The Global Review - Farid Novin

The Bank of Canada at a Monetary-Policy Crossroads

Macklem, Warsh, and the Reweighting of the Lucas Critique

Farid  Novin

I. Introduction

Central banking in 2026 is being conducted under conditions that the textbook version of inflation targeting did not anticipate: a live regional war disrupting a critical oil chokepoint, a bilateral tariff conflict between two economies bound by the world's most integrated trading relationship, and a newly installed Federal Reserve chairman who has made the outright rejection of forward guidance into a stated philosophy of central banking. The Bank of Canada's September 2 decision to hold its policy rate at 2.25% is, on its face, the least interesting part of the story — it was unanimously expected by economists surveyed beforehand. The more interesting story is what Governor Tiff Macklem's press conference reveals about how a modern, transparency-oriented central bank thinks about the relationship between its own communication and the expectations of the people it is trying to influence. That relationship is now being tested from two directions at once: an energy shock that will not resolve itself, and a Federal Reserve chairman next door who has deliberately chosen to say less.

This essay argues that the September decision should be read as a case study in what might be called Bayesian adaptive forward guidance — an approach that neither promises a fixed future rate path (the vice Kevin Warsh rightly warns against) nor withholds the central bank's reaction function altogether (the vice toward which Warsh's own communication style tends). Macklem's remarks, read carefully, show a central bank that has already found a workable middle path between these two failure modes, even though it has never formally named that path as a doctrine. Warsh's Jackson Hole speech of August 28, by contrast, illustrates the costs of the alternative: less guidance did not produce less market anticipation, only more volatile and more self-generated anticipation.

II. The September Decision: A Deliberate Pause, Not a Neutral One

The Bank of Canada left the overnight rate at 2.25%, with the Bank Rate at 2.50% and the deposit rate at 2.20%, marking a seventh consecutive hold dating back to December 2025. Every one of the thirty-five economists surveyed ahead of the decision expected exactly this outcome. What was not fully anticipated was the tone of the accompanying language, which was noticeably firmer on the inflation side than in July.

The Bank is managing an unusually asymmetric set of shocks. Canadian GDP rebounded 3.3% in the second quarter after a very weak first quarter, with the recovery broad-based across consumer spending, housing, exports and business investment. Hiring has picked up, and unemployment eased to 6.4% in July — still elevated, with the Bank's own language pointing to continued excess supply in the economy. At the same time, headline CPI inflation has held near 3%, driven almost entirely by gasoline and refined-product prices tied to the ongoing conflict in Iran and the curtailment of shipping through the Strait of Hormuz. Inflation excluding gasoline stood at 2.2% in July, and core measures have stayed close to the Bank's 2% target. Layered on top of this is a fresh round of American tariffs and Canadian counter-tariffs, which Macklem estimates affect roughly 5% of Canadian exports to the United States directly but which carry a second-order risk: renewed uncertainty about the trade relationship may cause businesses more broadly to delay hiring and investment regardless of whether their own products are directly tariffed.

This is a textbook supply-shock dilemma. Tightening policy would help prevent the energy shock from generalizing into broader inflation, but risks choking off a recovery that is not yet proven durable. Easing policy would support that recovery, but risks allowing an energy-driven price shock to become embedded in expectations and wage- and price-setting behaviour. Macklem's decision to describe the current rate in conditional terms — rather than repeating July's language that policy was "at the right level" — is the clearest signal that the Bank now sees the distribution of risks as having shifted, even though the level of the policy rate has not moved.

III. The Most Revealing Passage: Markets as Part of the Transmission Mechanism

The most theoretically important moment in the press conference is not the rate decision itself but Macklem's answer on how the Bank thinks about market expectations of future tightening. In substance, his argument runs as follows: markets have enough experience with the Bank's reaction function to infer that persistently higher oil prices raise the probability of future tightening, and price that probability into bonds well before the Bank acts. If the Bank subsequently fails to deliver the policy path that its own stated objectives require, markets will simply reprice around that failure.

