Housing, consumption, and the
economy: Why do house prices
become misaligned, and what are
the consequences?
Andrew Farlow
University of Oxford
Department of Economics, and Oriel College
Guest Lecture
(with expanded post-lecture notes)
The London Business School
2 June 2005
This lecture is based on papers to be found at:
www.economics.ox.ac.uk/members/andrew.farlow
Today’s presentation
draws off a five-part project

Part One: “UK House Prices: A Critical Assessment”

Part Two: “Bubbles and Buyers”

Part Three: “UK House Prices, Consumption and
GDP in a Global Context”

Part Four: “Risk Premia and House Prices”

Part Five: “Mortgage Banks and House Prices”
forthcoming

Book: Spring/Summer 2006, Publisher: Constable
Robinson
Real house prices against trend
House prices against gross earnings
House price inflation
35%
30%
25%
20%
15%
10%
5%
0%
-5%
-10%
-15%
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
Source: Halifax UK Real House Price Growth (annualized)
UK historically strong housing
cycles



Strong housing market cycles, linked to volatile
consumption, have been an overriding feature of the
UK economy for over three decades.
OECD (2004): 1971-2002, UK the strongest
correlation of housing wealth and GDP of any
country surveyed.
IMF(2004): The UK (along with Finland, Ireland, and
Switzerland) has had one of the most procyclical
housing markets in the world.
Something more global this time?


Something extra this time: The “first global house price
bubble” (The Economist)?
BUBBLE: “a situation in which temporarily high prices
are sustained largely by investors’ enthusiasm rather than
by consistent estimation of real value.”
Shiller, R.J., ‘Irrational Exuberance’ 2000 page xi




Small-probability large-loss events – they matter.
Even if only a small chance, we should analyze the
possibility.
IMF, 2003 World Economic Outlook, 40% of booms
followed by busts, and 8% cumulative GDP loss (though
important caveats about the data).
State of housing markets, part of a bigger picture of
global imbalances?
UK House price falls and GDP
falls

No fall in real house
prices (blue line) not
followed by major fall in
UK GDP (gray bands).

Will the pattern repeat?
Source: Bank of England Inflation Report, May 2004, p7, Chart 1.10.
Market efficiency

‘Efficient markets’ view of asset prices:


Financial prices always reflect all information (defined at
various levels of stricture) – current price is an unbiased
predictor of the future value of the asset if information
arrives randomly.
Explaining the way economists cling to ‘efficient
markets’ thinking:
“A peculiar psychological disorder known as ‘physics
envy'...We would love to have three laws that explain
99% of economic behaviour; instead, we have about
99 laws that explain maybe 3% of economic
behaviour. Nevertheless, we like to talk as if we are
dealing with physical phenomena.”
Andrew Lo, MIT, financial economist
Market efficiency


Following Fama, E.F., (1970), discrete case.
Let the expected return on an asset between t and t+1
be: r = Et[rt+1] = Et[Pt+1-Pt + dt+1)/Pt]




where E is the expectations operator
P is the per unit price of the asset
d is the dividend or rent
If Et[rt+1]-ρ = 0 is a fair game* where ρ is a constant
term, then, by substitution, Pt = Et[(Pt+1+dt+1)]/(1+ρ)
* Stochastic process xt is considered a fair game if Et[xt+1]=0. In the
case here, Et[xt+1]=Et[rt+1]-rt.
Market efficiency

Solving for n periods recursively, using iterative
expectations (something that economists easily slip in
without thinking too much what it actually implies about
those, such as house buyers, supposedly doing it):
Et [d t 1 ] Et [ Pt  n ]
Pt  t 0

 Pt *  B
n
n
(1   )
(1   )
n


If the second term converges to zero for large enough n,
then the equation gives the long-run equilibrium price,
and the value of the asset is the sum of expected future
dividends or rents.
If the second term does not converge to zero, then the
asset price is said to include a ‘bubble’ term.
‘Efficient Markets Hypothesis’:
Assumptions about investors



1) Either all investors are rational and this is ‘common
knowledge’* or…
2) To the extent that some are not rational, their trades are
random and cancel each other out so as not to affect price,
or…
3) To the extent that some are not fully rational and their
trades are not random but are serially correlated, rational
arbitrageurs eliminate their influence on price. Market
forces work to eliminate the wealth of the irrational types,
and they die out in the population.
* Several recent interesting models of bubbles have all agents rational,
but that this not common knowledge, with the rest of the framework
essentially classical. See Abreu, D. and Brunnermeier, M. K. 2003.
Market efficiency


Sharpe et al.* defines a perfectly efficient market as
“one in which every security’s price equals its
investment value at all times”.
* Sharpe, W.F., G.L. Alexander and J.V. Bailey, 1995
Strongly ‘efficient’ market:




Asset price changes should be unpredictable since they
only respond to new information – which by definition is
itself unpredictable.
Prices changes should follow a ‘random walk’.
Returns unforecastability – where the mathematical
expectation of returns is based on all publicly available
information at time t.
Efficiency tests essentially become tests of forecastability
of price changes.
Problems testing efficient markets





Power of statistical tests in distinguishing efficient
markets hypothesis from alternatives is weak.
Null hypothesis of market efficiency not well
defined.
This is the ‘joint hypothesis’ problem: Market
efficiency is always tested jointly with some model of
equilibrium – i.e. some asset pricing model.
Whatever the result, it can always be claimed that the
asset pricing model was misspecified.
In particular, it can always be argued that a higher
return could simply be compensation for accepting
higher risk.
Tulipmania

To give some idea of how this problem bites: In testing
the theory of 17th century tulipmania, Peter Garber
concludes that we cannot clearly declare it a case of
‘irrational’ pricing; gyrations in price could have been
based on information revelations at the time about
which we know relatively little now, and we may
misuse our benefit of hindsight. Garber concludes
however that he is “hard pressed to find any market
fundamentals explanation” (Garber, P. M., 2000, quoted
from Garber, P.M., 1989). As John H. Cochrane puts it (in
a review of “Famous First Bubbles”): “Garber suggests
fundamental explanations, but he does not nail the case
shut. If it were easy, the events would not have passed
into legend.”
More on problems testing
efficiency



Furthermore, all models of asset pricing used in tests
require current expectations concerning future paths of all
variables in the pricing formula – no mean feat.
And no model is any good at dealing with non-linear
feedback loops (even chaos) since this makes the tests –
based on linear models – impossible to use to prove lack of
a bubble (never mind the fact that agents are supposed to
form expectations over the future paths of such systems).
We never get definitive answers.
Efficient house prices?




If housing markets were efficient, housing prices
would anticipate optimally the stream of real returns
– including housing services – that housing will pay
in the future.
Inefficiency = serial correlation of house price
changes.
Or, another way to think of this, inefficiency = future
house price movements can be predicted from
information available now, namely deviations from
the long-run trend and recent price increases.
The general conclusion is that housing markets are
not efficient.
Specific Problems in Testing
Housing Market Efficiency 1


Similar problems to testing for equity price bubbles –
but probably much worse.
We do not have many long or high-quality time series
on prices or rents (the ‘dividends’ for real estate) for
owner-occupied dwellings:



The implicit rent of owner occupiers is never directly
observed – there is no market to derive an exact valuation.
We end up using proxies from rental indexes in government
statistics.
But, if the measure is based on properties different from
those that are owner-occupied – which it often is – the tests
are biased.
Specific house test problems 2

House price series are hard to construct:

Real estate is not a standard commodity (size, quality,
depreciation, expenditure on improvements, etc.).

