Bumps in the road? − speech by Sarah Breeden

Speech

Introduction

Good afternoon and thank you for the invitation to speak here at Cardiff Business School. It’s a real pleasure to be here in Cardiff – a city that has shown the world the value of teamwork, whether through rugby at the Principality Stadium or in its history as a hub of global trade, as well as the importance of innovation, as exemplified by the regeneration of Cardiff Bay. Similarly, Cardiff Business School has been at the forefront of nurturing future leaders and thinkers, equipping them with the skills to navigate the ever-changing economic landscape. It is the economic landscape that I wish to focus on today. And specifically, the path that inflation has taken over the past year.

Inflation has come down significantly since it hit double figures in late 2022, supported by the restrictive stance of monetary policy. But the road has not been smooth. Having fallen back to our 2% target last year, headline inflation has risen again, back up to 3.8% in the latest data for August. And we think it will have increased again to around 4% for September.

This is too high. While we expect inflation to begin to fall back from next month and gradually return to target thereafter, it is right that we question the risks around this projection. Specifically, does this latest “hump” in inflation represent a mere “bump” in the road? Or is it a sign that the disinflationary process is veering off track – that inflation is proving sticky not bumpy, meaning our hump is in fact a plateau?

As I do, I’ll look at this through three lenses:

1. What new shocks have we faced and are they likely to generate additional second-round effects?
2. Are the old shocks playing out as expected or is there news we need to respond to?
3. What have we learnt about the impact of our policy? I will then look to bring this all together and set out what it means for my policy approach going forward.

What new shocks have we faced?

This “hump” has not been a surprise. Indeed, the MPC had been forecasting a pick-up in inflation over 2025 since the time of the February MPR, with inflation then expected to rise to 3.5% in June and to peak at around 3.7% in Q3 (Figure 1). And the increase reflected factors largely unrelated to domestic inflationary forces – what I will call “external” factors – such as administered prices like water, tax and previous falls in energy prices dropping out of the annual comparison.

Figure 1: Projected cumulative change in CPI inflation from December 2024 in the February MPR^(a)

Footnotes

• Source: Bloomberg Finance L.P., Department of Energy Security and Net Zero, ONS and Bank calculations.
•  
• (a) Component-level Bank staff projections from January to June 2025 made at the time of the February 2025 MPR. For further detail see footnotes for Chart 2.19 in the February 2025 MPR. MPC projection for CPI inflation in 2025 Q3 made at the time of the February 2025 MPR.

We have received some upside news since then, with inflation now expected to peak a little higher, reaching 4.0% in September. This news too has largely reflected external factors. Using the May MPR for comparison^{footnote [1]}, upside news to energy and food prices accounts for almost all of the surprise (Figure 2), with the stronger than expected outturns for services inflation in April and July explained by administered or erratic factors, such as recent volatility in airfares^{footnote [2]}.

Figure 2: Decomposition in inflation news since the May MPR^(a)

Footnotes

• Sources: ONS and Bank calculations.
•  
• (a) Component-level decomposition of the news between the latest ONS CPI data and Bank staff projections from April to July 2025 made at the time of the May 2025 MPR. The projection for September shows the component-level decomposition of the news between the Bank staff projection at the time of the August 2025 MPR and the May 2025 MPR.

The rise in UK food price inflation, which reached 5.1% in August, reflects in large part a sharp increase in some agricultural commodity prices and so provides little signal on domestic inflationary pressures. Part of that rise in agricultural prices reflects adverse weather events. As I set out in a speech earlier this year (Breeden, 2025), severe weather events can have significant impacts on agricultural commodity prices. And, as an economy that imports a high proportion of its food – around 40%^{footnote [3]} – UK inflation is highly susceptible to such changes. Indeed, over the past year droughts in major coffee-producing regions like Brazil and heavier than usual rainfall and plant diseases in some cocoa producing regions in Africa have pushed up on prices.

There are likely to be some UK-specific factors at play too behind the rise in food prices. The increase in the National Living Wage and the change to the threshold at which employers start paying NICs are thought to have pushed labour costs higher. A new framework to reduce packaging waste (EPR) has also provided some upward pressure.

Are these likely to lead to additional second-round effects?

Externally driven cost shocks like these matter most when they lead to inflation persistence – or second round effects – when wages and other prices increase in response, keeping inflation higher for longer. Such second-round effects were important following the 2022 inflationary episode.

It has long been our central case that this “hump” in inflation would not lead to additional second-round effects. In comparison to our experience in 2022, the peak is expected to be much lower (almost a third of the size), the labour market is much looser and the demand environment weaker, all of which mean that the chance of follow-on wage and price rises is much lower. Notwithstanding the recent upside surprise in inflation I just described, all three of these things still hold true.

