Josh Martin
The Monetary Policy Committee has recently looked at wage growth as an important indicator of inflation persistence. One way that wages matter for price inflation is as a cost for businesses, who may raise their prices in response to higher wages. For this channel, the wage measure needs to reflect the coverage and composition of the Consumer Prices Index (CPI). However, most wage measures do not. This blog explores a wage growth measure which is re-weighted to better match the CPI.
What’s the link between wages and inflation?
There are at least two reasons to care about wages for inflation. First, wages are a source of income, which earners can then spend. So higher wages increase demand, putting upward pressure on prices. Second, wages are a cost to businesses. Higher wages increase business costs, who might raise their prices to maintain their profit margins.
In the first story, all labour income in the economy is relevant for inflation, since all workers earn and all workers can spend. A measure that reflects total labour income, including bonuses for instance, would be appropriate.
In the second story, only wages that produce items in the CPI basket matter for inflation. Higher wages in a firm which only produces exports are not relevant for CPI prices, since exports are not in the CPI. By contrast, wages in a firm which produces something for household consumption in the UK, like restaurant meals, are very relevant for CPI.
How to match wage data to the CPI
Most wage data, including the Average Weekly Earnings (AWE) published by the Office for National Statistics (ONS), is weighted by employment. That means it gives more importance (weight) to industries with more employees. This allows the statistics to measure the average (mean) wage growth of all employees in the economy, and within each industry.
To construct a wage measure that best reflects the composition of the CPI basket, we need to adjust the weights. We want to give more weight to industries which produce consumption products, and less to industries that produce things not in the CPI basket, like exports, government output, and investment goods. Since we are thinking about wages as a cost, we also want to give more weight to industries that are more labour-intensive, since wages will be a more important cost for those industries.
What about industries that produce intermediate goods and services, like raw materials or business services? Firms that make consumer products buy those things, so the wage costs might get passed along the supply chain and be relevant for CPI too. For instance, if an accountancy firm raises wages, and a restaurant buys accountancy services, then the higher accountancy wages might lead to more expensive restaurant meals.
That’s possible, but requires several steps – the accountancy raises wages, they must also raise their prices, the restaurant then must also raise its prices because of the higher accountancy costs. In reality, either accountancy firm or restaurant might not raise prices and instead accept a temporarily lower profit margin given higher costs. There are also likely long lags between accountancy wages and restaurant prices. So, given uncertainty and time lags, I won’t factor in the wages of industries that produce intermediate inputs, only those producing products directly sold to consumers.
To figure out the right weights for our CPI-weighted wage measure, I use data from the supply and use tables (part of the National Accounts) to spread the CPI weights to industries. First, I match the CPI weights to the detailed ‘product’ categories in the supply and use tables, spreading them out where necessary. I have to account for the difference in coverage of the CPI and household consumption in the National Accounts – for instance, the CPI excludes gambling, but the National Accounts includes it.
Second, I split apart the CPI weight for goods into that which reflects the good itself, and that which reflects the retail and wholesale services required to get the good to consumers. For instance, when you buy a banana in the shop, you are paying partly for the banana itself, partly for the wholesaler who got it to the UK, and partly for the retailer who put it on the shelf. Consumers don’t buy retail services directly, only indirectly through other goods, so retail doesn’t have an obvious weight in the CPI – it needs to be separated from the weight of goods.
Third, I account for which CPI products are imported and which are produced domestically. Consumers may buy lots of bananas, but if most of those are imported, then the wages in the domestic banana industry aren’t so important after all. Finally, I account for the share of wages in total costs of the industry. In industries that are more labour intensive, wages will be a more important cost, and so more relevant for the price.
Putting all of that together and the summing up by industry gives us a new set of industry weights for our wage measure. This should, in theory, better reflect the importance of each industry’s wages in the CPI.
Does the re-weighting make much difference?
Using these new weights to aggregate the industry AWE regular pay growth rates published by ONS gives a CPI-weighted wage measure. Chart 1 shows the annual growth in this measure between 2001 and 2023. The chart also shows the annual growth in AWE whole economy and private sector regular pay for comparison.
Chart 1: Measures of annual regular pay growth, January 2001 to December 2023
Source: ONS and author’s calculations.
Notes: Rolling three-month averages of annual growth. Latest period October–December 2023.
Over the long run there is little difference between the CPI-weighted AWE and the headline measures published by ONS. That suggests that the measures we usually look at do a good job of capturing the key information for understanding wages as costs for businesses. The new measure is just a re-weighted version of the same data underlying the other measures, so it is perhaps unsurprising that they are similar.
In the past year or so, there is a little more difference between the measures, as shown in Chart 2, which is the same data as in Chart 1 but zoomed in on the period since January 2019. The CPI-weighted AWE grew slower than the headline AWE measures during most of 2023. But in the past few months, while the headline measures have slowed sharply, the CPI-weighted measure has been flatter. That’s because the industries driving the fall in the headline measures include professional services and construction. These industries don’t produce many consumer products, so get much lower weights in the CPI-weighted AWE measure.
Chart 2: Measures of annual regular pay growth, January 2019 to December 2023
Source: ONS and author’s calculations.
Notes: Rolling three-month averages of annual growth. Latest period October–December 2023.
Chart 3 shows the difference between the industry weights in 2023 in the AWE private sector measure and the CPI-weighted AWE measure described in this blog. Green bars show industries with more weight in the CPI-weighted measure, such as wholesale and accommodation and food services. Industries that get less weight (shown in red) include professional services, construction, and admin services – all business-facing industries. Some of these industries would likely get a greater weight if also factoring in industries producing intermediate inputs for use in making consumer products.
Chart 3: Difference in weight between CPI-weighted wages and AWE private sector, 2023
Source: ONS and author’s calculations.
Notes: Industries are defined by SIC 2007, consistent with AWE breakdowns. Positive (green) bars show more weight in the CPI-weighted measure than AWE private sector, and negative (red) bars show less weight. Units are percentage points; for instance, accommodation and food services is weighted 12.7 percentage points higher (22.1% versus 9.4%).
Other people have also thought about this issue. Former MPC-member Silvana Tenreyro, in a speech in 2020, constructed a CPI-weighted measure of unit labour costs (labour costs per unit of output). This used National Accounts data on labour costs and productivity, so is slightly different to the measure in this blog, but done for the same reasons. She found that CPI-weighted unit labour costs were growing slower than whole economy unit labour costs between 2017 and 2019, mostly due to differences in productivity growth.
In a recent series of blogs, the White House Council of Economic Advisors constructed a wage measure to match the composition of core non-housing services inflation. They have far more detailed industry wage data available than we do in the UK. They suggest that this measure is a slightly better predictor of future core non-housing services inflation than other private sector wage measures.
Summing up
Overall, it seems like re-weighting wage data to match the CPI is a good idea in theory, but doesn’t make very much difference in practice, at least not so far. That might be because the available industry breakdown of wage growth from the AWE is quite limited, so there isn’t very much scope to pick out the key industries. But the re-weighting might be relevant in future. For instance, the increase in the National Minimum Wage in April 2024 will affect some industries more than others, and as we know, not all industries are equally important for CPI.
Correction (15 February 2024): This post has been corrected due to a calculation error in the weighting of the retail and wholesale industries. This principally affects Chart 3, though the other charts have also been updated. The author apologises for the error and any inconvenience caused.
Josh Martin works in the Bank’s External MPC Unit.
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.
