- Marginal propensities to consume with behavioural agents [Nov 22]
I use survey questions on spending in hypothetical scenarios to study differences in marginal propensities to consume out of positive and negative income shocks. I find a sizeable asymmetry in spending responses driven by large adjustments following income losses compared to small adjustments following income gains. This asymmetry occurs irrespective of the position in the wealth distribution and is not driven by liquidity constraints. As such, it stands in stark contrast to predictions of traditional models of consumption behaviour. I develop a model with dissaving-averse households, a behavioural feature consistent with mental accounting, that rationalizes this asymmetry and simultaneously generates low MPCs out of income news, low MPCs out of wealth and high MPCs out of income gains for households that are not liquidity-constrained. I show through the lens of a quantitative model that asymmetric MPCs dampen the effectiveness of redistributive fiscal policy.
Work in progress
- The macroeconomic implications of coholding liquid assets and debt, with M. Boutros [Draft available upon request]
This paper highlights the importance of the joint distribution of liquid assets and debt in understanding the consumption response of households to income changes. We show that grouping households across the distribution of liquid wealth, as is typically done, confounds two very different types of households. True hand-to-mouth households with low liquid wealth due to low liquid assets, and households with low liquid wealth due to high debt. The former type has a high marginal propensity to consume while the latter type has a high marginal propensity to repay debt. We add a cash-in-advance constraint to a standard consumption-savings model which generates the co-holding of liquid assets and debt observed in the data and matches the empirically observed marginal propensities to consume and repay debt. We apply our model to the study of stimulative fiscal policy and illustrate the role that the joint distribution of assets and debt plays in the aggregate marginal propensity to consume.
- Domestic inequality and global imbalances, with J. Mazza [Draft available upon request]
We document that countries with higher income inequality have larger current account surpluses. A 1 percentage point increase in the share of income earned by the top 1 percent is associated with a 0.4-0.7 percentage point higher current account. We rationalize this finding through a two-country heterogeneous agent model and show that, all else equal, capital flows from unequal to equal countries. Non-homotheticities in preferences generate higher savings rates by rich households which translate into current account surpluses under international capital mobility. We illustrate how developed financial markets can offset the effects of inequality on the current account as evidenced by the persistent deficit of the United States.
- Equity duration and monetary policy transmission, with J. Graeb
- The Reliability of Equilibrium Exchange Rate Models: A Forecasting Perspective,
with M. Ca’Zorzi, A. Cap and M. Rubaszek
International Journal of Central Banking (2022)
ECB Working paper version, Replication files, Slides for ISF21
Coverage: Econbrowser, Macrohive
In this paper we evaluate the predictive power of the three most popular equilibrium exchange rate concepts: Purchasing Power Parity (PPP), Behavioral Equilibrium Exchange Rate (BEER) and the Macroeconomic Balance (MB) approach. We show that there is a clear trade-off between storytelling and forecast accuracy. The PPP model offers little economic insight, but has good predictive power. The BEER framework, which links exchange rates to fundamentals, does not deliver forecasts of better quality than PPP. The MB approach has the richest economic interpretation, but performs poorly in forecasting terms. Sensitivity analysis confirms that changing the composition of fundamentals in the BEER model or modifying key underlying assumptions in the MB model does not generally enhance their predictive power.