Oracle, Prisoner, Navigator
A Bayesian Examination of the Federal Reserve, Forward Guidance, and Credibility under Radical Uncertainty
Farid Novin
September 28, 2026
Abstract
Commentary on the Federal Reserve often reaches for two images from Greek myth: the Delphic oracle, to mock the confident forecast, and Odysseus bound to the mast, to explain why a central bank might wish to tie its own hands. This essay argues that both images mislead, and for different reasons. The oracle metaphor confuses a fallible, evidence-based, conditional forecast with a claim to transcendent knowledge. The Odysseus metaphor treats a narrow problem of time-inconsistent commitment as though it described all central-bank communication. Drawing on the Federal Reserve’s own documentation, on the academic distinction between Delphic and Odyssean forward guidance, and on episodes from 2008 to 2026, I propose that credibility attaches to the process by which a central bank updates rather than to any particular forecast, and that the apt figure is the navigator. I add a qualification the metaphor invites but does not supply: updating that is late, unexplained, or erratic destroys the very credibility it is meant to protect. The inflation episode of 2021–22 and the present Committee’s retreat from forward guidance are treated as tests of the argument.
I. The Category Error: The Fed Is Not an Oracle
A word about vocabulary comes first, because the two myths are already technical terms. Campbell, Evans, Fisher and Justiniano (2012) call forward guidance “Delphic” when it merely forecasts macroeconomic performance and likely policy, and “Odyssean” when it publicly commits the Federal Open Market Committee (FOMC) to a future action, as Odysseus committed himself to his ship by having himself bound to the mast. They are careful to say that they do not use “Delphic” to evoke the oracle’s famously ambiguous utterances. This essay borrows the same two figures but asks a wider question: what does each myth imply about the epistemology and the commitment structure of a modern central bank?
Begin with the oracle. The Delphic tradition claimed access to knowledge that was, by definition, unavailable to ordinary reasoning. Its authority rested not on a reproducible forecasting procedure but on the supposed special status of the institution. To accept the prophecy was to accept the priestess.
A modern central bank operates under an entirely different epistemology. The Federal Reserve builds its projections from observed data, statistical relationships, economic models, financial conditions, surveys and informed judgment. Its projections are fallible hypotheses about the future, not revelations about it, and the institution says so plainly. The projection materials released after the FOMC meeting of September 15–16, 2026 describe each participant’s projections as the most likely outcome given information available at the time of the meeting. The same materials acknowledge that the models used are necessarily imperfect descriptions of the real world, that the path of the economy can be altered by unforeseen developments, and that the confidence intervals built from historical forecast errors may not capture participants’ current assessment of uncertainty and risk (Federal Reserve Board 2026a). The published fan charts are themselves an admission of fallibility: they are constructed from the root-mean-squared errors of forecasters over the previous twenty years, precisely so that readers can see how wrong such forecasts have tended to be (Reifschneider and Tulip 2017).
The distinction matters. A forecast can be wrong without the forecasting institution having behaved irrationally. If a central bank projects inflation of two percent and an unforeseen geopolitical shock delivers five, the projection does not retrospectively become a false prophecy. It was a conditional estimate made with the information then available.
This is why a Bayesian rear-view-mirror image is far closer to the economics of forecasting than any Delphic one. Economic forecasting necessarily begins from information about the past and the present: inflation, employment, wages, productivity, financial conditions, expectations. The forecaster infers from these the probable evolution of the system, ceteris paribus. The difficulty is that the vehicle is moving forward, and the forecaster therefore faces a basic asymmetry. The past is observable; the regime governing the future is not. That asymmetry is the essence of radical uncertainty.
II. The Rear-View Mirror Does Not Mean the Fed Is Blind
The Fed does not, of course, drive by the rear-view mirror alone. Modern forecasting draws on leading indicators, market prices, surveys, structural models, alternative scenarios and probability distributions. Yet even the mirror is imperfect. Orphanides (2001) showed that policymakers act on real-time data that are later revised, so that the picture of the recent past is itself provisional. The road behind is blurred before one even begins to extrapolate it.
