Learning about earnings: Can we rely on labour market slack to mitigate the latest supply shock? − speech by Megan Greene

Speech

Introduction

Thank you for inviting me to the LSE to speak. It’s nice to be back here—I often studied in your library when I was an undergraduate in the US working in London for the summer. Come to think of it, I may have snuck in so please don’t alert the authorities.

Over the past few months, a lot has happened in the world of geopolitics…and yet very little has fundamentally changed. Passage through the Strait of Hormuz remains beset by attacks and potential new chokepoints in shipping have emerged such as the Bab el-Mandeb Strait. Although Middle Eastern oil flows have at times returned to near pre-war levels, expensive and inefficient workarounds have kept energy prices high (Figure 1). Wholesale gas prices are close to triple their level at the start of the year, with oil spot prices just under double their early-January level.^{footnote [1]} And the prices for refined products have remained elevated despite volatile spot prices. Even if the conflict in Iran ends tomorrow, we have already been hit by a protracted energy shock.

Figure 1: Evolution of energy prices since start of 2026

Index, 100 = price at 1 Jan 2026

Footnotes

• Source: ONS, Bloomberg Finance L.P. and Bank calculations.
• (a) Dashed vertical lines denote selected key events. Dates may not be exact due to frequency of data. Latest datapoints are for 19 October 2026.

I have previously laid out the various avenues via which such a negative supply shock can propagate through the UK economy, highlighting the important role of second-round effects in embedding inflation in the economy. As I set out in my last speech in June, they can take hold through the price- or wage-setting channel, or both. For today’s purposes, I will focus specifically on the wage-setting channel.

Wages represent a material cost for firms, particularly in the services sector where labour accounts for a large share of total costs. Wages are also a significant determinant of incomes, thus underpinning consumption. As a result, measures of wage growth offer useful insight into domestically-generated inflation. Of course, not all wage growth is inflationary. If wages rise because productivity improves, the implications for inflation may well be limited. But wages can also act as the mechanism through which temporary negative supply shocks become embedded in domestic inflation, particularly when workers seek to protect real incomes and firms then raise prices to protect their margins.

The MPC has judged that there is slack in the labour market, as evidenced by the ratio of vacancies to unemployment relative to equilibrium (Figure 2). Some of my colleagues and I have argued that the wage-setting channel poses a smaller risk of generating second-round effects than the price-setting channel. If firms are facing higher costs and have experienced years of margin compression, they are likely to pass on some of their higher costs by raising prices. This is consistent with intelligence that we’re hearing from our Agents. But if there is slack in the labour market, worker power is relatively weak. Even if workers demand higher pay as their costs go up, employers may deny that request given the large pool of talent from which to choose.

Figure 2: Vacancy-unemployment (V-U) ratio and V-U equilibrium

Ratio^(a)

Footnotes

• Source: Advertising association/ World Advertising Research Centre Expenditure Report, ONS and Bank calculations. Latest data point is for 2026Q2.
• (a) The equilibrium V/U ratio is estimated using an error-correction model over the period 1982–2024. The real cost of vacancy posting and hourly labour productivity are included as long-run determinants for the level of vacancies. The model also includes controls for short-term movements in these variables (Stelmach, Kensett , & Schnattinger, 2025)

Given our monetary policy strategy has relied partly on slack mitigating the impact of second-round effects, it’s worth exploring how much labour market slack there really is. The wider economy has proved surprisingly resilient, implying there is less slack than some labour market indicators suggest. If so, there is a greater risk of second-round effects emerging through the wage-setting channel than we initially thought.

We may not have a clear steer on whether the shock from the Iran conflict gives rise to material second-round effects through wage- or price-setting until early next year when our Agents’ pay survey is out in February and the brunt of 2027 pay settlements are agreed in January and April. But if we wait for this evidence to lean against second-round effects with a tighter monetary policy stance, we will already be too late given the lags with which monetary policy affects the economy and inflation.

As outlined in our Forecast Evaluation Report and the Macro-technical paper alongside it, predicting wage growth has been a challenge for the Bank, particularly since the Global Financial Crisis. As shown in Figure 3, Bank forecasts tended to systematically predict higher wage growth than what we experienced until the Covid period, after which our forecasts systematically underestimated the strength of wage growth.^{footnote [2]}

Figure 3: MPR forecasts and outturns of wage growth^(a)

Annual growth (%)^(a)

Footnotes

• Source: ONS and Bank calculations.
• (a) The grey lines show forecasts from different vintages. The aqua line shows the outturns k=12 quarters after the first data release.

Our suite of wage models, like many macro forecasting models, are largely driven by historical relationships. The recent persistent undershooting of wage growth forecasts may suggest something structurally shifted in the labour market during the pandemic. As I argued in my 2024 speech (Greene, 2024), this could reflect reduced matching efficiency, lower search intensity, labour hoarding, persistent recruitment difficulties and weaker participation. It could also reflect stronger real wage resistance following the 2022 energy shock than historical relationships have implied (Pill, 2025). Wage growth has generally decelerated since 2023, suggesting these upward pressures have been unwinding, but they may not have unwound entirely.

