## Friday, July 31, 2009

### Building a better credit policy speech

I have an article in Financial Express today titled Building a better credit policy speech.

## Thursday, July 23, 2009

### Debt issuance this year, and setting up an independent debt management office

Writing in Financial Express, Mahesh Vyas and Jayanth Varma analyse the extent to which the 6.8% fiscal deficit of this year is hard' to fund through bond issuance.

In FE today, Jayanth Varma has an article on the merits of setting up an independent debt management office, and they have an editorial on this subject.

## Wednesday, July 22, 2009

### What risk models are useful?

by Ajay Shah.

Risk management failures have clearly taken place. It has become fashionable to criticise risk models.

A fair amount of the naive criticism is not well thought out. Too many people today read Nassim Taleb and pour scorn upon hapless economists who inappropriately use normal distributions. That's just not a fair depiction of how risk analysis gets done either in the real world or in the academic literature.
Another useful perspective is to see that a 99% value at risk estimate should fail 1% of the time. If a VaR implementation that seeks to find that 99% threshold does not have actual losses exceeding the VaR on 2-3 trading days each year, then it is actually faulty. Civil engineers do not design homes for once-in-a-century floods or earthquakes. When the TED Spread did unbelievable things:

the loss of a short position on the TED Spread should have been bigger than the Value at Risk reported by a proper model on many days.

The really important questions lie elsewhere. Risk management was a new engineering discipline which was pervasively used by traders and their regulators. Does the field contain fundamental problems at the core? And, are there some consequences of the use of risk management which, in itself, create or encourage crises?

### Implementation problems

There are a host of practical problems in building and testing risk models. Model selection of VaR models is genuinely hard. Regulators and boards of directors sometimes push into Value at Risk at a 99.99% level of significance. This VaR estimate should be exceeded in one trading day out of ten thousand. Millions of trading days would be required to get statistical precision in testing the model. In most standard situations, there is a semblence of meaningful testing for VaR at a 99% level of significance [example], and anything beyond that is essentially untested for all practical purposes.

Similar concerns afflict extrapolation into longer time horizons. Regulators and boards of directors sometimes push for VaR estimates with horizons like a month or a quarter. The models actually know little about those kinds of time scales. When modellers go along with simple approximations, even though the underlying testing is weak, model risk is acute.

In the last decade, I often saw a problem that I used to call the Riskmetrics illusion': the feeling that one only needed a short time-series to get a VaR going. What was really going on was that Riskmetrics assumptions were driving the risk measure. Adrian and Brunnermeier (2009) emphasise that the use of short windows was actually inducing procyclicality: When times were good, the VaR would go down and leverage would go up, and vice versa. Today, we would all be much more cautious in (a) Using long time-series when doing estimation and (b) Not trusting models estimated off short series when long series are unavailable.

The other area where the practical constraints are onerous is that of going from individual securities to portfolios. In practical settings, financial firms and their regulators always require estimates of VaR for portfolios and not individual instruments.

Even in the simplest case with only linear positions and multivariate normal returns, this requires an estimate of the covariance matrix of returns. Ever since atleast Jobson and Korkie (JASA, 1980), we have known that the historical covariance matrix is a noisy estimator. The state of the art in asset pricing theory has not solved this problem. So while risk measures at a portfolio level are essential, this is a setting where our capabilities are weak. Realworld VaR systems that try to make do using poor estimators of the covariance matrix of returns are fraught with model risk.

As an example, when we look at the literature on portfolio optimisation, there is a lot of caution about the complexity of jumping into portfolio optimisation using estimated covariance matrices. As an example, see this paper by DeMiguel, Garlappi, Nogales and Uppal, which is one of the first papers to gain some traction in trying to actually make progress on estimating a covariance matrix that's useful in portfolio optimisation. This paper is very recent - it appeared in May 2009 - and highlights the fact that these are not solved problems. It seems easy to talk about covariance matrices but obtaining useful estimates is genuinely hard.

Similar problems afflict Value at Risk in multivariate settings. Sharp estimates seem to require datasets which do not exist in most practical settings. And all this is when discussing only the simplest case, with linear products and multivariance normality. The real world is not such a benign environment.

### With all these implementation problems, VaR models actually fared rather well in most areas

There is immense criticism of risk models, and certainly we are all amazed at the events which took place on (say) the money market, which were incredible in the eyes of all modellers. But at the same time, it is not true that all risk models failed.

My first point is the one emphasised above, it was not wrong to have VaR models being surprised at once-in-a-century events.

By and large, the models worked pretty well with equities, currencies and commodities. By and large, the models used by clearing corporations worked pretty well; derivatives exchanges did not get into trouble even when we think of the eurodollar futures contract at CME which was explicitly about the London money market.

Fairly simple risk models worked well in the determination of collateral that is held by futures clearing corporations. See this paper by Jayanth Varma. If the field of risk modelling was as flawed as some make it out to be, clearing corporations worldwide would not have handled the unexpected events of 2007 and 2008 as well as they did. These events could be interpreted as suggesting that, as an engineering approximation, the VaR computations that were done here were good enough. Jayanth Varma argues that the key elements that are required are the use of coherent risk measures (like expected shortfall), fat tailed distributions and nonlinear dependence structures.

