Two new indicators for pay

A comprehensive review of earnings statistics has resulted in a number of developments to the measures of earnings in the UK. We report on the introduction of two new earnings indicators.

Key points

  • Following a review of earnings statistics, the Office for National Statistics has introduced two new short-term earnings indicators.
  • Average weekly earnings supplies a monthly indicator of the level of earnings in pounds and pence.
  • The index of labour costs per hour includes non-wage costs as well as salaries.

A programme of work to improve the availability of statistics on all aspects of earnings has resulted in the introduction of two new experimental earnings indicators. As reported in the August issue of Labour Market Trends, this has led to "a more robust series of earnings data"1.

In 2002, National Statistics published the Review of distribution of earnings2, which aimed to "review National Statistics earnings statistics in the context of user need, methodology, best practice and burden on data suppliers".

A key focus for the report was the New Earnings Survey (NES), which was the main source of information on the distribution of earnings. As a result of the review, the NES was replaced by the Annual survey of hours and earnings (ASHE) from 2004. The new survey boasts a number of improvements on the NES, including more reliable earnings estimates, as a result of extended coverage and methodological changes to reduce bias, and better estimates at the lower end of the earnings distribution. The ASHE also covers Northern Ireland for the first time.

Long- and short-term indicators

The report examined user needs in relation to two categories of earnings statistics produced by National Statistics - structural statistics and short-term indicators.

Structural statistics are "more detailed" and are typically used to analyse trends over long periods. The main source of this information is now the ASHE.

Short-term indicators are used "mainly for macroeconomic analysis and policy, where timely, monthly observations are more important". As a result of the higher frequency than the structural statistics, these indicators are less detailed. Until now, the main source of information in this area was the average earnings index.

The report noted that "one of the top priorities for earnings statistics was the provision of a monthly estimator of inflationary pressures emanating from the labour market." It also comments that "there are a number of complementary indicators that might be developed to augment the short-term data currently available [the average earnings index], and these should be explored further." Two new sets of statistics have been developed by National Statistics to enhance the provision of data in this area:

  • average weekly earnings; and
  • the index of labour costs per hour.

These two new indicators are being published on an experimental basis, with further work planned to develop them into national statistics. Here, we review each of the new measures in turn, including how the surveys are compiled and what data is published.

Average weekly earnings

The average weekly earnings indicator supplies a monthly level of earnings in pounds and pence, previously only available on an annual basis from the ASHE. It also differs from the average earnings index, which only publishes the rate of increase in earnings, rather than the actual level of earnings. It is the first short-term indicator of this kind. The data source for average weekly earnings is the Monthly wages and salaries survey (see box).

A key development with the new statistic is that it will be published as total pay, and regular pay, which excludes bonuses and arrears payments. The series has so far been produced back to January 2000 on a consistent basis. To date, bonuses and arrears pay has been found to make up around 4% of total pay, although during December to March each year, when annual bonuses are typically paid, this rises to more than 10%. Average weekly earnings are published monthly for the whole economy, public sector, private sector, manufacturing, production, services and private sector services. The series is also available at a more detailed industry level.

The new indicator can also estimate the effect on pay growth caused by movements of employees between industries. This is because the industry weights are updated each month (rather than on an annual basis, as with the average earnings index). Since November 2002, changes in employment between industries have had a negative impact on the series, as a result of the shift in employment to lower-paid industries such as retail and education.

Further work is planned on the series, including producing a seasonally adjusted version to give a better picture of the underlying trend of average weekly earnings.

Index of labour costs per hour

The second of the new experimental earnings series is the index of labour costs per hour. This differs from other indicators primarily due to its inclusion of non-wage costs (see table 1). The other fundamental difference is that it measures earnings per hour, while the average earnings index and average weekly earnings collate earnings per job.

The index is based on the following components of costs of labour:

  • wages and salaries, which make up approximately 83% of total labour costs;
  • sickness, paternity and maternity costs;
  • benefits in kind;
  • national insurance contributions; and
  • pension contributions.

