Searching for signposts − speech by Alan Taylor

It is very much an honour and a pleasure to be here with you today at NIESR, and my thanks to David Aikman for inviting me to give the 2026 Dow Lecture.

I am conscious that I follow in the footsteps of a number of my esteemed current and former colleagues at the Bank of England, including Swati Dhingra, James Talbot and Silvana Tenreyro. The Dow Lecture has provided an important forum for thinking seriously about the major macroeconomic challenges facing the United Kingdom, and I am grateful for the opportunity to contribute to that tradition.

Today I want to focus on one such challenge facing monetary policy right now: the ongoing and still unpredictable energy shock.

I should emphasise that this is only one of the challenges currently facing policymakers. The world economy is again being shaped by geopolitical uncertainty, by shifts in global trade and production, and by changes in financial conditions. But time is scarce, so I will not spend this lecture trying to predict geopolitical events or venturing too far down those other avenues.

Instead, as we look down the macroeconomic road ahead, I want to ask a narrower but, in my view, more tractable question: how do inflationary energy shocks pass through the economy; how far and wide do their effects subsequently persist; what role do expectations play in that propagation process; and what does all of this mean for monetary policy?

Unfortunately, that road ahead is not clearly illuminated, even at the best of times. The signposts policymakers would like to rely on are often sparse, partial and difficult to interpret. That is especially true after a large inflation shock, when the accumulated evidence can point in different and sometimes contradictory directions and the distinction between temporary pressure and longer-term persistence is hard to draw in real time.

So where should we look?

The answer, I will argue, is not to search for a single decisive indicator that will reveal the route ahead. Rather, the better approach is to look across a set of imperfect signposts: the initial energy-price impulse; the response of inflation expectations, especially among households and firms; the extent to which those expectations feed into wages and prices; and the wider state of demand, slack and policy credibility that will shape that transmission. Together, this array of signposts can help us judge whether an energy shock is likely to fade, or whether it risks becoming embedded in more persistent inflation.

A useful starting point is to recap the distinction the Bank has often drawn between so-called direct, indirect and second-round effects in the transmission of energy price shocks.

The direct effects are the mechanical impact of higher energy prices on headline CPI. For example, increases in petrol prices, household gas and electricity bills feed directly into measured inflation as faced by consumers in items they directly buy.

The indirect effects occur when higher energy prices raise the costs of producing and distributing other goods and services, leading firms to pass some of those costs through to consumers. This is particularly likely in energy-intensive sectors or where margins are under pressure. These effects can be powerful, but they are still relatively easy to conceptualise. They tend to work through the economy over a fairly short horizon, and input-output tables give us a benchmark for judging the relative magnitude of the pass-through.

The more difficult issue concerns the second-round effects. This phrase is often used rather broadly, even loosely. I think it best to use it narrowly and precisely to describe the behavioural, general-equilibrium dynamics that can follow an initially temporary rise in inflation and make the inflation process more persistent.

These dynamics should not surprise us. Most of the prices and wages in our economy are largely determined by private decisions without policy intervention. If inflation rises sharply, workers may seek higher nominal wages to recover lost real income. Where labour demand is strong, firms may have to accept those costs to recruit and retain workers. Firms may also raise prices not only because their own input costs have risen, but because they expect competitors, suppliers and customers to behave differently in a higher-inflation environment.

The key policy question is not whether these dynamics exist, but how large and persistent they are likely to be, for a given shock, and how monetary policy should respond.

Our flexible inflation-targeting mandate is central to that judgement. Like most central banks, we have not been assigned a price-level target; our task is not to reverse the increase in prices caused by an energy shock, but to return the rate of inflation to target.

And the flexibility embedded in the mandate instructs us to do so over a period that avoids undesirable volatility in output. Together with the lags in monetary transmission, this means that policy can 'look through' short-lived relative-price shocks, which is the relevant case for most external commodity price shocks in the most normal times.

But monetary policy cannot look through the emergence of more persistent inflationary pressure. Here, its influence lies not in reversing the original price-level increase, but in preventing that shock from propagating into an ongoing inflation process.

The question is therefore how much policy should lean against that risk, given uncertainty about its magnitude and the trade-off between stabilising inflation and real economic activity.^{footnote [1]} This is the nature of the policy problem created by second-round effects that we now face.

And inflation expectations are central to that problem. If households and firms expect a rise in inflation to be short-lived, they are less likely to respond in ways that sustain it. If instead they expect inflation to remain high, workers may seek greater compensation for future price increases and firms may raise prices to protect their margins. Those responses, all else equal, can then make the original shock more persistent, making matters worse.

This is why inflation expectations matter for second-round effects. Let me now make it concrete with an example.

The experience of 2022 is instructive. If we look first at the evolution of headline CPI and energy CPI, the basic pattern is clear (Chart 1). Headline inflation initially moved with energy inflation but remained persistently high even as energy inflation dropped sharply. That divergence is important. It suggests that the initial shock was doing more than mechanically raising the energy component of CPI. Rather, broader forces were contributing to a more persistent inflation process that unfolded gradually in waves (Taylor, 2025; Ramsden, 2024).

We can see the same pattern more formally by decomposing the drivers of inflation over this period (Chart 2). Using the Bank's forecast, we can separate the contribution of direct and indirect energy-price effects from the broader general-equilibrium dynamics that followed. The decomposition suggests direct and indirect effects (orange bars) account for much of the initial rise in headline inflation. But they cannot fully explain the persistence of inflation through 2023 and into 2024. Beyond additional global supply disruptions (light blue), contributions from persistence judgments (purple), slack (green), and the unaccounted-for component (pink) are consistent with broader general-equilibrium dynamics having played a material role.

