When marking to market fails

The idea of "valuing" a firm or a portfolio originated with accounting. Many centuries ago, accountants came up with a few remarkably simple ideas which first made systematic valuations possible -- these ideas included distinguishing stocks (balance sheet) from flows (profit and loss), and keeping track of the price at which all assets were purchased. This gives us the "accountants model" of value, where the value of a firm is measured using the book value of all assets and all liabilities. This was undoubtedly a great advance in the history of capitalism.

Book value is a crude measure when the "fair value" of every asset changes over time. The price at which the asset was once purchased becomes a historical detail of little importance given inflation, and the volatility of asset prices of the modern world. Modern economics suggests that the best measure of value is the price on a liquid, transparent market.

Indeed, when book value is used as the measure of value, we obtain the strange phenomenon of people who are willing to sell an asset at well below market value, as long as the sale price exceeds the price at which it was purchased (which assures them a "profit" in accounting terms)! Most Indian banks have made such errors not so long ago.

A respect for secondary market prices as a measure of value leads to the need for "marking to market", or revaluing assets using prices observed from a secondary market. It is common to use the last traded price (LTP) from the market for this purpose. Over the last decade in India, the idea of marking to market has increasingly become well established. The only notion of value that many people would respect is that which comes out of a full mark--to--market (MTM).

Traders who keep score based on the purchase price tend to wait too long on a position that has made losses, because they are unwilling to sell and take a book loss. In reality, the moment market prices drop, a loss has taken place. Keeping score using the MTM value instead of the book value ensures that traders look to the future and not at the past.

An accurate assessement of value is of essence in open-ended mutual funds. When a new unitholder comes into the fund, fairness demands that the price that he pays for the units reflects the true value of these units. If the incoming unitholder pays a price below the true NAV, he is being subsidised by existing unitholders, and vice versa. In addition, performance evaluation procedures would only make sense when comparing funds which use consistent valuation methodologies. Hence, the methods used in calculating the NAV assume considerable importance. In most mutual funds today, book value would play no role in calculating NAV.

Our focus so far has been on the new idea in India in the 1990s, the emphasis upon MTM value as opposed to book value. However, we should never lose sight of the foundation that is required for the MTM value to be the best estimator of value: a liquid and transparent market. If the market is illiquid, or if market prices are not accurately observed, then observed market prices can be poor estimators of value.

Stale prices.
When the last traded price (LTP) of a security is not recent, it is said to be "stale". When we observe the price of Reliance on the market, it is likely to accurately reflect the valuation of the market, since Reliance trades once each second on NSE. When we observe the price of an illiquid stock, like SAIL, on the market, the trade is likely to have taken place a while ago. This price is stale.

The bid--ask bounce.
A stock has a bid/ask of 90/100. Suppose no news breaks on the market, but trades take place at random at the bid or at the ask. Then we will see LTPs randomly fluctuating between 90 and 100 (movements of +11% and -10%). These changes do not reflect changes in value, they are simply noise. Defining the market price as (bid+ask)/2 instead of LTP is an effective way to eliminate this problem.

The price of an illiquid security is easily manipulated. When the closing price is used in important procedures, there are incentives to manipulate the closing price. In Calcutta, payoffs on the stock options market are based on stock prices at 3:15, this leads to efforts to manipulate the 3:15 price. A procedure which reduces this problem, used at NSE, defines the closing price as the average price from 3:00 to 3:30.

A spectacular example of stale prices was seen on 16 January 1998, when the RBI raised the bank rate by two percentage points, a massive one--day interest rate move by world standards. The standard "disaster" scenario used in risk management in the US is a one--day change of one percentage point.

When interest rates rise, bond prices drop. The event should have generated a sharp drop in all bond prices. In reality, most corporate bonds did not trade on the 16th - they are illiquid and only trade infrequently. Mutual funds which used the LTPs of corporate bonds in their valuation had an amazing outcome where their NAVs were actually unscathed across 16 January! By world standards, it is unthinkable to have income funds with NAVs untouched across the largest--ever one--day rise in interest rates.

Unitholders who withdrew their money on 17 January at these incorrect prices were being subsidised by the unitholders who did not. This was an incredible arbitrage opportunity -- people could have withdrawn their money on the 17th at high NAVs, waited a few days for NAVs to drop, and come back into their old positions at lower prices. The profits of these arbitrageurs were at the expense of the innocents who attempted no such arbitrage.

Stale prices are hard to deal with. If a stock does not trade, nothing can remove the staleness from the LTP. Remarkably enough, in the case of the bond market, there is a fairly sound alternative to MTM, called "model--based valuation". The prices of a few liquid government securities would be used to estimate the treasuries yield curve. An offset would then be applied to obtain the AAA yield curve, which would be used for valuing all AAA corporate bonds. In similar fashion, a "fair value" for all bonds can be estimated.

When MTM is done to such theoretical prices, it is called "marking to model". Marking to model is widely used in the worldwide derivatives industry in dealing with illiquid derivative products.

Funds in India do not have a choice in adopting such ideas -- SEBI guidelines require the use of LTPs in valuation as long as they are less than 60 days old. Sixty days is an extremely long period; generally stale prices from even six hours ago should be considered unacceptable. This is clearly an area where SEBI and the fund industry need to improve practice. In the meantime, the easy profits described above are there to be had the next time interest rates move sharply. Such blemishes serve to keep `innocents' away from mutual funds.

Back up to Ajay Shah's media page