Featured Publications
Journal of Economic Behaviour and Organisation (2023)
School’s author: Kevin Lee
We study the relationship between health and changing economic conditions in local areas using a GVAR model that allows for dynamic and interdependent responses to local and national economic conditions. We examine quarterly British data for 2002–2016 for 131 local areas, which displays considerable heterogeneity in economic conditions.
We find robust evidence that health improves as the local economy (employment) expands, but that it takes over two years to realise the full effect. This relationship holds for musculoskeletal, cardiovascular, respiratory, and mental health conditions. We find considerable response heterogeneity at the local area level with the strongest relationship between changes in economic conditions and health found for areas with more traditional industrial structures.
American Economic Review (2023)
School’s author: Mattia Bertazzini
We test between cooperative and extractive theories of the origins of government. We use river shifts in southern Iraq as a natural experiment, in a new archeological panel dataset. A shift away creates a local demand for a government to coordinate because private river irrigation needs to be replaced with public canals. It disincentivizes local extraction as land is no longer productive without irrigation. Consistent with a cooperative theory of government, a river shift away led to state formation, canal construction, and the payment of tribute. We argue that the first governments coordinated between extended households which implemented public good provision.
Economica (2023)
School’s author: Giovanni Facchini
Economic incentives play a key role in the decision to run for office, but little is known on how they shape immigrants' self-selection into candidacy. We study this question using a two-period Roy model, and show that if returns to labour market experience differ between migrants and natives, then this will affect the relative likelihood to run for office for the two groups. We assess this prediction empirically using administrative data from Norway, a country with a very liberal regime for participation in local elections. Our results strongly support our theoretical model and indicate that immigrants' political and economic integration are closely intertwined.
Games and Economic Behaviour (2023)
School’s authors: Maria Montero and Martin Sefton
We use laboratory experiments to study communication games with partially verifiable information. In these games, based on Glazer and Rubinstein, 2004, Glazer and Rubinstein, 2006, an informed sender sends a two-dimensional message to a receiver, but only one dimension of the message can be verified. We investigate the effect of evidence and verification control using three treatments: one where messages are unverifiable, one where the receiver chooses which dimension to verify and one where the sender has this verification control.
First, we find that evidence helps the receiver. Second, despite significant differences in behavior across the two verification treatments, receivers' payoffs do not differ significantly across these treatments, suggesting they are not hurt by delegating verification control. We also show that a theoretically optimal receiver commitment strategy identified by Glazer and Rubinstein is close to being an optimal response to senders' observed behavior in both treatments.
Journal of Economic Behaviour and Organisation (2023)
School’s author: Simon Gaechter
We examine the role of cooperative preferences, beliefs, and punishments to uncover potential cross-societal differences in voluntary cooperation. Using one-shot public goods experiments in four comparable subject pools from the US and the UK (two similar Western societies) and Morocco and Turkey (two comparable non-Western societies), we find that cooperation is lower in Morocco and Turkey than in the UK and the US. Using the ABC approach – in which cooperative attitudes and beliefs explain cooperation – we show that cooperation is mostly driven by differences in beliefs rather than cooperative preferences or peer punishment, both of which are similar across the four subject pools.
Our methodology is generalizable across subject pools and highlights the central role of beliefs in explaining differences in voluntary cooperation within and across culturally, economically, and institutionally diverse societies. Because our behavioral mechanisms correctly predict actual contributions, we argue that our approach provides a suitable methodology for analyzing the determinants of voluntary cooperation of any group of interest.
Economic Journal (2023)
School’s author: Markus Eberhardt, Giovanni Facchini and Valeria Rueda
Academia, and economics in particular, faces increased scrutiny because of gender imbalance. This paper studies the job market for entry-level faculty positions. We employ machine learning methods to analyze gendered patterns in the text of 12,000 reference letters written in support of over 3,700 candidates. Using both supervised and unsupervised techniques, we document widespread differences in the attributes emphasized. Women are systematically more likely to be described using 'grindstone' terms and at times less likely to be praised for their ability. Using information on initial placement we highlight the implications of these gendered descriptors for the quality of academic placement.
Management Science (forthcoming)
School’s author: John Gathergood
Using data from an online brokerage, we examine the role of investor logins in trading behavior. We nd that a new reference point is created when an investor logs in and views their portfolio. We observe this as a disposition eeect on returns since last login, in addition to the traditional disposition eeect on returns since purchase. Further, these reference points produce a strong interaction such that even a small loss since last login nulliies the positive eeect of a gain since purchase. This interaction follows if investors select the higher, more aspirational price as a reference point.
