Imagine you are running the recruitment process for a government agency and you are trying to attract high quality, public service oriented staff to work in difficult agencies. How should you do this? If you offer higher wages, maybe it will get you higher quality folks, but will you lose public service motivation? And how do you get these high quality folks to go to remote and dangerous areas?
This is a question tackled with some elegance in a new paper by Ernesto Bo, Frederico Finan and Martin Rossi. They are working with a recruitment experiment by the Mexican Regional Development Program, which places coordinators and community development agents within particularly marginalized municipalities to help them build up the provision of public goods.
Bo and company start with the question: will higher wages get you better applicants? This would seem obvious, but spending some time in certain African countries and looking at parliamentary salaries might make you question how obvious this is. And heck, this is definitely not confined to certain African countries. But empirically, this is clearly a hard thing to disentangle. So how do they do it?
The set up of the experiment is where the elegance shows up. First, they allocate two wages: 3750 pesos a month and 5000 pesos a month (about $500), randomly across the sites that were seeking to recruit a total of 350 community development agents. Positions were then advertised and applicants invited to apply via phone or email (but the wage wasn’t revealed to them until they got in touch to register to avoid early sorting).
Then, applicants were invited into a screening process. This consisted of a three hour exam including questions on aptitude (including a Raven test to measure IQ), personality (including the “Big 5” – openness, conscientiousness, extraversion, agreeableness, and neuroticism -- that Francisco and I talked about two weeks ago), motivations (with a focus on the public sector), and job history. (For those interested in how to measure these issues, this section of the paper is well worth reading.)
After going through this process, eligible candidates were grouped into those with normal IQ levels (a Raven score between 7 and 9) and those with high IQ scores (10-12). This gives them four strata of applicants : 1) high wage and high IQ , 2) high wage and normal IQ, 3)low wage and high IQ and, 4) low wage and normal IQ. Each of the 350 vacancies was then randomly assigned to one of these types and applicants were randomly selected to fill these vacancies conditional on type, region of residence and whether or not the applicant was indigenous.
All of this random assignment lets them get at a number of questions of interest to public sector hiring (and labor markets more generally). First off: do you get better applicants when you offer a higher wage? The answer is yes. Applicants in higher wage situations were better in a number of dimensions. First, they had higher previous wages by around 22% (so the market thinks they’re better). Second, they had higher IQ scores by 0.19 standard deviations on the Raven test – and interestingly this seems to be driven both by higher number of high IQ applicants and lower number of lower IQ applicants. Moreover, those offered higher wages were also more likely to score higher on some measures of the Big 5 personality traits.
What about public sector motivation? This is a bit trickier because quality and public sector motivation are possibly correlated – and maybe even deterministically related. And indeed, they find evidence that the pool of applicants in higher wage areas had more indicators of public sector motivation. So while they can’t definitively nail down what’s going on, they can say that the hire wage offer did not lead to any sacrifice in terms of public sector motivation.
The second question they can answer is by how much the higher wage turns into higher acceptance of job offers. Indeed, this is the first randomized setup of the question of the elasticity of labor supply I have seen. So for the folks offered 3750 pesos, 42.9 accepted. The 5000 peso offer got them an additional 35% in terms of acceptances. When they take into account the higher application rate, they end up with an arc-elasticity estimate of the labor supply facing the employer of 2.15 – an estimate that is in line with work using quasi-experimental methods.
Finally, keep in mind that these are the most marginalized communities in Mexico. Consequently they are tough places to work – when you measure this by a human development index or by the number of drug-related deaths. And individuals might have to travel a fair distance to get there. So the final aspect Bo and co-authors look at is how the higher wage may entice applicants to take jobs in the tougher places. Distance has a significant effect on uptake overall: while 80% of those offered a job less than 100 kilometers away took the job regardless of salary, this fraction drops to 25 percent among the low wage group for distances of 200 kilometers or more. However, for the high wage group, acceptance remains around 80 percent for these bigger distances. In terms of violence and lower levels of human development, worse conditions lower the acceptance of the job offer. Again, these are mitigated by the higher wages (although for drug-related deaths, the result is significant at 10 percent).
So in the end, if you pay them more you will get better people. And they’ll take the tough jobs, far from home. It seems to work for civil servants, now we just need to make this work with politicians.