Nevertheless, the significant rise in household inflation expectations since the recent lows in 2024 has given me some pause for thought (Figure 3).

Figure 3: Household inflation expectations^(a)

Footnotes

• Sources: Bank of England/Ipsos Inflation attitudes Survey, Citigroup, YouGov and Bank of England calculations.
•  
• (a) Bank/Ipsos data show the median responses from the Bak/Ipsos IAS. A methodological break occurred during the Covid pandemic that means a degree of caution should be taken when making long-run comparisons with these data. The methodology notes linked in the latest IAS release for February 2025 provide more information. Data are not seasonally adjusted and the latest data points are for 2025 Q3. Since August 2022, the YouGov/Citigroup survey has been based on updated response buckets. Data are not seasonally adjusted and the latest data points are for September 2025. Dashed lines represent the series averages over 2010-19.

That household inflation expectations should rise in response to realised inflation is not unexpected (Weber et al, 2022), particularly during periods of high inflation (Pfauti 2024). But we have seen, in addition, an increase in the sensitivity of expectations to food prices. Indeed recent Bank staff analysis extending Anesti et al (2025) shows that the sensitivity of household inflation expectations to food prices has almost doubled in recent years^{footnote [4]} (Figure 4). Since household inflation expectations can influence pay demands in the wage-bargaining process, if expectations were to continue to rise with further increases in food prices, this could be a cause for concern.

Figure 4: Households’ near-term inflation expectations have been particularly sensitive to developments in food price inflation^(a)

Footnotes

• Sources: Citigroup, ONS, YouGov and Bank calculations.
•  
• (a) The chart shows the estimated coefficients from regressions of changes in Citi/YouGov one year ahead household inflation expectations on changes in the contributions of food price inflation and core CPI inflation to headline CPI inflation. The pre-2022 estimation period runs from 2005 Q1 to 2021 Q4, while the post-2022 estimation period runs from 2022 Q1 to 2025 Q2.

Whether the rise in household inflation expectations ultimately passes through into higher wages depends also on the outlook and behaviour of firms. Evidence from the literature suggests that, similar to households, firms’ expectations are sensitive to current inflation, and this sensitivity can increase during high inflation episodes (Yotzov et al, 2024). Recent moves in firms’ inflation expectations have though been much more muted than those of households (Figure 5).

Figure 5: DMP respondents’ CPI and own-price inflation expectations^(a)

Footnotes

• Sources: DMP Survey and Bank calculations.
•  
• (a) CPI expectations are based on the questions “As a percentage, what do you think the annual CPI inflation will be in the UK, both one year from now and three years from now?”. Own-price expectations are based on the question ‘Looking ahead, 12 months from now, what approximate percentage change in your average price would you expect in each of the following scenarios: lowest, low, middle, high, highest’. Respondents were then asked to assign a probability to each scenario. A point estimate is constructed by combining the five scenarios with the probabilities attached to them. Firms that do not set regular prices are not asked about their own-price expectations. All series are presented as three-month moving averages.

There is little evidence in the literature to suggest conclusively whether firms’ or households’ expectations ultimately matter most for wage growth. That said, only around 10% of households in the latest Bank/Ipsos survey say they will respond to higher inflation expectations by pushing for higher pay with their current employer. This share has been broadly unchanged for the past ten years, despite the recent rise in wage growth, suggesting perhaps that firms’ rather than households’ expectations matter most.

The relative importance of firms’ and households’ inflation expectations is likely also to be state dependent. I find it plausible that the bargaining power of firms is greater when the labour market is looser. And so it is important context that a range of indicators point to a significant degree of slack having opened up in the labour market recently. Covid-aside, the unemployment-to-vacancies ratio is at levels not seen since 2015 and the balance between staff availability and demand in the REC survey is at levels not seen since 2008. Net desired hours wanted by workers are now also at levels not seen since 2015.

We need also to consider how firms would react to any additional wage pressure. The weaker demand environment, compared to 2022, reduces their ability to pass wage increases through to prices, making it more likely that any increase will be absorbed in firms’ margins. Indeed, the Business Insights and Conditions Survey (BICS) suggests firms are now more concerned about demand than inflation. In late 2022, around twice as many firms in the BICS were concerned about rising inflation as were concerned about falling demand. As of September this year, that ratio has inverted.

Are old shocks playing out as expected?