Models inevitably simplify reality, and their performance depends critically on the quality of the signals they are trying to capture. One important development over the past decade has been an expansion in the range of wage indicators we look at, reflecting efforts to measure underlying wage pressures from multiple perspectives.

But this richer information set has brought its own challenges. At the moment, wage data are like a Rorschach Test; they can be used to confirm anyone’s preconceived view of slack in the labour market because they are giving mixed signals (Figure 4). Different indicators often tell different stories, reflecting differences in coverage, frequency, composition and methodology (Villar, Moyon, & Ahn, 2025). The task for policymakers is therefore not simply to monitor wages, but to understand the dynamics underpinning different metrics.

Figure 4: Measures of private sector wage growth

Annual growth (%)^(a)

Footnotes

• Sources: Bank of England settlements database, DMP survey, HMRC, Indeed, ONS and Bank calculations.
• (a) The private sector regular pay growth line shows the ONS measure of private sector regular AWE growth, measured as the three-month average on the same period a year earlier. The DMP lines show average realised pay growth and average expected pay growth (shifted 12 months forward) from the DMP survey on the same basis. The HMRC RTI line shows a proxy of growth in median pay in the private sector, also measured as the three-month average on the same period a year earlier. Bank staff construct the HMRC RTI measure by taking median pay for each industry – excluding public administration and defence, education, health and social work – and weighting them by industry shares in payrolled employment. Annual growth is then calculated using the Laspeyres formula, holding industry weights fixed at their levels a year earlier to reduce the effect of changes in workforce composition. The Indeed Wage Tracker line shows annual average job title matched pay growth for UK job vacancies. The latest data points are for the three months to August 2026 for private sector regular pay and the Indeed Wage Tracker and September 2026 for the HMRC RTI private sector median, BoE settlements and the DMP survey.

This brings me to my main focus today. Given the importance of understanding wage-setting behaviour and what it might mean for slack and second-round effects, I want to dig into some of our key wage indicators to parse signal from noise. I will start by examining how these indicators differ and where in the wage-setting process they are likely to be most informative. I will then consider what they can tell us about the recent developments and what signal we might glean about wage dynamics going forward. Finally, I will examine how confident we can be that a material degree of slack will restrain second-round effects and inflation via the wage-setting channel.

Getting into the weeds isn't always glamorous, but in policymaking as in gardening, sometimes you just have to get your hands dirty.

Measures of wage growth

When confronted with a range of indicators, it is often useful to take a step back and consider exactly what information each one conveys. After all, wage indicators measure different things and are collected in different ways. This is a feature, not a bug - the differences between measures allow us to build a richer picture of wage developments within the domestic economy to assess the degree of current and emerging wage pressures.

In light of this, Table 1 brings together some of the key metrics used by the Bank of England to monitor UK wage growth. While not exhaustive, the selection illustrates the breadth of information available to policymakers and the different insights a variety of measures can provide, as demonstrated by the range of questions presented in the second column: each of these metrics has something different to offer when it comes to understanding wage pressures.

Table 1: Wage growth measures^(a)

Footnotes

• Sources: ONS, Bank of England settlements database, DMP, Indeed Hiring Lab
• (a) Table 1 gives an overview of wage measures. Coverage estimates are approximate and reflect monthly figures. PAYE stands for Pay As You Earn. Estimates relating to the Bank’s pay settlements database refer to private sector firms. Number of Indeed job vacancies is not publicly available. However, previous research suggests that there were 16 million job postings with wage data over 2018-2022 for the UK (Adrjan & Lydon, 2023).

Naturally, some measures receive more attention than others. Chief among these is the average weekly earnings (AWE) series, which helps us answer the question “what is the growth of the average wage?”. AWE is an accredited official statistic published with a two-month lag by the Office for National Statistics (ONS), derived from the Monthly Wages and Salaries Survey (MWSS).^{footnote [3]} The MWSS is issued to around 9,000 businesses and covers approximately 15 million employees across Great Britain. While adjustments are made to account for smaller firms, the survey includes only businesses with 20 or more employees. Nevertheless, AWE provides a broad measure of the change in pay levels across the economy and is thus useful for gauging domestic inflationary pressures stemming from changes in wages.

At the Bank, we tend to focus on private sector regular AWE growth as opposed to the whole economy (including the public sector) for a few reasons. Private sector pay feeds more directly into the CPI measure of inflation, particularly when it comes to services, where labour costs make up a large chunk of total costs. In addition, UK public sector pay is often determined by government policy decisions. As such, wage-setting in the private sector is generally more closely tied to labour market conditions, whereas institutional pay setting arrangements and fiscal constraints can often influence public sector pay outcomes. Monetary policy therefore has more influence over private sector pay. Furthermore, private sector regular pay excludes volatile bonuses while they’re included in total pay. For these reasons, I’m going to focus on private sector regular AWE pay growth for the rest of this speech.

