Thursday, 20 December 2012

An exploratory evaluation of the Next Step Service

A research paper (No 97) by London Economics for the Department for Innovation and Skills (November 2012) Ref: URN 12/279

Key Findings
  • The Next Step service is a nationally branded careers and skills advice service that is available free to adults in England aged 19 and over irrespective of their prior skills, qualifications and employment status. Since April 2012, Next Step has been rebranded as the National Careers Service.
  • Of the 809,000 Next Step customers in the first year of service between August 2010 and July 2011, almost seven in ten were White British (68%); 54% were male; and half were aged between 19 and 34.
    Almost two thirds (63%) of Next Step customers had a qualification at Level 2 or below, reflecting the relatively low level of prior attainment. The majority of Next Step customers (65%) reported that they were unemployed, of which just over a quarter had been out of work for less than 6 months, while just over one in five Next Step customers had been unemployed for more than three years. More than half of Next Step customers (55%) reported that they were in receipt of JSA, Income Support or Employment and Support Allowance.
  • Four-in-five people who accessed Next Step self-referred to the service, with the remainder being referred through Jobcentre Plus. Compared to self-referral, a higher proportion of males are referred to Next Step through Jobcentre Plus (62% compared to 52%). Additionally, proportionately more White British customers and those aged between 25 and 54 were referred through Jobcentre Plus compared to the self-referral route.
  • Support can either take place on the telephone, through a web-based service such as e-mail or face-to-face. Overall, almost three-quarters of first intervention sessions were face-to-face, 27% were via the telephone while approximately 1% were web-based. Almost all referrals from Jobcentre Plus led to face-to-face support sessions (98%) compared to only two thirds of self-referrals.
  • To undertake a robust analysis, we compared the outcomes of individuals in receipt of Next Step service to a sample of individuals that were not in receipt of the Next Step service but had similar personal and socioeconomic characteristics as the treatment group. We recognise that identifying a control group is particularly challenging because the reason people choose Next Step will possibly be to do with motivation to change one’s circumstances which is particularly difficult to assess from existing data. Nonetheless, we thought that using Propensity Score Matching would be a useful first step. A number of Propensity Score Matching models were constructed using different variables to match the treatment and control groups, as well as using different samples of Next Step customers. The use of this econometric approach reduces the potential for bias and results in a better comparison between those in receipt of Next Step support and those individuals not-in-receipt of support. The PSM modelling approach was successful. Across all variables considered, the difference in the means between the treatment and control (matched) groups was substantially less than between the treated and untreated (unmatched) groups. As such, we have some confidence in the subsequent results.
  • Looking at the entire sample of Next Step customers, for which approximately 528,000 have employment and benefits records, 55% of Next Step customers were in employment 12 months prior to the intervention, compared to 59% in the control group. The rate of employment increases for the control group to 63% at the time of intervention, while the employment rate for Next Step customers remains relatively constant at 55% (corresponding to a gap of 8 percentage points (pp) at the time of intervention). Post-support, the rate of employment for the control group increases further with 65% in employment six months after Next Step customers receive support. In contrast, for those Next Step customers, employment rates increased by 9 percentage points from 55% to 64% in the six months post intervention. The gap in employment rates between the treatment and control groups stood at 1 percentage point six months post-support, implying that approximately 85% of the employment gap had been erased following the receipt of support.
  • Adopting an equivalent approach in relation to JSA, the analysis suggests that the control group exhibited a steady downward trend in the proportion claiming JSA (from 13% twelve months before the intervention to 9% at the time of the intervention and 8% six months post-intervention). Next Step customers experienced an increase in the proportion claiming JSA up until a peak of almost 39% at the point of receiving support, which demonstrates the rapid decline in labour market outcomes prior to engaging with Next Step. After the receipt of Next Step support, the proportion of Next Step customers in receipt of JSA decreased rapidly to 27% and 21% three and six months post-support respectively. The overall gap between the treatment and control groups was greatest at the time of intervention (almost 30pp), but declined to between 12 and 13 percentage points six months following the intervention. Although the benefit dependency gap was not eliminated, nor does it return to the level that existed 12 months pre-support, from the highest point, the size of the JSA dependency gap was reduced by approximately 59% in the six months post intervention.
  • Despite our best efforts, the control group is too different from the Next Step group in both analyses above to show clear conclusions about the impact of careers advice interventions. However, the analysis describes what has happened to the group of people who accessed Next Steps both before and following the Next Steps intervention, and that is interesting, for example showing a substantial reduction from 39% claiming JSA at the time of the intervention to 21% 6 months later. There will be many reasons why people obtained employment or otherwise left JSA benefit, to which the Next Steps intervention has contributed. Because of the nature of the control group, it has not been possible at this stage to say how large that contribution was. The findings here are clearly consistent with a positive impact of Next Step on moving from JSA benefit. But they are also consistent with there being a negligible effect. At this stage we cannot clearly distinguish between these outcomes. Further work will be needed to obtain a better understanding of the additional benefit of Next Step, as indicated at the end of the Executive Summary.
  • We replicated this initial analysis using a range of different ‘cuts’ of data and different Propensity Score Matching models and found that the results were qualitatively unchanged.
Full report (PDF 116pp)

No comments: