• Battigalli, P., Panebianco, F., and Pin, P. 2023. “Learning and selfconfirming equilibria in network games”, Journal of Economic Theory, 212, 105700. 

Consider a set of agents who play a network game repeatedly. Agents may not know the network topology. They may even be unaware that they are actually interacting with other agents in a network. Possibly, they just understand that their optimal action depends on an unknown state, that is, actually, an aggregate of the actions of their neighbors. In each period, every agent chooses an action that maximizes her instantaneous subjective expected payoff and then updates her beliefs according to what she observes. In particular, we assume that each agent only observes her realized payoff. A steady state of the resulting dynamic is a selfconfirming equilibrium given the assumed feedback. We identify conditions on the network externalities, agents’ beliefs, and learning dynamics that make agents more or less active (or even inactive) in steady state compared to Nash equilibrium.

Indeed, Nash equilibrium action profiles are limit outcomes of learning paths when agents have perfect feedback about the payoff-relevant aspects of others’ behavior. Yet, as we argue, such perfect feedback hypothesis may be too strong for some social networks applications and, if learning is based on imperfect feedback, non-Nash action profiles may result as the steady-state limits of learning paths.

In our analysis we assume that the only feedback agents receive is their realized payoff. This implies that they do not always identify the payoff-relevant aspects of the actions of others, represented by a payoff state. We analyze how agents use the feedback they receive to update their conjectures about the payoff state and best respond to them, and we characterize their limit behavior under different settings of local and global externalities. We show that if externalities are only local, active players are always able to infer the correct payoff-relevant state, whereas inactive players fall into an inactivity trap, and stay inactive even when it would be objectively optimal not to do so. We study sufficient conditions over the network topology to ensure existence and stability of the selfconfirming equilibria. Conversely, if also global externalities are at play, these externalities play a confounding role and a multiplicity of equilibria arise where an arbitrary set of agents can be inactive and consistently decide to stay so, and the level of activity depends on the level of each agent’s perceived centrality in the network. In particular, players may be more active if they think that they are more linked in the network than they actually are, and this can be welfare improving for the whole society. Also, agents with excessive perceived connectedness may prevent convergence of best reply paths to interior equilibria.

All these conditions are based on the network structure and on the type of externalities, on the conjectures that agents have, and on the rules that they use to update their conjectures. Thus, in some applications, knowing these conditions, a social planner or the owner of the network can try to change the beliefs of people to induce them to increase their activity levels.

• Parodi, F. “Taxation of Consumption and Labor Income: a Quantitative Approach”, forthcoming at AEJ: Macroeconomics.

The uniform commodity taxation theorem by Atkinson and Stiglitz (1976), one of the most influential results in Public Economics, implies that dfferentiated consumption tax rates are not needed if an optimal non linear income tax is in place. Yet, in practice, most countries heavily rely on differentiated commodity taxation even if they have sophisticated progressive income tax systems. 
In order to bridge this gap between tax theory and tax practice, in this paper I quantitatively characterize the optimal consumption and income tax rates using an estimated structural life-cycle model of households' consumption, saving and labor supply choices. The model relaxes the key assumptions of the Atkinson-Stiglitz theorem that could be driving the uniform commodity taxation result. First, I allow for multiple consumption goods with different degrees of necessity and durability. In particular, I include durable goods that are partially irreversible and enter the borrowing constraint. Second, I consider rich preference heterogeneity. Third, I take into account horizontal equity objectives of the social planner. 
I derive the optimal tax system under the assumption of a utilitarian social welfare function. I find that durables should be subsidized and non-durables should be taxed at a uniform rate. The durable subsidy is driven by the life-cycle features of the model together with durables irreversibility and borrowing constraints. Uniform taxation on non-durables holds under exogenous and endogenous fully or weakly separable labor supply and it relies on homogeneity of intertemporal preferences. Lastly, allowing for government’s horizontal equity concerns, I show that the model can rationalize the tax practice under a high degree of government’s inequality aversion.

• Sandi, M., with T. Kirchmaier, S. Machin, and Robert Witt. 2020. “Prices, Policing and Policy: The Dynamics of Crime Booms and Busts”, Journal of the European Economic Association, 18 (2): 1040–1077.

In many historical episodes, criminal activity displays booms and busts. One recent, very clear, example of crime boom and bust has been the case of metal crime. In the UK, starting from the mid-2000s metal theft more than doubled, reaching unprecedented levels by 2011. However, as with other (crime) booms, a bust followed this as metal theft fell very rapidly since the start of the current decade. 
This paper researches the dynamics of metal crime. The empirical analysis starts by examining whether the changing prices of metals affect the level of crime over time. Administrative records of metal crime from the British Transport Police (BTP) and the Metropolitan Police Service of London combined with data on metal prices from 2007-2015 reveal sizeable crime-price elasticities and a strong sensitivity of metal crimes to scrap metal prices. 
The subsequent parts of the paper investigate the impact on metal crime of the responses by the institutions in the UK. In 2011, the BTP conducted an anti-metal crime police initiative called “Operation Tornado” (OT). The UK government also responded by introducing the Scrap Metal Dealers Act (SMDA) in 2013. Our analysis uncovers a significant negative effect of OT on metal crime, showing the critical role of policing in shaping crime dynamics. The analysis also shows the significant impact of the SMDA 2013 on both the economic activity of Scrap Metal Dealers (SMDs) and metal crime. Following the SMDA 2013, SMDs experienced a significant reduction in their economic activity. A genuine reduction in metal crime also originated from the stricter regulation regime of the SMDA 2013. 
Therefore, prices, policing and policy matter for crime, a finding of relevance not only for the market of metals, as lessons can be learnt also on the crime-reducing potential of institutional responses (e.g. stricter regulation) in other industries.