Pay As You Earn (PAYE) Real Time Information (RTI) data provides an alternative read on wage growth. This is published by the ONS on a monthly basis, in conjunction with HMRC, and can be used to help answer the same question as the AWE: “what is the growth of the average wage?”. Median estimates of the data are available with less of a lag than the official AWE measure, though they are often heavily revised. In addition, its production does not rely on survey responses. Instead, the information comes directly from HMRC’s PAYE RTI system, which covers the whole population of employees payrolled by employers, allowing for more detailed breakdowns of the data. The data is presented as gross pay, which includes bonus pay and other one-off payments (but not pension contributions) unlike private sector regular AWE. Arguably, this can make it harder to gain a clear view on wage growth and makes it more difficult to compare RTI and AWE.

We can also draw on wage growth measures derived from responses to the Decision Maker Panel (DMP) survey to answer the question “what assessment do firms make of their own wage growth?”. The DMP is a large, monthly panel survey of Chief Financial Officers from small-, medium- and large-sized UK businesses. Since 2022, the survey has included a number of questions focused on firms’ wages. This means that the back data available for wage analysis is limited. In 2026, the survey included wage growth data from approximately 700 firms, covering around 300,000 employees each month. Responses are used to construct measures of both realised and expected wage growth. The former tells us what has actually happened to firms’ own wage growth over the last year, while the latter gives us an idea of what firms anticipate will happen to their wage growth over the next year. Forward-looking indicators are hard to come by – an unfortunate truth for those of us in the forecasting business – but they are particularly useful given that both shocks and monetary policy take their time to work their way through the economy. As such, I am monitoring this measure closely as the current shock unfolds.

Pay settlements data can also provide us with a glimpse into the future path for wage growth. Such data allows us to shed some light on the pay decisions firms are making with regard to their current employees, helping us answer the question “what pay increases are firms agreeing for their existing employees?”. However, settlements data captures changes in basic pay and does not capture all sources of earnings growth, such as progression, merit and performance-related pay, or bonuses. We can access pay settlements data through the Bank of England’s settlements database. This database brings together information collected from businesses across the UK economy by the Bank’s Agents and external data providers, collected at various frequencies. Given that these figures can be updated retrospectively as more settlements are finalised, they are subject to some revision. We also get a sense of how pay settlements are likely to evolve over the next year in the Agents’ pay survey, released in February but with intelligence coming in from our Agents in the lead up to that.

Settlements data can only tell us so much. By its nature, its focus is on those workers already employed. But what are firms willing to pay new hires? The Indeed Wage Tracker provides insight into this, measuring the median annual change in advertised wages and salaries found on the Indeed platform across job title, region and salary-types. The sample size of each is reliant on a number of different factors, including the amount of total job vacancies, the proportion that is posted online and the share of these postings that contains wage or salary information (Adrjan & Lydon, 2023). This data is collected in the UK and a range of developed countries, allowing for cross-country comparison.

This is a whistle-stop tour of wage growth metrics. Each tells us something slightly different about wage growth pressures in the economy. I’ll now put them to use to explore present and future developments in wage growth.

Recent and current wage growth: two measures, different stories

Having gone through the various wage metrics, let’s use them to try to work out how much slack there is the labour market now and whether the wage-setting process has changed. For this, I’ll look at private sector regular AWE and its median RTI equivalent (proxied for the private sector), which both answer the question “what is the growth of the average wage?”. The good news is that both metrics show wage growth broadly continuing to moderate since recent peaks in 2023 (Figure 5). But these two indicators tell very different stories about the extent of that progress.

Figure 5: Private sector pay growth according to AWE and RTI

AWE private sector regular pay, RTI private sector proxy median^(a)

Footnotes

• Source: ONS and Bank calculations.
• (a) See note (a) under Figure 4 for an explanation of how HMRC RTI private sector proxy is constructed.

After almost 3 years of gradual easing, recent AWE data broadly suggest a further and more rapid leg down in wage growth, coming in at 3.0% in August (the aqua line in Figure 5). In contrast, our primary RTI metric of private sector wage pressure – growth in median pay – remains at a considerably higher rate of 4.4% in September (the orange line in Figure 5). Historically, these measures have tracked each other reasonably closely, albeit over a relatively short sample given RTI data are only available from 2014. Since early 2026, the divergence between them has become unusually large by historical standards.

RTI and AWE measures are answering the same question and both are constructed using sectoral wage growth and weighted according to the level of employment.^{footnote [4]} So what's behind their divergence now? To decompose the wedge between these different measures, I’ll use a measure of mean pay growth derived from the RTI data (showing annual wage growth of 5% in August), which is more comparable to mean private sector regular pay growth from AWE (showing annual wage growth of 3% in August).

