The Devil is in the Details

That is the title of a recent paper in the Journal of Development Economics (NBER version here, 2013 ungated version here), and although the piece does not feel dramatic at first it is one of my favorite articles of the year.  It pins down some critical features of economic underdevelopment better than any study I know.  The subtitle, by the way, is “The Successes and Limitations of Bureaucratic Reform in India,” the authors are Iqbal Dhaliwal and Rema Hanna, and the work is set in rural Karnataka.

It is not easy to excerpt from, so I will summarize the narrative:

1. Using biometric technology — thumbprints — to monitor absenteeism induces staff attendance for public health workers to rise by almost 15 percent.

2. That in turn leads to a reduction in low-birth weight babies.

3. Yet the government proved not so interested in monitoring attendance on a more regular basis, not even to enforce their pre-existing human resource policies.  Potential penalties against late or absent doctors were not, for the most part, enforced.

4. Following the implementation of monitoring, the doctors showed the least improvement in attendance of all the workers, in fact virtually no improvement.  The entire positive effect came from nurses, lab technicians, and lower level staff.

5. The government was reluctant to continue the monitoring because it feared staff attrition and staff discord, especially from the doctors.  There is growing private sector demand for doctors, and many doctors are considering leaving these clinics for superior pay elsewhere, and perhaps also superior location.  Therefore the doctors are given, de facto, a very lenient absence and lateness policy, in lieu of a pay hike.

6. It is already the case that many of these doctors moonlight on the side, or have separate private practices, and that spending more time at the public clinic is not their major priority.

7. It is not easy for the underfunded local government to pay these doctors more, and thus a high level of lateness and absenteeism continues.  I wonder also what would be the morale costs on the non-doctors, if the monitoring were to be continued to be enforced in this differential manner over a longer period of time.


I think perhaps that "absenteeism" is not a bug but a feature: the doctors/staff consider this a form of 'non-monetary compensation' since they think they would make more in the private sector. We who have worked in the private sector know that's not true, but that's their thinking. I'm pretty sure I read somewhere the most incompetent doctors end up in the USA working for the state (the Veterans Administration for example).

Who would work for government in the US other than people with no other options?

Here in tax cut fever NH, the number of prison guards is down 25% since 2000 while the number of prisoners is up 8%. A large number of positions are open, but very few sane and honest and fit people are qualified from those who apply. All the neighboring States and the Feds pay higher wages and benefits with better working conditions.

Thus NH prison guards are mandated to work overtime in the range of 60-70 hours per week.

I'm sure there are a few conservatives who want to privatize so the private operators can hire psychopaths and dishonest prison guards. Which is ironic since private prison operators accept only low risk prisoners, ie drug use related crimes, so the private prison guards make a lot of money and trade dealing drugs to prisoners.

Of course, private prisons produce paroles who quickly return to prison so private prisons increase costs by increasing crime and prisoners.

Conservatives seem to reject Deming, believing that the best results happen by first cutting costs. The evidence of higher costs never deters the cost cutters, thus the ongoing war on drugs.

Unlike their close relatives, cops, prison guards are seldom the subjects of exciting and interesting television fare so the young are unaware of the many rewards of the profession. Since US incarceration rates are high and large numbers of prison guards are required, it would seem to make sense that prison administration courses be part of the high school curriculum instead of obsolete subjects like home economics and civics.

One way to make money in the private sector is to be good at something that people want. Sorry about your experience.

Its sad that those in power actually have to start implementing unrealistic rules in order for them to confront reality. This is a very old story. Why it is such a struggle to pay attention to the contexts within which "problem" behaviours are occuring is the real mystery, not how to stop them.

I don't see a mystery. Blame shifting and unwillingness to confront tough problems are human nature.

If we think about the doctors as setting norms for the rest of the group. And, we think those norms have strong effects on outcomes. Then might it not be a good move to reduce the headcount of doctors, raise the pay of the remaining doctors and then institute monitoring. Indeed, the last step might not be needed if headcount is reduced enough to fund an effeciency wage.

Or just change the hours the doctors are officially expected to work down to the current average, and then start enforcing the new rule. The doctors you lose at that point are probably mostly dead weight. Granted, you'll have some doctors who were putting in more hours now working less, but the new enforcement will ensure that this is offset by doctors who were previously working less either work more or leave.

Ok, I should have read the Thomas Sewell post first.

Step one of removing corruption is sunlight as disinfectant.

Recognize the current situation. i.e. officially change attendance policies to match what the staff actually does on average (you can take off X hours a week without per-arrangement, or whatever) and institute the biometric monitoring permanently at the same time.

Then when you have the resources to pay the staff more, you can make enforceable agreements to do that in exchange for spending more time working. In the mean time, you've moved the benefit to all staff on average, rather than incentivizing the staff who are most willing to take advantage of you by lying about showing up for work or not and penalizing the more honest staff.

Doctors see this as humiliation, they resist, doctors are supply limited high status individuals. Also doctors are in effect the senior managers on site, a whole topic in itself. When the boss is ritually humiliated, which this effectively is, authority is undercut, satisfaction declines etc...

