Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The...

55
Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris, Dec 3-4 2009

Transcript of Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The...

Page 1: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

Housing, Credit and Consumer Spending

John Muellbauer, Nuffield College, Oxford

Conference ‘The Macroeconomics of Housing Markets’

Banque de France, Paris, Dec 3-4 2009

Page 2: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

1. Introduction: Implications of some evidence-based macro research

1. Introduction: institutional diversity.2. Aggregate UK consumption evidence for role of

expectations vs. collateral, wealth, unemployment etc.

3. Aggregate evidence (UK, US, Japan) for failure of Euler equation.

4. Japan is very different.5. US evidence for solved out consumption.6. Systems method for AUS with common latent

variable.

Page 3: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

Co-authors

Janine Aron, CSAE, University of Oxford

John Duca, Federal Reserve Bank of Dallas and Southern Methodist University

Keiko Murata, Tokyo Metropolitan University

Anthony Murphy, Hertford College, Oxford

David Williams, New College, Oxford

Page 4: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

4

Motivation

The global economic crisis of 2008-9 originated in a credit crisis.

Core to a credit crisis is asymmetric information between lenders and borrowers.

The household credit channel played an important part in the preceding boom, as well as in the crisis which began in the sub-prime mortgage market.

Page 5: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

5

Figure 1: The Channels of Transmission of the Mortgage and Housing Crisis

12

Lower Demandfor Housing

SlowerGDP Growth

↓Home Prices & Wealth, SlowerConsumption

Mortgage andHousing Crisis

Less HomeConstruction

Lower Capital ofFinancial Firms

↑ Counter-PartyRisk, Money &Bond Mkts Hit

Credit StandardsTightened

on All Loans

Page 6: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

6

Financial accelerator neglected

by DSGE models popular with central banks and with main-stream macro economists.

-even with ‘New Keynesian’ frictions, mainly price stickiness and adjustment costs.

Micro assumptions usually ignored asymmetric information revolution of 1970s and 1980s for which George Akerlof, Michael Spence and Joe Stiglitz shared the 2001 Nobel Prize in economics.

Page 7: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

7

Don Kohn, Vice Chair FRB, Speech Nov 12, 2008 “The recent experience indicates that we did

not fully appreciate how financial innovation interacted with the channels of credit to affect real economic activity-both as credit and activity expanded and as they have contracted. In this regard, the macroeconomic models that have been used by central banks to inform their monetary policy decisions are clearly inadequate. These models incorporate few, if any, complex relationships among financial institutions or the financial-accelerator effects and other credit interactions that are now causing stresses in financial markets to spill over to the real economy”.

Page 8: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

8

No model currently fully captures linkages and feedbacks of financial acc Hence work is needed on the individual

elements of such a model without, initially at least, a general equilibrium solution.

This paper presents estimates of consumption functions for four major economies in the tradition of Modigliani and Brumberg (1954, 1980) and Ando and Modigliani (1963) but more explicitly incorporating income expectations, uncertainty and credit channel influences, the latter differing across countries and over time.

In Cameron et al (2006), Duca et al (2009) we explore credit, expectations etc. in house price determination.

Page 9: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

9

SOLVED OUT CONSUMPTION FUNCTION

The Friedman-Ando-Modigliani consumption function requires an income forecasting model to generate permanent non-property income.

Unlike Euler equation, it does not throw away long-run information on income and assets.

Evidence by Campbell-Mankiw 1989, 1991 and us is of huge rejection of martingale implication of Euler eq – ‘excess sensitivity’ rules.

Confirmed by our evidence on UK, US, AUS, Japan.

Page 10: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

Log-linearizing the consumption function

The basic aggregate life-cycle/permanent income consumption function has the form:

Dividing by yt and a little manipulation shows that this implies:

Pttt yAc 1*

]/)(1/)/*[(/ 1 ttPttttt yyyyAyc

Page 11: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

Log-linearization cont’d

RHS of this eq. has the form 1+ x, where x is usually a fairly small number. Then take logs, using the fact that log (1+x) ≈x, (2nd order approx. would use x-0.5x2) and

.We then see that

where The last term captures income growth expectations. The

log formulation is good for exponentially trending macro data, since residuals are likely to be homoscedastic.

