Showing posts with label education. Show all posts
Showing posts with label education. Show all posts

Wednesday, January 02, 2013

The rise of high-end finance work in India

by Shashank Bansal.

Until recently, outsourcing by global financial firms to India conjured up an image of commoditised low end services outsourcing: call centres, peripheral systems programming, and testing and maintenance. However, in recent years, there is a new rise of more sophisticated work. This reflects supply and demand factors. Global financial firms are keen to cut costs. Capabilities of operations in India -- both captives and independant firms -- have grown for many reasons:

  • The individuals involved in this field in India have gained experience ("learning-by-doing") and credibility.
  • New management practices and improved telecommunications technologies have improved the extent to which teams and projects are handled in a more non-local way.
  • The Indian diaspora has been rising to senior management levels in global firms, and is better able to envision what can be done in India and to obtain execution.

A European investment bank was among the first to experiment by bringing in teams in India into critical projects. This was a landmark change as a lot of inertia about confidentiality was overcome. Other banks followed suit. New management practices, higher pay, greater meritocracy came in, which helped Indian teams make the transition from low-end work where the HR and management techniques used are quite different. Demand for high skill labour has helped induce greater supply, with a lag, as individuals were more inclined to tool up with advanced degrees and high-end knowledge.

Alongside the developments in finance, parallel developments were taking place in the field of offshoring which have driven up skill levels, and helped create a high skill ecosystem in India. Top tier consulting firms launched `centres of excellence' in India, hiring grads from IITs, IIMs, IISc, statisticians, economists. While education in India has huge problems, the raw talent available in India was of good quality, particularly when we focus on individuals who were able to read on their own and reinvent themselves ("never let your school come in the way of your education"). This process has been helped by globally recognised certification exams such as the FRM and the PRM.

IT firms have have been evolving from core development and maintenance to an entire gamut of IT strategy and consulting for financial firms. Many smaller KPO firms with specialised domain knowledge in finance have emerged, who cater to smaller hedge funds, trading houses, not just outsourcing increasingly complex pieces of work, but also advising them on the entire outsourcing strategy. All this has helped create a pool of high skill labour which is moving between multiple employers in India and able to build knowledge through diverse kinds of experience.

The most impressive development of recent years has been the growth of offshore trading units of global brokerages and trading houses, where people sitting in India take independent trading decisions in international financial markets based on their own skills and judgement. In some ways, this is the highest level of transfer of decision functions to India, albeit at relatively low monetary stakes.

In this fashion, within a period of 15 years, India had graduated from doing repetitive low value tasks to Knowledge Process Outsourcing (KPO) for the global financial system. While these activities are primarily in Bombay, they are also taking place in Gurgaon and Bangalore. The number of high-end finance workers in Bombay has never been greater than it is today. It is estimated that there are now 50 individuals working in Bombay doing work for global financial firms who have Ph.D. degrees in quantitative fields. This is starting to become a big enough number for them to talk with each other and get network effects going. From an employer's point of view, it is now possible to shop in the labour market in Bombay and recruit a 10-man team all with Ph.D. degrees so as to get a new group going. This is a sea change when compared with conditions just a few years ago.

To appreciate this change a little further, it was interesting to take a look at some of the capabilities of finance focussed KPOs, divided mainly into 4 broad categories, catering to Sales and Trading, Middle office and Back office:

  1. Quantitative Research and Analytics Support:
    1. Equity and FICC Analytics: Model Validation, Price Verification jointly with clients: these are pretty quant heavy functions which require in-depth understanding of products.
    2. Technical and Fundamental Analytics.
    3. Index and Portfolio Analytics: Index maintenance, design, construction, operations and after sales, Portfolio tracking, decomposition and correlation analysis, performance measurement and attribution support.
    4. Derivatives and Risk Analytics: Measurement of derivatives Greeks, Value at Risk, Tolerance checks.
  2. Research:
    1. Equity and FICC Research: Company research, Credit Research, Economics research etc. to augment senior analysts in money centres.
    2. Trade idea generation and back testing: Sales pitches for clients and internal trading desks.
    3. Country, Sector, Company profiling, trends, news and projections: Pitch book generation and support.
    4. 24x7 weather patterns tracking for global energy trading outfits
    5. Overnight trade and market tracking to feed in summary reports, Market Dashboards, news letters, morning meetings and agendas
    6. Market Research: Pre-entry market research and positioning survey for bank's clients.
  3. Data Analysis and Modelling:
    1. Data sourcing from multiple heterogeneous sources, refining and maintenance: Static data, Live and Historical market data maintenance. Data research and statistical studies feeding into trading strategies.
    2. Data Mining solutions.
    3. Data modelling, smoothing: Providing data solutions for Algo trading desks.
  4. Operations and Control
    1. Derivatives trade processing and documentation: Trade review of structured trades and complex documentation. End to end life cycle management of trades e.g., matching, broker confirmations and fee calculations.
    2. P&L and balance sheet control: Generation and reporting of P&L for vanilla products. Some banks have started moving exotics P&L functions to India. This is quite a significant milestone as such activities require high degree of confidentiality and direct user (e.g., traders) interaction who have zero tolerance for mistakes.
    3. Risk Stress testing, VaR back testing, Risk reporting to senior management.
    4. Auditing: external auditing of valuation marks of trading desks and control processes around it.
    5. It should be noted here that since the funding crisis of 2008, these jobs have become quite complex as most banks have built more sophistication into their analytics. For example, most yield curves would now have multiple basis spreads (like tenor basis, xccy basis) and not just rates desks but even credit and equities desk have been using such advanced discounting curves.)

What's next

The biggest push probably has been in quantitative middle-office functions with an ever increasing emphasis on valuations and counterparty risk management. Given the way markets have adopted collateral based pricing of derivatives, and the regulatory push on managing counterparty default risk, some captives have started building quantitative teams who will develop and manage CVA, DVA, etc. processes for all trading desks.

The new regulatory climate (Dodd Frank, Basel III etc) has lead to a substantial increase in costs due to additional checks and reporting requirements e.g., centrally cleared OTC trades, real time trade reporting to regulators, exhaustive risk reporting - all of which can are leading to fresh volumes of activity in offshoring.

All high quality banks have a team of techno-quants who work closely with the sales/trading desk, risk managers etc, on their day to day needs as well as on strategic projects. It is now feasible to move such high impact roles to India. It would be possible to have "extended front office teams" where dedicated staff support traders in money centres, doing real time risk analysis and client profiling, while the trade is being dealt overseas.

For a back-of-envelope calculation, if we think of internal billing rates of $100,000 per person per year, and if there are 10,000 persons at this average price, then this is services export of $1 billion a year, which is a sizeable amount. It appears that the early beach-head is in place, and this area will grow dramatically now.

This blog post reflects my experience, which is in investment banking and money management. A similar escalation of complexity of work in India is taking place in retail banking, insurance, etc., reflecting similar compulsions and opportunities.


There is a certain tension between the push towards offshoring to India, and the activities that regulators consider `key in-house activities' that cannot be outsourced.

There are serious constraints with education in India. The top institutions are producing some quantitative skills (e.g. fluency with matrix algebra, fluency in numerical computation). On one hand, there are weaknesses of broad intellectualisation that shapes cognition, creativity and malleability. On the other hand, there is essentially nothing in place by way of a finance education in India. A small amount of high-end finance research is taking place (example) but for the rest, there isn't much capacity in the existing academic campuses. New approaches to learning and training need to be devised through which high quality individuals, with strong quantitative skills, can be converted into full fledged participation in high-end global finance work. A mix of public and private initiatives are required in order to jump to the next level.

