The 4th fastest supercomputer, continued

In continuation to the email from Viral Shah on my blog entry titled Interpreting the 4th fastest supercomputer in the world, Subhomoy Bhattacharjee got Viral to write an article in Financial Express on the subject. In addition, Business World has a story describing the `Eka' project.

The wheels of global general equilibrium adjustment are in motion

Ila Patnaik has an article in Financial Express today on the global rebalancing that's underway. It's a very interesting time in international economics. On this subject, also see:

• Uncertainty shifts to China, by Brad DeLong.
• A full blown collapse of the dollar? in The Economist.
• Linkages between the US and India, by Ila Patnaik.
• The woes of the dollar, in The Economist.
• The world does not shake China, in The Economist.
• Wake up to the dangers of a deepening crisis, by Larry Summers in FT.
• Is the US in the throes of a sudden stop? by Brad Setser.

Percy Mistry has questions

In Business Standard today, Percy Mistry asks questions which will make many of us in the policy community squirm.

R. H. Patil

Outlook Money has given a `Hall of Fame' award to R. H. Patil, and C. B. Bhave talks about it.

New Pension System (NPS) coming to life

The Pension Fund Regulatory and Development Authority (PFRDA) has put out an important press release today. Some highlights:

• National Securities Depository (NSDL) has been contracted to become the Central Recordkeeping Agency (CRA).
• CRA operations will commence by 1 June 2008.
• NSDL's price will be : Rs.350 a year plus Rs.10 per transaction upto 1 million accounts, dropping to Rs.250 a year plus Rs.4 a transaction beyond 3 million accounts.
• Three fund managers are expected to be up by 15 December - State Bank of India (SBI), Unit Trust of India (UTI) and Life Insurance Corporation (LIC). The fees charged by these managers range between 3 and 5 basis points of assets under management (AUM) a year! [link]
• The equity fraction can go up to 15% right now.

Update: Business Standard has an edit on this subject:

The reform of the pension system has achieved two important milestones. First, after an unfortunate delay induced by the Securities and Exchange Board of India (Sebi), the contract for the Central Recordkeeping Agency has been signed with the National Securities Depository (NSDL). The administrative competence and computer systems of NSDL are now of supreme importance in making the new pension system (NPS) work. In many other countries, well-meaning pension reforms have floundered because of weak administration. While NSDL has a strong track record of running large and complex computer databases, record-keeping for the NPS is a daunting task. The devil lies in the details of rounding up information for central government and state government employees from all over the country, creating accounts for them, delivering their money into the relevant accounts, and getting account balance statements back to them.

The most important MIS reports that NSDL and the pension regulatory authority should be releasing every day to the Press consist of two numbers: How many accounts are in place with full reconciliation, and the assets to be found in these accounts. This graph will portray the success of rolling out the NPS. Four factors will be at work in shaping the graph. First, how quickly is the NPS extended from the central to the state governments? Second, how quickly is the backlog of past recruits taken into the new administrative system? Third, there will be an ongoing process of new recruitment which will steadily accrete the number of NPS accounts. Finally, returns on pension assets will swell the total assets in the NPS. Within two to three years, it is possible to get the NPS past a million accounts and Rs 5,000 crore of assets, at which point it will be a significant pension system by international standards.

The second major announcement that has been made is about fund managers. Earlier, the pension regulator had announced that UTI, SBI and LIC had been recruited through a competitive bidding process that was restricted to public sector bidders. Now, the regulator has announced that the prices that will be paid to this trio of pension fund managers range from 3 to 5 basis points of the assets under management, and will reflect the sum of fees and expenses. By way of comparison, mutual funds in India charge over 100 basis points, and insurance companies charge even more. Compared with these benchmarks, price points of 3 to 5 basis points are a very good achievement. The NPS has delivered on the promise of auction-based procurement obtaining very low costs when compared with the private retail market. At the same time, these bids are higher than the 1 basis point level that is visible internationally. It is to be hoped that, as the assets of NPS build up, and when private competition comes in, these prices will drop further.

In December 1997, or 10 years ago, the ministry of social justice tapped Surendra Dave to head Project Oasis, which designed the New Pension System. It has taken 10 years for the idea to wind through the complex policy-making processes of India, across both NDA and UPA administrations, to get to this starting point where ideas are now turned into execution. Over the next 10 years, the goal of the NPS should be to get to 100 million accounts, so as to make an important dent in the problem of India’s ageing workforce.

Great article on advisors and fund managers

An article by Michael Lewis that will make most employees of most financial firms squirm. The most interesting firm in the story is Dimensional Fund Advisors.