The wedge between these measures is explained primarily by three factors:

• Differences in wage growth within industries: for a given industry wage growth in AWE is different to that in RTI. This is possible due to the different data underlying the two (a survey in the case of AWE, administrative data for RTI).
• Between-industry compositional effects: This arises because the industry shares underlying the two differ and can change over time as employment shifts between industries.
• Differences in flexible pay such as bonuses: AWE regular pay doesn’t include bonuses and arrears, whereas RTI does.

Differences in wage growth between industries and within industries can both help us understand underlying wage dynamics. Wage growth driven by changes in industry shares has implications for aggregate wage pressure and can give us a sense of overall slack. Wage growth within sectors reflects firms’ pay-setting decisions and can reveal whether they have changed in the face of a shock.

Figure 6 isolates the contributions to the wedge from the three sources. In August the wedge was 2 percentage points and has grown since late 2025 due to steadily rising compositional differences (pink area), fluctuating contributions from within industry differences in pay growth (purple area) and bonuses (blue area).

Figure 6: The wedge between RTI and AWE has been growing

Decomposition of the wedge between RTI mean private sector proxy and private sector regular AWE ^(a)

Footnotes

• Source: ONS and Bank calculations.
• (a) ‘Wedge’ is the difference between mean RTI pay, private sector proxy and private sector regular AWE. In August the RTI metric was 5% and AWE was 3%, giving a wedge of 2%

Of the three main components explaining the wedge between AWE and RTI, let’s first address the easiest to interpret: flexible pay. AWE private regular pay data exclude bonuses and arrears, while private sector RTI data include them. We want to compare apples-to-apples when we are gleaning signals from the two series. To do this, we assume that the bonuses (and arrears) captured in the AWE measure of total pay are also present in the RTI data and subtract their contribution from the RTI measure. This is necessarily a simplifying approximation and so the results should be interpreted with some caution. Bonuses and cost of living payments played a significant role in explaining the wedge between the two series in 2022-23 and earlier this year before waning more recently. During 2022-23 bonuses and one-off payments were common across sectors, whereas more recently these have been concentrated primarily in finance and professional services.

Now let’s turn to the role of compositional effects in creating a wedge between AWE and RTI. We can do this by comparing the changes in sectoral employment weights in private sector AWE and private sector RTI over the past year (Figure 7). This chart shows the year-on-year changes in the latest data, but it reflects trends that have persisted throughout 2026. AWE shows employment shifting towards lower-paid sectors, with rising weights for industries such as accommodation and food services and falling weights for higher-paying sectors such as financial and professional services. RTI, however, paints the opposite picture, with employment shifting slightly away from lower-paying sectors towards higher-paying ones.

Figure 7: According to AWE employment has been shifting to lower-paid sectors, but RTI suggests the opposite

Change in sectoral shares across RTI and AWE over past year ranked from lowest to highest average monthly earnings^(a)

Footnotes

• Source: ONS and Bank calculations.
• (a) The change in each sectors’ share of total employment is calculated from August 2025 in the case of AWE and September 2025 in the case of RTI. Ranking is based on the level of non-seasonally adjusted regular AWE averaged between August 2025 and August 2026, rather than RTI.

What signal can we draw from this? There is really no silver bullet here, and this is exactly why we look at multiple indicators to get a complete picture of the labour market. But on balance, I’d put slightly more weight on the compositional story from the RTI. The RTI series is drawn from administrative data rather than a survey. RTI doesn’t require imputing employees who work in small firms (under 20 employees) as the AWE series does. Finally, the narrative from the RTI data chimes with the Labour Force Survey (LFS), which also indicates that employment has shifted to higher paid industries over the past year. There are reasons to treat the LFS with caution; it suffers from low response rates and the industry classifications are self-reported and may not always be accurate. On balance, I place slightly more weight on the RTI’s narrative that workers are shifting towards higher paying industries. This means that the AWE may be giving us an artificially low reading of wage growth because of the specific compositional effects it incorporates.

Let’s now take a look at differences in pay growth within industries in the AWE and RTI series. Figure 8 shows wage growth within various sectors in the latest AWE and RTI releases. This chart shows year-on-year changes in the latest data, but this is consistent with the trends we have seen over the past year. The AWE data suggests that, on balance, wage growth in lower paying sectors was running hotter than for higher paid sectors. The RTI data, however, suggest wage growth was faster in higher paying sectors.^{footnote [5]} For the reasons I’ve just explained, I have a bit more faith in the RTI data. But again, these discrepancies point to the importance of making use of a range of metrics.

Figure 8: Wage growth has been stronger across most sectors according to RTI

Annual pay growth by industry in AWE and RTI, ranked from the lowest to highest in average monthly earnings ^(a)

Footnotes

• Source: ONS and Bank calculations.
• (a) Ranking is based on the level of non-seasonally adjusted regular AWE averaged between August 2025 and August 2026, rather than RTI.