American Economic Review (2023)
School’s author: Silvia Sonderegger
A large theoretical literature argues laws exert a causal effect on norms, but empirical evidence remains scant. Using a novel identification strategy, we provide a compelling empirical test of this proposition. We use incentivized vignette experiments to directly measure social norms relating to actions subject to legal thresholds. Our large-scale experiments (n = 7,000) run in the United Kingdom, United States, and China show that laws can causally influence social norms. Results are robust across different samples and methods of measuring norms, and are consistent with a model of social image concerns where individuals care about the inferences others make about their underlying prosociality.
Economic Journal (2023)
School’s author: Giordano Mion
We develop a novel framework that simultaneously allows recovering heterogeneity in demand, quantity total factor productivity and markups across firms while leaving the correlation between the three dimensions unrestricted. We accomplish this by explicitly introducing demand heterogeneity and systematically exploiting assumptions used in previous productivity estimation approaches. In doing so, we provide an exact decomposition of revenue productivity in terms of the underlying heterogeneities, thus bridging the gap between quantity and revenue productivity estimations. We use Belgian firms’ production data to quantify total factor productivity, demand and markups, and show how they are correlated with each other across time and with measures obtained from other approaches.
In doing so, we find quantity total factor productivity and demand to be strongly negatively correlated with each other, so suggesting a trade-off between the quality of a firm’s products and their production cost. We also show how our framework provides deeper and sharper insights on the response of firms to increasing import competition from China. In particular, we find that changes in revenue productivity materialise as the outcome of complex and sometimes offsetting changes in quantity total factor productivity, demand, markups and production scale.
Opening heaven’s door: Public opinion and congressional votes on the 1965 Immigration Act
Journal of Economic History (forthcoming)
School’s author: Giovanni Facchini
The 1965 Immigration Act represented a radical shift in US policy, which has been credited with dramatically expanding the volume and changing the composition of immigration. Its passing has often been described as the result of political machinations negotiated within Congress, without regard to public opinion. We show that congressional voting was consistent with public opinion on abolishing the country-of-origin quotas but not with the desire to limit the volume of immigration. While the former initially reflected attitudes over civil rights, the latter is consistent with contemporary expectations that the expansion in numbers would be modest.
Journal of Business and Economic Statistics (2023)
School’s authors: David Harvey and Steve Leybourne
We develop tests for predictability that are robust to both the magnitude of the initial condition and the degree of persistence of the predictor. While the popular Bonferroni Q test of Campbell and Yogo displays excellent power properties for strongly persistent predictors with an asymptotically negligible initial condition, it can suffer from severe size distortions and power losses when either the initial condition is asymptotically non-negligible or the predictor is weakly persistent.
The Bonferroni t test of Elliott, and Stock, although displaying power well below that of the Bonferroni Q test for strongly persistent predictors with an asymptotically negligible initial condition, displays superior size control and power when the initial condition is asymptotically nonnegligible. In the case where the predictor is weakly persistent, a conventional regression t test comparing to standard normal quantiles is known to be asymptotically optimal under Gaussianity.
Based on these properties, we propose two asymptotically size controlled hybrid tests that are functions of the Bonferroni Q, Bonferroni t, and conventional t tests. Our proposed hybrid tests exhibit very good power regardless of the magnitude of the initial condition or the persistence degree of the predictor. An empirical application to the data originally analyzed by Campbell and Yogo shows our new hybrid tests are much more likely to find evidence of predictability than the Bonferroni Q test when the initial condition of the predictor is estimated to be large in magnitude.
Management Science (2023)
School’s author: Sourafel Girma
Laws mandating vaccination against Covid-19 as a condition of employment in health and care sectors were commonplace during the pandemic. Using weekly data at the local authority level, we examine the impact of the vaccine mandate for elderly care homes in England on vaccine take-up, staffing levels, and mortality. Our identification strategy involves 1. comparing take-up and staffing in English elderly care homes relative to other social care settings unaffected by the mandate; 2. comparing take-up and staffing in English elderly care homes relative to those in Wales where the mandate also did not apply; 3. comparing Covid-19 mortality among English elderly care home residents relative to mortality in domestic homes in England and to care homes in Wales.
Our results suggest that the mandate substantially decreased the proportion of care home workers who remained unvaccinated (equivalent to between 28,000 and 41,000 fewer unvaccinated staff), but this came at the cost of a reduction in staffing levels of between 3 and 4 percent (equivalent to 14,000 to 18,000 staff). We observe this effect most strongly in areas of low unemployment. Our results do not provide evidence that the vaccine mandate was successful in its primary aim of reducing care home deaths. Relatively wide confidence intervals mean inferences regarding mortality are more uncertain than for vaccination uptake and staffing.
American Economic Journal: Economic Policy (2023)
School's author: Guillermo Cruces
The canonical model of Allingham and Sandmo (1972) predicts that firms evade taxes by optimally trading off between the costs and benefits of evasion. However, there is no direct evidence that firms react to audits in this way. We conducted a large-scale field experiment in collaboration with a tax authority to address this question. We sent letters to 20,440 small- and medium-sized firms that collectively paid more than US$200 million in taxes per year. We find that providing information about audits significantly affected tax compliance but in a manner that was inconsistent with Allingham and Sandmo (1972).
