Why peformance pay is difficult to implement in social sectors?

Conventional wisdom would have it that performance-based pay is the most effective strategy to improve outcomes in any organization. Corporate sector payouts accordingly have two parts, a fixed salary and a variable performance-linked bonus. Can the same strategy succeed in public sector bureaucracies like education, health care, and other regular government departments?

As I have already blogged about it here and here, I am sceptical of its success when implemented in scale in a country like India. Here are a few reasons why performance-based pay can run into problems in real-world implementation on scale.

1. Quantification and measurement problems

The fundamental pre-requisite for any credible performance-based pay system is its ability to quantify and measure outcomes. To start with, there is a strong case that not all dimensions of performance in social sectors can be quantified. Complicating any measure of quantifiable outcome is the role of widely varying exogenous factors like social and economic background of student/patient/customer, family environment, historical and legacy factors etc, all of which exert considerable influence on the final outcome. Furthermore, it is important to have some level of broad consensus about the variables used to measure performance outcomes and the actual measurement process itself. Such consensus is rarely forthcoming.

For example, how do we reliably measure learning outcomes among students? Even assuming some level of consensus among all stakeholders about what should be measured, what are the instruments available for its measurement? How do we ensure that instruments like examinations are not subverted when implemented in scale? How do we capture the personal interest shown by certain teachers, that encourages parents to send children in larger numbers? These are questions for which there may not be answers which have some reasonable level of acceptability.

2. Monitoring problem

Once there is a broad acceptability of the performance measure and the process of extracting it, its credibility also depends on the rigor of monitoring. Too much rigour in the measurement and data collection process also runs up against the problem of cost-effectiveness. It is natural that any such measurement system, especially if deployed for not-so-low stakes decisions, is bound to attract attempts at subversion.

Even assuming the availability of a credible enough performance measurement framework, like in case of maternal and child health interventions, the challenge of reliable data collection remains. How do we ensure that the ANM is not reporting inaccurate figures? Super-checks and sample validations by third party agencies, while useful, may not be credible enough when done on a state-wide or national scale.

3. How much bonus is appropriate?

A bonus is effective only when its magnitude is beyond a particular threshold. Too low a bonus fails to evoke the desired or even any performance response. This is more likely given the already high salary levels among government employees. However, too high a bonus, apart from being not cost-effective, also distorts incentives by raising stakes. Complicating matters, the optimal bonus level varies both across sectors and with each sector from place to place. For example, what works in education may not be the same in health care. In education itself, performance bonuses may differ from primary to secondary school, and from one geographical area to another.

Given all these factors, calculating the bonus with any reasonable level of accuracy, becomes a very complicated task. Further, if there are too many bonus slabs, then that creates another set of dynamics.

4. Cultural socialization

Any performance-based payment initiative has to be institutionalized both administratively and culturally. But the latter is difficult to achieve when implemented in scale in countries with wide social and cultural diversity.

In fact, its widespread cultural or social acceptance as an incentive to reward good performance underpins the success of the administrative implementation. Such consensus plays the critical role of creating the stakeholder pressures and institutional vigilance that are vital to ensuring that forces to subvert the system from within are foiled.

In the absence of socialization about its benefits among the stakeholders, willy-nilly subversive tendencies creep in. Once a significant share of the employees have partaken of bonuses at some time or the other, it is only a matter of time before they come to view these bonuses as entitlements and unions enter the fray.

5. Maintaining credibility

The success of such initiatives is, to a large measure, dependent on its acceptability among the large portion of its stakeholders. This credibility rests on tenuous foundations. It is easily shaken by a few jolts, which most often ends up giving a convenient excuse for opponents to question its reliability.

Once a few lapses get highlighted, especially high-profile ones, a downward spiral is never far away. Loss of credibility inevitably follows. It suits the vested interests to publicise such shortcomings to add credence to their opposition.

6. Political constraints

Any government service delivery channel, especially in democracies, is embedded in a political system. It is therefore natural that the processes and administering stakeholders are exposed to political dynamics. The most visible manifestation of this are trade unions. There are far too many areas where political considerations can take precedence in the conceptualization, implementation and sustainability of performance-based pay initiatives. Once such considerations creep in, it dilutes the program's objectivity and raises hackles among rival political and social groups, thereby denting the credibility of the process.

In fact, it should be sine-qua-non that all new public policy initiatives are analyzed in terms of political constraints to see the practicality of its implementation and examine whether there are strategies to overcome the identified political hurdles. A program that appears logically sound, but fails this test may not be worth pursuing.

7. Scale dynamics

Much, if not all, of the evidence of success with performance-based pay in social sectors comes from small scale experiments. Such studies fail to account for the dynamics that emerge once a performance-based pay program is scaled up. Apart from the logistical exercise of managing the collection of reliable data, there are also issues arising from socio-political factors. In fact, all the aforementioned factors have the potential to manifest in the most unexpected manner when the initiative is implemented in scale.

The administrative challenges are the most formidable. For example, how do we address the problem of teachers and administrators in a remote village or taluk colluding to subvert both the measurement and its collection. And imagine the problem when there are a large number of such taluks.

It is anybody's guess as to how these factors interact with each other and contribute to the emergent system. But most often the emergent dynamics are detrimental to the sustainability of the initiative.

In conclusion, I am inclined to believe that any performance-based pay system foir government officials, while unobjectionable at a theoretical level, may be very difficult to implement, most certainly for political and administrative reasons, in the prevailing environment in countries like India. While it may succeed in a limited area, scope and time, it may not yield the desired results with a more ambitious scope and pan-Indian area of implementation.

Observations on the Aarogyasri program

Aarogyasri is a hugely popular health insurance program initiated the Government of Andhra Pradesh. Administered by a government-run Aarogyasri Trust, it covers all the below poverty line (BPL) citizens, and provides for pretty much the entire spectrum of high-value tertiary treatments. In the language of insurance, the Aarogyasri is a single-payer (government), mandatory coverage (for all BPL families), pure community rated (same insurance rate for all those covered) insurance scheme.

Its supporters point to four features of the program as proof of its widespread appeal. One, it covers all the major medical conditions, with a generous coverage of upto Rs 2 lakh per family every year. Two, it provides un-paralleled choice to patients, giving them the freedom to choose any hospital, government or private, for their treatment. Three, it provides for completely cashless treatment in any of the empaneled hospitals. Four, the scheme incentivizes government doctors by earmarking a share of the payments recieved by their hospital for treating Aarogyasri cases to the doctors and staff.

However, it is precisely these four attractions that form the basis of concerns about its long-term sustainability.

1. The universal coverage is a red herring. In reality, the supply-side is severely constricted by the available treatment facilities. In fact, even with the spurt of private hospitals in the wake of the program, less than a quarter of patients suffering from a covered medical condition are likely to be treated under the scheme.

Herein lies one of the biggest challenge for the scheme. If the present trend continues, more private hospitals will crop up, if only to exclusively service patients covered by the scheme. This will in turn increase the available treatment facilities and thereby the actual claims processed by the insurer. It is inevitable that premiums will keep going up for years to come, merely due to the addition of new treatment facilities.

As the numbers of private hospitals increase, there will also be increased pressure to expand the pool of covered procedures. This too will drive premiums north. Adding to all this will be the universal trend of rapidly increasing medical treatment costs. Will the government budget prove deep and resilient enough to meet all these upward pressures?

2. The level of patient choice in Aarogyasri is simply unprecedented, a luxury not available to even patients in many developed economies. Given the state of government hospitals and the incentives of private and government hospitals (the former have no incentive to chase patients), patients are more or less certain to prefer the former. This would be a shame since most government secondary and tertiary care hospitals have well qualified doctors and adequate diagnostic and surgical devices, though the quality of service delivery is questionable. Questions will invariably have to be asked about whether it is possible to leverage the Aarogyasri program to improve the quality of service delivery in government hospitals.

3. Related to the previous point, the prevailing government policy on secondary and tertiary healthcare provides for no synergy between the government's own single-payer Aarogyasri health insurance program and its existing secondary and tertiary care facilities. In fact, they are each considered distinct and mutually exclusive. This is unlike the health insurance model in most western countries, where there are strict protocols for referrals, with cases being referred to private hospitals only when government hospitals are unavailable.