A revolution in payments could be in the pipeline

Imagine being able to pay a taxi driver by `sending him an SMS'. A lot of people have cell phones, so person -> person and person -> establishment payments can be achieved with very low frictions, once mobile phones are properly in harness for the purpose of payments. I believe the cost of production of an SMS in India is now roughly Rs.0.02, so perhaps a P2P payment can get done for a flat cost of Rs.0.04. That makes it viable when buying something worth Rs.5 or Rs.10. (Rs. 10 is roughly $0.25 at the present exchange rate).

At the simplest, a payment could go hit a bank account. But when you have a prepaid phone, a payment could dip into the cash that's been deposited with the mobile phone vendor. And, once the mobile phone vendor has got some of your history of payments, he might be willing to take some credit risk and let your account balance go negative, thus giving you credit. This credit could get securitised and sold off on the bond market just as is the case with credit cards. All this could be merely a relationship between the customer and the mobile phone company, backed by bulk investors for securitisation paper, without a role for a bank.

Hence, this will be opposed by banks, who will substantially lose their role and revenues from payments. Poorly run banking regulators will be defensive about banks and will try to come in the way. How this political economy plays out is of critical importance, in judging whether this new technological opportunity will be properly harnessed.

A press release from GSM Association shows new work and focus on this subject. The GSM Association (GSMA) is the global trade association representing more than 700 GSM mobile phone operators across 218 countries and territories of the world. In addition, more than 200 manufacturers and suppliers support the Association's initiatives as key partners. I was very happy to see Indian telecom vendors playing a role in this thinking. I guess India is very big on the global scale in the world of mobile phones (In October there were 217 million mobile phones in the country, with growth of 8 million a month).

Leslie D'Monte has an article in Business Standard about these developments.

OTC equity derivatives in India, and their applications

An article by Rajesh Abraham in Business Standard talks about Nifty-linked bonds being sold by financial firms. These kinds of products are best achieved when an OTC equity derivatives market is available.

You should also read Jayanth Varma on this subject.

Phrase watch

In 1973, Erica Jong wrote a book titled Fear of flying. The instant success of the book made the phrase famous and recognisable: Google shows 981,000 occurences.

In 2000, Guillermo Calvo and Carmen Reinhart used Fear of floating as the title of a paper. The paper was very influential, and the phrase `fear of floating' along with the acronym FoF have gone mainstream in international economics. Google shows 58,900 occurences - roughly one-sixteenth of `fear of flying'.

At Suman Bery's seminar on MIFC a few months ago, I thought I came up with the phrase `fear of finance' on the fly, but googling shows 143,000 occurences - even more than `fear of floating'. All three phrases have a nice f sound, but I guess `fear of finance' is a more mainstream issue than `fear of floating'.

Another mainstream, and evocative, phrase is fog of war for which which google reports 972,000 hits. The wikipedia entry says it goes all the way back to Clausewitz.

Now I have seen a new phrase fog of finance trying to make its way. Right now google shows 1600 hits. I'm not that enthusiastic about it. I use `fear of floating', `fog of war', and `fear of finance' in my ordinary speech, but I don't yet feel the same about `fog of finance'.

First research meeting of the NIPFP-DEA Research Program on Capital Flows and their Consequences

The program is up on the website.

If you read this blog, you must be a

13th Finance Commission

Vijay Kelkar will head the 13th Finance Commission, and Business Standard has an interesting edit framing the issues that 13FC faces. You might like to see Kelkar's FRBM Implementation Task Force report, and this feature on him in Economic Times. In a timely piece in Economic Times, Amaresh Bagchi and Satya Poddar argue in favour of a lower GST rate than that proposed in the FRBM Task Force report.

Implications of new international financial centres

I was at the FT/DIFC World Financial Centres Summit in Dubai, which helped me to write a Business Standard article titled Implications of new international financial centres.

Authoritarianism 2.0

There is a considerable consensus that USSR-style authoritarianism does not work - it delivers low material wealth, and generally leads to political breakdown. At the end of the 20th century, there was a great deal of `end of history' optimism that the breakdown of this model would lead to a universal movement towards liberal democracy.

However, a series of countries have now embarked on what I term `Authoritarianism 2.0' : a blend of free markets with authoritarian politics. This aims to produce good resource allocation and high growth rates through sound economic policy, while retaining one-party (or one-person) rule, denying the rule of law, and supressing the rights of individuals. Rowan Callick has written an article titled The China Model which reviews this style of design of the State, where individuals are seen as consumers but not citizens; where individuals have some economic flexibility but weak political rights. This model is being attempted in Russia, China, Vietnam, UAE, etc. In some ways, Singapore also exhibits some of these features. These countries do not see their political systems as illegitimate way stations on the road to democracy. The effort is to make Authoritarianism 2.0 work; liberal democracy is not even the long-term goal.

