What Can Experiments Tell Us About How to Improve Governance?

This study by Rachel M. Gisselquist & Miguel Angel Niño-Zarazúa 

. . . identified 139 relevant papers grouped into three major types of policy interventions that aim to: (1) improve supply-side capabilities of governments; (2) change individual behaviour through various devices, notably incentives, and (3) improve informational asymmetries. We find that randomized controlled trials can be useful in studying the effects of some policy interventions in the governance area, but they are limited in significant ways: they are ill-equipped to study broader governance issues associated with macro-structural shifts, national level variation in institutions and political culture, and leadership. Randomized controlled trials are best for studying targeted interventions, particularly in areas of public goods provision, voting behaviour, and specific measures to address corruption and improve accountability; however, they can provide little traction on whether the intervention is transferable and ‘could work’ (and why) in other contexts, and in the longer run.

To be sure, the authors discuss relevant papers, for example

Banerjee et al. (2012) test the impact of four low-cost reforms across police stations in eleven districts in Rajasthan. Results suggest that two of these reforms – freezing staff transfers between police stations and providing in-service training in investigation skills and ‘soft’ skills like communication and leadership – were effective in improving police effectiveness and public satisfaction, while the other two reforms – placing community observers in police stations and a weekly duty rotation – were not effective. P. 10. 

Other relevant studies, like this one, are not cited. 

Does Money Burn Fat?

This paper provides comprehensive evidence for the usefulness of financial incentives to encourage obese people to reduce body weight. The analysis has improved upon the existing literature by conducting a large randomized experiment, distinguishing between two different premium levels as well as gender, and accounting for non-random sample attrition. The results are highly relevant for both employers and health insurance companies that search for effective measures to reduce obesity among their employees or their insured. Financial incentives for weight loss could be easily integrated in existing bonus programs for healthy behavior, disease management programs, and other health promotion programs.

We find that a reward of e 150 [€] doubles the success in reducing body weight. The control group who did not receive a reward for achieving an individual-specific weight-loss target of between 6 and 8 percent of initial body weight within four months has lost, on average, 2.4 percent. Patients who were rewarded by up to e 150 lost about 5 percent. The causal effect of the reward amounts to 2.6 percentage points, with the estimate being insensitive with respect to attrition. The size of the effect is in line with most recent medical trials on this topic, even though these are not able to rule out the possibility of biased estimates of the treatment effect due to non-random sample attrition.

[It does not seem like a large effect (?) . . . ] 
That is from the new working paper "Does Money Burn Fat? Evidence from a Randomized Experiment" by Augurzky et al (September 2012). 
A graph from the paper:

Financial and non-financial incentives to improve students' performance

A new paper by Levitt et. al. (June 2012) explores the effect of financial and non-financial incentives on the performance of high-school students in standardized tests. They find the usual - that incentives matter, and that "incentives framed as losses have more robust effects than comparable incentives framed as gains." 

Two findings stand out:

1) "Rewards provided with a delay have little impact on student performance." And, 

2) "The introduction of rewards does not crowd-out future effort."

The title of the paper is "The Behavioralist Goes to School: Leveraging Behavioral Economics to Improve Educational Performance."

May be grades were not good enough incentives after all? But something does not sound right about coming to class with awards, chocolates, trophies, and money. In the longer run, once in a new equilibrium, will larger incentives be necessary? Where will this stop?  

Experimental Political Science

The rise of experiments as one of the most prominent empirical strategies has led to new advances in the study of democracy and development. So far, some experimental results have confirmed previous arguments, such as the effectiveness of clientelism as a mobilization strategy and the prevalence of political and social inequalities despite institutional innovations. Other experiments have revealed relationships that only a randomized control trial could uncover, like the fact that clientelist appeals reinforce ethnic voting and not the other way around. Finally, some experiments are revolutionizing the measurement of core concepts in the field. For example, we now know that vote buying measured experimentally is more prevalent than what observational studies suggested. 

Going forward, field experiments in collaboration with policymakers, governments, and NGOs are a promising line of research. The next round of experiments, however, faces considerable challenges, including those we have highlighted throughout this chapter. First, researchers must find creative ways to design interventions that are attractive to potential partners but that still speak convincingly to theoretically relevant questions. In doing so, researchers must pay special attention to internal validity issues. Second, a more analytical approach would help guide researchers to design experiments that enable significant accumulation of knowledge to take place. Finally, as the scope of25experimentation expands, the tradeoff between external validity and ethical concerns will become more salient.Despite these challenges, experimental research on development and democracy is a productive and exciting endeavor. As insightful as the experimental research has been up until now, numerous substantive questions remain unanswered. Hopefully the selection of studies covered in this chapter illustrate how experiments can be used as a research tool to study broader and more central questions about the relationship between democracy and development.

That is from the paper by De La O and  Wantchekon. It is here.