An application of the same model would have brought in the government hospitals as a major health service providers in the Aarogyasri scheme through a similar protocols-based sharing of cases between them and private hospitals. It would also have enabled resource-strapped Government hospitals to access payments from the Aarogyasri program. This cash flow becomes all the more important since the state government reduced its budgetary allocation to all these hospitals in lieu of the Aarogyasri allotment. In simple terms, the budgetary allocations to Aarogyasri and existing government hospitals being a near zero-sum game (net allocation being more or less the same), the private hospitals benefitted at the cost of the government hospitals.

4. Further, once the patient is admitted by the private hospital, given the pay-per-intervention payment system, their incentives are strongly aligned towards over-treatment. Since the treatment is cashless, the incentives of the patient are aligned towards accepting the "best" available treatment. Unfortunately, in the prevailing model, the incentives of the doctors are aligned towards projecting expensive invasive surgical procedures as the "best" option. For example, irrespective of the medical condition and the age profile of the patient, irradiation therapies are generally preferred (by both doctors and patients) over medication. In simple terms, the most aggressive treatments have become the standard of healthcare.

In standard insurance schemes, insurers have to keep a strict vigil on the pre-authorization process (when the tests are done and the patient is screened for a particular surgery/therapy) so as to minimize over-treatment. This is all the more so since the payments to health service providers (doctors and hospitals) are on a pay-per-procedure/intervention basis, as against the less distortionary fixed payment for treatment of a medical condition.

The Aarogyasri program too makes payments to hospitals based on a pay-per-procedure basis. In fact, the tender premiums quoted by the insurers are based on this premise. The Trust prefers this approach since it believes that its in-house pre-authorization process is rigorous enough to effectively screen patients and prevent over-treatment. In fact, effective pre-authorization is the forte of the best Third Party Administrators (TPAs) hired by the insurers. If the Aarogyasri Trust does this effectively, then it has to be counted among the most effective TPAs. In any case, as the program expands, maintaining such rigorous pre-authorization process will become difficult.

However, unless it moves away from the in-house pre-authorization process to a purer insurance model, it may not be possible to change the payment model. A medical condition based payment approach is much more complex to administer and riskier too and may not be possible with an in-house model of pre-authorization.

5. In simple terms, the incentives under the Aarogyasri scheme offered a cash reward top-up to doctors for doing much the same procedures which they were doing through their regular hospital in-patient channel. This has the potential to create a moral hazard - the doctors who internalize the incentive and do these procedures come to slowly view these incentives as entitlements.

This turn of events can damagingly distort the incentives facing doctors, especially if at some point in time the government decides to abandon Aarogyasri and decides to revert back to the old model of government institutions based health care. Further, it cannot be denied that atleast some doctors are likely to be disincentivized in taking proper care of patients not covered by Aarogyasri. Also, what about the cash incentive crowding out intrinsic motivation?

Aarogyasri incentive structuring is a powerful example of the need to exercise great caution when we introduce performance-based pay systems into government bureaucracies. Unless carefully structured, cash incentives not only distorts the current implementation, but it also generates adverse expectations which come in the way of future implementation of performance based pay. In some ways, this is similar to a situation where a doctor abruptly replaces a commonplace but effective drug with a powerful new medication against a particular virus/bacteria, only to find after some time that the second generation drug too is losing sting, leaving us with limited available options to effectively treat the microbe.

So what can be done to make the Aarogyasri program more cost-effective without radically tinkering with its existing model?

For a start, it is imperative that there be a clear protocols-based system of referrals, so that the existing government facilities are more closely integrated into the Aarogyasri scheme. The government hospitals benefit by way of accessing more funds and thereby better diagnostic and surgical facilities. It will also help the government accommodate the massive budgetary support that is inevitable in the coming years as the scheme grows.

A treatment facility wise mapping of government hospitals can help route Aarogyasri patients to those hospitals for specific medical conditions. Only those cases which cannot be treated in these hospitals (for either lack of bed space or lack of required facilities) should be referred to private hospitals. Simultaneously, there should be a vigorous campaign to improve service delivery standards in secondary and tertiary hospitals.

The incentive system for government doctors provided for under the Aarogyasri scheme has to be either dismantled or be made more nuanced. If the later is preferred, the incentives should kick-in only after a certain performance benchmark is breached.

Under the Aarogyasri scheme, the insurance premium quoted by the insurer is a function of the number of procedures/therapies covered, N, the respective price (to be paid to the hospital) fixed for each surgery/therapy (or medical condition) i, Pi, the number of empaneled hospitals (or number of available treatment beds for each surgery/therapy i), Ei, and the disease incidence risk among the population pool insured for each medical condition i, Ri.

In other words, Premium, Pr = f(N)+g(Pi)+h(Ei)+q(Ri)

Insurers seek to ensure that their expenditure due to claims and administration costs is lower than the premiums collected.

Of these, the most important parameter is the prices of procedures. Neither the insurer nor the health service providers have an incentive to control it. The health service providers are the direct beneficiaries of higher procedure rates and therefore lobby hard for maximizing procedure prices. The insurers merely pass on these higher prices on to the consumers by way of higher premiums.

The insurer seeks to minimize his claim outgo either by limiting the number of empaneled hospitals (so that the numbers of cases that can be treated is controlled) or turning away (on some pretext or other) those who claim treatment. Both these problems can be addressed. The former can be mitigated by defining the list of empaneled hospitals in the tender itself, including those which are likley to be added each year and details of when they will become operational. Since the premiums are revised each year and it takes atleast an year for establishing any hospital, such up-front disclosure is not likely to create any problems. The later can be overcome by making it mandatory to treat all the patients pre-authorized by the Aarogyasri Trust.

Both the aforementioned conditions, coupled with upfront disclosure of number of surgeries/therapies, transparent fixation of prices for each procedure, and government-run pre-authorization can substantially align the incentives of all parties. If these conditions are fulfilled, the insurer's bid would be determined purely based on his actuarial risk calculation for the insured risk pool and their administration costs. Such bids are more likely to generate efficient outcomes, since it increases the likelihood of the successful bidder also being the most efficient insurer.

Performance-based pay for teachers

Econ 101 teaches that "incentives matter", and there are numerous real-world examples of its application, none more so than with structured cash-incentives. So why not introduce performance-based pay for teachers?

A recent study of country-level performance-pay measures from the PISA-2003 (does not include India) international achievement micro data by Ludger Woessmann of the University of Munich estimated student-level international education production functions. After controlling for various cross- and within-country biases and other salary adjustments, he finds that math, science, and reading achievement scores in countries with performance-related pay are about one quarter standard deviations higher. He writes,

"The results of cross-country education production functions that extensively control for student, school, and country background factors suggest that students in countries that make use of teacher performance pay perform significantly better in math, science, and reading than students in countries that do not use teacher performance pay. The size of the association between use of performance-related pay and student achievement is about one quarter of a standard deviation on the math and reading tests and about 15 percent of a standard deviation on the science test.

Given the well-established nature of teacher bonuses in many countries, these associations are likely to capture general-equilibrium effects of performance-related pay which are likely to combine long-term incentive effects of motivating current teachers with sorting effects of changing the pool of applicants for the teaching profession."

I had blogged earlier about a study by Prof Karthik Muralidharan about evaluation of incentives-based teacher performance in 500 government primary schools in Andhra Pradesh which found that "there could be significant gains from moving to a system of hiring teachers on fixed-term contracts and then using performance measures to pay bonuses on an ongoing basis and to inform the tenure decision after a longer period of performance measurement".

I am inclined to believe that any teacher performance-based pay system, while unobjectionable at a theoretical level, may be very difficult to implement, both for political and administrative reasons, in the prevailing environment in countries like India. While it may succeed in a limited area and time, it may not yield the desired results with a more ambitious scope and pan-Indian area of implementation.

So, a "good teacher" it is!

One of the most frustrating deficiencies in the education sector, at the pre-matriculation and more so at primary school level, is the virtual absence of reliable measures of learning outcomes. In many respects, education stands alone as the only sector without any objective yardstick whatsoever for evaluating the value-addition from the massive resources poured in.

Alone among all professionals, teachers are not subjected to any objective and reliable, direct or indirect, measure of performance evaluation. In every other sector, there are atleast some standard measures of evaluation, though it is a different matter that they are rarely used in any meaningful manner.