)/log( tPt yy ( ) /P

t t tapprox y y y

)/log(/loglog 10 tPttttt yyyAyc

0* / logand

Page 12: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

Habits or adj costs implies partial adjustment

0 1 1log ( / log( / ) log log )P

t t t t t t tc A y y y y c

Page 13: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

13

Solved out approach, relaxing parameter restrictions

Robust to limited rationality – just need household common sense about the budget constraint and a concern about sustaining consumption.

Does not require strong assumptions of typical DSGE models: rational expectations common to all agents, representative agent, efficient (financial and credit!) markets, no asymmetric information, no agency problems.

Page 14: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

14

But Friedman-Ando-Modigliani model needs modification for housing and credit

Classical life-cycle theory suggests the ‘housing wealth effect’ on aggregate consumption (including imputed housing) is small or negative.

The credit channel is crucial to explain a positive impact of house prices on consumption via 2 mechanisms: Down-payment constraint affecting mainly

consumption/income. Ability to borrow against home equity, affecting

mpc out of housing collateral.

Page 15: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

15

Implications….

Poorly developed credit markets (e.g. Italy’s) imply aggregate consumption falls when house prices rise. Future first time buyers (and renters) save more for a deposit (or higher future rents), and home-owners have limited access to home equity loans.

Deep mortgage markets imply the opposite. A lower ratio of down-payments to value applies, so future first time buyers will save little and not respond much to higher house prices. Higher collateral values boost spending.

Page 16: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

16

Encompassing Friedman-Ando-Modigliani

Consumption Function and Credit Channel:

Many studies of housing wealth effects suffer from poor controls, but not this one:

0 1 2 3

1 1 2 1

3 1 1

1 2 1

log ( log( / )

/ /

/ log log )

log ( / )

t t t t t t t t

t t t t

t t t t t

t t t t t t t

c r E yperm y

NLA Y IFA Y

HA Y y c

y nr DB Y

Page 17: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

17

Encompassing the Friedman-Ando-Modigliani Consumption Function:

c is real per capita consumption, r is the real interest rate, θ is an uncertainty indicator, and y is real per capita non-property income;

measures income growth expectations;

NLA/y is the ratio of liquid assets minus debt to non-property income, IFA/y is the ratio of illiquid financial assets to non-property income, HA/y is the ratio of housing wealth to non-property income;

1 11 1log( / ) ( log / ) logk s k s

t t t t s typerm y E y y

Page 18: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

18

Encompassing the Friedman-Ando-Modigliani Consumption Function:

∆nrt.(DBt-1/Yt), where nr is the nominal interest rate on debt and DB is debt, measures the cash flow impact on borrowers of changes in nominal rates (Jackman and Sutton, 1982);

the speed of adjustment is β and the γ parameters measure the mpcs for each of the three types of assets.

Page 19: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

19

The credit channel features through:

the different mpcs for net liquid assets, illiquid financial assets (larger for net liquid, Otsuka, 2006) and for housing;

through the cash flow effect for borrowers; by the possibility of parameter shifts with

credit market liberalisation, index CCI. CCI is modelled as latent variable in model

of 10 credit indicators with rich controls, Fernandez-Corugedo and Muellbauer (BOE, wp 2006)

Page 20: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

20

Credit market liberalisation should raise: the intercept α0, implying a higher level of

log(c/y) the real interest rate coefficient, α1

the impact of expected income growth, α3

the mpc for housing collateral, γ3 . But lower the cash flow impact of the change

in the nominal rate, β2.

Page 21: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

2. UK application: first, income forecasting equation

We use average of naïve and sophisticated: naïve has trend reversion, change in short term interest rate, and log real house price x post Thatcher dummy.

(incl. real hp avoids charge that housing collateral effect on consumption is omitted income expectations).

Sophisticated model also includes tax rate, government deficit/GDPxPostThatcher dummy, stock market, union density, credit growth, real oil prices.

Page 22: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

22

UK Empirical Evidence on Consumption 1967-2005

The CCI (credit conditions) level effect is important: in partial equilibrium, it lowers the current household saving rate by 6.5 percentage points, compared to 1980.

The housing collateral effect on consumption rises with CCI, and appears to be close to zero before 1980, Figure 2.

The cash flow impact of nominal rate rises weakens with easier access to credit.

Page 23: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

23

UK Empirical Evidence 1967-2005

The marginal propensities to consume: for liquid assets is 0.11 (close to micro

evidence by Gross & Souleles, 2002); for illiquid financial wealth is around

0.02, Figure 3; for housing wealth is 0.032 from 2001

(at the CCI maximum).