The index will produce, on a quarterly basis, four measures, for the whole economy and broken down by public and private sectors, manufacturing, production and services:

  • average total labour costs per hour worked;
  • average wages and salaries per hour worked;
  • average other labour costs, primarily national insurance contributions and occupational pensions, as well as sickness and maternity pay, per hour worked; and
  • average total labour costs, excluding bonuses and arrears, per hour worked.

To date, information is available from the first quarter of 2000 to the first quarter of 2005, although National Statistics is working on producing the index back to 1996.

Growth rates compared

Table 2 shows that earnings growth figures produced by the average earnings index, average weekly earnings and the index of labour costs per hour follow the same general trend, but that differences are apparent. The main explanation for this lies with the number of hours worked. When employees are on holiday, fewer hours are worked in the pay period but pay rates remain unchanged so the labour costs per hour are increased. This effect is particularly apparent in the first quarter of each year, which National Statistics notes could be due to "the increasing prevalence of winter holidays such as skiing, and extended Easter breaks".

Elsewhere, the index of labour costs per hour, which includes non-wage costs, was affected by the new national insurance contributions rate introduced in April 2003. Therefore, when the data for the four quarters from quarter two 2003 are compared with the same quarters in the previous year, there is a significant difference.

1Labour Market Trends, August 2005, available at www.statistics.gov.uk.

2Review of distribution of earnings statistics, National Statistics, available at www.statistics.gov.uk.

Monthly wages and salaries survey

The data source for both the two new experimental indicators, and the average earnings index, is the Monthly wages and salary survey. This is a survey of 8,500 companies with 20 or more employees. Companies that are eligible for sampling are allocated one of four size bands: 20-99 employees; 100-499 employees; 500-999 employees; and 1,000 or more employees. These companies are then used to weight the sample for each size band, and to produce estimates for companies that were not sampled. This is then applied to the pay and employment data collected to calculate estimates for each size band.

Table 1: Characteristics and purposes of the earnings indices

Average earnings index Average weekly earnings Index of labour costs per hour
What it measures Monthly change in average earnings. Average weekly earnings. Average hourly labour costs.
Denominator Employment. Employment. Hours worked.
Non-wage costs None. None. Includes employer NI and pension contributions, sick leave, paternity and maternity payments and benefits in kind.
Source of data Monthly wages and salaries survey. Monthly wages and salaries survey. Monthly wages and salaries survey.
Frequency Monthly. Monthly. Quarterly.
Index reference period Updated every five years. Not an index - no reference period. Updated each year.
Weighting Each company represents a number of similar companies, based on employment. This number is updated annually. Each company represents a number of similar companies, based on employment. This number is updated monthly. Each company represents a number of similar companies, based on employment. This number is updated monthly.
Seasonally adjusted Yes. In progress, expected autumn 2005. Yes.
Sample size (number used) About 8,500 (7,500) companies. About 8,500 (8,000) companies. About 8,500 (8,000) companies.
Periods available January 1990-present. January 2000-present. January 1996-present (from January 2000 on a consistent basis).
Delay between response period and publication 6-7 weeks. 7-8 weeks. 7-8 weeks.

Source: Labour Market Trends.

Table 2: Annual growth rates for earnings indicators, Great Britain, not seasonally adjusted

Average earnings index Average weekly earnings Index of labour costs per hour
2003
Q1 3.6% 3.1% 3.9%
Q2 3.0% 2.6% 3.4%
Q3 3.8% 3.3% 5.1%
Q4 3.1% 3.7% 4.9%
2004
Q1 5.3% 5.0% 6.8%
Q2 4.4% 4.2% 6.1%
Q3 3.8% 3.9% 7.5%
Q4 4.3% 3.9% 4.1%
2005
Q1 4.8% 5.0% 5.1%

Source: Labour Market Trends.