Surely, any system where people wake up to suddenly discover that a significant number of people who have been exposed to it for nine years are unable to pass the most basic of examinations and are, to be charitable, semi-literate, or a major share of its products are downright unemployable, deserves to be completely revamped.

In this context, Freakonomics points attention to an analysis of seven years of Los Angeles state standardized test-score data in Math and English from 6,000 state teachers by the L.A. Times and the Rand Corp., which finds that teacher effectiveness is three times more influential than school attendance on student performance.

The study using longitudinal student-level achievement data found that greater variations existed in the quality of teachers within each school than between schools in affluent and poorer neighborhoods. It found that highly effective teachers, the ones who consistently and dramatically raise their students' scores, are fairly evenly distributed among schools and across different levels of experience and education. Strikingly, it found that after a single year with teachers who ranked in the top 10% in effectiveness, students scored an average of 17 percentile points higher in English and 25 points higher in math than students whose teachers ranked in the bottom 10%.

Put differently, though parents obsess with picking the right school for their child, it matters far more which teacher the child gets. Yet parents have no access to objective information about individual instructors, and they often have little say in which teacher their child gets. Further, contrary to widespread belief, many of the commonly assumed factors responsible for improving teachers' effectiveness - experience, education and training - had little bearing on improving students' performance. Most interestingly, the students' race, wealth, English proficiency or previous achievement level played little role in whether their teacher was effective.

In fact, they also find that the commonest distinguishing characteristics of effective teachers were a tendency to be strict, maintainance of high standards, encouragement of critical thinking, and the engagement of his or her students. See an FAQ on the methodology adopted here.

The graphic compares the contrasting performances (in terms of raising the percentile of students able to do specified level of Math and reading) of two teachers teaching the same lessons at two different fifth grade classes at the same school



Another graphic demonstrates how the difference between the student's expected growth (each student's past test performance is used to project his performance in the future) and actual performance is the value a teacher adds or subtracts during the period. The projection based on past performance means that no teacher is hampered by the presence of low-performing students. The value-added compares students to themselves in previous years, rather than to other students with different backgrounds. For all the aforementioned reasons, this methodology can therefore be used for longitudinal tracking of students and teacher value-addition.



The study finds that many important teacher qualifications have little effect on student outcomes and "more experienced or better educated teachers are no more effective in the classroom than inexperienced teachers with only undergraduate diplomas".

It draws attention on the need to "focus on measuring teacher skills and preparation that predict subsequent teacher performance in the classroom". They write,

"Districts could consider developing policies that place importance on output measures of teacher performance. Current policies emphasize teacher qualifications that are inputs to student learning. These inputs are costly to produce and sustain in terms of hiring and salary costs, but they have little consequence on student achievement outcomes. A better approach would be to incorporate value-added measures of teacher effectiveness into teacher assessments. Teachers and administrators should have access to value-added measures of teaching effectiveness.

These measures would provide useful feedback for teachers on their performance and for administrators in comparing teacher effectiveness. Merit pay systems would realign teaching incentives by directly linking teacher pay with classroom performance. Merit pay is 'results oriented' in the sense that compensation focuses on the production of specific student outcomes. The challenge for designing a merit pay system for teachers is in defining an appropriate composite of student learning (output) and in measuring teacher performance in producing learning...

We find that teachers with better nominal teaching tools (e.g., experience, education, licensure scores) perform no better than teachers with weaker qualifications, but the current system provides little reward for better classroom performance. Perhaps teachers with extra teachings skills have too little incentive to fully utilize those skills in a compensation system that rewards their measured inputs and ignores their outputs. By realigning the incentive system and rewarding student achievement gains, we might find a different ordering of teacher effectiveness and improved overall levels of student learning."

Here are a few observations

1. The critical challenge will be in the administration of the standardized tests. How do we manage the logistics of standardized examinations, given the wide geographical spread and massive numbers of students being tested? How do we ensure the purity of both the examination invigilation and paper valuation? In other words, how do we ensure the administration of the massive exercise of standardized tests without compromising on the purity of its results?

One way would be to outsource the process itself. However, its cost and more importantly, the perception and resultant salience of an externally outsourced assessment process will amplify opposition from the unions. Administering it through internal arrangements, for example by shuffling teachers across schools, too will raise formidable administrative and supervisory challenges. However, in the initial stages, this appears to stand the best chance of success.

2. A perception that such value-addition analysis would be used to assess teachers will naturally raise political opposition from the unions. It may therefore be necessary to completely de-link its use from high stakes decisions like punishing teachers.

In fact, mere disclosure of teacher-wise value-addition for each student and the entire class, will go a long way in contributing towards increasing performance outcomes. Appropriately designed student report-cards aimed at parents, teacher-report cards aimed at administrators, and school-report cards intended for community at large, can play an important role in getting all stakeholders to respond in a manner that will nudge teachers to improving their performance.

3. In order to buy acceptance among teachers, such value-addition analysis should be spun-off as say, "teacher enhancement feedback programs". Analysis of classroom data and student learning outcomes can be used to deduce specific skill-deficiencies of teachers. This can in turn be used to objectively design training programs and impart focused trainings to teachers based on their respective deficiencies.

4. The biggest source of last-mile challenge will be in ensuring that the data collected, analyzed and presented is acted upon. It is commonplace in government to have massive data being collected and not being utilized in any meaningful manner. And the sheer volume of longitudinal data collected only increases the probability of policy-making getting buried in the small detail of numbers.

As aforementioned, this last-mile problem can be overcome with effectively designed and institutionalized policies that uses the data to simultaneously inform parents about their students' performance, administrators about the respective value-addition (and value-subtraction) of teachers and performance of schools, and teachers about where they and their students are lagging behind.

This information disseminated in the most cognitively effective manner (well designed report cards), through platforms like school management committees, and utilized to design training programs for teachers and remedial classes for students, can go a long way towards improving the quality of our education system.

Update 1 (7/9/2010)

See this collection of LA Times stories on the teacher value-addition study. And this, this, and this from NY Times.

See this website of SAS EVAAS, the most comprehensive reporting package of value-added metrics available in the educational market, which provides valuable diagnostic information about past practices and reports on students’ predicted success probabilities at numerous academic milestones.

Update 2 (23/10/2011)

A study of New York City schools by Jonah E. Rockoff and Cecilia Speroni explored the power of objective (student achievement data) and subjective (evaluations from both applicant interviews for a certification program and mentors who worked with teachers their first year) measures of teacher evaluations finds considerable merit in the later. They write,

"We find evidence that teachers who receive better subjective evaluations of teaching ability prior to hire or in their first year of teaching also produce greater gains in achievement, on average, with their future students. Consistent with prior research, our results support the idea that teachers who produce greater achievement gains in the first year of their careers also produce greater gains, on average, in future years with different students. More importantly, subjective evaluations present significant and meaningful information about a teacher’s future success in raising student achievement even conditional on objective data on first year performance. This is an especially noteworthy finding, considering that variation in subjective evaluations likely also captures facets of teaching skill that may affect outcomes not captured by standardized tests."

As Freakonomics writes, "Among the many knocks on the new push for objective evaluation measures is that they fail to capture the nuances of teaching, which the authors believe traditional subjective methods do much better."

Incentivizing school attendance and performance

I have argued earlier here and here that conditional cash transfer (CCT) is a more economically efficient way of transferring welfare benefits to the poor. The latest post of Gary Becker and Richard Posner debate on paying children to attend school.

As Prof Posner writes, Milton Friedman was one of the earliest proponents of direct cash transfers to replace welfare programs. His contention being that people have a better sense of their needs than government bureaucrats, so that if the government simply gave poor people money they would allocate it more efficiently than the welfare bureaucracy would do. This is the philosophy underlying the US Government's program of Earned Income Tax Credit. Any moral hazard concerns can be taken care by making the cash transfers conditional to the recipients achieving certain pre-specified outcomes.

The Mexican Government's Progressa (and Oportunidades) program of mid-nineties sought to reduce child labour and improve school enrollment rates by paying poor parents to keep their children in school and to take them for regular health check-ups. It was argued that if the children remain in school and performs well instead of going to work, the families could be compensated for the loss in their children's earnings by direct cash transfers. Studies by economists in the United States and elsewhere clearly show that Progressa has succeeded in inducing the mainly rural parents in the program to keep their children in school longer than they would have.

