Documents de travail

In order to investigate strategic interactions between a “global north” and a “global south” we introduce a two-country extension of the model in Golosov et al. (2014). We consider different transfers between the two regions, including transfers that can improve the abatement technology. Our model can accommodate several kinds of heterogeneity, including in preferences, time discount rates, and damages resulting from the stock of accumulated GHG. We solve for both planner’s solutions and non-cooperative equilibria. We then calibrate our model in order to study quantitative differences between these solutions and to quantitatively explore the role of heterogeneity and Knightian uncertainty. We characterize emissions, damages, consumption, transfers, and welfare by computing the Nash equilibria of the associated dynamic game. We then compare these to efficiency benchmarks. Further, we investigate how (deep) uncertainty affects climate outcomes. We develop a general model for the study of optimal control and differential games that are linear-in-state, which we term the Integral Transformation Method (ITM), which encompasses several existing models as special cases.

We link French households’ tax records to the corporations they control, and build a payout-policy–neutral income measure, with corresponding tax burdens including those of "billionaires": the top 0.0002%. De- fined as such, income is more concentrated than taxable income, it better predicts rich-list membership, and persists more among billionaires. Personal taxes remain progressive until the top 0.1%, but eventually decline to 2% of income. Corporate taxes are an imperfect progressive backstop, as total tax rates fall from 45% at the 0.1% threshold to 25% for billionaires. Among these, the tax burden is global and tax-efficient pyramidal control over businesses ubiquitous.

This paper tests experimentally how preferences for redistribution of members of the general public depend on how money is earned. An experiment was designed to form of “microparticipatory-democracy”where redistribution from winners to losers is decided through a sequential strategy-proof majority voting procedure. Based on five distributive justice theories, we elicit people’s preferences for redistribution when their earnings come from four factors: effort, social circumstances, brute luck, and option luck. In the aggregate, our results show that a relative majority of people agree with Dworkin’s cut, namely, to compensate for social circumstances and brute luck but not effort and option luck. Participants with bad outcomes are more likely to engage in a self-serving vote, but on average, the dominant concern in voting remains people’s fairness view. The knowledge of the distribution of earnings and petition for equality of opportunity make participants vote more in favor of redistribution.

During the Covid-19 pandemic, the Omicron wave was notable for its highly transmissible and contagious variant of concern, coinciding with the availability of a vaccine that has been rolled out well earlier. In this paper, we address two key questions. First, we seek todesign a simple epidemiological model that can best capture the dynamics of Omicron infections. We demonstrate that combining the SIRDand SISD models provides an adequate solution. The second question examines the benefits of vaccination, in terms of both economicactivity and lives saved, once the model is implemented. Our results show that without vaccination, the human cost would have been fivetimes higher, and production losses would have doubled, due to stricter con- finement measures and a higher death toll. We also quantify the cost of vaccine hesitancy at more than 8,000 extra deaths.

Models of social learning conventionally assume that all actions are visible, whereas in reality, we can often choose whether or not to advertise our choices. Inthis paper, I study a model of sequential social learning in which social agents choose whether or not to let successors see their action, only wanting to do so if they are sufficiently confident in their choice (they are timid), and noise agents act randomly. I find that in sparse networks, this produces a form of unravelling to the effect that noise agents are overrepresented. This can damage learning to an arbitrary extent if social agents are sufficiently timid. In dense networks, however, no such unravelling occurs, and the combination of noise and timidity can facilitate complete learning even with bounded beliefs.

We propose a new random method to minimize deterministic continuous functions over subsetsSof high-dimensional spaceR K without assuming convexity. Our procedure alternates between a Global Search(GS) regime to identify candidates and a Concentrated Search (CS) regime to improve an eligible candidate in the constraint setS. Beyond the alternation between those completely different regimes, the originality of our approach lies in leveraging high dimensionality. We demonstrate rigorous concentration properties under theCSregime. In parallel, we also show thatGSreaches any point inSin finite time. Finally, we demonstrate the relevance of our new method by giving two concrete applications. The first deals with the reduction of theℓ1−norm of a LASSO solution. Secondly, we compress a neural network by pruning weights while maintaining performance; our approach achieves significant weight reduction with minimal performance loss, offering an effective solution for network optimization.

