduce
debt.” (Fama & French 1997)
– “The source of financing most correlated with investment is longterm debt… These correlations confirm the impression that debt
plays a key role in accommodating year-by-year variation in
investment.” (Fama & French 1998)
• In words, change in debt equals investment minus profits
• As an equation: d
dt
D = I −Π
Sensitive dependence on initial conditions..
• Two equilibria: “good” with positive employment, incomes
• Which one depends on initial conditions:
– Close to good equilibrium, convergence
– Close to bag equilibrium, convergence too: a “Black Hole”
• “Event Horizon” boundary: many initial combinations can
lead to Depression outcome
Basic Minsky Model: Divergence
Basic Minsky Model: Convergence
3
1.1
2.5
1
2
λ c ( t)
0.9
λ d( t)
ω d( t )
ω c ( t)
1.5
0.8
1
0.7
Employment Rate
0.5
Employment Rate
Wages Share of Output
Wages Share of Output
0.6
0
0
20
40
60
80
100
0
20
40
60
t
t
Years
Years
80
100
Sensitive dependence on initial conditions..
• Debt dynamics behind very different outcomes: • No price dynamics
in this model
• Strictly monetary
model of capitalism
developed to
explore price
dynamics
• Outcome: deflation
arises from falling
wages
• Dynamic price
equation derived
from financial flows
Basic Minsky Model: Divergence
20
19.131
18
16
14
12
10
d.c ( t)
d.d( t)
8
6
4
2
0
2
Employment Rate
Wages Share of Output
− 3.313
4
0
0
5
10
15
20
25
30
35
40
45
50
t
Years
55
60
65
70
75
80
85
90
95
100
T
dP
1
=−
dt
τP

W 
⋅  P −

a ⋅ (1 − s) 

Explicitly Monetary Minsky Model
• Monetary macroeconomic models devised from accounting table:
• Model generates system of coupled ODEs for analysis, simulation
Given
d
BP( t )
dt
FL( t)
τR
−
BE( t)
τL
BP( 0)
EqInit
AT
AT
Liab
Liab
Liab
Equity
 "Priv. Bank"
FL( t) "Workers" "Capitalists" "Bk Equity"
BE"Firms"
( t)
 "Account"
"Loans" "Bk Reserves"
d
F
(
t
)
−
FL( 0) 0
L

τ−LInit τ R
InitLoan
"Value"
dt0
0
0
0
Loan


"Symbol"
FL( t )
PB
( t)
FD( t )
W D( t)
CD( t )
PBE( t )
dR
B
(
t
)
0
B
(
0
)
0

V
V
dt
"Gov
Spend"
0
Gov
0
−
Gov
0
0


"Tax"
0
−dTax
BE( t) 0 FL( t )
0
Tax
0
FD( t )
−
FD( 0) 0
 "Make Loan"
Loan
0
0
0
τ L−Loanτ R
dt0

SP3 := 
"Repay"
−Repay
0
Repay
0
0
0
d
W D( 00) 0

"Wages"
0
0W D( t ) 0 Wage
−Wage
0
d
t

0
0
Div
0
−Div
0
 "Dividends"
d
CD( t) 0
CD( 0) 0
 ":Charge Interest"
Int
0
0
0
−Int
dt0
 "Pay Interest"
−Int
0
0
0
0
FL( t) Int BE( t)
d

−
BE( 0) EqInit
BE( t )
0
Cons W
Cons W
0
0
τ
τ−
dt0
 "Consume"
L
R
 "Consume"
0
0
−Cons C
0
Cons C
0