It is worth being clear about what kind of uncertainty is at issue. Knight (1921) separated measurable risk from uncertainty that cannot be reduced to a known probability distribution, and Kay and King (2020) have argued forcefully that many of the decisions that matter most are taken in that second world, where the honest question is less “what is the probability?” than “what is going on here?”. A Bayesian framework respects this point rather than evading it. Its priors are disciplined judgments, not pretended frequencies, and its purpose is to state how beliefs should move when evidence arrives, not to claim that the evidence will ever be complete.
A more precise formulation than “driving by the mirror” is therefore available. The central bank drives forward using the rear-view mirror, the windshield, the dashboard instruments and a probabilistic map, but none of these reveals the road beyond the next bend.
The contrast with Delphi can be put in two sentences. The oracle says:
“This is what will happen.”
The Bayesian forecaster says:
“Given what we know now, this is how the probabilities have changed.”
That difference, between announcing a future and revising a belief, is the hinge of the whole argument.
III. Odysseus Answers a Different Question
The Odysseus image concerns not epistemology but commitment. In Book XII of the Odyssey the hero knows that he will be unable to resist the Sirens’ song, and so instructs his crew to bind him to the mast and to ignore any order to release him. The arrangement is deliberately designed to prevent his future self from overriding his present intention. Economists more often use the Latin name, following Elster’s (1979) study of precommitment in Ulysses and the Sirens, and I use the two names interchangeably.
In monetary economics the analogy addresses the problem of time inconsistency identified by Kydland and Prescott (1977). A policy plan that is optimal when announced may cease to be optimal once agents have acted on it, so a policymaker who re-optimizes at every date can end up in a worse equilibrium than one who could bind herself in advance. This point requires a correction to the way the analogy is often stated. The danger is not that tomorrow’s policymakers will behave irrationally. It is that they will behave rationally, and that what is rational for them tomorrow differs from what was best to promise today. Odysseus fears not a fool but a captive of the song.
Seen this way, the classification of Campbell et al. (2012) is exact. Odyssean guidance changes private expectations by committing the FOMC to a future deviation from the policy rule it would otherwise follow; circumstances will tempt the Committee to renege precisely because the rule describes its preferred behavior. The canonical case is the argument of Eggertsson and Woodford (2003) that, at the zero lower bound, optimal policy keeps rates low for longer than the usual rule would require once the shock has passed. Everything else is Delphic in the technical sense: it forecasts.
It follows that most of what the Fed says about the future is not a mast at all. Suppose the FOMC states that, given current conditions, it expects the federal funds rate to remain at a stated level for the foreseeable future. Inflation then accelerates, or an energy shock arrives, or financial conditions change sharply. If the Fed subsequently changes course, that is not a failure of commitment. It may be state-contingent policymaking working as intended. The Fed itself describes the appropriate future policy rate as highly uncertain because it depends on how real activity and inflation evolve (Federal Reserve Board 2026a). Even the Fed’s most explicit efforts to bind itself have been conditional in form. The December 2012 statement, which replaced calendar dates with numerical thresholds, kept the exceptionally low range in place “at least as long as” unemployment stayed above 6½ percent and inflation and inflation expectations behaved, while reserving the right to weigh other information (Federal Reserve Board 2012).
A qualification is owed here, because the literature is not one-sided. Campbell, Fisher, Justiniano and Melosi (2017) find that in the first years after the financial crisis a purely rule-based policy would have done better than the guidance actually given, but that from late 2011, after the introduction of calendar-based communication, Odyssean guidance appears to have raised real activity and moved inflation closer to target. Limited commitment can therefore add value when conventional policy is constrained. My claim is accordingly not that the Fed should never be bound. It is that it should not, in ordinary circumstances, be bound to the mast, and that when it does take a commitment it should be a conditional one that it can loosen in public, with reasons, when the evidence changes. The proper anchor is the mandate and the reaction function, not a forecast.
IV. Credibility Does Not Require Inflexibility
This brings us to what I take to be the strongest part of a Bayesian argument. A Bayesian holds that markets expect the Fed to keep its promises in normal circumstances but understand that extraordinary circumstances can dissolve the premises on which a promise was made. The proposition can be sharpened. Market credibility does not require a central bank to preserve an obsolete forecast. It requires the bank to behave consistently with its stated objectives and to explain why changing information warrants a change in policy.