Taking cue from the Progressa experience, private foundations and individuals have started experimental programs in New York (New York City Opportunity program) and few other American cities that directly pay poorly-performing, older children (and not parents) to incentivize them to remain in school and improve their performance.

As Prof Becker writes, "Rewarding these poor students for better performance is similar to the tuition scholarships and stipends that colleges award to students with good grades. To earn the "pay" offered, students involved will skip school less often. They will also pay closer attention to their teachers during classes and do more homework, so that they can do better on the standardized tests that are being used to judge their performance."

The moral hazard concern with such conditional cash transfer programs is that it will encourage some children who have been doing well to lower their school performance, so that they can qualify for the program. The program therefore ends up rewarding even those children who would have achieved good performance even without the cash incentive. Therefore the challenge with administering such programs would be to identify the right target group of children.

Prof Posner sees significant incentive distortions, in both parents and children, arising from such cash transfer schemes. He claims that such programs will continue the neglect of public schools, which may be the cause of the poor performance and drop-outs. He also foresees substantial transaction costs associated with implementing and monitoring such programs. Further, such programs do not mandate any sunset clauses which will end the cash support to the child.

However, these cash transfer programs can be more effective in promoting education in socio-economic contexts like in many Indian states, without stoking off incentive distortions. In many parts of the country, especially among specific communities, girl children drop out from schools very early. Further, during the harvest season, the parents have an incentive to take their children out of school so as to work in the fields. It is also true that many of these practices and trends cover the major portion of children in such areas that it may not be worth the transaction costs to target and exclude the small minority who attend school.

In such circumstances, it is important that the incentives be structured appropriately to meet the objectives. There are many different ways in which the cash transfers can be structured. The amount of cash transferred can increase with every class, and can culminate as a scholarship to attend professional courses. The cash transfer can be graded into a few categories, so as to incentivize children to perform well. Therefore while the best performers get the maximum cash transefr, the worst get the least. It may also be worth increasing the cash incentives for the worst performers in each class, so as to incentivize them to perform better in the next higher class. Further, in many backward areas of certain states, the enrollment rates are so low that merely keeping children at school is itself a challenge.

The transaction costs associated with such programs can be minimized by involving the women Self Help Groups (SHGs). Apart from the child attending school and performing well, such cash transfer can also be made conditional to the mother being a member of an active SHG. The cash transfer can then be made to the bank account of the SHG. This transfer can be made quarterly, based on the attendance and test results of the child. In order to avoid incentive distortions like grade inflation, there should be relative grading of the performance of children.

Such policies will have to be formulated by carefully analyzing the statistics available and tailoring programs to suit the specific local requirements. For example, the cash transfer can kick in at those classes where girl children normally drop out or for those months when children drop out for harvest. But given the different social context, unlike the New York program, all the cash transfers should be made to the parents.

More on bureaucratic incentives and corruption in China and India

I blogged earlier here about Yuen Yuen Ang's analysis of corruption in China and India. I had also blogged here about how President Xi Jinping's anti-corruption drives had created some form of risk-averse survival strategies among bureaucrats. 

In a recent article, she pointed to the paradox of a "corrupt meritocracy",

Corruption and competence do not just coexist within China’s political system; they can be mutually reinforcing. Ji Jianye, the former leader of Nanjing and Yangzhou, in Jiangsu Province is a case in point. Through massive demolition and urban-renewal projects, he rapidly transformed Yangzhou into an award-winning tourist destination, and over the course of his career has earned the nickname “Mayor Bulldozer.” Under his leadership, the city’s GDP surpassed the provincial average for the first time ever. Meanwhile, Ji’s long-time cronies made a fortune during his tenure. In exchange for lavish gifts, bribes, and company shares, Ji awarded their businesses near-monopoly access to government construction and renovation projects. One of these companies, Gold Mantis, saw its profits grow fifteenfold in just six years. The more Ji pushed for growth, the more spoils he produced. This paradox is not limited to Ji. In a forthcoming book, China’s Gilded Age, my study of 331 CPC city-level secretaries’ careers, I find that 40% of those who have fallen to corruption charges were promoted within five years, or even just a few months prior to, their downfall... Yet corruption is more of a feature of the system than a bug. This should come as no surprise. The CPC controls valuable resources – from land and financing to procurement contracts – and individual CPC leaders can and do command immense personal power. Hence, CPC leaders find themselves constantly inundated with requests for favors, many of which are accompanied by graft.

But the other side of corruption is the salient role of competition and performance.

... the symbiotic relationship between corruption and performance in China’s fiercely competitive political system. For political elites whose formal pay is low, cronyism not only finances lavish consumption but also helps advance their careers. Wealthy cronies donate to public works, mobilize business networks to invest in state construction schemes, and help politicians complete their signature projects, which improve both a city’s physical image and the leader’s track record.

See also this article.

Tyler Cowen has a crisp summary of Ang's book,

1. Access money dominates.

More concretely, politicians prosper by getting things built, not by preventing things from getting built.

2. China’s political system operates on a profit-sharing model.

3. Capacity-building reforms have curtailed damaging forms of corruption.

4. Regional competition checks predatory corruption, spurs on developmental efforts, and ratchets up deals.

This is a good podcast on the book. 

There are two aspects of performance that are of relevance. One, performance in the Chinese case is measured in terms of positive contribution to regional economic growth. The metrics used to measure performance were reasonably credible measures of aggregate progress. This is important because there is a difference between cronyism that confers disproportionate privately shared benefits while inflicting significant net long-term social costs and one which shares around both private and social benefits. The later is net welfare enhancing.

The actions of an official who takes initiative and executes a project effectively in time and with good quality or expedites a critical process or proactively eases long-pending constraints on a project are all productivity and welfare enhancing. In contrast, actions which are by way of rent-seeking on statutory permits or licenses, or merely favouring one bidder over another in a tender, or turning a blind eye to a developer compromising on the quality of a project are all productivity and welfare decreasing. The former can be described as 'good' corruption and the latter as 'bad' corruption.

Two, what underpins this performance dynamic in case of China is the institutional incentive system. For the higher officials, positive performance, measured in terms of some aggregate growth, is rewarded with promotions and movement up the Party and government hierarchies. The street level bureaucrats are incentivised with salary increases linked to performance of their agencies or provinces and not insignificant fringe compensation. Then there are the associated rents. These rents act as an attractive efficiency wage to incentivise effort as well as discourage officials from becoming extractive.

An important disciplining factor is the intense competition among officials and party functionaries at all levels, and across provinces. The latter in particular is important since the higher level officials leading the provinces compete with each other to rise up the party and government. This has meant that they have been supportive of the reforms introduced early this decade to constrain theft and predatory practices by lower officials. After all this contained petty and 'bad' corruption and increased their control over their subordinates.

As Joe Studwell has nicely documented, this incentive system also underpinned successful industrial policy among North East Asian economies. The private enterprises which received government benefits had to compete with each other and deliver on export performance. As long as they adhered to this, the system condoned other business practices, including corruption and generation of negative externalities. In short, in China, both development and corruption are competitive.

The formal nature of the performance-promotion relationship is also important. The connection between promotions and performance means that even with the risk of the anti-corruption investigations, there is enough incentive remaining to sustain the symbiotic relationship. Also, officials can mitigate the risk by being part of well-connected patronage networks. In such networks, higher officials act as patrons to lower level officials (clients) - they identify promising clients, cultivate and nurture them, and support their progress up the ladder, thereby creating norms of trust and loyalty.

However, even with the best incentive structure and oversight, in such massive and complex systems there are multiple good and bad equilibriums possible. China's success till date has been in maintaining a balanced equilibrium. For example, the corruption at the provincial and county levels appear not to have tipped over (which could easily happen, even with the best incentive structure) to become aggregate welfare reducing. Or the performance metrics have not been extensively gamed to cover up inefficiencies and corruption. These problems, while doubtless present, appear not to have become the norm. That's some achievement. But how long can it last?

In contrast, in India, there is little symbiotic relationship between corruption and performance. Yes, among some officials (and as a general bureaucratic culture in some states) there is a relationship which is tolerated because they also deliver irrespective of whichever party is in power. It is the classic efficiency wage. And in a system with very few who can actually deliver, such officials command a premium among all parties in power. 

