Resource centre

Earnings top-up evaluation: employers' reactions

A hard copy of this report summary can be obtained by contacting Paul Noakes  [E-Mail: Paul.Noakes@dwp.gsi.gov.uk] or by writing to him at the 'Social Resaerch Division, Department for Work and Pensions Security, 4th Floor, Adelphi, 1-11 John Adam Street, London WC2N 6HT'.

top of page

Research Report No. 132

Part One: Surveys of Employers

By Stephen Lissenburgh

Part Two: Econometric Analysis

By Chris Hasluck and Anne Green

Earnings Top-up (ETU) was an in-work benefit available to low paid workers without children. ETU was piloted from October 1996 to October 1999 in eight areas across Britain. This volume is part of a set of seven final reports from the evaluation of the ETU pilot. The evaluation was conducted by the Policy Studies Institute (PSI), the Centre for Research in Social Policy (CRSP) at Loughborough University and the Institute for Employment Research (IER) at the University of Warwick. Part One explores employer’s experiences of ETU focussing on wage effects and hours worked. Part Two considers the effects of ETU on employers’ behaviour and the recruitment process.

The main findings include:

Surveys of employers

Econometric Analysis

Part One – Surveys of employers

top of page

Research methods

The focus here is on one part of the evaluation: the surveys of employers. The initial sample was interviewed in summer 1996 and again in summer 1997. In 1999, employers who had been interviewed in both 1996 and 1997 were re-interviewed. These employers constitute a ‘panel’ from whom data was obtained at three points in time. The 1999 survey also interviewed a cross-section of employers with the same characteristics as the initial sample interviewed in 1996. The report discusses the panel sample and also compares the 1996 and 1999 cross-section surveys.

top of page

Main characteristics of the employers in 1999

Type of industry and sector

As in 1996, most establishments were in the private sector (82 per cent). A quarter of the employers (27 per cent) were in distribution industries, while 14 per cent were in hotel and catering and 12 per cent were in manufacturing. The majority of establishments (61 per cent) were independent, in that they did not form part of a larger organisation, while a quarter (27 per cent) were branches of a larger organisation and the remaining 12 per cent were the Head Office of a larger group.

The workforce

Employers were asked in detail about employment in three relatively low-paying groups:

Employers in the hotel and catering industries had a predominantly semi-skilled/unskilled workforce (57 per cent). Those in agriculture, mining or construction had the highest proportion of skilled/craft employees (41 per cent), while employers in distribution and public administration each had a high proportion of their workforce made up of clerical/sales employees (41 per cent).

Wages

Whether ETU had any effect on wage offers would depend partly on the way in which pay levels are set and the influences on this. More than half of the employers interviewed in 1999 (55 per cent) said ‘the pay individuals are willing to work for’ influenced their wage-setting behaviour; four out of ten (42 per cent) mentioned ‘the pay offered by other local employers’ and ‘national agreements and national pay trends’; and a similar proportion (38 per cent) said ‘the availability or scarcity of labour’ had some influence on the rates of pay they offered to employees. With the exception of ‘national agreements and national pay trends’, these most commonly mentioned influences on levels of pay offered by employers could allow ETU to have an effect through the workings of the local labour market.

Another factor that could influence the potential for ETU to have a ‘wage effect’ is the extent to which employers have wage policy autonomy. Overall, eight out of ten employers (79 per cent) had local autonomy to determine pay levels.

Recruitment

The ability of ETU to have an effect on recruitment through a ‘substitution effect’, whereby employers may target their recruitment on those eligible for the benefit at the expense of existing non-eligible employees, would depend partly on employers’ autonomy in recruitment procedures. Overall, eight out of ten employers (80 per cent) had local autonomy in their recruitment procedures.

top of page

Changes since 1996 – a descriptive analysis

Workforce and recruitment

Among the job groups where ETU would be expected to have its greatest effect – semi/unskilled and clerical/sales jobs – there was mixed evidence regarding employment effects among the panel sample of employers. Levels of semi/unskilled employment were maintained between 1996 and 1999 in the pilot areas but fell substantially in the Control areas, a difference consistent with an ETU effect. However, clerical/sales employment grew more strongly in Control than in pilot areas, contrary to expectation.

It was also expected that ETU may have an effect on labour turnover. Recruitment would be expected to increase at the lower-end of the wage distribution in response to ETU and some employees may find they are able to stay in jobs longer because of the new benefit, thereby reducing termination of employment. Termination rates in semi/unskilled employment were indeed lower in the ETU pilot than in the Control areas in 1999. Whereas termination rates in this job group were 42 per cent in Control areas, they were 29 per cent in Scheme B areas and 26 per cent in Scheme A areas. Contrary to expectation, however, recruitment rates were also lower for semi/unskilled employment in the pilot areas, running at 35 per cent in both Scheme A and Scheme B areas compared with 47 per cent in the Control areas.

Hours of work

Entitlement to ETU begins at 16 hours per week and there is an extra credit available to those who work for 30 or more hours per week. It might be expected, therefore, that employers in ETU areas would experience an increase in the proportion of their employees working 16 to 29 hours per week or 30 hours or more per week, rather than 1 to 15 hours per week, compared to employers in the Control areas. The only clear evidence to support this contention related to maintenance of full-time working in Scheme A areas, compared to Scheme B and Control areas. In both Scheme B and Control areas the proportion of semi/unskilled employees who worked 30 hours or more per week fell quite sharply from 1996 to 1999, a decline from 63 per cent to 54 per cent in Scheme B areas and from 65 per cent to 52 per cent in Control areas. No such decline took place in Scheme A areas, however, so that by 1999 the proportion of semi/unskilled employees who worked full-time in these areas (61 per cent) was higher than in Scheme B and Control areas.

Pay and the extent of low pay

The introduction of ETU was expected to constrain wage growth among recent low-paid recruits in the pilot relative to the Control areas. Overall, the average starting wage of employers’ most recent low-paid recruit increased from £3.18 per hour in 1996 to £3.47 in 1999. This wage growth was more limited in Scheme A areas than elsewhere, rising from £3.24 per hour to £3.39 (an increase of 5 per cent), compared with wage growth of 11 and 12 per cent respectively in Scheme B and Control areas. As expected, the starting wage paid to employers’ most recent low-paid recruit was heavily influenced by the National Minimum Wage, introduced in April 1999. Four out of 10 employers (38 per cent) who recruited someone at £4 per hour or below in the past year in 1999 did so at the adult minimum wage of £3.60 per hour (Table 3.19). This proportion did not vary by evaluation area, but Scheme A employers were more likely than those in other areas to pay their low-paid recruits less than £3.60 per hour and less likely to pay them more than £3.60 per hour (29 per cent, as against 41 per cent in Control areas). These findings are suggestive of an ETU ‘wage effect’ for low-paid recruits in Scheme A areas.

Social security benefits

There was a high level of general awareness of in-work benefits among the employers interviewed but less than a third (32 per cent) reported actual experience of their lower paid employees being entitled to in-work benefits.

Awareness of ETU increased between 1996 and 1999, especially in Scheme B areas. However, only six per cent of employers in the pilot areas interviewed in 1999 reported any actual experience of ETU. Knowledge of the eligibility criteria for ETU was low even among those who said they had heard of ETU and was no higher than it had been in 1997.

Assessing the impact of ETU – multivariate analyses

Multivariate analyses, which control systematically for differences between employers, were carried out to provide a robust test of ETU’s impact on wages and employment.

Change in hourly wages of semi-skilled/unskilled and clerical/sales employees

ETU had no impact on the change in wages of semi-skilled/unskilled employees between 1996 and 1999, nor on the change in wages of clerical/sales employees between 1996 and 1999. Where employers had become “aware” of ETU by 1999, however, whereas they had not been aware in 1996, this was associated with the payment of lower wage increases to clerical/sales employees. This suggests that where employers show increased awareness of ETU, the benefit is more likely to restrain wage growth.

Change in the proportion of low-paid semi/unskilled and clerical/sales employment

If ETU had exerted a ‘wage effect’ this should be manifested in an increase in the proportion of low-paid employment within the semi/unskilled and clerical/sales job groups in the pilot relative to the Control areas. ETU had no effect on the change between 1996 and 1999 in the proportion of semi/unskilled employees who were low-paid. Increased awareness of ETU, however, was again associated with a change consistent with theoretical expectations – where employers were aware of ETU in 1999 whereas they had not been in 1996, this was associated with an increase in the proportion of low-paid employment with in the semi-skilled/unskilled job group.