Consider two hypothetical central banks. Central Bank A announces a policy path and refuses to alter it after a major shock, believing that a change of course would damage its reputation. Central Bank B announces a path conditional on the information then available, receives radically different information, revises its forecast, and changes policy while explaining the revision. Under a naive conception of commitment, A appears the more credible. Under a rational-expectations or Bayesian conception, that conclusion is far from obvious. If economic agents understand that policy responds to incoming information, they should expect Bank B to change course when the state of the economy changes, and should be alarmed by Bank A’s refusal to do so. The credibility of an institution lies not in “we will never change our minds” but in “we will change our policy when the evidence changes, according to a reaction function you can understand.”
V. Four Episodes
2008–2014: guidance that evolved
The global financial crisis offers the clearest illustration. In December 2008, with the target for the federal funds rate cut to a range of zero to one-quarter percent, the FOMC stated that weak economic conditions were likely to warrant exceptionally low levels of the funds rate “for some time” (Federal Reserve Board 2008). Forward guidance became an instrument in its own right because the conventional instrument had reached its floor.
The important point is what happened next. The language did not stand as a fixed promise. It was recast in 2011 in calendar terms, then in December 2012 in terms of economic thresholds that the Committee described as consistent with its earlier date-based guidance (Federal Reserve Board 2012), and later still, as unemployment approached the numerical threshold, into a broader qualitative assessment of progress toward the mandate. At each stage the wording was adapted to the state of the economy and to what the Committee had learned about how the public read it. This is exactly where the Odysseus analogy strains. Odysseus prevents his future self from changing the decision. The Federal Reserve has always retained the ability to respond to information. Its commitment was conditional, not absolute.
2019: patience
The 2019 shift is instructive because it was small in wording and large in meaning. Through late 2018 the FOMC had signalled that some further gradual increases in the target range would be appropriate. In January 2019 that language was dropped. The Committee said instead that, in light of global economic and financial developments and muted inflation pressures, it would be patient in deciding what future adjustments might be appropriate (Federal Reserve Board 2019). Nothing in the mandate had changed. The Committee had received information, revised its assessment of the balance of risks, and altered its signal accordingly. This is Bayesian updating conducted in public.
2020: the prior is overturned
The pandemic was the extraordinary shock. In December 2019 the median FOMC projection for 2020 real GDP growth was 2.0 percent and for the year-end unemployment rate 3.5 percent. By the June 2020 projections the medians were a contraction of 6.5 percent and an unemployment rate of 9.3 percent, and individual participants’ projections for 2020 growth ranged from a fall of 10.0 percent to a fall of 4.2 percent (Federal Reserve Board 2020). Almost every participant also expected the policy rate to stay unchanged through 2022 (Money and Banking 2020). The earlier projection had not been an oracle’s lie, and it would have been absurd for the Committee to defend it. The underlying stochastic process had changed, and the only rational response was to update, at once and visibly.
The episode also shows why forecast credibility and forecast accuracy must not be confused. No central bank could credibly have promised that its December 2019 projections would survive the spring of 2020. What it could credibly promise, and did, was to revise them openly and to explain the revision.
2021–2022: when updating comes late
The fourth episode is the hard one, and an honest version of the argument must confront it. Through 2021 the Committee’s projections and public messaging treated the surge in inflation as largely transitory. On November 30, 2021, with consumer prices up 6.2 percent over the preceding year, the highest reading since 1990 (Scripps News 2021), Chair Powell told the Senate Banking Committee that it was probably time to retire the word, that inflation would persist at least through the middle of 2022, and that policy had to address the range of plausible outcomes rather than only the most likely one (Fox Business 2021; Scripps News 2021). The first increase in the target range followed in March 2022.
It is reasonable to read this episode in two ways, and the argument should not pretend that only one is available. On the first reading, the Fed behaved as a good Bayesian should: it began with a prior that supply disruptions would fade, observed accumulating contrary evidence, revised, and then acted. On the second, the updating was too slow, the prior too stubborn, and the public was left to conclude that the reaction function was less responsive than advertised. Both readings are compatible with the thesis of this essay, but the second imposes a discipline on it. Credibility attaches to the updating mechanism, and a mechanism that updates with a long lag, or that re-labels its own earlier language without a clear account of what changed, damages the asset it is meant to protect. Speed and transparency of revision are part of the process, not extras.