To isolate wage growth within industries, we can adjust AWE and RTI for compositional effects and strip out bonuses from RTI (Figure 9). These adjusted measures are shown in the solid lines, whereas the original series are represented in the dotted lines. Comparing the solid and dotted aqua lines shows that compositional effects are currently pushing down on the AWE by 0.4pp. Doing the same for the purple lines shows that compositional effects and bonuses are pushing up on RTI by 0.6pp (0.2pp due to compositional effects and 0.4pp due to bonuses). These adjusted measures indicate wage growth of 3.4% according to the AWE and 4.5% for the RTI measure.^{footnote [6]} In my view, both remain above levels consistent with an inflation target of 2% and suggest there may be less labour market slack than we thought.^{footnote [7]}

Figure 9: Our adjusted measures suggest pay growth is higher than implied by AWE but lower than implied by RTI

Private sector AWE and RTI mean private sector proxy with constant weights and removing bonuses^(a)

Footnotes

• Source: ONS and Bank calculations.
• (a) Dotted lines show headline measures; solid lines show private sector regular AWE adjusted for compositional effects and a private sector proxy measure of mean RTI adjusted for compositional effects and bonuses and arrears.

Where does this leave us? Our adjusted measures of wage growth remain above levels consistent with our inflation target of 2%. The most frequently cited indicator that we use, private sector regular AWE, seems to have been consistently undercooking actual private sector wage growth since late-2025.

Looking ahead: what can guide us?

Over the past year, different metrics of wage growth have told different stories. However, there is strong evidence to suggest that underlying wage growth is higher than that reported by private sector regular AWE. What we ultimately want to know, however, is how wage setting will evolve going forward, particularly in light of yet another negative supply shock. For this, I’ll look at some forward-looking indicators—specifically the DMP, pay settlements and the Indeed Wage Tracker.

The DMP has been an invaluable source of information on wages since wage questions were introduced in 2022. The survey directly asks firms about wage changes for their employees, and so arguably provides the clearest window into firms’ wage-setting processes. Realised wage growth has broadly tracked the trends in AWE and RTI, though there have been some periods of divergence. The DMP recorded a lower peak in 2023 and indicated more gradual moderation in wage growth thereafter. We can see this in Figure 10, which plots DMP realised wage growth and one-year ahead expected wage growth, at the time of the survey, alongside the RTI and AWE measures.

Figure 10: Realised wage growth according to DMP has tended to track RTI and AWE

Private sector regular AWE, private sector proxy median RTI, expected and realised DMP annual wage growth

Footnotes

• Source: ONS, DMP and Bank calculations.

Expected wage growth and realised wage growth have also diverged, most notably when the wage questions were introduced in 2022. When asked in the middle of 2022, DMP respondents reported year-ahead expected wage growth of around 5%. A year later, realised wage growth was close to 7% according to the DMP and 8% according to AWE.

The DMP wasn’t alone in underestimating the strength in wage growth during this episode—we also undercooked the rise in wage growth in 2022/23. Figure 11 shows the DMP own-wage growth forecast errors and our MPRs’ private sector regular AWE forecast errors. Positive values indicate that realised wage growth exceeded expectations or forecasts. The DMP errors appear smaller than the MPR forecast’s; firms have seemingly been better at anticipating wage growth than we have. In the three months to September, the DMP's one-year-ahead wage growth expectations point to a levelling-off in wage growth at 3.4% rather than a material further easing.

Figure 11: DMP firms have tended to have lower forecast errors for wage growth than the MPR

Forecast errors in wage growth ^(a)

Footnotes

• Source: ONS, DMP and Bank calculations.(a) The chart starts in 2023 because that’s the first year in which we were able to see how accurate answers to the DMP expected wage growth question was.

Pay settlements, both expected and realised, provide a narrower view of wage-setting than the DMP. They capture the pay increases employers agree, or expect to agree, with existing employees over the coming year and only cover basic pay.^{footnote [8]} Settlement expectations collected by the Bank's Agents—including in the Agents’ pay survey—have tended to provide a decent guide to subsequent realised settlements, including during the recent period of elevated wage growth in 2022/23.

Figure 12 shows that median pay settlements across sectors (the purple line) also tracked AWE growth (the aqua line) closely from the early 2000s until 2020, though this relationship has weakened post-Covid. The range of median settlements across sectors (shaded purple area) has also varied over time, falling back more recently from Covid-era highs.

Figure 12: Pay settlements have tended to track private sector regular AWE growth

Private sector regular AWE pay growth and pay settlements^(a)

Footnotes

• Source: ONS, Bank of England settlements database and Bank calculations.(a) The private sector regular pay growth line shows the ONS measure of private sector regular AWE growth, measured as the three-month average on the same period a year earlier. The range and median of sector settlements are estimated from the Bank of England’s settlements database. Classification of sectors into SIC codes involves some judgement. Sector settlements are measured as the three-month weighted average on the same period a year earlier, the median refers to the median sectoral settlement, the range is the range of sectoral settlements. Latest datapoint is for August 2026.

