| ||||||||||||||||||||||||||||||||||||||||||||||||||||
This article appears in the following Oxford Review of Economic Policy issue: HOUSING MARKETS AND THE ECONOMY [View the issue table of contents]
Housing markets and the economy: the assessment
* Nuffield College, Oxford, e-mail: john.muellbauer{at}nuffield.ox.ac.uk
** Hertford College, Oxford, e-mail: anthony.murphy{at}hertford.ox.ac.uk
 |
Abstract |
|---|
Housing markets have multiple interactions with the rest of the economy and these are surveyed in this paper. The drivers of house prices include income, the housing stock, demography, credit availability, interest rates, and lagged appreciation, the latter a potential mechanism for overshooting. There is rather less agreement on the determinants of new construction, though planning constraints are widely seen as a major issue and one of the causes of the UK housing affordability problem. The paper argues that housing collateral and downpayment constraints are the key to understanding the role of house-price variations in explaining medium-term consumption fluctuations. Institutional variations between countries and over time account for major differences in linkages between house prices and economic activity. This illuminates debates about how monetary and other policy should react to house-price variations. The paper also discusses the role of housing markets in explaining regional migration and location decisions, intergenerational inequality, and restricting access of the less affluent to public goods, such as good schools, which are capitalized in local house prices.
Key Words: house prices • consumption • monetary policy • credit conditions • institutional change
This special issue of the Oxford Review of Economic Policy was jointly conceived by Gavin Cameron, John Muellbauer, and Anthony Murphy. Gavin's death on 9 September 2007 deprived us of a deeply valued friend and colleague. We dedicate this issue of the Review to his memory. Financial support from Cemex for the one-day seminar held on 12 September 2007 is gratefully acknowledged. We are grateful for comments from Tim Jenkinson and Steve Nickell, but take full responsibility for views expressed here.
1 The exceptions were the 2001 recession caused by the internet bubble collapse, and the 1953 recession after the end of the Korean war. The longer history of such fluctuations is discussed by Abramovitz (1964).
2 Indeed, the word ‘bubble’ for many economists is like a red rag to a bull. Many regard the rational bubble, as explained, for example, by Blanchard and Watson (1982), as the definition of a bubble. Many others, however, reject the idea that all agents are rational and share the same beliefs. Then there can be persistence in high valuations if optimists have sufficient wealth or access to credit, while pessimists are constrained by the inability to ‘short’ the asset (see Harrison and Kreps (1978) and Scheinkman and Xiong (2003) on the role of short-sale constraints). These are even more powerful for housing than for shares, see Stein (1995) for similar explanations for the inefficiency of housing markets. Models with ‘noise’ or momentum traders coexisting with ‘fundamentalists’ (see DeLong et al., 1990) take a similar ‘non-rational’ view of mispricing. In such models, the word ‘bubble’ has a somewhat looser meaning.
3 See DiMartino and Duca (2007), Green and Wachter (2007), Mian and Sufi (2008), and Keys et al. (2008), inter alia.
4 In the upswing, consumer spending and bank profits expand in part because of the rise in house prices, boosting growth and the willingness of lenders to advance funds, while in a downswing, the opposite holds, with bad loans constraining the willingness to lend.
5 Inverse demand functions have a long history, particularly in the analysis of markets for natural resources. Theil (1976) refers to a 1909 Danish study as the first empirical study of inverse demand functions.
6 An alternative is the limited rationality approach advocated by Demery and Duck (2007), in which it is assumed that households have access to a limited range of information. Then a model for 
is estimated using this information set which satisfies the usual econometric criteria of parameter stability, good fit, etc. The fitted value from the model can then be taken as a proxy for expected appreciation. Muellbauer and Murphy (1997) use this approach.
7 Comprehensive reviews of the regional house price literature for the UK can be found in Muellbauer and Murphy (1994) and Meen and Andrew (1998), as well as CMM (2006). The model fulfils eight criteria for a satisfactory model set out by the last of these, including: the model is data consistent; incorporates spatial lags and errors; has some spatial coefficient heterogeneity; has a plausible long-run solution; includes a full range of explanatory variables.
8 This matches the empirical findings in the model of debt in Fernandez-Corugedo and Muellbauer (2006) and the UK consumption functions in Aron et al. (2007).
9 The CCI was estimated by Fernandez-Corugedo and Muellbauer (2006) from data on ten consumer-credit and mortgage-market indicators. It is intended to measure the shift in the credit-supply function to UK households, especially since 1980. See section V(iii).
10 Note that the estimated effects are merely the products of estimated coefficients and explanatory variables and not a variance decomposition or stochastic simulation.
11 In London, this was the result both of higher per-capita income growth and of population growth, driven by net foreign immigration. Since 2002 or so, the net change in population has altered, with net outflows from London to other regions partly offsetting immigration.
