Thursday, July 11, 2024
HomeBankUsing sectoral data to estimate the trend in aggregate wage growth –...

Using sectoral data to estimate the trend in aggregate wage growth – Bank Underground


Tomas Key

Nominal wage growth has increased markedly in the UK in recent years, reaching levels that haven’t been seen for more than 20 years. Although growth has moderated a little in recent months, it remains significantly above its pre-pandemic level. An assessment of whether this strong rate of wage growth will persist is a key input to the monetary policy decision, given the important link between the cost of labour and firms’ pricing decisions. In this post, I will outline a new measure of the trend – or underlying – rate of wage growth which is estimated using data from many different sectors of the economy and which can help with this assessment.

The recent elevated levels of price and wage inflation have spurred a renewed interest in estimates of the trend, or underlying, growth rates of these series. While there are now a large number of approaches to estimating trend price inflation – such as by excluding volatile components or by using statistical models – there are fewer examples of methods to estimate underlying wage growth. In the UK, the most common approach is to exclude a volatile component of pay, bonuses, from the headline ONS average weekly earnings (AWE) series and to smooth the data. In particular, most focus is placed on the annual growth rate of regular pay, smoothed using a three-month moving average. This approach has the advantage of reducing the volatility of the resulting series when compared to measures based on total pay or which use non-smoothed data or which use higher-frequency growth rates. A disadvantage of this approach is that it can be slow to register turning points and can be persistently affected by one-off changes to the level of the series, known as base effects.

Estimating underlying wage growth

My approach to estimating the trend in aggregate wage growth exploits disaggregate data on sectoral pay. In particular, I apply a multi-sector statistical model developed by Stock and Watson (2016) to quarterly AWE data for 24 industrial sectors. The model decomposes sectoral AWE growth into a trend component, which captures persistent variation in wage growth, and a transitory component. In order to account for the comovement of wage growth across sectors, the trend and transitory components are further decomposed into components that are common to all sectors and components that are sector-specific. The resulting sectoral trends are then weighted using employment shares to produce an estimate of the trend in aggregate AWE growth.

The influence that new data from each sector has on the estimate of the trend in aggregate AWE growth depends on two factors. First, the volatility of the growth rate in the sector. Less importance is attached to sectors in which the average wage fluctuates a lot from quarter to quarter. Second, the employment share of the sector. Sectors are more influential if they account for a larger share of employment. The balance of these two factors determines the overall influence of the incoming data from each sector on the aggregate trend.

To account for changes to the structure of the economy, the model incorporates time variation in parameters such as the volatility of each component. It also allows for large one-off shocks, or outliers. These features are likely to be especially important when estimating the model using data from recent years. The volatility of wage growth has increased materially following the pandemic. That likely reflects the impact of the introduction and withdrawal of the furlough scheme in 2020 and 2021, as well as the impact of the very tight labour market in the past couple of years. It is therefore important to allow for increased volatility in both the persistent and transitory components of wage growth, as well as large one-off shocks in the periods in which wage growth was most heavily affected by the furlough scheme.

The estimated trend in aggregate AWE growth produced by this framework is shown in Chart 1, alongside annual whole economy total AWE growth for comparison. There are a few notable differences between the estimated trend and the aggregate data. First, it is less volatile, particularly in the periods following the financial crisis and the pandemic. This is facilitated by the inclusion of time-varying volatility and outlier adjustment in the model. Second, it sometimes leads the aggregate AWE series, particularly during the turning points associated with the financial crisis and the subsequent recovery. That is due to the model being estimated using annualised quarterly growth rates as opposed to the annual growth rate of the aggregate series.

Chart 1: Trend wage growth

Sources: ONS and author’s calculations.

Notes: Red line is the estimated trend in aggregate AWE growth; shaded red area is the 68% probability interval that captures the uncertainty associated with the estimate; blue dashed line shows annual whole economy total AWE growth (quarter on same quarter a year ago). Latest observation is 2024 Q1.

Finally, the estimated trend tracked below the aggregate data in 2023. This suggests that some of the acceleration of AWE growth during the middle of 2023 reflected data volatility and helps to reconcile the difference between the AWE data and other sources of pay information, which were generally weaker in 2023. The estimated trend currently lies a little above the headline annual growth rate series. There is considerable uncertainty about the exact position, however, as illustrated by the shaded area in Chart 1. As is the case with many similar approaches to estimating trends, the latest reading from this model is particularly prone to revision as new data is received. We can be more confident that the trend currently lies significantly above its pre-pandemic level. This means that further moderation in wage growth will likely be required in order for price inflation to return sustainably to target, unless the rate of productivity growth is materially higher.