We would like repeat sales on the same properties where
the kind and quality is known.

It is after-tax returns we want to know about:




But agents vary in tax liability.
So, after-tax returns are pretty impossible to measure.
We need decades of data for analysis of volatility.
CONSOLATION: While there may be little chance of
proving definitively that a market is not efficient, this
conclusion may not apply in periods of extraordinary price
rises…

And these are usually the periods we are most worried about.
Efficient Market




OBSERVE: The the same problems that economists
have, will equally apply to ordinary investors.
YET we presume that these investors think about
these issues in order to work out when and how to
invest efficiently… including in housing!
CONCLUSION – It is a bit of a red-herring
presuming that we can rely on tests of housing market
efficiency for definitive answers.
Like a good doctor dealing with a difficult patient, we
might want to treat tests as one of our possible
diagnostic tools, and look at other analysis and at the
logical reasons for why markets may find it hard to be
efficient… for example, due to arbitrage failure.
The bottom line
“The bottom line...is that theory by itself does not
inevitably lead a researcher to a presumption of
market efficiency. At the very least, theory leaves a
researcher with an open mine on the crucial
issues....market efficiency only emerges as an
extreme special case, unlikely to hold under plausible
assumptions.”
Schleifer, A., 2000, p16.
Simple test of housing efficiency

Treat housing as just another other asset (something that ultimately
gives utility via consumption, including housing services) and
compare the change in the theoretical price of housing with the
actual price. The theoretical optimal price (P*) equals the sum of the
discounted future income that it will generate. Here, R is the rent
level, g the expected growth rate of rents, r the mortgage interest
rate, τ the risk premium. Those who own are treated as renting from
themselves – which is how their ownership is treated in the National
Accounts (Blue Book) under ‘imputed rent’*
 1 g 
P*  t 1 R

1 r  
so long as  0

t
R

r  g 
* Or, since the alternative to owning is renting, the opportunity cost of owning (via payment of interest) is
compared to the alternative of renting. So if the typical rent is falling as house prices are rising, then buying
housing and renting it to yourself is getting more expensive than letting someone else buy it and you rent it
from them. There are, however, some weighty problems in dealing with the tax advantage of housing and the
Housing market efficiency test

Hence
P
P
  r  g   
P*  R
If P/P* falls, real estate prices fall below their
initial point (not necessarily the optimal reference
point itself). If P/P* rises significantly (exceeding 1
with caveats) this is described as a ‘bubble’.

To use this to work out the degree of overvaluation
of a property market we need to know some moment
at which the market was actually in equilibrium.

Bubble builder-bubble burster 1




Another popular (empirically-based) method used for
checking for mispricing/‘bubbles’ in property
markets, and applied in various countries, is the
methodology of Abraham and Hendershott 1996.
Methodology is also useful for locating the moment
at which a property market is in ‘fundamentals
equilibrium’.
(The nomenclature here is identical to that of
Abraham and Hendershott)
*Abraham, J.M., and P.H. Henderschott, 1996.
As an alternative, Hall, R.E., 1978, runs a stochastic switching regime model in real
house prices. Booms in house prices are associated with an unstable regime. The
probability of the system staying in the regime fall as deviation from equilibrium
increases.
Bubble builder-bubble burster 2



The equation for property prices contains a fundamentals longrun equilibrium term (based on the efficient markets
hypothesis, using a standard asset pricing model), drppit*, and
an error term:
drppit=drppit*+ut
Concentrating on the fundamentals term for a moment. The
growth rate of the fundamentals term is modelled as a function
of the growth rate of rental prices, real interest rate, real
construction costs and real effective exchange rate, real GDP,
real GDP per capita, nominal wage, real wage, population, and
various supply factors (with all prices, rent, etc. deflated by the
CPI). All variables (except interest rates) are in logs, so that
rppi = log real property price index, RINT = real interest rate:
drppit*=a0+a1RINTt+a2drrisat+a3dpcrcgdpt+...+lagged
Bubble builder-bubble burster 3

The error term is further specified to capture the
dynamic adjustment. The term is the sum of a bubble
‘builder’ (expected future appreciation), a bubble
‘burster’ (the possibility of a price drop if price
exceeds the fundamental price by a certain limit) and
a random variable:

ut= λ0 +λ1drppit-1+λ2(rppit-1*-rppit-1)+εt
Bubble builder-bubble burster 4



This analysis of course requires some estimate of the
parameter rppi*, which needs also to be consistent
with drppi* and the estimates of parameters ai. This is
done as follows (and is worth explaining since it
indicates a method for working back to the moment
when the UK property market was last in
fundamentals equilibrium):
i) Estimate without λ2 (i.e.. ignore λ2 for now)
ii) Construct a first-pass estimate of rppi* by
cumulating the drppi* over time using the parameter
estimates derived from i);
Bubble builder-bubble burster 5



iii) Calibrate (not always a sound idea) the rppi*
series on the assumption that actual property prices
were, at some particular point in time, in equilibrium;
iv) Re-estimate drppi, this time including the λ2;
v) Keep repeating steps i) to iv) until the ai estimates
stabilise, at which point the drppi* and rppi* will also
have stabilised. This also pins down the year when
the market was approximately in fundamental
equilibrium. In the UK this would be about 1994. In
the original paper, in Hong Kong it was about 1990.
Bubble builder-bubble burster 6




If λ1 > 0, the lagged growth of real property prices
acts in a way to perpetuate the growth of real property
prices, i.e. a bubble. λ2 > 0 captures the notion that the
bubble bursts when the actual price level λ1drppit-1
exceeds the equilibrium price level rppit-1*.
Putting all this together we get real property price
growth:
drppit* = (a0+λ0)+a1RINTt+a2drrisat+a3dpcrcgdpt+...
...+λ1drppit-1+λ2(rppit-1*-rppit-1)+εt
Some results – nominal interest
rate


These studies often start with the nominal interest rate
included as a variable, but find that it drops out as
insignificant in the long run.
This is a rather striking revelation. It backs up the
claim made and analysed in Part One, and found in
many studies, that ultimately – and contrary to much
of the mortgage bank emphasis and media coverage –
it is real interest rates that matter.
Some results – US, Hong Kong

Hong Kong and US metropolitan cities*, tendencies
towards bubbles


Bubble builder parameter:




Estimated bubble builder and the bubble burster parameters were
significant and comparable:
Hong Kong: 0.3
US metropolitan cities: 0.5
US remainder: 0.2
Bubble burster parameter:



Hong Kong: 0.05
US metropolitan cities: 0.1
US remainder: 0
* Based on 30 US Metropolitan cities for the period 1977-1992 (Abraham, and
Hendershott, ibid).
Some results, cont.



Specifically, the coefficient on the one-period lag in
ARIMA(1,1,0) was found to be large, positive and
significant: in other words appreciations tend to be
followed by further appreciations.
Over the period covered, in US metropolitan cities a
1% increase in real property prices in a quarter tended
to be followed by a 0.6% increase in the next quarter.
In Hong Kong, the absence of volume changes in the
number of apartments and of the real construction
cost index as explanatory variables in the estimated
equations was “striking” and confirmed the relative
importance of demand side factors in explaining
short-term property price movements.
Some results, cont.