So, what drove those general-equilibrium dynamics? They reflected the interaction of the original shock with wage and price-setting behaviour, inflation expectations, slack in the economy, and the stance of monetary policy. No single factor explains the persistence of inflation. Rather, the key point is one of state dependence: the effects of an energy shock depend not only on the size of the shock itself, but also on the economic environment into which it arrives.

In other words, the economy responded as an interconnected system in which households, firms and financial markets, in the UK and globally, formed views about what that particular shock meant in that particular context, and acted on those views.

That experience points to the first major signpost: inflation expectations. What do economic actors think is coming?

During the 2022 energy shock, expectations among financial markets, firms, and households rose markedly. Chart 3 compares CPI inflation expectations among households and firms, and a market-implied measure of inflation compensation over the coming year.^{footnote [2]}

We see that the inflation expectations evolved differently across agents. Market implied expectations normalised relatively quickly as energy prices fell and policy tightened. By contrast, firm and household expectations, declined more slowly and remained elevated for longer. The heterogeneity in dynamics across measures shown in Chart 3 is therefore important not simply because expectations rose, but because it illustrates that expectations can move at different speeds and remain elevated even after the original shock begins to unwind.

So, what do these indicators on that first signpost tell us? Well, most models, even those that are purely statistical, attribute a role to inflation expectations in explaining inflation persistence. For instance, a machine-learning model produced by Buckmann et al., (2025), summarised in Chart 4, attributes a notable share of the persistence in inflation, reflected in a relatively slow-moving trend component, to what we might call the inflation-expectations channel.

The sceptic would call that a black-box explanation, and I would not want to overstate the precision of any single model-based decomposition. But together the divergence between energy and headline inflation, the Bank's forecast decomposition, the behaviour of expectations and the statistical evidence suggest that the 2022 shock propagated well beyond its direct and indirect effects.

But a model decomposition does not reveal the mechanism. Expectations rising is not the same as expectations becoming inflationary in practice, in a causal sense. To affect realised inflation, they must change wage bargains, spending or price-setting, and the ability of households and firms to act on them will depend on labour-market conditions, demand, competition and margins.

This is why the second major signpost we come to is transmission: whether expectations actually feed into realised behaviour.

For monetary policy today, that distinction is central. The UK is again facing an energy shock, but in a different environment from 2022: the labour market is looser, excess demand has given way to excess supply, and monetary and financial conditions are restrictive. Those conditions may not fully eliminate the risk of second-round effects, but they should serve to limit their magnitude.

In the remainder of this speech, I assess the evidence across the two stages of the expectations channel. Stage 1 asks how inflation expectations did, and likely will, respond to the ongoing energy shock. Stage 2 asks whether those expectations are likely to be transmitted through into wages, prices and other economic decisions, making inflation more persistent. I then sum up by considering the implications, in my view, for policy strategy, given the evidence and uncertainties.

Stage 1: Drivers of inflation expectations and dynamics

Inflation expectations are relatively easy to obtain, but much harder to interpret. They matter for policy not simply because people report a number in a survey, but because those numbers may shape the decisions households and firms go on to make. So let me begin with the first part of the mechanism: how expectations are formed.

Over the past decade, we have learned much more about how different groups form inflation expectations. A key message from that work is that financial markets, professional forecasters, firms and households are not simply interchangeable sources of the same signal (Coibion & Gorodnichenko, 2025; Weber et al., 2025; D'Acunto et al., 2024; Reis, 2023). They pay attention to different information, update at different speeds, display different biases, and bring different experiences to bear when thinking about inflation.

Financial markets are, in some sense, the easiest case to understand. They are strongly incentivized to process information as quickly and as accurately as possible. They react to data releases, central bank communications, geopolitical developments, changes in the expected path of policy, and form expectations that are, by construction, highly forward-looking and information-sensitive.

In levels, households and firms look rather different to financial markets. Both form expectations that are highly dispersed in the cross-section, and importantly they tend to biased upwards with a skew (Chart 5). This upwards bias translates into means that typically sit above realised inflation, at least in the pre-2022 period, and since 2024. For households, this caveat applies not only to expected future inflation but also to perceptions of inflation today. Those perceptions also tend to be biased upwards, meaning that expectations are formed from a subjective starting point rather than directly from the same benchmark as official statistics (D'Acunto & Weber, 2026; D'Acunto et al., May 2024 NBER).

Of course, these surveys are not perfect instruments; responses are noisy, question wording matters, and the average can conceal a great deal of disagreement (Assenza et al., 2026; D'Acunto et al., 2024). But that is precisely why the distribution of beliefs, not just the mean, might be informative.

For policymakers, however, the key signal may lie less in the level of expected inflation than in how expectations respond to shocks over time. On this dimension, we can see that firms appear closer to households than to financial markets, while still occupying an intermediate position.

Neither households nor firms, in general, continuously process every piece of macroeconomic news in the same way as financial market participants. One of the key lessons from recent research is that they often use relatively accessible information when forming and updating expectations. That is not irrational. Attention is scarce and information is costly to acquire and process. And, for most households and firms, the returns to doing so are much smaller than for financial markets, particularly in a world with low and stable inflation.

Indeed, low attention to inflation is not a failure of credibility; but a reflection of it. One of the achievements of the inflation-targeting era was that inflation became sufficiently stable that most people did not need to think about it very much.

Whether the recent inflationary episode disrupted that low-attention equilibrium is an empirical question to which I will return. To interpret that evidence, however, we first need to understand how households and firms form expectations when inflation does attract their attention.

