1. "Beyond Signaling and Human Capital:
Education and the Revelation of Ability" with Peter Arcidiacono
and Patrick Bayer. 2010, American Economic Journal: Applied
Economics. 2(4): 76-104. (Appendix)
In traditional signaling
models, education provides a way for individuals to sort themselves by
ability. Employers in turn use education to statistically discriminate,
paying wages that reflect the average productivity of workers with the
same given level of education. In this paper, we provide evidence that
education (specifically, attending college) plays a much more direct
role in revealing ability to the labor market. Using the NLSY79, our
results suggest that ability is observed nearly perfectly for college
graduates. In contrast, returns to AFQT for high school graduates are
initially very close to zero and rise steeply with experience. As a
result, from very beginning of the career, college graduates are paid in
accordance with their own ability, while the wages of high school
graduates are initially completely unrelated to their own ability. This
view of ability revelation in the labor market has considerable power in
explaining racial differences in wages, education, and the returns to
ability. In particular, we find no racial differences in wages or
returns to ability in the college labor market, but a 6-10 percent wage
penalty for blacks (conditional on ability) in the high school market.
These results are consistent with the notion that employers use race to
statistically discriminate in the high school market but have no need to
do so in the college market.
2. "The Effect of Mortgage Credit Availability on House Prices and Construction: Evidence from a Frontier Estimation Approach", with Elliot Anenberg, Edward Kung and Raven Molloy (2015). Working Paper.
We present new evidence on changes in mortgage credit availability from 2001 to 2014 and estimate its effects on house prices and residential construction. To isolate changes in mortgage credit supply from changes in credit demand, we construct a new measure of supply based on production frontier estimation. This ``loan frontier'' allows us to examine changes in credit availability for different types of borrowers in different housing markets, dimensions that have yet to be fully explored. Exploiting the disaggregated nature of our measure and national trends in the amount of credit extended to various groups, we construct an instrument for mortgage credit supply. The exogenous variation in the loan frontier can explain 27\% of the total variation in price appreciation. We find that a one percent increase in the change in the aggregated loan frontier increases the change in house prices by 0.9 percent and the change in the single-family housing stock by 0.09 percent.
3. "Risk in Housing Markets: An Equilibrium Approach", Working Paper.
Homeowners are overexposed to city-specific house price risk and income risks, which may be very difficult to insure against using standard financial instruments. This paper develops a micro-founded equilibrium model that transparently shows how this local uninsurable risk affects individual location decisions and portfolio choices, and ultimately how it affects prices in equilibrium. I estimate a version of this model using house price and wage data and provide estimates for risk premia for different cities, which imply that homes are on average about $20000 cheaper than they would be if owners were risk-neutral. This estimate is over $100000 for volatile coastal cities. Next, I simulate the model to study the effects of financial innovation on equilibrium outcomes. Creating assets that hedge city-specific risks increases house prices by about 20% and productivity by about 10%. The average willingness to pay for completing the market per homeowner is between $10000 and $20000. Welfare gains come both from better risk-sharing and from more efficient sorting of households across cities.
4. "Momentum and Neighbourhood Level Differences in the Residential Housing Market" with Alex Chinco, 2015, Draft Coming soon.
5. "The Common Variation in Housing Price Returns" Working Paper.
I show that a real estate counterpart to the Fama-French Three Factor model fits the yearly house price returns very well. Using yearly housing return and average wage data at the metropolitan area level I construct three risk factors. The first factor is the yearly house price returns for the whole US housing market. The second factor replicates a diversified portfolio that holds houses in low priced metropolitan areas and shorts houses in high price ones. The third factor replicates a diversified portfolio that holds houses in metropolitan areas with high price to wage ratios and shorts houses in areas with low price to wage ratio. Remarkably, these three factors explain a nearly 90 percent of the time series and cross-sectional variation in returns for twenty five diversified housing portfolios constructed by sorting metropolitan areas on price level and price over wage ratios. As these portfolios are well diversified, I find that idiosyncratic risk is not priced in the cross-section. These results mean that an investor would not have to worry about location specific house price risk if they could hold these portfolios of houses instead of only owning in one particular metropolitan area. On the other hand, when the same analysis is carried out using individual metropolitan areas instead of portfolios, the three factors explain a much lower share of the time series and cross-sectional variation in returns. In addition, idiosyncratic risk is priced in the cross-section. Households that own only in one metro area are exposed to a significant amount of local idiosyncratic risk and this is reflected in house price returns.
6. "Credit Conditions and Biases in the Case-Shiller Repeat Sales Index", with Kristoph Kleiner. In Progress.