VI. Commitment to an Outcome versus Commitment to a Process
These cases point to a distinction between two kinds of credibility. The first is outcome commitment: “We promise that X will happen.” It is vulnerable to any shock that makes X undesirable or impossible. The second is process commitment: “We promise to evaluate incoming information according to a known framework and to adjust policy consistently with our objectives.” This form is far more compatible with radical uncertainty, and it is arguably the more appropriate form for a modern central bank.
The Fed cannot control every future economic outcome. It can, however, influence expectations by showing that its decisions come from a recognizable procedure. In Bayesian terms, credibility attaches to the updating mechanism, not to the unconditional prediction.
VII. Why a Market Can Understand a Broken Forecast
This resolves an apparent paradox. Suppose the Fed projects two percent inflation and the outcome is four. The forecast was wrong. But suppose the Fed promptly acknowledges the new information, revises its projection and adjusts policy accordingly. Has it necessarily lost credibility? Not necessarily. Markets may reasonably read the revision as evidence that the central bank responds to information instead of clinging to obsolete projections.
The more dangerous situation is the opposite one. The Fed discovers that its forecast has become implausible but declines to revise it, fearing that admitting uncertainty would cost it credibility. That can convert a forecasting error into a credibility error, and the second is much the more expensive of the two. The qualification from 2021–22 applies here as well: the sooner and the more clearly the revision comes, the cheaper the error.
VIII. Forward Guidance as a Bayesian Signal
Forward guidance is not merely a prediction. It is also a signal about the central bank’s reaction function. The Fed explains that forward guidance matters because households and firms build expectations about the future course of policy into current decisions on spending and investment. Two components can therefore be distinguished. The forecast component answers the question “What do we currently expect?” The strategic component answers “How should you interpret our future response to economic developments?”
The second component is the more durable. “We expect inflation to decline” is a forecast. “If inflation remains above target, we will take the steps needed to restore price stability” is a statement about the reaction function. The second can remain credible even after the first has been proved wrong.
There is a further Bayesian subtlety, and the empirical literature supports it. The public does not read Fed announcements only for what they say about policy. It also reads them for what they reveal about what the Fed knows. Campbell et al. (2017) find that puzzling responses of private-sector forecasts to movements in federal funds futures on announcement days are attributable almost entirely to Delphic guidance: forecasters treat a statement about the future rate path as news about the economy itself. Andrade and Ferroni (2021) draw the same distinction for the euro area. A central bank that signals a lower path may therefore be heard as reporting weaker prospects. Guidance thus carries two messages at once, one about policy and one about the state of the world, and the audience updates on both.
This gives a sound theoretical basis for resisting an over-literal reading of the Odysseus analogy. What the audience most needs from the central bank is not a promise about a number but a legible account of how information will move policy.
IX. The Lucas Critique and the Reflexive Forecast
One more dimension distinguishes monetary forecasting from forecasting a physical system. Lucas (1976) showed that relationships estimated on historical data can break down when the policy regime changes, because agents adapt their behavior to the regime they expect. The Fed is therefore not forecasting a fixed mechanism. The system is strategic. Markets observe the Fed, the Fed observes markets, households and firms alter their behavior in response to expected policy, and the Fed then observes that altered behavior and revises its own assessment.
Its communication is consequently several things at once: a forecast, a policy signal, an instrument for managing expectations, a description of the reaction function, and an input into the very system being forecast. This is why the Delphi analogy is insufficient and the Odysseus analogy incomplete. A prophecy stands outside history; a central-bank forecast is a move within it.
X. A Better Metaphor: The Navigator
I propose replacing the two metaphors with a third. The Fed is neither Delphi’s oracle nor Odysseus permanently lashed to the mast. It is better understood as a navigator working under incomplete information. The navigator observes the sea, the instruments, the weather, the currents and previous experience. She sets a course and communicates it to the crew. But if the weather turns, holding the announced heading merely because it was announced would demonstrate not competence but vanity. The credible navigator says: “This was our course given what we knew when we set it. Conditions have changed. Here is the evidence. Here is the revised course, and here is why it serves the same destination.”
