Improving public sector performance

Why are people in developing countries poor?

• Note that this question is frequently framed in terms of countries

​

• Implicitly acknowledges an important role for the state

​

• Indeed, we have seen in last two lectures importance of state behaviour for resource misallocation and regulation enforcement

​

• How can developing countries improve state behaviour?

​

From policy to people

• Policy important but increasingly economists are worried about enforcement and implementation
• Increasing evidence that individuals matter
• E.g. Barteska & Lee (2024, JMP)

​

• Relatively new field for economists so important positive research agenda as well as normative one
• E.g. Gathen (2024) estimates public sector worker skill using household survey data

​

• Work on this field is increasing at a very fast rate
• It’s hard to do RCTs / get data on public agents, but not impossible

Key questions

​

1. How does the state recruit/promote the right people?

​

• How do we improve their performance once they are there?

​

Besley et al. (2022, ARE) provide a survey of the literature

​

1. How should the state recruit and promote?

Two key questions:

​

A. How do policies affect the pool of candidates?

​

B. How should we choose from amongst these candidates?

How do policies affect the pool of candidates?

Typical focus is on how job characteristics impact pool:

• E.g. framing, pay level, peformance related pay
• I.e. where papers running RCTs often make contribution – see survey

​

Also interesting to look at how other policies impact the pool:

• Cruces et al. (2023, AER:I) - encouraging dishonesty encourages public service in Argentina!
• Xun (2024, JMP) – auditing Brazilian municipalities reduces share of high-ability students aiming to go to public service

​

And how policies impact other sectors:

• Mangal (2024, JDE, JMP) shows competitive exams worsen candidates labour outcomes

​

How should we choose from amongst the candidates?

Often framed as “how should recruiters’ discretion be constrained”?

​

Standard view is that discretion allows patronage, which leads to lower quality / less motivated employees:

• Several papers cited in survey find evidence for this
• Li (2020, JMP) finds good teachers quit and perform less well after unfair ‘connected’ promotions

​

​

​

​

​

​

​

​

​

But is discretion always bad?

But now several papers suggesting that, in some circumstances, discretion can lead to relatively good selection:

• See survey paper
• Salgado (2024, JMP) colonial officials raise more taxes when connected to superiors in Spain

​

Open question – when/how can recruitment processes have positive effects of discretion without patronage?

2. How can we improve public servants’ performance ?

​

Massive literature in labour about how to motivate workers to peform – so why should this be different in developing country states?

​

• Extrinsic vs intrinsic motivation?

​

• Can we rely on intrinsic or social motivations?

​

• Do extrinsic motivations (i.e. pay, promotions) not work in the public sector?

​

​

​

Does intrinsic or social motivation help?

Generally, little evidence on social motivation (see Ashraf & Bandiera, 2018 for a survey)

​

Several papers don’t find much evidence for it

• E.g. Vanden Eynde et al. (2018, AEJ EP) find police officers perform worse when the same ethnicity is ruling, apparently for behavioral reasons

​

But it surely does exist sometimes, so how can we exploit this?

How can we harness social motivations?

Some recent examples of RCTs that manage this:

• Azuli et al. (2020) run national training with Ghanain civil service focused on changing motivation & norms – improved team performance
• Harris et al. (2024) train Ghanain police officers in ethics and find cheat less 20 months later
• Agte & Begoya (2024) find Peruvian psychologists randomly sent to poor areas are more likely to stay in public service & serve poorer areas later

​

But certainly at the moment we know much less well how to do this than how to use extrinsic motivations – so do extrinsic motivations work?

Does pay for perfomance motivate?

We now have quite a few examples of pay for performance in the public sector (see survey) – and it seems to generally work.

​

Now question is more on how should we best design such programs:

• Gaduh et al. (2022) pay teachers in Indonesia either as a function of attendance or community scorecards – both work
• Deserranno r al. (2024, JPE) find incentives improve health worker performance – particularly when shared between worker and supervisor

​

​

Do career incentives motivate ?

Here there is less evidence, but still generally consistent with expectations:

• Several examples in survey
• Deserrano et al. (2024, AERI) randomize meritocratic promotion – increases productivity (especially when employees randomized to have higher beliefs of pay scale)

​

​

​

​

​

So why don’t we see stronger incentives?

My reading of the evidence here:

• Evidence on using intrinsic motivation instead is weak – evidence of crowd-out even weaker
• Standard incentives (pay and promotion) seem to work
• Not much evidence of serious multitasking problems
• Although this may be biased by the contexts studied
• Little evidence suggesting negative effects on recruitment

​

Of course, incentives depend on monitoring. So is the problem in developing countries that monitoring is poor?

​

​

What’s the impact of indirect monitoring?

A lot of papers on community based monitoring but results are heterogeneous – perhaps not surprising given link to incentives is unclear?

​

Examples:

• Christensen et al. (2020, QJE) show community monitoring in Sierra Leone increased service quality and trust, reducing Ebola deaths
• Raffler et al. (2024, JOP) scale up Bjorkman & Svensson community based monitoring and find very little impact
• Buntaine et al. (2023, AER) show citizens reporting on pollution violations decrease violations

​

​

​

​

​

​

​

​

​

What about monitoring inside the bureacracy?

Perhaps clearer to think of improving information flow within government:

​

• Audits:
• Wong (2023, JMP): Explores how corruption adjusts as a function of audit threat
• Wolfram r al. (2023): Audits improve performance (but contracting constraints slow process down)
• Schaner et al. (2023): Even social audits work
• Information systems:
• Mattson (2024, AEJ:AE, JMP)
• Dodge et al. (2023): Giving supervisors access to e-governance systems improves performance
• Gulzar et al. (2024): Flagging bad performance to central unit had no impact
• Complaint mechanisms:
• Kumar & Sharan (2023, JMP) randomly assist villages in sending complaints and finds to be effective

​

​

Paying more attention to the monitors

However, internal monitoring can have negative effects:

• Gerardino et al. (2023, AEJ:AE): Government audits reduce competitive auctions in procurement, increasing price paid
• Bandiera et al. (2021, QJE): Reducing the power of monitors leads to better outcomes (probably because monitors extort)

​

So question is, how can we get better (not just more) monitoring?

• E.g. Raffler (2023, APSR) finds training local politicians to be better monitors improves performance (except for when politicians are aligned with the ruling party)

​

Still lots of unanswered questions here of understanding what works…

​

​

​

​

​

​

Conclusions

Some outstanding questions:

​

• Does intrinsic motivation matter much to policy design?

​

• Are the positive results we find of p4p interventions just because we are tinkering at the small space where monitoring is easy?

​

• How do we deal with ‘bad’ monitors?

​

• Is there really a developing country difference in this field? Or a public sector one?

​

​

​