My experience says that firms that impose such things on senior managers have a very high attrition of said managers.

My suggestion require this monitoring for all staff, with a set of tiers of rigor. Cleaners get time clocks, technicians, and aides get this button thing, put unregistered nurses over them and require them to give status reports. Then make RNs register simple reports on shifts that are set intervals which will provide same information to a central system, you can even require a thumb print. Senior Nurses can be required to log status reports on more junior staff at regular intervals, and so forth. Doctors, pharmacists, etc... can be given some sort of monitoring function over senior nurses and PAs. The trick is to give each status step a noticable reduction in humiliation while still keeping them under supervision. This can be easily done with the justification that says we trust you but don't trust your juniors.

If you want to be especially thorough you can give senior nurses a way of reporting on doctors attendance. And if you ever get a case of collusion between doctors and senior nurses things are either so bad you are screwed anyway or you have gotten very lucky and found a team that will probably be far more efficient

This sort of human engineering is why we have squads, platoons, companies, battalions, etc... with large divides between certain steps, soldier-NCO, petty officer-Chief, Chief-WO-Jr commissioned officers, and so forth in armies and navies. Each of these steps includes seperate facilities, reduced fraternization, signifigant uniform differences, etc...

Biometrics? Attendance at work has been monitored through brass tokens or time cards for a long time. How are thumbprints an improvement?

One worker can't punch clock for another.

Or cook numbers

In fact, it is not hard to reproduce a thumbprint. It is not so hard to steal someone else's print from a glass.

What exactly is the problem? The government fears attrition of doctors; but the entire positive effect came from nurses and other low-level staff. Doctors appear to be irrelevant to the problem of low birth-weight.

The solution therefore is to enforce biometric timekeeping, but ignore doctors' transgressions (unless they're really bad).

I recall the controversy in America over the use of a time clock. As for physicians in India, Cowen has addressed that in prior blog posts, in particular the discrepancy in what physicians are paid in the public sector and what patients pay (and it is mostly a cash pay system in the private sector) them in their private practices. Of course, the reason for the enormous disparity in pay is the enormous inequality in India. India's conundrum is an exaggerated version of the conundrum in America (think Medicaid pay for physicians vs. the cash pay to a cosmetic surgeon). Here's something an economist might study: studies have revealed that fewer Uber drivers are available during peak demand when their rates go up (the justification for higher rates at peak times is that there will be more drivers available), the reason being that Uber drivers set a target for how much income they wish to make in a given day, and once they reach the target, they quit; with higher peak rates, they reach the target sooner. If that phenomenon applies to Uber drivers, I suspect it applies to physicians. Thus, a physician specializing in Medicaid patients will see many more patients and (this is the kicker) order many more diagnostics, etc. than the physician specializing in cosmetic surgery. In other words, the low reimbursement rate for Medicaid induces physicians to waste more resources: raise the reimbursement rate, lower the waste. Of course, it could be that Uber drivers and physicians come from different planets, but I suspect not.

The better example would be Medicare (as opposed to Medicaid), but I didn't want to shine too bright a light on Medicare, given my age and all.

Is this a parable about illegal immigration and farm labor?

Yes, it is all well and good to enforce rules, but make sure you have rules suited to your reality.

The relatively sub micro nature of these findings contrast strongly with the post by Alex on MR today about self-employment and constraints on big firms. These industry level constraints are not easily amenable to RCTs yet impose orders of magnitude larger effects on hurting Indian development than the kinds of issues that the RCTs discuss.

Remove many of the barriers to competition to trade between regions in India, allow more foreign firms to enter, increase the possibility for growth of large firms, soften the role of intrusive commercial taxation, simplify and lower tariff barriers, lessen the effects of regulatory land reform, and limit regulation. These effects would dwarf those of all the policies driven by RCT findings dramatically. But politically, it seems they are all effectively out of bounds. And worse, they can't be RCT'd so the elite schools don't do so much research on them today.

Fifteen percent?! That's 3 days a month! Presumably in addition to regular vacation time? Insane.

The problems is that the true 'principal' is not the State Govt. because, in Karnataka, it is the local politicians who have salience. A bureaucrat who takes action against a delinquent Doctor will get transferred.
Kerala's health system previously had much more accountability to the community but then Kerala had some distinctive features- e.g. high education and assertiveness.
The rise of a rural middle class makes it more difficult to have top-down mechanisms. The Doctor may belong to an influential farming family with influence in the State Legislature.

That is why India needs to replace Public Sector entitlements with cash transfers- preferably through Ebanking. If people can pay for their treatment, demand will create its own supply. Moreover, resources currently improperly utilized by the State can be released in a productive fashion. In education, we see that continued support for failing State Schools means that they aren't enough qualified teachers, or proper buildings and other facilities, for the Private Sector. Even if India can muddle through on Education, there are more serious information asymmetries in Health.

We are getting to a very dangerous situation because of over-prescription of anti-biotics, improper use of medical technologies- e.g problem of female foeticide- leakage of prescription medication into the drug culture which in turn creates demand for addiction clinics- one can multiply such instances. Side by side with private provision there has to be robust Govt. action. Muddling through is not an option.

Comments for this post are closed