Liquidity effect helps clarify role of money – great confusion reigns currently!

The t-ratios for the mpcs are at least 5.

Page 24: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

24

UK Empirical Evidence 1967-2005

Co-integration analysis suggests that, given CCI, there is a single vector linking log(c/y) and the three ratios of assets to income.

Parameter stability is excellent – recursive betas.

Simpler models, e.g. with single net worth, omitting CCI, omitting change in unemployment rate and in nominal r, are far worse.

Page 25: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

Estimated UK consumption function 1967Q1-2005Q4

Dependent variable = ln c (1) (2) (3) (4)

1ln lny c 0.16 (4.9)

0.23 (6.3)

0.31 (8.5)

0.33 (8.9)

Credit conditions index CCI - - 0.020 (3.5)

Net liquid assets/income 0.0036 (4.8)

0.026 (4,3)

0.033 (5.6)

0.038 (5.6)

Illiquid financial assets/income Ditto 0.0076 (5.9)

0.0071 (5.1)

0.0061 (4.7)

Housing assets/income Ditto Ditto 0.0111 (5.7)

-

Housing assets/income and CCI interaction

- - - 0.0106 (6.3)

Expected income growth 0.21 (4.9)

0.18 (4.4)

0.22 (5.5)

0.10 (2.4)

Expected income growth and CCI interaction

- - - 0.15 (1.6)

Real mortgage interest rate - - -0.03 (1.4)

-0.04 (1.5)

Change in unemployment rate 4ur - - -0.56 (7.8)

-0.64 (8.6)

Debt/income and Δ4 nominal interest rate interaction

- - -0.0029

(3.8) -0.0072

(3.1)

Debt/income, Δ4 nom. interest rate and CCI interaction

- - - 0.0057 (1.9)

Page 26: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

26

Figure 2: Long-run contributions to log consumption/income of the credit conditions index and its interaction with housing wealth/income.

1970 1975 1980 1985 1990 1995 2000 2005

0.000

0.025

0.050

0.075

0.100

0.125

0.150

log ratio

log (consumption/non-property income) Credit conditions index (CCI) Housing wealth/income x CCI

Page 27: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

27

Figure 3: Long-run contributions to log consumption/income of net liquid assets/income and illiquid financial assets/income.

1970 1975 1980 1985 1990 1995 2000 2005

0.000

0.025

0.050

0.075

0.100

0.125

0.150

log ratio

log (consumption/non-property income) Illiquid assets (mov.av.)/income

Net liquid assets/income

Page 28: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

Some policy implications

Model gives 3.5 times larger weight to debt than to housing wealth so net housing wealth is wrong concept for consumption fn. (see Fig 3).

In short run, hard to reduce debt while asset prices and income are collapsing.

Credit crunch has direct and interaction effects. Model made coming UK recession in 2008H2

obvious in summer 2008. But mortgage rate reductions help

consumption. Some evidence for partial ‘Ricardian’ effects in

UK means fiscal policy effects are limited.

Page 29: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

Income growth expectations as the driver of consumption vs. credit, asset prices, uncertainty etc.

King, Pagano 1990 vs Muellbauer-Murphy 1990 What caused the fall in the UK personal saving

ratio in the 1980s? Take perfect foresight view. Define permanent

non-prop income over 10 year horizon. (assume historical growth rate continues beyond 2007).

Plot log c/y against log yperm/y…. Cannot account for 1984/5 to 1988/9 rise in c/y

Page 30: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

Log (c/y) vs. log (yperm/y) with discount rate 1.25% per quart.

Page 31: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

Log (yperm/y) is close to fitted trend – log y, hence easy to predict

Page 32: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

Log (c/y) vs. log (yperm/y) with discount rate 10% per quart.

Page 33: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

Income growth expectations (perfect f., rational or ‘ols learning’) cannot alone explain log consumption/income

Regression of log change in consumption on lagged log c/y and income growth expectations finds latter effect is not significant, 1967-2005

Including lagged A/y, as in Ando-Modigliani (1963), gives sensible long run solution (col 1). Our extended model is even better (col 4).

Strong evidence against DSGE view that asset prices are JUST proxy for expected future growth. Frictions, inefficiencies, non-RE matter.

Page 34: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

Section 3: failure of Euler equation

Centrepiece of all DSGE models is consumption Euler equation.

