Working Papers

Does Rent Regulation Affect Tenant Unemployment? Evidence from New York City.  with Hanchen Jiang and Xi Yang. Under Review. 

This paper estimates the effect of rent stabilization on tenant unemployment, an unintended consequence of rent regulation, using the New York City Housing and Vacancy Survey (NYCHVS) from 2002 to 2017. This policy is ubiquitous and imposes binding restrictions to rent growth in New York City. Our results show that rent stabilization increases the unemployment of tenants by approximately six percentage points, more than double the average unemployment rate during the study period. Moreover, the effect is concentrated in traditionally privileged groups: non-minority tenants and those with high education levels. These results are consistent with the predictions from a job search model with rent regulation we develop. We address the potential endogeneity of the policy with an instrumental variable that captures the relative vacancy availability of rent-stabilized units in the rental market for each neighborhood at the time of moving into the unit, leveraging rich data on vacancies for nearly four decades. We also document that around a third of policy beneficiaries are unaware of the stabilized status, a novel empirical fact, and use this awareness to identify mechanisms further.

The Long and Winding Roads: Roads, Inequality, and Growth in Colombia. with Guillermo Sinisterra. 

We measure road improvement and road construction on production and inequality in Colombia from 1993 to 2012, taking into account network effects using a market access approach. We found that roads, by changing market access, have an important effect on GDP growth and all the sectors. We also find that GDP increases with distance to the intervention. We address endogeneity in multiple ways. We use exogenous variation based on the likelihood of receiving a road improvement based on pre-colonial (indigenous) roads least-cost cost path counterfactual road networks that use estimated construction costs; we also build alternative market access measures that focus on quasi-random market access changes stemming from exposure to markets of smaller cities. We find that roads concentrate the land close to that infrastructure in fewer hands. Also, roads have an important effect on municipal development indicators. Roads also seem to have important spillover effects on municipalities located at 35km or closer to the intervention.

The Financial Fragility of For-Profit Hospitals: Evidence from the COVID-19 Pandemic with Ge Bai, Daniel Jiménez, Phillip Phan, and Alessandro Rebucci. R&R at Journal of Financial Stability.

We estimate the likelihood of financial distress of U.S. hospitals in 2020 due to the COVID-19 pandemic using AHA Annual Survey data for 2011-2019 and smartphone mobility data for 2020. We find that while the average likelihood of distress across all hospitals is 28.53 % in 2020, slightly increasing from 2019, for-profit hospitals are much more likely to be distressed. Their average likelihood of financial distress is 39.13 %---a 6.93 percentage point increase from 2019. For-profit hospitals are the main providers of specialty health care services, such as psychiatric and acute long-term care, so their increased likelihood of distress poses a risk to service provision in these specialty areas, and particularly in rural communities. Our prediction model based on mobility data performs very well in sample against actual data and can potentially help policymakers and hospital administrators to monitor financial distress in real-time when case mixes change, or other large shocks materialize.

Unequal Response to Mobility Restrictions: Evidence from COVID-19 Lockdown in the City of Bogota ́. with David Castells‐Quintana, Paula Herrera-Idárraga, and Guillermo Sinisterra. R&R at Spatial Economic Analysis.

In this paper, we study the efficacy of government-mandated mobility restrictions on curbing urban mobility, and estimate the spatial heterogeneity in lockdown compliance. We explore the role of cash subsidies disbursed during lockdown as well as socioeconomic differences across neighborhoods in explaining their unequal response to mobility restrictions. We rely on novel data showing changes in movements at highly disaggregated spatial units in Bogota ́, before and during the first wave of the COVID-19 pandemic, matched with data on socioeconomic characteristics as well as data on Non- Pharmaceutical Interventions (NPIs) implemented in the period of analysis. We find that the general lockdown imposed in the city significantly reduced mobility (by about 41pp). When looking at the unequal response across locations, we find that low-income areas, with higher population density, informality and overcrowding, reacted less to mobility restrictions. We also find that cash subsidies were not sufficient to make compliance easier in low-income neighborhoods. See slides. Policy version from the UNDP Latin America and the Caribbean Working Paper Series here

Staggered Adoption of Nonpharmaceutical Interventions to Contain Covid-19 Across U.S. Counties: Direct and Spillover Effects. with Vadim Elenev, Alessandro Rebucci and Emilia Simeonova. R&R at Management Science.

