Working papers
WARNING: Section under construction.
"Corrigendum: Long-Term Contracting With Time-Inconsistent Agents," with D. Gottlieb, P. Siconolfi and X. Zhang, Econometrica online 2023.
"Contracting with Time-Inconsistent Consumers Under Constraints," 2025, with D. Gottlieb, P. Siconolfi and X. Zhang (a revised version coming soon).
"Naive stochastic present bias and credit," with P. Siconolfi (Supplemental Appendix)
The first note shows that, in models with consumption sets unbounded above, existence of perception-perfect equilibrium is restricted to low levels of lifetime income, or unbounded utility. It characterizes the exact range where existence occurs. As shown in Gottlieb and Zhang (2021), when the equilibrium exists, inefficiencies due to naive present bias vanish as the contracting length grows arbitrarily large. Naivete may help a present-biased consumer.
In the second paper, we want to study how the naive present-biased consumers' income affects their borrowing behavior and welfare. Data suggest higher income consumers have disproportionately higher credit card balances (are more "revolvers") than lower income borrowers. Higher income consumers are also the target of financial firms lending. However, as the note above shows, previous work cannot generally deal with high income consumers. We introduce upper bounds to the consumer's consumption set, and show that the equilibrium contract is characterized by the presence or absence of an imaginary max out. The imaginary max out occurs when firms cannot back load perceived consumption any further into the future (cannot offer repayment schedules that look attractive since the consumers believe they will not procrastinate), and instead start offering earlier cash incentives or additional borrowing lines. The imaginary max out is more likely to occur for 'higher' payroll incomes, and is associated with over borrowing. When this is the case, we show that inefficiencies due to naivete may not asymptotically vanish. Upper bounds can arise, for example, from existing regulations which limit the amount of credit a lender can offer (single-line limits) due to prudential concerns. They can also result from physical limits to per period consumption due to time capacity (as in Becker's 1965 theory of time allocation), or by the observation that consumers may be satiated at unaffordable but finite consumption levels (a standard assumption in microeconomics since Ramsey's 1928 work). Affected consumers have relatively high income levels but still insufficient to make the credit limit affordable.
The third paper is a general model of borrowing under quasi-hyperbolic (beta-delta) preferences and false beliefs. Following well-known 'state/trait' duality in IIVP theories, or epigenetic theories of genes expression, or simply allowing for measurement errors in dual-self theories of procrastination, we allow for the procrastination parameter beta to be stochastic. Within this framework, we show that the prudence-to-risk aversion ratio determines borrowing behavior and welfare. With a high-enough ratio, naive consumers over-save. With low ratios, they may over-borrow or over-save, depending on how optimistic they are: When their beliefs are sufficiently close to the truth, naive consumers over-borrow no matter their lifetime income; when their beliefs are far from the truth, and they are close to fully naive beliefs, they may end up over-saving if their income is low. We provide a new, and robust, reason for market inefficiency: Long-run inefficiencies arise irrespective of income levels or of whether there is imaginary max out. Naivete does not help palliate the negative consequences of present bias when it is stochastic. In the process, we define pessimism, and characterize pessimistic beliefs.
Recently I have gone back to working on optimal fiscal policy. The current version is here:
"Do taxspots matter?," with M. Tvede.
It is understood that a general principle of optimal taxation is to smooth out consumption in the face of government spending fluctuations, aiming at tax certainty. In fact, tax uncertainty is pervasive, in and outside the circle of OECD countries, as shown by an IMF/OECD 2017 survey report. Generally, tax uncertainty has many sources (from random audits, tax laws interpretations, tax procedures to legislative uncertainty, as the political process produces changing tax environments), and is considered to negatively affect investments and welfare. However, evidence of the negative welfare effects of tax uncertainty is ambiguous (see, e.g., Bizer and Judd (1989), or references in the IMF/OECD report).
In this paper, we study if and when fiscal policy should introduce uncertainty in the economy -a phenomenon we dub 'taxspots'. Our focus implicitly is on legislative uncertainty, which affects all taxpayers at the same time. We find that capital taxes should be uncertain when prudence is large enough relative to the economy's risk aversion -a condition that requires, but is stronger than, 'nonconvexities' in the Ramsey problem. In economies with hand-to-mouth workers, taxspot uncertainty is persistent when a redistribution towards workers is desirable. Taxspots are not necessarily substitutes for positive average capital taxes, though 'front loading' taxspot policies Pareto improve on positive taxspot-free capital taxes. When workers are imprudent, average tax decreases with taxspots, other things equal.
Our taxspots are reminiscent of, but not equal to, 'sunspots', as fiscal policy is made to depend on 'higher-order beliefs' or 'sentiments' as defined in, e.g., Angeletos and La'O (2020); beliefs which otherwise would have no relevance in equilibrium.
The work on optimal fiscal policy is somewhat related to my previous work on taxes when markets are incomplete, and to my more recent excursions on the topic of bubbles. There the focus was on explaining credit volume collapses (such as occurring during the 2008 financial crisis). With G. Bloise (JET 2019) we show that credit collapse is not a necessary consequence of a recrudescence in agency problems ('manipulation of ARM mortgages') or of the arrival of bad news. Instead, the effect on credit volumes actually depends on how much the economy needs to escape autarky. Thus, when there are some minimal gains from trade at autarky, deterioration of agency problems or the arrival of bad news are not sufficient to create credit market crashes. Our result extends similar conclusions obtained by Fahri and Tirole (2012) in the context of an OLG economy. Our theory of bubble formation does not require abrupt changes in the equilibrium or other discontinuities for bubbles to arise. Our result also shows how studies of liquidity-constrained markets via approximation using no-trade equilibrium (see, e.g., Werning, 2015) can be misleading. [Strangely, you will not find reference to our JET paper in the sequel, appeared on TE, by Bloise, Polemarchakis and Vailakis (2021)...]
"Envy-free Pseudo-Markets for Indivisible Goods," with M. Tvede (very preliminary version).
In this paper develop a pseudo-market mechanism for allocating indivisible goods in environments with limited or no monetary transfers, complementarities, endowments, and assignment constraints. Our design delivers computable, fully envy-free, individually rational and ex-post feasible probabilistic allocations which are always envy-free-efficient, approximately efficient, and approximately incentive compatible in large markets. We also prove a novel result on the existence of envy-free Lindahl equilibria, and show how to use it to compute approximately efficient equilibrium allocations. Thus, our work improves on market designs of Echenique, Miralles and Zhang (AER 2021), Gul and Pesendorfer (REStud 2025), Nguyen and Vohra (JPE 2025).