Tuesday, March 31, 2015

The hot-hands fallacy revisited

Aeon has a nice essay that points to recent empirical studies which appear to question one of the most famous cognitive biases - the 'Hot-Hand fallacy', or the mistaken perception of lucky streaks among sports persons.

In the first study, Andrew Bocskocsky, John Ezekowitz, and Carolyn Stein, examined videos of 83000 shot attempts from the 2012-13 US NBA season, and argue that a player in form is emboldened to take more difficult shots,
They showed, first of all, that players who felt 'hot' did in fact start taking harder shots. And, after controlling for the difficulty of each shot selected, they found a small yet significant hot-hands effect - that is, those who did well began to do even better over time. 
Another study, by Jeffrey Zwiebel and Brett Green, analyzed twelve years of data from Major League Baseball, and found that
how a player performed the most recent 25 times at bat was a significant predictor of how he would do the next time. They also calculated that a hot player was 30 per cent more likely to get a home run than if he were not on a winning streak. Lucky streaks are real and not just an illusion, they said. 
Yet another study by Juemin Xu and Nigel Harvey, which analyzed about half a million sports bets, found that those on winning streaks were much more likely than not to keep winning, and those on losing streaks were more likely to keep losing. They tried to understand the reasons why these streaks persisted,
As soon as bettors realized they were winning, they made safer bets, figuring their streaks could not last forever. In other words, they did not believe themselves to have hot hands that would stay hot. A different impulse drove gamblers who lost. Sure that lady luck was due for a visit, they fell for the gambler's fallacy and made riskier bets. As a  result, the winners kept winning (even if the amounts they won were small) and the losers kept losing. Risky bets are less likely to pay off than safe ones. The gamblers changed their behaviors because of their feelings about streaks, which in turn perpetuated those streaks. 

Sunday, March 29, 2015

China cement consumption fact of the day

An old graphic, from Vaclav Smil, via Bill Gates, that puts in perspective the fantastic growth of China
In 2013, China's cement production was 2.42 bn mt. In contrast, India, the 2nd largest producer globally, produced just 280 mt!

Friday, March 27, 2015

Debt laden corporate sector and India's recovery

Following on from the dismal story of debt-laden balance sheets of real estate firms, Business Standard has a nice graphic that captures the magnitude of indebtedness in corporate India,
It is not so much the share of corporate debt as percentage of GDP that is frightening. While rising, non-financial corporate debt is still below that in many other large economies. The more worrying parameters are the high debt-to-equity ratios - in highly unsustainable double-digits for some of the largest infrastructure firms - and the low debt-service coverage ratios (DSCR) - at end-2013 nearly 40% of the corporates had DSCR lower than one, or inadequate earnings to cover even interest payments. Add to this the rising exposure to unhedged short-term external debt due to the liberalization of External Commercial Borrowings (ECB) inflows, and corporate India has to navigate very strong headwinds.
Their high indebtedness, and the resultant applications for corporate debt restructuring has led to many of the largest infrastructure companies from being barred from bidding for new projects. Of the 45 core infrastructure companies, 20 are currently in CDR and have been barred from contracts in many states. It is not just infrastructure, automobile sales, especially those of cars, has been stagnant for five years. With interest rates being sticky downwards, the sales growth of consumer durables and vehicles face a daunting challenge.

Corporate de-leveraging is therefore an essential requirement. But it runs counter to the government's expectations that predominantly privately financed infrastructure spending will do the heavy lifting for restoring economic growth to near double-digit rates. In any case, with banks hobbled with rising non-performing and distressed assets, credit availability for corporate India looks extremely uncertain. 

Wednesday, March 25, 2015

The "re-nationalisation" of Network Rail in UK and lessons for India

The Business Standard reports that the Bibek Debroy Committee constituted to make recommendations to reform Indian Railways is likely to advocate private participation in the running of passenger and freight trains. In this context, it may be instructive to take stock of the latest problems being faced by Britain's controversial rail privatization.

For the record, the British rail privatization had unbundled the system into three parts. An infrastructure company, Network Rail (earlier Railtrack) owns the track, rolling stock operating companies (ROSCOs) own the trains, and train operating companies (TOCs) run the trains. There are currently 17 TOC franchises (down from the original 25), operating both long distance inter-city routes as well as urbanized commuter lines. The ownership of the franchises have changed hands many times. 