Households appear to pay particular attention to price changes in salient items of the consumption basket, rather than headline CPI per se. Food, energy and fuel prices are clear examples: they are visible, often purchased or closely noticed, and feature prominently in people's sense of the cost of living (Anesti, Esady, Naylor, 2025; D'Acunto et al., 2021; Coibion & Gorodnichenko, 2015).

This indicates that the effect of inflation on expectations may depend on where that inflation shows up. An increase in the price of something households buy frequently may shift expectations more than an equivalent increase in a less visible component of the index (D'Acunto et al., 2021). A risk is that expectations may respond asymmetrically to changes in inflation in these components: rising quickly when visible prices increase but falling slowly when those prices stabilise (Anesti, Esady, Naylor, 2025). In that case, a temporary shock can leave a longer imprint on beliefs than its direct arithmetic contribution to inflation would suggest.

Firms, too, appear to respond to realised inflation, but with a greater focus on Headline CPI. Recent high-frequency evidence from the UK Decision Makers' Panel (DMP) finds that firms' inflation perceptions and expectations respond around official CPI releases (Chart 6). Those responses appear related to the reported change in CPI inflation, with little evidence of an additional response to the surprise relative to expectations, such as professional forecasts.

Putting these pieces together, we see a common theme. While financial markets are more forward-looking, both households and firms form expectations partly by looking backwards - to realised inflation, to perceived current inflation, to observed price changes (with a bias toward salient ones), and to the economic environment they have just experienced. That distinction captures an important contrast between "Wall Street" and "Main Street".

That backward-looking behaviour is not necessarily mistaken: where inflation itself has a (possibly time-varying) inertial component, recent inflation contains information about its likely path.

In the UK data, this backward-looking component shows up clearly (Chart 7). A one percentage point increase in perceived inflation is associated with a rise in expected inflation one year ahead of around 0.4 to 0.5 percentage points. Put differently, expectations appear to be partly conditioned by what households and firms believe inflation to be today, with the implication that biases in the latter can be passing through into the former.

However, expectations are not only backward-looking. Experimental evidence shows that both households and firms do update their inflation expectations when they are provided with clear forward-looking information, including information from central banks (Coibion, Gorodnichenko & Weber 2022; Coibion, Gorodnichenko & Kumar 2018).

That matters for monetary policy communication. It tells us that economic agents are not unreachable. But it also tells us that communication must compete with expectations that may already have taken shape from subjective experience, media coverage and observed prices. The question is not whether central bank information can matter. It can. The question is how much it matters relative to the information people have already gathered from what they see around them.

A simple way to frame this is as a competition between signals from the recent past and signals about the likely path ahead. The UK evidence is consistent with financial market expectations being more strongly associated with forward-looking information, and households and firms responding to elements of both channels (Chart 8).

The balance also varies within groups. That cross-sectional variation is important. Among households, those with more accurate perceptions of current inflation appear more responsive to both backward- and forward-looking information, while those in the tails respond less to macroeconomic signals. The distribution may therefore contain information obscured by the average.

For policy, the distinction between attentiveness to backward- and forward-looking signals matters: the backward-looking component may warrant particular attention. If a temporary increase in inflation leads households and firms to revise up their expectations persistently (Stage 1), and those expectations then feed through into wage and price-setting (Stage 2), the original shock can become more persistent and the force of transmission could mean that perceptions self-validate.

So did the inflationary episode of 2022 disrupt that low-attention equilibrium and take us some way down that path? There are reasons to think that it did, and that expectations may now be more sensitive to inflation shocks than before 2022. Whether that makes them more useful as a guide for policy is less clear.

That 2022 inflationary surge was large, visible and prolonged. It not only affected prices across the economy; it also generated intense media coverage and made inflation a normal topic of conversation again. We see this behavioural shift through online data; with Google searches and media references to inflation rising sharply in 2022 (Charts 9a and 9b).

The recent econometric evidence is consistent with that shift. While sharp inference is difficult, household expectations appear to have become more sensitive to realised inflation since the 2022 shock, particularly to visible components such as food and energy (Chart 10).

For firms, the picture is similar: own-price expectations appear to have become more responsive to current inflation during the high-inflation period (Yotzov et al., 2025).

Importantly for policy, greater attentiveness appears to have increased the relative weight placed on backward-looking information, such as realised inflation, while reducing the weight placed on forward-looking information (Chart 11).

We should be careful not to overstate precision here. Some of the post-2022 evidence is based on short samples, and it is difficult to separate cleanly the effects of inflation itself, media attention, energy prices and broader uncertainty. But the broad direction of travel is clear enough to matter for policy: today's shocks may now have larger effects on expectations than similar shocks would have had before 2022.

I would not call this de-anchoring. But I would call it a change in the environment in which expectations are formed. A central bank that could previously assume a high degree of inattention among households and firms may no longer be able to do so.

That observation cuts both ways. Greater attention means that clear and credible central bank communication may have more traction. But it may also allow adverse price news to travel more quickly into beliefs. Louder noise is not necessarily offset by louder signals, especially if attention to recent inflation crowds out the forward-looking information policymakers seek to convey.

The conclusion I draw is deliberately cautious. Household and firm expectations may now respond more strongly to inflation shocks than before 2022. We should not treat every rise in those expectations as evidence of a wage-price spiral, but neither should we dismiss them simply because they are noisy or imperfectly measured.

But expectations are only the first stage of the mechanism. Whether they ultimately matter for inflation depends on whether they feed through into behaviour. I turn to that second stage next.

Stage 2: The importance of inflation expectations in generating persistent inflation

Survey expectations are not behaviour. They are an input into behaviour. For monetary policy, the crucial question is whether they feed into wage bargaining, price setting, consumption and investment and, ultimately, inflation itself.