Assets such as real estate, large transportation equipment, and luxury artwork sell infrequently and are not interchangeable. As a result traditional price indices, such as the Case-Shiller Index for Metropolitan House Prices, are the dominant method to measure price fluctuations of these illiquid and heterogeneous assets. We first argue that changing financial constraints affect the market composition and demand, resulting in mismeasured traditional price indices. Second, we correct for this bias by introducing the locally-weighted repeat sales technique, a novel estimation procedure for estimating a distinct price index for any given asset; by aggregating the results we can define a new properly measured index. Third, we highlight the advantage of our technique by examining the US housing market. By matching transaction-level data with household credit data we find that beginning in 2004 changing credit conditions to low income households drove strong demand and high turnover in low tier housing. As a result of these changing market conditions, Case-Shiller overstates the peak of the housing bubble by 20% in the Los Angeles Market and understates the decline by nearly half in Cleveland. Our econometric results highlight the effect of market conditions on traditional price indices and the applicability of our new econometric methodology to overcome these concerns.
7. "Hedging Housing Risk with Stock Indexes from Local
Employers" (Slides), In Progress.
Using detailed data on the location of homes and employment centers,
this paper documents how much of the local house price risk can be
hedged using traded financial assets. While local house prices returns
are generally not correlated with aggregate stock price indexes, they
show considerable correlation with local stocks. Exploiting these
correlations can allow for the construction of location specific optimal
retirement portfolios that do take into account the large idiosyncratic
risk that homeowners face by living in a particular neighborhood. I use
detailed housing transaction level data to construct a house specific
price index. This indexes are estimated by a local-linear repeat-sales
estimator that only uses transactions of similar homes. Using business
establishment data from the NETS database, I find the closest largest
employers to a particular home and construct a location specific stock
index. This local stock index is created by weighting the individual
stock returns of local employers by the distance to the particular home
and by the number of employees. At the metropolitan area level this
procedure yields a correlation of about .35 between the house price
index and the local stock index. Residents of these areas would benefit
from shorting the local stock index to hedge housing price risk.
8. "Mortgage REITs and Reaching for Yield" (Slides), with Stijn Van Nieuwerburgh and James Vickery. In Progress.
Mortgage REITs provide secondary market investors an opportunity to invest in mortgages or MBS in a structure similar to the structure mutual funds provide for investing in stocks. Between 2010-2013, assets under the management of mortgage REITs have increased by 350%. Given the poor historical performance of mortgage REITs, why have investors been increasingly attracted to them? We provide evidence that retail investors are attracted to the high dividend yields that mortgage REITs offer, especially in a low interest rate environment. Using data on 13-F filings, we show that most of the recent growth of MREITs has been driven by retail investors. MREITs that experience increases in dividend yields issue more new shares and experience higher inflows from retail investors relative to institutional ones. The effect of higher dividend yields on stock issuance and retail investors is more pronounced in low interest rate environments. While dividend yields are associated with higher future returns for most stocks, this is not the case for MREITs. MREITs that have higher dividend yields and that issue more shares have lower expected returns both in the short and the long run. Take together, we interpret these findings as evidence that investors may be chasing yields, especially when the Fed keeps the interest rates low.
9. "The Dynamics of Gentrification in Housing Markets", with Patrick Bayer, Marcus Casey and Eduardo Jardim. In Progress.
We develop and estimate the first dynamic search equilibrium model with social interactions that incorporates the real geography of a city. In particular, the model features forward-looking, heterogenous buyers and sellers sorting across neighborhoods in an equilibrium search framework. The individual location decisions of both buyers and sellers are based on current and expected neighborhood amenities that evolve endogenously over time. Further, the model allows individuals to use information on amenities from the neighborhood of interest as well a set of nearby neighborhoods to form their forecasts. To mitigate the typical curse of dimensionality problem we use approximations of high-dimensional models with sieve value function iterations.
Using data from 1990-2013 for Chicago, we fit and simulate a fully dynamic sorting equilibrium and compare its predicted behavior to a model where households are myopic. We find that without forward looking behavior we cannot simultaneously explain the speed of gentrification that we see in the data and the cross-section distribution of income and of neighborhood quality. Gentrification occurs faster in a dynamic model since high income individuals understand that if they move into a neighborhood, they will make it more desirable for other high income individuals to live there, increasing the speed of gentrification. The forward looking feature of the model combined with social interactions help richer agents coordinate and cause the neighborhood to transition.
We conduct a series of policy experiments that explore how the preferences over socioeconomic composition and amenities interact with city structure to change the speed and character gentrification of the city. We evaluate various strategies a central planner can use to revitalize certain parts of the city without causing the neighborhood to fully gentrify.