Abstract - Public in-kind provision of education, healthcare, or culture accounts for a large share of government spending, even though redistribution can be achieved with cash transfers and externalities addressed with subsidies. We propose a new macroeconomic theory of in-kind provision, grounded in two features: (i) these goods are luxuries, with consumption rising more than proportionally with income; and (ii) they generate externalities that depend not only on the total level of consumption but also on its distribution across households. In a tractable heterogeneous-agent model, we show that these features make direct in-kind provision welfare-improving, even when cash transfers and subsidies are available. Using household- and country-level data, we document that most publicly provided goods are luxuries and present evidence consistent with distribution-sensitive externalities. We then apply the framework to the design of fiscal consolidation, showing in a quantitative model that optimal consolidation falls primarily on goods without private substitutes, while preserving in-kind provision of goods with private substitutes. Finally, our framework yields a welfare-based valuation that reduces the measured equalizing effect of in-kind benefits in France by about one third.

Prizes - Finalist at the 2026 CESifo Public Economics Affiliate Award, Best Paper Award at the 2025 Franco-German Fiscal Policy Seminar

Abstract - How do central bank losses affect output and inflation, and how should monetary policy respond? When rates rise, QE portfolios shift capital losses from households to the central bank, a debt-financed transfer that raises demand, especially if losses are deferred. Repayment through labor taxes is inflationary, and the monetary response can reverse the stimulus. In a US-calibrated HANK model, a loss of 3.8% of GDP causes a small expansion, then a persistent contraction costing 1.4% of GDP in present value. Deferral mostly shifts this cost in time. Fully offsetting inflation raises it to 2.2%; optimal policy cuts it to 0.4%.

Abstract - The distributive effects of carbon taxation are critical for its political acceptability and depend on both income and geographic factors. Using French administrative data, household surveys, and matched employer-employee records, we document that rural households spend 2.8 times more on fossil fuels than urban households and work in firms that emit 2.7 times more greenhouse gases. We incorporate these facts into a spatial heterogeneous-agent model with migration, housing and wealth accumulation, bridging spatial and macroeconomic approaches. Carbon taxes generate 56% larger welfare losses for rural households than for urban households. While uniform or income-based rebates reduce inequality along the income dimension, they leave large spatial disparities that must be offset through location-based transfers. These results suggest that carbon policies should account for spatial differences to improve political feasibility.

Prize - 2023 CEPR Philippe Martin Prize