That is Stage 2, and it is harder to establish empirically.

In theory, the mechanisms are straightforward. On the nominal side, households expecting higher inflation may seek higher nominal wages to protect real incomes; and firms may raise prices. In the extreme, as Lorenzoni and Werning (2023) formalise, when workers' real-wage aspirations and firms' desired margins are jointly inconsistent with the real income available, attempts to protect those positions through wage and price setting can generate persistent inflation.

On the real side, expectations may also affect household spending decisions, with two competing effects: higher expected prices can bring consumption forward, while concern about future real incomes can push households in the opposite direction (Christiano, Eichenbaum, Evans, 2005). For firms, the effect on employment and investment depends not only on price expectations but also on what they expect to happen to demand.

Identification is difficult because inflation influences expectations, as we saw in Stage 1, while expectations may also influence inflation, and both respond to wider economic conditions. As a result, the evidence base is thinner than the importance of the question might suggest.

I would organise the evidence in two buckets: microeconomic studies that ask whether changing expectations changes behaviour; and aggregate evidence on whether expectations help explain or predict inflation dynamics. Considered as signposts, neither bucket is perfect. But together they help us assess which expectations matter most, and why.

Micro evidence

The micro evidence is strongest for firms' price setting. Experimental and survey studies find that higher inflation expectations lead to higher intended price increases. Estimated pass-through is not one-for-one, but economically meaningful at around 20 to 30 per cent in (Abberger et al., 2025; Coibion, Gorodnichenko, Ropele, 2019), and larger in some UK survey evidence (Ghassibe et al., 2025; Yotzov et al., 2025). Expectations are therefore not merely epiphenomenal: in at least some settings, they enter directly into decisions that matter for inflation.

The wage evidence is positive but weaker, and much of it predates the recent inflation surge (Guerreiro et al., 2024; Jain et al.,2024; Savignac et al., 2024; Hajdini et al., 2023). Crucially, the ability to translate an increase in inflation expectations into wage outcomes is likely to depend on labour-market conditions, institutional arrangements and bargaining power.

Evidence on real decisions is more mixed. Higher expected inflation can bring some households purchases forward (D'Acunto et al., 2022a; D'Acunto et al., 2022b),^{footnote [3]} but it can also make them more cautious if they associate inflation with weaker real income or greater uncertainty (Coibion, et al., 2023; D'Acunto et al., 2023; Coibion et al., 2022),^{footnote [4]} as growing evidence suggest they do (Zhang, 2026, Debortoli & Iovino, 2025). Similarly, firms' employment and investment decisions similarly depend on whether higher inflation is interpreted as stronger demand or as a cost shock squeezing margins (Coibion et al., 2019; Coibion et al., 2018). That ambiguity is not a nuisance in the data; it is part of the economics.

Aggregate evidence

The aggregate evidence begins with a deliberately simple question: do expectations predict inflation? Using UK data, we can look at the relationship between headline inflation and the one-year-ahead expectations reported one year earlier by different groups. We focus short-horizon expectations, which evidence suggests matter more for transmission to aggregate dynamics (Reis, 2026; Coibion & Gorodnichenko, 2025; Werning, 2022).

Chart 12 summarises the results. Financial-market expectations have the strongest correlation with future headline inflation and explain the largest share of the variation. Firms come next, followed by households. Among households, those with more accurate perceptions of current inflation have relatively greater predictive power, as one might expect.

But forecasting performance is not the same thing as policy relevance. A forecasting regression cannot tell us whether an agent matters because their expectations influence inflation, or because they are good at forecasting inflation that would have happened anyway. That distinction is central. Financial-market participants are paid to process macroeconomic information, and they incorporate forward-looking signals quickly. It may not be surprising that their expectations look like better forecasts, even if they have no direct part to play in setting goods prices and wages.

Chart 13 illustrates this point in another way, through simple cross-correlations. Financial-market expectations are most strongly related to inflation several quarters ahead, while household and firm expectations are more closely related to current inflation. Consistent with the Stage 1 evidence, markets appear more forward-looking, whereas households and firms place greater weight on realised and observed inflation.

Put another way, an expectations measure can matter for two different reasons. It may contain useful information about inflation that would occur anyway, or it may influence the decisions that determine inflation. A measure can therefore be a good forecast without being causally important, or a poor forecast in levels while still affecting the direction of wage and price-setting. Policymakers need to distinguish between those roles.

We can get closer to the hypothesised mechanism by focusing on inflation measures that households and firms are more likely to influence, such as domestic, underlying and wage-sensitive inflation, rather than imported energy prices or agricultural shocks. Chart 14 shows that household and firm expectations tend to be more informative for these measures than for headline CPI. The faded bars show results for headline inflation and the solid bars for domestic (core) inflation. The left-hand panel shows the strength of the relationship between inflation and expectations, while the right-hand panel shows how much of inflation's variation those expectations explain. The results are more consistent with a behavioural channel operating through domestic price-setting, although they do not identify such a channel causally.

Whether expectations influence inflation is also likely to depend on the state of the economy. Households have more scope to translate inflation expectations into wage outcomes when labour markets are tight. Firms have more scope to translate them into prices when demand is strong, and pricing power is greater. A given movement in expectations may therefore have considerably more traction in some economic conditions than in others.

One suggestive test of that state-dependence is to compare the periods before and after 2022 (Chart 15). The later period combined high inflation and elevated attention with a tight labour market and pressure on firms' margins. During that period, the predictive power of household and firm expectations appeared to have increased.