There are strong synergies between the sophistication of the Indian financial system and the work that is done for global financial firms. There is a two-way feedback loop here: Better domestic capabilities will help do sophisticated offshore work, and the brainpower built for offshore work will strengthen domestic capabilities. The best example of this is found in the equity derivatives market, where India has a world-class market. The individuals with a domestic background here are ready for offshore jobs in fields like algorithmic trading, and individuals with capabilities built in offshore work are useful in the domestic setting. This is where India can set itself apart from Malaysia and the Philippines. To the extent that Indian financial reform makes progress, this will fuel the rise of high-end outsourcing to India.


I am grateful to Anand Pai, Paul Alapat and Gangadhar Darbha for useful discussions.

Tuesday, August 21, 2012

The widget illusion

The Economist runs a discussion forum titled The Economist By Invitation. In this, they recently setup a discussion about an opinion piece by Dani Rodrik about the future of manufacturing-led growth in emerging markets. I wrote a response there which is reproduced here.

The role of manufactures

I agree with a small element of Dani Rodrik's argument, but mostly for different reasons. Rodrik says:

Except for a handful of small countries that benefited from natural-resource bonanzas, all of the successful economies of the last six decades owe their growth to rapid industrialization.

I have seen this kind of thinking among some policy makers in India also: that industrialisation is somehow special and good when compared with services. I would question this proposition, that I term `the widget illusion'. What matters to a country is having sophisticated firms that have a high marginal product of labour. We should not care whether this happens in services or in manufacturing. If anything, the opportunity to do it is perhaps better in services.

India is a good example of a country which embarked on its catchup by connecting into globalisation late: from 1991 onwards. It was probably the last country in the world to shed autarkic policies. This has given a remarkable growth acceleration. Sustained growth of 7 per cent is pretty good by world standards. These achievements have been significantly driven by services production in India within global supply chains (whether within production facilities owned by global MNCs who are operating in India, or contracted-out by global MNCs to Indian firms). If your null hypothesis was that industrialisation is essential to growth, then you would not have predicted what happened in India, where manufacturing was hobbled by an array of policy mistakes.

This illustrates the limitations of manufacturing-focused thinking, which seems a bit out of date in today's world economy where most output is services. Agriculture and manufacturing have wilted away in the consumption of the global representative agent: to succeed in the world economy today requires prime attention upon services.

Rodrik says:

Consider India, which demonstrates the limitations of relying on services rather than industry in the early stages of development. The country has developed remarkable strengths in IT services, such as software and call centers. But the bulk of the Indian labor force lacks the skills and education to be absorbed into such sectors. In East Asia, unskilled workers were put to work in urban factories, making several times what they earned in the countryside. In India, they remain on the land or move to petty services where their productivity is not much higher.

As Rodrik points out, there are important gaps between the skills of the great unwashed masses in India versus China, where elementary technical training reached a larger mass of humans. In addition, China did better on core economic policy choices about (a) Removing protectionism; (b) Removing barriers to FDI; (c) Building hard infrastructure; (d) Labour law and (e) Rationalising taxation.

What policy advice would flow from this? India should not have have made these six mistakes in economic policy (low training for the masses, protectionism, barriers to FDI, weak investments into infrastructure, labour law and mistakes in tax policy). At the same time, this does not recommend a bias in favour of manufacturing. It is hard to discern a meaningful choice about emphasising services versus manufacturing in Indian economic policy. Participation in all global production is good. Governments should remove all barriers that inhibit global integration whether in goods or in services - e.g. the six mistakes in Indian policy sketched above.

A paragraph earlier, Rodrik says:

To be sure, some modern service activities are capable of productivity convergence as well. But most high-productivity services require a wide array of skills and institutional capabilities that developing economies accumulate only gradually. A poor country can easily compete with Sweden in a wide range of manufactures; but it takes many decades, if not centuries, to catch up with Sweden's institutions.

I would point out the contradiction: "A poor country can easily compete with Sweden in .. manufactures" but earlier it was asserted that the gaps in Indian skills inhibited India's ability to compete with Sweden in manufactures.

Doing things that push skills and institutional capabilities

I would go further to say that it is good to go after fields which require a wide array of skills and institutional capabilities.

I am reminded of Ricardo Hausmann's `Good Cholesterol' argument about financial globalisation as opposed to mere FDI. When a poor country operates in an institutional vacuum, foreign investors are uncomfortable, and the only thing that can happen is FDI. To obtain financial flows, the country has to build institutions: laws, regulators, property rights, and so on. This is a good thing! A country that gets to FDI and gets stuck there should ponder what is going wrong. In similar fashion, no country aspires to have low-wage production; every country wants to understand the secret sauce through which a part of the labour force can earn high wages by world standards.

As a country rises out of poverty, it is essential to build up skills and institutional capabilities. If policy makers hinder services and/or favour manufacturing, there is a greater chance of being stuck in low skills and low institutional capabilities. I am not proposing industrial policy in favour of services. I am only proposing the absence of industrial policy; we should avoid a `widget illusion' and foster more global integration without trying to push towards one industry or another.

In India, with 7 per cent growth, GDP doubles every decade. As a thumb-rule, I feel that a comprehensive transformation of skills and institutions is required across each doubling of GDP, which is roughly each decade for India. A country that is stuck in low-skill manufacturing will find it difficult to achieve the reinvention of this `soft infrastructure' of the mind. If policy makers tried to push a country towards doing low end grunge work, it would be harder to obtain these repeated transformations of institutions and the furniture of the mind, which would lead to growth decelerations.

As an example, in the article New wave of deft robots is changing global industry, John Markoff says:

Foxconn has not disclosed how many workers will be displaced or when. But its chairman, Terry Gou, has publicly endorsed a growing use of robots. Speaking of his more than one million employees worldwide, he said in January, according to the official Xinhua news agency: ``As human beings are also animals, to manage one million animals gives me a headache.''

The project of economic development requires sophisticated interactions between firms and workers. The laws, human rights and management practices that are required when dealing with humans are different from those required when running a firm with `one million animals'. I would hence argue that it is limiting for a country to focus on the political, legal and institutional requirements to produce a la Foxconn. It is better to confront the complexities of high skill, high wage production, and to build the environment for this to happen: in the political and legal system, in management practices of firms, and in the power structure embedded in a conversation between two citizens who are co-workers within a firm. Services production is a valuable learning ground where the complex management practices that involve high skill humans can be learned.

The new world of manufacturing

Rodrik correctly points out that manufacturing has become more sophisticated in recent years. This has some fascinating dimensions:

  • The rapid improvements in capabilities and declining costs of robots.
  • The rise of open source design coupled with 3-d printers. If a 3-d printer in the US fabricates a part close to its usage in an assembly line, while the labour-intensive design work ("services") that controls the 3-d printer is done in India, does this entail manufacturing or services work in India?
  • The world economy is likely to be in a low interest rate environment for a long time, which will encourage capital intensity worldwide (robots, 3-d printers), thus blunting the value of low wages.

Momentous changes are afoot, which challenge our traditional notions of manufacturing versus services. To some extent, we are even seeing some manufacturing go back to the US.

Things that might `go wrong'

Finally, Rodrik talks about reduced willingness in the West to tolerate unfair tactics like the Chinese exchange rate regime. I would generally consider this to be a good thing, both for developing countries and for the world. In any case, the Asian `Bretton Woods II' episode seems to be subsiding. As an example of the disenchantment with exchange rate distortions: From 2004 to 2007, India debated exchange rate rigidity, and walked away from it. The links between undistorted exchange rates and growth have not been adequately emphasised in the discourse. A developing country builds up inferior skills and institutional capabilities by exporting under a subsidised exchange rate: it is better to force firms to confront the market price and achieve the productivity required to participate in globalisation when facing an undistorted price vector.