In addition to understanding new shocks, we must of course also monitor whether disinflation from the ‘old’ 2022 shock is continuing as we expected, particularly given the noticeable stalling in the downward trajectory of services price inflation over 2025.

Services consumer price inflation has been broadly flat over recent months, at a little under 5%. This remains notably above target consistent levels (around 3%) and above levels recorded in other advanced economy regions, such as the Euro Area, where it is running closer to 3%. This could be taken as a signal that the disinflationary process from the 2022 shock is veering off track in the UK, with persistence taking longer to dissipate and with our hump turning into a plateau.

Much of this apparent slowing in disinflation can in fact be explained by the administered prices I described earlier. And as my colleague Swati Dhingra points out in an recent op-ed, these can also help explain a chunk of the difference in services price inflation between the UK and the Euro Area (Figure 6, LHS). Inflation in services set by administered and indexed prices, such as water bills and education, combined with private rent growth, is three times higher in the UK than in the Euro Area, while prices in “market-based” services such as leisure and hospitality grew at similar rates. Indeed, the recent increase in UK administered pries, combined with the lingering strength in private rent growth, explains over 90% of the EA-UK services inflation gap.

Figure 6: UK and Euro Area services price inflation by category in August 2025 and measures of underlying services inflation^(a)(b)

Footnotes

• Sources: ONS and Bank calculations.
•  
• (a) Administered and indexed prices as sewerage, dental, vehicle excise duty, tram, underground and bus transport, communications, TV licence fees, social rents, and education which comprise one-quarter (24%) of the UK services basket. Euro Area inflation statistics do not separate out private and social rents in HICP. We typically consider social rents to be administered, so include both private and social rents with administered prices for both the UK and Euro Area. Including rents within market-based services does not materially alter the story.
• (b) The low variance measure is calculated by weighting each component of services inflation by the inverse variance of the change in 12-month inflation of that component from 12 months previously. The maximum adjusted weight is capped at twice its original value. The trimmed mean measure excludes the 10% largest and 10% smallest price changes. The latest data points shown refer to August 2025.

We can also take some comfort that the disinflation process is continuing from measures of underlying services price inflation, which have continued to fall back (Figure 6, RHS). But even these measures remain at levels above what would be considered target consistent.

This strength in services inflation appears to reflect persistence in wage growth, rather than additional persistence in firms’ pricing behaviour. A simple Autoregressive Distributive Lag (ARDL) model estimated by Bank staff, shows that the level of core services inflation (excluding rents) seen since mid-2023 can be well explained by wage growth and producer price inflation (Figure 7). There was some unexplained strength in 2025 Q2, but this can be rationalised by those unusually strong increases in administered prices I described earlier.

Figure 7: ARDL estimates of contributions to quarterly core services inflation, excluding rents^(a)

Footnotes

• Sources: ONS and Bank calculations.
•  
• (a) Estimated contributions to core services inflation excluding rents are based on autoregressive distributed lag regressions of services inflation on metrics of pay growth and manufacturing PPI inflation.

While wage growth continues to fall back, it remains above levels that can be explained by economic fundamentals. The latest intelligence from our Agents however continues to suggest that pay growth will fall back to around 3.7% by the end of the year – a level that is consistent with the latest data on pay settlements too. This outlook is pretty much unchanged on the start of the year. And this suggests to me that there is little to suggest that persistence in wage growth will prove any more sticky in light of the “hump” than we had anyway expected.

Looking further ahead, wage growth is expected to fall back further over 2026. Year-ahead expectations for wage growth in the DMP Survey have crept down further to around 3.6%.and I expect this in turn to pass through to lower services price inflation.

Taken together, therefore, I judge the underlying disinflationary process from past shocks to remain on track. Wage growth has fallen back and I have little reason not to believe it will continue to do so as we look ahead, bringing services inflation lower alongside it.

What have we learnt about the impact of policy?

Of course past increases in Bank Rate have contributed to this disinflationary process by weighing on demand.

We set out our latest model-based estimates for the impact of Bank Rate in the August MPR (Figure 8). Precise estimates of the impact are inevitably uncertain, and there is evidence to suggest that the relationship can change over time (Mumtaz, 2010). This could result from non-linear or asymmetric effects of monetary policy (Stenner, 2022) or from underlying changes in the structure of the economy.