Hall (1978) established that consumption growth should be unforecastable.

Most DSGE literature ignores inconvenient truth: Campbell & Mankiw (1989, 1991) multi- country evidence on ‘excess sensitivity’ i.e. forecastability, of consumption growth.

New evidence for UK, US and Japan of dramatic rejection of Hall hypothesis, consistent with failure of simple RE.

Supports less restrictive consumption model of Sections 1 and 2.

Page 35: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

35

Section 4. Japan is different: no credit shifts, stable parameters for 1961-2006, and... Positive real interest rate effect on

consumption Negative real land price (or physical

wealth/income) effects. Net financial wealth: mpc≈0.06 Deposits + shares + pension wealth - debts

can be aggregated -- unlike UK or US, where illiquid financial assets have lower mpc.

Unlike UK and US, (liquid assets –debt)/income failed to decline.

Page 36: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

1960

1962

1964

1966

1968

1970

1972

1974

1976

1978

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

2004

2006

Net financial assets/ incomeDebt(excl. uninc. bus)/ incomeDeposit/ income

Note: Income means non-property disposable income.

Ratios to income of household deposits, total debt and liquid assets (Japan)

Page 37: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

Theory suggests +’ve direct real r effect on consumption…

If households have large share of liquid assets in life-cycle wealth

And intertemporal elasticity of substitution is low, i.e. consumers are averse to consumption fluctuations.

Likely to hold in Japan.

Page 38: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

38

Model for Japan (with different income uncertainty measure and annual data)

Income growth expectations

Consumption

2

1 1 2 3 40 1

5 3 6 1

log log 73

log

t t

t

t t t t

BGy a y a Trend a Trend a

GDP

a i i a USGDP

1 1 2 3 1

14 1 5 6

7 1

log log log log log

( log )( )

log( / )

t t t t t t

tt t t t

t

t t

c b y c b y b E y

NFAb E y y b r b y

b pland pc e

Page 39: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

39

Stability and other tests

The Japanese consumption equation passes recursive stability tests.

The LR parameters obtained in the consumption equations are not statistically different from those obtained in system cointegration analyses.

Weak exogeneity tests were accepted for Δlog y and income volatility.

Results also supported by IV estimates.

Page 40: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

40

1990 2000

-0.10

-0.05

0.00

1990 2000

0.4

0.6

1990 2000

0.2

0.4

1990 2000

0.2

0.4

0.6

1990 2000

-7.5

-5.0

-2.5

0.0

1990 2000

0.3

0.4

0.5

0.6

1990 2000

0.025

0.050

1990 2000

-0.03

-0.02

-0.01

0.00

1990 2000

0.00100

0.00125

0.00150

1990 2000

-0.01

0.00

0.01

Figure 11: Recursive stability tests for the equation, Table 2a, col. 4.

Note: order of variables is constant, logy-logc-1, log change in y, forecast income growth, interaction between forecast income growth and income volatility, real interest rate ,net financial assets-1/income, log real land price-1. The cumulative sum of squared residuals and a recursive Chow-test for structural breaks are shown last.

Page 41: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

41

1960 1965 1970 1975 1980 1985 1990 1995 2000 2005

-0.10

-0.05

0.00

0.05

0.10

0.15

0.20

0.25

Long-run contribution to log consumption/income of real interest rate, net financial assets/income and log real land price

Log (cons/non-prop income) Log relative price of land

real interest rate net financial assets (-1)/income

Page 42: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

Policy implications

Japan’s ‘lost decade’ was partly due to weak overall consumption response to lower interest rate.

US scare in 2002-4 about ‘US lost decade’ could have been avoided if special nature of Japan’s monetary policy response had been understood.

Could have avoided one of the causes – too low US rates- of the recent global crisis.

Page 43: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

‘Ricardian’ effect is strong in Japan

Rise in govt debt/income lowered growth rate of disposable income and raised saving rate, contributing to policy ineffectiveness.

Page 44: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

44

5: US Results

We construct CCI back to 1966 from Senior Loan Officer Survey response on willingness to extend credit on consumer installment loans.

Not ideal, since likely to miss some, more specifically mortgage market related financial innovations.

Important to ‘exogenize’ index to remove some of the cyclical economic conditions.