In the US metropolitan cities, changes in market
fundamentals and adjustment dynamics (including the
bubble builder and bubble burster components)
together explained about 60% of the variation in price
movements and, separately, about 40%.
On the assumption that Hong Kong prices were
broadly at equilibrium in the early 90s, the upswing
peaked at about 40% to 45% above fundamentals
level: “This estimate is broadly consistent with
market perceptions at the time.” In 1998, property
prices declined by an average of 40%.
Arbitrage

“To make a parrot into a learned financial economist
it needs to learn just one word – arbitrage.”
Stephen Ross (a learned financial economist).

Arbitrage = “the simultaneous purchase and sale of
the same, or essentially similar, securities in two
different markets for advantageously different prices”
Sharpe, W., and G. Alexander, “Investments.” 4th Edition, quoted in
Shleifer, A., and R. W. Vishny, 1997.



Arbitrage is at the heart of the theory of financial
market efficiency.
Its failure is at the heart of cases of market
inefficiency and bubbles
Inability to arbitrage especially at times of excess.
Arbitrage




Arbitrage is a game of co-ordination.
The outcome presumed in classical finance is based
on the common knowledge of all players that they are
all synchronising their arbitrage strategies and that it
is riskless.
In theory, the desired outcome involves no risk of
capital loss.
Arbitrage:




Removes influence of idiosyncrasies
Removes importance of institutional detail
ties price to fundamental value
Hence mispricing does not last
Arbitrage in practical reality…


We cannot presume that agents converge on the
optimal outcome in this game if this common
knowledge is lacking.
Player needs to be sure that enough other players are
also arbitraging






Otherwise individual acts of arbitrage are too individually
risky and costly;
Most positions involve capital and risk of loss;
Arbitrage limited by risk-bearing capacity of arbitrageurs in
the aggregate;
These risks vary across markets and institutional structures;
Arbitrageurs are using “other peoples’ money”…So face a
principle-agent problem;
Arbitrage = very broad interpretation, to include, decision on
house buy/sell, bank lending strategies, estate agent tactics.
Arbitrage Failure – Why no-one
corrects a bubble


Behavioural finance
 Idiosyncrasies matter, psychological motivations,
thought processes about each others’ thought
processes, etc.
 Institutions matter… especially at times of excess.
Combining psychological with institutional… harder
to rule out arbitrage failure… hence bubble-type
behaviour.
Two-sided arbitrage failure




Customers don’t arbitrage
Mortgage banks don’t arbitrage each others’
behaviour either
Mortgage banks don’t arbitrage consumers’ behaviour
In times of excess it is better to exploit momentum of
customers than to try to fight against it.
Risks to Arbitrageurs - general

Risk at various levels – the levels interact


Fundamentals risk
Financing Risk/Noise Trader Risk
Horizon Risk
 Margin Risk
 Short covering risk


No time to discuss most of this today


Fundamentals risk (next few slides)
Horizon Risk

Without fundamentals risk


E.g. Royal Dutch Shell, etc…
With fundamentals risk

Housing…
Fundamentals risk

Classical finance


The greater the deviation from fundamentals the more
aggressive the arbitrage since potential returns are higher.
BUT arbitrage is a game of coordination:


The less that players collectively understand that they have
deviated from fundamentals, the less aggressively they
collectively arbitrage.
If an arbitrageur takes a position and then some good
fundamentals news is released that justifies what had
originally seemed a mispricing, the position must be closed
at a loss.
Fundamentals risk

Inability to define a fundamental value = two problems.

More difficult to test market efficiency


One can never reject the null hypothesis of market efficiency
when it may be that the model was misspecified.
Hard-to-define fundamentals make it easier for mispricing to
develop in the first place.



With no unique fundamental value from which players can
know (in the common knowledge sense of the word) that they
have deviated, it is impossible to achieve common knowledge
of deviation from fundamental value.
This increases the uncertainty of players regarding the acts of
other players and, hence, the uncertainty of gains from
arbitrage;
Which reduces the expected payoff from any given position;
which reduces the incentive to take those positions in the first
place
Fundamentals risk…housing

Housing added problems:






Very heterogeneous asset.
Highly segmented market. There is no central exchange
and information about the value of the underlying asset is
much more imperfect.
Behaviour of intermediaries. Very few (do any?) estate
agents price according to macroeconomic or other
fundamental factors – instead they price almost always in
comparison to recent local sales.
Consumers - better at pricing relative to fundamentals.
Only very imperfect substitutes, since the risk will have
been poorly, if at all, hedged.
Small transaction and information costs become
serious impediments to arbitrage.
Greenspan 1

"I don't think we have a bubble in house prices. First,
let's remember it's very difficult to get one. Unlike
stocks, where you have a single market, low
transaction costs and an ability of people to pile on
nationally and cumulatively, residential housing
markets are all local."
Federal Reserve Chairman Alan Greenspan, 17 April
2002, Testimony before the Joint Economic
Committee.
Greenspan 2


Greenspan emphasises entry of speculators – perhaps daytraders with low transactions costs, able to ‘pile on’, as
Greenspan puts it, and generate herd behaviour and bubbles.
But entry is not the only issue in a bubble phase; correction,
via acts of arbitrage, is important too.




Day traders can also easily ‘pile out’, correcting the market.
We can’t immediately presume that the power of the first
dominates the latter (the very knowledge of the ability to ‘pile
out’ may itself act as a corrective influence).
Arbitrage is a double-edged sword; just as rational agents
arbitrage away inefficient pricing, irrational traders arbitrage
away efficient pricing.
If each group has significant risk-bearing capacity, both will
influence the price.
The ability of the first group to offset the second group is very
weak in housing markets.
Substitutes, Short-Run Supply,
and Short Sales

Arbitrage needs close substitutes
Arbitrageur goes short the expensive security and
long very similar but cheaper securities
 Works even if some agents are not fully rational
 Even if demands are correlated
 Price determined relative to close substitute and
NOT the absolute supply of the asset


So, may achieve “Law of One Price” in shares
to one company even if not of whole market.
Equities substitutability

Equities weak substitutability
Roll, 1988
 Amount of variation in return on large stocks
explained by aggregate economic factors, returns
on other stocks in the same sector, and firmspecific publicly-known news.
 35% on monthly data.
 25% on daily data.
 In other words, there is very little substitutability
between even equities.

Substitutability…



Even if individual securities have better substitutes
than the whole market, many broad categories do not
have substitute portfolios.
Risky to get in and out of shares as a class given
given high historical rate of return.
Housing – only strategy if there is suspected
overpriced is to get out altogether, with intent to get
back in later when price is lower:


But very risky.
Housing also has no (or low) ‘release valve’ of
international trade, unlike equity.
Inability to go short in housing

Creates an asymmetry
Can’t exploit profit opportunities if market
expected to decline (c.f. can in principle in equity
markets).
 Buy-and-hold in periods of excess positive return
 Asymmetry feeds bank expectations of how to
exploit volatility.
 Feeds bubbles.

Replicability of asset


The more easy it is to create substitutes to current assets,
the less likely it is to get a bubble.
Affects interpretation of slow supply response of
housing … as not just a determinant of ‘fundamentals’
… but as a ‘bubble’-generating feature too.



Normally the ex ante knowledge that excess price relative to
fundamentals will draw forth extra supply of underlying asset,
acts to numb price pressure
Slow supply response feeds episodes of mispricing
Supply lags – long planning and build process exaggerate boom-bust cycle.

