Housing: Microdata, macro problems

Global Housing Watch Newsletter: May 2016

*Below is a conference summary prepared by Philippe Bracke (Bank of England).


On May 4-5, the Bank of England hosted the conference Housing: Microdata, macro problems (program here) in London, organised by Philippe Bracke (Bank of England), Jonathan Halket (Cemmap and Essex) and Lars Neshiem (Cemmap, UCL, and IFS). The conference was co-sponsored by the Bank of England, CeMMAp and the Brevan Howard Centre for Financial Analysis; it also benefitted from a Transformative Research Grant by the UK Economic and Social Research Council.


First day


Alex Brazier, director of Financial Stability at the Bank of England, gave the opening remarks. His speech stressed the importance of academic research and, in particular, micro data analysis to inform policy decisions.

Morris Davis (Rutgers Business School) started the academic presentations by talking about his research with Erwan Quintin (Wisconsin Business School) on Default when Current House Values are Uncertain. People’s evaluation of their own house lags aggregate price trends as measured by the Case-Shiller indices. The authors construct a model based on this empirical finding to replicate aggregate default numbers in the US. The discussant Colin Caines (University of British Columbia) highlighted the need for having more data on who defaulted in 2007-2009 and recommended the insertion of an unemployment shock in the model.

Allen Head (Queen’s University) presented Default, Mortgage Standards and Housing Liquidity, a joint work with Hongfei Sun (Queen’s University), and Chenggang Zhou (University of Waterloo). Their quantitative model with directed housing search and mortgages, replicates the empirical association between debt level and asking price/time to sell, highlighted by the previous literature. Households’ decision to sell may be triggered by either a relocation shock or a financial-distress shock. In his discussion, Erwan Quintin suggested a greater emphasis on what search models one can add (persistence/propagation of shocks) on top of standard non-search setups.

Aaron Hedlund (University of Missouri) presented Failure to Launch: Housing, Debt Overhang, and the Inflation Option During the Great Recession, where he sets up a general equilibrium model to evaluate the effects of an inflation increase (assuming policy makers are able to make this happen). The model predicts three effects: (1) inflation reduces the burden of debtors (standard Fisher effect); however, (2) the Fisher effect is offset by banks tightening credit to avoid a loss in profits; (3) the housing market becomes more liquid because of less indebted homeowners. The discussant Wei Cui (UCL) focused on the role of nominal rigidities in mortgage markets and the broader debate on the effects of monetary policy on housing markets.

Kurt Mitman (Stockholm University) presented Consumption and House Prices in the Great Recession, joint with Greg Kaplan (Princeton) and Gianluca Violante (NYU). Their quantitative model allows for three kinds of shocks: credit conditions, productivity, and beliefs in future house price movements. They find that beliefs in future house price movements are the most important driver of house prices booms and busts. This is consistent with other recent papers which (a) stress the need to incorporate nonstandard house price expectations in quantitative models, and (b) admit that the credit channel is perhaps less important than we used to think. The discussant Ralph Luetticke (University of Bonn) suggested that the authors explore a bit more the differences in boom-bust amplitude between US states to say something more about the role of the different shocks.

Ethan Ilzetsky (LSE) presented Interest Rates, Debt and Intertemporal Allocation: Evidence from Notched Mortgage Contracts in the UK, co-authored with Michael Best (Stanford), James Cloyne (Bank of England) and Henrik Kleven (LSE). This paper exploits a feature of the UK housing market whereby salient LTV thresholds (such as 90%) are associated with discrete jumps in the interest rate paid (which applies to the entire loan). The authors can estimate the elasticity of intertemporal substitution by looking at how many people decide to stop below the threshold as compared with a counterfactual “smooth” LTV distribution. The discussant Ben Etheridge (Essex University) asked for more information on the dynamics in the data, in particular on the remortgaging, and equity extraction patterns.

Mariassunta Giannetti (Stockholm School of Economics) presented her work with Giovanni Favara (FED Board) Forced Asset Sales and the Concentration of Outstanding Debt: Evidence from the Mortgage Market, which addresses the question of why banks do not renegotiate more with their mortgage borrowers, since defaults are so costly. This paper shows that banks are more willing to renegotiate when they hold a higher fraction of total mortgages in a given area, i.e. when they can internalize the default externality. The discussant David Miles (Bank of England) focused on the different definitions of concentration in the mortgage market, and whether these are based on flows of new loans or the stock of outstanding mortgages.