Page 45: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

45

Figure 7: Credit Conditions Index for the US based on SLO

1970 1975 1980 1985 1990 1995 2000 2005

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0 CCI from SLO Survey

Page 46: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

46

US income forecasting models, naïve and sophisticated have big role for Michigan expected fin. conditions

For

Naïve model with no trend reversion also driven by recent av. growth rate, change in nominal T-bill rate, change in log real raw mat. price index.

Sophisticated model has reversion to 1968 split trend and labor productivity; change in T-bill rate (as in UK and Japan); change in log real raw mat. price index; change in unemployment rate; govt. surplus/GDP (as in UK and Japan); and per capita housing starts.

1 1

1 1log( / ) ( log / ) logk s k s

t t t t s typerm y E y y

Page 47: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

47

US Consumption Function 1966 Q3 -2008 Q3

Traditional net worth version has very poor fit, residual autocorrelation and a low adjustment speed of 0.05.

Adding change in unemployment rate and in nominal auto finance rate improves fit and adj. speed 0.07.

Adding CCI (t=4.7) and splitting assets into three categories greatly improves fit and adj. speed to 0.26.

But mpc for housing collateral or wealth at 0.05 (t=3.5) exceeds mpc for net liquid assets.

Page 48: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

48

US Consumption Function (Cont’d)

Adding housing wealth dynamics doesn’t change this.

(Liquid assets - debt)/income trends down almost monotonically since 1980.

So correlation with ‘true’ CCI induces downward bias.

Need to introduce specific mortgage credit conditions indicator

Progress on systems approach where MCCI is latent variable in consumption, mortgage, equity withdrawal, refi equations, also lowers est. housing collateral effect.

Page 49: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

Implications for recent fiscal policy discussion

Marty Feldstein and John Taylor wrongly concluded that 2008 temp tax cut had no effect on consumer spending.

With Blinder-Deaton temp tax adj. our model has largest outlier in 13 years in 2008Q2, approx ¾% of consumption. Back on track for 2008Q3.

2008Q2 would have been a lot worse w/o the tax cut!

‘Ricardian’ effects weaker in US than Japan or UK.

Page 50: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

6. Further progress on systems…

In current work on Australia 1978-2008 with David Williams we model consumption, mortgage stock, mortgage equity withdrawal and hp jointly using common spline function to pick up credit liberalisation specific to the mortgage market.

The first paper to obtain sensible long-run solutions for hp and consumption in Australia, and first to model AUS equity withdrawal.

Page 51: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

Mortgage credit conditions index as common latent variable

1980 1985 1990 1995 2000 2005

0.1

0.2

0.3

0.4

0.5

0.6Mortgage credit conditions index for Australia

Page 52: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

Long-run solutions for AUS

Long–run coeff. on MCCI: for log real hp is 1, for log real mortgage stock is 2.3

(t=4.9) for equity withdrawal/income 0.62

(t=5.0) for cons/income 0.16 (t=2.6) Speed of adj for cons function 0.31

(t=9.5)

Page 53: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

Long-run solutions cont’d

Mpc for net liquid assets 0.17 (t=5.8)

illiquid financial assets 0.033 (t=2.5)

housing collateral 0 before 1978, 0.04 at max MCCI (t=2.8)

Coeff on log yperm 0.29 in 1978, 1 at MCCI max. (less with naïve forecasting model)

Page 54: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

54

Conclusions The ‘housing wealth effect’ on consumer

expenditure works via the credit channel. Evidence on Japan suggests no collateral or H-

wealth effect & little consumer credit market liberalisation.

For US, collateral effect is larger than for UK or AUS (tax subsidy, low interest rate risk, walk-away default option)

There have been major shifts in behaviour with credit market development: a fall in the saving rate (given income, wealth etc) and an increase in the housing collateral effect on expenditure.

Page 55: Housing, Credit and Consumer Spending John Muellbauer, Nuffield College, Oxford Conference ‘The Macroeconomics of Housing Markets’ Banque de France, Paris,

55

Conclusions (Cont’d)

Only part of a larger system, but highly relevant for policy and short term forecasting, given consumption is about 70% of GDP.

US and UK equations imply large rise in household saving rates for 2008 Q3-2009Q2: fall in HW/Y, reduced credit supply, fall in stock market, rise in unemployment rate.

Other system feedbacks need modelling, but mainly amplify direction of these effects – though policy offset.

BOE is currently “up the creek without a model” since BEQM misses most of this.