ETU also had no effect on the change in the proportion of low-paid clerical/sales employment between 1996 and 1999. Increased awareness was again associated, however, with an increase in the proportion of low-wage employment, although in this case it was increased awareness of in-work benefits in general rather than ETU in particular that was important.

Change in hourly wages of most recent low-paid recruit

The change in hourly wage offers to the most recent low-paid recruit between 1996 and 1999 was subject to an ETU effect. After controlling for other variables influencing wage change, employers located in Scheme B areas reported slower wage growth than was the case in Control areas. This is the type of effect ETU would be expected to have if the benefit was making low-paid work more attractive for eligible employees. There was no such effect in Scheme A areas. Awareness of ETU and in-work benefits in general was also associated with slower wage growth, albeit not strongly. If employers were aware of ETU in 1999 having not been in 1996, hourly wage offers to the most recent low-paid recruit rose more slowly than was otherwise the case. The same was true if employers became aware of in-work benefits in general during the course of the evaluation period.

top of page

Conclusion

Employment and new jobs

Among the job groups where ETU would be expected to have its greatest effect – semi/unskilled and clerical/sales jobs – there was mixed evidence regarding employment effects among the panel sample of employers. Levels of semi/unskilled employment were maintained between 1996 and 1999 in the pilot areas but fell substantially in the Control areas, a difference consistent with an ETU effect. However, clerical/sales employment grew more strongly in Control than in pilot areas, contrary to expectation. There was no evidence that recruitment difficulties had eased following the introduction of ETU.

Employment Policy

Awareness of in-work benefits was high, although less than a third of employers reported any actual experience of these benefits. Awareness of ETU rose between 1996 and 1999, especially in Scheme B areas, though only six per cent of employers in the pilot areas reported any experience of ETU by 1999.

Wages

Of the three most common influences on the levels of pay offered by employers, two of them – the pay individuals are willing to accept and the pay offered by other local employers – gave considerable scope for an ETU ‘wage effect’. No such effect emerged clearly from the descriptive analysis, however, which instead produced a lot of conflicting evidence. For example, while among the panel sample of employers wages grew more slowly for semi/unskilled and clerical/sales workers between 1996 and 1999 in Scheme B areas than in Scheme A or Control areas, the average proportion of each job group earning £4 per hour or below in 1999 did not vary by evaluation area. The one area in which the descriptive evidence was a little more consistent was in relation to recent low-paid recruits. Thus, analysis of the cross-section surveys showed that recent low-paid recruits were paid more in 1999 than equivalent workers in 1996 in all areas, but this wage growth was slower in Scheme A areas than in Scheme B or Control areas and, among the panel sample, the average pay of recent low-paid recruits rose more quickly between 1996 and 1999 in Control areas than in pilot areas.

The multivariate modelling produced limited evidence of an ETU effect on wages. There was no evidence of an ETU effect on change in hourly wages of clerical/sales and semi/unskilled employees between 1996 and 1999, nor on the change in the proportion of employees in these job groups who were low-paid. But in relation to the change in wages of the most recent low-paid recruit a clear ETU ‘wage effect’ did emerge, albeit only in relation to Scheme B areas, where wage growth was slower for recent low-paid recruits than in Control areas.

Part Two – Econometric analysis

top of page

Research methods

The research drew upon data from two sources. These were, first, the time series of vacancies notified to the Employment Service (ES) in the ETU pilot and control areas. Second, the research examined survey data relating to recruitment collected from employers as part of the evaluation of ETU.

A number of different analytical methods were used, including inspection of time series, descriptive statistical analysis and, finally, some multivariate analysis.

top of page

Background to the study

As an example of an in-work benefit (a welfare benefit paid to people who are in paid work), ETU had two main objectives. First, to increase the incomes of low paid workers. Second, to strengthen the incentive to take low paid jobs. By shifting the balance away from out-of-work income towards in-work income, an impact on employment could be expected. However, increased competition for low paid jobs or overt actions by employers could negate such gains and the overall impact of ETU on employment levels remains uncertain and is a central concern for any evaluation of ETU.