Figure 8: Estimated contribution of the overnight index swap curve (OIS) to the level of GDP^(a)

Footnotes

• Sources: Bank of England, ONS and Bank calculations.
•  
• (a) The estimates for the contribution of the OIS curve to the level of UK GDP are the sum of the impacts on consumption, business investment, housing investment and the net imports offset in the MPC’s baseline projections. The net imports offset component reflects that part of the reduction in consumption from higher interest rates will arise from lower spending on imported goods, which mechanically pushes up GDP. These estimates are based on the standard treatment of the impact of changes in Bank Rate and the OIS curve in the Bank’s forecasting models. Bank staff estimate the impact of changes in the OIS curve between each Monetary Policy Report on the level of GDP, and these are then summed to obtain estimates of the total contributions. The estimates incorporate an impact from changes in the OIS curve on the level of GDP that persists over several years. The level of the aqua bars in the first data point in 2021 Q3 therefore reflects the estimated impact on demand of changes in the OIS curve from several years prior to that date. The effects of the estimated impact of changes in the OIS curve on sterling exchange rates have been excluded. The final data are for 2025 Q2.

These uncertainties have made it important for the MPC to consider bottom-up analysis alongside top-down model-based estimates of monetary policy transmission. By doing so we look to judge how restrictive we are with policy. One of the clearest indicators that higher interest rates have weighed on demand has been the rise in the household savings ratio which now stands at just over 10% (although other factors - precautionary motives related to heightened uncertainty or structural factors – could have played some role too).

As set out in the August MPR, several of these bottom-up indicators suggest the impact of past increases in Bank Rate on the level of GDP is now around its peak. For example, survey evidence from the NMG and the DMP show that the numbers of households and firms reporting higher interest rates as a reason for higher saving or constraining capital spending are falling. Nonetheless, the impact of some channels continues to build. For example, 3.4 million mortgage holders expect to see an increase in payments over the next three years, compared to 2.8 million expecting a decrease (Figure 9).

Figure 9: Proportion of owner-occupier mortgages by estimated changes in monthly mortgage costs, from June 2025 to 2028 Q3^(a)

Footnotes

• Sources: Bloomberg Finance L.P., FCA Product Sales Data and Bank calculations.
•  
• (a) There are around 8,908,000 mortgages in the UK. There are 100 squares, each representing 1% of the total current stock of UK mortgages (around 89,080 mortgages), rounded to the nearest 1%. The projection uses the OIS curve as at 31 August 2025 and the latest available data (2025 H1) on the stock of outstanding mortgages. Changes in payments on variable-rate mortgages are calculated using the implied change in the OIS curve, and changes in payment on fixed-rate mortgages are calculated by assuming that mortgagors refinance onto a typical fixed rate implied by the OIS curve at the point that their fixed-rate contract ends. Mortgages with less than £1,000 outstanding are excluded. These data do not include buy-to-let mortgages or mortgages that are off balance sheet of authorised lenders, such as securitised loans or loan books sold to third parties. 

It is partly through demand, which in turn feeds through into inflation expectations, that the restrictive stance of monetary policy is expected to lean against remaining persistence. But monetary policy restrictiveness may also plausibly influence inflation expectations directly, by reassuring the strength of the inflation anchor which in turn can move inflation via price or wage setting decisions.

Looking ahead, although the impact of past rises in Bank Rate looks to be around their peak, they will still be weighing on the level of demand and so continue to contribute to the disinflationary process. While the degree of restrictiveness has fallen over the past 18 months as the MPC has reduced Bank Rate, I still judge the current monetary policy stance to be restrictive and so continuing to squeeze persistence from the system.

The inflation and policy outlook

Taken together, what does this mean for my outlook for inflation and so policy?

I do not see evidence that the disinflation process is veering off-track. Instead it remains my central case that the “hump” will prove just a bump in the road. As I set out:

• The current “hump” has been driven by external shocks that are not a reflection of domestic inflationary pressures.
• With a looser labour market and weaker demand than in the 2022 inflation episode, I continue to see it as unlikely that we will see additional second-round effects.
• The underlying disinflationary process from past shocks remains on track. Wage growth has fallen back and I have little reason to believe it will not continue to do so. This should, in turn, pass through to lower services price inflation next year.
• Monetary policy continues to contribute to the ongoing disinflationary process. While the degree of restrictiveness has fallen over the past eighteen months, I still judge the monetary policy stance to be restrictive and to be squeezing persistence from the system.

But there are, of course, risks around these conclusions. And I am mindful of the risk of moving to a high inflation track, particularly after four years of above target inflation.

First, it is possible that new shocks will materialise. As my colleague Meagan Greene set out last week, there are reasons to believe we have entered an age of supply shocks (Greene, 2025). We have experienced a series of repeated supply shocks over recent years, and transformational trends such as climate change and geoeconomics mean it is possible (perhaps even probable) that we may see more going forward.