I have already blogged about a comprehensive study on British rail rolling stock private franchising which found large private profiteering without capital investments and costing large subsidy outflows from the government. It found that the franchising policy boosted the profits of privately owned TOCs, without investments in network expansion or upgradation, and loading debt into the publicly guaranteed Network Rail. Faced with rising train ticket prices, the British government was forced to intervene by capping commuter fares in 2013.

As regards rail tracks, the British rail regulator, Office of Rail Regulator (ORR), has criticized the recently "re-nationalized" monopoly rail infrastructure operator, Network Rail (NR), of poor performance and declining standards in all aspects of its business. The ORR has said that NR is failing badly in meeting its new investment commitments on electrification and signalling, and standards on punctuality, service reliability, and track and signal maintenance. This criticism comes on top of a study few years back that found that privatized British rail performed the worst in comparison to its mostly government controlled peers in seven other European countries.

In September 1, 2014, the 1994 British railways privatization turned full circle as NR, the private entity which operates and develops Britain' fixed railways infrastructure - over 2500 railway stations, railway tracks, signalling systems, bridges, tunnels, and railway crossings - was formally "re-nationalised" as a public entity. This followed the government assuming the £38bn debt that NR had accumulated. 

NR, the successor to the failed Railtrack, is listed on the London Stock Exchange and has a private market sourced management. It was set up in 2002 as a private company with no shareholders, though its finances were guaranteed by the government. Though now public, its governance structure provides for an arms-length relationship with the government, thereby ensuring commercial and operational freedom. NR is regulated by the Office of Rail Regulator (ORR).

NR's operating revenue comes mainly from three sources - 60% as grants from Department of Transport, 27.8% from track access charges paid by train and freight operators, and 10.6% from income from property and shops. While it has previously raised debt directly from the market, the re-nationalization means that it will borrow directly from the government, by way of a £30.3 billion loan facility for its current (2014-19) five year investment plan. In the current plan, NR intends to spend £35 billion - £12.5 billion on new and upgraded stations, electrification, and platform lengthening, a third on track renewals, and the remaining third on running the railways. It spent £36.2 billion in operating, reviewing, and maintaining tracks in the previous five year period (2009-14). 

Its management argues that NR is being squeezed by onerous cost reduction targets and rising service standards at a time when passenger traffic is growing at record rates. It claims to have reduced operating costs by 40% over the past decade and says that any further reductions are difficult. The result has been rising ticket prices, which has generated popular resentment and political opposition.

As it proceeds with reforming Indian Railways and introducing private participation, its administrators would need to take a more nuanced view of the whole process rather than a simple strategy of allowing franchising in passenger and freight traffic segments. 

Tuesday, March 24, 2015

Widening inequality and declining returns to labor

The conventional wisdom on increasing inequality across the world lays the blame on rising returns to capital complemented by declining (or stagnant) returns to labor.

These two trends revolve around three causal factors - trade, technology, and politics. Globalization and trade liberalization has increased the global supply of labor relative to capital; technological changes have lowered the price of capital and increased the substitutability of capital and labor, especially less skilled labor, thereby lowering the demand for labor; and political ideological shifts have undermined labor bargaining capacity. Four recent papers lend credence to the aforementioned factors.

1. Holger Mueller, Elena Simintzi, and Paige Ouimet, using firm-level data, find that "wage differential between high and either medium or low-skill jobs increase with firm size". This coupled with the increasing share of employment by the largest firms appears to point to rising firm-size as contributing to widening inequality. The benefits of size are appropriated by the skilled and senior workers at a firm - a cleaner or salesman in a small retail shop or big chain retailer gets similar salaries whereas the wage differential for managers is substantial. And the big chain retailers have been squeezing out the small shops.

2. Albert Bollard, Peter Klenow, and Huiyu Li examined manufacturing sector in US, China and India over 1982-2007 and find a rising trend towards larger sized firms. They reason that bigger firms have higher productivity, which in turn raises the barriers to entry for new and smaller competitors and entrenches the incumbents. The increased entry cost stems from "rising cost of labor used in entry, as well as higher output costs of setting up a business to use more sophisticated technologies".

3. Lukas Karabarbounis and Brent Neiman document a secular (across countries and industries) decline in global labor share of income over the past 35 years. They credit it to decrease in the relative price of investment goods, often attributed to advances in information technology and the computer age, which in turn induced firms to shift away from labor and toward capital.