Again, I would not claim that this proves a causal mechanism. The stronger relationship could reflect greater inflation persistence or improved forecasting, rather than a greater influence of expectations on wage and price setting. Causal proof is a high bar, but this evidence is at least consistent with the hypothesis that household and firm expectations gained traction during the high-inflation episode.

Structural US evidence points in the same direction. Coibion & Gorodnichenko (2025) find that short-horizon household expectations help explain inflation dynamics in a New Keynesian Phillips Curve, while professional expectations play a smaller role. That result clearly depends on details of the model specification, especially its inertial pieces, and it should not be imported mechanically to the UK setting; but it cautions against focusing only on the most sophisticated forecasters.

To sum up, financial-market expectations are timely, forward-looking and often informative at the policy-relevant horizon. For the questions we face, they would be the first place I would look. But household and firm expectations matter for a different reason: they are closer to the decisions that can make inflation persistent. They need not be superior forecasts and are often biased. But in the wrong state of the economy, however, they can become part of the propagation mechanism itself.

That brings us to the current juncture. Stage 1 showed that households and firms may now respond more strongly to shocks; Stage 2 suggests that this matters only if those expectations gain traction in wages, prices and demand.

A broader historical perspective is also useful. After the experience of the 1970s, the lesson was not simply that inflation expectations could rise, but that they could become embedded in wage bargains, price-setting rules and contracts, and thereby become part of the inflation process itself (Taylor, 2026a; Taylor, 2026b)

The lesson of 2011 is almost the mirror image. Headline inflation rose well above target and household expectations moved with it, but weak demand and a softer labour market limited the transmission into wages and domestically generated inflation.

Indeed, one of the achievements of the modern monetary framework is precisely to weaken that second-stage transmission: credibility does not mean that near-term expectations never respond to shocks, but that temporary shocks are less likely to be built mechanically into wage claims and pricing plans, and so less likely to persist.

The question as to what happened in 2022 and later is whether a very different transmission environment - tighter labour markets, stronger nominal wage growth, disrupted supply and greater pricing flexibility - temporarily gave expectations more traction than they had in earlier episodes, and whether we are still in that regime now.

That is where the final part of my discussion begins, on the policy problem we face today. Higher attentiveness means that shocks can still move expectations. But whether those movements matter for policy depends on the state of the economy. It is that configuration of risks, rather than expectations in isolation, that I now turn to.

Where do we go from here?

There are many signposts and they can often conflict. Some of them are true signs, an informative picture of the road ahead; others are ghost signs, a fading image of past structures.

Expectations remain central to monetary policy, but not all expectation measures answer the same question. Some are better forecasts. Some are closer to behaviour. Some are more sensitive to recent inflation. And the importance of each depends on the wider economic environment.

The task for policymakers is not to follow any single measure mechanically, but to understand what each measure is telling us, whether it is signalling information, influence, or both, and how that signal should shape the risk calculation for inflation persistence.

Recap of Stages 1 and 2

The argument so far can be summarised briefly. Inflation expectations are indispensable signposts for monetary policy, but they are not straightforward guides. Large and salient shocks can move the expectations of households and firms, perhaps more readily now than before 2022. But those movements become inflationary only if they affect wage bargaining, price-setting and other economic decisions. Their policy significance therefore depends not only on the expectations measure itself, but on whether it is influencing behaviour and on the economic conditions governing that transmission.

The policy problem is therefore not to follow any one expectation measure mechanically, but to judge whether it is signalling information, behavioural influence, or both.

Second round effects = State x Shock

Second-round effects depend on the interaction between the shock and the state into which it arrives. A large or persistent energy shock is more likely to generate enduring domestic inflation when expectations, bargaining conditions and pricing behaviour allow it to propagate. That is why the difficult interpretive work around expectations is not merely academic: it is central to judging whether today's shock will fade, or whether it may start to reproduce the dynamics that made the 2022 episode so persistent.

Chart 16 provides direct evidence of state-dependence. The cumulative inflation response to an oil price shock is larger when the labour market is tight and, more strikingly, when firms adjust prices frequently. The confidence intervals are wide, so the precise magnitudes should be treated cautiously. But both exercises point in the same direction: energy shocks propagate more strongly when workers have greater bargaining power and firms have more scope to adjust prices.

Chart 17 reinforces this point for wages. More than three quarters of the response to a gas price shock depends on labour market slack. The mechanism operates through a common cost-of-living channel, rather than the direct exposure of energy-intensive sectors. That channel is likely to be weaker when labour market slack is greater and workers have less bargaining power.

State

A useful way to see the importance of initial conditions is in JPMorgan's cross-country Taylor-rule comparison. Table 1 points to a striking asymmetry: the UK is the odd one out.

That matters for the policy judgement. The UK is not simply different numerically; it faces a different policy trade-off. For those asking why we might not move in the same way as other central banks, this evidence supplies the main reasons.

The table makes the point clearly. For most economies shown, the Taylor-rule estimates move higher by mid-2027, reflecting positive or rising output gaps and inflation still above target. The UK does not fit that pattern. Its projected inflation gap is still sizeable, but its output gap is negative and larger in absolute terms than in most of the other economies shown. In other words, the UK faces an inflation problem in a weaker real economy.

UK policy also starts from a more restrictive position: our actual policy rate in 4Q25 is, along with that of the US, above that of all other the economies shown. It is also above the Taylor-rule estimate they project for the UK in 2Q27: unlike all of the others, the UK Taylor-rule estimate falls from 4.2% in 4Q25 to 3.5% in 2Q27, rather than rising.

As I noted in my voting paragraph in the latest Monetary Policy Summary, both Bank Rate and the OIS curve remain substantially restrictive relative to my estimate of nominal R*, of 2.75% to 3% (Chart 18). And beyond the two-year point, market rates are more restrictive than they were when we delivered the final hike of the previous cycle in August 2023, at the time when inflation stood at 6.8%. My assessment is that the current stance is more than sufficient to weigh on demand and inflation and provide the degree of restrictiveness needed to return inflation sustainably to target.

Shock

The shock side of the equation is unusually hard to pin down at present. Since March, spot Brent has moved sharply, ranging from roughly $70 to around $110, and we should be humble about what comes next. Geopolitics is now driving economic outcomes - and therefore monetary policy - to an unusual degree, and remains highly unpredictable.

But the task is not to forecast the next tick in energy markets; rather it is to judge the distribution of risks and the conditions under which those risks would generate persistent inflation.

The central scenario, in my view, is likely one of oscillation, as in the last six months, a continued stalemate followed by modest de-escalation later in the autumn. Under this outcome, Persian Gulf exports gradually recover, oil prices remain elevated but manageable, and the global economy avoids a major energy shock. A more optimistic scenario assumes some form of a resolution in the near term, allowing energy flows to normalise by year-end and easing inflationary pressures. By contrast, a pessimistic scenario involves further escalation and renewed disruption to shipping and production, substantially reducing energy flows and generating another sharp rise in oil prices and energy-driven inflation.

For the UK, it is also important not to look only at the crude oil curve. UK inflation risks can also come through other energy-market channels. Crack spreads have risen, pointing to tightness in refined products such as diesel and petrol, as well as jet fuel, which matter directly for transport costs and indirectly for distribution and production costs. LNG is another risk factor, as it is central to UK electricity pricing and costlier to arbitrage across locations. Even if oil markets stabilise, relatively low inventories mean gas prices could remain sensitive to cold weather, stronger demand or further supply disruptions.

The overall message is that energy risks into winter are two-sided but asymmetric. The downside for prices appears limited in the near term, while the potential for renewed upward pressure remains significant. That does not mean the shock must inevitably become macroeconomically large enough to generate second-round effects. But it does mean the first part of the risk calculation - the size, duration, salience and breadth of the energy shock - cannot be dismissed. But that brings us to the next point.

Implications for second-round effects

We can see some signposts about the state and the shock. But what do we know about the risk of transmission?

The Bank is monitoring a broad set of signposts, including inflation expectations, wage settlements, firms' pricing behaviour and margins, alongside demand conditions and labour-market slack. The key is to distinguish between a temporary rise in inflation and genuine transmission into prices, wages and persistent inflation. High-frequency indicators can help us trace that process in real time, but no single measure is decisive, so we need to assess whether these indicators collectively show that the energy shock is spreading beyond its direct effects and becoming embedded in domestic inflation.

The sectoral evidence suggests that some propagation beyond direct energy costs may already be present (Chart 19). Since the 2026 shock, prices have risen somewhat faster in sectors historically associated with quicker second-round effects than in sectors where those effects have been weaker. But the overall increases remain much smaller than after the 2022 shock and closer, so far, to the 2011 episode. This is consistent with second-round effects that are present but more limited in magnitude, reflecting both a smaller shock and a less permissive economic environment.

The DMP evidence gives a similar message (Chart 20). Realised own-price growth among energy-intensive firms has so far responded less strongly than during the 2021-22 episode, even though expected own-price growth has risen. That gap is consistent with firms intending to raise prices but facing greater constraints on their ability to do so. The comparison remains provisional because the follow-up period is much shorter.

Policy

To conclude,the recent surge in gas and oil prices has clearly increased near-term inflation risks and could push headline inflation considerably higher over the winter. But the key policy question is not the current level of energy prices alone. It is whether those prices remain elevated for long enough, and become broad enough, to generate more persistent inflationary pressure.

On the evidence so far, higher energy costs still appear largely concentrated within the energy complex itself rather than spreading widely through the economy. Monetary policy should respond to the risk that a relative-price shock becomes a general inflation process, not mechanically to every movement in the relative price itself. This is about judgements on propagation, not about denying the shock.

Evidence for significant second-round effects remains scant at present. Inflation in many energy-intensive goods and services categories has not accelerated as one might have expected after such a visible energy shock; food inflation has in fact fallen markedly and unexpectedly, though could rebound; and survey and producer-price indicators point to only modest pass-through so far. The labour market tells a similar story. Underlying wage growth appears broadly consistent with inflation returning to target over the medium term, and there is little sign of an emerging wage-price spiral.

Taken together, these developments suggest that the economy is proving less susceptible, at least so far, to a repeat of the inflation dynamics seen in 2022.

Nevertheless, there remains a nontrivial risk. Where views may differ on the Committee is on the extent to which we lean against these risks. Now, by holding Bank Rate fixed while financial conditions have tightened, policy has already become materially more restrictive relative to the outlook, and especially compared to the pre-war configuration in February. My view is that the current stance is also sufficiently restrictive, given the initial conditions of the economy in which the shock is hitting.

Against that backdrop, the case for further rate increases is not compelling to me unless energy prices remain high for an extended period and also generate clearer signals of a transmission into broader inflation persistence, as revealed by the signposts that we are actively monitoring.

The central judgement, then, must be whether the current energy shock remains a relative-price shock or evolves into a more general inflation process. That judgement has to combine both sides of the framework: the magnitude, duration and salience of the shock, and the initial conditions that determine whether it transmits. The available evidence currently points more toward the former interpretation.

That does not remove the need to stay alert. However, it does suggest, to me, that the burden of proof for additional tightening should rest on evidence that second-round effects are actually gaining traction, rather than on the existence of the energy shock alone.

