Performance related pay in public services: theory and evidence - - PowerPoint PPT Presentation

Performance‐related pay in public services: theory and evidence

Simon Burgess

Centre for Market and Public Organisation

• Public services are a sizeable part of UK
• utput and deliver key outcomes.
• Efficiency and responsiveness are essential.
• Performance pay is common in private sector

– about half of workplaces have some form of performance pay.

• Why not in the public sector?
• Performance pay is a powerful tool, but … it is

a powerful tool.

PIRU/ LSHTM, Feb 2013 www.bris.ac.uk/cmpo 2

• Theory

– Why have PRP? – Why not? – What’s special about the public sector?

• Design issues in PRP schemes
• Evidence

– Education – Health care – Public servants

• Summary

PIRU/ LSHTM, Feb 2013 www.bris.ac.uk/cmpo 3

Theory 1: Why have PRP?

• Core issue: cannot pay people for the effort they

put in because we can’t observe it

• So instead pay for output, proxy for effort
• PRP does two things:

– Motivates, incentivises effort – Selects, attracts higher performing workers

• Aligns incentives for worker with those of the
• rganisation
• Work out the optimal ‘gearing’ of the PRP

– Trade‐off between incentives and income risk – Other factors

PIRU/ LSHTM, Feb 2013 www.bris.ac.uk/cmpo 4

Theory 2: Issues and problems

• Optimal PRP rate may be very low.
• Multiple tasks
• Multiple tasks and differential measurement
• Inter‐temporal aspects

– ratchet effects – career concerns

• Intrinsic motivation
• De‐professionalistion
• Design issues

– design is hard to get right and poor design can have adverse unintended consequences

PIRU/ LSHTM, Feb 2013 www.bris.ac.uk/cmpo 5

Theory 3: What’s special about the public sector?

• PRP much less common in public sector: optimal rate is

low, or union power?

• Measurement issues

– Common occupations – Different occupations

• Working in teams, collaboration not rivalry
• Intrinsic motivation
• Multiple principals
• No single factor appears special, but the combination is

unique once we add:

• Fuzzier goals at organisation level

PIRU/ LSHTM, Feb 2013 www.bris.ac.uk/cmpo 6

Design Issues

• Linear (per‐unit) or threshold?
• Individual or team?
• Objective or subjective?
• Relative or absolute performance?

Tournament?

• Monetary reward? Or resources for clients?
• Who for? Workers? Bosses? Clients?
• Framing and the use of loss aversion

PIRU/ LSHTM, Feb 2013 www.bris.ac.uk/cmpo 7

Evidence

• For a policy discussion we need the answer to a causal

question:

– If we introduce PRP, how will productivity change?

• Evidential requirements for causality are hard.
• There is little robust evidence on the impact of

performance pay in the public sector

– even though this sector employs as many people as manufacturing does.

• Interested in the main outcome, and potential

unwanted side‐effects

• Consider: education, health care and public service.

PIRU/ LSHTM, Feb 2013 www.bris.ac.uk/cmpo 8

Evidence 1: Education

• International evidence is mixed:

– Some studies show positive effects (eg Lavy) – Others show no effects (eg Fryer) – Often strong effects in developing countries – On balance, positive but a lot of variation

• Differences in design may be important
• Little evidence of gaming or undesirable effects
• Study for England:

– significant effect of the scheme, about half a GCSE grade per pupil – despite poorly designed scheme.

PIRU/ LSHTM, Feb 2013 www.bris.ac.uk/cmpo 9

Evidence 2: Health care

• Systematic reviews find different effects but little

evidence in favour of performance pay.

• Study of Advancing Quality in NW England

(Sutton et al 2012):

– Team‐based (hospital‐level) performance pay; tournament basis; multi‐task safeguards. – Rewards not salary but invested in clinical care – Strong data infrastructure and support – Significant fall in mortality – Different strategies to achieve the gains; collaboration not competition.

PIRU/ LSHTM, Feb 2013 www.bris.ac.uk/cmpo 10

Evidence 3: Public servants

• Again, not a lot of robust causal evidence. Two examples:
• Jobcentre Plus

– incentive scheme based on team performance, covered five different targets – Incentives had a substantial positive effect in small offices. Peer monitoring

• vercame free rider problems in small units

– No impact of performance pay on quality measures.

• HMCE (as was)

– Again team‐based scheme; teams include the manager. – The incentive structure raised individuals’ tax yield and productivity. – The most successful team allocated more of all its workers’ time to the incentivised tasks, but disproportionately reallocated the time of its efficient workers. – This reallocation was the more important contributor to the overall outcome.

PIRU/ LSHTM, Feb 2013 www.bris.ac.uk/cmpo 11

Conclusions

• Incentives work

– Private sector, public sector; unskilled occupations, skilled manual, professionals.

• But be careful!

– “On paying for A whilst hoping for B”

• Design matters crucially

– match scheme to production; content of the incentive; framing – loss aversion; measurement; multi‐tasking safeguards; ...

• Requires clarity on the organisation’s goals.

PIRU/ LSHTM, Feb 2013 www.bris.ac.uk/cmpo 12