Option based component to when high uncertainty about
fundamentals… resist supplying till some of the uncertainty is
resolved
Diversification, Liquidity and
Arbitrage

Portfolio theory




Should be well-diversified.
In the limit (though taking into consideration transactions
costs) should participate in all securities markets
Should not over-invest in assets that correlate with own
income
 Plenty of equity evidence that workers do oveinvest in
own firm stock)
Standard arbitrage notion is that small number of marginal
positions, with limited extra exposure
Diversification, Liquidity and
Arbitrage…

Not work well in housing


Most players not well diversified… heavily exposed to this
one asset
Most decisions are not marginal…


Encouraged by tax system


We tend to rely on large-value decisions by small number of
arbitrageurs (first-time buyers, etc…. one of the lest
experienced categories of investors)
Principle residence is tax advantaged, buy-to-let sales
generate capital gains liability
Besides, even well-diversified arbitrageurs find that
markets tend to fall in a correlated fashion.
Liquidity





Liquidity - smart traders can arbitrage mispricing.
But liquidity makes it easier for foolish traders to
arbitrage away efficient pricing.
Those in housing are less liquid.
Irrational entrants “more liquid”.
Falling market = forces sales into highly illiquid
market.
Housing market problems: Summary






The market is dominated by individuals who only trade in their
own home.
High transactions costs, carrying costs, and tax considerations
make it relatively difficult for professionals to take advantage
of profitable arbitrage opportunities when the market is
overvalued.
Markets are local (in the sense that house purchase decisions
are related to many other aspects of the buyer’s location)
reducing the willingness to arbitrage across markets.
And the inability to hedge – no futures or options market.
All this dramatically reduces the aggressiveness of
arbitrageurs.
With arbitrageurs discouraged, markets become more open to
momentum and panic-based inefficiencies
Central banks exacerbate arbitrage
problems…bubbles worse?


Usual rationally-determined, ‘put’ option generated by central
bank asymmetric behaviour.
In the housing market context this refers to that part of house
prices that factors in the value of the downside insurance that
buyers and lenders perceive (rightly or wrongly) on account of
the central bank’s willingness to act vigorously when prices fall,
though being less willing to act when prices rise. Just as an
investor can set a floor to the price of a security by buying a put
option, so the central bank can set a floor for house prices by
cutting interest rate when prices go too low for their liking. At
the same time there is no call option, or ceiling, on house prices
such that a rise in interest rates would be triggered. Of course,
the ability of the central bank to make good on the ‘put’ option
is another matter altogether. What consumers believe and what
the central bank is able to deliver may be very different.
Central banks exacerbate
arbitrage problems?



Extra, behavioural, angle.
Central bank asymmetrically alters the uncertainty of
those (house buyers/mortgage banks) who would
arbitrage = increase incidence of behaviourally
generated bubbles.
Arbitragers, when trying to correct a bubble, find that
the central bank effectively works against them, and
this is, ex ante, factored into the strength of their
arbitrage and their ability to correct momentum
behaviour, and this is, ex ante, factored into the
behaviour of those engaging in bubble-type
behaviour.
Psychological biases of house buyers

Biases
feed bubble-type behaviour.
 Make arbitrage harder.


Optimism and similar biases
98% of confidence intervals only contain the true
value 60% of the time
 “Illusion of control”



exaggerate ability to control own situation
“Self-attribution/confirmation bias”

Tendency to interpret investment successes as confirming
own abilities
Optimism...

“Availability cascades”

Ballooning number of TV programs reassure that prices
an only ever go up.
A few optimists (buy-to-lets?) capable of driving
prices above fundamentals…given no short selling
possible, few active traders.
 Seek confirmatory information. Hirschliefer
evidence that excessive attachment to activities on
which we have expended a lot of resources.

Why don’t optimists die out?

Sort of Darwinian selection


It doesn’t work
Upward trend in prices means optimists survive and those
being ‘more rational’ fall as percent of population



Note role of banks in this… they have to feed the ability to
borrow against collateral value
So…this is also a story about ‘optimist banks crowding out
‘rational’ banks
And hence a story about ‘bank bubbles’ and equity
valuation of banks.
Momentum reasoning




Empirical/experimental evidence that agents ignore
laws of probability.
Kahnemann and Tversky 1974, ‘representative
heuristic’… subjects extrapolate short time series.
Currently low nominal interest rates and ‘burden of
debt low’.
Forget not repaying debt so quickly as in the past.
Some loan rates and loan sizes
Percent burden of monthly payments after
1
2
3
4
Salary
Mortgage
Mortgage
rate
Monthly
repayment
Inflation
rate
Wage
inflation
1 year
5 years
30,000
100,000
10%
£918
7%
9%
34%
30,000
155,300
5%
£918
7%
9%
30,000
155,300
5%
£918
2%
30,000
100,000
5%
£591
2%
10years
25years
lifetime
24% 16%
4%
15%
34%
24% 16%
4%
15%
4%
35%
30% 25%
14%
23%
4%
23%
19% 16%
9%
15%
Shift in loan repayment schedule
Lower inflation. Lower real repayments now,
higher real payments later.
Source: Bank Of England Quarterly Bulletin, Spring 2003, p129
Consumers bid out debt
Percent income spent of mortgage against year of
repayment
Row 1 The initial situation
Row 3 Real rates the same but take on more
debt.
Row 4 Real rates the same and take on same
debt.
Anchoring, regret, and disposition
effects

House price anchoring



Northcraft and Neale 1987
Regret
The disposition effect and house price stickiness




Dislike losses more than the enjoy gains
Genesove and Mayer (2001) find a ‘disposition effect’ in
housing.
Falling downtown Boston housing market in the 1990s,
owners facing negative equity tended to set prices at 25% 35% of the difference between the property’s expected selling
price and their original purchase price
consequently, they held out too long and made an even greater
loss
Anchoring, regret, and
disposition effects…



See similar behaviour in private rentals markets, when
owners would rather take another month without a tenant
then chip 5% of the rent
Helps explain part of the fall in volumes of housing market
trade at a peak before prices come down, and why house
price falls in the early stages of a collapse can be slow.
Evidence too of price stickiness:
Sellers compare their selling price to the selling price of
other homes nearby.
 Hard to coordinate price drops if each seller does not want to
be the first to cut.
 This, too, feeds the potentially slow onset of the unwinding
from a bubble.
 Less resistance to price rises in the bubble phase, this
generates a ratchet effect.

Transaction costs and mispricing

Fama (1991) efficient market is one where “deviations
from the extreme version of the efficiency hypothesis are
within information and transactions costs”. We need to
understand, therefore, whether transactions costs can
cause a degree of volatility that is even greater than the
efficient level of volatility. In a world with momentum
traders, it can be.
Transaction costs and housing
market ‘frenzy’

High and lumpy transactions costs







Brokerage fees.
Buyers’ and sellers’ search costs.
Moving costs.
Capital gains costs (for buy-to-lets who decide to temporarily
sell out in an overvalued market).
Tax on interest earned from holding cash rather than property
while waiting for correction, etc.
This reinforces the illiquidity of the market.
Empirical studies further find that lumpy transaction
costs lead to important nonlinearities in price dynamics.
Transaction costs and housing
market ‘frenzy’…



The large fixed element of transactions costs
influences the choice between buying and renting,
and whether to trade up.
In periods of greater appreciation of house prices,
more households are pulled over the transactions cost
hurdle to engage in momentum trade, in the
expectation of capital gain.
At these times of heightened activity – of ‘frenzy’ –
increased demand feeds back into higher prices and
further demand.
Frenzy effect

Behavioural/arbitrage failure framework.