That judgement, as ever, is conditional: if pressure builds and second-round effects begin to gain traction, the policy assessment would have to change. But symmetrically, if the economy evolves broadly as expected, with demand weakening and underlying domestic inflationary pressures continuing to moderate, then at some point, once energy risks abate, policy will need to move in the other direction - not because the risk has disappeared, but because maintaining an unnecessarily restrictive stance would itself carry costs.

The right policy response is therefore vigilant but disciplined. We should not deny the shock, nor dismiss the risk that it could propagate. But neither should monetary policy react mechanically to movements in energy prices if those movements remain primarily relative-price shocks.

The views expressed in this speech are not necessarily those of the Bank of England or the Monetary Policy Committee.

Acknowledgements

I would like to thank Sofia Carollo, Vitor Dotta, and Matthew Naylor for their help in preparing this speech.

I would also like to thank Andrew Bailey, Maria Balgova, Marcus Buckmann, Jenny Chan, Rohan Churm, Hannah Copeland, Ruslana Datsenko, Lucio D'Aguanno, Mridula Duggal, Katie Farrant, Nowrin Hossain, Clare Lombardelli, Josh Martin, Gaspar Montenegro Calvimonte, Galina Potjagailo, Nades Raviraj, Carleton Webb, Tim Willems, and Ivan Yotzov for their comments and help with data and analysis.

References

Abberger, K., Funk, A.K., Lamla, M.J., Lein, S. and Siegrist, S. (2024) 'The pass-through of inflation expectations into prices and wages: Evidence from an RCT survey'. SSRN Electronic Journal. Available at: The pass-through of inflation expectations into prices and wages (Accessed: 24 September 2026). doi: 10.2139/ssrn.5073673.

Andre, P., Pizzinelli, C., Roth, C. and Wohlfart, J. (2019) 'Subjective models of the macroeconomy: Evidence from experts and a representative sample'. CESifo Working Paper Series, No. 7850. Available at: Subjective models of the macroeconomy (Accessed: 24 September 2026).

Anesti, N., Esady, V. and Naylor, M. (2025) 'Food prices matter most: Sensitive household inflation expectations'. Bank of England Staff Working Paper, No. 1125. Available at: Food prices matter most (Accessed: 24 September 2026).

Assenza, T., Huber, S., Mogilevskaja, A. and Schmidt, T. (2026) 'When wording changes what we find: The impact of inflation expectations on spending'. Deutsche Bundesbank Discussion Paper, No. 13/2026. Available at: When wording changes what we find (Accessed: 24 September 2026). doi: 10.71734/DP-2026-13.

Buckmann, M., Potjagailo, G. and Schnattinger, P. (2025) 'Blockwise boosted inflation: Non-linear determinants of inflation using machine learning'. Bank of England Staff Working Paper, No. 1143. Available at: Blockwise boosted inflation (Accessed: 24 September 2026).

Carney, M. (2017) 'Lambda'. Speech given at the London School of Economics, 16 January. London: Bank of England. Available at: Lambda (Accessed: 24 September 2026).

Christiano, L.J., Eichenbaum, M. and Evans, C.L. (2005) 'Nominal rigidities and the dynamic effects of a shock to monetary policy'. Journal of Political Economy, 113(1), pp. 1-45. Available at: Nominal rigidities and the dynamic effects of a shock to monetary policy (Accessed: 24 September 2026). doi: 10.1086/426038.

Coibion, O., Georgarakos, D., Gorodnichenko, Y. and van Rooij, M. (2023) 'How does consumption respond to news about inflation? Field evidence from a randomized control trial'. American Economic Journal: Macroeconomics, 15(3), pp. 109-152. Available at: How does consumption respond to news about inflation? (Accessed: 24 September 2026). doi: 10.1257/mac.20200445.

Coibion, O. and Gorodnichenko, Y. (2025) 'Inflation, expectations and monetary policy: What have we learned and to what end?'. NBER Working Paper, No. 33858. Available at: Inflation, expectations and monetary policy (Accessed: 24 September 2026). doi: 10.3386/w33858.

Coibion, O., Gorodnichenko, Y. and Ropele, T. (2020) 'Inflation expectations and firm decisions: New causal evidence'. The Quarterly Journal of Economics, 135(1), pp. 165-219. Available at: Inflation expectations and firm decisions (Accessed: 24 September 2026). doi: 10.1093/qje/qjz029.

Coibion, O., Gorodnichenko, Y. and Weber, M. (2022) 'Monetary policy communications and their effects on household inflation expectations'. Journal of Political Economy, 130(6), pp. 1537-1584. Available at: Monetary policy communications and their effects (Accessed: 24 September 2026). doi: 10.1086/718982.

D'Acunto, F., Charalambakis, E., Georgarakos, D., Kenny, G., Meyer, J. and Weber, M. (2024) 'Household inflation expectations: An overview of recent insights for monetary policy'. NBER Working Paper, No. 32488. Available at: Household inflation expectations (Accessed: 24 September 2026). doi: 10.3386/w32488.

D'Acunto, F., Hoang, D. and Weber, M. (2022) 'Managing households' expectations with unconventional policies'. The Review of Financial Studies, 35(4), pp. 1597-1642. Available at: Managing households' expectations with unconventional policies (Accessed: 24 September 2026). doi: 10.1093/rfs/hhab083.

D'Acunto, F., Hoang, D., Paloviita, M. and Weber, M. (2023) 'IQ, expectations, and choice'. The Review of Economic Studies, 90(5), pp. 2292-2325. Available at: IQ, expectations, and choice (Accessed: 24 September 2026). doi: 10.1093/restud/rdac075.