The Review of Economics and Statistics (2023)
School's authors: Simon Gaechter and Chris Starmer
We introduce "group cohesion" to study the economic relevance of social relationships in team production. We operationalize measurement of group cohesion, adapting the "oneness scale" from psychology. A series of experiments, including a pre-registered replication, reveals strong positive associations between group cohesion and performance assessed in weak-link coordination games, with high-cohesion groups being very likely to achieve superior equilibria. In exploratory analysis, we identify beliefs rather than social preferences as the primary mechanism through which factors proxied by group cohesion influence group performance. Our evidence provides proof-of-concept for group cohesion as a useful tool for economic research and practice.
Review of Economic Studies (2023)
School's author: Martin Jensen
Rich behavioral biases, mistakes and limits on rational decision-making are often thought to make equilibrium analysis much more intractable. We establish that this is not the case in the context of one-sector growth models such as Ramsey-Cass-Koopmans or Bewley-Aiyagari models. We break down the response of the economy to a change in the environment or policy into two parts: the direct response at the given (pre-tax) prices, and the equilibrium response which plays out as prices change.
Our main result demonstrates that under weak regularity conditions, regardless of the details of behavioral preferences, mistakes and constraints on decision-making, the long-run equilibrium will involve a greater capital-labor ratio if and only if the direct response (from the corresponding consumption-saving model) involves an increase in aggregate savings. One implication of this result is that, from a qualitative point of view, behavioral biases matter for long-run equilibrium if and only if they change the direction of the direct response. We provide detailed illustrations of how this result can be applied and generates new insights using models of misperceptions, self-control and temptation, and naive and sophisticated quasi-hyperbolic discounting.
Review of Economics and Statistics (forthcoming)
School's authors: Paul Mizen and Gregory Thwaites
We analyse the impact of Covid-19 on productivity using data from an innovative monthly firm survey that asks for quantitative impacts of Covid-19 on inputs and outputs. We find total factor productivity (TFP) fell by up to 6% during 2020-21. The overall impact combined large reductions in 'within-firm' productivity, with offsetting positive 'between-firm' effects as less productive sectors, and less productive firms within them, contracted. Despite these large pandemic effects, firms' post-Covid forecasts imply surprisingly little lasting impact on aggregate TFP. We also see significant heterogeneity over firms and sectors, with the greatest impacts in those requiring extensive in-person activity.
Journal of the European Economic Association (2022)
School's author: Seung-keun Martinez
Understanding the structure of time preferences allows for accurate predictions of the effects of changing intertemporal incentives. Behavioral models of present bias are used to rationalize field data seemingly at odds with exponential discounting, leveraging additional degrees of freedom to improve in-sample fit. Largely lacking to date are the critical out-of-sample tests necessary to ensure predictive accuracy.
This paper contrasts exponential discounting with present-biased procrastination for around 22,000 tax filers, advancing the literature in this domain by providing novel out-of-sample tests for both theories. Present bias provides qualitatively better in-sample fit, matching substantial increases in filing probability as the tax deadline approaches. Present bias also has improved out-of-sample predictive power for responsiveness to the 2008 Stimulus Act, and experimental data demonstrate a link between present bias and filing times. Without present bias, predicted responses to changed incentives are inaccurate, demonstrating its necessity in research and policy applications.
Review of Economics and Statistics (2022)
School’s author: Giordano Mion
We investigate how managers help firms grow by entering a new export market. We conduct an event study on the decision to export to Angola using data on Portuguese firms and workers. We evaluate the impact of the presence of managers with experience in exporting to the Angolan market on a firm's entry success in the aftermath of an exogenous shock: the sudden end of the Angolan civil war. We show that the presence of managers doubles the probability of a firm entering the market. We do not find any significant impact on the intensive margin of exports.
Journal of Business and Economic Statistics (2022)
School’s author: Lorenzo Trapani
We propose a family of CUSUM-based statistics to detect the presence of changepoints in the deterministic part of the autoregressive parameter in a Random Coefficient Autoregressive (RCA) sequence. Our tests can be applied irrespective of whether the sequence is stationary or not, and no prior knowledge of stationarity or lack thereof is required. Similarly, our tests can be applied even when the error term and the stochastic part of the autoregressive coefficient are non iid, covering the cases of conditional volatility and shifts in the variance, again without requiring any prior knowledge as to the presence or type thereof. In order to ensure the ability to detect breaks at sample endpoints, we propose weighted CUSUM statistics, deriving the asymptotics for virtually all possible weighing schemes, including the standardized CUSUM process (for which we derive a Darling-Erdős theorem) and even heavier weights (so-called Rényi statistics). Simulations show that our procedures work very well in finite samples. We complement our theory with an application to several financial time series.