This is a far more significant statement than it appears. It is, in effect, an admission that expectations about future policy are themselves part of Canada's monetary transmission mechanism — that bond yields, mortgage pricing and financial conditions today are already partly doing the work that a future rate change would otherwise have to do, precisely because market participants believe they understand how the Bank will respond to incoming data. The Bank of Canada does not publish calendar-based forward guidance of the kind used during the pandemic. But it benefits, continuously, from the fact that markets construct their own forward guidance out of the Bank's demonstrated reaction function. That distinction — between guidance the Bank issues and guidance the market infers — is the conceptual hinge on which the rest of this analysis turns.

IV. Macklem versus Warsh: Two Philosophies of Expectations

This is where the September decision becomes genuinely interesting in a comparative sense. Kevin Warsh used his first Jackson Hole address as Federal Reserve chairman, delivered August 28, to lay out an explicitly different philosophy. Warsh argued that transparency is not an unqualified virtue in central banking, and that in ordinary circumstances a central bank does better to limit forward guidance because overly explicit signalling can constrain its own freedom of action and can cause households, businesses and markets to form expectations that later prove difficult to walk back. His stated preference is for the Federal Reserve to draw relatively unfiltered information out of market prices — Treasury yields, exchange rates, credit spreads, commodity prices — rather than to try to shape those prices through announced policy paths. He was blunt about the asymmetry he wants: the Fed should not, in his words, encourage a dynamic in which market participants look to it for their next trading signal.

Macklem's framework runs in something closer to the opposite direction, even if it converges on some of the same inputs. He too says the Bank incorporates bond yields, equity markets, exchange rates and commodity prices into its assessment of financial conditions, which in turn feeds into the forecast that drives the policy decision. But rather than treating those market signals as a substitute for central-bank communication, Macklem treats them as one further input alongside a continuously updated, if informally delivered, account of the Bank's own reasoning. Put simply, there is an important difference between forward guidance in the literal sense and what might be called market-based guidance, in which the market does the work of formulating a policy expectation because the central bank has been consistent enough about its own logic to make that inference possible. The Bank of Canada appears comfortable operating within the second mode. The new Federal Reserve chairman, at least rhetorically, is trying to avoid both.

V. The Puzzle Inside the Lucas Critique

Warsh's underlying theoretical justification draws, whether explicitly or not, on an argument closely related to the Lucas critique: once private agents understand a policymaker's rule, their behaviour adapts to that rule, so historical relationships estimated under one policy regime cannot be assumed to hold once the regime — or the public's belief about the regime — changes. That proposition is correct as far as it goes. But a second implication tends to receive far less attention in policy discussion than it deserves: if expectations genuinely matter for economic outcomes, then deliberately shaping those expectations is not an illegitimate manipulation of the public but a policy instrument in its own right.

The Lucas critique establishes that economic agents respond to the policy regime they expect, not merely to the policy setting of the moment. It does not follow from this that central banks should therefore avoid trying to influence expectations altogether. If anything, the opposite conclusion is more defensible: because expectations feed directly into consumption, investment, wage-setting, pricing decisions, bond yields and exchange rates, a central bank has every incentive to understand — and, within limits, to shape — those expectations rather than treat them as an exogenous nuisance. The genuinely useful question is therefore not whether central banks should use forward guidance at all, but what form of guidance remains credible, and welfare-improving, once the public understands that the central bank is aware of its own influence over their beliefs.