The metaphor has more to offer than a flattering contrast. Brainard (1967) showed that when the effects of policy are themselves uncertain, the optimal response is often more cautious than certainty would suggest: the navigator moves the rudder by degrees. Greenspan (2004) described modern policymaking as a form of risk management, in which the central bank weighs the distribution of possible outcomes and the costs attached to each and not only the most likely path. The Fed’s current projection materials say almost the same thing: in setting policy, participants consider not only what appears most likely but also the range of alternatives, their likelihood and their potential costs (Federal Reserve Board 2026a). A navigator, in short, steers by the chart of dangers as much as by the destination.
XI. The Present Test: A Committee That Has Dropped Forward Guidance
The argument now meets a live case. At the June 2026 meeting, his first as Chair, Kevin Warsh did not submit a rate projection of his own, said the Committee had dropped forward guidance, and announced a review of Federal Reserve communications, including press conferences, economic projections, transcripts and minutes, to be completed by the end of the year (The Hill 2026). The Committee’s June projections were accordingly compiled from participants other than the Chair (Federal Reserve Board 2026b). Reactions among economists were mixed, with some welcoming the change as overdue and others reading it as a retreat from transparency (Yahoo Finance 2026). In September the Chair again declined to submit a projection (J.P. Morgan Asset Management 2026; Bloomberg 2026).
The September projections came in a period of revision. The median participant now expects the policy rate to end 2026 at 4.1 percent, holding there through 2027; median core PCE inflation for 2026 stands at 3.4 percent, with a return to two percent only in 2029 (FRED Blog 2026; J.P. Morgan Asset Management 2026). The Committee raised its target by a quarter point at the September meeting (Bondsavvy 2026), and one market commentary describes the projected path as having moved since March from about half a point of cuts to about half a point of hikes. Whatever one makes of those particulars, the direction of revision is precisely what a Bayesian account would lead one to expect when growth is stronger and inflation higher than previously projected.
How does the framework of this essay read the retreat from guidance? Two answers pull in opposite directions. In favor of the change, Delphic guidance is a forecast published in advance, and a forecast published in advance can become an embarrassment or an unintended commitment when the world moves. In a period of unusually wide uncertainty, refusing to publish a point path lowers the risk that the Committee is held to a stale one. Against it, the strategic component of guidance, the account of how the Committee will respond to data, is exactly what makes a bank legible when its forecasts fail. Withdrawing the forecast component does not remove the need for the strategic one. It raises the burden on statements, press conferences and the consistency of decisions to communicate the reaction function, because the market will still attempt to infer it, and it will infer it from less. One market commentary suggests the remaining dots are better read as a measure of the range of views on the Committee than as a guide to the path of rates (Lord Abbett 2026).
I do not treat this as a settled matter, and the thesis makes a testable claim about it. If the Committee’s reaction function proves legible, so that persistent inflation above target reliably leads to firmer policy and softer data to easier policy, and if decisions are explained in those terms, then process credibility can substitute for point guidance at little cost. If it does not, the likely symptom is greater sensitivity of market pricing to each data release. The year-end communications review will be the occasion on which this is decided, and the outcome will be informative about how far credibility really can rest on process alone.
XII. The Necessary Qualification: Flexibility Must Be Rule-Like
It would be too strong to claim that markets will forgive any deviation from prior guidance whenever the Fed invokes exceptional circumstances. If a central bank repeatedly changes its guidance without a coherent explanation, markets will come to regard its communications as unreliable. Flexibility must itself be rule-like. The Fed needs to make clear what information can change its assessment, which objectives remain invariant, which indicators matter, how risks are weighed, and why the new policy follows from the new information. The first of these is the easiest to state and the hardest to honour: a reaction function that cannot be described in advance cannot be seen to have been followed afterwards.
The Fed’s institutional machinery is already moving in this direction. The projection materials separate the central projection from uncertainty, from historical forecast errors, and from participants’ own judgments on whether risks are balanced (Federal Reserve Board 2026a). Staff research is pushing further. Herbst, Konzem and Scofield (2026) document 1,265 alternative scenarios presented to the FOMC between 1968 and 2020 and find that the most accurate of them often anticipated major developments even when they missed the magnitudes. Adrian, Giannone, Luciani and West (2026) propose a Scenario Synthesis that places narrative scenarios and predictive distributions within a single Bayesian framework and assigns probabilities to scenarios consistently with a reference distribution. These are staff working papers and do not represent the views of the Board or the Committee, but they show the institution’s own analysts converging on the position argued here: under deep uncertainty, what should be communicated is a disciplined way of weighing possibilities, not a single prediction.
