Use of profiling for resource allocation, action planning and matching

Size: px
Start display at page:

Download "Use of profiling for resource allocation, action planning and matching"

Transcription

1 Use of profiling for resource allocation, action planning and matching Regina Konle-Seidl PES to PES Dialogue Conference Brussels, May 2011 D - Institute for Employment Research (IAB)

2 Profiling More than just predicting who is at risk of becoming long-term unemployed Method of allocating employment services based upon client characteristics Allows for an early identification of clients needs within the scope of a customized approach Wider definition allows for a wider range of country experiences 2

3 The presentation addresses the following questions: What are the main approaches to profiling? In selected European countries: DE, DK, FR, NL, SE compared to experienced profiling countries AUS and USA Statistical profiling versus or in conjunction with qualitative assessment tools How is profiling used? Nature and intensity of linking profiling to targeted assistance Use of jobseeker profiles for automatic matching How effective is profiling, targeting and early intervention? Predictive power of statistical profiling? Do profiling and targeting improve the quality of the matching process? Deadweight effects? What has to be taken into account for implementation? 3

4 Main approaches to profiling Profiling methods: no clear trend across European countries shift from statistical to qualitative assessment tools in DE, DK to no profiling per se in NL statistical profiling in FIN, IRE, SE (trials) as part of an expert system (soft profiling) caseworker usually makes final decision diagnosis: prediction of LTU risk or benefit exhaustion, but also job search capacity; probable duration of UE variety of variables and factors included; only few countries include soft skills in an integrated way (AUS, DE) timing: - initial face-to-face interview (3 to 4 weeks after registration) or when registering for benefits (AUS, USA) - regular re-judgment (dynamic profiling) 4

5 Use of profiling calculation of risk scores decisive for allocation of re-employment resources in AUS and USA JSCI scores determine allocation to 4 different stream services; provide providers are rewarded according to stream services client differentiation in Europe: opposing developments; DE: from 4 to 6 profiles, DK: from 5 to 3 matches weak direct link to allocation of staff resources 5

6 Use of profiling Targeted assistance (action planning) link to coherent streaming policies (sequence of interventions) very dissimilar DK: match groups rather rough indication for referral to services DE (more) and FR (less) standardized action planning characterized by 4 phases : 1) IT-based profiling 2) goal definition according to profiles (needs) 3) selection of appropriate measures/software guided service strategy 4) mutually agreed personalized action plan 6

7 Jobseeker profiles in Germany Client profiles Market profile Activation profile Promotion profile Development profile Stabilisation profile Support profile Exit into regular LM < 6 months regular LM < 6 months regular LM < 12 months regular LM >12 months improve employability >12 months improve employability >12 months Prognosis Close to the labour market Complex profiles Profiling in Denmark 7

8 DE: Use of profiling for automatic matching Automatic Matching Applicant Vacancy 1 Angebot Fähigkeiten Schulabschluss Matching Angebot Arbeitszeit Region Profiling Ausbildung... Branche... Nachfrage Nachfrage Arbeitszeit Fähigkeiten Region Schulabschluss Branche Ausbildung Strength analysis Job-related and holistic competencies relevant for bidirectional matching + Potential analysis = Profile (Profillage) ELISE 8

9 Statistically assisted targeting alternative way to allocate individuals to re-integration services referral to services based upon a regular assessment of what works for whom sophisticated targeting systems have been developed in Canada (SOMS), CH (SAPS), Germany (TrEffeR) and US /FDSS) and partially in DK (Job Barometer) proven to be effective (CA, CH) but nowhere currently in use, mainly due to caseworker resistance 9

10 How effective is profiling, targeting and early intervention? Predictive power of statistical profiling models USA (length of benefit spells) and AUS (outflow from UE) is high satisfactory in European countries (~ 70% in DK, SE) evaluation of WPRS in six US states: cost-effective (savings in UI benefits) and increase in yearly earnings Quality of the matching process general acknowledgment that profiling provides a more systematic framework for caseworkers increase in transparency of service delivery DE: validity and plausibility of profiles and applied service strategies increased but tools are not well accepted by caseworkers (complaints about documentation workload, inflexibility, loss of autonomy ) little reliable information on whether caseworkers really use profiling tools adequately 10

11 How effective is profiling, targeting and early intervention? Efficiency gains of early intervention and personalized delivery models general evidence gap on the impact on central performance indicators (shortening of UE spell, UE off-flow rates) technically difficult to identify casual relations between services assessed, services accessed and outcomes Cost-benefit calculations of early intervention are rare evidence from an UK segmentation exercise (JSA claimants): prediction model was correct in 70% of cases, but impact of false predictions would negate any savings but also empirical evidence that more caseworkers reduce individual UE duration (DE: by 10 days, workload 1:70) change in contact and activation regime (DK) reduced UE duration by 3 weeks; net benefits of ~ 2000 per UE spell 11

12 Lessons for implementation Obstacles high risk of miss-classifications too early transfer to re-employment services diminish search efforts focus on strengths and opportunities instead of weaknesses and barriers high initial set-up costs of sophisticated profiling and targeting systems large investments in development (good data and software) Profiling can help PES to improve service delivery making full use of available information (incl. information on soft skills) teaching caseworkers how to use tools and targeting results ensuring frontline staff use instruments as intended 12

13 Implications for practitioners (managers and caseworkers) practical experience not sufficient to make good decisions about effectiveness of re-integration measures further professionalization is desirable (esp. concerning counselling and guidance skills) management tools at the team level can help to optimize professional help more leadership from jobcentre managers is required uniform standards may help to improve equal treatment across different groups of benefit recipients early consultation of major stakeholders 13

14 Conclusions soft profiling is common practice but opposing developments across European countries concerning the intensity of using profiling and early intervention strategies work first and mandatory activation seem to replace more differentiated segmentation and early intervention in some countries (e.g. NL, DK) more differentiation required in countries where PES is now in charge of more heterogeneous (insured and non-insured) client groups (e.g. UK, AT) with pressure on budgetary and staff resources, upgrading of Internet job search and e- services across client groups high on PES agenda balancing intensive support with a self-help strategy is still a crucial challenge better use of customer profiles for different purposes (e.g. for automatic matching and client segmentation) stretched budgets require more knowledge about what works 14