The question we really face is about the soundness of the `Capitalism and Freedom' hypothesis. Does the growth of a free market economy inevitably generate pressures to modernise the political system? Or is it possible to combine 21st century economic policy with pre-enlightenment politics?

One facet of Authoritarianism 2.0 is the rise of large Sovereign Wealth Funds (SWFs) controlled by the State. Many people are benign about the mingling of SWFs into global capitalism. I am not. I am pretty uncomfortable with ownership of shares of Indian firms by the Indian government; by extension I should be even more uncomfortable with shares held by foreign governments, who are not even accountable in the domestic political system. In Business Standard today, Deepak Lal has a suggestion on how liberal democracies should confront the SWF question.

Interpreting the 4th fastest supercomputer in the world

In response to my blog entry (and comments about it posted on this blog) about `Eka' making rank 4 in the Top500 list of the world's supercomputers, I got this email from my friend Viral Shah who knows a bit about the field:

Ajay, it seems that the ranking of Eka at number 4 on the Top500 list has resulted in quite a lot of excitement. Hats off to the folks at CRD Labs for achieving the feat of assembling such a large computer in a short amount of time. As some of your readers noted, Eka is a cluster. It is roughly 2,000 nodes, consisting of roughly 15,000 processors and connected by Infiniband. Some readers noted that the benchmark is not representative of real scientific applications. 

Firstly, making a small cluster is quite easy. However, constructing such a large cluster, and operating it is no easy task. It requires some serious skills to administer it, tune the hardware and software for performance, and run scientific applications on it. Second, the Top500 is an interesting benchmark. Sure, it is not representative of a realistic workload, but over the years, the bar has been set quite high. If a general purpose computer does not achieve a good LINPACK score (the top500 benchmark), it is safe to conclude that something is terribly wrong. I am of course excluding special purpose computers that are built to solve specific problems, rather than get a high LINPACK score. 

That said, one needs to think this through clearly. Why was Eka built? To simply show that we can do it, and place a computer in the Top 10 supercomputers? To run specific scientific applications? I am guessing that the answer is "a bit of both". Almost always, it is safe to conclude that the full supercomputer is never used to solve one problem. What are the largest problems that will be run on Eka? What percentage of peak will they achieve? Would it have been a better idea to buy an "off the shelf system" such as the Cray XT4 or the SGI Altix and focus on programmer productivity, instead of getting a high LINPACK score? 

Computers such as Eka achieve extremely high and unrealistic flop rates on the LINPACK benchmark.Typically, they can achieve over 70% of the peak flop rate (Number of floating point operations per second). However, real applications often run at below 5% of the peak flop rate. Let's examine some other possibilities. 

Note that the software industry has been one of India's strong points. It is becoming increasingly clear, that, software is the key. For example, Apple's success with the iphone and ipod have as much to do with well designed software, as with the hardware. If you ask me, the big event at Supercomputing'07 was not that Eka placed at No. 4 on the Top500 list. For me, the most exciting event was one that you will not hear about in media - it has to do with the other part of the HPC Challenge, often called the beauty contest. Instead of asking "which computer can run LINPACK the fastest", it asks, "which programming language implements the benchmarks elegantly". 

The winners of the class II challenge this year were IBM's X10, and Interactive Supercomputing's Python Star-P. For me, the most surprising, and the coolest event was the revelation that some of the compiler work for X10 was done at IBM's research labs in India. This is cutting edge compiler technology, and the fact that part of the team was based in India is a strong statement about HPC innovation in India. 

Not to belittle the effort that went into Eka, but we should be asking the hard questions. I think it is fantastic that we can afford to build a $30 million supercomputer. But how are we going to program it? What applications will run on these large computers? Will we be able to address some important problems such as better meteorological forecasts for our farmers, or better groundwater modeling to solve our long term water problems, or allow our companies to gain that extra edge in the international arena? Are we better off buying computers from those who know how best to make them, and focusing our skill sets on what we are good at - developing software? 

Our universities are indeed not be up to this challenge yet, but, perhaps, we don't need to wait till they catch up. With all the resources available online, a hungry young person can learn this game - the material is all online, after all, like these classes at MIT, at UC Berkeley and at UC Santa Barbara.