But this is not the case with the vast majority of corrupt officials. Their corruption is of the 'bad' or welfare subtracting kind, largely petty or grand theft without any productive growth enhancing contribution. In fact, a significant portion of the corruption is of the preventing-things-from-getting-done or destructive variant. Besides there is neither any competition nor performance relationship to productively channel such corruption. It also does not hurt anyway to be corrupt. And the occasional efforts to capture performance have been gamed extensively as to lose all credibility.

The highly centralised nature and ubiquitous presence of the government in China, without the clear rule of law restraints that characterise democracies, means that bureaucrats not only have significant power but also the discretion to exercise them. Authoritarian officials can make unilateral decisions and grant exclusive access to privileges and profits. They can make big changes very quickly. They can confer large benefits with the stroke of a pen. So the actions of a high-level provincial bureaucrat can have a significant positive influence on the fortunes of a business. Officials can help businesses move their projects by offering special deals, cheap land, regulatory exemptions, and other perks. These are the access money corruption.

In contrast, in a democracy like India with its rule of law and checks and balances, there are limits to the powers of individual bureaucrats or even politicians. In fact, in most instances, no amount of access money can assure businesses of such preferential privileges. In the Indian context, access money can at best expedite processes, and most often only to get a service within the time that the business is already entitled to. It is often said that officials in India, even at very high levels, can stop a file far more easily that they can move the same file. 

Assessing India's economic growth prospects

Ruchir Sharma offers a nice dose of realism about Indian economy, the relationship between economics and politics, government's contribution to growth and more. He makes some interesting points. 

1. Lower growth rates are the new global normal. There is not a single region in the world which is growing faster now than a decade back. Exports, which provided the boost for the last two decades of growth has been stagnant this decade. To that extent, any growth rate above 5 per cent should be good for India. 

2. India has been out-performing its emerging market peer basket by about two percentage points for 2-3 decades, and that out-performance is likely to continue. It helps that the country's GDP per-capita at $2000 is only a fifth or so of the peer basket average, and therefore the boost from low base is significant. 

3. Empirical analysis of state elections shows that even in the past decade, anti-incumbency has been the dominant trend in Indian politics, even when states deliver impressive economic performance. Anti-incumbency has only marginally declined from 65% to 58% in the past decade.

4. The findings of the World Values Survey over the past 2-3 decades show that among the major countries India shows the highest increase in public preference for authoritarian leaders (as against one more answerable to the Parliament and favours more democracy). 

5. Economic growth in recent years has been driven by consumption than investment, as reflected in the buoyant performance of consumer stocks as against the poor performance of manufacturing and infrastructure sector stocks. To the extent that the former is less dependent on government policies than the latter, the government role in contributing to the high growth rate is marginal. 

6. The arrival of a new leader in emerging economies is associated with elevated stock market performance for 2-3 years, with an out-performance of 20-30%. Over the past three years, Indian stocks have out-performed emerging market peers by 10-15%. 

7. The quality of private sector companies, in terms of having consistently delivered 15% or more earnings growth on a steady basis for more than five years, driving equity markets in India is the best in the world. This is encouraging sign both for the sustainability of the bull market as well as for future growth. 

He writes,

The path has been one of incrementalism for a very long period of time and that is the path that we should expect over the next few years and therefore... to pay attention to politics in India is, from an economic perspective, is really a waste of time... this is a country... that consistently disappoints both the optimist and the pessimist. And so, that realism is what we need. You can always be an optimist and always say that this is going to happen. But for me, that is a money losing strategy and that is one thing which is also, we have not appreciated that if you look at what has happened over the last three years, had you bet on the government to deliver, look at it from a pure stock market perspective, you would have been a loser... the evidence suggests that there is really no connection between politics and economics in this country. On the other hand, it is this sentiment of nationalism and you see this, you see this across social media, you see it across television channels, it is this sentiment towards nationalism and stride at hyper-nationalism at times, it is this sentiment which is what is buoying Modi and the current administration... 

if you look at what has worked in the stock market over the last three years, you will find that there is nothing to do with the government or politics. The best performing sector since May 2014 has been the consumer staples sector. This reflects the fact that what is really driving the Indian economy over the last three years has been consumption. As we know from instances across the world, that the government really doesn't have much of an impact on consumption as much it has on investment. Investment is what the government can really drive by creating the investment environment for investment to pick up or pushing that. Instead if you look at investment related stocks in India, those have done quite poorly on a relative basis especially over the last three years.

And this is interesting and I agree,

So, my simple point being here that the connection between politics and economics in this country is rather limited and the stock market's behaviour over the last three years since this government came to power has only reinforced that notion. If you look at both the internals of the market, in terms of what has done well and also the overall market, neither the pessimists nor the optimists on the Modi government would have made any money based on a political view. So, in this country if you want to do well you have tune out the politics and be an internal exile as far as politics is concerned because that only interferes with sort of making money in this nation.

But, in favour of the government, it has to be argued that the counter-factual is impossible to have. To its credit, the government has not done any harm, and has largely been pursuing stable fiscal and macroeconomic policies. This is more than can be said of governments, not just in India previously, but also globally in emerging markets. 

It is here that I disagree with Ruchir, 

There are countries like China, Korea, Taiwan which have been able to grow at 10 percent plus. But my point has always been that to expect big bang reforms in India, to expect that some major big bang reforms will take place in India has always been a bad bet because that never happens. Our culture is one of incrementalism. We do things in incremental steps and I think that is what should be the operating assumption.

I do not think that there are big-bang reforms in India's context. When we talk of big-bang reforms, we think of one-off decisions like deregulation, privatisation, promulgation of new laws, and so on.  Take a decision and you are done. The sort of stuff like privatisation of banks and public sector units, while necessary, on their own, are unlikely to unlock massive growth energies.

The real reforms in India's context, as we laid out in our book, Can India Grow?, are more in the nature of steady and focused accumulation of human, physical and institutional capital, whose base is astonishingly low for an economy of India's size. Deficiencies in these are the binding constraints to sustainable high growth rates. No amount of big-bang can make up for them. For sure, there are some big-bang stuff there, as we outlined, but those are not the stuff the markets associate with big-bang reforms. They are in the nature of pulling complementary levers and persistent follow-up for long periods to address deep-rooted problems.

They would include reorientation of school education single mindedly towards learning outcomes; restructuring of UGC, MCI etc; facilitating the development of financial savings  instruments and enabling access to them, so as increase the savings rate; transition to outcomes-based financing in health care, away from line-items health funding; policies to provide tenure stability and address politicisation of officials postings; across the board standardisation, e-procurements, and third party quality audits; reforms to address decision paralysis and so on. Not the sexy stuff like repeal of Section 25N of Industrial Disputes Act 1947 or the privatisation of Air India or Indian Railways!

"Wage ceilings" for executives!

The ongoing financial crisis has seen some of the hallowed principles of free-market capitalism taking a beating and being put aside and government interventionist policies being embraced. First came the acknowledgement that markets are not always efficient in both allocating resources efficiently and discovering prices, as the sub-prime bubble built up and then burst in spectacular fashion. Then came the more distressing conclusion (or a re-affirmation) that markets are not self-correcting and require massive government interventions through monetary, credit and fiscal policy support. This has been followed by the more dramatic measures like creeping nationalization through liquidity injections to ailing banks, protectionism, and the latest being the modern day equivalent of rent controls and minimum wages - ceiling on executive compensation!

The Obama administration has finally bowed to populist pressure and taken the plunge by placing limits on executive compensation for Wall Street firms (both those bailed out in exceptional circumstances and those receiving general bailout money) taking the bailout assistance. The objective is "to ensure that the compensation of top executives in the financial community is closely aligned not only with the interests of shareholders and financial institutions, but with the taxpayers providing assistance to those companies".

The Plan limits the total executive compensation to no more than 500,000, including bonuses and except for restricted stock awards or other long term awards, that vests till the government has been repaid with interest, thereby aligning their incentives with both the long-term interests of shareholders as well as minimizing the costs to taxpayers. It also contains more pro-active disclosure and certification requirements by the company boards on executive compensation. There are provisions to clawback bonuses for top executives engaging in deceptive practices, after the wrong-doings come to light. Other highlights include - restrictions on severance payouts (golden parachutes) to less than one year; restrictions on luxury expenditures which would include adoption of a company policy on such expenditures and require certification by the CEOs etc.