He worries about a rise in protectionism in the West, but we have to admit that the 2008-2012 experience has been pretty good in this regard: by and large the West has not succumbed into protectionism. In 2008, all of us worried about Smoot-Hawley. Today, things seem to be be going well.


In summary, I would argue that we should avoid a `widget illusion'. There is nothing special about manufacturing or industrialisation: as long as people in India get high wage jobs, this is good. Getting there requries deep integration into the world economy, which includes policy battlefronts such as:

  • Openness to the Internet
  • Use of English
  • Inbound and outbound FDI
  • The array of cross-border financial services that are the enablers of complex globalised production of both goods and services
  • Globalisation-compatible tax policy on both trade and finance
  • The absence of either protectionism or mercantalism
  • Fostering high quality human skills, and
  • Infrastructure.

To the extent that globalised production of goods and services happens in areas which involve high skills and complex institutional development, this is a bonus, since any high growth country needs a rapid pace of reinvention of laws and institutions.

Most of this is the old orthodoxy. Policy makers worldwide are generally focused on these issues, as they should be. From the 1960s onwards, dirigisme has generally subsided, with the twilight of policies like fixed exchange rates, industrial policy, capital controls, protectionism, etc. These key lessons remain intact in the 21st century.

Saturday, June 09, 2012

Is building and running the IIT JEE a public goods problem?

What should government do?

In the question "What should government do?", economists have one big answer "do the public goods". A public good is something that is non-rival (the consumption by one does not come at the cost of consumption by another) and non-excludable (it is not possible to exclude someone from benefiting from the public good).

The regulation of air pollution is the favourite example which illustrates a public good. Clean air is non-rival (if you breathe clean air, it does not diminish my supply of clean air) and non-excludable (if the air is cleaned up, nobody can prevent me from breathing it in). Indeed, nothing that one person can do can make a difference to air pollution. Only the government can regulate pollution and this deliver clean air.

Similar issues arise with defence, police, judiciary, monetary policy, financial regulation, public health (though not the health of the public), biodiversity, etc., all of which add up to the economists' vision of what government should be doing.

What should government do in the field of education?

Education is substantially a private good. I study, I benefit. There are spillovers ("externalities") to others, and so there is a case for a government subsidy. But barring that, this is a field where the incentives are well aligned for each person to be the main one in charge of his own education.

Public funding solves the problem of externalities. At the level of elementary education, vouchers are a nice way to deliver public funding that is large enough to pay for the externalities. At the level of higher education, public policy can focus on economies of agglomeration alongside some public funding, nudging the outcome in India so that there are 100 high quality broad-based universities.

As I read The delicate technology of creating excellence by Pradip Ghosh in the Telegraph, I was reminded of the public goods character of testing and curriculum development. As he says:
in this very large country with a multitude of school boards and their non-uniform curricula and examination standards, it would be inappropriate to go by board grades because that would yield unreliable, undesirable results — we would not get the best students. And, such a course, therefore, would be unfair both to the aspiring students and to the institutions they would be entering. A single post-high school examination with a well-defined syllabus and a centrally administered paper-setting and grading system was thought to be the best alternative
The production of education services is a private good problem, to be sorted out between one student and one education provider. However, the problems of curriculum and testing have a public goods character. Let's run the tests of a public good, for a nationwide system for standardisation of curriculum and testing.

Is it non-rival? Does the consumption of the services of this system by one person diminish the amount of this system available for another? With computerised testing, there should be full scalability (though Pradip Ghosh argues, in the article above, that there are problems with this).

Is it non-excludable? High quality curriculum is non-excludable in that once curriculum documents are on a website, everyone can download them. Testing is excludable if you want to be cussed about it, but for the rest it should not be possible to exclude anyone from taking a nationwide test.

This argument guides us in thinking about what government should be doing in the field of education:

  1. Funding (calibrated to overcome the externalities)
  2. Curriculum development
  3. Testing
  4. Information infrastructure about education service providers (i.e. schools but also all sorts of new ways of organising education service delivery) so as to assist choice by parents and students.
The entire focus of management time, and the entire resources available for the task, should be devoted to these 4 problems.

Central government or local government?

Once we know that testing and curriculum are public goods, we have to ask who should do it.

If an important outcome (getting into the IITs) was linked to regional board examinations, that is a recipe for grade inflation. This is a reason for doing this at the central government.

There are economies of scale. A curriculum only needs to be developed once. This is a reason for doing this at the central government.

In conclusion, the IIT JEE has many problems, but the building and running of high quality examinations is an important task of the central government and should not be diluted or abandoned. The fraction of management time, and resources, that are devoted to curriculum and testing need to go up.

Tuesday, April 10, 2012

Path-breaking rules under the Right to Education Act, in Gujarat

by Parth Shah.

One major initiative of the Indian government, in the field of education, was the Right to Education Act of 2009. This act has major problems, as has been argued by numerous observers and experts in the field. This Act focuses on the interests of incumbent public sector education providers, instead of focusing on the interests of children and parents. It is focused on inputs into the educational process, regardless of the outcomes which are coming out. It penalises private schools that have weaknesses on inputs, regardless of the fact that these schools often induce better learning outcomes when compared with public schools.

At the same time, the translation of the Act into benign or malign outcomes critically hinges on the Rules under the Act, which are notified by State governments. Thus, now that Parliament has chosen to enact the RTE Act, the critical frontier that matters is how state governments choose.

In recent weeks, Gujarat notified its Rules for the implementation of the Right to Education Act (RTE) 2009. It has introduced some of the most innovative ideas for recognition of existing private unaided schools. The Committee in charge of drafting the Rules in Gujarat, that was headed by the former Chief Secretary Mr.Sudhir Mankad, has broken new ground in understanding the policy issues faced in education in India today.

Instead of focusing only on input requirements specified in the Act like classroom size, playground, and teacher-student ratio, the Gujarat RTE Rules put greater emphasis on learning outcomes of students in the recognition norms. Appendix 1 of the Gujarat Rules is the one which has a path breaking formulation for recognition of a school: this will be a weighted average of four measures:

Student learning outcomes (absolute levels): weight 30%
Using standardised tests, student learning levels focussing on learning (not just rote) will be measured through an independent assessment.
Student learning outcomes (improvement compared to the school's past performance): weight 40%
This component is introduced to ensure that schools do not show a better result in (1) simply by not admitting weak students. The effect of school performance looking good simply because of students coming from well-to-do backgrounds is also automatically addressed by this measure. Only in the first year, this measure will not be available and the weightage should be distributed among the other parameters.
Inputs (including facilities, teacher qualifications): weight 15%
Student non-academic outcomes (co-curricular and sports, personality and values) and parent feedback: weight 15%
Student outcomes in non-academic areas as well as feedback from a random sample of parents should be used to determine this parameter. Standardised survey tools giving weightage to cultural activities, sports, art should be developed. The parent feedback should cover a random sample of at least 20 parents across classes and be compiled.

This is one of the first times in India's history that public policy has focused on children and parents, instead of focusing on the public sector producers of education services.

Furthermore, the Gujarat RTE Rules have taken a more nuanced and flexible approach in other areas too. For instance, both class size and teacher-student ratio have not been defined in absolute terms, but in relative terms. The required classroom size is 300 sq feet but in case classrooms are smaller, then instead of re-building them, the Rules allow for a way to accommodate that with a different teacher-student ratio. The formula is: Teacher Student ratio = (Area of the classroom in sq feet-60)/8. This approach not only allows smaller classrooms to exist but also gives schools a more efficient way to manage physical infrastructure.

