Public management is a curious mix of uncertainty and dogma. Uncertainty — about how to structure public sector organizations, about how to link the budget to performance, about how to motivate employees — is quite appropriate given the weak theoretical basis and the even weaker empirics, and thus the frequent changes of direction on these topics are hardly surprising. What is perhaps more surprising is the dogmatic belief in certain policy prescriptions at any given point in time. It seems that what we believe in changes rapidly; but our certainty about the fashion of the day runs deep.
Take the case of performance-related pay (PRP). The current dogma of public administration specialists, in academia and policy circles, is to reject PRP. This opposition however, bumps up against the awkward reality that a growing number of countries are introducing PRP and a growing body of rigorous research on PRP for teachers and medical staff is showing that, under specific conditions, outcomes can indeed be very positive.
Public administration experts draw their certainty from largely sociological studies that have focused on PRP in OECD countries and which show little association with improved public sector performance, and in fact often very negative results (see OECD 2005, Cardona 2007, Behn 2004). The literature highlights the many problems of implementing PRP and the perverse outcomes that it can generate — the difficulty in determining who should make the decision on PRP, the potential negative impact on motivation of those who don’t get the bonus and the jealousies it fosters, the difficulties in determining the optimal size of the performance bonus, and the more fundamental problems of how to measure performance in a complex bureaucracy. Given these problems, it is not surprising why many managers distribute performance bonuses equally among staff, thus defeating their purpose. The bottom line then is that civil services are ill equipped to make PRP work and that, instead of PRP, policy-makers should focus on introducing a performance-oriented culture in the civil service by better linking longer term staff careers to performance. However, it is unclear from the literature as to why the civil service would be better at designing performance-based promotion arrangements if it cannot get the design of performance-based pay right. In particular, given that the space for promotions is so limited, and that pay is compressed, how would a better performance appraisal system on its own motivate staff?
By contrast, policy-makers in many OECD and middle income countries are, today, equally robustly convinced of the merits of PRP. Maybe they are more concerned about the signaling device to the public that PRP provides ("we're really getting a grip on these lazy civil servants"), or maybe they know more about ground realities than the researchers and see that PRP has real but intangible benefits such as fostering more positive collaboration between staff and their supervisors (see Marsden 2009).
More research is doubtless needed. One comprehensive retrospective by Perry et al (2008) of 57 studies over the past three decades concluded that we will not get a clear picture until we move beyond simple assessments of employee attitudes and perceptions. Lessons emerging from the education and health research, which is more grounded in economics and is more empirically rigorous, including a small but growing body of “gold standard” randomized controlled trials (RCTs) of performance-related pay for teachers and health workers, is also much more mixed in its conclusions. For example, Esther Duflo and collaborators found that linking teacher pay to attendance reduced teacher absenteeism by 21% in the control group. A study by Muralidharan and Sundararaman (2011) in Adhara Pradesh, India found that both individual and group bonuses for teachers had a positive impact on student achievement, as measured by standardized tests. Lavy (2009) found positive effects of both group and individual incentives for high school teachers on student achievement in Israel. By contrast, Glewwe, Ilias, and Kremer (2010), found that while students in treatment schools initially had higher test scores these gains disappeared after one year. Recent studies in the US have also failed to find an impact of PRP on student learning outcomes.
To sum up, this mixed evidence gives every reason for an open-minded, case by case, assessment of what is likely to work, backed up by a strong research agenda to figure out precisely what design features of PRP can work under what conditions, what are the mechanisms by which they work, and what features of the local institutional and political context are important. Rather than rejecting it off-hand, or welcoming it in every case, let’s start helping countries selectively design PRP schemes that could work and let’s do it in a way in which we can measure the results.
Photo Credit: flickr user kolix