Research Seminars

Our aim is to examine behaviour and equilibrium selection in two similar, indefinitely repeated games, Stag Hunt and Prisoner’s Dilemma under anonymous random matching. We are interested in the role that historical precedents may play for equilibrium selection between these two repeated games. Duffy and Fehr (2018) find no significant spillover effects in this setting but they only vary the temptation parameter. We revise their experiment through the approach of Mengel (2018) and the theory of LiCalzi and Mühlenbernd (2019) and want to explore whether varying the efficiency parameter can alter their results.

We discuss the effects of offshoring on the labor market in a matching model with endogenous adjustment of educational skills. We carry out a comparative statics analysis and show that offshoring leads to a restructuring of the economy through skill-biased technical change (SBTC) where overall welfare is improved. In a policy exercise we show that, if offshoring were to be opposed by a protectionist agenda, labor market flexibility can bring about the same welfare gain. In addition, we offer an empirical analysis aimed at verifying the correlation between offshoring and SBTC in US manufacturing industries in recent years. Our results show that different offshoring strategies affect SBTC differently. In particular, the evidence suggests that while high-skill offshoring strategies open the skill gap, low-skill offshoring strategies tend to work in the opposite direction.

Website

Have you ever wondered how are colonies of ants able to efficiently search for food, birds or fish collectively defend themselves against predators by forming flocks or schools, and why crowds of people behave like a fluid? Recent advances in automated tracking technology resulted in high-resolution recordings of individual trajectories and behavior of groups, often complemented by identification of stereotypical behaviors. But the main hurdle still remains to be data analysis and inference of informative models. I will talk about a class of probabilistic models, which is general enough to apply to a broad range of systems, incorporating individual and collective behavior, spatial and temporal dependence, discrete and continuous variables, deterministic and stochastic components and internal cognitive or behavioral state dependence. Our approach has two desirable features: (1) the maximum likelihood inference is tractably solvable by gradient descent, (2) model selection can be used to adjust model complexity to data. Multiple toy/real examples will be shown during the talk.​

This paper studies the origins and consequences of the Russian Mafia (vory-v-zakone). Using a web scraping method, I obtained a unique dataset that contains detailed biographies of more than 5,000 mafia leaders operating in 15 countries of the (former) Soviet Union at some point between 1916 and 2018. Using this data, I first show that Russian Mafia originated in the Gulag - Soviet system of forced labor camps which housed around 18 million prisoners between 1920s-1950s. Second, I document that the distance to the nearest camp is a strong negative predictor of mafia presence in Russia's communities in the mid-1990s. Finally, using an instrumental variable approach which exploits the spatial distribution of the gulags, I examine the effects of mafia presence in mid-1990s on local crime and violence. In particular, I show that the communities with mafia presence experienced a dramatic rise in crime driven by turf wars which erupted among rival clans around 1993 and lingered on until the late 1990s. This is suggested by a sharp increase in attacks against members of Russia's economic elite in places under mafia control.  Further heterogeneity analysis reveals that mafia presence led to a spike in violence against businessmen, fellow criminals, as well as law enforcement officers and judges, while the attacks against politicians remained unaffected.

Website

In a 1-year randomized controlled trial involving thousands of university stu- dents, we provide real-time private feedback on relative performance in a semester-long on- line assignment. Within this setup, our experimental design cleanly identifies the behavioral response to rank incentives (i.e., the incentives stemming from an inherent preference for high rank). We find that rank incentives not only boost performance in the related assignment, but also increase the average grade across all course exams taken over the semester by 0.21 standard deviations. These beneficial effects remain sizeable across all quantiles and extend beyond the time of the intervention. The mechanism behind these findings involves social learning: rank incentives make students engage more in peer interactions, which lead them to perform significantly better across the board. Finally, we explore the virtues of real-time feedback by analyzing a number of alternative variations in the way it is provided.

Paper
Website

We study a real effort environment, where a delegator has to decide if and to whom to delegate a task. Delegating to a person who is better at the task increases welfare. Potential delegatees send cheap-talk messages about their past performance before the delegator decides. We experimentally test the theoretical prediction that information transmission cannot occur. In our experiment, we vary the message space available to the delegatees and compare the information transmitted and the level of efficiency. Depending on the treatment, the sender can either submit a number indicating how many tasks she solved previously, an interval in which the number of tasks falls, or a free text message. We observe that messages contain information in all treatments. Interestingly, information transmission occurs only in the treatments where messages are intervals or free text but not if messages are exact. The highest efficiency is obtained in the free-text treatment, as delegators are able to extract information contained in the different styles of messages sent by subjects with different abilities.