D'Acunto, F. and Weber, M. (2026) 'Information and macroeconomic expectations: Global evidence'. NBER Working Paper, No. 35511. Available at: Information and macroeconomic expectations (Accessed: 24 September 2026). doi: 10.3386/w35511.

Duca-Radu, I., Kenny, G. and Reuter, A. (2021) 'Inflation expectations, consumption and the lower bound: Micro evidence from a large multi-country survey'. Journal of Monetary Economics, 118, pp. 120-134. Available at: Inflation expectations, consumption and the lower bound (Accessed: 24 September 2026). doi: 10.1016/j.jmoneco.2020.03.005.

Ghassibe, M., Wanengkirtyo, B. and Yotzov, I. (2025) 'Intertemporal pass-through'. Bank of England Staff Working Paper, No. 1130. Available at: Intertemporal pass-through (Accessed: 24 September 2026).

Guerreiro, J., Hazell, J., Lian, C. and Patterson, C. (2024) 'Why do workers dislike inflation? Wage erosion and conflict costs'. NBER Working Paper, No. 32956, revised February 2026. Available at: Why do workers dislike inflation? (Accessed: 24 September 2026). doi: 10.3386/w32956.

Hajdini, I., Knotek, E.S. II, Leer, J., Pedemonte, M.O., Rich, R.W. and Schoenle, R.S. (2023) 'Low passthrough from inflation expectations to income growth expectations: Why people dislike inflation'. Federal Reserve Bank of Cleveland Working Paper, No. 22-21R. Available at: Low passthrough from inflation expectations to income growth expectations (Accessed: 24 September 2026). doi: 10.26509/frbc-wp-202221r.

Jain, M., Kostyshyna, O. and Zhang, X. (2024) 'How do people view wage and price inflation?'. Journal of Monetary Economics, 145, article 103552. Available at: How do people view wage and price inflation? (Accessed: 24 September 2026). doi: 10.1016/j.jmoneco.2024.01.005.

Kamdar, R. (2019) 'The inattentive consumer: Sentiment and expectations'. Society for Economic Dynamics 2019 Meeting Papers, No. 647. Available at: The inattentive consumer (Accessed: 24 September 2026).

Reis, R. (2023) 'Four mistakes in the use of measures of expected inflation'. AEA Papers and Proceedings, 113, pp. 47-51. Available at: Four mistakes in the use of measures of expected inflation (Accessed: 24 September 2026). doi: 10.1257/pandp.20231033.

Reis, R. (2026) 'Why did inflation rise and fall in 2021-24? Channels and evidence from expectations'. Centre for Macroeconomics Discussion Paper, CFMDP 2026-03. Available at: Why did inflation rise and fall in 2021-24? (Accessed: 24 September 2026).

Roth, C. and Wohlfart, J. (2020) 'How do expectations about the macroeconomy affect personal expectations and behavior?'. The Review of Economics and Statistics, 102(4), pp. 731-748. Available at: How do expectations about the macroeconomy affect personal expectations and behavior? (Accessed: 24 September 2026). doi: 10.1162/rest_a_00867.

Savignac, F., Gautier, E., Gorodnichenko, Y. and Coibion, O. (2024) 'Firms' inflation expectations: New evidence from France'. Journal of the European Economic Association, 22(6), pp. 2748-2781. Available at: Firms' inflation expectations (Accessed: 24 September 2026). doi: 10.1093/jeea/jvae015.

Taylor, A.M. (2025) 'The last half mile'. Speech given at Leeds University Business School, 15 January. London: Bank of England. Available at: The last half mile (Accessed: 24 September 2026).

Taylor, A.M. (2026a) 'Central reservations'. Speech given at the Barclays-CEPR Monetary Policy Forum, London, 23 June. London: Bank of England. Available at: Central reservations (Accessed: 24 September 2026).

Taylor, A.M. (2026b) 'Stopping for gas'. Speech given at the Exante Data 10 Year Anniversary Conference, New York, 26 March. London: Bank of England. Available at: Stopping for gas (Accessed: 24 September 2026).

Vellekoop, N. and Wiederholt, M. (2019) 'Inflation expectations and choices of households'. SAFE Working Paper Series, No. 250. Available at: Inflation expectations and choices of households (Accessed: 24 September 2026). doi: 10.2139/ssrn.3383452.

Weber, M., Candia, B., Afrouzi, H., Ropele, T., Lluberas, R., Frache, S., Meyer, B., Kumar, S., Gorodnichenko, Y., Georgarakos, D., Coibion, O., Kenny, G. and Ponce, J. (2025) 'Tell me something I don't already know: Learning in low- and high-inflation settings'. Econometrica, 93(1), pp. 229-264. Available at: Tell me something I don't already know (Accessed: 24 September 2026). doi: 10.3982/ECTA22764.

Lorenzoni, G. and Werning, I. (2023) 'Inflation is conflict'. NBER Working Paper, No. 31099. Available at: Inflation is Conflict (Accessed: 25 September 2026). doi: 10.3386/w31099.

Werning, I. (2022) 'Expectations and the rate of inflation'. NBER Working Paper, No. 30260. Available at: Expectations and the rate of inflation (Accessed: 24 September 2026). doi: 10.3386/w30260.

Yotzov, I., Bloom, N., Bunn, P., Mizen, P. and Thwaites, G. (2026) 'The speed of firm response to inflation'. Journal of the European Economic Association, 24(2), pp. 736-768. Available at: The speed of firm response to inflation (Accessed: 24 September 2026). doi: 10.1093/jeea/jvaf044.

Zhang, Y. (2026) 'Rational inattention choices in firms and households'. Working paper, July. Available at: Rational inattention choices in firms and households (Accessed: 24 September 2026).