Local policies can have substantial spillovers both across geographies and markets. Despite the developed literature documenting spillovers from local tax policies, little is known about the impact of public health regulations across borders. The COVID-19 pandemic has demonstrated the need to better understand the effects and mechanisms through which public health policies operate across time and space. We estimate the direct and spillover effects of Stay-at-Home-Orders (SHO) on mobility measures of social distancing measures and interaction to contain the spread of COVID-19 at the U.S. county level. Adopting counties should experience a decline in mobility due to the direct effect, while neighbors may experience a spillover, the sign of which is ambiguous from a theoretical perspective. We propose a modified difference-in-difference contiguous-county triplets regression design, comparing both a county that adopted the SHO and its neighbor that did not to a neighbor's neighbor (``hinterland'') county. We find that mobility in neighboring counties declined by a third to a half as much as in the directly treated county. These spillovers are concentrated in triplets sharing the same media sources of news and information. Using directional mobility data, we decompose the neighboring counties' decline in mobility into a reduction in external visits from the treated county and a comparably larger voluntary reduction in the neighboring county's own traffic. Together, our results provide strong evidence that SHOs operate through information-sharing and illustrate the quantitative importance of voluntary social distancing. The finding that the estimated spillovers are in the same direction as the direct effects casts doubt on the prevailing narrative that a more nationally coordinated policy response would have accomplished a greater reduction in mobility and contacts. See it in the NBER's Research Spotlight.

Cities and Productivity: Evidence from 16 Latin American and Caribbean Countries. with Mark Roberts. R&R at JUE.

This paper explores the roles of agglomeration economies and human capital spillovers in accounting for productivity variations across sub-national areas in 16 Latin American and Caribbean countries. We find positive elasticities with respect to density. While heterogeneity exists across countries, the estimated agglomeration economies for the region are comparable to others in high-income countries. Including human capital measures further reduces the estimated agglomeration economies for several countries in our sample, which is consistent with high congestion levels. We also find that human capital externalities play a stronger role in explaining productivity than agglomeration economies in the region. By providing comparable estimates of the strength of agglomeration economies and human capital externalities for many countries, the paper considerably expands knowledge on the determinants of urban productivity in non-high-income countries. World Bank Policy Research Working Paper 8560 version.

Fewer players, fewer homes: concentration and the new dynamics of housing supply. 

We investigate the impact of increasing concentration in local residential construction markets on housing cycle dynamics. We show that the increase in concentration has led to greater unit price volatility, less production, and fewer vacant unsold units. Our results imply that the greater concentration has decreased the annual value of new housing production by $144 billion. Because housing is a determinant of the business cycle these findings provide further evidence that the secular decline in competitive intensity in the American economy is altering macroeconomic dynamics. Previous draft circulated as Market Concentration in Homebuilding. Also see a press release, media coverage, the slides presented at the AEA meetings, and a recent presentation at the ReCapNet 2020.

Measuring the Value of Rent Stabilization and Understanding its Implications for Racial Inequality: Evidence from New York City with Ruoyu Chen, Hanchen Jiang. 