The noticeable gap between private regular AWE and pay settlements over the past five years can be explained by the state of the labour market. Faced with recruitment difficulties and elevated staff turnover, firms had stronger incentives to raise pay to attract and retain workers. As a result, there was ‘pay drift’; actual wage growth often exceeded that implied by pay settlements, reflecting pay for new workers, retention-related pay increases and other adjustments outside formal settlement rounds. This is consistent with academic literature that finds that workers achieve faster pay growth when outside job opportunities improve, and that the threat of higher turnover can prompt firms to raise pay more broadly across their workforce (Jäger, Roth, Roussille, & Schoefer, 2024) (Sokolova & Sorensen, 2021).

Figure 13: Post-pandemic wage growth for existing employees diverged from settlements

Wage growth of stayers and baseline pay settlements ^(a)

Footnotes

• Source: ONS, New Earnings Survey (NES) and Annual Survey of Hours and Earnings (ASHE) and Bank of England settlements database.
• (a) ‘Existing employees’ are people who remain in the same employment in successive waves of the survey.

As shown in Figure 13, this meant that basic pay settlements became a less reliable guide for the wage growth of existing employees. In 2023, pay settlements were 5.2% whereas wage growth for existing employees was 8%.

Figure 14 shows the contributions of job switchers and existing employees to wage growth. Relative to historical experience, retention pressures meant wage growth has been driven disproportionately by pay increases for existing employees rather than job-switchers since 2022.

Figure 14: Existing employees have made an even larger contribution to wage growth in the past 5 years

Contribution of existing employees and those switching jobs to wage growth ^(a)

Footnotes

• Source: ONS, New Earnings Survey (NES) and Annual Survey of Hours and Earnings (ASHE).
• (a) ‘Existing employees’ are people who remain in the same employment in successive waves of the survey. Those switching jobs are those who are in a different employment in the subsequent wave of the survey.

As I outlined in a recent speech (Greene, 2026), the labour market is considerably weaker now than it was in 2022. As aggregate demand has softened, demand for labour has also eased, reducing recruitment difficulties and, in turn, the pressure on firms to grant large pay increases. Expected pay settlements may therefore provide a more reliable guide to future wage growth than they did during the post-pandemic period.

At present, our Agents report that expected settlements for 2027 will be similar or only slightly lower than in 2026, suggesting the wage disinflation process may be stalling. For the past two years, actual settlements have come in above what was projected in the Agents’ pay survey (4% in 2025 vs 3.7% in the pay survey; current estimate of 3.6% in 2026 vs 3.4% in the pay survey), so I expect the risk to the latest guidance is to the upside. Moreover, the vast majority of pay settlements will be agreed between January and April (Figure 15), when we expect inflation to be above 4%. This is well above the threshold in which households are more sensitive to inflation outturns in forming their expectations.

Figure 15: Settlements over the year

Millions of employees

Footnotes

• Sources: Bank of England settlements database and Bank calculations. The latest datapoints are for August 2026.

Finally, the Indeed Wage Tracker offers a highly forward-looking and high frequency insight into wage pressures. The Indeed Wage Tracker captures the pay firms are willing to offer new recruits and therefore captures those workers that the pay settlements data omits. Their main limitation is that they – typically – represent a small proportion of the work force who have in recent years made a relatively small contribution to overall wage growth. However, as retention pressures have eased and firms are less reliant on large pay increases to retain staff, wages offered to new recruits may play a somewhat larger role in explaining overall wage developments as they did in the years before the pandemic.

Since late 2023, the Indeed Wage Tracker has tended to imply slightly higher wage growth than private sector regular AWE (on average about 0.8pp) (Figure 4). In August, the UK Indeed UK Wage Tracker stood at 4% (Figure 15) and while UK wage growth remains high relative to many of its peers, it has become less of an outlier in recent months. However disinflationary wage momentum here also seems to be levelling off, with the wage tracker remaining stuck at around 4% since the turn of the year.

Taken together, I place considerable weight on the DMP and pay settlements, while monitoring the Indeed Wage Tracker as a higher frequency indicator of broader labour demand conditions. Collectively, these measures suggest that the wage disinflation process is stalling out with pay growth likely to stabilise over the coming year closer to 3.5% than 3%. I believe this remains above levels consistent with our inflation target and suggests there is less labour market slack than we thought.

What will I be monitoring for the emergence of second-round effects?

There is a plethora of wage growth indicators that can help inform our view of slack in the labour market. Our investigation of these different measures leads me to believe that underlying wage growth is running above rates consistent with our 2% inflation target and this looks set to continue. This suggests there is not as much slack as we had expected, and so there is a greater risk of second-round effects emerging from the wage-setting channel.

