Policymakers around the world often worry about decreases in real-estate prices and other asset prices, and take measures to prevent them. For example, in the aftermath of the financial crisis, the Federal Reserve has engaged in large-scale asset purchases – especially of mortgage-backed assets – to support the housing market and, in turn, the overall economy.
The reasoning behind this policy stems from a large literature in financial economics and macroeconomics that studies the positive effects of increases in asset prices on productive real investment.1 The idea is that firms can borrow and invest more when their assets are worth more, because they can provide more valuable collateral, and also because they have more ‘skin in the game’, which reduces moral-hazard concerns. This is the so-called 'balance-sheet channel’. There is also a ‘lending channel’ operating through the banking system – when assets on a bank’s balance sheet are worth more, the bank should be able to lend more, thus amplifying a positive initial asset-price shock.2 Finally, an increase in the value of consumers’ assets creates a positive wealth effect, leading to higher demand and higher consumption. This demand effect has a further positive impact on productive real investment, as firms respond to higher consumer demand with more investment.
Over the years, empirical evidence has been provided to support an optimistic view about housing-price appreciations. Recently, Chaney et al. (2012) have shown that firms are able to borrow and invest more when the value of their real-estate collateral increases. Others have shown that a decrease in housing prices hurts banks’ balance sheets, leading to a decrease in real investment.3
The crowding-out effect of housing-price appreciation
Are housing price appreciations always desirable for the real economy? Unfortunately, the answer to this question is negative. As discussed in the theory of rational bubbles, an increase in housing market activity may crowd out commercial and industrial lending through increased interest rates. As a result, one sector of the economy that is receiving liquidity and experiencing bubbles may overheat, and crowd out other sectors of the economy.4
In our new paper, we provide the first empirical evidence on the crowding out effect of housing price appreciation (Chakraborty et al. 2013). Looking at the period from 1988 to 2006, we find that firms that borrowed from banks located in stronger housing markets paid higher interest rates, received lower loan amounts, and ultimately invested less compared to similar firms that borrowed from banks located in weaker housing markets.
The normative implications for the economy are significant – if monetary policymakers are actively supporting one sector of the economy, such as the housing market, they are causing a detrimental effect for other productive sectors.
To begin the empirical analysis, we need to find a proxy for banks’ exposure to real estate prices – not an easy task, given the opacity of banks’ balance sheets. We obtain our proxy using data on the geography of banks’ branches. By weighting the housing prices in each state with bank branch deposit data, we generate a bank-specific housing-price exposure variable.
Figures 1 and 2 show the relationships between bank assets and housing prices. Figure 1 shows that as housing prices increase, banks on average invest more in real estate-related loans. However, Figure 2 shows that as housing prices increase, banks on average reduce their commercial and industrial loan portfolios as a percentage of assets. It appears that in reaction to housing-price appreciations, banks increase real estate lending and decrease commercial lending.
Figure 1. Housing prices and real estate loans
Figure 2. Housing prices and commercial and industrial loans
This decrease in commercial lending is interesting if it is caused by a reduction in the supply of capital, as opposed to a reduction in firms’ demand for capital. We estimate that a one standard-deviation increase in housing prices (about $79,700 in year 2000 dollars) that a bank is exposed to decreases investment by firms related to that bank by almost 6.3 percentage points, which is approximately 12% of a standard deviation for firm investment. Banks also increase the interest rate charged by 9 basis points, reduce outstanding loans by approximately 9%, and reduce loan size by approximately 4.5%. These results are consistent with banks reducing the supply of capital to firms in response to increased housing prices.
An important endogeneity concern is that housing-price changes may be picking up unobserved changes in economic growth prospects in a certain area. If this bias is present, it should weaken our estimates – an unobserved positive economic shock will both increase housing prices and firm investment opportunities, giving a positive bias to the relation between bank housing prices and firm investment. To address this concern more rigorously, we instrument the banks’ housing price exposure using the percentage of land unavailable for development in the bank’s area, the state-level mortgage interest rate, and their interaction.5 These instruments are relevant because for a given increase in housing demand (potentially driven by a drop in mortgage interest rates), areas with less developable land should see larger changes in housing prices. Importantly, geographic constraints to housing supply elasticity are unrelated to changing economic conditions. With instrumentation, our estimates of the housing price effect on firms’ investment are consistently more negative.