How widespread has the recent increase in wage growth been?

Chart 2 displays a decomposition of the trend in aggregate AWE growth into the common and sector-specific components. This reveals that most of the fluctuations in trend wage growth that we have seen in recent decades have been due to changes to the trend that is common across sectors. Conversely, changes to sector-specific trends have contributed only a small amount, first to the reduction in trend growth following the financial crisis, and then to the increase in trend growth during the subsequent recovery and in recent years. This prominence of the common trend has also been found in studies of US wage growth, and might help to explain the finding that alternative weightings of the sectoral AWE data make little difference.

Chart 2: Aggregate trend and contributions of common and sector-specific components

Sources: ONS and author’s calculations.

Notes: All series are expressed as deviations from their full-sample mean. Common (sector-specific) component is the weighted average of the persistent common (sector-specific) components for each sector. Shaded areas are 68% probability intervals that capture the uncertainty associated with the estimates. Latest observation is 2024 Q1.

The trend growth rates for selected sector groupings are shown in Chart 3. This visually corroborates the finding that there is strong comovement in the trend rate of wage growth across sectors. However, it also reveals that the minimal contribution of the sector-specific component to fluctuations in the aggregate trend masks some offsetting movements in relative wage growth across sectors. For example, wage growth in low-paying business and other services sectors (LNRS) shifted from the bottom to the top of the pack between the mid-2000s and the mid-2010s, while wage growth in the non-market services sectors (OPQ) moved in the opposite direction.

Chart 3: Trend AWE growth for selected industrial sector groupings

Sources: ONS and author’s calculations.

Notes: Grouped sector labels are combined one-digit SIC codes, as outlined below. 24 industrial sectors are used in the estimation of the model, not these groupings. Latest observation is 2024 Q1.

Definitions:
ABDE: Primary sector and utilities (Agriculture, forestry and fishing (A); Mining and quarrying (B); Electricity, gas and water supply (D and E)).
C: Manufacturing.
F: Construction.
GHI: Trade, accommodation and transport (Wholesale and retail trade; repair of motor vehicles and motorcycles (G); Transport and storage (H); Accommodation and food service activities (I)).
JKM: High-paying business services (Information and communication (J); Financial and insurance activities (K); Professional, scientific and technical activities (M)).
LNRS: Low-paying business and other services (Real estate activities (L); Administrative and support service activities (N); Arts, entertainment and recreation (R); Other service activities (S)).
OPQ: Non-market services (Public administration (O); Education (P); Health and social work (Q)).

Trend wage growth has accelerated in all sectors in recent years, but to varying degrees. The contribution that different sectors have made to the increase in the aggregate trend is a combination of the estimated trend growth rate in the sector and its employment share. The production and construction sectors (ABDE, C, F) account for only a small amount of the increase in trend growth during this period. That is because these sectors have seen both the smallest increases in trend growth and account for a relatively small share of employment. Instead, the bulk of the increase in the aggregate trend has been due to higher trend growth in the services sectors, with the largest contribution from the trade, accommodation and transport sectors (GHI).

Conclusion

The likely persistence of domestically generated inflation is currently one of the key considerations for the appropriate setting of monetary policy. This post has outlined a new measure of one aspect of that – the trend in aggregate wage growth. It uses disaggregated data on sectoral pay to produce an estimate of the aggregate trend and to unveil the sources of fluctuations in trend growth. The estimated trend currently lies a little above the headline rate of wage growth – although there is considerable uncertainty about the exact position, which may well be revised as we receive more data – and significantly above its pre-pandemic level. Updated estimates of the persistent component of wage growth are therefore likely to continue to be of interest in the coming quarters.


Tomas Key works in the Bank’s International Surveillance Division.

If you want to get in touch, please email us at bankunderground@bankofengland.co.uk or leave a comment below.

Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.

RELATED ARTICLES

LEAVE A REPLY

Please enter your comment!
Please enter your name here

- Advertisment -
Google search engine

Most Popular

Recent Comments