It also offers a few suggestions for long term restructuring of executive pay regulations - greater disclosure and more proactive explanations for higher payouts; restrictions on stocks cash-out, so as to encourage long-term perspective; and shareholder inputs on these payments. It will apply only to those now seeking assistance from the government, and not to the 350 odd financial institutions who have already sought government assistance. There are also certain flexibilities - there is no restriction on the amount of long term bonuses/stocks that can be awarded; companies that seek aid but do not need exceptional government assistance can waive the $500,000 pay cap, as long as they submit their executive pay policies to a nonbinding shareholder vote etc.

The $500,000 cap has to be seen in light of the spectacular increases in executive compensation over the past few years. In 2007, the latest year that figures are available, the largest participants in the bailout program paid their chief executives an average compensation of $11 million, including salary, bonus and benefits. Of that amount, only about $844,000 was cash salary, about $2.5 million was in a cash bonus, with the bulk — $7.4 million — in stock awards, and the remainder in benefits and perks. New York based financial institutions, most of whom received bailout money, paid out $18.5 bn as bonuses in 2008, even as these institutions tottered on the edge of precipice.



Politically, it had become almost impossible for the Obama administration to stave off the demands to impose restrictions on executive compensation, following news reports of excesses in many firms receiving tax-payer bailout money. There are a few issues that are likely to get debated here. Lawrence Katz and Claudia Godin studied the career choices of Harvard undergraduates since the 1960s, and found that the share entering banking and finance rose from less than 4% to 23% or so in recent years, and those who chose careers in finance made three times the pay of their peers, adjusting for grade-point averages and test scores.

Will the pay caps drive away the most talented people from the sector, just as they attracted these people during the boom years of the past two decades? Will these restrictions discourage institutions away from seeking bailout assistance? This in turn could have two contrasting effects - it could incentivize the firms to seek out the best possible deals from the market to salvage their companies, thereby promoting most efficient allocation of scarce tax payer money, or it could lead to executives delaying seeking government assistance till they face the inevitable, when the costs of any bailout will naturally be much larger. On an even more distortionary note, will the pay caps drive these talented people to those relatively less regulated areas of the financial markets, like private equity and hedge funds in the shadow banking system?

Update 1
Two articles by Uwe Reinhart and Robert H Frank, has some interesting points. Economists refer to two issues in support for higher supply-demand driven executive compensation - some executives are truly endowed with exceptional qualities that deserve a high enough premium (this in turn is a derivation from the premise that individuals can have disproportionately large influence on the fates of companies), and that the supply of such individuals is limited thereby magnifying the premium. Now both these assumptions have been questioned in numerous studies, thereby leaving these fundamental assumptions in doubt.

Update 2
Lucien Bebchuk feels that the Obama administration's execitive compnesation cap is too modest and liberal, in so far as it leaves companies, including those receiving exceptional assistance, free to increase performance-based compensation to make up for any salary reduction. Reed Hastings calls for a 50% tax on executive compensation, instead of pay caps.

Update 3
NYT has this account of the deal when Merrill was taken over by Bank of America, which conveys a tale of greed and deception.

Update 4
NYT has this article which discusses the disclosure by New York Attorney General that Merrill Lynch doled out $3.6 bn in bonuses in 2008, an year it lost $27 bn and then disappeared into the stables of Bank of America in December. About 696 out of the 39,000 employees received $1 million plus bonuses, while 149 of them received more than $3 million, amounting to $858 mn. Payments were made just before the takeover by BoA.

Update 5
Uwe Reinhardt marshals impressive statistics to prove that Jack Welch may after all not be the Lone Ranger CEO he is made out to be.

Temporary recruitment in government

One of the proposals on the problem of job creation in India is the idea of using government employment as a combination of temporary employment and skill acquisition opportunity for youth at the start of their labour market journey.

On these lines here's one proposal about temporary employment in government.

One is, I think, the government could expand the number of people it employs if it didn’t offer them the kind of terms they’re offered. For example, in China — I haven’t checked the data — there are three people with a bachelor’s degree in every village working for the village government. That changes the world. These are people with certain skills and a certain amount of knowledge of the world. So one thought is that the government should start introducing maybe a transitional mechanism where you take a job, and then it’s only if you’re good at it, that you keep it otherwise you can work for some years, and then you don’t. It’s like a tenure system. That will still create more jobs. We need more people on the ground. I don’t think our government is big enough. There’s a lot of people who will say that we have a big government, but in fact, we have a small government that’s trying to keep control, which looks like a big government, a heavy government, but it’s not actually so. The size is small (but) the hands are heavy, as a result partly, because it can’t do anything new because it has so little bandwidth. So rethinking the shape of the government, having more young people in government, as a trial as a way to start your life, but then you can go out and do something else. And somehow getting the court systems to agree to not ex-post turn everybody into a government official. I think it needs a set of tough decisions. But I think without that, our employability issues are going to be fraught all the time.

There have been other ideas like recruiting local youth as apprentices, training them, giving performance-based marks, and offering exit payments if they do not get into regular recruitments. 

This is a logically good idea and has already been tried out in some form of other. The most salient example is that of the vidya volunteers in education, who were the local educated and who were recruited with a honorarium wage under the Sarva Sikhsha Abhiyan. The idea was that these local educated youth would meet local teacher shortages and also get some experience while also acquiring their BEd/DEd qualifications and preparing for their Teaching Eligibility Test (TET) examinations. If they qualified, they would get recruited as a regular teacher, and if not, they could move on to elsewhere. 

The only problem is that in general the movement to elsewhere rarely happen. The volunteers stay on, and then the political economy takes over to mount pressure to regularise them. And it's a matter of years (sometimes decades!) before they get regularised. Or be absorbed into something like the minimum time scale on the principle of equal pay for equal work enunciated by the Supreme Court. 

Once recruited on contract or any other temporary employment mode into the government, the political economy makes the retrenchment very difficult. And the strength of the political economy factors increase with the size of the group under consideration. 

Further, since the recruits internalise this expectation at the time of recruitment itself, very few of them end up leaving on their volition for better opportunities. After all why seek better opportunities, when you have a strong chance of being regularised, sometime or the other. 

Another example is contract faculty in higher education institutions. Thanks to the Supreme Court orders, all of them are now paid at the minimum time scale. When originally conceived, the idea was that these faculty would be only for a short time till the regular faculty got recruited. But for a variety of reasons, the contract faculty have continued. Very few among them leave after getting better opportunities. 

An emerging category of large non-regular recruitment that's happening in governments is that of data entry operators, who have largely replaced the Junior Clerks or Junior Assistants. Here too the logic of temporary recruitment was that these people will pick up some skills and experience and move on to the private sector. This rarely happens. And there are already pressures to regularise them in several states. 

As an empirical validation, it would be useful to look at examples of more than 100 people recruited by a government department or agency on some temporary mode across states and see how many of those recruitments got formally terminated. My guess is that there'll be just a few, if at all. We could start with examples of large scale temporary (or non-regular) recruitments in governments - vidya volunteers, contract teachers, work inspectors, anganwadi workers, ASHA or community health workers, municipal sanitation workers, and home guards. What proportion of those recruited got into regular government jobs, what proportion left for the private sector, what proportion got regularised, and what share are now left? This would be a great PhD thesis for a young scholar - the political economy of contract recruitments in Indian states! It would unpack several aspects of public recruitments hitherto unknown to outsiders. 

Similarly, all government recruitments come with a 1-2 year probation period. In fact, most states recruit teachers with a two year probation. And there have been numerous instances of clear irregularities and moral turpitude by probationers. But there'll be hardly any instance of terminations during probation in any state. 

I believe there are three very strong reasons why this logically appealing solution will struggle in the Indian context. One, there's a very large premium associated with local employment, even if it's on a contract mode and has limited career progression opportunities. Two, government employment, especially in the same district or region, is economically and for social status considered the best among all employment opportunities. And there is the strong moral hazard that the political economy will ensure regularisation of contract (and nowadays even outsourcing) recruitments even if it might take time. Finally, there are limited comparable opportunities available in the private sector. For a start, educated youth prefer white collar jobs. But in these kinds of jobs, till the middle-levels, private sector pays far less than their public sector counterparts. Also, unlike government jobs, good private sector jobs are found only in the larger cities, where costs of living too are much higher. Further, it also does not help that the vast majority of these youth would anyways be unemployable in the positions they aspire to work in the private sector. 