top of page

ETU and the recruitment process

This report examined the evidence relating to the impact of the introduction of ETU on employers and employment. The research considered the mechanism by which ETU might bring about employment effects. If ETU made employment in low paid jobs more attractive to some jobseekers than hitherto in the absence of the ETU benefit payment, employers will have experienced an increase in the flow of jobseekers willing to work in low paid jobs. The consequence of such an improvement in labour supply might have been to allow employers in areas where ETU was on offer to fill job vacancies more quickly and, in more extreme cases, to fill job vacancies that would have previously remained unfilled. While the overall number of jobs may not have been changed as the result of ETU, it“ ”is possible that the proportion of jobs that were filled had increased, representing an increase in the effectiveness of the job matching process performed by the labour market.

If ETU had an effect on employment of the form outlined above, this suggests that it is through an examination of vacancies and recruitment that evidence of an ETU effect will be found. Changes in the number of unfilled vacancies and the time taken to fill them before and after the introduction of ETU, and comparisons of ETU pilot and control areas, may provide indicators of the impact of the introduction of ETU. This report has examined the evidence relating to these aspects of ETU. Two types of analysis were undertaken. In“ ”the first, a broad perspective was taken by examining vacancy flows at the level of the ETU pilot and control areas. The second approach considered evidence from individual employers about their experience of the recruitment of workers to low paid jobs.

top of page

ETU and notified vacancies

The analyses of Employment Service (ES) data on notified vacancies revealed no clear evidence of ETU having had an impact on:

in Scheme A and Scheme B areas relative to either the control areas or relative to Great Britain as a whole.

This does not necessarily mean that ETU had no impact on the recruitment process and the structure of vacancies. Rather, it may mean that any impact was small relative to other impacts on the notification of vacancies to ES and other influences on the manner and speed by which vacancies were filled/remained unfilled. No evidence was available on the precise nature of other influences nor were such influences quantified. However, two possible factors other than ETU that might have affected vacancy notifications were:

Vacancies in a local economy may have been generated by employment growth, skills shortages and labour turnover, as well as by policy initiatives such as ETU. Moreover, the relative importance of these factors may vary between local areas, hence the relative impact of each was difficult to assess.

top of page

ETU and recruitment at establishment level

The second element of the research considered evidence relating to employment, recruitment and vacancy durations from a sample of employers in the three types of ETU area. Some evidence of an ETU effect was found but such evidence was, at times, contradictory.

Inspection of evidence relating to employment change provided little clear indication that the introduction of ETU was associated with an increase in employment in ETU pilot areas. The employment share of low paid workers was declining in all areas and decreased at a higher rate in the ETU pilot areas than in the control areas. However, the increasing share of skilled manual employment in Scheme B was in marked contrast to this general picture and might be indicative of some ETU effect (especially as ETU was paid at a higher rate in Scheme B areas). Evidence relating to recruitment, again, provided only weak and contradictory support for the hypothesis that the introduction of ETU had led to an increase in the rate of recruitment to low paid skilled manual and clerical & sales jobs. Where found, this effect was most marked in Scheme B areas.

In order to isolate ETU effects from other influences on recruitment, multivariate analysis was used to model the recruitment process. The model found that whether or not an establishment recruited a low paid worker was significantly influenced by the type of activity undertaken (industry) and size of establishment. While no direct evidence of an ETU effect was found, employers were more likely to have recruited to low paid jobs after ETU was introduced if they were aware of in-work benefits. Such awareness might be an indication of actual experience of ETU by such employers. Restricting the analysis to employers who did not recruit a low paid worker in 1996 but did so in 1997 found similar factors to be associated with a switch in recruitment but also found some weak evidence that employers located in Scheme A areas were more likely to have recruited a low paid worker when not having done so in the previous year (before the introduction of ETU). As before, awareness of in-work benefits was the most strongly related influence on such a switch in recruitment.