In stable periods both consumers and banks are more likely
to behave in ways that arbitrage the market.
Many speculative agents keep out and the market is less
likely to deviate from fundamentals for long periods of time,
and arbitrageurs – both banks and house buyers – are less
likely to be ‘punished’ for doing their job.
Unstable, ‘frenzy’, periods agents stop arbitraging the
market since it is too dangerous
Prices may deviate even further from fundamentals as many
speculative agents enter, and arbitrageurs are much more
likely to be ‘punished’ for trying to correct the market.
Frenzy effect…

Phasing out of mortgage tax relief over the 1990s
should have helped to cool market by giving less
benefit to owners than to renters



Recent ‘frenzy’ must have been particularly severe.
Low nominal rates have fed agents’ ability to ‘go with the
frenzy’.
The number of houses on estate agents’ books is a
function of the ‘frenzy’ effect.



In ‘frenzy’ periods the flow off estate agents’ books is greater
than the flow on.
In bust periods the resistance to price cuts means that the flow
off the books falls relative to the flow on.
Casually asserting that stock on the books is a measure of
fundamentals supply and demand conditions is simply wrong
Frenzy effect…


The ‘disposition effect’, price stickiness and ‘frenzy’
effect, together mean that housing on estate agents’
books and trading volume are good leading indicators
of future price changes.
Econometric evidence for ‘frenzy’ behaviour in
housing markets:

Muellbauer, J., and A. Murphy, 1997; Hendry, D.F.,
1984; Quigley, J.M., 1999; Case, B., and J.M. Quigley,
1991; and Hall, R.E., 1978.
Frenzy effect
“Without such a nonlinearity or dummies for the spikes in
the data, the equation standard error more than doubles”.
They find that by taking the frenzy component and
downside risk term out, the shift in the income growth
component and the 1980s real interest rate effect become
“quite insignificant”. In conclusion they comment:
“These results suggest that the omission of a non-linear
‘frenzy’ effect is a major specification error. The omission
worsens the fit and fails to support the predictions of
economic theory regarding the consequences of financial
liberalisation, which are supported by a better specified
model including a ‘frenzy’ effect.”
Muellbauer, J., and A. Murphy, 1997
Frenzy effect…


But most mortgage bank analysts use models based
on efficient markets reasoning that ignores the
possibility of such effects.
This also explains why zero house price growth can
be serious news for a market that has recently
experienced ‘frenzy’. Price rises are no longer high
enough to draw many over the threshold and the
market grinds to a halt…before declining.
Survey evidence



So how do house buyers think? In ways likely to
make the market more or less easy to arbitrage?
The nearest thing we have to a controlled laboratory
experiment in housing was performed by Case and
Shiller* in the US on two boom cities, a post-boom
city and a control city with fairly constant
macroeconomic fundamentals.
Identical questionnaires of actual home buyers, and
found the following results.
*Case, K.E., and R. J. Shiller, 1988 (also found in Shiller, R.J.,
1989, p403-430).
Survey evidence – four cities

Four cities facing fairly similar macroeconomic
fundamentals:

Two boom cities, a post-boom city and a control city.



[Boom cities: Anaheim and California. Post boom city:
Boston. Control city (where no price changes in past five
years): Milwaukee]
Identical questionnaires of actual home buyers
Big issues were expectations about housing asset prices and
risk of capital loss.
Miles Report

Interesting reading alongside recent evidence on
assessments of interest rates and ultimate cost of
housing – including the ‘Miles Review’ of the UK
housing market (H.M. Treasury, December 2003):




House buyers overly-optimistic about future interest rates
Under-influenced by risks of future changes in interest
charge
Irrationally influenced by current-period cost of loan and
current period nominal interest rate deals
‘Miles Review’ describes people as openly ‘laughing’,
according to some studies, when asked to predict the cost
of loans by forming assessments of future interest rates
What triggered the booms?


There was no exogenous trigger for the housing price
booms.
In all four cities interest rate changes were cited as a
major factor. However, at the time, interest rates
were virtually identical. The same interest rates got
the credit for the boom in California and the blame
for the stagnation in Boston. Price changes could not
have been driven in all four cities by interest rates!
Knowledge of fundamentals

There was evidence of strong investment motivation
and higher price expectations in the boom cities than
the control city, but these were “expectations that they
could not show any ability to justify...Since most
people expressed a strong investment motive, one
would assume significant knowledge of underlying
market fundamentals. The efficient markets
hypothesis assumes that asset buyers make rational
decisions based on all available information and
based on a consistent model of underlying market
forces...The survey reveals little real knowledge of or,
agreement about, the underlying causes of price
movements. Rather then citing any concrete
evidence, people retreat into clichés....”
Local economy?


The second most-cited feature was a “strong local
economy.”
Yet “None...cited any specific evidence of such
strength or any detail about its character...people
look to observed price movements to form their
expectations and then look around for a logic to
explain and reinforce their beliefs.”
Quantitative evidence?



Many (in the boom city and post boom city, but not in
the control group) cited that there was “not enough
land”. But this is not news and cannot explain a
sudden boom.
“An especially striking feature.... not a single
respondent referred to explicit quantitative evidence
relevant to future supply or demand for housing... one
would expect some to volunteer such evidence if it
figured prominently in their views.”
People seem to form expectations on the basis of past
price movements rather than any knowledge of
fundamentals. Essentially, home buyers become
destabilising speculators.
Price feedback




This, and stock market studies, conclude that the feedback
of price changes on price dynamics is important.
In stock markets, the feedback is very rapid – a
downward price movement attracts attention leading to
further price declines within a day or so.
In real estate, price increases attract attention and
contribute to increasing prices over a year or so.
Asked whether price increases influenced their decision
to buy:





Boom cities: 90%
Control city: 84.8%
Post-boom city: 77.8%
25% overall expressed a fear of never being able to get
into the market.
65% in the boom market expressed this fear.
Attitude to risk



Willingness to pay for an asset is related to perceived risk.
“Very few of the home buyers in any of the four cities
thought that the housing market involved a great deal of
risk... (even where openly speculating about crash)”. The
degree of risk perceived was lowest in the boom markets.
“ Rising prices seem to dampen fears, and that may well
fuel the boom.” Practically all buyers in the two boom
markets and the vast majority in the post-boom and
control groups believed that prices were bound to
increase, reflecting the popular myth that “one cannot
lose in this market; houses are always a safe investment,
so long as one holds out long enough.”
More on risk