In credence goods markets, experts have better information about the appropriate quality of treatment than their customers. Experts may exploit their informational advantage by defrauding customers. Market institutions have been shown theoretically to be effective in mitigating fraudulent expert behavior. We analyze whether this positive result carries over when experts are heterogeneous in their diagnostic abilities. We find that efficient market outcomes are always possible. However, inefficient equilibria can also exist. When such inefficient equilibria are played, a larger share of high-ability experts may lead to more inefficiencies relative to the efficient equilibria.

Website

Over the last years, the use of experimental research has received an increasing attention from Management academics and journal editors. While experimental research is popular among social psychologists (actually my background) it is still rare in the field of International Human Resource Management. The objective of this seminar is not to explain how to design an experimental research, but instead, present my own experience of using experimental designs to answer a few common questions in the management and business fields, such as: (1) Does the academic performance (GPA) and the participation in extracurricular activities (ECAs) affect the perceived employability of business graduates? (2) Does the effect of GPA and ECAs vary with the characteristics of the respondents and the cultural context? (3) Does a facultative internship affect the perceived employability of marketing graduates? What about the effect of an international versus a domestic facultative internship? (4) Does the use of a facial piercing influence the perceptions of interpersonal attraction, confidence and job suitability of hospitality receptionists? Starting from my first published paper (Pinto & Ramalheira, 2017) employing an experimental design, I then illustrate how this method was applied more broadly to examine the perceived employability of business graduates in other cultural contexts (e.g. China, Brazil, Italy) and to advance research in other fields, such as hospitality management, higher education and leadership. The seminar ends with a discussion of the challenges and learnings from employing (quasi) experimental designs.

Website

Reporting a peer’s wrongdoing to an authority is a morally controversial decision: re-porting will likely result in a punishment for the peer (and harming others is unethical), but the punitive action might prevent further harm to the victims of the misconduct (which justifies reporting on a moral ground). A policymaker might want to encourage the solution of this moral dilemma in the direction of reporting, through financial incentives. Yet, material incentives sometimes backfire, especially in domains involving moral concerns. Using a field experiment with employees of the Ministry of Education in Afghanistan who are asked to confidentially report their colleagues’ absence, I show that reporting is lower among participants who are offered a monetary reward for their reports. This is the case, however, only for participants who expect their reports to be consequential (making their choice morally-charged). Among employees who are told instead that their reports are inconsequential and won't result in any penalty for their peers (making the choice more morally-neutral), incentives do not backfire.

Website

Abstract: OECD and other international organizations have been very keen in recommending principles and institutional safeguards to curb corruption and to enhance transparency and integrity in public procurement. Despite the fact that Portugal is being considered a good example of e-procurement policies and practices among European countries, this is a very sensitive issue.
Based on the the literature that provides evidence of a politicized administration of public procure- ment contracts, but limited to a specific municipality, this paper extends the analysis of political effects to other municipalities. Specifically, it asks if there is a relation between the political parties in power in a given municipality and the frequency of contracts awarded to a given firm?
Our results show that for political reasons private firms are more likely to win a contract in a given municipality if they have already won contracts in other municipalities led by the same political party. We rely on a dataset (’base.gov’) with information on all bids by private firms and all contracts awarded by the 308 Portuguese municipalities in the period between 2008 and 2017. This includes three electoral cycles and more than 250,000. The empirical results - the political proximity - is robust to a number controls, including geographic proximity. This result has political and public governance implications.

Download paper
Website

Foreign-born individuals come with different health needs and different level of vulnerability. In the public debate today, immigration is often viewed as a threat to the access and the quality of health care services. The health needs of immigrants and refugees pose new challenges to health care systems. We discuss the main findings of the effects of immigration on demand and supply of health care in host countries. Moreover, immigrant inflows could have large effects on labour markets which can in turn affect natives’ health and their demand for health care. Understanding the health trajectories of immigrants are paramount to provide a correct assessment of the costs and benefits of migration.

website

There is increasing acceptance of the role played by mood in decision-making. However, the evidence of mood effect in the field, as proxied by external factors such as weather and weekday, is limited. Using two large datasets containing all car inspections in Sweden and England during 2016 and 2017, we study whether inspectors are more lenient on days when their mood is predicted to be good, and if car owners exploit this mood effect by submitting lower quality cars to be inspected on these days. Different sources of good mood are studied: Fridays, last working days before public holidays, paydays, sunny days, and days following major sports results, in a decreasing order of the car owner’s ability to plan for inspection, and hence the likelihood of selection bias. We find limited evidence to support the existence of mood effect in this domain, despite survey results showing belief to the contrary. In Sweden, failure rate is in fact higher on Fridays and last days before holidays, suggesting that the selection effect is stronger than any mood effect.

