# The Liquidity Channel of Fiscal Policy

*Christian Bayer, Benjamin Born, Ralph Luetticke*

## Abstract

We provide evidence that expansionary fiscal policy lowers return differences between public debt and less liquid assets—the liquidity premium. We rationalize this finding in an estimated heterogeneous-agent New-Keynesian model with incomplete markets and portfolio choice, in which public debt affects private liquidity. This liquidity channel stabilizes fixed-capital investment. We then quantify the long-run effects of higher public debt and find little crowding out of capital, but a sizable decline of the liquidity premium, which increases the fiscal burden of debt. We show that the revenue-maximizing level of public debt is positive and has increased to 60 percent of GDP post-2010.

Local projection specification: $x_{t+h}=β_{0}+β_{1}t+β_{2}t_{2}+ψ_{h}gg_{t}+Γ(L)Z_{t−1}+u_{t+h}$

$Z$ includes vector of controls, including four lags of everything. Note the use of a quadratic time trend.

Makes the point that including lags means that the variable in question represents a shock and is therefore plausibly causal in its influence, a point also made by @blanchardEmpiricalCharacterizationDynamic2002:

“Under the Blanchard and Perotti (2002)-predeterminedness assumption, the coefficient $ψ_{h}$ provides a direct estimate of the impulse response at horizon h to the government spending shock in t.” (Bayer et al., 2020, p. 6) (pdf) “This is equivalent to a two-step approach, where $g_{t}$ is first regressed on lags of itself and additional covariates and the residual is then included in step 2 as the shock measure.” (Bayer et al., 2020, p. 6) (pdf)

This work uses levels of all the variables along with their lags and seems to arrive at sensible results, which I’ve generally found to be harder to achieve:

Transclude of Local-projections#c91898