Digging deeper, we find that the negative investment effect is stronger for firms that borrow from smaller, more regional banks, and for firms that have limited sources of external financing (e.g. no access to bond markets). This finding is intuitive because larger banks are presumably able to obtain sufficient capital to allocate in all productive sectors, while smaller banks may have to substitute capital between sectors. Likewise, unconstrained firms should be better able to maintain investment using internal reserves, or to obtain financing from other capital sources (e.g. public debt and equity markets), as they have greater access compared to constrained firms.
It is important to note that our results do not contradict those of Chaney et al. (2012). They show that firms are able to borrow and invest more when their own real estate collateral increases in value. We confirm their result, but find that the overall effect of housing price appreciation is more complex. The bank lending channel that we document generates an opposite effect, and in fact we show that our bank lending channel is of a similar order of magnitude. Thus, the positive effects of the collateral channel may be completely mitigated by the negative effects of the bank lending channel.
While prior research has investigated the effects of a contraction of bank balance sheets on firm activity, our paper is the first to investigate the role of banks in capital allocation when asset prices are rising in a specific sector of the economy. We find that it is incorrect to assume that an expanding balance sheet leads to positive spillover effects across all sectors of the economy. There is a crowding out effect in which banks divert resources across sectors – in the case studied in our paper, rising real-estate prices lead banks to cut commercial loans and increase real-estate loans.
If the change in relative prices is market-driven, then banks can be seen as reallocating resources across sectors to support the growing sector. However, if the price change is policy-driven, then the channelling of assets to one overheated sector of the economy at the expense of other (potentially more productive) sectors may not be the consequence policymakers have in mind.
Bernanke, Ben S (1983), “Nonmonetary Effects of the Financial Crisis in the Propagation of the Great Depression”, The American Economic Review 73: 257–276.
Bernanke, Ben S and Mark Gertler (1989), “Agency Costs, Net Worth, and Business Fluctuations”, The American Economic Review 79: 14–31.
Bleck, Alexander and Xuewen Liu (2013), “Liquidity Flooding, Asset Prices and the Real Economy”, HKUST and UBC Working Paper.
Chakraborty, Indraneel, Itay Goldstein, and Andrew MacKinlay (2013), “Do Asset Price Bubbles Have Negative Real Effects?”, University of Pennsylvania and SMU Working Paper.
Chaney, Thomas, David Sraer, and David Thesmar (2012), “The Collateral Channel: How Real Estate Shocks Affect Corporate Investment”, The American Economic Review 102.
Cuñat, Vicente, Dragana Cvijanović, and Kathy Yuan (2013), “How Did US Banks React to Capital Losses Induced by Real Estate Prices?”, UNC and LSE Working Paper.
Farhi, Emmanuel and Jean Tirole (2012), “Bubbly Liquidity”, Review of Economic Studies 79: 678–706.
Gan, Jie (2007), “The Real Effects of Asset Market Bubbles: Loan- and Firm-Level Evidence of a Lending Channel”, Review of Financial Studies 20: 1941–1973.
Holmstrom, Bengt and Jean Tirole (1997), “Financial Intermediation, Loanable Funds, and the Real Sector”, Quarterly Journal of Economics 112: 663–691.
Kiyotaki, Nobuhiro and John Moore (1997), “Credit Cycles”, Journal of Political Economy 105: 211–248.
Loutskina, Elena and Philip E Strahan (2013), “Financial Integration, Housing and Economic Volatility”, Boston College and UVA Working Paper.
Miao, Jianjun and Pengfei Wang (2013), “Sectoral Bubbles, Misallocation, and Endogenous Growth”, Boston University Working Paper.
Saiz, Albert (2010), “The Geographic Determinants of Housing Supply”, Quarterly Journal of Economics 125: 1253–1296.
Tirole, Jean (1985), “Asset Bubbles and Overlapping Generations”, Econometrica 53: 1499–1528.
1 This literature is exemplified by the papers of Bernanke and Gertler (1989), Holmstrom and Tirole (1997), and Kiyotaki and Moore (1997).
2 Loutskina and Strahan (2013) investigate the role of integrated banks in amplifying housing-price shocks.
3 See Bernanke (1983), Gan (2007), and Cuñat et al. (2013).
4 See, for example, Tirole (1985), Farhi and Tirole (2012), Bleck and Liu (2013), and Miao and Wang (2013).
5 Land unavailability data is based on Saiz (2010). Our approach is similar to that of Chaney et al. (2012).