No idea or innovation can wish away these fundamental problems. All these ideas end up providing backdoor for virtual public recruitments without any of the eligibility qualification requirements of formal recruitments. Governments are left with poor quality employees.

In general, it's observed that such approaches work best for higher skilled contract posts like individual consultants or technical personnel, and where the recruitments are done in small numbers. In these categories, the pressure groups will be too small to mount pressures to demand regularisation.

Designing incentives to improve teacher performance

How do teachers respond to financial incentives and contract tenures? Karthik Muralidharan and Venkatesh Sundararaman have examined the results of a randomized evaluation of various policy options to improve the quality of primary education implemented across a large representative sample (500) of government-run rural primary schools in five districts of the Indian state of Andhra Pradesh and found several interesting results.

They found that providing an extra teacher with fixed-term renewable contracts to a randomly-chosen subset of primary schools led to an improvement in student performance in math and language tests by 0.16 and 0.10 standard deviations (a treatment impact of 0.1 SD is equivalent to saying that an average child who received the treatment would have improved his/her rank by 4) respectively over the untreated control group schools after two years. The treatment effects were found to be the largest in remote schools and for students in the first grade. The contract teachers were significantly less likely to be absent than employee teachers, and more likely to be engaging in teaching activity when observed during unannounced visits to schools.

As the authors point out, unlike that of contract teachers, the wages of regular teachers are pushed up by their higher education qualifications and availibility of opportunities, premium to work in remote and rural locations, and the effect of unionization. The wage differential is often a multiple of more than five. The biggest criticism against the employment of contract teachers has been the on the grounds that it would lead to dilution of learning standards. Without claiming that a contract teacher is more effective than a regular teacher, they write,

"The combination of low cost, superior performance measures than regular teachers on attendance and teaching activity, and positive program impact suggest that expanding the use of contract teachers could be a highly cost effective way of improving primary education outcomes in developing countries... expensive policy initiatives to get highly qualified teachers to remote areas may be less cost effective than hiring several local contract teachers to provide much more attention to students at a similar cost... there may be an extent to which quantity can more than make up for the lower qualifications of contract teachers (especially for primary schooling) and do so in a cost-effective way."

In order to avoid the problems associated with mass recruitment of un-trained contract teachers and the resultant dilution of professional standards, the authors propose a performance-linked tenure track to integrate contract and regular teachers into a career progression. Under this arrangement, only contract recruitments are made and consistently high-performing contract teachers could be promoted to regular civil-service rank at the end of a fixed period of time.

Another policy experiment that involved providing bonus payments to teachers based on the average improvement of their students' test scores in independently administered learning assessments (with a mean bonus of 3% of annual pay) found that students in treatment schools performed significantly better than those in control schools by 0.28 and 0.16 SDs in math and language tests respectively at the end of two years. They found that the gains in test scores represented an actual increase in learning outcomes, and that the treated schools also performed better on subjects for which there were no incentives, suggesting positive spillovers. They also find that teacher incentive programs were three times as cost effective in raising test scores than unconditionally provide additional schooling inputs. The challenge with this is to arrive at the optimal level of bonus which incentivizes effort without creating distortions. Combining these two results, the authors suggest that

"there could be significant gains from moving to a system of hiring teachers on fixed-term contracts and then using performance measures to pay bonuses on an ongoing basis and to inform the tenure decision after a longer period of performance measurement. Such a system could both move the entire distribution of teacher effectiveness to the right (by increasing effort) and further increase average teacher effectiveness by not renewing the contracts of ineffective teachers... The use of teachers on fixed-term renewable contracts can be a highly cost effective policy for improving student learning outcomes, especially if placed in the context of a long-term professional career path that rewards effort and effective teaching at all stages of the teaching career."

I am inclined to believe that the performance-based pay system being proposed may be more difficult to scale up than anticipated. While it may have succeeded in the limited context and time over which it was implemented, it may not yield the desired results with a more ambitious scope and larger area of implementation. As the program is expanded to cover all the teachers, the incentive award risks becoming an routine entitlement and getting diluted.

There are likely to be four major problems associated with the determination of bonuses. First, what would be the most optimum bonus? It should neither be too small as to have limited incentive effect nor too large as to waste scarce resources and generate incentive distortions. Second, as the program coverage expands, it will become difficult to cover all the teachers under a single bonus formula with broad acceptance among the stakeholders. Third, there will be increasing conflict between the search for accuracy of the formula in approximating the actual impact of teacher efforts and its transparency so as to generate broad-based acceptance. Finally, and most crucially, there will always be the problem with administering accurate data collection on such a massive scale. How do we guard against potential problems like grade inflation and data manipulation?

Efforts to incentivize teachers with performance based-pay and choice in transfers have been attempted in many states across India, albeit on smaller scales, with not so satisfactory results. The problem ultimately boils down to administering such massive programs and guarding against the dilution of its standards as it expands to cover all the schools. Given the standards of school supervision that prevails and the massive numbers of supervisors at different levels involved, it will be a major challenge to ensure the quality of data collection even for a district, leave alone a state. It will be very difficult, even with extensive computerization to effectively address this challenge. Once the programs are institutionalized, there is an ever-present danger that the performance incentive system loses its sanctity.

Further, in the absence of a sunset clause, these incentives are liable to be distorted. With time, there is the imminent danger, especially given the influence wielded by unions, that the incentives could become internalized as part of the salary structure. Ultimately, this could merely add to the substantial premium that government teachers enjoy over private school teachers.

Since behavioural economists have shown that people are more averse to losses than attracted to similar sized gains, penalties are more likely to be effective than incentive rewards. However, penalties raise opposition from the unions and would therefore be difficult to implement.

The policy of renewable fixed-term contract appears more practical and easier to implement. It opens up the possibility of a tenure track for newly recruited teachers that incentivizes them to perform and not shirk work. But the challenge with this approach would be to actually fire or discontinue the contracts of the poor perfroming contract teachers.

There is already a provision in the existing teachers' recruitment process in many states, that takes in these teachers initially as probationers whose services will be regularized after two or three years on receipt of a "satisfactory service certificate" issued by their supervisory officer. It is therefore legally possible to discontinue the probations of poor perfroming teachers. But there have been no or very few instances of such terminations, so much so that the process of declaration of probations has become superfluous and degenerated into a routine activity.

One way to overcome this problem is to keep the contract periods relatively short, two or three years, so as to leave the discharged candidates enough time to seek an alternative career. A policy on the recruitment process assumes significance in view of the current proposals in a number of states to recruit teachers in large numbers.

Individuals matter, and more so in public bureaucracies

I blogged here about the importance of strong public oversight and in-house expertise for the successful execution and management of infrastructure projects. 

Mainstream development discourse focuses disproportionately on institutional and systemic challenges, and overlooks the important role played by individual officials in effective public services delivery and the realisation of policy outcomes. 

Specifically, I am referring to the commitment and expertise of individual public servants in important positions at all levels of the government. By important, I’m not confining to leadership positions, but any position where they can make significant contributions to influence the agenda. 

The importance of individual officials in the success of development interventions is generally overlooked amidst systemic and institutional factors. This also comes from the belief that the success of development interventions is determined by good ideas, comprehensive planning, and rigorous monitoring. This theory of change overlooks the reality that successful development interventions tend to emerge iteratively over the course of their implementation. Such iteration, in turn, requires capable and committed leadership, especially important given weak state capabilities. 

In fact, it is no exaggeration to argue that capable, proficient, and committed officials are perhaps the most important ingredient of state capability. 

The importance of capable individual officials is also borne out by academic research. I blogged here, drawing attention to the work of Philipp Barteska and Jay Euijung Lee, who examined the impact of the bureaucratic capabilities (of export promotion officers) on the effectiveness of industrial policy in terms of export performance in South Korea. They found the following:

We exploit the three-yearly rotation of managers of South Korea’s export promotion offices in 87 countries between 1965 and 2000 to show that a one standard deviation increase in bureaucrat ability boosts exports by 37%. Under higher-ability bureaucrats, South Korean exports respond more strongly to a country’s import demand, suggesting a more effective transmission of market information.

An increase in exports by nearly two-fifths with just one standard deviation increase in bureaucratic capability tells us that the quality of officials might matter more than (or at least as much as) fiscal incentives and regulatory changes in trade promotion efforts. 

