![Sample our Urban Studies journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110826/TOC-Subject-Sample-2021-Urban-Studies.jpg"})
Abstract
In this study, we examine whether the fluctuation of residential housing prices affects the stability of financial institutions in 31 Organization for Economic Co-operation and Development (OECD) and emerging Asian countries. Utilizing 272,530 unique observations from 1990 to 2017, we explore effects of residential financing on bank stability mostly channeled by housing prices as a collateral effect, and by a type of moral hazard due to asymmetric information between borrowers and lenders. We first apply pooled mean group (PMG) estimators to isolate housing market price deviations from the market’s fundamental equilibrium, and then we conduct analysis using bank stability measures including return on assets (ROA,) and non-performing loan ratios (NPL). Results indicate that price deviation in the housing market from fundamental equilibrium is statistically significant for bank stability measures and persists in OECD countries over longer time horizons compared to emerging Asian countries. The findings also imply that loan growth represents a critical factor determining the level of bank stability while result from individual countries vary according to the relative maturity of their economy, level of interdependency between housing market and banking sector, and regulative environment for residential housing finance.
Acknowledgments
The authors gratefully acknowledge valuable comments and suggestions from discussants and participants, especially Dr. Jon Wiley during the 2018 ARES meeting in Florida. The authors modified the unpublished PhD dissertation of Sangjun Lee with a research grant from Hanyang University.
Notes
1 Bernanke and Gertler (Citation1989) assert that firms’ ability to borrow depends on the net worth of their underlying assets. Due to information asymmetry, lenders tend to have little information about the reliability of any given borrower such that banks require borrowers to collateralize their assets to secure the ability to repay. Thus, a decrease in asset value deteriorates the net worth of the underlying asset on the bank’s balance sheets. The resulting deterioration of their ability to borrow has a negative impact on their investment. In response, the feedback cycle of falling asset prices tightens lending provisions, worsening financial conditions. This financial feedback loop occurs from a small change in financial markets that interplays with a large change in economic conditions.
2 Case and Shiller (Citation2003) attribute the excess housing returns exceeding predicted values to an increase in sentiment. Shiller (Citation2007) argues that fundamentals do not explain the recent housing boom. Rather, a type of psychological theory better explains the phenomenon: Wheaton and Nechayev (Citation2008) note that the increase in housing prices from 1998 to 2005 could not be explained by demand fundamentals,. Jin et al. (Citation2014) document that market sentiment has a sizable economic impact on residential real estate markets.
3 Two econometric techniques have been adopted to estimate nonstationary dynamic panels where the parameters are heterogeneous across groups: the mean-group (MG) and pooled mean-group (PMG) estimators. The MG estimators are the average coefficients from single time-series regressions for each component of the panel, whereas the PMG estimator depends on the combination of pooled coefficients for the long-run relationship and average coefficients for the short-run dynamics. The PMG estimator posits a homogeneity restriction on the long-run relationship between variables, whereas the MG estimator does not. The main advantage of using the PMG estimator in the context of real estate price analysis is that it relies on economic theory that links real estate price developments to their fundamental value and generates more efficient estimates compared with the MG estimator if the homogeneity restrictions imposed by the theory holds (Koetter & Poghosyan, Citation2010).
4 During and immediately subsequent to the global financial crisis (GFC), unconventional monetary policy also became commonplace, as financial and real estate markets in many countries experienced large financial losses and experienced a heightened risk environment.
5 A potential limitation of the MG model is that it is quite sensitive to outliers and small model permutations. We have annual data from 1990 to 2017 and thus we consider the time series of data included in our study satisfactory for reliability.
6 Similar to previous studies,our work adopts the measurement of ROA as a proxy for bank profitability and stability. We consider the ROA measurement to be unbiased measure from interest rate fluctuation and change of local or government tax policy on corporate taxation. Alternatively, ROE measure is reflective of country-level tax policy changes, and net interest margin is problematic due to management internal interest rate policy for each financial product. Thus we consider ROA to be an appropriate indicator for bank stability that is unbiased from capital structure preference, operational expenses, and individual bank-level internal interest policy (Kohlscheen et al. 2018).
7 The 8 Asian countries included in our study are China, Hong Kong, Indonesia, Philippines, Malaysia Korea, Thailand, and Singapore, and the 24 OECD countries are the United States, Canada, Japan, Australia, Israel, New Zealand, and European countries (Austria, Belgium, Denmark, France, Finland, Germany, Greece, Ireland, Iceland, Italy, Luxembourg, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, and the United Kingdom).
8 Analysis includes the country-level PMG estimates, however we only present results from full sample of both aggregated OECD, and emerging countries, respectively.
9 The results of panel cointegration estimator guided by Kao (Citation1999) are omitted for concise presentation and are available upon request.
10 Results from the Hausman test indicated that the PMG model results are potentially more reliable.
11 We opted for panel fixed-effect models because pooled OLS is considered to be a highly restrictive model that assumes common intercept and coefficients for all cross sections without consideration of any individual heterogeneity. And the panel random effects model considers the model to be time invariant. The advantage of the fixed effects model is that is assumes the estimator has common slopes and variance, but it allows country-specific intercepts.
12 Liow and Ye (Citation2017) suggest “regime change” during the financial crisis in public real estate market, increasing market volatility and risk spillover across the international market due to the contagion effect.
13 Clarification and elaboration of issues raised by the December 2004 meeting of the Advisory Expert Group of the Intersecretariat Working Group on National Accounts. International Monetary Fund. June 2005.
14 Since the financial recession (and even before), banks’ equity-to-asset ratios have been increasing. The combination of higher minimum capital requirements and stringent stress tests has led to a sharp rise in banking system capital. This focus on bank capital is an international regulatory effort governing bank capital adequacy and stress testing, manifesting in the series of Basel Accords beginning in 1988. This to some degree also impacts the income-to-asset ratio due to higher capital requirements, i.e., less leverage. Both OECD and emerging Asian banks experienced a negative NPL response to increased equity-to-asset ratios during our sample period that begins shortly after the first of three Basel Accords.