Bulletin – June 2010 Commercial Property and Financial Stability – An International Perspective

Introduction

In the United States, but also to some extent in other countries, housing and mortgage markets have played an important role in the most recent crisis. International experience suggests, however, that exposures to commercial property markets have usually posed greater direct risk to the balance sheets of financial institutions. Bank exposures to commercial real estate (CRE), along with other corporate lending, have historically been one of the main sources of loan losses during episodes of banking sector difficulties. This was true both for industrialised economies – for example the banking crises in Scandinavia and Japan in the early 1990s, as well as the US savings & loan crisis – and for the emerging markets most affected by the 1997–1998 Asian financial crisis. Conditions in global commercial property markets have been especially challenging in the current cycle. In most countries, with the notable exception of the United States, losses on CRE lending (including on loans to developers of residential property) currently account for a much greater share of actual and prospective loan losses than do residential mortgages to households.

This article briefly discusses recent developments in commercial property markets in a number of industrialised economies. It also draws out some of the characteristics of CRE and aspects of the way it is financed, which contribute to its greater cyclicality and capacity to threaten financial stability compared with residential mortgage lending.

Developments in Major Commercial Property Markets

In most countries, the price cycle in CRE markets has been more severe during the recent economic downturn than that in housing. Except for France, the recent decline in commercial property prices has been much greater than for residential property, even though the run-up in commercial property prices in some cases was much less marked (Table 1 and Graph 1). The difference in the sizes of commercial and residential property price cycles has been particularly stark in the United States, the United Kingdom and Ireland where, on some measures, CRE prices have fallen by between 20 and 30 percentage points more than residential property prices from the peak to the trough.

The downturn in the commercial property market has coincided with a contraction in credit flows to the sector; most lenders in the United States and United Kingdom have reported that lending standards have tightened over the past couple of years (Graph 2). Reported demand for CRE loans remains subdued in the United States, although there are tentative signs of stronger demand in the United Kingdom. Funding from capital markets has become more difficult and global issuance of commercial mortgage-backed securities is well below the levels of a couple of years ago.

Characteristics of Commercial Real Estate

Although housing construction contributes significantly to cyclicality in total economic output, activity in the commercial property and development sector has usually contributed more to the build-up of financial vulnerabilities in previous cycles. Fluctuations in CRE activity and prices propagate rapidly through financial institutions' balance sheets, for several reasons:

• the CRE loan book is more exposed to the construction cycle than is the housing loan book;
• imbalances can build up further because of construction lags;
• borrowers in the CRE sector do not have as much of a disincentive to default as home mortgage borrowers; and
• the typical features of CRE financing induce greater correlations in defaults on these loans.

The share of (relatively risky) construction and development loans in the banking sector's CRE portfolio is greater than for mortgage lending. CRE lending is generally taken to include construction and development (C&D) loans to corporates, as well as loans for buy-and-hold investment in non-residential property (e.g. real estate investment trusts). Owners of existing CRE, unlike household owners of residential property, are often able to access capital market funding, thereby reducing their call on bank-originated funding. In addition, some C&D projects for new residential property (housing construction) are undertaken by firms and are treated as CRE exposures. Both these market features cause banks' CRE exposures to be relatively more concentrated in financing new development than in financing buy-and-hold investors, compared with the mix of construction versus purchase loans in the housing mortgage book. Among US institutions insured by the Federal Deposit Insurance Corporation (FDIC), for example, C&D accounts for around 20 per cent of CRE loans, compared with 3 per cent for residential real estate. This mix matters because C&D loans are inherently more prone to becoming impaired in a downturn than (secured) loans for buy-and-hold investment. The value of the project is not realised until it is complete, and as discussed below, this can result in sizeable losses-given-default on the associated loans. Property developers also sometimes have undiversified business models, so they are more likely to fail in economic downturns.