## Sunday, July 05, 2009

### A TV show on the Bombay attacks

A TV show with amazing footage on the Bombay attacks has come out. It's very painful to watch, but we have no choice. Part 1 Part 2 Part 3 Part 4 Part 5.

It reminded me of the glorious melting pot that is Bombay. The people in the show speak in Hindi, Gujarati, Marathi, Malayalam, English. I was surprised at how little of the language of the terrorists I could understand. I feel I do better at parsing the local language when I'm in Pakistan.

Read Irfan Husain in Dawn. You might like to see this which I wrote at the time, and this collection of readings from that time.

## Saturday, July 04, 2009

### Microsoft inside the exchange

Microsoft has long faced by a credibility gap in getting into mission critical, enterprise settings. One initiative they embarked on was the TradElect' system which did trading at London Stock Exchange. This trading system was built by Microsoft and Accenture who were keen to prove that it could work. It utilised a series of Microsoft technological components. It was used in ad campaigns by Microsoft [image credit] who claimed that if they could handle London Stock Exchange then they are ready for Serious applications [example, until they take down the page].

This is not as much of a big deal as meets the eye. The London Stock Exchange is a famous and well known brand name, but it's not particularly a big exchange by world standards. That is, it's not a really demanding IT problem. Here's some data, from the June newsletter of the World Federation of Exchanges. At page 39, they show the number of trades through order matching that are seen on all member exchanges for Jan-May 2009, a period of five months. I have added one column where I translate this into trades per second under the assumption that there were 110 trading days in these five months and trading took place for six hours a day.

NYSE Euronext 1403 590
Nasdaq OMX 1167 491
Shanghai 794 334
NSE 602 253
Shenzhen 456 191
Korea 371 156
BSE 222 93
Taiwan SE 108 45
London SE 72 30
NYSE Euronext (Europe) 70 29
Hong Kong Exchanges 53 22

This shows NYSE and NASDAQ at 590 and 491 trades per second, which is a challenging IT problem. The two big Indian exchanges -- NSE (rank 4) and BSE (rank 7) -- are also difficult problems at 253 and 93 trades per second.

These are averages for the system load; in this business there is an extreme peak-to-average ratio. E.g. NSE routinely exceeds 1000 trades/s and occasionally does a lot more. There are days when half of the days activity happens in the last 30 minutes. So the IT challenge is much bigger than the average trades/s seems to suggest.

In this ranking, London Stock Exchange is not that big; it's ranked 9th in the world and does an estimated 30 trades per second on average. So it was a good choice for a certain kind of vendor who tries to make a point using a toy problem which does not sound like one. When sizing an IT system, it is peak load that matters, of course. But the ratio of peak to average is likely to be similar for all the above exchanges. Hence, NSE is likely to be a much bigger problem than LSE whether you measure by average load (as shown above) or by peak load.

The story seems to have gone badly wrong for Microsoft. LSE consistently failed to match rivals like Chi-X, which run Linux, in becoming a credible choice for algorithmic trading. Then there was a day when the trading system collapsed (9 Sep 2008).

This played a role in the CEO of LSE, Clara Furse, getting sacked. The new CEO, Xavier Rolet, is said to have decided to dump TradElect. Here's the story, by Steven J. Vaughan-Nichols.

To be sure, complex engineering projects can fail for many reasons. But it's ironic that the marquee adoption at an exchange, that was advertised by Microsoft as proof that they had arrived, should have flamed out like this despite direct staff involvement from Microsoft.

## Friday, July 03, 2009

### What if India had a Hong Kong

Comment by Anonymous:

Interesting. Have a look:

### Measuring the consequences for developing countries, of open access to the literature

Comment by bagdu:

Then I contacted their coordinator Nicole Hunt who told me that CEPR papers are free as well like NBER. Here is the gist of that communication:

On reading this post, I communicated with CEPR coordinator Nicole Hunt. Here are my findings:

1. One needs to provide one's address and email address to CEPR by sending a mail to CEPR and creating a profile at CEPR's website.This will enable free access to their papers.
2. I suggested to CEPR to make this process user friendly similar to NBER and this suggestion is under consideration.
3. It seems like this is not a defined process yet and the access is on an ad-hoc basis. This being weekend, I have not yet got the access to their papers. Create a profile at CEPR website, try downloading a paper, it results in a failure and an email address to contact. Follow up on that id and you might get the access. I will confirm it here once I get it myself.

### Measuring the consequences for developing countries, of open access to the literature

Comment by bagdu:

This is to confirm that further to creating a profile at CEPR's website and my communication (as an individual from a developing country) with CEPR I have been granted free access to CEPR's papers.

One feels a little lump in the throat when one gets these high quality materials free by virtue of being a citizen of a developing country.

This is one privilege one would like to let go of. Let us quickly become developed!!

## Wednesday, July 01, 2009

### Great men versus well functioning institutions

I have an article in Financial Express today titled Great men versus well functioning institutions.