d0
 "Consume"
0
0
Cons B
BR( t) 0 −Cons B
BR( 00
) 0

dt






















Explicitly Monetary Minsky Model
• Monetary macroeconomics model reproduces stylized facts of crisis
Model output
Smoothed
US Data
15
Unemployment
Inflation
Debt to GDP
10
250
7.5
5
200
2.5
0 150
0
− 2.5
−5
0
1980
5 198510
15
20 1995
25
1990
30
2000 35
40
2005
www.debtdeflation.com/blogs
452010 50
100
55
2015
Debt to
to GDP
GDP
Debt
Inflation
Inflation &
& Unemployment
Unemployment
12.5
300
Aggregate debt overview
• Monetary macroeconomics redefines aggregate demand & supply
– Necessary consequence of endogenous money
• Debt not a “zero sum game” but net addition to demand
• Change in debt finances investment & speculation
• Theoretical Outcome
– AD is income plus change in debt;
• Mathematically proven here (pp. 15-16; 23-25)
– AS is goods & services plus asset sales
• Empirical consequences
– Strong causal (with feedback) relations between
• Change in debt & macroeconomic performance
– Hypothesis: macroeconomic effect at all times
• Acceleration in debt & change in asset prices
– Hypothesis: drives change in growth, asset prices
Aggregate demand, income & debt
• Hypothesis: change in debt has macroeconomic effect at all times
Debt contribution to demand & unemployment
30
Debt change
Unemployment Rate
0
1
20
2
15
3
10
4
5
5
0
0 6
− 5
7
− 10
8
− 15
− 20
− 25
− 30
1980
• Correlation:
1. -0.63 after ZLB—Strong
2. -0.76 1980-Now—Very Strong
3. -0.84 before ZLB—Even Stronger
1985
1990
1995
2000
2005
Sources: As for Figure 3 plus BEA GDP
2010
9
10
11
2015
Percent unemployed (inverted)
Percent of GDP p.a.
25
ZLB
Change in Debt & Aggregate Demand
• Today—compared to Then
US
USAggregate
AggregateDemand
Demand1980-2012
1920-1940
US $ million p.a.
US $ million p.a.
7
1.9
120000
×10
7
115000
Nominal GDP
1.8
×10
7
110000
1.7×10
+Change in Private Debt
7
105000
1.6×10
+Change in Public Debt
100000
7
1.5×10
95000
7
1.4
×
10
90000
7
1.3
×10
85000
7
1.2
×10
80000
7
1.1
75000
×10
7
170000
×10
6
965000
×10
6
860000
×10
55000
6
7×10
50000
6
Nominal GDP
645000
×10
6
+Change in Private Debt
540000
×10
6
+Change in Public Debt
435000
×10
6
330000
×10
1980
1920 19821922
1984 1986
1924
1988 1990
1926 19921928
1994 1996
1930
1998 2000
19322002 1934
2004 2006
1936
2008 2010
19382012 2014
1940
www.debtdeflation.com/blogs
www.debtdeflation.com/blogs
Acceleration in Debt & Change in Employment
• Now (compared to then)
20
15
10
3
15
10
7.5
2
10
5
5
1
5
0
2.5
00
0
−5
−5
00
−1
− 2.5
− 10
− 10
−2
−5
− 15
− 15
−3
− 7.5
− 20
− 20
−4
− 10
− 25
Credit Acceleration
Employment Change
− 12.5
5
− 30
− 15
6
1920
1922
1924
1926
1928
1930
1932200220042006
1934
1936
1938
1940
1980198219841986
198819901992
1994199619982000
200820102012
2014
www.debtdeflation.com/blogs
Change in 100 minus unemployment rate p.a.
Change in 100 minus unemployment rate p.a.
Private Debt Acceleration p.a. as percent of GDP
Credit Acceleration & Employment Change (Corr=0.76)
(Corr=0.69)
Share Prices—the long view
• Dow Jones deflated by the CPI
DJIA deflated by the CPI
1200
1100
1000
Index (2012/02/01: 979)
Mean 1915-1995 (245)
Trend 1915-1995 (2012/02/01: 584)
Index 1915 = 100
900
800
700
600
500
400
300
200
100
0
1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 2020
www.debtdeflation.com/blogs
Acceleration in Debt & Change in Dow Jones
20
200
100
15
150
75
10
100
50
5
50
25
00
0
−5
50
− 25
− 10
100
− 50
− 15
150
− 75
− 20
200
− 100
− 25
Credit Acceleration
DJIA Change
250
− 125
300
− 30
− 150
1920 1988
192219901924
1928 1998
193020001932
1936 2008
1938
1942
1986
1992 1926
1994 1996
2002 1934
2004 2006
2010 1940
2012 2014
www.debtdeflation.com/blogs
Annual Change Deviation from Trend DJIA Percent
Acceleration
p.a. Percent
of (1
GDP
CreditCredit
Acceleration
p.a. Percent
of GDP
year lag)
(Corr=0.65)
Credit Accelerator & DJIA Deviation from Trend (Corr=0.34)
House Prices deflated by CPI—the long view
• NO trend; long term average 98
Real House Price Index
300
275
Greenspan
Index (2006/04: 262; 2012/01: 173)
Mean 1890-1997 (98)
Index 1890 = 100
250
• “a "bubble" in home prices does not appear likely
225
• home price declines, were they to occur, likely would not
200
have substantial macroeconomic implications.”
175
(Greenspan to Congress, August 2005)
150
125
100
75
50
25
0
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
www.debtdeflation.com/blogs; Case-Shiller Index
2010
2020
Acceleration in Mortgages & Change in House Prices
7
21
6
18
5
15
4
12
3
9
2
6
1
3
0
00
−1
−3
−2
−6
−3
−9
−4
− 12
−5
−6
Mortgage Accelerator
Change in Real House Prices
− 15
− 18
−7
− 21
1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014
www.debtdeflation.com/blogs
Percent change in real Case-Shiller Index p.a.
Percent of GDP
Mortgage Acceleration & House Price Movements (Corr=0.78)
How long to recovery?
• On historical trend, could be 15 years…
USA Private Debt to GDP
320
300
280
260
240
Percent of GDP
220
2027
• Sounds implausible?
• If I had said in 1990 that Japan
would enter a 2 decade long slump,
would you have believed me?
200
180
160
140
120
100
Debt to GDP ratio
12.5% p.a. decline
9% p.a. decline
7.5% p.a.
Extrapolated
80
60
40
20
0
1920
1930
1940
1950
1960
1970
1980
1990
www.debtdeflation.com/blogs
2000
2010
2020
80
2030