• Ghisolfi, S. 2022. “Market Access and Quality Upgrading: Evidence from Four Field Experiments”, forthcoming in American Economic Review

Smallholder farming in many developing countries is characterized by low productivity and low quality output. Low quality limits the price farmers can command and their potential income. We conduct a series of experiments among maize farmers in Uganda to shed light on the barriers to quality upgrading and to study its potential. First, we document that quality is indeed low but partly observable with easy and quick procedures by the buyers, therefore it would be possible to reward farmers based on the quality of their produce. Second, we experimentally vary the quality of the maize produced and show that in the existing market the causal return to quality is zero, suggesting that the market for quality maize is effectively missing. Third, to understand if farmers would be willing and able to produce high quality maize when rewarded, we contract with a Ugandan vertically integrated agro trader to generate experimental variation in access to a market for premium quality maize. Following 200 farmers in 20 villages for 7 seasons (i.e. 3.5 years), we document that access to a market for high quality maize, combined with training on agricultural best-practices, produces large increases in maize quality (+60%) and in both farm productivity and income, increasing yields by 15% and mean profits by $63-$98 per season or 36%-81%. While the intervention also increased prices for lower quality maize in the treatment villages, the evidence suggests that the increase in output in the treatment group can fully be accounted for by the subgroup of farmers selling premium quality maize to the high quality buyer. Fourth, in a separate trial following 170 farmers for 6 seasons, we investigate the impact of the training program alone and find no evidence that farmers changed their farm practices, productivity or profits. Our findings reveal the importance of demand-side constraints in limiting rural income and productivity growth.

• Cipullo, D. 2021. “Political Fragmentation and Government Stability. Evidence from Local Governments in Spain”, American Economic Journal: Applied Economics

Unstable governments have been historically associated with fragmented parliaments. Topical cases include the Weimar Republic in Germany and the Fourth Republic in France. Despite the abundance of examples, to date, we lack rigorous evidence on whether the association between fragmentation and stability is indeed causal. The main empirical issue lies in finding exogenous variation in fragmentation, especially when using national-level data. Also, occurrences of instability events at the national level are too uncommon to apply reliable econometric techniques.

A recent work by Davide Cipullo (Università Cattolica), with Carozzi (LSE) and Repetto (Uppsala University), forthcoming in the American Economic Journal: Applied Economics exploits data from 50,000 local elections in Spain (featuring more than 1,000 events of unseated governments), where parties can only enter the municipality council if they receive more than 5% of the votes. Using a Regression-Discontinuity design, the authors compare municipalities where a party receives just less than the required quota – and remains out from the council – with municipalities where a party receives a few more votes than the required quota – and may enter the council.

The empirical results document that the entry of an additional party causes a 5- percentage points increase in the probability that the incumbent mayor is unseated via a vote of no confidence. In relative terms, for each additional party that enters the municipality council, incumbent mayors are 140% more likely to be replaced during the term. Using the empirical results and a simulation method, the authors also show the policy impact of admission thresholds on the stability of incumbent governments. Low entry thresholds (e.g., those requiring a party to receive at least 3% of the votes to participate in the seat distribution) do not reduce instability compared to not having an admission threshold at all. 2 On the contrary, relatively high thresholds (e.g., those requiring at least 5% of the votes) may reduce occurrences of instability substantially.

• Le Moglie, M. 2020. “Revealing “Mafia Inc.”? Financial Crisis, Organized Crime, and the Birth of New Enterprises”, Review of Economics and Statistics

Organized crime has strong negative effects on the economy, the society, and the institutions of many countries, but despite these detrimental effects, and the dramatic increase in the public resources deployed to contrast organized crime, it has followed to proliferate. A possible explanation for this persistence relates to the fact that organized crime also represents an important actor within the economy of many countries, generating gains for a part of the population that partially counterbalance its negative impact and reduce the effectiveness of law enforcement.

A recent work by Le Moglie and Sorrenti, now forthcoming at the Review of Economics and Statistics, exploits the 2007-subprime mortgage crisis to detect Mafia's presence in the Italian legal economy. The credit contraction homogeneously hit all the geographic areas, while Mafia presence was stronger in some provinces and weaker in others; Mafia sources of profit and capital were barely dented, as UNODC records, allowing organized crime to invest as usual. If the Mafia invests in the legal economy, the number of newly-established enterprises should thus have decreased less in Mafia-ridden areas than in the rest of the country. That’s exactly what happened, according to the number of firms newly registered in each province in the Registry of Enterprises. The drop was 4% lower, corresponding to a provincial average of 241 enterprises established every year in the post-crisis period due to Mafia investment in the legal economy (28 for every 100,000 inhabitants), and it lowers even further when considering the sector (construction) and legal form (limited liability) most preferred by Mafia.

The analysis suggests that standard repression policies against organized crime need to be complemented by institutional interventions, e.g. provision of credit or programs to enhance employment opportunities, to undermine the roots of the social consensus obtained through its investment in the legal economy.