If a private school is unable to meet recognition norms, then the RTE Act de-recognises the school and forces it to close down. This sudden forced closure would create serious problems for the students and parents who would have to find a new school in the neighbourhood. The Gujarat Rules allow for the State to takeover the school, or transfer management to a third party, and create a genuine possibility for the school to continue and meet the norms. This, once again, shows the focus of the Gujarat Rules upon the interests of students and parents.

This approach is significantly better to that of the other states where recognition norms are based solely on input requirements and that are also rigid (like playground, classroom size and teacher-student ratio). The Gujarat approach recognises the substantial contribution made by budget private schools in urban and semi-urban areas where land and buildings are very expensive. Actually many government schools themselves would not be able to meet the rigid input norms that RTE has mandated.

Tuesday, February 21, 2012

CNBC Awaaz symposium on India's education crisis

India's education crisis is now high on our consciousness. In response to the recent developments, CNBC Awaaz did a four-hour symposium on national television (in Hindi/English) titled Kala Akshar: The Great Education Crisis. This brings together an interesting and diverse set of voices on the subject.

10:00 - 11:00 AM 1 2 3 4
11:00 - 12:00 PM 5 6 7 8
12:00 - 1:00 PM 9 10 11 12
1:00 - 2:00 PM 13 14 15 16

Tuesday, January 24, 2012

Education in India: A compact reading kit

With the first release of OECD PISA results for India, and with the release of one more year of Pratham data, there has been an upsurge in interest in education in India. The following set of materials are a useful reading kit to get a grip of the field.

Elementary education

Higher education

Education in India at the crossroads

The debate

Roughly one decade ago, there was a strong debate in India about how we should tackle the problem of education. There were two views:
On one side were those who felt that nothing was fundamentally wrong; all that was needed was more money. So we should just continue building more government schools and hiring more civil servants to act as school teachers, and we'll be fine.
On the other side were the reformers, who argued that the basic incentives in Indian education were wrong. Putting more money down a dysfunctional system was pointless.
The Intensifiers won this debate. An informal coalition of educationists (i.e. the incumbent education system) and leftists came together, supported by the World Bank, which pushed for mere enlargement of Indian education, without questioning the foundations.

All of us are involved in this story at many levels. At the simplest, we are the customers of the education establishment. We pay income tax and VAT and a few other taxes. On top of this, we pay the 2% education cess. In return for this, we get certain educational services. These influence our kids, and they influence all the young people that we encounter in this young country. Trillions of rupees have been spent, and more than a decade has gone by. It is time to assess the performance of this strategy.

Three blocks of evidence are now visible, which tell us that the Intensifiers were wrong. The old strategy, which was invigorated by a vast rise in spending, was the wrong one.

Evidence #1: OECD PISA results for India

This story is well told in a recent blog post by Lant Pritchett. Bottom line: The first internationally comparable measurement of what children learn has been done. The sample correctly includes urban and rural children; it correctly includes children going to private or public schools; there are no first order mistakes in what was done. It tells us that Indian education policy has failed miserably: the results have come out at the bottom of the world.

Evidence #2: ASER 2011 results

Pratham has been running surveys which measure characteristics of children and schools in rural India (only). Their latest survey results, for 2011 show the following facts.

First, rural kids learn less at public school. Here's a simple example of what the evidence shows. Surveyors ask kids in class III to recognise numbers upto 100. Here are the numbers, for the proportion of kids in class III who cannot recognise numbers upto 100:

In 2008, the failure rate with private schools was roughly 17 per cent. Government schools were much worse at over 30 per cent. A short three years later, conditions had deteriorated sharply in government schools. The failure rate had gone up to 40 per cent. Private schools had also worsened slightly, to a failure rate of 20 per cent. By 2011, a big gap had opened up between the two: private schools are failing to teach 20 per cent of the kids while government schools are failing with a full 40 per cent of their kids.

Parents in India face the choice between sending their children to a government school, which is free and serves a mid-day meal, versus sending them to a private school where they pay fees. Yet, an increasing fraction of parents choose to send their children to a private school, paying tuition fees from their own pockets, while government schools are free. The relationship between a parent and a private school is a transaction between consenting adults. The relationship between a parent and a government school involves all of us, because we are paying for it.

Given the low income of parents in India, their use of private schools is a striking indictment of what the Intensifiers have wrought:

At class II, the fraction of rural children in private school went up from 19 per cent (2007) to 23 per cent (2011). At class VII, this rose more slowly to levels slightly above 20 per cent.

Evidence #3: CMIE household survey

CMIE has data for the year ended March 2011 about the behaviour of 169,492 households, about their expenditure on school/college fees and tuition fees. Here's the picture for the quarter ended September 2011; all values as percent of overall expenditure:

Income class School/college fees Private tuition fees
Rich - I 4.79 0.66
Rich - II 3.79 0.51
High Middle Income - I 3.54 0.63
High Middle Income - II3.12 0.65
High Middle Income - III2.44 0.68
Middle Income - I 1.93 0.59
Middle Income - II 1.62 0.45
Lower Middle Income - I1.38 0.49
Lower Middle Income - II1.05 0.60
Poor - I 0.76 0.58
Poor - II 1.13 0.28
Overall 2.10 0.57

If parents chose to stay within public sector schools, their expenditure on fees would have been zero. The table shows that across all income groups of India, there is movement towards private provision of education, both by paying fees at schools and by paying for private tuition classes. These two elements add up to 2.67 per cent of overall expenses of households. (The CMIE household survey separately measures expenses on books, journals, stationary, additional professional education, education overseas, hobby classes and other education expenses. This helps us gain confidence in the extent to which the two fields in the table above narrowly pin down the feature of interest).

These decisions of well intentioned parents are the strongest indictment of education policy in India. The product being given out by the Intensifiers is such a terrible one, the parents of India are walking away from it even though it is free and the alternative is not and the parents are poor.


For more than a decade, the Intensifiers have controlled Indian education policy. They have said: Leave education to the education establishment, do nothing radical, just give us more money, we will deliver results. Now we know that they were wrong. They took the money, but failed to deliver the results.

Kapil Sibal has said that his ministry should not be held responsible for the stream of bad news that is coming out. To me, this seems to be dodging accountability. His ministry is responsible for Sarva Shiksha Abhiyaan, for the Right To Education Act, for blocking OECD PISA from being done in India, etc. The bureaucratic consensus of his ministry represents the education establishment.

The key phrase that needs to be emphasised today is accountability. If a contractor took money from you, and failed to deliver on building your house, you would sack him. (You would also take him to court, to recover the money that was paid to him, for services not delivered). In similar fashion, education is too important to be left to the educationists. We need to start over.

What is to be done

  • We need to start over in the field of education, with a fresh management team, one that is not a part of the status quo, one that is rooted in the worlds of incentives, public policy and public administration.
  • In 2004, we were told that in return for a tax rate increase of 2%, in the form of an education cess, we would obtain improvements in education. We now know that those improvements did not come about. Hence, that tax rate increase should go. (Even if sharp improvements in educational outcomes had been obtained, the education cess was a mistake in terms of basic public finance, and needs to go. Public expenditures on education should simply come out of general tax revenues; there is no need to have a cess.)
  • The flow of public money into the status quo needs to go down sharply. There is no reason to put money into something that fails to deliver the goods. First we must prove that a mechanism delivers results, and only after that should we put money into it. This is the common sense that a housewife would apply. She would not spent gigabucks on promises from people who have failed to deliver.
  • OECD PISA measurement needs to take place every year at every district. The production of this data is a public good that the government can and should do. It can be fully contracted out to private firms so as to avoid the problems of public sector production. Datasets about student characteristics and school characteristics should be released, covering every district and every year, so as to enable research.
  • Civil servant teachers, who have tenured (permanent) have no incentive to teach well, regardless of their qualifications or high income. We can't sack them, but what we need to do on a massive scale is to stop recruiting them. The existing stock can be reallocated to other civil servant functions where staff is in short supply. Through this, it would become possible to whittle away at the accumulated stock over the coming 20 years.