Given that much CRE lending is for specific projects, and that loans are frequently syndicated, it is often an easier market for new entrants (including foreign and second-tier domestic banks) than retail mortgage lending and relationship-based commercial lending. And within the CRE lending market, the entering lenders are more likely to end up with the more marginal, less-established borrowers. This dynamic adds to the concentration of risk around CRE lending.

The United States is a good example of the consequences of these factors for the financial system. Small US deposit-taking institutions (those with assets less than US$10 billion) are relatively highly exposed to the commercial property sector. At small deposit-taking institutions, commercial property lending accounts for 40 per cent of their on-balance sheet gross loans, compared with 21 per cent for mid-sized banks and 13 per cent for large banks (Graph 3). Within this, the smaller institutions are also more exposed to loans for C&D, providing almost half of C&D loans in the United States. Similarly in Australia, much of the increase in exposures and non-performing commercial property loans has been seen among the smaller and foreign-owned banks.

Construction lags are longer and construction cycles are lumpier for commercial developments than for housing construction. Dwellings – particularly detached single-family houses – can be erected more quickly than a typical commercial development such as a shopping centre, office building or apartment building. Average construction lags are therefore inherently longer in the commercial property sector than for residential property. As a result it is easier for ‘hog cycles’ to arise, where building work in the pipeline turns out to have been unnecessary because demand has already turned down. If the market turns down before the project is finished, the result can be an unfinished site with little residual recovery value other than the land. Depending on the remediation or completion costs, the loss to the bank can be 100 per cent of the loan or even more.

Commercial property can be more cyclical than residential real estate because its construction is lumpier. Commercial property developments are larger relative to the size of the existing stock of CRE than is the case for housing. New (CRE) developments therefore have a greater local supply effect than for residential property. Vacancy rates tend to remain high long after the economic downturn and well into the subsequent recovery, because it takes a considerable period of time for excess supply to be absorbed. For example, in the recession of 2000–2001, US office vacancy rates increased steadily from 7.7 per cent in the September quarter 2000 to a peak of 16.8 per cent three years later, and remained above 16 per cent until late 2004, long after the turnaround in GDP growth.

In the current cycle, the National Association of Realtors expects US office vacancy rates to peak at 17.4 per cent in 2011, around two years after the trough in GDP growth (Table 2). Industrial and retail vacancy rates are forecast to increase to around 14½ per cent and 13 per cent respectively this year and remain roughly steady in 2011. In contrast, the average vacancy rate on multifamily residential real estate is believed to have already peaked in 2009 and is expected to decline to around 6 per cent by 2011.

Similar trends in vacancy rates are evident in the euro area, especially for Germany. Private sector bodies forecast office vacancy rates in major German cities to peak in 2011, two years after the trough in German GDP. For the other major euro area economies, they expect vacancy rates to peak in either 2010 or 2011, consistent with the relative speeds of economic recovery, but later than the recoveries in GDP. In the United Kingdom, office vacancy rates in London are forecast to decline more rapidly than in other major markets but are expected to remain above 10 per cent in 2011.

Default probabilities are arguably more cyclically sensitive for CRE loans than for residential mortgages. Defaults on CRE lending tend to be bunched in cyclical downswings, more so than defaults on residential mortgages. The main reason for this is the cyclicality of defaults on lending for construction projects described above. In addition, defaults by buy-and-hold investors are also highly cyclical, because they are usually not owner-occupiers. They therefore face the risk of a sudden loss of rental income should the tenant move out, which is more likely in an economic downturn, when more firms are failing or otherwise shedding labour. If the value of the property has also fallen below the size of the loan, these borrowers might make themselves better off by defaulting. Owners of CRE are thus typically more likely to default in a downturn than home-owning households, who derive the same real benefit from living in their home regardless of its price. Only a small minority of home mortgage borrowers in negative equity actually default, even in the United States where lenders frequently do not pursue defaulters for any deficiency between the collateral value and the loan amount (despite the law allowing them to do so in most jurisdictions).^[1]

The importance of C&D lending in making the CRE loan book riskier than home loans can be seen from the poor performance of C&D loans made by US deposit-taking institutions. Non-performing C&D (predominantly commercial property) loans were 17 per cent of gross C&D loans while net charge-offs were 6 per cent of gross loans in the March quarter 2010 (Graph 4). The level of non-performing loans for CRE excluding C&D and on residential mortgages was much lower, even though these have risen sharply as a share of gross loans during the crisis. The quality of the collateral securing the loan is crucial to the recovery on the loan following a default, and for C&D loans, the riskier nature of the collateral also leads to much higher net charge-offs for a given rate of non-performing loans.

