I just finished an intense period of training and evaluation for my project staff, and made final selections for my team of supervisors and enumerators. They are a very talented group, and I’m quite happy with my final team, but getting here was a pretty stressful journey. One reason for the stress is that I used a fairly standard approach of starting with more people than I needed and constantly evaluating them over the first couple of weeks before making cuts.
Another was deciding what to pay people. The Malawi Kwacha was devalued earlier this year, altering the official exchange rate and allegedly changing many prices. In actuality, though, most imports had already been bought nigh-exclusively on the black market, or using foreign currency bought at black market rates, for quite some time, so that the devaluation actually lowered some prices. For example, fuel was almost all black-market before the devaluation, and the going rate was 500 kwacha or more; now it’s at 485 kwacha, and until three days ago it was 441. I consulted extensively with other researchers on how to adjust my pay and in the end I kept it flat relative to what I paid last year. This puts me at a slightly lower pay scale than most local research organizations, which to my thinking is justified due to my lower budget and the fact that I’m selecting people with lower qualifications and less experience (but hope to move them up if they do well).
This means that my enumerators, for example, earn around ten times what they could otherwise make doing piecework labor, assuming such jobs are available. Some people I’ve talked to have argued that foreign researchers are morally obligated to pay even more than that, since the pay for foreign researchers’ Malawian staff is really low relative to our own incomes. On the one hand, I sympathize with that argument – I’m a development economist for a reason! – but on the other, it doesn’t strike me as an unmitigated good to set a really high pay floor for my research team. High incomes are awesome, and you’ll never hear me say otherwise, but high wages have other consequences besides increasing my workers’ incomes.
One such unintended consequence is similar to what you’d see with a binding minimum wage in a competitive market: the more I pay, the fewer workers I can hire. From the perspective of benefits to Malawians, is it better to hire one worker at MK 50,000 per day, or 50 and MK 1,000 per day? There’s actually an answer, at least from an economist’s viewpoint – the latter almost surely maximizes social welfare, since the marginal utility of money is decreasing. If you have no income at all you will likely die, but after you have 49,000 kwacha on a given day, another 1,000 doesn’t matter too much.
Another consequence is that there is a strong tendency for the very best talent in the country to gravitate toward working as a researcher. My UM colleague (and co-advisee under Rebecca Thornton) Susan Godlonton has done some descriptive work on this as background to her research on labor markets here: fully one third of people at the high end of the skill distribution in urban Malawi have worked on research projects. That’s great for the specific Malawians in question – they’re seeking those jobs since doing so maximizes their own well-being – but again, less obviously good for society as a whole. As I noted elsewhere, tons of super-smart, super-educated people who might otherwise start their own businesses and eventually employ their less-educated countrymen are instead trying to get jobs doing stuff like driving cars for the UN or conducting surveys. The more wages for research rise, the stronger this tendency is likely to be.
The other aspect of my choice is a deliberation on how much I’m actually paying my staff, in real terms. The nominal exchange rate is probably misleading, since a dollar goes farther here than it does in the US. But I’m dubious that the official purchasing-power adjustments are right either. I’m going to come back to this in the near future, and discuss how we should think about the value of pay in a developing country.
I agree with your analysis, but I’m curious about one feature of it:
“From the perspective of benefits to Malawians, is it better to hire one worker at MK 50,000 per day, or 50 and MK 1,000 per day? There’s actually an answer, at least from an economist’s viewpoint – the latter almost surely maximizes social welfare, since the marginal utility of money is decreasing.”
As I understood it, this statement would actually be quite controversial to most economists since it relies on an interpersonal utility comparison. (That is, for person A, the move from 0 to 1000 > 49,000 to 50,000, but how can you know how it compares to person B’s move from 0 to 1000? Just to restate what I understand to be the standard formulation.) At least, it would have been controversial – or just outright rejected – starting in the 1930s or so with the ordinalist revolution and the rejection of psychological models of utility, and with them the idea of comparing utilities between individuals. While many economists slip into such comparisons, I had thought that such language was still verboten.
What’s your take on interpersonal utility comparison? Is that part of an “economist’s viewpoint” or an impermissible step into the realm of picking social ends (rather than means)?