Prior work finds that individuals are often less prosocial when they can exploit uncertainty as an excuse. In contrast to prior work that largely explores the relevance of excuses in the gain domain, this paper investigates the relevance of excuses in both the loss and gain domains. In our laboratory experiment, participants evaluated risky payoffs for themselves and their partners in either the gain or loss domain, with or without interpersonal trade-offs. We found that participants exhibited excuse-driven risk behaviors in both domains. We also documented significant individual heterogeneity in the degree of excuses, influenced by factors such as individuals’ risk preferences, beliefs about others’ risk preferences, and the size of the risk.We present a self-signaling model that incorporates self-image concerns to explain our experimental findings. We show that excuse-driven risk behavior arises because people misattribute their selfish behavior to risk preferences rather than a reduced level of altruism.

Political campaigns influence how voters prioritize issues, which in turn impacts electoral outcomes. In this paper, we study how candidates’ communication shapes which issues prevail during the campaign, through which mechanisms, and to what extent. We develop an electoral competition model with two candidates, each endowed with exogenous platforms and characteristics. Candidates allocate strategically their communication time across two issues to maximize their expected vote shares. We find that when one candidate holds similar comparative advantages on both issues, the disadvantaged candidate communicate on a single issue to saturate the campaign with one topic and then increases the randomness of the election. The advantaged candidate has the opposite incentive and communicate on both issues, creating an asymmetry in the campaign. We show that in some cases, the campaign can become entirely centered on a single issue.

Between 1954 and 1998, the tobacco industry funded more than 1,900 research projects at a total cost of $355 million, on topics such as the roles of heredity and nutrition in cancer. Even though legitimate, this research was intended to divert attention from the harmful effects of tobacco. We provide the first formal analysis of such diversion research. We show that special interests may have strong incentives to affect the scientific agenda, even when the research itself is unbiased. This form of scientific lobbying yields large welfare losses and raises concerns about the private funding of research.

This paper introduces the African Debt Database (ADD) -a new, comprehensive dataset that traces both domestic and external debt instruments at a granular level. The main innovation is a detailed mapping of Africa's domestic debt markets, drawing on rich, new data extracted from government auction reports and bond prospectuses. The database covers over 50,000 individual government loans and securities issued by 54 African countries between 2000 and 2024, amounting to a total of USD 6.3 trillion in debt. For each instrument, it provides harmonized micro-level information on currency, maturity, interest rates, instrument type, and creditor. The data reveal the growing dominance of domestic debt in Africa -albeit with substantial cross-country variation. Four stylized facts stand out: (i) the rapid expansion of domestic debt markets, especially in middle-income countries; (ii) the wide dispersion in borrowing costs and real interest rates; (iii) large cross-country differences in maturity structures and associated rollover risks; and (iv) a rising debt-service burden, particularly due to international bonds. Generally, this project shows that debt transparency is both feasible and valuable, even in data-scarce environments.

This paper extends the standard two-period prevention model by incorporating anticipatory emotions. We introduce an additional cost, referred as the emotional load, which is endogenously determined by future risk but can be mitigated by current preventive effort. We show that a more intense emotional load incentivizes the emotional agent to increase investment in either self-insurance or self-protection. By contrast, greater uncertainty sensitivity has an ambiguous effect: It depends on the curvature of the emotional load function and wealth. When savings are substitutes, the effect of these parameters may diverge, whereas they align when savings are complements to risk prevention. Finally contrasting our setting with a setting without uncertainty or emotions, we show that, under prudence, the introduction of a zero-mean risk leads to a higher optimal level of self-insurance. Anxiety amplifies the incentive to reduce risk by lowering present well-being.