State capacity in 1893

I came across a fascinating story of public administration in the colonial era:

Gohna village is located on the true right bank of the Birahi Ganga, not far from Chamoli town. The nearest road head is Nijmula village, 2 kms down the true left of the river. A further 9 kms by road from there is Birahi town, where the river merges with the Alaknanda. Chamoli town is a 7 km drive from Birahi.

A massive landslip in September, 1893 created a natural dam across the Birahi Ganga near Gohna village, 900 ft high, 11000 ft wide at the base, and 2000 ft high at the apex. The trapped waters of the river collected upstream of the dam, forming a lake that came to be known as Gohna Tal, although several locals seem to refer to it as Durmi Tal, after the village of Durmi 3 kms upstream the Birahi Ganga.

The first intimation that the world received of this catastrophe was a message from the patwari (local administrative official) to the Deputy Commissioner of the district, in which he merely reported that a mountain had fallen. This unlikely news was ignored by the DC. Fortunately, the district surveyor and the executive engineer were touring in the area, and gave a detailed report to the administration. This resulted in a visit by a brilliant army engineer, Lt. Col. Pulford, who gave the opinion that there was no danger until the accumulated waters topped the dam, at which time there would be an enormous flood down the Alaknanda valley. Pulford's view was contested by other experts, some saying that the dam will burst due to pressure well before the waters flow over the top, and some others saying that nothing will happen at all because the waters would top the dam and flow peacefully downstream. Fortunately, again, the Government listened to Pulford.

An Assistant Engineer from the army, Lt. Crookshank, was sent to Gohna, with the task of watching the lake, and sounding the alarm when it was about to top the dam. A telegraph cable was installed for this purpose. Observation posts to monitor the levels of the Alaknanda and the Ganga were established in various places, from Chamoli town all the way downstream to Haridwar. Pillars were erected in many places in the Alaknanda and Ganga valleys to mark the danger limits of the expected deluge, and the inhabitants were directed to evacuate as soon as the water crossed these levels. The pilgrim routes were diverted, and suspension bridges across the Alaknanda were dismantled.

Based on Lt. Crookshank's data, the army engineers very accurately predicted the date when the overflow would occur. On 22 August, 1894, nearly one year after the formation of the lake, Lt. Crookshank declared that the flood would start in the next two days. Quite creditable, since the river topped the dam in the wee hours of the 25th August, the barrier collapsed with a bang at about midnight, and the flood ended early in the morning on the 26th August. It was found that over 10000 million cubic feet of water had escaped, and that the level of the lake had descended by 390 ft. Srinagar was completely swept away, and there was extensive loss of property everywhere. However, there was almost no loss of human lives, the only exception being the rather foolhardy family of a mendicant who returned to their home near the downstream face of the dam after being evacuated from there. I find it an amazingly successful case of disaster management.

It made me wonder how State effectiveness in India today compares against this.

Knowledge without universities

It's a mystery to me, how India manages to make significant progress (the 4th fastest supercomputer in the world) without having credible universities.

The woes of administered prices

While there is a lot of focus on two administered prices - the rupee-dollar and petroleum products - there are many other products where the government also has an extensive engagement. Business World has an interesting edit on the story unfolding with cotton.

Bias in favour of exchange traded

Stephen Cecchetti expands and enlarges on his previous arguments about the importance of exchange-traded instead of OTC in achieving a sound and stable financial system. He also has some ideas on how a policy bias in favour of exchange-traded might be implemented.

Influence of Malaysia on Indian thinking on economic policy

Where do India's policy makers distill their frenzy from? Mahathir Mohamad?

Malaysia	India
Runs an elaborate system of discriminatory policies designed to favour Malays (the `bumiputra') over the economic elite - the Indians and the Chinese.	Runs a less elaborate system of discriminatory policies designed to favour the `lower caste' over the economic elite - the `upper caste'.
The former prime minister indulged in vitriolic outbursts during which he said "immoral" currency speculation should be banned and rumour-mongers should be shot. He called George Soros, the US financier, a "moron" and suggested that a Jewish "agenda" might be behind the assault on the ringgit, Malaysia's currency. He ascribed great power to `currency speculators' : They can speculate with any currency, and their speculation is so designed that they can either revalue a currency or devalue a currency to any level. They hold this power, and they can literally make or break you by just by doing that.	The petroleum secretary says that in the global crude oil market, supply and demand are not imbalanced. Rather, trading on exchanges like the New York Mercantile Exchange, or Nymex, is contributing enormously to high prices and the solution is to ban crude oil related futures trading at Nymex.
Was the only country affected by the Asian Crisis which tried to address the problem using capital controls.	Is known to resort to capital flows hatao when facing implementation difficulties with a inconsistent monetary policy regime.
The former prime minister was suspicious of short selling: As far as the stock market is concerned, we know that players in the stock market can also destroy the stock market simply by short selling. And to short sell you don't even need to have the shares. That is why we decided that we would stop that, and as a result, the market has recovered.	Policy makers have blocked short selling on the equity spot market.
The former prime minister said: We have welcomed foreign capital a long time ago, and they came in, they built factories, and they exported goods from our country, creating jobs for our people, improving our economy. Yes, that is the kind of globalization that we want. At the same time, of course, they respected our laws and our policies and allowed our own companies to be protected until such time when they were able to compete with the foreign countries. That kind of globalization, yes. But the sudden inflow and outflow of currency is too destabilizing: That we cannot accept as a part of globalization.	Traditional policy thinkers have the same `goods illusion' -- support FDI but work to block financial globalisation.