website

This paper focuses on the relationship between experimentally measured willingness to compete and field data. Our sample consists of more than 1500 Swiss compulsory school students from the whole ability distribution. We elicit their willingness to compete in one of two tasks which differ in associated gender stereotypes. We relate our experimental measure to field data on ability and to students' choice of post-compulsory education. This enables us to make contributions in three directions. 1. We examine how willingness to compete varies with field-measured ability and whether this differs by gender. 2. We investigate the relationship between gender, willingness to compete and study choices along the whole ability distribution, in a comprehensive framework that includes specialization options both in the academic and the vocational track. 3. We examine whether the results depend on the task that is used to elicit willingness to compete. Our main findings are: 1. High-ability boys are more willing to compete than low-ability boys while the relationship between ability and willingness to compete is flat for girls. 2. Willingness to compete predicts choices both of academic specializations and of vocational careers. 3. The results are similar across our two experimental tasks.

website

In recent years, there has been growing opposition against the principle of free movement of labor within the EU. Criticism is often based on the belief that immigrants hurt residents' employment opportunities. Despite these discussions, there is little ex-post research on the impact of the increased immigration after the EU enlargement on old Member States' labor markets. In our paper, we first provide a descriptive analysis of employment from the new EU Member States in the Austrian labor market with a focus on the period around EU entry and free labor market access. Second, we exploit the observed patterns in immigration from the EU8 countries to identify the causal effect on workers and firms in the Austrian border region. More precisely, we use variation in the EU8 worker density across communities over time and over the distance to the closest EU8 border. We find that the share of EU8 employees among all employees in Austria increased by a factor of four from 1997 to 2015. With free access, we see a shift in the composition of migrants toward lower-qualified and younger groups. We can further show that the largest inflow of EU8 employment occurred in communities closer to the border.

website

This paper provides strong evidence supporting the long-standing speculation that decision-making in groups has a dark side, by magnifying the prevalence of anti-social behavior towards outsiders. A large-scale experiment implemented in Slovakia and Uganda (N=2,309) reveals that deciding in a group with randomly assigned peers increases the prevalence of anti-social behavior that reduces everyone’s but which improves the relative position of own group. The effects are driven by the influence of a group context on individual behavior, rather than by group deliberation. The observed patterns are strikingly similar on both continents.

Download paper

This paper illustrates how a historical shock to political preferences can translate into observable electoral support as the political landscape evolves. During World War II, the Third Reich annexed the French eastern borderlands and their inhabitants were forcibly conscripted into the Wehrmacht.
In the first stage, survey data evidence is used to show how this forced conscription reduced political trust, affecting policy preferences. The data is then used to estimate the impact of conscription on municipality-level support for radical candidates and on abstention in elections during the 1965-2017 period. Identification exploits the fact that different birth cohorts were affected in each annexed region by using eligible births as an instrument for conscription. In earlier elections in which platforms were more similar, both radical and moderate candidates were penalized in municipalities where more men were conscripted, resulting in higher abstention. In more recent elections which were more polarized, conscription increased support for radical candidates.

website

We investigate how New York City taxi drivers respond to positive changes in labor demand. Exploiting high-frequency variations in taxi demand due to subway service disruptions, we show that drivers work more when earnings opportunities are greater. We also explore whether income-targeting affects drivers’ behavior. Results show that drivers’ response to demand shocks is 40% smaller once they have reached their daily income target. Overall, while drivers’ behavior seems largely consistent with the standard model of labor supply, the large difference between below-target and above-target responses suggests that targeting behavior does nevertheless play an important role in determining drivers’ decisions.

website

This paper analyzes the link between effort and distributive preferences in an environment, in which effort does not affect the amount to be distributed. We propose a model that suggests that such a link is bidirectional. People adapt their distributional choices to their performance in a self-serving way, but they also exert effort in line with their distributive preferences. The literature has documented a link running from effort to distributive preferences. We provide evidence of the reverse relationship: individuals who make egalitarian choices later make less effort than people who behave selfishly. Our results thus provide one explanation for self-serving assessments of fairness documented in the literature and place distributive preferences among the determinants of effort and productivity.