XIII. The Central Proposition
The central thesis can be stated as follows. A credible central bank does not promise to be right about the future. It promises to decide rationally when the future turns out to differ from what was expected.
This resolves the apparent contradiction between forecasting and flexibility. The Fed’s forecast can be wrong without its communication being dishonest. Its guidance can change without its institutional credibility collapsing. An extraordinary shock can invalidate a previous policy path without showing that the original judgment was irrational. The real test of credibility is therefore not “Did the Fed do exactly what it said it would do?” but “Did the Fed respond to new information in a manner consistent with its mandate, its stated reaction function, and a transparent explanation of why its earlier assessment had changed, and did it do so in time?”
The final clause is the price of the argument and should not be dropped. Credibility resting on process must be earned by conduct: by revising promptly, by explaining plainly, and by keeping the reaction function stable enough to be learned.
Delphi represents impossible certainty. Odysseus represents commitment bought at the price of flexibility, a price worth paying occasionally at the lower bound and rarely otherwise. Bayesian navigation represents rational policy under radical uncertainty. The Federal Reserve’s projections belong to the third category. Its forecasts are empirical rather than oracular; its commitments are conditional rather than absolute; and its credibility should derive less from never changing course than from showing, each time the economic regime shifts, that the process of updating remains intelligible, disciplined and consistent with its mandate.
References
Sources marked “staff research” are working papers that do not represent the views of the Board of Governors or the FOMC. Web addresses were checked on September 28, 2026.
Adrian, Tobias, Domenico Giannone, Matteo Luciani, and Mike West. 2026. “Risks and Uncertainty in Monetary Policy.” Finance and Economics Discussion Series 2026-061. Washington: Board of Governors of the Federal Reserve System (staff research). https://doi.org/10.17016/FEDS.2026.061
Andrade, Philippe, and Filippo Ferroni. 2021. “Delphic and Odyssean Monetary Policy Shocks: Evidence from the Euro Area.” Journal of Monetary Economics 117: 816–832.
Bloomberg. 2026. “Why Kevin Warsh Again Skipped the Fed’s Dot Plot.” September 16. https://www.bloomberg.com/news/articles/2026-09-16/why-kevin-warsh-again-skipped-the-fed-s-dot-plot
Bondsavvy. 2026. “September 2026 Fed Dot Plot Sees Low 4% Fed Funds in 2027.” https://www.bondsavvy.com/fixed-income-investments-blog/fed-dot-plot
Brainard, William C. 1967. “Uncertainty and the Effectiveness of Policy.” American Economic Review 57 (2): 411–425.
Campbell, Jeffrey R., Charles L. Evans, Jonas D. M. Fisher, and Alejandro Justiniano. 2012. “Macroeconomic Effects of Federal Reserve Forward Guidance.” Brookings Papers on Economic Activity, Spring: 1–80. https://www.brookings.edu/articles/macroeconomic-effects-of-fomc-forward-guidance/
Campbell, Jeffrey R., Jonas D. M. Fisher, Alejandro Justiniano, and Leonardo Melosi. 2017. “Forward Guidance and Macroeconomic Outcomes Since the Financial Crisis.” NBER Macroeconomics Annual 2016, vol. 31. https://conference.nber.org/confer/2016/Macro16/Campbell_Fisher_Justiniano_Melosi.pdf
Eggertsson, Gauti B., and Michael Woodford. 2003. “The Zero Bound on Interest Rates and Optimal Monetary Policy.” Brookings Papers on Economic Activity 2003 (1): 139–211.
Elster, Jon. 1979. Ulysses and the Sirens: Studies in Rationality and Irrationality. Cambridge: Cambridge University Press.