There is the same curious attitude towards risk in housing
markets as can be found in overvalued stock markets.
Recent research suggests that a recent period of good
times leads investors to have an extraordinary high
tolerance for risk, and accept relatively low expected
returns to holding stocks (Campbell and Cochrane 1999)
and hence equity valuations rise. Unfortunately, this
contrasts with survey evidence on investor expectations in
that they expect high returns on stocks in the future, not
the low returns that they supposedly accept because they
are especially risk-tolerant (Durell 99). The two things are
totally irreconcilable – one of them has to be give. As
Case and Shiller put it in housing markets: “Prices are
high because investors expect them to go even higher not
because they are ready for prices to go down.”
Residential yields
9.0
8.5
8.0
7.5
7.0
6.5
6.0
5.5
5.0
4.5
4.0
Q2 04
Q4 02
Q2 01
Q4 99
Q2 98
Q4 96
Q2 95
Q4 93
Q2 92
Q4 90
Q2 89
Q4 87
Q2 86
Q4 84
Q2 83
Q4 81
Q2 80
Q4 78
Q2 77
Q4 75
Q2 74
Q4 72
Q2 71
Q4 69
Q2 68
Net UK average residential yields (GSISG figures)
Risk premia
20.0
15.0
10.0
5.0
0.0
-5.0
Q3 04
Q1 03
Q3 01
Q1 00
Q3 98
Q1 97
Q3 95
Q1 94
Q3 92
Q1 91
Q3 89
Q1 88
Q3 86
Q1 85
Q3 83
Q1 82
Q3 80
Q1 79
Q3 77
Q1 76
Q3 74
Q1 73
Q3 71
Q1 70
Net U.K. Average Residential Rental Yield less real 20 Year Government
Bond Yield (GSISG calculation)
Thoughts on risky assets
Long run returns to a risky asset – stocks
Andrew Smithers and Stephen Wright, Valuing Wall
Protecting Wealth in Turbulent Markets, McGraw Hill 2000
Street,
Volatility of house prices
Volatility in real house prices HMT p41
‘Self-awareness’ of psychological
influences

In stock market surveys people are much more aware
of possible investor psychology explanations for price
rises, whereas “most participants in housing markets
do not attribute market events to psychology of other
investors” (Not surprisingly, the highest percent who
did, 18%, was in the post-boom city). “Perhaps we
should conclude that social psychology is an
important factor in transmission of [housing] booms,
but that individuals’ perceptions of the psychology of
others are less so.”
After the survey

Two years after the survey, a major bust started in the
boom cities, wiping 20+% off prices (the typical
down-payment of first-time home buyers in those
cities, 98% of whom had been convinced prices
would rise and 63% of whom had said they faced
little or no risk).
Consumption and House Prices


1970-mid 1990s, rapid
increase in house prices
(green line) accompanied
by rapid growth of
consumption (red line)
Recently, house price
inflation has accelerated
but the rate of growth of
consumption has steadied.
Source: Bank of England Inflation Report May 2004, p12, Chart 2.1.
A breakdown?




Misleading causation.
Both driven by income expectations? Not
directly observable.
Credit constraints?
A very different central bank response next
time?
More evidence: durables and
non-durables


Real durable and
semi-durables
consumption is
highly pro-cyclical.
But need to adjust
for rapid price falls
of durable goods
(global price falls,
etc.)…
Source: Bank of England Quarterly Bulletin, Spring 2004, J Power, Chart 1.
Nominal ratio of durable to nondurable consumption
Source: Bank of England Quarterly Bulletin Spring 2004, J Power, Chart 11.
Real house prices and share of
durable spending in consumption



Source: Bank of England Inflation Report November 2004
Durables more likely
purchased on credit.
Share previously
very correlated with
house prices.
Correlation has
broken down since
the late 1990s.
More Evidence


Swings – in all periods
– in spending relative to
income of homeowners
is nearly as great as
renters.
But: risk premia and
options thinking of
owners, etc.
Source: BOE Inflation Report May 2004, p12, Chart 2.3, based on the Family Expenditure Survey.
Possibilities


Credit constraints less important?
No upwards revision in expected future
earnings and wealth?
This would have driven higher housing demand
(and house prices) and higher desired stock of
durables.
 Flow increase in durables expenditure on the path
to new desired stock level.

Awkward Conclusions




If real income expectations are pretty stable, how
could these have driven house prices so much higher?
Not enough ability of demographic factors and the
slow rate of house build to explain house price rises.
Great deal of weight placed on a credit constraint
story for house price rises.
But difficult to create a consistent story if credit
constraints are highly important for housing
consumption but not for non-housing consumption,
especially durables.
Or…


Very high real house prices (and/or low interest rates)
are not regarded as long-term sustainable by the
general public?
Also some distributional issues – since rapidly rising
house prices (relatively) redistribute wealth from
asset-poor to already asset-rich, from young to old.
And this feeds subtle differences in aggregate
consumption behaviour.
Mortgage Equity Withdrawal




Source: Bank of England Inflation Report May 2004, p 13, Chart 2.4.
0% of household
income in the late 1990s
Now over 8% today
Yet, consumption as a
percent of household
disposable income has
hardly changed (1997today)
What if MEW were to
collapse this time?
Correlation between annual
house price inflation and annual
consumption growth
Source: Bank of England Inflation Report November 2004, p12 Chart B.

10 year rolling correlation coefficient has collapsed
Need to correct for…



Demutualisation of building societies. Windfall
payments (fungible with MEW) of £35 billion, or 7%
of annual consumption, helped the jump from 90% to
96% in two years with hardly any change in MEW.
Introduction of self-assessment.
The relationship is still weaker than in the past but
nevertheless is more positive than it first appears.
The big consumption story?



Levels v. stocks: The surge in the level of
consumption from 86% to 96% of household
disposable income created high rates of growth of
consumption in the 1980s.
Recent consumption has run consistently at a much
higher level for much of the last 7-8 years (with 3%5% consumption growth per year).
The big consumption story of the late 1990s and early
2000s is the historically high level of consumption
from disposable income, low levels of savings, and
deteriorating pension provision.
More on MEW




Insignificant in the UK
up to 1980.
1980s liberalization.
No other country in the
EU ever managed
anywhere near to 8%.
1979-1999 Germany,
France, and Italy net
injection of 6% of
household income into
housing.
Source: Bank of England, Office for National Statistics and HM Treasury calculations. Chart 5.5, HMT 2004 p52, not updated.
Key MEW findings





50%-60% is last-time sellers and those trading down,
i.e. those most likely to pay off debt and save.
A large proportion that is spent goes on ‘home
improvement’ and ‘new goods for the home’.
But, house prices and MEW are partly endogenous to
any price bubble (family transfers, home
improvement).
Borrowers who withdraw to spend are concentrated at
higher incomes, but a sizeable proportion are on low
incomes, and their borrowings are relatively highlevel.
Low levels of serial remortgaging.
MEW and housing market
Transactions


UK has one of the
highest rates of
housing market
transaction in the EU.
Transactions volumes
matter if most current
MEW is released
through last-time
sales and trading
down.
Source: HMT Table 5.4, p51. Source: Bank of England and Office for National Statistics.
MEW cont.




MEW is more lumpy than many credit constraint
stories suggest, which is why so much MEW is
immediately saved and then generates a later flow.
Much of the consumption flow from recent MEW is
still to come. More stabilizing?
If MEW plummets, there is no new flow into the stock
of assets to generate new consumption flow. Less
stabilizing?
In ‘turning’, or stagnant, housing markets, prices do not
fall heavily at first; transactions do. The liquidity of the
market and ability to ‘release’ equity falls.Those
wanting to trade-down are heavily dependent on chains
of buyers. Their ability to release MEW falls.
But…






Only element of MEW that seems not cyclical is
remortgaging (but evidence is hard to interpret).
For those forms of MEW that involve
borrowing…debt bites much more in a low inflation
environment.
What if those borrowing on MEW are excessively
buying into the house ‘price rise’?
If MEW replaces ‘more expensive’ forms of
debt…this dries up if house prices fall.
A small consumption response can still be magnified
by a big collapse in MEW.
If prices fall, there will be a collapse in MEW.
Demographics of MEW





Withdrawal of equity by last-time sellers is much
greater than injection by first-time sellers. Stock of
debt naturally rises over time.
If house prices are overvalued, those at the top have
been removing more equity than they would have
done in a less overvalued market, leaving behind on
average more indebted households.
Price bubbles are highly redistributive.
Price bubbles are popular with voters, and politicians.
Element of being a ‘Ponzi’ game.
MEW is not the main story
Savings and pensions are!