VI. Why the Critique Can Strengthen, Rather Than Undermine, the Case for Guidance

Consider a simple illustration. Suppose households come to believe that the Bank intends to keep interest rates elevated for several years. Absent any other information, they would rationally postpone housing purchases, trim consumption and raise precautionary saving. Now suppose the Bank privately expects that the current inflation overshoot is driven by a temporary energy shock that will fade once the Middle East conflict eases and Strait of Hormuz shipping normalizes. If the Bank can credibly communicate that it intends to maintain a restrictive stance only so long as inflation expectations and underlying inflation remain elevated, and that it will ease as underlying inflation returns sustainably to target, household and business expectations can adjust accordingly — mortgage pricing, bond yields, investment plans, wage demands, corporate pricing behaviour and the exchange rate can all move in the desired direction without the Bank having to move the policy rate itself. This is the practical reason that expectations management can make monetary policy more, rather than less, effective: it allows part of the necessary adjustment to happen through beliefs rather than through the blunt instrument of the overnight rate.

VII. Unconditional Versus State-Contingent Guidance

The theoretical distinction missing from the conventional version of the Warsh critique is the difference between two fundamentally different kinds of forward guidance.

The first kind is unconditional: a promise such as "rates will remain at 2.25% until 2028." This is precisely the form of guidance that creates the time-consistency problem Warsh is right to worry about. If circumstances change, the central bank must either break its promise — damaging its credibility — or persist with an outdated policy simply to preserve the appearance of consistency, at real economic cost.

The second kind is state-contingent: a statement such as "policy will remain restrictive while inflation expectations and underlying inflation stay above target, and the appropriate rate will decline once underlying inflation returns sustainably toward 2%." This does not commit the Bank to any specific rate at any specific date. It communicates the reaction function itself, leaving the ultimate rate path to be determined by how the data actually evolve. That is a fundamentally different, and far more robust, form of guidance — and it is, in substance if not in name, close to what Macklem is already doing.

VIII. Macklem's "Beacon" as Rule-Based Guidance

Macklem's repeated description of the 2% inflation target as the Bank's "beacon" is not merely rhetorical colour. Combined with his statement that future decisions will be guided by the inflation forecast and the balance of risks around it, this constitutes a form of rule-based forward guidance quite distinct from the calendar-based guidance used during the pandemic era. The Bank is effectively telling economic agents: if a specified set of conditions materializes, the Bank will reassess a specified set of considerations, with a view to a specified objective. This is not a promise about a number. It is a disclosure of the variables that determine how the Bank's assessment will evolve — which is, in practical terms, a considerably more durable form of guidance than any fixed-date commitment could be.

IX. Two Agents, Each Updating on the Other

The framework that makes sense of all of this is a Bayesian one, described here in words rather than in formal notation, in keeping with the convention that this analysis avoids explicit mathematical apparatus for a non-technical audience. The central bank begins each decision cycle with a working view about the likely path of inflation, based on everything it currently knows. As new information arrives — oil prices, tariff developments, wage growth, unemployment, survey measures of inflation expectations, exchange-rate movements, bond yields and broader financial conditions — the Bank revises that working view, and the revised view feeds into the policy decision.

The crucial point is that private economic agents are doing exactly the same thing in parallel, except that the object of their belief-updating is not the economy directly but the central bank's own behaviour. Households, firms and market participants observe the same incoming data and use it to refine their beliefs about how the Bank is likely to respond. Monetary policy is therefore better understood as a two-sided process of mutual belief revision: the central bank updates its understanding of the economy, while the public simultaneously updates its understanding of the central bank. The second half of that process — how the public's model of the central bank evolves — is the part policymakers, and much conventional monetary economics, tend to underweight.

X. Communication as a Policy Instrument in Its Own Right

Under the conventional account, the central bank's objective is simply to minimize the joint deviation of inflation from target and output from its sustainable level. But once the public is understood to be running its own model of the central bank, a second and less obvious problem emerges: the Bank must also manage the information set from which households and firms construct their expectations, because that information set will shape their consumption, investment and pricing behaviour independently of where the policy rate actually sits at any given moment. Communication policy, understood this way, is not simply a public-relations exercise layered on top of the "real" decision. It is itself an instrument of monetary policy, capable of shifting outcomes even when the interest-rate setting does not move — which is precisely the mechanism through which forward guidance, properly designed, can help economic agents make better-informed decisions and thereby ease the central bank's own task of hitting its objective.