Tuesday, March 08, 2011

Education research opening at Centre for Civil Society

The Centre for Civil Society (CCS) is an independent public policy think tank based in New Delhi working on research, advocacy and outreach on critical public policy issues that affect India. CCS is currently looking for researchers, advocacy specialists, and program coordinators to help take the Centre to a higher level. We are recruiting for a Research Coordinator to manage our research agenda, focusing on our ‘Education Reforms Initiative’ and the ‘Law, Liberty and Livelihoods' campaign. The Research team at CCS is responsible to help develop, capacitate, and deliver high-quality research focusing on education reforms and livelihoods in line with the goals of the Centre. For more information on this and other open positions, please look us up on the web.

Thursday, January 29, 2009

First measurement about crime and police

To economists, the role of government is to deliver `public goods'. A public good is something that is non-rival and non-excludable. Something is non-rival when my consumption of it doesn't detract from your consumption of it. Non-excludability is the inability to exclude a person from benefiting from the public good.

Law and order is a public good. When a city is safe, everyone benefits: the safety that one person enjoys doesn't detract from the safety that another person enjoys. And, it isn't possible to exclude any one person from benefiting from this safety.

In recent decades, economists have pushed for greater government involvement in `human development' covering issues such as health and education. Most activities in this field are actually not public goods. The bulk of educational services and health services are private goods: I get healthier or smarter, I benefit, it is a private good. Hence, the economic logic of pushing the government into this field was always suspect [link].

In India, we have had this great government interest in human development while - at the same time - we've had a worsening of the police and judiciary. While law and order is a public good, the performance of the State in this field has been getting worse. While most of human development is not a public good, we've seen an increasing focus and resource outlay in those areas.

The terrorist attacks in Bombay brought forth an outpouring of interest in stopping terrorism. But it is not possible to do this through superficial changes; genuinely achieving safety requires fundamental change to the police and judiciary.

I find it useful to compare and contrast the judiciary against the election commission. If the election commission came to us and said that conducting a general election would take 20 years, there would be an uproar and the entire leadership of the election commission would be sacked. It has not been easy, but the election commission has figured out how to run a fairly clean general election within roughly two months. Compare and contrast this with the non-performance of the judiciary. Why are we willing to accept non-performance from the judiciary of a kind that we would never accept from the election commission? If anything, as Fareed Zakaria has emphasised, a healthy democracy is more about good courts than it is about good elections.

The challenge in India is that of reining in the State, getting it to perform on the police and judiciary, and shift the focus of the State away from the pleasures of recruiting teachers who do not teach.

One of the first pieces of work that I have seen on the police in India was done by the MIT `Poverty Action Lab', and is written up by Jacob P. Koshy in Mint [slideshow].

I was very impressed at the fact that the research involved conducting a `crime survey'. Once again, there are a zillion surveys in India on issues like poverty or health, but none on the critical business of police and judiciary. In my knowledge, this is the first survey evidence on the experience of citizens with crime and the police. In the US, such surveys are called `crime victimisation surveys'.

Here are some gems from this. There are four districts that they report (Dholpur, Kota, Chittorgarh, Jaipur) where one in ten households (or worse) reports direct experience with one crime. I'd be curious to know what comparable international values are. In districts like Udaipur, Nagaur and Barmer, this value is 5% or lower.

Only 21% of the victims reported the crime and got an FIR registered.

13% of the victims were completely satisfied and 14% of the victims were satisfied with what the police did. The remaining vast majority were unsatisfied at what the police did. This is a massive vote of no confidence in the police. A full 82% reported that no beat constable ever visits their village or neighbourhood.

The slideshow goes on into statistical measurement of a few innovations in how the police could be made to function better. As with the bulk of this `development policy through randomised evaluation' literature, I'm underwhelmed at the usefulness of such simplistic schemes for making government work better. I have worked in government, and I have worked on reforming government from outside government, and this is not the way fundamental change is achieved.

In short, I was very impressed at (a) the crime victimisation survey, which is a big step forward, and (b) the fact that more people are taking interest in the most important public good of all. All of us should be pushing the top leadership to put more time and focus and resources into true public goods (e.g. law and order) at the expense of areas which are not public goods (e.g. most of human development).

Wednesday, January 14, 2009

Right to Education Bill, 2008

Read Ila Patnaik on this proposed law. While on this subject, you might like to see my five-part classification scheme on education policy. I guess the education bureaucracy is keen to go from #3 (State production but do no harm to the private sector) to #4 (State production while trying to actively damage the private sector).

Saturday, August 23, 2008

Certifications, IT

In the early 1990s, if you had looked at the potential for an IT industry in India, the answer could easily have been: "Not possible, because there aren't enough decent colleges and universities around". Yet, the IT industry happened. I find the build up of the requisite human capital of this industry to be rather interesting. Going beyond IT, more generally, if you look at the tripling of India's GDP and the immense internationalisation in this period, it seems to have taken place against daunting odds given that higher education is dysfunctional. For almost all students in India, the knowledge they command when stepping into the labour market is not substantially different from the knowledge they had when finishing 12th grade; the undergrad / masters degrees add next to nothing. This could have held back India's growth. It didn't.

How do we explain the tripling of GDP? The bulk of human capital in India has been built by self-motivated people. This was largely learning-by-doing - as economists have always known - supported by a little timely training. This process can be speeded up through the use of high quality certification exams, that really challenge the exam-taker, and that push forward to international standards of quality. Both processes - learning by doing and certification - are largely independent of formal education.

I read a column by Robert Cringely which links up to these issues. He talks about a certification in computer networking : ``Cisco Certified Internetwork Expert'' (CCIE). As he says:

The CCIE is Cisco's top certification category and VERY hard to earn. Being a CCIE doesn't mean you have Len Bozack on speed dial, but it might as well. Cisco products dominate the Internet and CCIEs are Cisco gurus, so if you are serious about the Internet as a nation you'll have CCIEs hanging about, or that's the theory. Conversely, if you just talk a good game as a country with technological aspirations, maybe you won't have many CCIEs at all -- maybe none. It's one way to determine who the posers are.

Cisco publishes the number of CCIEs by country every quarter. Here's what it shows for India:

2000 4
2001 17
2002 39
2003 72
2004 102
2005 115
2006 144
2007 200
2008 410

It shows a nice exponential build-up, despite roughly zero progress on higher education in this period.

In the case of finance, I know that the headcount of FRM certification is a useful measure of the number of people in India who actually know some finance (does a time-series exist?). I don't know anything about the CCIE. Is it a very Cisco thing? If this is the case, fluctuations in the number of CCIE would partly merely reflect the success or failure of Cisco in India. My informal impression is that Cisco is much less important in price-sensitive-India than they are elsewhere in the world.

The Cisco data also has numbers for other countries. I'm curious what blog readers think about (a) The extent to which the headcount of CCIE can be interpreted as a measure of the presence of high quality skills in the country and (b) A common sensical interpretation about inter-country comparisons - e.g. China's count is 8x bigger than India but China hasn't made the same impression in the international software market.