Federal Reserve Board. 2008. “FOMC Statement.” December 16. https://www.federalreserve.gov/newsevents/pressreleases/monetary20081216b.htm
Federal Reserve Board. 2012. “FOMC Statement.” December 12. https://www.federalreserve.gov/newsevents/pressreleases/monetary20121212a.htm
Federal Reserve Board. 2019. “FOMC Statement.” January 30. https://www.federalreserve.gov/newsevents/pressreleases/monetary20190130a.htm
Federal Reserve Board. 2020. “FOMC Projections Materials, June 10, 2020.” https://www.federalreserve.gov/monetarypolicy/fomcprojtabl20200610.htm
Federal Reserve Board. 2026a. “FOMC Projections Materials, September 16, 2026.” https://www.federalreserve.gov/monetarypolicy/fomcprojtabl20260916.htm
Federal Reserve Board. 2026b. “Summary of Economic Projections.” Monetary Policy Report, July 10, 2026 (June 16–17 meeting). https://www.federalreserve.gov/monetarypolicy/2026-07-mpr-part3.htm
Federal Reserve Bank of St. Louis. 2026. “FOMC Summary of Economic Projections, September 2026.” FRED Blog. https://fredblog.stlouisfed.org/2026/09/fomc-summary-of-economic-projections-september-2026/
Fox Business. 2021. “Powell admits Fed got it wrong on inflation, says they should stop calling it ‘transitory.’” November 30. https://foxbusiness.com/politics/powell-fed-wrong-inflation-not-transitory
Greenspan, Alan. 2004. “Risk and Uncertainty in Monetary Policy.” American Economic Review 94 (2): 33–40.
Herbst, Edward, Scott Konzem, and Cristina Scofield. 2026. “Alternative Scenarios at the Federal Reserve from 1968 to 2020: Data, Interpretation, and Evaluation.” Finance and Economics Discussion Series, May (staff research). https://www.federalreserve.gov/econres/feds/alternative-scenarios-at-the-federal-reserve-from-1968-to-2020-data-interpretation-and-evaluation.htm
J.P. Morgan Asset Management. 2026. “FOMC Statement: September 2026.” https://am.jpmorgan.com/us/en/asset-management/adv/insights/portfolio-insights/fixed-income/fixed-income-perspectives/fomc-statement-september-2026/
Kay, John, and Mervyn King. 2020. Radical Uncertainty: Decision-Making Beyond the Numbers. London: The Bodley Head.
Knight, Frank H. 1921. Risk, Uncertainty and Profit. Boston: Houghton Mifflin.
Kydland, Finn E., and Edward C. Prescott. 1977. “Rules Rather than Discretion: The Inconsistency of Optimal Plans.” Journal of Political Economy 85 (3): 473–491.
Lord Abbett. 2026. “September Fed Meeting: From Signals to Action.” https://www.lordabbett.com/en-us/financial-advisor/insights/markets-and-economy/2026/september-fed-meeting-from-signals-to-action.html
Lucas, Robert E., Jr. 1976. “Econometric Policy Evaluation: A Critique.” Carnegie-Rochester Conference Series on Public Policy 1: 19–46.
Money and Banking. 2020. “The Fed’s Crystal Ball: Looking Beyond the COVID-19 Recession.” June 15. https://www.moneyandbanking.com/commentary/2020/6/15/the-feds-crystal-ball-looking-beyond-the-covid-19-recession
Orphanides, Athanasios. 2001. “Monetary Policy Rules Based on Real-Time Data.” American Economic Review 91 (4): 964–985.
Reifschneider, David, and Peter Tulip. 2017. “Gauging the Uncertainty of the Economic Outlook Using Historical Forecasting Errors: The Federal Reserve’s Approach.” Finance and Economics Discussion Series 2017-020 (staff research). https://www.federalreserve.gov/econresdata/feds/2017/files/2017020pap.pdf
Scripps News. 2021. “Jerome Powell, Janet Yellen ditch ‘transitory’ term to describe inflation.” https://san.com/cc/fed-chair-jerome-powell-ditches-transitory-term-to-describe-inflation
The Hill. 2026. “Federal Reserve shifts away from forward guidance under new chief Kevin Warsh.” June 17. https://thehill.com/business/5929155-warsh-ends-fed-forecasts/
Yahoo Finance. 2026. “No Dot Plot, No Forward Guidance: Kevin Warsh’s First Fed Meeting Draws Mixed Reactions From Economists.” June. https://finance.yahoo.com/economy/policy/articles/no-dot-plot-no-forward-113128646.html