4. Francisco Rodriguez and Arjun Jayadev document a similar secular decline in the labor share of income in manufacturing sector over the past three decade. They also find that "this decrease is driven by declines in intra-sector labor shares as opposed to movements in activity towards sectors with lower labor shares", thereby appearing to corroborate the findings of Karababounis and Neiman. 

Saturday, March 21, 2015

India real estate market fact of the day

From Livemint, adding to the long list of badly indebted corporates, comes debt-laden property developers, whose problems are exacerbated by unsold inventory,

At brokerage Kotak Securities, analysts estimate unsold inventory held by a group of leading Mumbai developers alone now stands at some Rs.53,400 crore—with an additional Rs.36,800 crore of project launches in the pipeline. That puts the current, unsold area in Mumbai at almost the value of the total sold in the 2014 calendar year. The backlog, analysts estimate, could take more than a decade to clear... It now takes developers about four-and-a-half years to turn property inventory into cash, more than a full year longer than it takes developers in China, according to Thomson Reuters Starmine data.
Another article points to a study by Knight Frank India which highlights the large volumes of inventory piled up across the largest cities. For example, in Delhi, even if no more properties are added to the market, the current inventory will take about 14 quarters to be sold.
However, this glut, mainly at the middle and upper end of the market, stands in stark contrast to the severe shortage at the lower middle income and lower income level (Rs 5-20 lakhs) markets. This trend points to a market failure in the housing market, demanding public policy action. 

Monday, March 16, 2015

The flawed orthodoxy - case of infrastructure contracting

The Economic Survey's advocacy for a strategy of "combining construction and maintenance responsibilities to incentivize building quality" in infrastructure asset creation is classic example of theory sans reality. It is also an example of how orthodox approaches can end up costing more than prudent strategies designed based on the existing market structure.

The fundamental assumption is unexceptionable - good building quality is critical to life-cycle costs, and a combination of building and construction aligns incentives to minimize life-cycle costs. But a first order requirement for this assumption to hold is that the construction contractor has a stake in operation.

In an ideal world, life-cycle costs minimization can be achieved by assembling a project development consortium of construction and operation and maintenance (O&M) contractors as well as financiers. As construction is completed, the construction contractor gradually exits, allowing the O&M contractor take over the project. But in the real world, especially in developing markets like India where the distinction between construction and O&M contractors is blurred, such tight alignment and seamless transitions rarely, if any, happen. Here, a GMR or HCC, after outsourcing the O&M to smaller contractors, sells their assets to infrastructure funds like SBI Macquarie or IDFC India Infra Fund and abruptly exits the project. This does little to align incentives towards minimizing life-cycle costs since the project developer predominantly performs the role of a construction contractor.

Conditional on the difficulty of combining construction and maintenance responsibilities, the most optimal strategy would be to de-couple the two activities. The road map,
  1. Establish a Special Purpose Vehicle (SPV)
  2. Put in place professional management and good governance practices (arms-length relationship with the government department/ministry)
  3. Construct, using mainly public funds, and off-load construction risk
  4. Stabilize the project and lower commissioning risks
  5. Contract out O&M as long-term concession
  6. Regulate the concession
The benefits are substantial. Public financing lowers the cost of capital for construction. Once the project is commissioned, real project information becomes available, which enables concessionaires to bid with far greater assurance. It also enables government agencies to write more complete contracts, which minimize re-negotiation risks. All this enhances transparency and limits political risks. On the other side of the risk assessment ledger, the biggest downside risk, which assumes significance given the country's state capability deficit, involves the governance of the SPV which is critical for both construction time and its quality. 

Sunday, March 15, 2015

Weekend visualizations

1. NYT has a series of night-time activity visualization of different Syrian cities before and after the break out of the civil war. The country is 83% darker at night than before the war began.
The civil war has so far displaced nearly half the country's population and killed more than 200,000 people. Even Damasacus, under government control, is 35% darker than before the war.

2. FT has this superb visualizations about Twitter activity in Venezuela which starkly shows the extreme levels of polarization of opinion in the country. The graphic below shows the Twitter activity  (about 91000 tweets) networks on the day of President Maduro's claim about having foiled a US sponsored coup attempt.
The red cloud on the right points to tweets from those aligned to President Maduro and that on the left represents the remaining vast variety of Venenzuelan and Spanish-language media and others. The lack of overlap is a striking illustration of the disconnect of the Maduro administration with the rest of the country.

3. The most fascinating is this one about which airports, airlines, and routes within the US are likely to get you fastest to your destination.