Jonathan Halket presented his joint work with Lars Neshiem and Florian Oswald (Sciences Po) The Housing Stock, Housing Prices, and User Costs: The Roles of Location, Structure and Unobserved Quality on why some properties are more likely to end up in the rental market or the owner-occupied market. Motivated by data from the English Housing Survey, where properties are less likely to be owner occupied if they are smaller and closer to the city centre, the authors construct a selection model with observable and unobservable housing features which they bring to London properties and interpret the results through a user-cost framework.


Second day


Anthony De Fusco (Northwestern) kicked off the second day with his presentation Homeowner Borrowing and Housing Collateral: New Evidence from Expiring Price Controls in which he empirically disentangles the two mechanisms behind the observed relation between house prices and household indebtedness: wealth effect and collateral effect. He exploits a rule in Montgomery Country (Maryland) where people who bought houses at an affordable discounted price cannot resell their property at full market value before a certain date; up to that date, banks can only lend against the discounted price. The identification stems from the jump in collateral value at this pre-established date, and the absence of a corresponding jump in lifetime wealth. The discussant Orazio Attanasio (UCL) suggested that future iterations of this paper on this line of research should include a theoretical model of the mechanisms at work.

Jose Fillat (Federal Reserve Bank of Boston) presented Portfolio Choice with House Value Misperception, joint work with Stefano Corradin (ECB) and Carles Vergara-Alert (IESE). Similar to Morris Davis’s presentation on day one, the first part of this paper collects evidence on the differences between households’ reported house values and actual house price indices. This evidence is used in a model of portfolio allocation with slow information acquisition; the predictions of the model are then compared with data from the Panel Study of Income Dynamics (PSID) in the US. The discussant Eric Smith (Essex University) pointed out that it is difficult to exactly define misperception in a market with frictions, where one could come up with several reasons for the observed discrepancy between valuation and actual prices on top of the mechanism highlighted in the paper.

Silvana Tenreyro (London School of Economics) presented her joint work with Philippe Bracke History Dependence in the Housing Market: Facts and Explanations. Using all residential transactions in England and Wales in the last twenty years, this paper shows that the aggregate house price level at the time a house was purchased influence its price and selling probability today. The authors disentangle the two main mechanisms (loss aversion and down-payment effect) by separating properties bought with cash from other properties, and using information on loan-to-value ratios for the latter group. The discussant Andreas Fuster (Federal Reserve Bank of New York) suggested to dig more into the mortgage data (for instance, by distinguishing between first-time buyers and other purchasers) and to use more local house price indices rather than aggregate ones.

Kyle Mangum (Georgia State University) presented Speculative Fever: Investor Contagion in the Housing Bubble, joint work with Patrick Bayer, and James Roberts (both at Duke University). The paper uses detailed micro data on housing transactions (including names of buyers and sellers) to check whether people living close to housing investors or investment properties are more likely to become investors themselves. The authors find evidence of the contagion mechanism that underpins many theories of bubble formation. The discussant Florian Oswald (Sciences Po) highlighted some of the difficulties of identifying investors in the data using names associated with multiple transactions.

Tomasz Piskorski (Columbia University) presented his joint paper with Sumit Agarwal (NUS Business School), Gene Amromin (Federal Reserve Bank of Chicago), Souphala Chomsisengphet (Office of the Comptroller of the Currency), Amit Seru (Chicago Booth) and Vincent Yao (Georgia State University) on Mortgage Refinancing, Consumer Spending, and Competition: Evidence from the Home Affordable Refinancing Program (HARP). The first part of the presentation contained analysis of the participation rate to HARP and its effect on durable consumption. The second part investigated why the participation rate was lower than expected and how competitive frictions in the banking market may have pushed down the overall benefits of the program. The discussant Joao Cocco (London Business School) focused on the convoluted political process that led to HARP and its many subsequent modifications, and on how the authors could incorporate this background information into their analysis.

Chamma Yoon (Baruch College) presented Residential Construction Lags and the Real-Options Channel of Housing Supply, joint with Hyunseung Oh (Vanderbilt). While many studies focus on the extensive margin of housing supply (housing starts), this paper concentrates on the intensive margin (how fast houses are completed). The data show an increase in construction lags during the housing bust and the authors build a real option model to match the facts. The discussant Yoannis Yoannides (Tufts University) commended the paper’s focus on a poorly understood cyclical property of housing supply and suggested to derive analytical solutions from the model, which at the moment is solved computationally.

To sum up

The conference included the perfect mix of sophisticated quantitative models, great micro data, and fascinating insights from behavioural economics (expectations and misperceptions were frequently mentioned). Housing research is alive and thriving. Hopefully the next months and years will see a similar flow of new research ideas and papers, some of which helped, perhaps, by the discussions that took place in these couple of days in London.

Posted by at 5:00 AM

Labels: Global Housing Watch


Subscribe to: Posts