Assessing rent discounts implied by rent regulation is challenging because the counterfactual rents of regulated units in the unregulated market are not observed. We estimate these counterfactual rents and predict the quality-adjusted rent discount for each rent-stabilized unit in New York City (NYC) using novel data from 2002 to 2017. We find robust average rent discounts of $410 per month (34% of contract rents of stabilized units). The aggregate size of these discounts in NYC is between 4 to 5.4 billion USD per year, roughly 10-14% of the federal budget on means-tested housing programs. We document that discounts: (1) increase linearly with housing tenure; (2) are not progressively distributed; (3) are larger in Manhattan and increasing in gentrifying neighborhoods; and (4) are three times larger for households correctly aware of being beneficiaries. We find that rent stabilization has disproportionately benefited White tenants. Not only are they more likely to occupy rent-stabilized units conditional on observables, but they also receive higher discounts. On average, Black stabilized tenants get $150, Hispanics $135, and AAPI $43 less on monthly rent discounts than White stabilized tenants. This racial gap, which has shrunk over time, is mainly explained by the uneven sorting of households of different races across locations. GLO Discussion Paper, No. 1102

Urban Decline in an Urbanizing World, with Paula Restrepo.

This article presents evidence on the striking phenomenon of population decline in Eastern Europe and Central Asia (ECA) using a novel dataset that spans more than 3 decades for all cities in the countries of the region. To explain the observed patterns of population redistribution in a context of strong population decline, we present a modified gravity migration model based on Brezis and Krugman (1997) model on life cycle of cities. Under a negative population shock, the model predicts concentration of population in larger cities, driven by a temporary wedge between productivity and living costs.  We test these predictions and find that indeed population distribution across the city distribution increased its concentration. In particular, there is a negative causal effect on population growth of having access to larger labor markets, and a causal positive effect of having a larger local market. This research was awarded the ECA Academy award 2020

The Efficacy of a Pre-Algebra Cognitive Tutor in Chile and Mexico, with Jason Imbrogno, Ignacio Casas, and Paul Goodman.

A math cognitive tutor (MCT) system widely used throughout the U.S. was adapted for use in Chilean and Mexican public middle schools. The curriculum requires the use of computers for individual students to progress through an extended pre-algebra program. We show that students enrolled in schools which were randomly assigned to adopt the MCT significantly im- proved their standardized math test scores as compared to control group peers. However, the implementation of the changes in the schools and classrooms was not perfect. Those schools which were better prepared to make changes, especially those with sufficient computers and technical support services, saw their students master more of the software part of the curriculum. Students and teachers generally viewed the MCT positively. Knowledge from this study regarding the structure and implementation required for schools to successfully exploit the unique teaching capabilities of the MCT should guide the future diffusion of this specific technology.

MCMC Approach to Classical Estimation with Overidentifying Restrictions.

I extend the Laplace estimators approach proposed by Chernozhukov and Hong (2003) for an overidentified system by decomposing the m moments into the identifying space and the overidentifying space, and using both to construct a transformed criterion function for a new just-identified system. Parameters and test statistics are estimated simultaneously using the entire equation do- main, not only the global minimum. As in Chernozhukov and Hong (2003), Markov-Chain Monte Carlo (MCMC) avoids the curse of dimensionality in this method. It is also applicable to non-smooth criterion functions. Incorporating the ORs in the objective function amounts to using economic theory as criterion for estimate selection when facing multiple local solutions. The proposed estimators outperform counterparts in simulation of an asset-pricing model in Hall and Horowitz (1996).

Stopping Asset Market Bubbles.

We test a policy that changes the information and payoff structure in a double auction experiment in order to prevent bubbles from forming. This cash out policy gives the investor all his asset holding (game money) and dividends at the end of each transaction period in the form of real cash. This intends to change the reference point and make the investor realize gains or losses as real. The results of the experiments support the hypothesis that, at least partially, the creation of bubbles is affected by mental accounting and framing. These phenomena affect investors’ willingness to hold or demand assets whose prices are unrealistically above its fundamental value. We have experiments with 3 groups of subjects (inexperienced, experienced, and mixed). The largest bubbles are formed for the inexperienced subjects, although bubbles do form in all groups. In all cases where strong bubbles are present, the cash out policy seems to have a strong effect in reducing or preventing the bubble formation.

Projects in Progress

  • Here Comes the Neighborhood: The microstructure of housing supply.  (with Caitlin Gorback).