Alessandra Fenizia studied the impacts of managers in the public sector in Italy using a dataset containing an output-based measure of productivity. 

Exploiting the rotation of managers across sites, I find that a one standard deviation increase in managerial talent raises office productivity by 10%. These gains are driven primarily by the exit of older workers who retire when more productive managers take over. I use these estimates to evaluate the optimal allocation of managers to offices. I find that assigning better managers to the largest and most productive offices would increase output by at least 6.9%.

Cristobal Otero and Pablo Munoz study government managers in public health provision in Chile. 

Using novel data from public hospitals in Chile, we document that top managers (CEOs) account for a significant amount of variation in hospital mortality. We then use a staggered difference-in-differences design, and show that a reform which introduced a competitive selection system for recruiting CEOs in public hospitals reduced hospital mortality by approximately 8%. The effect is not explained by a change in patient composition and is robust to several alternative explanations. The financial incentives included in the reform—performance pay and higher wages—do not explain our findings. Instead, we show that the policy changed the pool of CEOs by displacing older doctors with no management training in favor of younger CEOs who had studied management. The mortality effects were driven by hospitals in which the new CEOs had managerial qualifications. These CEOs improved operating room efficiency and reduced staff turnover.

Michael Carlos Best, Jonas Hjort, and David Szakonyi analyse data on public procurements in Russia and find the value of bureaucratic effectiveness. 

Using data on 16 million public purchases in Russia, we show that 39 percent of the variation in prices paid for narrowly defined items is due to the individual bureaucrats and organizations who manage procurement. Low-price buyers also display higher spending quality. Theory suggests that such differences in effectiveness can be pivotal for policy design.

R D Metcalfe, A B Sollaci, and C Syverson

In this setting, managers move between stores but management practices are set by firm policy and largely fixed, allowing us to hone in on managers’ personal roles in determining store performance. We find: (i) managers affect and explain a large share of the variance of store-level productivity; (ii) negative assortative matching between managers and stores, which may reflect both firms’ decisions and a selection-driven bias that we characterize and argue might apply in other settings using movers designs; (iii) managers who move do so on average from less productive to more productive stores; (iv) female managers are less likely to move stores than male managers; (v) manager quality is generally hard to explain with the observables in our data, but is correlated with the ratio of full-time to part-time workers; (vi) managers who obtain high labor productivity also tend to obtain high energy productivity, revealing some breadth in managers’ skills applicability; (vii) high-performing managers in stable growth times are also high-performing during turbulent times; and (viii) exogenous productivity shocks improve the quality of initially low quality managers, suggesting managers can learn. We explain implications of these findings for productivity research.

Ricardo Dahis, Laura Schiavon, and Thiago Scot investigated the performance of state judges in Brazil. 

We investigate this question focusing on state judges in Brazil. Exploring monthly data on judicial output and cross-court movement, we estimate that judges account for at least 23% of the observed variation in number of cases disposed. With novel data on admission examinations, we show that judges with higher grades perform better than lower-ranked peers. Our results suggest competitive examinations can be an effective way to screen candidates.

On a slightly different note, Kevin Hawickhorst shows how technical expertise built up within public systems in the US allowed for the nurturing and flourishing of capable officials, created public confidence, and thereby enhanced the credibility of governments. 

However, over time, expertise has come to be crowded out due to the conscious shift in the way government bureaucracies came to be organised. 

At the turn of the twentieth century, agencies followed a distinct blueprint: they were organized by subject matter, not by abstract function. Each bureau focused on a single domain—such as soils, mines, or forests—and combined research, regulation, and grants under one roof. In the U.S. Department of Agriculture (USDA), the Bureau of Entomology, for example, studied insect-borne diseases, issued rules to contain them, and funded farmers to protect their crops, all as part of a single mission. This structure helped agencies recruit experts by offering broader, more meaningful work than corporations could, and it built a shared sense of mission rooted in a vocational community.

Today’s agencies look very different. After World War II, reformers dismantled the integrated subject matter bureaus and reorganized government along what they called “functional” lines. In this system, regulation is one bureau, research another, and grant administration still another; each bureau covers a wide range of subjects and is defined by its activity rather than its mission. It is the model we now take for granted. The Bureau of Entomology is gone, and USDA now houses all agricultural research in a single unit. New agencies were built this way from the outset: the Department of Housing and Urban Development, created in 1965, was designed as a grantmaking machine, never a vocational community.

The shift was a well-intentioned one and backed by a wide coalition of reformers, businessmen, and interest groups. Functional departments looked modern, rational, and efficient: they simplified charts, tightened chains of command, and promised to reduce duplication. But what seemed like sensible reform gradually hollowed out the structures that had made expertise durable. Once government agencies lost their vocational missions, they stopped drawing on networks of expertise and started looking like paper mills, less able to command political respect, and more vulnerable to capture and drift… logic of the Progressive-era model: that research and administration had to remain intertwined within a unified mission if expertise was to thrive.

However, this wealth of internal technical expertise has, over time, given way to the tribe of generalist managers. 

The core mistake was a shift in what we thought expertise was. The Progressive reformers built vocations that were tied to missions, visible to the public, and legible to politicians. Their successors redefined expertise as a credential: the knowledge of process rather than mastery of a craft. To businessmen and academic reformers alike, competence meant general managerial skill, not professional vocation. As this view took hold within the bureaucracy, “expertise” came to mean knowing the procedures rather than knowing the work. We have traded the civil engineer and the entomologist for the program analyst, the management consultant, and the diversity officer—experts who know how to manage the process but not how to do the work.

This redefinition of expertise hollowed out our idea of representation. We now equate representation with participation and diversity, as if the state were legitimate only when citizens can see themselves in its officials. The Progressives, by contrast, recruited from the country’s varied vocations and made that work visible to the nation. Expertise was representative not because it resembled the public but because it served the public, visibly and competently… Their institutions were built to make expertise endure, by recruiting promising candidates from vocational schools and professional societies, dressing them in uniform, and sending them to work alongside state engineers, agricultural agents, and university researchers…

The Navy cannot build ships. In 1940, faced with the same problem, Congress did the obvious: it created a Bureau of Ships, put engineers in charge, and got ships built. That bureau is gone, and we treat its return as unthinkable. Yet the remedy remains the same. If we want ships, we should once again have a Bureau of Ships to build them… Repairing our institutions will ultimately require returning to the vocational conception of expertise… We have built institutions that valorize process in place of vocation, producing a bureaucracy that neither embodies skill nor commands respect. What matters now is not saving “expertise” in the abstract but rebuilding the institutions where it can serve visibly and credibly.

Hawickhorst’s essay points to several individual public leaders in the US who built institutions and brought credibility and confidence in public agencies through their careers - George Uhler (headed Steamboat Inspection Service for 20 years from 1903), Logan Page (Office of Public Roads, founded in 1905), Joseph Kinyoun (headed the Hygienic Laboratory, a precursor to the National Institutes of Health), and Gifford Pinchot (founded the Forest Service). Every country has such leaders across levels. 

In their search for better outcomes in public policy, governments tend to expend effort and resources on interventions involving financial support, regulatory enablers, and technology adoption, while overlooking personnel choices. This bias is also reflected in public commentary and academic research that shapes public narratives. 

However, as the case of the South Korean export promotion officers starkly demonstrates, for governments intent on reform and impact, personnel choice decisions may be the lowest-hanging fruit. In most policy areas, the range between the opportunity cost of a bad personnel choice and that of a capable personnel choice may be much greater than that for any other policy intervention.

The private sector addresses the issue of the importance of capable individuals by incentivising them with extrinsic material motivations like financial rewards and fast-tracked promotions. While neither of these instruments is available to governments, it can appeal to the intrinsic motivation of public-spirited officials. 

This would require acknowledging capabilities and merit (as borne out strictly by performance track record, not merely in some narrow quantitative sense). This requires differentiating capable bureaucratic leaders from their larger peer group by entrusting them with higher responsibilities, appointing them to identified important positions, drawing on their expertise in various forms, recognising their work through different non-financial channels, and generally signalling their differentiation. 

An explicitly professed intent, let alone a rigorously implemented process, that seeks to differentiate among officials at all levels, can be a powerful force to shape expectations and align incentives within public systems.