Thursday, August 21, 2008

Emphasise certifications as a means to building human capital

Given India's terrible elementary and higher education system (ponder the implications of this also), a greater emphasis on certification rather than education might make a lot of sense. Employers want to know that a person has the correct knowledge. An exam should give employers the confidence that this is indeed the case. In India, young people are very good at studying on their own for examinations. Once you start thinking like this, school is a low quality use of your time. I always tell young people in India: never let school come in the way of your education.

Think of the best 50,000 students at the IIT JEE. If they have a good certificate in their hands saying they are really top quality high school students by world standards, that's useful in signalling. And, this knowledge was largely picked up outside of the school system, which is calibrated to a much lower standard.

An excellent example of a certification mechanism which has helped remove entry barriers into the labour market is NSE's NCFM examination. I have met so many young people with weak educational backgrounds, who were able to jump this barrier and start off into successful careers in the financial sector, thanks to this certification exam. And, I have written about the GARP FRM before: it's an internationally recognised certification which delivers better knowledge than any MBA in India.

Saturday, August 16, 2008


Michael Walton has a good opinion piece in Financial Express on the subject of inequality in India. He starts out with the eminent good sense on the subject:

Should we care about inequality? Many would see this as a dumb question. But there are two, opposing views on why it may be considered dumb. The first is that inequality is so obviously central an issue, so pervasive in Indian society that it is transparently the case that tackling inequality is central to the development process. Indeed, the long history of both rhetoric and action by the Indian state is, ostensibly, in line with this view.

A second view is that inequality is a big diversion. India is still a poor country. The first order question is sustaining growth, while ensuring the poor participate in that growth. (And, indeed, in most growth episodes, most of the time, the income of the poor grows more or less as fast as the average.)

Moving from Latin America to India a year ago (as I did) casts this question into relief. At first glance the contrast seems to support the view that inequality is a second order question here. Latin America is the region of inequality par excellence. Measured inequality in India (from the National Sample Survey) is way below the Latin American average, and even below the most equal society theretiny Uruguay, famed for its extensive social insurance system. It is also lower than in China, Malaysia and Thailand. Sustaining rapid growth looks much more important than any feasible redistribution, not least for the poor.

On the scale of decades, the only thing that matters for poor people is GDP growth. Every percentage point of the growth rate that is given up in the quest of reduced inequality today does permanent damage of the trajectory of consumption of poor people in coming decades. As an example, suppose a family starts out in poverty in India today earning Rs.1000 a month. Here is what 30 years of growth does at a few alternative growth rates:

2.5% 2,097

6% growth over the coming 30 years gets a family making Rs.1000 a month in India today up to Rs.5,744 a month. But if the growth rate gets up to 7%, then this same family gets up to Rs.7,612 per month. These large differences are induced by small improvements in the growth rate. And, if we sink into socialist stasis of 3.5% GDP growth which would yield 2.5% per capita growth, then this family gets up to Rs.2,097 a month in 30 years.

I think 30 years is the right horizon to think about these issues, for it corresponds to the change that a person sees from age 0 to age 30, and then again from age 30 to age 60. It is very important for poor people in a country to see such massive changes in their well being coming about within such timescales.

Policymakers who care about the consumption of poor people have to have a very short-sighted discount rate (in addition to certain kinds of preferences) in order to espouse policies that emphasise equity at the price of growth.

Manish Sabherwal has a piece in Economic Times today, where he points out the logical fallacies of simplistic beliefs about inequality using a natural experiment that recently took place in India:

India has become substantially more equal since January 8, 2008. About two hundred billionaires have turned into millionaires. The drop in stock market and real estate values means that the top 5% richest people may have lost about 40% of their wealth and making the rich poor increases equality. But does this exponential increase in equality help India’s poor? Do they even care?

Michael Walton then runs through the criticisms of this perspective. First, inequality off NSSO data is underestimated. I feel that inequality using household survey data is underestimated everywhere. It's as hard for an investigator in Mexico City filling out forms to get into one of the haciendas of the rich, or an investigator in New York City filling out forms to get into a penthouse in Manhattan, as it is for him to get into a prosperous home in Bombay.

Second, the really important thing is inequality of opportunity. I fully agree. Here, it is possible to obtain first best with reduced inequality of opportunity and higher GDP growth, by undertaking fundamental reform of education and health in India. This is a rare situation where policy makers with an interest in one kind of inequality can actually assist growth. But so far, the gigantic spending programs in health and elementary education (e.g. Sarva Shiksha Abhiyaan) have been dominated by the interests of their respective civil servants, and not the interests of the users of these services. Higher education is another great opportunity to make a difference to equality of opportunity, but so far in India, the policy framework is systematically designed to disadvantage a person born in a family with weak financial and human capital.

Third, he says that inequality is likely to get worse, and that this will have implications for the political system and society. I agree with this outlook. Inequality in India will get a lot worse before it gets better, given that the top decile is in the process of plugging into globalisation, and generating tremendous wealth in the process. Over time, the remaining nine deciles will handsomely reap the fruits of this transformation, but in the short term, inequality will worsen.

Vijay Kelkar has always emphasised that inequality matters because it influences the political foundations of liberal economic policies. I see this as shaping up to be a big challenge for Indian politicians over the coming 25 years. So far, the early indicators are bleak.

In her article The first globalization: Lessons from the French, Suzanne Berger emphasises one of the `great surprises in history' : despite the inequalities created by capitalism, no electorate has ever voted in free national elections to overturn it. She argues that while anti-capitalist political parties have been strong, they have never won the day.

I'm not so sure. Democracy does contain the possibility of the demos coming together and expropriating private property and trampling on individual freedoms - as Indira Gandhi did and as a number of `idiotic' regimes in the world have shown. Berger emphasises one element of what holds such dangers in check:

...the constitutional engineering of Madison and the founders of other liberal democratic societies did work to protect the rights of individuals and the functioning of a market economy. Institutions like the Bill of Rights, the Supreme Court, and federalism did in fact build dikes that protected property and markets against democratic majorities

To the extent that the `constitutional engineering of Madison' is important in throwing up these dikes, we are in trouble in India, given the extent to which constitutional engineering (and re-engineering) have systematically damaged the rights of individuals and the functioning of the market economy.

Sunday, May 04, 2008

What value does a university add?

In judging what a university does to a student, a key problem lies in distinguishing differences in raw material vs. differences in training. Do IIT graduates do well because they were smart to start with, or because they learned something at IIT? The resources being expended by society on universities are misplaced if the universities aren't adding value. Michael Spence showed us the value of signaling in the labour market in 1973: it may be rational to employ IIT grads even if the university teaches them nothing.

It's possible to get stuck in an equilibrium where employers value a university because it attracts bright entrants, and the university attracts smart people because this is a good signal on the labour market. The university can then be a prodigal waste of resources, that doesn't add value to students, but the situation can be an equilibrium.

How does one break out of these situations? If there was a strong signalling value to getting through the CAT, but ISB taught me more than IIM, then I could put my CAT rank on my resume, and go on to study at ISB instead. This gives me the best of both worlds: the signalling value of the CAT coupled with the education of ISB.

David Friedman has a blog post in which he shows a situation where a small change in policy design can possibly obtain a big impact on the way the world works. His setting is law schools in the US. The innate ability of students is measured using the LSAT examination, and then they try to go to various law schools. But all this is a means to an end: The end is winning at the bar examination.

One proposal to measure the value of the school, which is at a proposal stage, is to require the school to disclose bar passage rates. Think of this as asking coaching classes to reveal the fraction of their kids that get into IIT.

