Federal Reserve · Policy Brief · 2018–2026
Two Measures. One Verdict.
Taylor Rule deviation (bars, left) and NGDP gap vs. 4% trend (line, right) · quarterly
I. Introduction
Kevin Warsh takes the Federal Reserve chairmanship with core PCE inflation above its 2% target and nominal GDP running 13% above its pre-pandemic trend path. His appointer has made clear that he regards the federal funds rate as a political instrument. Larger than the rate decision itself is the question before the new chairman: what framework he will use to decide, and whether he will commit to one at all.
This piece applies four competing monetary benchmarks to the current environment. Two of the oldest branches of the rules tradition contribute the Taylor Rule and Milton Friedman's k-percent rule. Both newer entrants share a different anchor: George Selgin's productivity norm and a nominal GDP level target evaluate policy against nominal spending rather than against inflation. None of the four agree on what Warsh should do with rates, and Figure 1 shows how far apart they sit. That disagreement is the point. A central bank operating under binding rules cannot select whichever rule is most convenient for the moment. A central bank operating under discretion always can.
II. Four Rules, Four Theories
The Taylor Rule anchors policy to two observable variables: the gap between actual inflation and the 2% target, and the gap between actual output and the economy's productive potential. Its institutional logic is that the central bank should respond systematically to what can be observed now and leave forecasts of unobservable future states out of the reaction function. Only as rules-based as the inputs that feed it, this is the rule's weakness. Both the neutral real rate and the output gap cannot be observed, only estimated, and the estimates carry enough uncertainty to absorb almost any policy preference.1 Set r* at 0.5% rather than 2.0% and the rule endorses rates more than 150 basis points lower at any given moment. That judgment is discretion wearing a formula's clothes.
Friedman's k-percent rule eliminates that problem by eliminating the real-time response entirely. Under this rule the central bank commits to a fixed annual growth rate for M2, in Friedman's own recommendation somewhere between 3% and 5%, and holds to it regardless of current conditions.2 Its premise is epistemological. Reliable real-time signals required to stabilize the economy lie beyond the Fed's reach, so it should not try. Its mandate under the k-percent rule is restraint. Precisely its indifference to the moment is the rule's virtue, which is also its political vulnerability.
Selgin's productivity norm addresses a problem the Taylor Rule mishandles. When supply-side productivity growth drives prices lower, the Taylor Rule signals that policy is too tight and prescribes accommodation, even though no monetary failure has occurred. Instead, the norm targets nominal spending and treats price changes driven by real shifts in productive capacity as benign. Its direct test case is the post-2020 episode. Pandemic supply disruptions the Fed read as a signal for continued accommodation would, under the norm, have been classed as a real adjustment and left alone.3
Of the four, the nominal GDP level target looks furthest forward. It evaluates policy against the cumulative path of nominal spending from a pre-crisis baseline; current inflation and the current output gap do not enter the assessment. Where the Taylor Rule asks whether the rate is right for this period, the level target asks whether the cumulative path of nominal spending has been made whole. A central bank that allowed NGDP to overshoot its level target in 2021 and 2022 owes a period of below-target growth to return to the path; one operating under discretion will find reasons to avoid collecting that debt.
III. What the Rules Prescribe Now
Federal Reserve · Policy Brief · 2018–2026
Federal Funds Rate vs. Taylor Rule Prescriptions
Effective federal funds rate and rules-based benchmarks · percent · quarterly
In its baseline specification with a real neutral rate of 0.5%, the Taylor Rule prescribes a federal funds rate of 4.8% for the first quarter of 2026. At 3.6%, the actual rate stood 1.1pp below that prescription. Under Taylor's original 1993 assumption of a 2.0% real neutral rate, the prescription rises to 6.3%, widening the gap to 2.6pp. Figure 2 plots both prescriptions against the actual rate. That spread across two plausible specifications is the problem. A discretionary institution will select the neutral rate assumption that makes current policy look least deviant. At 1.1pp, the baseline gap is the optimistic reading; the rule itself does not specify which reading to use.
Figure 2 shows where this divergence ran earlier. In the first quarter of 2022, with the effective federal funds rate still near zero and core PCE inflation at 5.6%, the baseline Taylor prescription stood 8.3pp above the actual rate. Steepest since the early 1980s was the hiking cycle that followed. A rules-based path would have begun the adjustment more than a year earlier and spread it across the cycle; the delay compressed it into roughly 14 months.
Federal Reserve · Policy Brief · 2019–2026
Nominal GDP vs. 4% Trend Path from 2019 Q4
Actual nominal spending vs. rules-based benchmark · billions USD · quarterly
Figure 3 provides the clearest view of the cumulative damage. Anchored to a 4% annual trend from 2019's final quarter, the level target shows actual nominal spending sitting 13.1% above the path as of the first quarter of 2026. That overshoot is still widening. A discretionary Fed can treat that as an acceptable new equilibrium; the level target has no provision for declaring the overshoot permanent.
M2 grew at 24.6% year on year in the fourth quarter of 2020, against Friedman's k-percent benchmark of 4%. At its widest, the gap reached 20.6pp. Current M2 growth of 4.6% sits near the benchmark, but the price level absorbed the earlier excess and has not recovered it. No mechanism exists within the k-percent rule for correcting a level deviation; it operates on the flow. Damage of that kind is in the stock.