Friday, March 13, 2015

High impact education interventions

What are the high impact interventions in education for developing countries? For those like India, which have largely surmounted the inputs and access challenges, and where learning outcomes are the primary concern, I can think of three high impact interventions.

1. Transition from "teaching to the class" to "teaching to the child" - It is now well established that children have differential learning trajectories. Therefore, the conventional approach of "teaching to the class", a large group at that, will not realize uniform learning outcomes. The different initial learning levels among students in a class exacerbates the "teaching-learning" gaps. It is therefore essential that class-room instruction has to be tailored around "teaching to the child" by remediating those lagging behind and bringing them upto speed with the rest.

In many developed countries, smaller class-rooms and highly motivated and well-trained teachers enable "teaching to the child". But in countries with large class-room environments, students at varying initial learning levels, and poorly-motivated teachers, the challenge lies with identifying the right strategy for "teaching to the child". One approach would be to group children across grades based on their initial learning levels, and moving them across groups based on their learning progress. A similar approach can be adopted within each grade too. Another approach would be to divide each class period into two halves, with the second half devoted to remediation of those lagging behind. Alternatively, the same remediation can be done through off-school hours instruction by trained volunteers. All these strategies are most likely to be contextual, to be selected based on the specific conditions available.

2. The adage that you cannot monitor what you cannot measure applies with great force to learning outcomes. Currently, monitoring of learning outcomes does not find place in the realms of data being collected by supervisors. Part of the reason is the sheer difficulty of quantifying primary school learning outcomes, especially in the aggregate, in a meaningful and reliable manner.

Any effective push towards improving learning outcomes requires the development of a credible quantifiable learning levels monitoring framework. This has to be multi-tiered, tailored to meet the functional needs and supervisory bandwidth available at each level of monitoring. A cognitively salient dashboard that enables effective supervision and can be disseminated through tablets and smart-phones can be a powerful instrument to improve learning outcomes.

3. Finally, a first order requirement for the success of all these measures is effective enforcement of accountability. Primarily, teachers and the school establishment have to be accountable to parents and the local community. One way to achieve this is effective functioning of School Management Committees. But SMCs have their limitations and cannot be a substitute for more institutional accountability of the school to local community. A fundamental requirement is to dismantle the current system, where teachers are employees of the state government, and make them employees of the local government. 

Thursday, March 12, 2015

Pharma industry fact of the day

Jeff Sachs illustrates the problem with the current patents regime by pointing to the example of Gilead Sciences' block-buster drug to treat Hepatitis C virus (HCV), sofosbuvir, sold under the brand name Sovaldi. He writes,
In December 2013, the Food and Drug Administration approved Sovaldi, and another formulation, Harvoni, which is sofosbuvir used in combination with another drug. Gilead set the price for a 12-week treatment course of Sovaldi at $84,000, amounting to $1,000 per pill. Gilead set the price of Harvoni at $94,000... In the first year of marketing, Sovaldi and Harvoni are already blockbusters, reaping a remarkable $12.4 billion of market sales in 2014, more in just one year than the $11.2 billion price that Gilead paid in January 2012 to buy sofosbuvir from a biotech start-up named Pharmasett. 
The real story is not about Sovaldi's pricing, exorbitant as it is, but that of its development. Sofosbuvir was developed by a team of biochemists led by Professor Raymond Schinazzi at Emory University using National Institute of Health (NIH). In fact, NIH even financed the Phase I and II of its clinical trials. As Sachs writes,
Prof Schinazi set up Pharmasset Inc. as a Delaware corporation in 2004 as his business to develop sofosbuvir and hold the patents on the new prospective drug. Pharmasset raised around $45 million in a 2007 IPO and used those funds and others to supplement the R&D. According to the company's SEC filings, the total Pharmasset R&D on sofosbuvir up through 2011 totaled around $62.4 million. In January 2012, with an eye on sofosbuvir, Gilead paid $11.2 billion to purchase Pharmasett. Schinazi pocketed an estimated $440 million for his shares in Pharmasett... The total private-sector outlays on R&D were perhaps $300 million, and almost surely under $500 million, meaning that the decade-long R&D outlays were likely recouped in a few weeks of drug sales. 
And about its real cost of production,
According to researchers at Liverpool University, the actual production costs of Sovaldi for the 12-week course is in the range $68-$136. Indeed, generic sofosbuvir is currently being marketed in India at $300 per treatment course, after India refused to grant Gilead a patent for the Indian market. In other words, the U.S. price-cost markup is roughly 1,000-to-1!
And Gilead's patent on Sovaldi runs until 2028. Talk about long-term rent-seeking! And it is no surprise that others are following suit. Pharma firm AbbVie (which sells autoimmune disorder drug, Humira), recently purchased cancer drug maker Pharmacyclics for an astonishing $21 bn. Pharmacyclics' only marketed product is Imbruvica, a treatment for chronic lymphocytic leukemia and two other rare blood cancers, which was approved in 2013 and is considered a major breakthrough with massive commercial potential.