12 As we have seen, the expansion in credit supply conditions in the USA from 2000 to 2005 might have been regarded as a permanent shift in 2005, but from the perspective of 2008 looks unsustainable.
13 In the USA, Himmelberg et al. (2005) also argue that conventional metrics, such as house-price-to-rent ratios, are misleading because they fail to account for long-run trends in real interest rates that have made housing rather more affordable.
14 For example, Ayuso and Restoy (2006) suggest that UK house-price-to-rent ratios were about 20 per cent above their equilibrium value in 2002. Weeken (2004), whose results imply that house-price-to-rent ratios were only a few percentage points above their equilibrium level in 2002, suggests that ‘because of data and model limitations, no firm conclusions can be drawn’.
15 CMM (2006) discuss a third strand of the housing bubbles literature which is more technical and involves estimating switching models. See Roche (2001) and Garino and Sarno (2004), for example.
16 A test of this type with a mis-specified model can give different results. The IMF (2005) estimates a dynamic, error-correction equation for log real house prices as a function of log real disposable income per household and real interest rates only. The resulting equation is not a reduced form, since no supply-side variables, such as construction costs, are included. It is also rather unstable since supply, the changing age structure of the population, nominal interest rates, and shifts in UK credit conditions play no role. Nevertheless, the IMF (2005) suggests that, on the basis of its model simulations, UK house prices were overvalued by 30 per cent or more in 2003.
17 Some of the fall in observed loan-to-value and loan-to-income ratios will, however, be due to higher rates of mortgage repossessions and negative returns in housing, as implied by Fernandez-Corugedo and Muellbauer (2006).
18 This is not meant to imply that other asset wealth is unimportant. However, falls in prices of equity as well as of houses and higher unemployment will increase the fraction of households in the vulnerable tail of the distribution of debt.
19 See Maclennan et al. (1998, 2000).
20 Quigley provides a useful review of the US literature, while Saiz (2008) is a good illustration of the sophisticated econometric modelling that can be undertaken in this area. Inter alia, Saiz uses GIS (Geographic Information System) data to give empirical content to the concept of land abundance and models the regulatory component of housing supply.
21 On the assumption that between 1987 and 2007 ‘excess’ house price appreciation was 1.9 per cent per annum, Weale (2007) argues that this appreciation was roughly equivalent to a government deficit per annum over these years of 4 per cent of GDP.
22 It is important that assets not appear in the widely used log form, which gives poor approximations for low levels of assets and breaks down for negative net worth.
23 This effect can arise from Campbell and Mankiw's (1990) aggregation of current income-constrained and -unconstrained households (see Muellbauer and Lattimore, 1995).
24 In further work, not reported in Muellbauer (2007), there are indications that the long-run effect is rather less than 0.07–0.09 but larger than the UK effect of 0.032, when capital gains in housing are included, the latter having proved insignificant for the UK. There is also some evidence for a short-term asymmetry: falls in house prices appear to change consumption less than do rises.
25 Martin Wolf in the Financial Times of 5 February 2008 discusses the particular problems of the banking system and credit markets and their wider economic importance, and reviews recent debates about the reform agenda with great insight.
26 See Hughes and McCormick (1987, 2000), McCormick (1997), and Oswald (1997), inter alia.
27 For example, De Graff and van Leuvensteijn (2007) suggest that the macroeconomic findings in Oswald (1997), Nickell (1998), and Green and Hendershott (2001) that home ownership constrains labour mobility and thus increases unemployment is due to the confounding effect of transactions costs.
28 See the literature review in MMC (2006), whose main conclusion is that empirical studies of inter-regional migration that have reasonable time variation in the data, and include fixed effects and relevant housing market variables, generally find significant and plausibly signed coefficients on the latter. Contiguity effects and the commuting–migration trade-off are important.
29 The migration rates are scaled by the overall rate of regional migration.
30 The labour- and housing-market variables are formulated so that there is an own-region effect relative to Great Britain as a whole, and a contiguous-region effect. The way the ‘own region’ and ‘contiguous region’ effects are combined varies for labour-market and housing-market variables, since the migration and commuting implications of these two types of variables differ. Relatively good labour-market propects (high earnings and low unemployment, etc.) in a region tend to increase both migration and commuting into that region. However, in the case of housing, relatively good prospects (low house prices, high expected capital gains, etc.) tend to raise migration and reduce commuting into the region.
31 For example, see Stein (1995) for reasons to do with credit and short sale constraints and lack of ‘deep pockets’ by traders.
32 For example, if building more houses or locating a business at the edge of a town or village reduces property prices, those harmed by the decision would be partly compensated by lower taxes. Similarly, building a good school would raise prices and the higher taxes would automatically help fund the public expenditure involved.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  S. Price Discussion of 'House prices, money, credit, and the macroeconomy' by Charles Goodhart and Boris Hofmann Oxf. Rev. Econ. Policy, March 1, 2008; 24(1): 206 - 209. [Abstract] [Full Text] [PDF]
|
|