Substack

Thursday, March 7, 2019

The problem with performance incentives in government

Much has been written about how financial performance incentives can help address state capacity weaknesses and improve outcomes in public systems, especially in certain sectors. 

The mainstream critique to this has revolved around the difficulty of credibly quantifying and then capturing outputs which are both reliable and proximate contributors to the desired outcomes. There is another equally important concern.

Consider this. Public servants, especially field-level functionaries (teachers, medical supervisors, bill collectors, tax inspectors, building inspectors, and so on) are among the highest paid workers (among comparable levels) in any developing country. This also partially explains the scramble for these jobs. In each of these cases, their actual performance is significantly lower that not just their formal responsibilities demand, but much lower than comparators in the private sector. 

So it is commonplace to have a tax collector or inspector assed and realise 30-50% of his estimated demand. Or an electrical engineer manning a substation have distribution losses in the range of 30-40%. Or have a teacher who attends the school only half the days, and whose student learning outcomes are abysmal. Or a primary care hospital doctor achieving just 30% universal immunisation and 40% institutional deliveries. This is business as usual in most low-income countries.

So, you have the highest paid officials achieving such low outcomes, and that too lower than even their much lower paid private sector comparators.  

In this context, economists recommend incentivising these functionaries with performance payments. They proffer evidence from sanitised RCT pilots. These results can appear alluring - logical and neat answers to a complex problem. 

But this is deceptive for at least two reasons. One, its general equilibrium incentive distortion effects. A solution like this immediately shifts the frame of reference from one where the challenge was to improve outcomes using existing levers to one where the challenge is to get the most optimal design of incentives. Implicit in the latter is an acknowledgement that the poor status-quo is inevitable and alright and we are now therefore looking at the performance incentive lever to improve outcomes. The moral hazard from this can create unexpected outcomes. Poor performance soon gets to becoming psychologically rationalised by truant officials. Further, not only the base salary but the structure of incentives too become factors in wage negotiations.

Second, implementation of financial performance incentives at scale can have unexpected effects. The most likely impact is that over time, performance incentives come to be viewed as an entitlement. Or for example, over time, it is reasonable to have expectations and attitudes that rebalance baseline performance downwards and revision upwards of incentives. When an incentive gets scaled up, employees tend to forget the baseline salary (which is then taken for granted) and there is the danger of the incentive becoming seen almost as an entitlement - if you want us to improve the performance, give us more incentives! 

There are several examples of distortions along these lines which have been engendered by financial performance incentives. The practice of bonuses is commonplace in public sector power (generation, transmission, and distribution) companies in India and their impact in terms of actually lowering transmission and distribution losses is perhaps nil. 

When thinking about such neat solutions which we encounter in our daily lives and how they could be transplanted to public sector, we overlook the numerous other formal institutional factors and informal norms that contribute to shaping expectations and attitudes in private sector which in turn contributes to the success of performance incentives there. And we forget that these factors are deficient or absent in public sector, not to mention the fact that public sector is never about selling widgets. 

Update 1 (13.11.2021)

A report by UK's Institute for Government assesses performance bonuses in public systems and finds little evidence of it being effective

1 comment:

Chinmaya Holla said...

On the question of premium, this may interest you: http://documents.worldbank.org/curated/en/398361551117216050/pdf/WPS8754.pdf