Choices, choices: Should IDOs give their employees more autonomy, or simply choose them better?
Dan Honig argues International Development Organisations (IDOs) should give their employees more autonomy in decision-making, particularly in unpredictable situations with hard-to-measure project aims. Crucially, this presumes motivated agents. The answer to this: better hiring.
Dan Honig recently published the book Navigation by Judgment – Why and When Top Down Management of Foreign Aid Doesn’t Work. In it, he argues IDOs are measuring outcomes too much, based on “the notion that counting everything in government programmes… will produce better policy choices and improve management”. Instead, he argues they should use “Navigation by Judgment” (NbJ) more – when “IDOs rely on the judgment of field agents to guide the strategic direction of development projects”.
An extreme example of Navigating by Judgment is a government school teacher who draws on their experience to draw up their own lesson plan, and could even teach a different subject if they thought it appropriate. At the other end, Navigating from the Top, is a Bridge Academy teacher reading from the lesson script drawn up by Bridge, using a tablet, which also tracks the teacher’s progress.
Honig’s argument: IDOs face pressure to prove themselves to their funders, hence use metrics more than they should
Why do IDOs measure so much? Honig argues this is due to their “authorising environment”, i.e. their sources of funding. IDOs “need to appear successful to the politicians to whom [they] report”, so they incentivise their employees to focus on hitting measurable targets, possibly at the expense of the actual project aim. He gives a real example of a project in East Timor measuring how many agricultural extension workers (i.e. trainers) were trained, rather than the actual impact on farmers. The trainers kept leaving for better jobs once they had been trained, so the project had no impact. However, the metrics looked good, so it was not changed. Hence, he argues, there can be a “tension between appearing successful and actually achieving success”.
Admittedly, Honig argues NbJ shouldn’t be used exclusively, but is “sometimes a good idea”, particularly in cases of:
- Low project external verifiability (the “tightness of the link between the best possible quantifiable output and project goals”).
- Low environmental predictability (“the extent to which the project environment is one in which there are ‘unknown unknowns’”).
Using a dataset of 14,000 development projects, and four case studies each for USAID and Dfid, Honig finds a systematic relationship between the propensity to NbJ and project success. The analysis suggests “greater returns to Navigation by Judgment in less predictable environments”, as NbJ “cushions the fall” of project performance decline in these contexts. Why would NbJ work?
- There can “be bad or inappropriate rules; even the best-designed, most logical controls may sometimes preclude good action in their desire to avoid bad action”. This argument is plausible, and he supports it with anecdotes of outcome measures “gone rogue”.
- NbJ allows agents to incorporate “soft information” – “information that field agents can observe but that cannot be easily codified or verified”.
- NbJ allows projects to be “more flexible and adapt to changing circumstances”.
Hence NbJ is “a strategy to employ when it is less bad than the distortions and constraints of top-down control”, a second-best strategy.
Is there a better way?
Honig’s argument has merit, but as he himself admits, there are issues with NbJ, even when used in the situations he suggests.
- NbJ requires trusting employees to do the right thing. But, they may not have the same goals as their bosses, may not know how to achieve the new, fuzzier outcome goal, or may make mistakes. Further, they will also find it easier to engage in illegal or undesired actions. Principal control means behaviour is more standardised, hopefully reducing bias and prejudice.
- NbJ cannot be paired with performance incentives. Otherwise, employees would put more effort towards the incentivised behaviours. However, incentives have been shown to work, sometimes even in hard-to-measure environments – e.g. incentivising teacher attendance improves learning outcomes, too. Evidence suggests they can leverage intrinsic motivation rather than crowding it out (Ashraf et al. 2014, Rasul and Rogger 2014).
Another plausible use of incentives when external verifiability is low are Development Impact Bonds (DIBs). Here, agents are incentivised on a final outcome (e.g. an existing DIB for girls’ learning in primary schools), without being told how to get there. They can still use their judgment in how to get to the goal, without necessitating the complete trust NbJ implies.
If an IDO weighs up the pros and cons and decides to use NbJ in a specific context, this requires employees who are intrinsically motivated to achieve project aims. To ensure this, IDOs should use recent evidence on how to hire pro-socially motivated employees.
Better hiring is crucial when giving more autonomy
A recent IGC Growth Brief by Bandiera et al. (2017) reported that offering incentives to employees allows motivating better job performance in the public sector, but that higher-quality hiring takes priority. Instead of focusing on incentives, they suggested that governments should focus on recruiting more qualified, motivated staff in the first place – by offering higher salaries, and opening up recruitment.
So how can IDOs get intrinsically motivated employees?
- Emphasise career promotion prospects: Job adverts for community healthcare workers in Zambia with a “go-getter” message helped to attract more qualified who were no less pro-social, compared with a “do-gooder” message (Ashraf et al. 2014). These were also more effective at delivering health services to communities.
- Pay for ability: Banuri and Keefer (2015) measured the effort put in by college students committed to joining the Indonesian public sector under different pay schemes. Pay-for-performance only made less-than-average prosocial candidates put in more effort. Under “Pay-for-ability”, where pay is based on some ability measure (e.g. entrance examinations) but then does not depend on effort, prosocial students put in as much effort as under pay-for-performance. The scheme was just as attractive to prosocial students, and is superior particularly when performance is difficult to measure.
- Otherwise, incentivise: Higher wages can help get smarter, more experienced candidates (Dal Bo et al. 2013). Once candidates are employed, reward them: both financial and non-financial rewards worked when incentivising health workers to sell condoms and prevent HIV (Ashraf et al. 2014), a fairly pro-social activity.
Honig is correct in identifying the pressures IDOs face from their authorising environments. His vision of IDOs giving their staff more autonomy is compelling – but using recent evidence to hire prosocial staff is an important part of this.
Additionally, he does not address the issue that, while this approach may lead to more successful outcomes, it does not counteract the political pressures faced by the IDO to prove themselves. It is difficult for the IDO to measure impact without also incentivising effort-substitution, so the IDO would struggle to prove to their funders that this approach is superior. Ideally, IDOs would use NbJ where they can e.g. through Development Impact Bonds, which marry autonomy and end-outcome measurement. In other cases, even if it were advantageous, it may not be politically feasible.
Ashraf, N, O Bandiera and K Jack (2014), “No margin, no mission? A field experiment on incentives for public service delivery”, Journal of Public Economics, 120: 1-17.
Ashraf, N, O Bandiera and S L Lee (2016), “Do-gooders and gogetters: Career incentives, selection, and performance in public service delivery”, IGC Working Paper.
Bandiera, O, A Khan, J Tobias (2017), “Rewarding bureaucrats: Can incentives improve public sector performance”, IGC.
Banuri, S and P Keefer (2015), “How to pay public servants: Rewarding ability and performance”, IGC
Dal Bo, E, F Finan and M Rossi (2013), “Strengthening state capabilities: The role of financial incentives in the call to public service”, Quarterly Journal of Economics, 1169-.
Rasul, I and D Rogger (2014), “Management of bureaucrats and public service delivery: Evidence from the Nigerian civil service”, LSE.
 Honig argues that accountability can be maintained through, for example, professional reputations – but this will not work in all contexts.
 Honig argues NbJ does not imply scrapping measurement entirely – he just thinks we should use it to evaluate impact, and not to project manage. However, this is to some extent implausible – if agents know their project will be evaluated based on an indicator, it is likely they will again gear their behaviour towards this indicator.