Why Bombay Matters Less

In the recent bestseller novel Dragon Strike by Humphrey Hawksley and Simon Holberton, when Chinese generals plan a war in the South China Sea, they plan to hit the futures markets as well. Just before the shooting begins, they will go long the dollar and oil, and short the yen.

This gives a whole new meaning to the term `insider trading'. Before Saddam Hussain gave the final command to invade Kuwait, did he make one last phone call to New York, to buy crude oil futures? In that case, the best stethoscope on the conflict would be the futures price; it would have better information than the CIA and the KGB. This scenario is not as far--fetched as it seems: senior government staff of many latin american countries are known to take positions on currency and commodity futures exchanges abroad.

The rise of liquid markets in modern times has opened up the possibility of obtaining vast profits by such means. In earlier times, market liquidity was the constraint: if Mr. Hussain tried to take a large position, the impact cost would wipe out his profits. Hence, in earlier times, positions in financial markets were an insignificant revenue opportunity for heads of state. The situation is different today.

Odious dictators are not the only ones attracted to liquid markets. A liquid market, which charges a small impact cost upon its users, attracts large investments into forecasting future prices, based upon information gathering and analysis. Impact cost is like a tax upon speculation, hence low impact cost attracts greater budgets for forecasting.

In our crude oil example, the liquid market produces one great public good -- a freely observed price -- which encapsulates an enormous amount of information (from the likes of Mr. Hussain) and research (from economists hired to forecast the markets). Others do not need to replicate this information and research, they get the prices for free in the newspaper.

We clearly see this in the EMS crisis of 1992. The British Sterling went off the EMS on September 16. However, the options markets forecasted a tripling of volatility from mid--August onwards.

Phillipe Jorion (Predicting Volatility in the Foreign Exchange Market, Jnl. of Finance, 1995) finds that all time--series models add no value as compared with the volatility forecast implicit in option prices. The reason for this is not hard to see: some extremely clever people who know time--series econometrics are extracting every bit of forecasting that these models offer, and trading on it, thus bringing this information into the prices.

This gives us the paradoxical situation where this best of forecasting services is available at zero cost. In India, lacking options markets, this public service is unavailable, and forecasting volatility requires hiring econometricians. Once the options markets arrive, the econometricians will be hired into trading on the market, and the results of their work will become publicly visible at zero cost. This "information revelation" function is one of the ways in which options markets enrich the economy.

The main thesis here is that liquid markets produce sharp, informative prices. When a market has low impact cost, it rapidly assimilates a great deal of information and analysis into prices. Economists call such a market "efficient".

A market is "efficient" when all publicly available information is swiftly subsumed in the prices. This happens because a lot of extremely smart people are watching for every scrap of information, and rapidly trading on it. Their profits pay for the research, and make the market prices fully absorb the research. Economists think of a "valuable forecast not yet contained in prices" as a 500 rupee note lying on the ground. It is not likely to stay there for long.

This idea has momentous implications for the activities on the market. The most widely available, public information about a security is the historical prices. Hence any forecasting of future prices based on past prices should be futile. This simple reasoning explains the low usefulness of "chartism", which is the effort of using historical prices to forecast future prices.

Many studies have taken purely random "stock returns" (obtained by flipping coins), converted them into price time--series, and given these prices to chartists (try this as a reliable practical joke on your neighbourhood chartist). The chartists discover "heads and shoulders" and "candlesticks" in these series exactly the way they do with prices from real financial markets.

Another consequence of market efficiency lies in thinking about market prices absorbing announcements. When Reliance announced its bonus issue, this information reached many traders late. Yet, they went ahead to trade on Reliance on the next trading day. It would be useful for them to appreciate that there is no security in India which is more watched than Reliance, so there is unlikely to be any opportunity left on the trading day after the announcement.

I often meet traders who are convinced that their trading losses are caused by cartels who manipulate the market. I have seen fluent use of phrases like "the HM group" (I had to ask to find out that HM stood for Harshad Mehta) applied to various cartels which are supposed to exist. I am quite skeptical about the extent to which these matter.

Cartels appear on markets where the identity of traders is known -- e.g. the equity market before 1994, and the inter--bank markets in foreign exchange and debt today. However, both NSE and BSE today are anonymous markets, and cartel members have no way of enforcing the agreements that they have made. If two people conspire to buy heavily, each has an incentive to fink and sell to the other on the anonymous market.

Steady trading losses are rooted in the futility of speculative trading based on inferior forecasts. I tell people outside Bombay that when they drive past a factory and see a fire, it is their best chance at go to an NSE terminal and trade on this unique information. In doing this, they should drive faster than anyone else who might have the same idea (the 500 rupee note won't lie on the ground for too long). In contrast, speculation on Reliance without any unique knowledge about it is unlikely to yield profits; it only generates a steady bleed of brokerage fees and impact cost.

The spread of NSE terminals outside Bombay has hence greatly improved the quality of information going into prices -- most companies have factories outside Bombay. The "fire in the factory" is an obvious case of a broader idea: good information is known close to the shop floor or the regional sales office well before it appears in the head office or in the board room. This helps explain how the share of Bombay in the trading volume on NSE has now dropped to only 36%. Bombay simply matters less in forming prices -- traders in Bombay have inferior access to ground reality -- as compared with the way things worked before NSE.

The next time you meet a fund manager in Bombay who thinks that Grasim will go up, remind him that NSE terminals are in Madhya Pradesh also. Understanding market efficiency consists of insisting on this sanity check: What unique information or research underlies his forecast?

Back up to Ajay Shah's media page