| ||||||||||||||||||||||||||||||||||||||||||||||||||||
This article appears in the following Oxford Review of Economic Policy issue: HOUSING MARKETS AND THE ECONOMY [View the issue table of contents]
House prices, money, credit, and the macroeconomy
* Financial Markets Group, London School of Economics, e-mail: caegoodhart{at}aol.com
** European Central Bank, e-mail: boris.hofmann{at}ecb.int
 |
Abstract |
|---|
This paper assesses the links between money, credit, house prices, and economic activity in industrialized countries over the last three decades. The analysis is based on a fixed-effects panel vector autoregression, estimated using quarterly data for 17 industrialized countries spanning the period 1970–2006. The main results of the analysis are the following. (i) There is evidence of a significant multidirectional link between house prices, monetary variables, and the macroeconomy. (ii) The link between house prices and monetary variables is found to be stronger over a more recent sub-sample from 1985 to 2006. (iii) The effects of shocks to money and credit are found to be stronger when house prices are booming.
Key Words: house prices • wealth effects • collateral • financial liberalization • money and credit
We would like to thank John Muellbauer, Massimo Rostagno, Frank Smets, and participants in the Oxford Review of Economic Policy Housing Markets Seminar at Saïd Business School on 12 September 2007, especially our discussant Simon Price, for helpful comments and suggestions. The views expressed in this paper do not necessarily reflect the views of the ECB or the Eurosystem.
1 Besides the housing wealth effect on money demand, there might further be a transactions effect, arising generally from higher demand for transactions balances when wealth increases, and specifically from higher transactions related to house purchase when house prices rise.
2 Aoki et al. (2004) and Iacoviello (2004, 2005) show, based on general equilibrium models, that a financial accelerator effect arises in the household sector via house prices, when households’ ability to borrow depends on the value of housing collateral.
3 For a more detailed exposition of the wealth and collateral effect of house prices on consumption, see Muellbauer (2007).
4 Bernanke and Gertler (1989), Kiyotaki and Moore (1997), and Bernanke et al. (1999) have developed modified real business cycle models wherein firms’ borrowing capacity depends on their collateralizable net worth, and show that fluctuations in firms’ net worth amplify macroeconomic shocks and can give rise to a powerful financial accelerator effect.
5 For example, Chen (2001) develops a general equilibrium model in which both borrowers’ and banks’ net worth influences the supply of credit. Just as borrowers’ net worth acts as an incentive mechanism and collateral for the banks, bank capital acts in these models as an incentive mechanism and as collateral for the bank's providers of loanable funds, e.g. depositors. So, the availability of loanable funds to banks depends on their capitalization.
6 Borio et al. (1994) investigate the relationship between credit-to-GDP ratios and aggregate asset prices for a large sample of industrialized countries. They find that adding the credit-to-GDP ratio to an asset pricing equation helps to improve the fit of this equation in most countries. Based on simulations, they demonstrate that the boom–bust cycle in asset markets of the late 1980s to the early 1990s would have been much less pronounced or would not have occurred at all had credit ratios remained constant. Goodhart (1995) investigates the effect of property prices on bank lending in the UK and the USA using long spans of historical data, and finds that property prices significantly affect credit growth in the UK but not in the USA. Hofmann (2004) analyses the role of property prices in explaining credit dynamics in industrialized countries since 1980. He finds that property prices are an important determinant of the long-term trend development in credit over this period and that increases in property prices have a highly significant positive effect on credit dynamics.
7 Hofmann (2003), Goodhart and Hofmann (2004a), and Goodhart et al. (2006) analyse the relationship between bank lending and property prices based on a multivariate empirical framework and find that causality does, in fact, seem to go in both directions, but that the effect of property prices on credit appears to be stronger than the effect of credit on property prices. Gerlach and Peng (2005) analyse the link between property prices and credit in Hong Kong and find that causality runs from property prices to lending, rather than conversely. Greiber and Setzer (2007) investigate the link between broad money and property prices in the USA and the euro area. They find that adding property prices to an otherwise standard money demand system restores a stable money demand equation in both economies. Based on a standard impulse–response analysis, they further show that causality runs in both directions: an increase in broad money growth triggers an increase in property prices and vice versa.
8 Gouteron and Szpiro (2005) investigate the effect of excess liquidity, measured by the ratio of broad money to GDP and, alternatively, the ratio of private credit to GDP, in the USA, the euro area, the UK, and Japan, but fail to detect any significant links except for the UK. Adalid and Detken (2007) explore the effect of broad money growth on house prices in a panel of industrialized countries and find that the link is significant and particularly strong in times of aggregate asset-price booms. They further find that private credit growth does not have a significant effect on house-price dynamics.
9 In a similar vein, Muellbauer and Murphy (1989), and more recently Muellbauer (2007), have argued that the housing collateral effect on consumption will be stronger when credit markets are liberalized.