State of the housing market, level of savings, and
‘pensions crisis’ are linked.
Global liquidity story too.
If consumers believe that rapidly rising prices are
sustainable…then they may believe that current
consumption can be run at very much higher levels
than in the past.
Shocking evidence of unrealistic price expectations.


(Case, K.E., Quigley, M., and Shiller, R.J. 2004)
Slack taken from other parts of the household balance
sheet, and over-reliance on housing to generate future
pensions provision.
What about non-MEW…

If many of those withdrawing housing equity are not
immediately spending, what are non-MEW
consumers doing to maintain consumption
consistently at 96% or more of income?




Low savings;
Eating into pensions contributions they ‘should’ be making;
Non-MEW forms of debt;
Relying on house prices? Some notion of ‘asset value
allusion’. Housing wealth fungible with other forms of
savings? The cheapest form of credit?
Saving ratio


Shows UK households’
saving ratio and house
prices (plotted with
negative of saving ratio)
Paradox of thrift too in a
major correction.
Source: HMT 2004, Chart 6.3. p. 60. Source: Office for National Statistics and Office of the Deputy Prime Minister.
Saving ratio cont.

Source: HM Treasury, 2004, p64.
Correlation
between saving
ratio and house
price inflation one
period earlier
Impact of interest rates on house
prices and consumption




EITHER house prices are less volatile than in the
past…
OR, if they are still just as volatile, the links to
consumption are weaker…
OR, interest rates can cope with any problems (ease
cash-flow problems, cushion/slow house price falls,
etc.)
Direct and Indirect effects of interest rate changes via
housing market:


Both have forces working for and against.
Both are difficult to be precise.
Conclusions on direct and
indirect effect




The direct effect of reducing interest rates after the last
house price crash was large. Likely more modest today.
The Bank of England should worry less this time when
raising (or holding high) interest rates, but also feel less
confident of the power of rate cuts to make a big impact
on aggregate cash-flow.
Indirect effect stronger, but not much interest rates can do
if large speculative aspect to house prices.
But many caveats, including:



Distribution and fragility of debt is not fully clear. Caution;
Credit conditions generally;
Dangers of moral hazard if perceived to be ‘bailing out’.
Redistribution effects of
falling/rising house prices



Rapid house price rises have similar consequences to
sustained budget deficits, depressing the current real
productive capital stock in exchange for current
consumption by the gainers.
House price booms, just like government budget
deficits, are popular with older consumers (and
younger bubble-motivated consumer who ‘know no
better’, or suffer asset price allusion).
Like deficits, older consumers benefit from the
‘borrowing’ from future generations, even as longterm income levels are reduced due to the crowding
out of real productive capital stock.
Distributional issues suppressing
long-run values

Depressing influences on the investment potential of
housing of the current generation (since the market
for the value of their homes is the next, smaller, more
burdened generation) include:





demographically aging population;
falling cohorts in younger generations;
record low savings;
deteriorating provision for retirement;
the burden of social security and health increasing over
time.
Rates of return…


Given current valuations, the real rate of return on
housing for, say, the next 20 years is much lower than
the historical average of about 2.5% (indeed, zero is
within the 95% confidence interval for the 20 year
real rate of return).
It also suggests that if correction is inevitable, it is not
obvious that correction should be resisted.
Global House price correlations
and global liquidity


Highly synchronized positive movements in house
prices, and global build up in mortgage debt.
An extremely recent phenomenon that does not affect
all countries equally:




US, Australia, UK, China, France, Ireland, New
Zealand, South Africa;
Relatively few EU countries.
The correlation between real house prices and output
(and consumption) has declined since the mid-1990s,
reaching unprecedented low levels by 2003.
Potential for instability from outside UK

Hence issues of timing, interest rate decisions, etc.
Variance decomposition of house
prices
Source: IMF 2004, data from: Haver Analytics; IMF, International Financial Statistics; national sources;
OECD; and IMF staff calculations.
UK case

House Factor and Global Factor over time, per
cent change, constant prices, demeaned
Source: IMF 2004, data from: Haver Analytics; IMF, International Financial Statistics; national sources; OECD;
and IMF staff calculations.
UK decomposition of house price
change over time




Source: Extracted from IMF 2004
Orange = Actual
Blue = Global Housing
Factor
Red = Global Factor
Black = Country Factor
Global Liquidity





Global driving forces – especially US house prices, US
interest rates, and global liquidity.
25% per year growth in the sum of America’s cash and
banks’ reserves held at the Fed, and in the foreign
reserve holdings of central banks around the world.
Excess liquidity in the past flowed into traditional
measures of inflation (goods and service prices).
Does extreme liquidity show up in asset price inflation
– house prices – and record high global levels of debt,
especially mortgage debt? Overreaction of housing
markets after stock market crashes?
US housing stock has risen in paper value by $5
trillion, almost precisely matching the $5trillion of lost
stock market wealth of the early 2000s.
Did Global interest rates go too
low?



Natural rate of interest: the rate at which the supply of
savings of households exactly balances the demand for
funds by firms for investment purposes.
Natural rate is roughly equal to the rate of inflation plus
the real trend rate of growth.
Natural rate moves about according to:




technological improvement;
changes in preferences;
the impact of demographics on the need for savings, etc.
If cost of capital set below this, we get
overcapitalization, excessive levels of borrowing and
investment, saving too low, and the chances of bubbles
greater.
Natural rate too low, continued





For UK about 5% (2% inflation plus 2% to 3%).
Global natural rate (difficult to work out precisely)
may even have risen (China, IT, global integration,
inflation success, etc.).
Yet, some of these forces have also lowered inflation
and even encouraged lower interest rates.
And bubbles encourage households not to save,
making things worse.
Low rates followed collapse of late 1990s and various
other crises/collapsing bubbles.
The US





US (especially) interest rate kept below ‘natural rate’
for too long? 3% to 5% too low? Maybe after
previous bubbles? Went from 6.5% in 2001 to 1% in
2003.
Past five years America’s national spending has
exceeded its income by about a fifth.
Private debt has boomed (nearly $10trillion).
Savings have hit 0.5% (compared to historical
average of 8%). Sometime, a reversion back to 8%?
Private debt service is historically high – even before
interest rates rise.
US cont.




US government has joined US citizens. Approx
$450bn budget deficit per year.
Externally held portion of debt risen from 20% to
45%.
Problems with ‘depth’ of US mortgage markets.
Problems with US ‘lender of last resort’.
US cont.



