Appendix 8:
Half Male Median Earnings Calculations

1.

A number of organisations give half male median earnings as the ‘target’ figure for the National Minimum Wage, and the question of the correct figure for half male median earnings has surfaced a number of times since we started our work in 1997; indeed, it predated the setting up of the Commission. In the first volume of this report we gave our estimate of half male median earnings as £4.13 per hour. Since then others have quoted a higher figure. We have therefore decided to address this issue in an attempt to put the record straight.

2.

In its written evidence the Low Pay Unit pointed out that it ‘has consistently argued for a minimum wage to be set at half male median earnings’. It believes the case for this ‘is based on clear principles of equity and fairness: a minimum wage linked to full-time, male earnings would immediately reduce income inequality, prevent low paid workers falling behind as average earnings rise, and above all ensure that people in low paid work earned enough to participate in the society around them’. And in a press release giving its reaction to our recommendations in the first volume of this report, the Low Pay Unit (2001) argued ‘that the minimum wage should be set at half male median earnings or £5.11 per hour’. Similarly, UNISON, in its evidence, argued that ‘the Low Pay Commission should recommend, and the Government should implement, a £5.00 per hour minimum wage as soon as possible, as a step towards a rate of half male median earnings, currently £5.10 an hour’.

3.

Both the Low Pay Unit and UNISON base their figure on the following formula: full-time median male earnings (including overtime) divided by total basic hours worked by men and women (excluding overtime). As Table A8.1 shows, using this formula and the ungrossed New Earnings Survey (NES) data produces a figure of £5.11 per hour at Spring 2000.

Table A8.1
Half Male Median Earnings, Low Pay Unit/UNISON Formula
	£ per hour
	1998	1999	2000
1. Ungrossed NES data	4.78	4.94	5.11
2. Grossed NES data	4.70	4.79	5.01
3. LFS data a	3.67b	3.73	3.99
4. ONS central estimate	4.18	4.26	4.50
Notes: a. The new LFS variable asks for the basic hourly rate, which should exclude overtime. b. This figure is calculated using a different method of imputation (regression) than estimates for later years (donor method); therefore, 1998 is not directly comparable with estimates for later years. ONS advise that the differences between the two methodologies are larger further up the earnings distribution, and since the methodology was developed to measure the extent of low pay, data are more reliable at the lower end of the distribution.

4.

One problem with this methodology is the data set used. The figure is based on ungrossed NES data. We explained in Volume 1 of this report that the NES under-samples individuals earning below PAYE levels, leading to an understatement in the (ungrossed) NES of the extent of low pay. The Office for National Statistics (ONS) has devised a methodology to adjust for this understatement and has produced a data set, the ‘grossed NES’, which better reflects the extent of low pay in the UK. It does this by grossing up the NES sample to all employee jobs in the UK. This mitigates, to a large extent, the under-representation of the number of young people and people working in small firms. Using these data, the estimate of half male median earnings, as calculated using the formula favoured by the Low Pay Unit and UNISON, is £5.01 per hour, as shown in Table A8.1.

5.

We explained in Volume 1 of this report that grossing cannot correct for all the under-representation of the low paid in the NES. We therefore also use the Labour Force Survey (LFS), which now has a revised methodology for measuring hourly earnings, in our estimates of low pay. Using the LFS, with the same formula, produces a figure of £3.99 for half male median earnings in 2000 (see Table A8.1). But, as we also explained in Volume 1, our and ONS’s preferred approach to assessing low pay is to combine the use of the grossed NES and LFS data into a central estimate. Using the formula favoured by the Low Pay Unit and UNISON with the ONS central estimate data gives a figure for half male median earnings of £4.50 per hour for Spring 2000 (see Table A8.1).

6.

Another, more fundamental, problem with the half male median earnings approach advocated by UNISON, the Low Pay Unit and Burkitt et al. (1999) is that the definition it employs does not compare like with like. By using male full-time earnings including overtime the definition overstates earnings; and by using basic hours for men and women (excluding overtime) it understates hours. Bain (1998) has referred to this definition as an ‘apples and pears’ formula.

7.

Burkitt et al. (1999) point out that male earnings have been used to avoid entrenching gender wage inequality. Similarly, overtime hours are used to avoid exacerbating inequality between manual and non-manual workers since these are an important contribution to pay for the former but not for the latter. They argue that the formula ‘is trying to establish a sensible basic wage for all workers’ and that workers should not have to work excessive hours to meet their basic subsistence needs. Moreover, they believe that using this formula is the most appropriate approach because low pay, like poverty, is a relative concept, and their approach would ensure that the low paid share in the gain in living standards arising from an increase in average earnings.

8.

Taking the last point first, we argue in Chapter 6 of this volume that the use of any formula — whether it is half male median earnings or any other — is problematic. We believe that a formulaic approach would be excessively rigid and would not take account of the dynamic nature of the labour market, particularly the challenges faced by business and the needs of workers in a changing economic climate. Nor would it consider the ‘wider economic and social conditions’ of a minimum wage, which has always been in our terms of reference.

9.

Turning to their point that the formula should meet workers’ subsistence needs, Burkitt et al. (1999) are shifting the argument from what is the empirical figure for one half male median earnings to what income workers require to meet their subsistence needs. Whatever income is required to meet these needs is a completely different and highly subjective concept to the empirical definition of half male median earnings. Whether the figure used by the Low Pay Unit, UNISON and Burkitt et al. (1999) is a subsistence figure, and whether it is the ‘right’ or ‘wrong’ figure for us to recommend, it is not half male median earnings: the earnings data on which it is based are biased upwards, and the definitions of earnings and hours used to calculate hourly earnings are inconsistent.

10.

Bain (1998) made the point that many advocates of the £5.11 figure accept that it is ‘a "campaigning tool", a slogan to mobilise employees and ensure as high a negotiating figure as possible’. We do not take issue with the Low Pay Unit and other organisations in campaigning for a higher minimum wage or for higher pay and incomes generally. Indeed, we welcome the part that these organisations play in helping the Commission fulfil its remit. But we need to ensure that recommendations we make are based on an objective and consistent analysis of the data. To do otherwise is to confuse and to mislead.

11.

The most consistent and reliable figure for half male median earnings is given by combining the formula — half male full-time median earnings excluding overtime divided by total basic hours worked by men, again excluding overtime — with the best available data set — that derived from the ONS central estimate methodology. Table A8.2 shows that this gives a figure of £4.13 per hour in Spring 2000.

Table A8.2
Half Male Median Earnings, Consistent Definition of Hours and Earnings, ONS Central Estimate
	£ per hour a
	1998	1999	2000
LFS half male median earnings (central estimate methodology)	3.61b	3.62	3.81
NES half male median earnings (grossed data)	4.15	4.32	4.44
Central estimate half male median earningsc	3.88	3.97	4.13
Notes: a. Estimates refer to employees aged 22+ and cover full-time earnings in the main job only. b. The estimate for 1998 is calculated using a different method of imputation (regression) than estimates for later years (donor method), and is therefore not directly comparable with estimates for later years. ONS advise that the differences between the two methodologies are larger further up the earnings distribution, and since the methodology was developed to measure the extent of low pay, data are more reliable at the lower end of the distribution. c. Other elements of pay such as regional allowances and waiting time cannot be separately identified and removed from these calculations to make them consistent with the definition of earnings used in the minimum wage legislation.

12.

This figure is still not entirely comparable to the definition of earnings used in the minimum wage legislation since it includes other elements of pay such as regional allowances and waiting time which cannot be separately identified in the data. Removing these elements would reduce the figure further still.