10 See Borio and Lowe (2004), Detken and Smets (2004), and Adalid and Detken (2007).
11 For example, The Economist (2006) recently stated that ‘(t)his link between money and asset prices is why the ECB's twin-pillar framework may be one of the best ways for central banks to deal with asset prices’. See also Mayer (2005), who characterizes the ECB's strategy as providing ‘a bridge between inflation targeting and a new paradigm which takes account of financial and asset market developments in monetary policy decisions’.
12 We used the CPI rather than alternative measures of the aggregate price level, such as the GDP deflator or the consumption deflator, mainly for the reason that central banks’ inflation targets or objectives usually refer to some kind of consumer price index. A drawback of using the CPI is that there are occasional changes in methodology, for example in the USA with regard to the measurement of home-ownership costs in 1983.
13 A short-term money-market rate was for most countries not available for the full sample period.
14 For the euro-area countries, there is a credit series in national currency until 1998Q4 and a series in euros from 1999Q1. Even after converting the national currency series to euros based on the irrevocable fixed exchange rates, some of the spliced credit series still displayed level shifts in 1999Q1. For this reason we performed a level-shift adjustment in this quarter for all euro-area countries.
15 The following level shifts were adjusted for: Australia 1989Q1, 2002Q1; Belgium 1992Q4, 1999Q1; Canada 2001Q4; Denmark 1987Q4, 1991Q1, 2000Q3; Finland 1999Q1; France 1978Q1, 1999Q1; Germany 1990Q2, 1999Q1; Italy 1999Q1; Ireland 1982Q4, 1995Q1, 1999Q1; Japan 1997Q4, 2001Q4; Netherlands 1982Q4, 1999Q1; Norway 1976Q1; Spain 1983Q1, 1986Q1, 1999Q1; Sweden 1983Q1, 1996Q1; Switzerland 1974Q4, 1982Q3, 1996Q4; USA 2001Q4.
16 Belgium: a missing observation for 1998Q4 was generated using the growth rate of a series for bank lending to the private sector taken from the BIS database. France: a missing observation for 1977Q4 was generated based on the growth rate of a series from the BIS Database named ‘Credit of a banking character to the economy’. Netherlands: missing observations for 1998Q1–1998Q4 were generated with the growth rates of a series for claims of monetary institutions on the private sector taken from the BIS Database. Norway: missing observations in 1987Q1–Q2 were generated from the growth rate of an IMF series for credit extended by non-bank financial institutions to the private sector (IMF IFS series code 42D). Sweden: missing observations in 2001Q1–Q3 were generated from the growth rate of a series for bank lending to the private sector from the Riksbank's website.
17 For the panel of 17 countries, the cross-correlation of the growth rates of nominal broad money and nominal bank credit is 0.56 for the year-on-year growth rates and 0.38 for the quarterly growth rates.
18 The results of the individual country analysis are available upon request.
19 Gavin and Theodorou (2005) find in their empirical application that, while the pooling restrictions are generally rejected in-sample, the panel model performs significantly better than the individual country model in out-of-sample forecasting.
20 As an alternative to instrument-based estimators, Kiviet (1995) has proposed a bias-corrected FE estimator which is based on a two-step procedure making use of the analytical bias expressions derived by Nickell (1981). Because of the problems associated with the practical implementation of this approach it is, however, almost never used in empirical applications.
21 The results of the fixed-effects panel VAR estimation with a full set of time dummies included are also available upon request.
22 The response of real house prices and real money and credit is given by the difference between the response of the respective nominal values and the response of the CPI. For example, if the CPI increases by more after a shock than nominal house prices, then real house prices fall.
23 The phenomenon of a positive response of the CPI to an interest-rate increase is known as the price puzzle and is attributed to forward-looking monetary policy, so that the impulse response captures in part also the reaction of monetary policy to expected future inflation.
24 These results are consistent with evidence presented by Ludwig and Sløk (2004) and Muellbauer (2007), suggesting that the effect of house prices on consumption has become stronger since the mid-1980s.
25 In a recent paper, Adalid and Detken (2007) have shown, also using a panel framework, that broad money growth has a particularly strong influence on real house-price growth when there is an aggregate asset-price boom, where the aggregate asset price is a BIS construct aggregating share prices, residential property prices, and commercial property prices weighted by their respective share in household wealth.
26 These results are broadly consistent with those reported by Adalid and Detken (2007). However, while they find that only money growth influences future house prices when there is an asset-price boom, we find that both money and credit matter. There are three potential explanations for this discrepancy. First, Adalid and Detken (2007) focus on the effect of monetary variables on real house prices in a single-equation framework, while our analysis is based on a multivariate framework. Second, Adalid and Detken investigate the link between house prices and money and credit during aggregate asset-price booms (see footnote 25), while we investigate the dynamics during house-price booms. Finally, there is a difference in sample periods. The sample period in Adalid and Detken is 1972–2004, while here it is 1985–2006.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  J. Muellbauer and A. Murphy Housing markets and the economy: the assessment Oxf. Rev. Econ. Policy, March 1, 2008; 24(1): 1 - 33. [Abstract] [Full Text] [PDF]
|
|