The CRE financing model is more procyclical than mortgage lending and more liable to produce correlated defaults. In most industrialised economies, residential mortgage lending generally takes the form of long-term amortising mortgages, as is appropriate to the typical expected holding periods for the asset. By contrast, much of the banking sector's lending to CRE is in the form of shorter-term project and syndicated finance, even for existing properties, which do not have the short-term horizon of a construction project. CRE borrowers therefore face more refinancing risk than mortgage borrowers in most countries; when that risk does crystallise, it is generally correlated across borrowers as the credit cycle turns. During the recent housing boom in the United States, the residential mortgage market became more similar to CRE finance, shifting towards products that all but required frequent refinancing, to avoid the sharp payment increases built into the contract (Gorton 2008). It is not a coincidence that mortgage defaults became more cyclical and more correlated following that shift – that is, more like the historical pattern of defaults on CRE loans.

Several other aspects of CRE finance contribute to its procyclicality. Unlike home mortgage lending, banks often impose covenants on CRE firms' gearing over the life of the project. But because CRE resale markets are usually quite thin, price appraisals and estimates are typically used instead of market prices. CRE property values are therefore marked down when rents soften or vacancies rise in that market. Borrowers can end up breaching loan covenants even if their own project is still profitable in an income flow sense; but because the project might now be more highly geared, refinancing becomes difficult. In a downturn, distressed sales add to the difficulties for borrowers that are still performing. Along with the more frequent refinancing normally built into loan contracts, the sensitivity of CRE loan covenants to current valuations results in more defaults and distressed sales. These in turn weigh on CRE prices elsewhere, generating correlated defaults. This contrasts with home borrowers, who are generally permitted to stay in a negative equity situation for as long as they can service the loan and do not want or need to move.

Conclusion

The outlook for commercial property markets and lenders in the major countries remains challenging as vacancy rates continue to rise and prices and rents are yet to recover. It is noteworthy, though, that loan losses on commercial property in Australia during the recent period have been relatively small compared with those in the United States and in some countries in Europe.

The crisis has severely affected global commercial property markets, with large declines in asset values and a sharp deterioration in asset quality in some major markets. In the current cycle as well as in previous ones, the downturn in commercial property and development markets has generally been more severe than in housing markets. The effect on bank loan losses is also generally greater. A number of factors contribute to this: banks' CRE loans are more concentrated in construction loans than housing lending is; imbalances can build up further because construction lags are longer; and borrowers in the CRE sector do not have the same disincentives to default as home mortgage borrowers. Developments in housing markets are also important for financial stability, but banks' related loan losses have historically been more concentrated in loans to (corporate) property developers, which are captured in CRE lending, than in loans to households.

Footnotes

The authors are from Financial Stability Department. [*]

Recent Federal Reserve research indicates that only about 1 in 10 US mortgage borrowers in negative equity actually default (Foote, Gerardi and Willen 2008). In the United Kingdom and other industrialised economies, industry reports suggest that this fraction is even smaller. [1]

References

Foote CL, K Gerardi and PS Willen (2008), ‘Negative Equity and Foreclosure: Theory and Evidence’, Federal Reserve Bank of Boston Public Policy Discussion Paper No 08-3.

Gorton G (2008), ‘The Panic of 2007’, in Maintaining Stability in a Changing Financial System, Jackson Hole Symposium, Federal Reserve Bank of Kansas City, Kansas City, pp 131–262.