XI. The Paradox in Warsh's Own Position

There is a genuine paradox embedded in Warsh's stated approach. He wants markets to supply the Federal Reserve with relatively unfiltered information about the economy's likely path. But market prices are themselves forward-looking — they reflect, in part, beliefs about what the Federal Reserve itself will do next. Federal Reserve communication shapes market expectations, which shape asset prices and financial conditions, which shape economic activity, which shapes inflation, which in turn shapes the Fed's own subsequent decisions. There is no way to step outside that loop simply by talking less.

The evidence from the weeks since Jackson Hole illustrates the point. Warsh's first two press conferences as chairman left many market participants genuinely uncertain about his reaction function, prompting a sell-off in longer-dated Treasury debt as investors demanded compensation for that uncertainty; a CNBC survey taken in the run-up to Jackson Hole found roughly four in five economists, strategists and investors wanted him to spell out his thinking in more detail. Bank of America's rates and currency strategists went so far as to argue, ahead of the speech, that the Fed could help contain long-end Treasury yields by offering clearer guidance on its reaction function — the opposite of Warsh's stated instinct. When Warsh did speak at Jackson Hole, he reaffirmed the 2% PCE target as what he called a firm, fixed objective, stated plainly that short-term interest rates remain the Fed's predominant tool, and said he did not want markets treating the Fed as the source of their "next trade" — but he stopped short of laying out anything resembling a state-contingent reaction function of the kind Macklem offers routinely. Analysts were left to read the speech as hawkish mainly by inference: Tiger Brokers' James Ooi read Warsh's upbeat description of the economy as reducing the case for near-term cuts, while Miller Tabak's Matthew Maley argued there was no empirical case for a hike at all and suggested Warsh was talking up an inflation threat he could later claim credit for taming. That split reading, from analysts working off the same speech, is itself evidence of Warsh's point turned against him: silence does not eliminate expectations, it merely leaves them more dispersed, more speculative, and arguably more volatile than a clearly articulated reaction function would have produced.

XII. Canada's Compounded Problem

The Canadian case is more complicated still, because the Bank of Canada does not set policy in isolation from what happens south of the border. Macklem was asked directly about financial-market conditions and discussed, in the same breath, US Treasury yields, global bond markets, Canadian yields, the exchange rate, commodity prices and market expectations generally — but at no point did he engage directly with the new Federal Reserve chairman's stated philosophy of reduced guidance. That omission is worth dwelling on, because Macklem elsewhere acknowledges that Canadian long-term yields have risen partly in sympathy with rising global yields, even though Canadian yields remain below their American counterparts. The Bank's reaction function is therefore not simply a mapping from Canadian inflation to the Canadian policy rate. It runs through Canadian inflation, the Canadian output gap, oil prices, tariffs, the exchange rate, global bond yields and the evolving stance of US monetary policy — with each of those channels itself partly a function of what the market believes Kevin Warsh is likely to do next.

XIII. Why Warsh Matters to Canada Even Without a Canadian Response

The transmission channel works even in the absence of any explicit Canadian reaction. If Warsh moves the Federal Reserve toward higher rates, US Treasury yields rise; global yields tend to follow; Canadian yields rise in sympathy even while staying below US levels; Canadian financial conditions tighten as a result; the Canadian dollar moves in one direction or another depending on the nature of the underlying shock; mortgage and corporate financing costs shift; Canadian demand adjusts; the inflation outlook changes; and the Bank of Canada's own reaction function is engaged, purely through the financial-conditions channel Macklem has already said the Bank monitors. This is a considerably more sophisticated form of policy dependence than the shorthand claim that "Canada follows the Fed." The Bank of Canada does not need to move in lockstep with Washington. But Washington changes the state variables to which Ottawa is already committed to responding.