Abstract: In this project, we aim to experimentally examine the trade-off between the loss of information and the loss of control that arises in organizations when subordinates are granted decision-making authority. Employees' discretion over the choice of a project is often justified in that it may allow the organization to make use of the employee's superior information about the right course of action. These benefits may be undermined by (i) the existence of a conflict of interest between the employer and the employee; (ii) the parties' inability to write a complete contract. The employer may be concerned about the loss of control, which arises when the employee abuses his authority to make decisions that are not in the best interest of the organization. The organizational optimal response will be that of limiting the agent's discretion, thereby giving rise to a loss of information, namely in some occurrences, the organization cannot make use of the agent's superior information. We argue that limiting the employees' discretion may signal the employer's distrust and may bring about a non-material loss to the employees, making them willing to retaliate by making suboptimal choices. Anticipating this, the employer may decide to grant the employees full discretion, signaling her trust in the hope that the employees will pursue the common good.
(Joint work with Philip Brookins, Claudia Cerrone and Alessandro De Chiara.) 

Personal website

Abstract: The effects of research and development (R&D) subsidies on patenting are heterogeneous in nature. Moreover, endogeneity caused by a multi-step selection of grant applicants makes empirical evaluation difficult. We use a unique dataset on the evaluation of grant applications and a novel instrumental variable identification strategy, which originates from the grant evaluation process. WE estimate the causal effects of R&D subsidies in the Czech Republic over 2011-2014 on patenting in the next three years. We exploit the random assignment of experts, who evaluate grant applications, and their leniency to give higher scores as an instrumental variable. As a result, we show that R&D subsidies have a strong positive mid-term effect on the propensity to apply for patents and the number of patent applications. R&D subsidies lead to the patent applications of higher quality, i.e. firms are more likely to apply for patents of invention than utility models. The main findings are driven to a large extent by firms with higher prior research intensity.

Personal website

We provide a new measure of work motivation and show that motivation shapes the effects of team incentives and observation by peers on performance. In particular, we measure motivation to work hard as the deviation from the money-maximizing benchmark in a real-effort experiment. While we find that average output increases in response to team incentives and observation, we find that highly motivated workers do not respond. The reason is that highly motivated workers already work hard and increasing effort even further is very costly to them.

Download paper

This paper examines the impacts of extremely hot and cold days on mortality in Russia, using a 25-year regional panel data. Unlike other studies, the sequence of those extreme days is also taken into account, that is, the impacts of both single and consecutive (i.e. heat waves and cold spells) extreme days are estimated simultaneously. We demonstrate the importance of accounting for the sequence of extreme days. We also disentangle the impacts of those extremes by age and gender. The findings suggest that single hot days increase mortality, while single cold days do not affect mortality. On the other hand, both consecutive hot and consecutive cold days increase mortality in females and males for all age groups, although males are affected more severely. Overall, consecutive days with extreme temperatures impose considerable costs to society in terms of years of life lost. Thus, ignoring the sequences of extreme days that are likely to increase in the future because of climate change may have critical implications for mitigation policies.

Download paper

Abstract: Regulatory investigations by Self-Regulatory organizations (SROs) have been recognized to usually be cheaper than investigations by the government. However, in practice, oversight by an SRO is mostly still supplied with forms of governmental oversight. The government may exert oversight over the SRO itself, a construction referred to as “meta-regulation" or "co-regulation", or over the members of the SRO. Indeed, the overall performance of SROs has been mixed and theoretical models show that SROs have incentives to set lax standards or cover up detected violations. However, some research indicate that meta-regulation, oversight of the SRO itself, may nonetheless not be necessary in some settings. Using a costly-state-verification model, DeMarzo et al. (2001; 2005) show that when the government implicitly threatens to perform additional investigations of the SROs members, a relatively "good" outcome can be established as an equilibrium. In this "good" outcome, the SRO chooses to follow high performance standards in order to pre-empt any of the (relatively costly) governmental investigations. As a result, no costly governmental investigations of the SRO's members take place, and no meta-regulation of the SRO is necessary.
I extend this model to include plausible settings where the actual rigor of oversight by the SRO can be verified only ex-post. I show that in such settings, the SRO may have incentives to announce stricter regimes than it effectively implements and that, as a result, a "bad", Pareto-inefficient outcome is established as an equilibrium. In the "bad" outcome, the SRO relinquishes all oversight to the government. The predictions of this model are supported by experimental tests. The "good" equilibrium can be re-established as an equilibrium with sufficient meta-regulation of the SRO. The results thus indicate a continuing need for meta-regulation in these settings. This form of meta-regulation may be of a relatively light-handed nature, limited to verifying and sanctifying that the SRO implements its announced policies.