If a country has strongly favourable investment
opportunities – that will ultimately make its
inhabitants much better off – it is economically
rational to consume some of the fruits now, borrow
from the rest of the world, and repay from higher
output later.
Meanwhile, run a strong currency and high trade and
current account deficits that generate a net capital
public and private consumption, rather than
investment flow equal to the current account deficit.
However, decomposition of US data shows that debt
is currently largely being used to finance.
Asia and US mutually reinforce





Global foreign exchange reserves have doubled since
the Asian crisis of 1998, to $3,800bn, with two-thirds of
this in dollars. Asia accounts for 80% of this growth and
now has 70% of global reserves.
Reserves account for 9% of global gross domestic
product, compared to less than 2% during the pre-1971
Bretton Woods era.
High demand for US debt may be a response to previous
crises. It has chipped 0.5% to 1% off US yields. It has
helped the US to run large government and trade
deficits.
China has chipped 0.1% to 0.3% of US inflation (maybe
as much as 1%).
In turn, China’s boom has been helped by low US rates.
House price contagion?

If sufficiently strong house price falls in one country
(for example, rebalancing in the US will require
lower spending on housing consumption) or several
countries generate a decline in consumption for them,
then it is more likely that consumption will fall in
other countries too:




Real contagion (via consumption).
Financial contagion (via, in particular, mortgage bank
and government balance sheets).
Equity-based bubbles less damaging than debt-based
bubbles. Has mitigating the first, led to the latter?
See Farlow, “UK House Prices, Consumption, and
GDP in a Global Context,” Section 6, for scenarios.
Impact on UK policy



How do these possibilities affect our attitude to
falling house prices in the UK and central bank policy
on interest rates?
If rebalancing of the UK housing market is inevitable,
this suggests allowing rebalancing to proceed sooner
rather than later – leaving the housing market in a
better position to withstand global disturbances
consequent on rebalancing elsewhere.
Reversion to fundamentals, while it harms
consumption, at least conceivably puts the economy
back on a footing that emphasizes real economic
activity over speculative housing activity and ends the
distortions that lead to long-term pension and saving
misallocation.
Lessons for the UK



Global instability might matter more for UK house
prices than is perhaps currently accepted. The fate of
the UK housing market may currently be one of the
less domestically controllable aspects of UK
macroeconomics.
The Bank of England has faced an unenviable choice
between trying to turn the tide of house prices and not
sacrificing growth and risking under-hitting its
inflation target.
Emphasis on controlling the housing market has to
include reforms and not just rely on interest rates.
Lessons for UK cont.



Interest rates may have little power to influence house
prices in a price collapse anyway. Trying to generate
a cash-flow affect to offset a wealth effect caused by
an unwinding bubble may simply not work very well.
Handling global surges in house prices might need
more of a coordinated response than it probably gets,
or is ever likely to get.
Stop-go cycle via house prices in place of stop-go
cycle via goods and services prices?
Asset price bubbles confuse inflation
signals

In fact recent asset price bubbles might have helped to
dampen up-front inflationary pressures, and led to confusing
signals. In case of equity market bubbles:
 Artificially boost profits as measured in standard accountancy
measures – firms adopt more aggressive pricing strategies;
 Positive feedback via capital accumulation and favourable
supply-side developments, especially productivity gains, the
spreading of technology, and ‘catch up’ in emerging
economies (c.f. China), with consequent lower inflationary
pressure;
 Firms able to make much lower contributions to pension
schemes (and employees willingly accept, c.f. US 1990s);
 Employees tolerate less inflationary wage claims given
perceived gains on stock market investments (especially
evident in the US in the late 1990s);
Bubbles confuse signals, cont.

Governments benefit from bubble-inflated asset
prices that inflate the tax yield (stock market taxes at
the end of the 1990s, housing transactions taxes,
consumption taxes, low use of pension tax
allowances, etc. in current housing bubble) and allow
them to run lower tax rates than otherwise would be
the case, even as their fiscal positions are
strengthened;

When the bubbles unwind, all these things go into
reverse – just at the wrong time.
A public sector generated cashflow shock?




There are dangers in running a fiscal policy that risks
a sudden tax increase in the face of a declining
housing market (maybe at the same time as a swing in
global credit conditions).
This time, the impact may show up more than usually
as a public sector cash-flow problem.
But burdens for the public sector are ultimately
private sector burdens. It is simply an issue of timing.
If house price falls are slow enough, this cash-flow
problem can be offloaded to the private sector at a
more timely pace.
Public sector cash-flow, cont.



If price falls are more rapid (and confidence takes a
greater hit) then the cash-flow difficulty (or even just
expectations of it) may be fed much more quickly to
the private sector – especially in the form of higher
taxes – reinforcing the private sectors cash-flow
problem and putting downward pressure on house
prices. The economy finds itself on an even higher
tax trajectory at quite the least opportune time for it.
And tax rises are hard to target on the relatively less
indebted (unlike interest rate falls).
Growing tension between Bank of England and
Treasury.
Precarious balancing act for
central banks





If rates are raised too slowly, any momentum-driven
bubbles will expand even further, financial imbalances
intensify, with inflationary pressures built up for the
longer term. Adjustment is simply delayed till a point
when the fragility is even greater and when the dangers
of triggering a switch to deflation are higher.
If rates raised too quickly, fragilities may unwind too fast.
Balancing act for debt holders too.
This time real price falls will need nominal price falls.
This time the Bank of England would not allow the runaway inflation needed to create the negative real interest
rates that would generate a similar situation today to that
easing previous price collapses.
Central bank challenges, cont.

Interest rates may have too much to do and be unable
to fall as far as the housing market might require:






If the pound were to weaken too quickly;
OR oil or commodities prices rise too strongly, feeding
UK inflationary pressures;
OR the US were to suffer a fiscal crisis and sudden
swing in sentiment forcing it to raise real interest rates
much higher in defense;
OR UK government finances were to deteriorate
quickly as lower confidence and economic activity
reduced the tax yield.
Rising spreads may keep interest rates off the floor.
System has not been stress-tested yet. Best to set so as
to minimize the fall-out from mistakes.
Central banks cont.




Policy in the late 1990s settled on price stability via
interest rate adjustments and not tax adjustment, since
easier to set up an institution independent of elected
government.
Danger that governments rely on – even over-exploit
– the policy credibility created by central banks.
A slow revision upwards of interest rate expectations
is better than a sudden unexpected tax rise. Gradual
reversions without surprises are best.
Openness about interest rates – and at least the chance
for households to think the scenarios through and
factor higher interest rates in – contrasts sharply with
the complete lack of openness about (and the political
nature of) the timing and level of future tax rises.
A Few Summary Thoughts





Global rebalancing at some point.
Expected consumption and GDP response will matter.
Much of the analysis of the consumption response is
based on efficiently operating bubble-free markets.
Once we allow bubble mispricing, consumption
responses are generally less well captured.
Soft landing is not completely certain when a broad
picture of consumption is considered, including
pension and savings, and when levels as well as
changes in levels are reviewed.
Continued on next slide…
A few summary thoughts cont.





Debt-backed housing price bubbles probably more
damaging than stock market bubbles.
The new stop-go cycle?
Politicians love to exploit price bubbles.
Aftermath of current period likely to focus attention
on a wide range of issues. Many of these require
coordination across diverse economies, something
usually achieved more easily in stable times. In less
stable times, something much easier to handle if some
thought has gone in ahead of time.
Having better educated citizens helps. Caricature of
‘optimists’ or ‘doomsters’, ‘experts’ or ‘pundits’ does
not help.
THIS IS THE LAST SLIDE