XIV. The September Decision as a Bayesian Wait-and-Update Strategy

The clearest way to characterize the 2.25% hold is as a deliberate, information-preserving pause rather than a settled judgment about the appropriate level of policy. The Bank has not yet seen evidence that the current inflation overshoot has become generalized: headline inflation near 3% is concentrated almost entirely in gasoline and refined-product prices, inflation excluding gasoline sits at 2.2%, core inflation remains close to 2%, and excess supply persists alongside an unemployment rate that, while improved, remains elevated at 6.4%. At the same time, the probability that the energy shock proves more persistent than assumed in July has clearly risen: oil markets are pricing meaningfully higher levels than the Bank's July forecast anticipated, reflecting the continuing conflict in Iran and constrained shipping through the Strait of Hormuz. The Bank's position, in effect, is that the posterior probability of a more persistent inflation problem has increased, but not by enough yet to justify an immediate change in the policy rate. That is a coherent and defensible position for an institution that is explicitly buying time to observe whether the shock stays contained to energy prices or begins to spread into broader price- and wage-setting.

XV. The Reaction Function Has Shifted, Even Though the Rate Has Not

The more consequential change since July is not the level of the policy rate but the shape of the distribution of plausible future rates. In July, a still-fragile recovery combined with what looked like a manageable inflation risk left room for the Bank to plausibly ease if conditions weakened further. By September, a stronger-than-expected growth rebound, a persistent energy shock, and continuing tariff uncertainty have together shifted probability mass toward tightening rather than easing. Macklem's language leaves little doubt that if inflation does not moderate as expected, the Bank is prepared to raise rates, and prepared to do so more than once if necessary. Independent economists reached a similar reading immediately after the decision: Capital Economics' Stephen Brown argued that Macklem's emphasis on rising inflation risk gives the Bank a bias toward hikes rather than cuts, while noting the Bank will likely want to see further improvement in unemployment or growth before actually moving; two of Canada's six largest banks, National Bank and Scotiabank, are already forecasting a rate increase by December. That is precisely the sort of shift in the perceived reaction function — rather than in the rate itself — that financial markets are now pricing.

XVI. Three Possible Explanations for the Bank's Silence on Warsh

There appear to be three plausible reasons the Bank of Canada has avoided commenting directly on the new Federal Reserve chairman's communication philosophy.

The first is institutional independence. Explicitly engaging with Warsh's framework risks creating the impression that the Bank of Canada is responding to the philosophy of a foreign central bank rather than to Canada's own domestic inflation outlook — a perception Canadian policymakers have long been careful to avoid, particularly at a moment of live trade friction with Washington.

The second is communication strategy. Discussing Warsh's approach explicitly could itself amplify the very market dependence Macklem seems reluctant to encourage, by inviting exactly the kind of comparative reading this essay is engaged in.

The third, and most interesting, possibility is that the Bank is adapting operationally without adopting the philosophy rhetorically. It is already monitoring precisely the variables Warsh emphasizes — bond yields, exchange rates, commodity prices, broader credit and financial conditions — and folding them into its forecast and, from there, into its policy decision. The Bank may simply not need to endorse Warsh's stated framework in order to absorb its genuinely useful informational content.

XVII. Market Dependence Versus Market Information

This distinction is, in this analysis's assessment, the conceptual core of the entire debate. "The central bank should follow the market" is a genuinely dangerous proposition — it would make policy hostage to sentiment, momentum and self-fulfilling speculation. "The central bank should extract information from the market" is something close to indispensable — market prices aggregate a vast amount of dispersed information that no single forecasting model fully captures. Macklem's framework sits much closer to the second proposition: he describes bond yields, equity markets, exchange rates and commodity prices as inputs into an assessment of financial conditions that feeds the forecast, not as a substitute for the forecast itself. Warsh, for his part, also emphasizes the informational value of market prices — the two governors are not as far apart on this specific point as their contrasting rhetoric about guidance might suggest. Where they diverge is on the second half of the equation: Warsh wants to minimize the central bank's own communication in order to preserve the market's role as an unfiltered information source, whereas Macklem preserves a comparatively transparent, rule-based communication style alongside that same reliance on market information.