David Friedman rightly points out that bar passage rates depend on a combination of the raw ability of students and characteristics of the school. This problem is easily solved: ask schools to disclose bar passage rates as a function of LSAT scores. This way, if I have a given LSAT score, I can shop amongst schools that give me the maximum bang for the buck in terms of bar passage rates at my LSAT score.

This is a rare situation which affords such a neat solution. In most situations, it is much harder to disentangle the raw ability of the student and the contribution (or lack thereof) of the school. I wonder if there are other such situations where clear measurement is possible.

More generally, I think that the conventional university (e.g. MIT or IIT) reflects a bundle of services that may have been appropriate in a different age. It is important to rethink this bundle from first principles, reflecting the way new technologies are making it possible to rearrange information and incentives. As an example, I'm fairly convinced that we should shift away from each university having its own lectures, just as we have shifted away from each teacher having his own lecture notes.

Wednesday, June 20, 2007

Crisis on skilled labour?

There is a chorus of protest amongst employers in the country about the high price of skilled labour and high staff turnover. Some people think that India's deep public policy mistakes on higher education will greatly hinder GDP growth. I like to point out that over the twenty year period 1987-2007, the higher education got worse (since salaries in universities fell behind those in the private sector), but GDP tripled all the same. There is clearly much more going on in achieving skilled labour than good universities. I wrote an article in Business Standard today titled Crisis on skilled labour? on this subject, where I offer an optimistic view about how the present crisis on skilled labour might play out.

Sunday, June 10, 2007

Reservations as cancer

Business World has an editorial titled Reservations as cancer where they say that reservations were always unjust -- now they create only misery and jealousy all round without political advantage to any party:

It is an historical accident that Meenas were included amongst scheduled tribes and Gujjars were not. There was a time when Meenas were like any other Rajput clan. They built Amber fort, which commands the approaches to Jaipur. In the 16th century, Baharmull Kuchhwaha, a Rajput king, transferred his allegiance to Akbar, and with his help, destroyed the Meena kingdom of Naed. The feud continued for four centuries. When the British came, Rajput kings allied themselves with them to defeat Meenas, who lost their kingdoms and turned to robbery. That is how they ended up in the list of criminal tribes. Later, when the British ceded power to nationalists, the label, “tribe”, proved a godsend. It brought Meenas reservations in the civil service and education, and proved to be their entry ticket to the lucky Indian middle class. Today, Meenas are well represented in the civil service, and are turning to business.

Gujjars have a less distinguished history. They were originally nomads spread across the dry tracts of western India and Pakistan from Kashmir to Karnataka; it is possible that they came from Central Asia, perhaps Georgia. There are two subcastes of Gujjars: dodhi and bakarwal, or buffalo-keepers and goat-herds. When India was sparsely populated, they used to take their animals up to the Himalayan foothills in summer and descend back into the plains of Punjab and Uttar Pradesh in winter. Now that settled population has grown and grazing grounds have shrunk, they are being forced to take to more sedentary occupations. But not being a landed community like the Meenas, they do not have steady incomes or family support and have not invested as much in education. As a result, they have found it difficult to climb up the social ladder.

When the mutiny broke out in 1857, Gujjars were amongst the most energetic rebels; as a result, they had their share of hangings and dispossession, and earned their place in the 1871 list of criminal tribes. But somehow the curse of the British did not turn into a blessing of the Congress on the advent of independence; Gujjars were not included amongst scheduled tribes.

It is thus an accident of history that Meenas are a scheduled tribe and Gujjars are not. Meenas did the right crimes in the 19th century to earn their place in the fortunate category of tribes; Gujjars somehow fell through the cracks of history. This is no justice; it is sheer chance.

This government does not believe in justice; it is prepared to take a chance. Its resolve to shower favours upon Other Backward Castes (OBCs) is a perfect example. OBCs are so close in their social parameters to the main population that if they deserve reservations, so does almost everyone. They are backward only in name; if they are backward, there are no forward castes, except politicians. But they are a big vote bank; reservations are the way the Congress hopes to get their votes. Hence, the government’s opportunistic move. Mayawati came to power by giving sops to the most forward caste; the Congress does not want to be left behind in opportunism.

But here too, Gujjars are unlucky. They are not numerous enough for the government to bother. There are many groups and castes related to Gujjars — after all, Gujarat calls itself the land of Gurjars — but they would rather hide their kinship to the poor Gujjars.

The looming civil wars of India are not over class as the Prime Minister claims. The working class may have fought bloodthirsty capitalists in the textbooks he once read; but in the India he rules, it is castes that are fighting over the right to undeserved jobs and places in educational institutions. The way to douse their wars is to leave caste behind, and to abolish reservations. Reservations were originally introduced for an opportunistic reason: the Congress wanted to wean away Untouchables from Ambedkar, and to persuade them not to convert themselves to Islam and Buddhism. So it gave them reservations — but only if they were Hindus. There are Muslim Meenas, called Meos; they were excluded from the reservations.

For reasons of political advantage, the Congress divided the people by caste and religion. But now those divisions it created are causing bloodshed and havoc. There is no more political mileage in them; instead, there is only trouble. Even sectarian political parties must see that the time for favouring vote banks has passed.

No politician likes to take a radical decision, least of all the Prime Minister. It may be beyond him to abolish reservations. But he should at least begin to reduce the reservation percentages. If he wants to profit politically, he can replace them at the margin by reservation for the meritorious poor. Let him practise inclusive growth.

Update: Arvind Subramanian wrote in Wall Street Journal Asia suggesting that the State shift from quotas to targeted education vouchers.

Thursday, April 26, 2007

Don't sweat the quant stuff

We generally think that workers in India are good at the quant stuff. But is this just stereotype, or is it real? On one hand, 12th standard CBSE mathematics - which the ordinary Indian worker has generally mastered - exceeds what high school students in the US know. India has a significant edge here against the US, though not against Europe. On the other hand, college education in India is pretty awful, and this has generated skepticism about the potential upside for BPO in India. Very bright MBAs in India tend to be underskilled in probability, statistics, optimisation, financial economics; they tend to use spreadsheets too much. Their counterparts - the best MBAs from top schools in the US - seem to be better trained in these four critical blocks of knowledge. (I am comparing like to like: MBAs from the 3 best IIMs and ISB against MBAs from the top 10 schools in the US).

Thanks to Tirthankar Patnaik, I saw mention of an intruiging study by Ravi Aron of Wharton (link):

Questions abound about job functions that are likely to gravitate to places like India sooner than others. Some insights into this can be had from the concept of a "mechanism for complexity arbitrage" that Aron and colleagues worked on in a recent research project funded by the Mack Center at Wharton. The project involved collecting data covering three years from several companies about various processes and the places where they were performed across the globe. Managers at these companies overseeing those processes where then asked to rate each of them on a complexity scale of one to seven, with seven being the most complex.

The researchers found a striking polarization in the way managers assessed the complexity levels in their processes. Those in the U.S. and the U.K. formed one group, while their counterparts in India, China and Singapore made up the other. "Whenever work involved number crunching, quantitative analysis, mathematical formulation, statistical data analysis and numerical calculations, managers in India, Singapore and China would say this is low complexity work," says Aron. "They could find the people to do those jobs and deliver quality on scale, month after month, year after year. But wherever work involves persuasion, communication, context-sensitive responses, interpretation, subjective judgment and cultural sensitivity, managers in the U.K and the U.S. will tell you that is low complexity work."

