Fiscal Progressivity and the Time Consistency of Monetary Policy

The paper asks how progressive fiscal policy influences the conduct of monetary policy in a tractable heterogeneous-agent economy, in explicit contrast to views promoting a separation of objectives and instruments between the central bank and the fiscal authority. The author quotes a former Fed Chair’s position that policies designed to affect the distribution of wealth and income are appropriately the province of elected officials rather than the Fed, and that monetary policy is a blunt tool. Against this, the analysis shows how fiscal policy, through its capacity to tailor the incidence of taxes, contributes to supporting efficient monetary decisions. The specific monetary problem is the time inconsistency identified by Kydland and Prescott (1977) and Calvo (1978): nominal quantities such as interest rates and money holding are sensitive to expectations, but policies are implemented once expectations are locked in, and this intertemporal conflict gives rise to an inflation bias that generates welfare losses. The paper studies a novel institutional response to it — what the author calls the commitment channel of progressive fiscal policy.

The model is an overlapping-generations economy with flexible prices, in which a unit mass of agents born each period lives two periods and differs in lifetime labour productivity z, distributed on a compact set. Agents supply labour and save when young, and supply labour and consume when old; because the consumption good is perishable, fiat money is the nominal asset available for storing wealth. Preferences are linear-quadratic and production linear, a specification the author adopts deliberately: curvature in utility would capture a desire for consumption smoothing, insurance or redistribution, and its absence lets the analysis cleanly disentangle policy choices with and without redistributive concerns. The government finances an exogenous real level of public good either by taxing labour income of the old or by printing money. The labour income tax schedule τ(y) = λy^(1+α) is parameterised by a progressivity parameter α ≥ 0 and a level λ ≥ 0; α = 0 implements a flat rate and any α > 0 is progressive in the standard sense that marginal rates exceed average rates at all income levels. These plans generate no positive transfers, which the author notes allows the analysis to focus on redistributive conflicts between labour tax and seigniorage rather than on conflicts driven by labour taxation itself.

Tax inertia is the assumption, following Farhi (2010) and Ferriere (2015), that fiscal progressivity α is set one period prior to tax collection — reflecting the idea that the legislative process is complex and some structural elements of the tax code, particularly those governing the incidence of fiscal policy, require more time to adjust. The author supports this assumption with empirical evidence in an appendix: using the Standardized World Income Inequality Database, country fixed effects account for 98% of the total dispersion in relative redistribution across 70 countries over 1981–2021 (97% for the 38 OECD countries), compared with adjusted R-squared of 85% (75% for OECD) when market-income Gini is the dependent variable. The author reads this as suggesting that fiscal redistribution is a strong country-specific attribute despite variation in market income inequality over time.

Lemma 1 establishes that in the static problem of financing a public good using labour taxes only, the optimal plan prescribes no progressivity in both homogeneous and heterogeneous agent economies. The reason, following Werning (2007), is that for a given level of taxes to be collected, progressivity is only costly: it increases marginal tax rates, labour supply distortions, and both individual and aggregate welfare losses. This result is what makes the paper’s central finding non-obvious — progressivity has to earn its place through a dynamic channel, not a static one.

Proposition 1 establishes that under commitment the benevolent policymaker chooses no progressivity and balanced taxation, λ_c = 1 − π̃_c, while under discretion inflation is strictly positive and the labour tax is zero for any predetermined level of progressivity, and lifetime welfare for every agent is lower under discretion than under commitment. Under commitment the policymaker spreads the burden of taxation equally across agents and over time, with revenue coming equally from labour taxes and seigniorage. That plan is time-inconsistent: because real money holdings of the old are predetermined to tax choices, ex post inflation is beneficial because it operates much like a non-distortionary lump-sum tax. Inflation is therefore higher under discretion — a classic illustration of the inflation bias — and the welfare losses stem from the anticipation of inflationary policies and its negative effect on young agents’ labour supply and money demand. Crucially, in both regimes progressivity is not desirable, neither to reduce the deadweight loss of taxation nor to mitigate the inflation bias, because the utilitarian planner here has a pure productive-efficiency objective.

Policy is set through a two-stage game under tax inertia: when young, agents decide the progressivity of labour taxes that will prevail next period; when old, majority voting over a standard two-candidate protocol determines the mix of labour taxes and seigniorage, given the predetermined progressivity and the distribution of money holding. Because the decisions of young agents internalise their effects on the outcome of the vote, the game is solved backwards. The choice of progressivity in young age is set behind a veil of ignorance — before agents learn their individual productivity — but it reflects individual preferences and their effects on next period’s vote.

An old agent’s marginal valuation of raising the labour tax decomposes into two terms: a tax-distortion cost, given by the marginal tax rate the agent faces, and a tax-shifting benefit, given by the agent’s relative money holding (proportional to z²/E(z²)) multiplied by the marginal aggregate labour tax revenue. The second term captures the strategic dimension: raising labour taxes reduces the magnitude of the inflation tax and thereby preserves money holding as a source of consumption. The author notes this expression makes clear that the shape of the agent’s value function is not sensitive to aggregate money holdings, the inflation rate, or the level of public spending — the willingness of a type-z agent to raise labour taxes is tied to the distributional consequences generated by different levels of progressivity. The analysis assumes throughout that the relevant levels of labour tax lie on the upward-sloping part of the Laffer curve.

Lemma 3 shows that individual preferences are single-peaked, and that when α = 0 all agents share the same bliss policy of zero labour taxes, whereas for α > 0 agents disagree and their bliss policies are ordered by productivity type. Under a flat tax, agents unanimously vote to rely exclusively on the inflation tax — a result the author notes is stronger than the discretionary optimum of Proposition 1, since not only does aggregate productive efficiency prescribe exclusive use of the inflation tax, but agents unanimously support seigniorage to take advantage of the inelastic tax base. Under progressivity there is a productivity cut-off z̄(α), strictly interior to the productivity support, such that all agents below it prefer strictly positive labour taxes (with the preferred level strictly decreasing in productivity) and all agents at or above it prefer zero. Lower-productivity agents favour labour taxation because it collects relatively more from higher-productivity agents at low individual cost — the tax-shifting effect — while high-productivity agents, standing on the receiving end of that effect, support relatively more inflationary policies to minimise personal exposure to distortionary taxation.

Proposition 2 establishes that majority voting selects a unique policy mix in which the decisive voter is the median-productivity agent, so that the implemented policy relies exclusively on the inflation tax when α = 0 but implements strictly positive labour taxes for any α > 0. Uniqueness follows because permanent lifetime productivity makes individual type and real money holding perfectly correlated, so agents differ de facto along one dimension only, and single-peaked preferences over a unidimensional policy space induce a unique Condorcet winner. The proof that the median agent falls below the cut-off uses Jensen’s inequality to show z̄(α) > E(z), combined with the maintained assumption that the median productivity level is no greater than the mean. The upshot is that any level of progressivity α > 0 contributes to curbing the inflation tax.

No. Lemma 4 establishes that the aggregate tax function evaluated at the vote outcome is positive for all α ≥ 0, admits a global maximum, and eventually converges to zero as progressivity goes to infinity. At α = 0 no labour tax is collected; a positive level of progressivity induces the median agent to exploit the tax-shifting effect and raise labour taxes; as progressivity increases further, increasing labour supply distortions reduce the total amount of labour taxes collected. The author cautions that these curves should not be read as standard Laffer curves, because individuals’ favourite levels of progressivity lie on both the upward- and downward-sloping portions.

Lemma 5 establishes that every agent, of any productivity type, would favour a strictly positive level of progressivity, and Proposition 3 concludes that when progressivity is set behind a veil of ignorance the chosen level α_p is strictly positive and finite. The marginal welfare effect of progressivity for a young agent has three terms — the direct disincentive effect of progressivity, the overall distortions induced by labour taxes (whose magnitude depends on the agent’s position in the productivity distribution and hence its exposure to tax-shifting), and the marginal effect on the inflation tax. In the limit α ≈ 0 the decisive voter next period would rely essentially on inflation, so an increase in α decreases inflation and mitigates its adverse effects on young agents’ labour supply and money demand; the author states that this third force dominates at low levels of progressivity for every agent. Individual favourite levels α_p(z) are non-monotonic in z: for the least productive agents the marginal tax rate is zero for any λ whenever α > 0, so they favour the progressivity that maximises total labour taxes collected; a slightly more productive agent supports higher progressivity to further exploit tax-shifting while minimising own exposure; higher-productivity agents support lower progressivity because they internalise bearing a large welfare cost; and the highest-productivity agent favours just enough progressivity to balance distortions from inflation against those from labour taxes.

Under the conflictual political protocol, lifetime welfare for a given generation is higher than under benevolent discretionary policymaking. The author’s argument is that the choice of progressivity under both decision processes satisfies the same welfare criterion, and that the political protocol could have set the same instruments as discretion (no progressivity, maximum inflation bias) but instead selects progressive labour taxes to reduce the inflation bias — so redistributive conflicts motivate the introduction of progressive labour taxes that enhance lifetime welfare under a lack of commitment. More broadly, the author characterises the political outcome as implementing an allocation similar to the commitment benchmark “in the sense that” the inflation bias is contained and the burden of policy distortions is distributed more evenly over time — not as replicating it exactly.