XVIII. Reassessing the Lucas Critique

Properly stated, the Lucas critique does not establish that forward guidance is irrational. It establishes that the effect of any given form of guidance depends on how economic agents revise their own behaviour once they understand the rule the central bank is following. The correct inference is not "abandon guidance" but "design guidance as an endogenous part of the policy regime rather than treating it as an isolated, one-off announcement." Once agents understand that the central bank is itself responsive to their expectations, the guidance the bank offers becomes part of an ongoing strategic interaction rather than a simple broadcast: the central bank chooses what to communicate; agents update their expectations in response; those updated expectations change economic behaviour; that changed behaviour alters the economic outcomes the central bank observes; the central bank updates its own assessment in turn; and policy adjusts accordingly. This is a dynamic, repeated game of mutual belief-updating, not a mechanical rule applied to a fixed set of inputs — which is exactly why a state-contingent, rule-based form of guidance survives the Lucas critique in a way that a fixed, calendar-based promise does not.

XIX. The Welfare Case for Guidance

If households and firms make decisions today based on their beliefs about future policy, then helping them form more accurate beliefs can reduce economic volatility that serves no one's interest. Credible, state-contingent guidance of the kind Macklem practices informally can reduce excessive precautionary saving, unnecessary delays in investment, abrupt and destabilizing repricing in bond markets, uncertainty in mortgage renewal decisions, needless exchange-rate overshooting, and inflated risk premia embedded in longer-term borrowing costs. In each of these respects, guidance that is well designed allows the central bank to achieve its inflation and output objectives with smaller, less disruptive movements in the policy rate itself — which is a genuine welfare gain, not merely a communications convenience.

XX. The Real Danger: Guidance as a Commitment Trap

Warsh's underlying objection nonetheless retains real force, and this analysis should not be read as dismissing it. Suppose the Bank had said, hypothetically, that rates would remain at 2.25% until inflation reached 2%, and markets had priced that promise into borrowing costs across the economy. An unexpected new oil shock — entirely plausible given the state of the Strait of Hormuz — would then force an impossible choice: honour the promise and risk letting inflation run persistently too high, or break the promise and pay a real credibility cost that would complicate every future communication. This is the genuine Lucas-critique, time-consistency problem, and it is not hypothetical; central banks around the world have run into versions of it. But the appropriate response to this risk is not to abandon forward guidance altogether. It is to make the guidance conditional from the outset, so that changing circumstances do not require breaking a promise but simply trigger the previously disclosed contingency.

XXI. Toward a More Robust Model: Bayesian Adaptive Forward Guidance

The alternative this analysis proposes — Bayesian adaptive forward guidance — does not ask the central bank to announce a fixed future policy path. It asks the central bank to disclose four things on an ongoing basis: its policy objective; the principal variables it is monitoring; the direction in which movements in those variables would tend to push policy; and the degree of uncertainty surrounding its own forecast, together with the conditions under which its assessment would change. The guiding principle can be summarized simply: communicate the rule, not the rate path. This approach is considerably more compatible with the logic of the Lucas critique than either extreme — it preserves the central bank's flexibility to respond to new information while still giving households, firms and markets enough structure to form expectations that are stabilizing rather than destabilizing.

XXII. The Bank of Canada Is Already Operating Close to This Model

What makes the September 2 press conference notable, in retrospect, is that it already contains most of the elements of exactly this framework, even though the Bank has never formally labelled it as a doctrine. Macklem tells markets, in substance, that the inflation forecast together with the balance of risks around it will drive the policy decision. He tells them that persistent oil-price inflation raises the probability of tightening. He tells them that continued excess supply exerts downward pressure on inflation over time. And he confirms that the forecast itself will be updated ahead of the next scheduled decision. None of this is conventional, calendar-based forward guidance. All of it is implicit, Bayesian, state-contingent guidance. The Bank's underlying message to the public is, in effect: do not expect us to announce tomorrow's interest rate; instead, infer the distribution of likely future policy from the reaction function we have shown you — which is very close to what markets are doing already.