Are there natural limits to the kind of work that can be offshored? Of course, says Aron, but he adds that Indian service providers will not encounter them "as long as they do not breach the complexity divide." He says certain job functions are a natural fit for the U.S. and the U.K. and not easily outsourced, because they require customer interactions and persuasion. These are activities that are "embedded in the context of the market and they need context-sensitive communication," he says. "The Indian market is nowhere as sophisticated as that in the U.S.; it does not have the range of derivatives like in the U.S. market, and futures and options are just beginning to catch on.

Also see here, another demo of the external perception of Indian quant prowess. How fascinating. Managers in India, Singapore and China think that the quant stuff is "low complexity". :-) I believe similar things are happening in Silicon Valley: many startups are organising themselves as a marketing front-end in the Bay Area, and the engineering team in India. If these effects are real, global capitalism will reshape the geographical allocation of production: activities involving the quant stuff are going to increasingly move to India, Singapore and China. This bodes well for MIFC. If you have the MIFC book handy, see page 172-176 on human capital which offers a gloomy perspective on this, and Chapter 5 on BPO. Both segments of the MIFC book paint a relatively gloomy picture: but maybe (in the light of the above) things are actually a bit better than portrayed there.

Once such a process gets going, it will be self-reinforcing, and it will induce effects that are reminiscent of the specialisation seen in evolutionary biology, where small mutations are accentuated by natural selection, and species separate themselves out in plying different trades. In the decision-making of the global firm, tasks that involve quant skills would be more likely to get sent to India/Singapore/China, while tasks that involve more cultural context are more likely to get done in an industrial-country setting. In the decision-making of the individual, people with quant predilections are more likely to move to India and people in India face a higher marginal product of investing in quant skills. Both these effects (at the firm and at the individual) would reinforce each other.

How will human capital build up in the context of terrible colleges & universities? I think learning-by-doing is the key. The last paragraph of the above quotation rings true to me. With weak colleges and universities, learning is dominated by on-the-job accumulation of skills. I would advise every victim of the terrible higher education system in India to intern, intern, intern and build up on-the-job skills, accompanied by a higher intellectualisation based on a self-study program of reading the best books and textbooks so as to not become a narrowly-specialised technician [link].

Index futures trading is active in India, so there is now plentiful talent that knows how to do index arbitrage. Interest rate futures trading in India is effectively banned, so naturally nobody knows anything about interest rate futures arbitrage. As the local financial system bulks up to higher levels of complexity, this feeds into the range of skills where there can be a global role for Indian professionals. There is thus a three-way synergy between (a) the BPO, (b) the local financial system and (c) the MIFC agenda: each feeds into the other. Conversely, a strategy of financial repression, where products and activities are banned in the domestic economy, has bigger costs than meets the eye: it hurts not just domestic finance but also the BPO and the prospect of MIFC.

Wednesday, March 28, 2007

Reforming the `universal services obligation' in Indian telecom

For many decades, telecom policy in India was distorted based on two excuses: the need to utilise this lever of power to "help poor people" and the need to "protect national security". What was really going on was the capture of an area of the Indian economy by the Department of Telecommunications, which was policy maker, regulator and service provider, all rolled into one. If you wanted to start a mobile telephone company, you had to come to DOT for permission, and they would say no. Extravagrant claims were made about systemic stability and the interests of poor people, in order to defend an entrenched status quo.

One of the singular achievements of the late 1990s was the political capital that was deployed at taking on DOT and breaking this monopoly. I don't think anything has happened to national security, and the results were miraculous for poor people. Decontrol, competition and the entry of private + foreign companies has done more for the interests of poor people than decades of socialism did.

That left the `Universal Services Obligation', a tax which has to be paid by all telephone companies in order to support telephony for poor people. I have previously argued that such taxation is unfair: if the government wants to run such a program, it must do so on-budget and not by charging an excise on just one sector. The income tax and the GST are the most efficient taxes; it is distortionary to have any other tax which penalises any one industry.

Earlier, USO was a cost-plus scheme where telephone companies were reimbursed absurdly high "costs" associated with placing a telephone in a village. Today Business Standard has a great edit about the benefits of shifting USO to an auction based and technology-neutral system:

Less than five months after the government amended the law to allow the Universal Service Obligation (USO) Fund to be used for subsidising even mobile phones and broadband services in the rural areas, the results are there for all to see. Mobile phone firms have come up with bids that ask for a subsidy that is a fraction of the benchmark price. In the case of Cluster 1, for instance, a benchmark price of Rs 394,967 had been set as the cost of building a mobile phone tower from which rural mobiles would be serviced; against that, the winning bid from GTL Infrastructure is for Rs 197,484. In the case of Cluster 10, to take another example, the benchmark price was Rs 423,860 while BSNL has won the bid by asking for a subsidy of just Rs 80,222. As for the other part of the subsidy (required for the electronics on the tower), most bidders have said they dont want itthat is, with the cost of the physical towers having been met, they are confident that they have a business model that can provide phones to rural India at costs low enough to bring in sufficient demand. Admittedly, this does not mean that the rural telephony model is robust enough to do away with subsidies altogether, but the heartening point is that the level of subsidy required has come down dramatically. It is reasonable, therefore, to expect a surge in rural telephony, with the attendant gains in connectivity and productivity that can be expected to follow when this happens.

Contrast this with the earlier situation in which the USO Fund was available only for fixed-line phones. This was exorbitantly costlythe Telecom Regulatory Authority of India (Trai) had estimated in 2005 that this model would need nothing less than Rs 30,457 crore by 2010 to set up all of 28 million rural phones, thereby reaching the rural tele-density target of a modest 4 per cent. Those numbers were forbidding, and that is when Trai came up with the model of subsidising the tower, which would service more users at a fraction of the cost. The other negative aspect of the earlier model was that, since fixed lines require physical digging and laying individual lines, progress was slow. Of the Rs 14,000-odd crore that will be collected by month-end through the five-year-old USO levy on phone services, just Rs 4,500 crore is expected to be spent! Since setting up mobile phones is faster, such backlogs can now be tackled.

The lesson this holds with regard to government subsidies in general should be obvious, namely that it is unwise to get married to particular technologies or policy prescriptionsif a new or different technology or method provides the same result, it should be adopted. To extend the logic to the education sector, for instance, if privately-run schools can ensure the same or a high quality of education as compared to government-run schools, channelling subsidies to private schools is clearly a good idea. The other lesson is that it pays to try and work through the market. Instead of fixing a subsidy based on normative cost, potential service providers should be asked to do a reverse-bid on what they require to provide the same servicethis is the only way to bring into play those efficiencies that are normally associated with free markets.

Tuesday, February 28, 2006

Five alternative frameworks in education policy

There has been much interest in scaling up programs like Sarva Shiksha Abhiyan and the midday-meal program. At the same time, the recent measurement of what children know, done by Pratham has shown a large-scale failure in what students actually know.

I wrote a column in Business Standard today where I describe five alternative frameworks of education policy:

  1. Do Nothing,
  2. Augment Purchase,
  3. State Production But Do No Harm,
  4. State Production While Damaging the Private Sector and
  5. Ban Private Participation

With higher education, we are on the 5th (ban). With elementary education, we are now veering from the 3rd ("do no harm") to the 4th ("damaging the private sector") by new efforts at having State-enforced quotas in private schools.

I argue that internationally standardised test scores need to be made the foundation of education policy, as opposed to efforts like SSA which have concentrated on spending more money on public sector education. The choice of which of the five frameworks is best should be based on which appears to deliver adequate test scores in a cost-efficient manner.

Our loyalty needs to be with the interests of students instead of the interests of the existing producers of educational services. There is an innate conflict of interest in the control of education policy with incumbent educationists.