Deviation from Selgin's productivity norm peaked at 15.5pp in the second quarter of 2021, when NGDP grew 17.4% year on year against potential output growth of 1.9%. Nominal spending under this norm would have been held to the rate at which the economy's real productive capacity was expanding. Instead, the Fed validated the overshoot as transitory and maintained accommodation. By the time the FOMC acted, the inflation the norm was designed to prevent had already arrived.
None of the four benchmarks converges on a single prescription for Warsh. Both Taylor variants put the appropriate rate between 4.8% and 6.3%, while the level target reads the cumulative path as uncorrected, which means the tightening is not finished. Dating the failure to 2020 and 2021, the k-percent and Selgin diagnostics imply that the price level will not surrender the excess without a deliberate path correction. Commitment to any one of these frameworks would give Warsh a defined problem. Without commitment, the four prescriptions are a menu.
Methodology
Sample. Quarterly, 2018 Q1 through 2026 Q1. Monthly series take the last observation of each quarter. All series are drawn from FRED.
Taylor rule.
$$i_t = \max\{\, r^* + \pi_t + 0.5(\pi_t - \pi^*) + 0.5\, y_t,\; 0 \,\}$$
with $\pi^* = 2\%$ and $y_t$ the CBO output gap, computed as $100 \times (\text{GDPC1} - \text{GDPPOT}) / \text{GDPPOT}$. Baseline variant: $r^* = 0.5\%$, consistent with post-crisis Laubach-Williams estimates; original variant: $r^* = 2.0\%$ (Taylor 1993). Prescriptions are floored at zero. Inflation is core PCE, PCEPILFE (year-on-year % change); the policy rate is the effective federal funds rate, FEDFUNDS. Full-precision 2026 Q1 values: baseline prescription 4.79%, original prescription 6.29%, actual rate 3.64%, gaps 1.15pp and 2.65pp.
k-percent rule. M2, M2SL, year-on-year growth against a fixed 4% benchmark, inside Friedman's 3% to 5% range. Peak deviation 20.55pp in 2020 Q4 (M2 growth 24.55%).
NGDP level target. Nominal GDP, GDP, against a 4% annual trend compounded quarterly from the 2019 Q4 level. Gap as of 2026 Q1: 13.09%, the widest in the sample.
Productivity norm. NGDP year-on-year growth minus potential GDP (GDPPOT) year-on-year growth. Peak deviation 15.48pp in 2021 Q2 (NGDP growth 17.37%, potential growth 1.89%).
IV. What Warsh Inherits
Three institutional conditions compound one another in what Warsh inherits.
First is damaged credibility. Visible as the 2021 transitory call, the underlying condition is the absence of any binding nominal anchor that would have forced earlier action regardless of the qualitative read on the shock. A central bank with a rules-based framework cannot choose to describe inflation as transitory and defer its response; the rule responds whether or not the explanation is satisfying. Sustained across four consecutive quarters of rising inflation, the Fed's latitude to make that call was a product of operating under discretion. Credibility cost, in this case, is the loss of the prior that the Fed will act on the price level rather than on its own narrative about the price level.
Second is political exposure. Trump's pressure on rates has historical precedent running from William Martin through Arthur Burns to G. William Miller. Absent now is any institutional framework robust enough to absorb it. A discretionary institution under political pressure is exposed by design: the same flexibility that allows it to respond to changing economic conditions allows it to respond to political ones. A rules-based institution can point to the benchmark. Warsh can point to nothing, unless he creates something.
Internal fragmentation runs underneath both. Powell's exit record makes clear that the committee's disagreement ran deeper than timing; it concerned inflation theory itself, beginning with where the neutral rate sat and whether the supply-shock framing was ever correct. Even the speed at which expectations would de-anchor was contested. Discretion leaves that disagreement unresolved and converts it into inconsistent signals and ex post rationalization. A binding nominal anchor would leave the theoretical question open. It would make the practical one answerable by anyone with the data in front of them: how far is the actual rate from the rule?
V. The Framework Question
Warsh has written favorably about rules-based monetary reform. His 2016 Wall Street Journal piece argued that the Fed had grown dependent on unreliable models and erratic judgment, and that the institution had left itself politically vulnerable in the process.4 Whether that conviction survives the institutional logic of discretion is the question his tenure poses. That logic rewards flexibility and punishes commitment, and it has absorbed every predecessor.
Four rules applied here do not converge on a number. That disagreement is the strongest possible argument for committing to one framework publicly, in advance, so that the next deviation can be measured against something fixed rather than against whichever rule happens to justify the convenient choice.
At the Sound Money Project's "Building a Better Fed Framework" conference in December 2024, the most serious recent record of what a rules-based reform agenda looks like in practice was produced.5 Warsh's transition is the first real test of whether that conversation produces institutional traction. Its answer will be visible in the framework, or in the continued absence of one.
Footnotes
- Holston, Laubach, and Williams (2017), "Measuring the Natural Rate of Interest: International Trends and Determinants," Journal of International Economics 108, S59–S75. Their confidence intervals around r* estimates are wide enough, in many periods, to make the Taylor prescription indeterminate within a range of several percentage points. ↩
- Friedman (1960), A Program for Monetary Stability, Fordham University Press. ↩
- Selgin (1997), Less Than Zero: The Case for a Falling Price Level in a Growing Economy, Institute of Economic Affairs, Hobart Paper 132. ↩
- Kevin Warsh, "The Federal Reserve Needs New Thinking," Wall Street Journal, August 24, 2016. ↩
- American Institute for Economic Research, "Building a Better Fed Framework" monetary conference, George Washington University, December 2, 2024. ↩
Get new essays and research by email — subscribe on Substack (opens in new tab)