But we need not only focus explicitly on wage measures in making our assessment of slack. Other labour market and activity measures can provide insights on the degree of slack in the economy, and as I have demonstrated previously, wider economic conditions play a role in determining the extent to which second-round effects are likely to take hold. Taking stock of these conditions suggests that it may be unwise to assume that labour market slack and weak demand will mitigate the risks. Other labour market indicators also suggest there is less labour market slack than we thought and that it may have peaked. As with wages, there is no one perfect metric of employment growth – particularly in recent years given issues with the Labour Force Survey. Bank staff produce an estimate of underlying employment growth drawing on a range of surveys and hard data. Figure 17 shows that employment growth was essentially flat in 2025 and below the rate of population growth—consistent with a degree of slack opening up. However, underlying employment growth has been steadily rising this year. This has recently been corroborated by the KPMG/REC jobs report, according to which measures of employment and labour demand have risen above pre-conflict levels.

Figure 17: Underlying employment growth has started to pick up in the past few months

Indicators of employment growth ^(a)

Footnotes

• Sources: Bank of England Agents, DMP survey, HMRC, KPMG/REC UK Report on Jobs, Lloyds Business Barometer, ONS, S&P Global and Bank calculations.(a) Bank staff’s indicator-based measure of underlying employment growth is constructed using a dynamic factor model, following the approach of (Doz, Giannone, & Reichli, 2011). The model extracts a common component from monthly survey indicators, capturing comovements across series. The common component is scaled to align with LFS employment growth between 2000–19.

Meanwhile, there are other signs that labour market slack may have peaked: the unemployment rate has fallen from 5.2% in the three months to January 2026 to 4.7% in August; redundancies have come down from their peak of 150,000 in the three months to October 2025 and are currently running at just above 100,000, in line with the average observed since 2012; and vacancies show signs of stabilising.

A stable or slightly improving labour market is consistent with more resilience in activity than we and other forecasters had been expecting. In April, HM Treasury’s summary of independent forecasts for the UK economy showed a median GDP forecast of 0.7% in 2026, roughly in line with our middle projection (Scenario B) in April of 0.8% growth this year. By September, HM Treasury’s summary of forecasts showed median growth of 1.3% this year and our central projection in July predicted 1.1% growth in 2026 . There has been a trend towards revising UK GDP forecasts for this year up, particularly as stronger-than-expected data has come in.

Surveys of firms and consumers tell a similar story. September's flash PMI ticked down modestly but remained in positive territory, reflecting an expansion of output. Purchasing managers are reasonably positive about the future as well, with the future output index at 67.4, just a little below the long-run average of 68.2.

The economy has proven surprisingly resilient even in the face of tighter financial conditions. Since the outbreak of the Iran conflict, financial conditions have tightened largely as short-term interest rates and mortgage rates have increased. But while the price of credit has risen, the impact on credit volumes has been muted. Although mortgages for home purchases have fallen since the conflict began, overall gross mortgage lending was £23.6bn in August—slightly below pre-conflict levels (an average of £24bn in the second half of 2025), but above the average level since 2024 (£21.9bn). Monthly consumer credit (credit cards, car finance and other non-mortgage credit) hit a high of £2.5bn in August, and over the past 6 months monthly net finance raised by corporates has remained above the average level since 2024. It is not obvious to me from this data that our monetary policy stance is meaningfully restrictive.

Meanwhile, household and firm inflation expectations remain elevated. As I have previously argued, attentiveness to inflation has risen after over five years of inflation remaining above target. Inflation is also currently already within the threshold at which outturns impact expectations. Greater inflation attentiveness is reflected in an increase in the frequency of UK Google searches for a variety of inflation-related terms in recent years (Figure 18).

Figure 18: UK google searches for inflation-related terms

Frequency^(a)

Footnotes

• Sources: Google and Bank calculations.(a) The series are three-month rolling averages. The frequency is a relative measure for each individual search term, over 2005 – September 2026, where 100 indicates the highest frequency over that time period. Latest data points are for September 2026 and data are not seasonally adjusted.

According to the DMP, firms’ expected year-ahead own-price growth remains elevated at 3.7%, compared with 3.4% in February before the outbreak of the Iran conflict. Similarly, households’ inflation expectations jumped following the Iran conflict, abated but remained elevated subsequently and picked up sharply in September to 4.5% for short-term expectations and 4.3% for longer-term expectations (Figure 19). The risk of elevated inflation expectations translating into second-round effects is greater if there is less slack in the economy.

Figure 19: Inflation expectations have risen since the conflict began

Household inflation expectations^(a)

Footnotes

• Source: Citigroup, YouGov and Bank calculations.(a) The data are from the Citi/YouGov survey and are based on responses to the questions: ‘How do you expect consumer prices of goods and services will develop in the next 12 months?’, and ‘And what do you think will happen to the prices of goods and services, on average, over the longer term – say five to ten years?’. The dashed lines represent the series averages over 2010–19. The latest data points are for September 2026.

