Limited foresight turns optimal public debt and taxes stationary
A Federal Reserve research paper demonstrates that bounded rationality invalidates Barro's random walk hypothesis for public debt and taxes. Researchers Martin Bodenstein and Junzhu Zhao show that households perceive debt as wealth, making post-shock debt reduction optimal.
The wealth channel behind mean reversion
Under full-rationality models with non-state-contingent debt, optimal labor taxes and debt levels follow unit-root dynamics after shocks.
Bodenstein and Zhao show that cognitive discounting breaks this property except under knife-edge parameterizations.
When private agents discount the future, they perceive government debt as net wealth, which stimulates aggregate demand and creates persistent inflationary pressure.
Consequently, leaving debt elevated generates cumulative welfare losses from price dispersion.
To mitigate these distortions, optimal policy requires debt and taxes to revert to their initial paths, shifting fiscal stabilization toward tax adjustments rather than permanent debt increases.
Decoupling wealth effects from discounting
By analyzing an overlapping generations New Keynesian model with intergenerational transfers, the authors isolate the transmission mechanism.
They demonstrate that the wealth effect of government debt alone drives stationarity, whereas discounting in the dynamic IS equation is neither necessary nor sufficient.
With global public debt exceeding $100 trillion and United States federal debt reaching 122 percent of GDP, the paper offers a distinct rationale for debt consolidation that complements conventional concerns about rollover risks and fiscal solvency.
Elegant theory meets fiscal reality
The paper provides a rigorous theoretical case for retiring public debt after crises.
However, attributing fiscal dynamics entirely to bounded rationality downplays political obstacles to raising distortionary taxes.
While mathematically elegant, the model offers little guidance for real-world debt consolidation battles.