Second, there might have been a structural change in firms’ pricing behaviour so that the “hand-off” from lower wage growth to services price inflation may not occur as we are expecting.

I described earlier how services price inflation has been explicable given wage growth over recent years. But there are some signs that suggest that the pricing behaviour of firms remains consistent with a high inflation regime. The price microdata underlying the CPI show that the net share of services prices (excluding catering and accommodation) increasing each month has not shown any meaningful moderation since 2022 (Figure 10). This might just reflect additional (one-off but repeated) cost shocks. But it might also indicate something about firms’ margins, or their ability and willingness to push through price changes.

Figure 10: Share of price changes (net up/down balance) as a fraction of the underlying bucket^(a)

Footnotes

• Sources: ONS and Bank calculations.
•  
• (a) Chart shows a 12-month rolling average. The latest data points refer to August 2025.

Third, it might take longer for the degree of slack in the economy to squeeze the remaining persistence from the system.

Indeed, as we set out in the August MPR, there remains significant uncertainty over the degree of slack in the economy. Model-based estimates of the output gap based on capacity utilisation and labour market data suggest a margin of slack has opened up. However, the signal from nominal indicators remains consistent with a margin of excess demand (Figure 11).

I place more weight on the capacity utilisation and labour market measures, and so judge that a degree of slack has opened up. However, this disconnect – and in particular its scale - is a reminder of the nominal puzzle that we face. Indeed, it is possible that the degree of slack is smaller; the degree of persistence larger; or the mechanisms between them weaker than we have assumed. In any of these cases, it might take longer for inflation to return sustainably to target.

Figure 11: Model-based estimates of the output gap^(a)

Footnotes

• Sources: ONS and Bank calculations.
•  
• (a) The model is estimated over 2000–2025 Q2 using the two-step estimator from Doz et al (2011), obtained from running the data through a Kalman filter and smoother once. The first factor of the dynamic factor model is interpreted as a measure of slack. The factor is then mean-variance adjusted to the MPC’s baseline output gap estimate over 2000–2025 Q2. The labour market block is estimated using survey indicators of slack and the vacancy gap. The capacity utilisation block includes a range of surveys of capacity utilisation also used in the left panel of this chart. The nominal block contains measures of pay and underlying inflation. The data are shown to 2025 Q2.

In such a world it may be tempting to wait to see the “whites of disinflation’s eyes” before looking to reduce the restrictiveness of policy further. But managing the upside risks to inflation in this way brings risk in the other direction: holding policy too tight for too long comes with costs to output and employment, which could then pull inflation below target .

More broadly, there are downside risks to demand which could also pull inflation below target further out. I have for some time seen risks to the outlook for consumption given that it is predicated on a substantial decline in the savings rate. Our previous projections of material declines in the savings rate have failed to materialise, possibly signalling a structural change in household savings behaviour. And heightened uncertainty whether for domestic or global reasons might weigh on demand too.

Conclusion

We expect inflation this month to peak at 4%, well above our target. It is too high and it is our job return it to target sustainably.

The good news is that this "hump" reflects external shocks, and in the current environment it is unlikely, in my judgement, to lead to additional inflationary pressures. Moreover, I have not seen any evidence yet to suggest that the underlying disinflationary process from past shocks is veering off-track. So far so good.

However, the path ahead is not assured. Indeed we face risks on both sides – that greater persistence means inflation turns out to be sticky not bumpy, and that unexpectedly weak demand causes an unanticipated easing in price pressures. Steering a path that manages both risks is not easy.

With this context, I will be focused on identifying those indicators that give me confidence that the future disinflationary process - in particular the “hand-off” from wages to services price inflation - is remaining on track. Indicators of pricing intentions, from surveys and intelligence from our Agents, will be important here, as will conversations like those I have had in Cardiff today underlining the important part you’re playing in the team trying to understand what’s going on in the economy. Indeed we might think of indicators such as these as signposts that help determine whether we are likely to veer off track. And they will therefore be key in determining when it might be appropriate to remove further restrictiveness.

I would like to thank Nicola Shadbolt for her assistance in drafting these remarks. I would like to thank Emily Fry and Swati Dhingra for their work comparing administered services price inflation in the UK and Euro Area. I would also like to thank Giulia Gardin, Matthew Naylor, Ivan Yotzov, Meghna Shrestha and Andrea Sisko, Rupert de Vincent-Humphreys, Carleton Webb, James Talbot, Martin Seneca and Andrew Bailey for their helpful input and comments. The views expressed here are not necessarily those of the Monetary Policy Committee (MPC) or the Financial Policy Committee (FPC).