In this project, we investigate the effect of children’s development on parental labor market outcomes. Using an instrumental variable approach and accounting for sample selection, we find that mothers of boys with development difficulties (but not fathers) significantly reduce weekly working hours during the first three years of school. In contrast, mothers of girls with development problems reduce working hours only during the first year of school but increase their hours later on. We find that the reduction in working hours is associated with a fall in weekly earnings, although our estimates lack precision. Investigating potential channels, we find evidence that mothers increase the general time spend with their children as a response to reduced working hours. We do not find evidence, however, that the extra time is of higher quality.

We ran a population-level randomized field experiment to ascertain whether a costless manipulation of the informational content (restricted or enhanced information) and the framing (gain or loss framing) of the invitation letter to the national breast cancer screening program affects the take-up rate. Our experiment involved more than 6,000 women aged 50- 69 targeted by the screening program of the Province of Messina in Sicily, randomly assigned to receive different invitation letter formats. Using administrative data from the Local Health Authority archives, we show that giving enhanced loss-framed information about the risks of not having a mammography increases take-up rate by about 25 percent with respect to all other treatments (no information; restricted gain-framed information; restricted loss-framed information; enhanced gain-framed information). Results are stronger for subjects living farther away from the screening site. For them, the manipulation may indicate higher perceived risks of negative outcomes that makes it worthwhile to participate in the screening program, in spite of longer travel time.

Download paper

In this paper we study the effects of acute stress on individual financial speculative behavior using a controlled laboratory experiment with 208 men and women. We employ a recently introduced measure that captures individual speculative behavior, the Speculation Elicitation Task, and an efficient stress-inducing procedure, the Trier Social Stress Test for Groups, and we pay special attention to the gender-specific effects. Our design allows for a separation of the main channels behind the treatment effects. We observe strong gender differences: The treatment – stress-inducing – procedure increases men's willingness to speculate compared to control men, but it decreases it for women by about the same amount. As we do not observe any change in the task-specific risk-preferences, concentration, and only a little change in the strategic expectations of shift in others' behavior and in beliefs, we conclude that the behavioral change is driven by the change in preferences, although in the opposite directions for both genders. The analysis of salivary cortisol and subjective mood shows that the subjects were under a considerable level of stress.

Download paper

Year 2018 marked the 40th anniversary of the first ever Tax Evasion Game published in the Journal of Public Economics by Friedland, Maital, and Rutenberg. This game has since been the subject of many other publications (more than 140 articles). In the present study, we collect more than 60 datasets recreating a Tax Evasion Game and run a meta-analysis on more than 220,000 observations and 15,000 subjects. The aim here is to set the impact of three types of variables (public policy, experimental context and sociodemographic of participants) on lab tax compliance.
The public policy variables are: tax rate, tax regime (progressive or proportional taxes); type of audit (endogenous or random audit); audit probability; fine size, and amnesty. The experimental variables are: framing of the experiment (loaded or not); way to ask for compliance in the instructions (relaxed or not); origin of income (earned or windfall); nature of income (self-employed or salaried job); redistribution to participants, public good fund, and pool of subjects (students or taxpayers). The sociodemographic variables are: age; gender; income; and studies.

What makes couples with daughters more likely to divorce than couples with sons? Using Dutch registry and U.S. survey data, we show that daughters are associated with higher divorce risks, but only when they are 13-18 years old. These age-specific results rule out explanations involving overarching son preferences and selection. Our findings are consistent with causal mechanisms involving relationship dynamics in families with teenage children. Survey evidence buttresses this interpretation. Subsample analyses show that the magnitude of the effect is linked to parental gender norms and that the effect is absent for fathers who grew up with sisters.

Personal website

Mathias Czaika is a Professor in Migration and Globalisation at Danube University Krems in Austria. Mathias was formerly Director of International Migration Institute (IMI) at the University of Oxford in the UK.

Personal website

Whether it is copyright infringement, hate speech or terrorist content, Internet intermediaries like Facebook, Twitter or YouTube are expected to essentially do the government's job -- enforcing the law. The legal scheme under which a lot of such delegated enforcement takes place is often referred to as notice \& takedown. According to empirical evidence, it leads to over-notification, over-compliance by providers and under-assertion of rights by affected content creators. We re-create these existing problems in a laboratory and then test a mechanism to address two of them: the overcompliance by providers, and the lack of complaints by creators. Through experiment, we show that our proposed solution gives more power to the users, who realize higher profits due to an improved accuracy of providers' assessment of content and leads to a significant reduction in over-compliance by providers.