Section 5 introduces incomplete markets and idiosyncratic productivity risk between young and old age, plus risk aversion and a lump-sum transfer in the government budget, in order to study how the commitment channel interacts with the classic insurance channel of progressive fiscal policy. Log productivity follows an AR(1) with persistence 0.96 and innovation standard deviation 0.169, calibrated to a US market Gini of 0.48; preferences are CRRA with risk aversion 2, a Frisch elasticity of 0.72 from Chetty et al. (2011), a discount factor of 0.96 at annual frequency, and labour disutility set so average hours worked equal one-third. Government parameters are progressivity of 0.499 (targeting a US after-tax Gini of 0.36), pure government expenses of 0.11 (US public consumption of 15%) and transfers of 0.05 (US public transfers of 7%). The model is calibrated so that policy choices are made by a benevolent policymaker under a lack of commitment; given the parsimonious OLG structure, the numerical solution delivers exact policy functions.

Under commitment, adding the calibrated progressivity raises lifetime welfare from the normalised 1 to 1.008 and reduces welfare dispersion from 1 to 0.951, isolating the insurance channel; under discretion without progressivity, the incentives to rely on inflationary finance generate a collapse of money demand, with lifetime welfare falling to 0.729 and welfare dispersion rising to 2.271, alongside output falling from 0.741 to 0.562 and average hours from 0.339 to 0.257. Discretionary policy choices with progressivity at 0.499 instead deliver lifetime welfare of 1.003 and dispersion of 1.019 — the author’s statement that progressivity contains the inflation bias and provides valuable insurance against productivity shocks. Under majority voting the pattern repeats: without progressivity the outcome is identical to discretion without progressivity, while majority voting with progressivity yields welfare of 0.996 and dispersion of 1.069, comparable to discretion with progressivity. The author reads this as confirming the qualitative equivalence between policy choices made by a planner with a concern for the distribution of consumption and those made under the political majority-voting protocol.

The author reports that the more persistent and volatile idiosyncratic shocks are, the more effective progressive labour taxes are at limiting the inflation bias, with similar conclusions applying to lower lump-sum transfers or to higher inequality aversion on the part of policymakers. The summary formulation is that the higher the need and desire to redistribute consumption, the more effective progressive fiscal policy is in containing the excesses of monetary discretion. Appendices additionally discuss how results generalise to intergenerational heterogeneity across multiple cohorts, to an ex-post cost of inflation, and to a richer asset structure.

The paper situates itself against three literatures. It builds on Albanesi (2007), which links income inequality and inflation as the outcome of a distributional conflict underlying monetary policy choices, and adds the claim that the incidence of fiscal policy — captured by labour tax progressivity — is critical to understanding monetary outcomes. It draws a contrast with Farhi, Sleet, Werning and Yeltekin (2012), where progressive capital taxation emerges as optimal because it contains the build-up of inequalities and the temptation to reduce them with a capital levy; the author notes that monetary decisions subject to similar time-inconsistency problems cannot be made progressive or targeted, and focuses instead on progressive labour taxes. And it distinguishes its commitment channel from the insurance channel of progressive income taxes studied by Conesa and Krueger (2006) and Heathcote, Storesletten and Violante (2017), while noting that the growing HANK literature — Auclert (2019), Kaplan, Moll and Violante (2018), Gornemann, Kuester and Nakajima (2021) — highlights how the effects of monetary decisions depend on the conduct of fiscal policy; the added claim here is that appropriate fiscal policy can curb the time inconsistency of monetary policy.

The results are derived in a stylized environment deliberately designed to isolate the distinctive forces at work, and the commitment channel of progressive labour taxes is effective in mitigating the inflation bias only if policymakers are concerned about the distribution of consumption. The linear-quadratic preference specification in the analytical sections rules out consumption smoothing, insurance and redistribution motives by construction — which is precisely what makes the benevolent-planner benchmark deliver no progressivity, and what makes the political-economy section’s distributional conflict the operative force. The within-generation budget constraint focuses the analysis on intragenerational conflicts. The analysis abstracts from the seigniorage Laffer curve indeterminacy by assuming private agents’ inflation expectations lie on the upward-sloping part of that curve, and imposes an upper bound on government spending to guarantee interior solutions.