However, the Mahathir breaks with the caricature in one respect: he criticises the system of reservations - it is not a long-term solution, it reduces the incentive for Malays to compete, moving towards a meritocracy is desirable. In India, reservations are, as yet, a holy cow.

I thought I was being clever with the phrase `goods illusion'. But it's been around since atleast 1959.

The weak rupee policy

The Fed cut rates, and India appears to have responded by raising the limit on MSS issuance by RBI. Let's chalk out the rough fiscal costs --

• Each month with over $10 billion of MSS issuance uses up Rs.40,000 crore of bond issuance. Roughly speaking, assuming a net cost of 4%, a stock of MSS of Rs.250,000 crore runs up a tidy cost of Rs.10,000 crore a year.
• In addition, assuming India has roughly $125 billion in USD assets, each 1% depreciation of the USD is a cost of roughly $1.25 billion or roughly Rs.5,000 crore on account of depreciation of the reserves portfolio.

Ila Patnaik looked at monetary policy in Indian Express a few days ago.

Martin Wolf on competition policy

Martin Wolf reminds us of the importance of competition policy.

A coming of age for private equity

I wrote an article in Business Standard today about what private equity funds are doing in India.

Reasons why the PN/ODI market is attractive

The edit in Business Standard today points out that one of the many reasons why global customers favour the PN/ODI market lies in the Indian tax treatment of derivatives, where profits are treated as ordinary income.

Jamal Mecklai makes monetary policy interesting

In Business Standard today, Jamal Mecklai makes monetary policy fun:

In the four weeks to October 12, the RBI bought over $24 billion — that’s more than the forex reserves of the Netherlands, Saudi Arabia or Kuwait! This enabled them to cage the rupee at about 39.30, with liquidity splashing around everywhere. With inflows unabated — indeed, accelerating — and the half-hearted (old, uncommitted and terrified) efforts to stimulate outflows completely ineffective, the RBI’s nightmarish screams were finally heard in Delhi and the ministry of finance pushed Sebi to try and raise the barricades.

Of course, the market tanked and the government backed down — welcome to Thailand. While markets have recovered much of their composure, it seems that the government continues to turn a deaf ear to the real issues — at least to judge from the politics-as-usual sentiment that prevails. This means that while markets will get back to business as usual soon enough — after all, this is India and it is going to grow at 10% a year for some time come — there will certainly be another, and then another, wobble, each one larger than the last.

The good news, however, is that, if you leave the political establishment out, change is in the air. Indeed, almost everyone would agree that there is a complete disconnect between the essence of the country today and that of the government, and, in this day and age of market determination, this means that sooner rather than later we will see substantial change in the nature of our government.

And here's the Indian improvement upon `google' as a verb:

Indeed, there are several straws in the wind. Note, for instance, the continuing impact of the Right to Information Act — in a few short years, it has become a part of life; indeed, my wife points out that “RTI” has become a verb as in “just RTI it.”

Improvement in Indian public finance

In the late 1990s, the #1 area of concern in India was the fiscal crisis. As an example, here is a gloomy piece that I wrote in 2001. A remarkable amount of progress has been made on that front. Tax policy is more rational; tax administration has improved; the FRBM Act has tied down the government; the Twelfth Finance Commission has triggered good improvements at the state governments also. In Economic Times today, Mythili Bhusnurmath talks about developments at the state governments.

In my reckoning, the fiscal consolidation is roughly half done. We aren't out of the woods yet, but a good down payment is visible. As Mythili points out, the focus has now shifted to monetary policy, where a comparable transformation of law and institutions has not begun.

Entry of foreign legal firms into India

Cyril Shroff, one of India's most famous lawyers and a great beneficiary of liberal economic policy, presents `infant industry' and other `pre-requisites are missing' arguments about why liberalisation of legal services is inappropriate.