I’m personally totally down with interpersonal utility comparisons of certain kinds. I was trying to pick numbers that I thought most economists would obviously agree with, along the lines of giving a dollar to the poorest person in the world or to Warren Buffett.
While I definitely recall being taught comparing utility across people is verboten at some point (maybe in undergrad?) comparing utility across people (or just adding it up) was a fairly common feature of my coursework at UM. Now, maybe the escape hatch is that in e.g. first-year macro we were using representative agent models where everyone was identical* – but I think my example of arbitrary Malawians fits right into that framework. I’m picking some number of people, fairly arbitrarily, to receive an even share of a fixed payroll budget.
You might be right about there being a stigma against openly comparing utility across people, but I’d venture that in their heart of hearts most economists would unambiguously prefer 1000 going to 50 people over 50,000 to 1, ceteris paribus. Teasing out their actual thoughts is however likely to be even harder than the kind of data collection I usually attempt.
*Another excuse for doing the addition thing is that sometimes were were just looking for Pareto optima.
I agree that most economists would prefer it, but I think when being disciplined, they would argue that this represents a normative preference for equality that should be separated out from positive analysis. What was fascinating to me, though, is that you precisely invoked the economist’s “viewpoint” for this sort of normative analysis – which again, I agree with, and bet most economists would too. What this says to me is that economists are, pragmatically, cardinalists not ordinalists, and that they recognize that ordinalism places extreme restrictions on the ability of economics to say anything of use. But then we’re back to the puzzle of why economists still claim in so many settings to be a strictly positive science – and how y’all get away with that, so to speak.
Agreed on all counts. And I’ll go further – we’re clever at fooling people, even ourselves, into confusing the normative and positive aspects of the discipline. Think about the third week or so of Econ 101, when we smoothly transition from comparative statics into measuring the area between a supply and demand curve and calling it social welfare.
Any economist who claims we’re practicing a strictly positive science is lying. It’s possible to do strictly positive things, and lots of research papers limit themselves to that – use an IV to estimate an elasticity, stuff like that. But we’re all trained to add up welfare triangles, and it becomes part of our instincts to do inter-personal comparisons.
Partially because I have long wanted to mention this and partially because it is relevant to the discussion of pay standards on development research projects:
Last year, the PI on my project tried to pull off a pay scale for the data entry team that looked something like this – x MK / day for entering up to y questionnaires (double blind, fully checked, corrected, & scanned) and then a bonus of z MK / day for each additional questionnaire that was fully entered. The theory (as supported by a successful trial run during a similar project in Kenya) was that this would help the data entry go faster by rewarding the faster data entry teams for their above-average work. However, shit hit the fan when we attempted to introduce the new system.
Instead of seeing the variable pay-scale as a bonus for speedy and accurate work, they saw two major flaws: first, that they would receive more money and have work for longer if they only completed the bare minimum and dragged out the project. Essentially, they observed that the minimum entry rate was optimal because clerks were guaranteed per-diem for the duration of their employment and because z x/y, most teams were only able to complete an additional 1-2 questionnaires over the minimum requirement and the bonus was so insubstantial as to be insignificant over the duration of the project – perhaps amounting to an additional $40 at the upper end.
From this whole episode, I learned an interesting lesson about pay structures and the importance of the manner by which international organizations communicate such information with their staff in developing countries. For example, things might have gone more smoothly if the clerks had just been told that they would receive some bonus (in addition to their unconditional regular wage & per diem) for working quickly and efficiently during the day. When is it a good idea to unpack the black box of pay-rates and when is it better to keep the lid firmly on?
Hey Jason,
Nice blog, by the way. And thanks for linking to mine… looks like we are are having very similar experiences. Two of my enumerators are college-educated and have some survey experience as well, so my flat rate market wage does not exactly allow them to retire after this project.
Anyway, I think the biggest under-reported consequence of donor funded research salaries is the externality. Many researchers (including dev economists) just don’t think about what will happen after they leave, they are thinking about managing the next six months of data collection efficiently. When it comes down to it, paying a lower salary in the short run, so that the externality of high expectations on donor-funded research projects is avoided in the long run (and thus creating more of a private-sector incentive) is worth it. I just hope that more researchers think about the consequences of their work in the long-run.