Update 1 (16/05/2015)

From Peter Singer,

One drug, Soliris, costs $440,000 per patient per year. In contrast, GiveWell estimates that the cost of saving a life by distributing bed nets in regions where malaria is a major killer is $3,400. Given that most of the lives saved are those of children, who even in developing countries have a life expectancy of at least 50 years, this equates to a cost of $68 per year of life saved. Should we really be valuing the life of a person in an affluent country at more than 6,000 times the value of the life of an impoverished child in a developing country?

Tuesday, March 10, 2015

Importance of Central Bank independence in a graph

Jim Hamilton points to this old Alesina-Summers graphic which shows that greater political control over monetary policy exercised by the legislature and government is associated with higher inflation.

Sunday, March 8, 2015

Financial market growth and resource mis-allocation

Gretchen Morgenson points to a recent paper by Stephan Cecchetti and Enisse Kharroubi which finds that overall productivity gains were dragged down in economies with rapidly growing financial markets. They studied 33 manufacturing industries in 15 countries and came to two interesting conclusions,
First, the growth of a country's financial system is a drag on productivity growth. That is, higher growth in the financial sector reduces real growth. In other words, financial booms are not, in general, growth-enhancing, likely because the financial sector competes with the rest of the economy for resources... This is a consequence of the fact that financial sector growth benefits disproportionately high collateral/low productivity projects. This mechanism reflects the fact that periods of high financial sector growth often coincide with the strong development in sectors like construction, where returns on projects are relatively easy to pledge as collateral but productivity (growth) is relatively low... 
Second, using sectoral data, we examine the distributional nature of this effect and find that credit booms harm what we normally think of as the engines for growth – those that are more R&D intensive... We report estimates that imply that a highly R&D-intensive industry located in a country with a rapidly growing financial system will experience productivity growth of something like 2 percentage points per year less than an industry that is not very R&D-intensive located in a country with a slow-growing financial system... By draining resources from the real economy, the financial sector becomes a drag on real growth.
In simple terms, when capital is available in plenty and cheap, it is more likely to get mis-allocated to less-productive sectors prone to asset-bubbles like construction and real-estate and away from more productive but riskier activities like start-ups and entrepreneurial ventures or research and development intensive sectors. They also find that "when finance is ascendant in an economy, it attracts an inordinate number of highly skilled workers who might otherwise take their productivity and brains to non-financial industries." A corollary to this is that real estate and construction asset bubbles generate financial resource mis-allocation effects that adversely affects the real economy.

This mechanism partially explains India's brief high-growth episode of 2003-08. It was associated with cheap and plentiful credit which fuelled a real-estate bubble and construction boom. This was amplified by sharply increased public and private spending on infrastructure sectors - roads, ports, airports, power projects, urban utilities, mining etc. The share of construction sector in the gross value added and total employment rose disproportionately. In contrast manufacturing's share of employment contracted and output remained stagnant. Instead of trying their luck in knowledge-based sectors and manufacturing, entrepreneurial talent flocked to high-return but less productive real estate and infrastructure sectors.

The net result was a growth episode whose foundations were laid on low productivity non-tradeable construction sector. This was unlikely to be sustainable and collapsed when faced with adverse shocks leaving behind distressed balance sheets in both the construction-intensive infrastructure sectors as well as among their creditors. An infrastructure and construction focussed revival of economic growth, without adequate contribution from productivity enhancing tradeable sectors like manufacturing, is only likely to repeat the story. 

Friday, March 6, 2015

Universal health coverage in a graphic

The debate on universal health coverage can be captured in the following graphic.

The best that India can hope for the foreseeable future is to achieve high people coverage, of reliable standards, for a defined basic set of medical conditions, including the more common high-cost  conditions (like heart ailments, dialysis etc). The conditions coverage should also include the provision of generic drugs and all basic diagnostic tests, made freely available for those below a certain income level through public facilities as well as by contracting in from private providers.