The period covering the introduction of ETU was a period of increasing demand for labour and this was reflected in the evidence of increasing vacancy durations across all areas. If ETU facilitated the recruitment of low paid workers, such increases might be expected to be lower in the ETU pilot areas than in the control areas. While this was true of Scheme B areas it was not true of Scheme A areas. Comparison of ETU ‘eligible’ and ‘ineligible’ jobs found that average vacancy durations increased least in ETU eligible jobs. While this was consistent with the presence of an ETU effect, the presence of this pattern in the control areas was not. However, vacancy durations increased less (recruitment was easier) amongst employers who were aware of in-work benefits than amongst employers who were unaware of such benefits. This might be seen as an indication that any ETU effect was concentrated on establishments in which employees were not only eligible for ETU but were also actually claiming it. Unfortunately, the employer data provides no direct evidence of actual take-up of ETU.

top of page

Conclusion

The lack of any evidence of a strong and significant effect on employers’ recruitment and vacancies arising from the introduction of ETU can be interpreted in a number of different ways. In the first place, any such effects were likely to have been of a small scale. Given that ETU was unlikely to increase the number of jobs that employers offer (their demand for labour), any effect would be limited to making the market for low paid jobs more efficient so that a greater proportion of jobs were actually filled at any time. If potential ETU effects were of such a modest scale, then they would be difficult to detect. This would be because, first, any ETU effects could easily be overwhelmed by the effects of other factors inducing change in employment and recruitment (such as changes in local labour market and macroeconomic conditions, other policy developments and so forth). Second, the detection of any small ETU effect required the data to be sufficiently sensitive to allow such effects to be seen and isolated from other factors. As acknowledged in this report, ES notified vacancies are a fairly crude measure of total vacancies in the local labour market while the ETU Employer Survey data, while very detailed, also has its limitations (as do all sample surveys).

The lack of any significant evidence of an ETU effect on employers is consistent with other research undertaken as part of the evaluation of the ETU pilots (see, for instance, Lissenburgh, 2001). However, it is worth considering whether such a lack of detectable impact matters. While one of the objectives of ETU was to increase the incentive of jobseekers to enter low paid jobs, a second objective was to raise the real income of members of low paid households. A common criticism made of interventions in the labour market is that any gains made will be at the expense of other people in the workforce (through substitution effects and the like). However, just as there is little strong evidence of a positive ETU effect on employment, there is no significant evidence of a negative impact either. The recruitment of people who were ineligible for ETU does not appear to have been adversely affected. This being so, the lack of any evidence of a significant labour market impact from the introduction of ETU can be seen as a very positive finding. It suggests that an incentive to enter low paid employment, and an improvement in the real incomes of low paid workers, may be achievable without the offsetting labour market consequences allegedly associated with such policy intervention.

top of page

Publication details

Lissenburgh, S., Hasluck, C. and Green, A. (2001) “Earnings Top-up Evaluation: Employers’ Reactions,” Department of Social Security Research Report No. 132 CDS: Leeds (£29.50)

top of page

Relevant publications

Smith, A., Dorsett, R. and McKnight, A., (2001) “Earnings Top-up Evaluation: Effects on Unemployed People”, Department of Social Security Research Report No. 131, CDS: Leeds.

Marsh, A., (2001) “Earnings Top-up Evaluation: The Synthesis Report”, Department of Social Security Research Report No. 135, CDS: Leeds.

Marsh, A., Stephenson, A., Dorsett, R and Elias, P., (2001) “Earnings Top-up Evaluation: Effects on Low Paid Workers” Department of Social Security Research Report No. 134, CDS: Leeds.

Heaver, C., Roberts, S. Stafford, B. and Vincent, J. (2001) “Earnings Top-up Evaluation: Qualitative Evidence”, Department of Social Security Research Report No. 133, CDS: Leeds.

Vincent J., Heaver, C., Roberts, S. and Stafford, B., (2001) “Earnings Top-up Evaluation: Staff Views”, Department of Social Security In-house Research Report No. 74, London.

Green, A. (2001) “Earnings Top-up Evaluation: Labour Market Conditions”, Department of Social Security Research In-house Report No. 75, London.

Marsh, A., Callender, C., Finlayson, L., Ford, R., Green, A and White, M., (1999) “Low Paid Work in Britain”, Department of Social Security Research Report No. 95) CDS: Leeds.

Finlayson, L., Ford, R., Marsh, A., Smith, A., and White, M., (2000) “The First Effects of Earnings Top-up”, Department of Social Security Research Report No.112) CDS: Leeds.

Vincent, J., Abbott, D., Heaver, C., Maguire, S., Miles, A., Stafford, D., (2000) “Piloting Change”, Department of Social Security Research Report No. 113 CDS: Leeds.

top of page