Conclusion

Our monetary policy strategy since the outbreak of the Iran conflict has relied to some degree on labour market slack mitigating the risks of second-round effects taking hold. But not all of our wage metrics support this argument. A closer look at the two main metrics we use to understand the recent evolution of wage growth—private sector regular AWE and private sector RTI—suggests wage disinflation has stalled at levels above those consistent with inflation at 2%.

If getting a better handle on the evolution of wage growth to date is tricky, devising a clear view of the future is trickier. The forward-looking indicators we consider—the DMP, pay settlements and the Indeed Wage Tracker–all point to wage growth that is likely to plateau closer to 3.5% than 3% over the coming year. The risks to this are skewed to the upside, in my view. Not only have we tended to underpredict wage growth since the pandemic, but realised pay settlements have been above those predicted in the Agents’ pay survey the past two years. Furthermore, the majority of 2027 pay settlements will be agreed at a time when inflation is above 4%.

Our broad range of wage indicators suggests there is less slack in the labour market than we thought and it may have peaked. This is reflected in broader indicators for the labour market and activity. All together, this suggests the risk of second-round effects taking hold is higher. This is even more so the case as energy prices have remained high and volatile for longer. In my view, this makes a risk management approach to our monetary policy strategy even more appropriate.

When it comes to managing risks around second-round effects, we face two main scenarios. In the first one, we proactively lean against second-round effects by hiking interest rates and later discover this was unnecessary and adjust our policy stance. In the second one, we are too sanguine about second-round effects and set policy accordingly before course correcting. I believe the evidence laid out in this speech shows the risk of second-round effects is greater than we thought. Furthermore, staff simulations suggest setting policy as if you were in a stronger second-round effects scenario and then adjusting policy upon learning you are in a minimal second-round effects scenario leads to better outcomes than doing the opposite. I therefore continue to believe we should proactively lean against the risk of second-round effects.

Acknowledgements

The views expressed in these remarks are not necessarily those of the Bank of England or the Monetary Policy Committee. Opinions, errors and omissions are my own.

I would particularly like to thank Emma Hatwell, Stephen Nelson-Clarke and Teresa Wukovits-Votzi for their help in the preparation of this speech.

The text has also benefitted from helpful comments, data and analysis from Andrew Bailey, Sarah Breeden, Huw Pill, Josh Martin, Krishan Shah, Sumer Singh, Ivan Yotzov, Maria Balgova, Ken Clark, Hassana Babangida, Harry Li, Harry Rigg, James Kensett, Zaar Khan and Vitor Dotta.

Bibliography

Abiry, R., Hurley, J., Labonne, P., Latto, D., Li, H., Moreira, A., Oyegoke, J., & Singh, S. (2026). Learning from forecast errors: the Bank's enhanced approach to forecast evaluation. Macro Technical Paper Number 6.

Adrjan, P., & Lydon, R. (2023). What Do Wages in Online Job Postings Tell Us about Wage Growth?

Bank of England. (2026). Forecast Evaluation Report – January 2026.

Bank of England. (2026). Monetary Policy Report - April 2026.

Bank of England. (2026). Monetary Policy Report - July 2026.

Box, G.E.P. (1976). Science and statistics. Journal of the American Statistical Association 71(356), 791-799.

Doz, C., Giannone, D., & Reichli, L. (2011). A two-step estimator for large approximate dynamic factor models based on Kalman filtering. Journal of Econometrics 164 (1), 188 - 205.

Greene, M. (2024). Two puzzles: recent UK labour market dynamics. Speech at Make UK.

Greene, M. (2026). Here we go again? Assessing the inflation risks of the recent energy shock.

HM Treasury. (2026). Forecasts for the UK economy. Retrieved from https://www.gov.uk/government/collections/data-forecasts

Indeed Hiring Lab. (2026). Indeed Wage Tracker. Retrieved from https://github.com/hiring-lab/indeed-wage-tracker.

Jäger, S., Roth, C., Roussille, N., & Schoefer, B. (2024). Worker Beliefs About Outside Options. Quarterly Journal of Economics 139 (3).

Office for National Statistics. (2026). Component approach to measuring labour productivity, UK: 1997 to 2024.

Pill, H. (2025). The courage not to act. Remarks on the monetary policy outlook at a briefing hosted by Barclays.

Sokolova, A., & Sorensen, T. (2021). Monopsony in labour markets: a meta-analysis. ILR Review 74 (1), 27-55.

Stelmach, M., Kensett , J., & Schnattinger, P. (2025). What can 40 years of data on vacancy advertising costs tell us about labour market equilibrium? Bank Underground.

Villar, D., Moyon, L., & Ahn, H. (2025). What do various wage measures tell us about underlying wage growth? FEDS Notes.