XXIII. October 28 as the Next Bayesian Updating Point

The Bank's next scheduled decision falls on October 28, 2026, alongside a new Monetary Policy Report. That meeting will matter less for whether the Bank moves the rate than for what it reveals about how far the Bank's own probability distribution has shifted. Between now and then, the Bank will absorb new information on the trajectory of oil prices, the duration of the Iran conflict and the state of the Strait of Hormuz, the practical implementation of the American tariffs and Canadian counter-measures, employment and core inflation data, exchange-rate movements, global bond yields, and — whether or not the Bank chooses to say so explicitly — the evolving posture of a Federal Reserve now led by a chairman whose own communication strategy is itself in flux. The household side of the ledger adds urgency to that assessment: mortgage delinquency balances rose sharply through the first quarter of 2026 on a year-over-year basis, consumer insolvencies reached their highest level since 2009, and the Bank's own Financial Stability Report has flagged that a meaningful share of Canadian borrowers — concentrated disproportionately in the Toronto area — may struggle to qualify for refinancing at 2027 rates and prices. None of this dictates the October decision, but it raises the cost of getting the underlying judgment wrong in either direction. The right question heading into October is therefore not simply "will the Bank raise rates," but whether the Bank's posterior view has moved far enough from optionality toward conviction to justify acting on it.

XXIV. Overall Judgment

The September 2 decision is best described as cautious hawkishness delivered inside a hold. The Bank did not raise rates because underlying inflation remains close to target, excess supply persists, the strength of the second-quarter rebound is not yet proven durable, and new tariffs threaten to weigh on growth going forward. But it also declined to repeat July's more reassuring language that the policy rate was "at the right level," replacing that formulation with a more conditional one because the balance of risks has genuinely shifted. The 2.25% rate should therefore be read not as a settled commitment but as a purchased option: the Bank has bought itself time to observe whether the energy shock stays contained to gasoline prices or begins to spread into broader inflation, while leaving itself free — and having said as much publicly — to raise rates, potentially more than once, if the evidence tips that way.

XXV. Conclusion: Pragmatic Bayesian Monetary Policy

The deeper theoretical conclusion is that the Lucas critique does not defeat forward guidance; it changes its optimal form. A central bank should not try to anchor expectations through unconditional promises about future interest rates — that path leads directly to the time-consistency trap Warsh is right to fear. But it can rationally, and usefully, shape expectations by communicating a credible, state-contingent reaction function, because once expectations are recognized as an endogenous part of the economic system rather than a fixed background condition, deliberately improving the quality of those expectations becomes part of the optimal policy itself, not a departure from it.

Warsh is correct that excessive, overly specific forward guidance can box in a central bank and force it into bad choices later. But it does not follow that less communication necessarily produces better policy outcomes — the market's reaction to his own Jackson Hole address suggests the opposite can just as easily occur, with silence simply replaced by dispersed and sometimes contradictory market-generated guesses. The workable middle ground is less promise and more reaction function: not committing to tomorrow's interest rate, but giving households, firms and markets enough insight into the central bank's conditional decision process that their own expectations become a stabilizing force rather than a destabilizing one. On the evidence of September 2, the Bank of Canada is already closer to that middle ground than its American counterpart — not because it has adopted a formal doctrine of Bayesian adaptive forward guidance, but because Macklem's habitual style of communication has arrived at something functionally similar to it. In one sentence: Warsh wants the Federal Reserve to learn from markets without letting markets dictate policy; Macklem is attempting something similar, but the Canadian experience so far suggests a central bank can — and arguably should — go one step further, by deliberately helping markets learn the reaction function on which their own expectations ultimately depend.