We examine how the 2015 relaxation of China's one-child policy affected marriage outcomes. Before the reform, some groups were already permitted to have two children. In China, where the sex ratio is heavily skewed toward men, being exempt from the one-child constraint may have been a desirable characteristic for marriage, increasing men's marriage odds. Using detailed policy data on exemptions and individual data from 2010-2018, we find that after the relaxation, men previously allowed a second child are less likely to marry compared to those not allowed. There is no effect for women. The results suggest that differential fertility constraints distorted who got married by advantaging certain men when there was a demand for a second child and strong marriage competition. Furthermore, suggestive evidence shows that the relaxation increased matching by education when exemptions were moderately widespread, indicating that fertility constraints also shaped who married whom.

This paper provides a simple uniÖed axiomatic framework for appraising the central tendency of distributions of a single attribute (pie) among a collection of individuals depending upon the available measurement of the attribute. Two types of measurement are considered: cardinal and ordinal. For each of them, three properties are posited on an ordering of distributions of numbers among individuals. The two first properties are the anonymity requirement that permutations of the same list of numbers be equivalent and the weak Pareto requirement that a strict increase in the value of the variable for all individuals be favorably appraised. The third property requires that inverting the numerical measurement of the variable leads to an inversion of the ranking of the any two distributions to which the inversion is applied. The mean of a distribution is shown to be the only ordering of distributions consistent with cardinal measurability that satisÖes those three requirements in the cardinal context while the median is the only such ranking consistent with ordinal measurability of the variable that satisÖes those same requirement if the number of individuals is odd. If the number of individuals is even, then those three requirements applied to the ordinal context are shown to be inconsistent.

Agricultural policies in poor rural developing countries have the potential to improve both household nutrition and agricultural income. But can these policy consequences be reconciled? This is not obvious because many policies are deficient. Moreover, in villages, mismatches have been observed between nutrition and profit indicators. However, incomes raised by such policies may generate nutrition improvement. In Niger, a major program directed to agro-pastoralists is the 3N Initiative. Do these policies enhance households’ agricultural profit and dietary intakes? And if so, is it because of an income effect, or through alternative channels? Using an agropastoral survey conducted in 2016 Niger, we find that livestock extension services that reduce calorie intake while improving diet diversity operate mostly through an increased household’s pastoral profit. In contrast, veterinary services and low-cost livestock feed programs improve diet diversity, but do not affect profit and calories. Because livestock extension services foster households specializing in cattle and sheep rearing and sometimes switching to transhumance, they restrict their access to energy-dense cereals. This generate a perverse consequence on caloric intakes, despite rising animal calories. Therefore, nutritional policy-makers should better account for agro-pastoralist access to cereal markets and monitor whether policies generate differential incentives, especially through profit, for specific specialization or lifestyle.

This paper develops a dynamic model of addiction on networks, where individuals’ consumption is shaped by peer influence. We analyze the longrun effects of social interactions by characterizing steady-state consumption as a function of both network position and forward-looking behavior. We also examine the welfare implications of network structure and evaluate the effectiveness of various public policies aimed at reducing the demand for addictive goods. In particular, we study a key-player policy—modeled as a targeted rehabilitation program—that leverages the network’s interpersonal influences to maximize impact.

Using data from a selection of Latin American countries affected by El Ni˜no-Southern Oscillation climate phenomena, we observe that extreme weather events can be highly disruptive for an economy, particularly in the agricultural sector, while also giving rise to inflationary pressures. Motivated by these findings, this paper examines the optimal stabilization policies for a climate-vulnerable economy with two segmented sectors: agriculture (producing food) and manufacturing. In response to climate disasters affecting agriculture, it is found to be optimal to increase fiscal transfers to farmers while maintaining core inflation at its target level. Deviating from the optimal policy mix results in smaller welfare losses as long as core inflation remains stabilized.

This paper analyzes how collateral quality shocks affect banks’ liquidity management and the risk-free rate. We develop a model where banks manage liquidity through near-cash assets and marketable securities subject to idiosyncratic and/or aggregate shocks. Collateral quality deterioration leads to non-monotonic changes in liquidity holdings: moderate declines reduce cash holdings via lower market returns, while severe declines cause precautionary hoarding and market freezes. Reduced collateral quality depresses the risk-free rate. Policy interventions, including liquidity regulation and negative interest rate policies can mitigate these effects. Our findings highlight the risks of collateral quality shocks and the importance of policy complementarities in addressing liquidity issues.