Prof. Jayanth R. Varma's Financial Markets Blog

Photograph About
Prof. Jayanth R. Varma's Financial Markets Blog, A Blog on Financial Markets and Their Regulation

© Prof. Jayanth R. Varma

Subscribe to a feed
RSS Feed
Atom Feed
RSS Feed (Comments)

Follow on:

Sun Mon Tue Wed Thu Fri Sat

Powered by Blosxom

Mon, 30 Jan 2006

FASB may kill fair value accounting by permitting it

I have long argued that the alternative to fair value accounting is unfair value accounting and so I should normally be cheering the proposal by the Financial Accounting Standards Board (FASB) to permit fair value accounting for financial assets and liabilities. But actually, I am not happy at all.

The FASB’s Exposure Draft entitled The Fair Value Option for Financial Assets and Financial Liabilities “would create a fair value option under which an entity may irrevocably elect fair value as the initial and subsequent measurement attribute for certain financial assets and financial liabilities on a contract-by-contract basis, with changes in fair value recognized in earnings as those changes occur.” There are two problems with this exposure draft. First is that the fair value option can be exercised on a contract by contract basis allowing the company to chery pick profitable contracts to show on fair value basis while showing the loss making contracts on historical cost basis. The requirement that the fair value election is irrevocable provides only partial protection against this. The second problem that aggravates the cherry picking danger is that there are no safeguards at all on how this option can be exercised. Comparing its proposal with International Accounting Standard 39 (IAS 39), the FASB states:

This Statement has no eligibility criteria for financial assets and financial liabilities, whereas IAS 39 (as revised in 2005) indicates that, for other than hybrid instruments, the fair value option can be applied only when doing so results in more relevant information either because it eliminates or significantly reduces a measurement or recognition inconsistency (that is, an accounting mismatch) that would otherwise arise from measuring assets or liabilities or recognizing the gains and losses on them on different bases, or because a group of financial assets, financial liabilities, or both is managed and its performance is evaluated on a fair value basis, in accordance with a documented risk management or investment strategy, and information about the group is provided internally on that basis to the entity’s key management personnel.

The FASB proposal thus threatens to make fair value accounting very attractive to the scoundrels. The market recognizing this would penalize any entity that exercises this option. Thus fair value accounting would be killed by a proposal that professes to permit it.

I think fair value accounting should be the default method for all financial assets and liabilities. Companies should be allowed to irrevocably elect historical cost accounting (on an asset class by asset class basis) if they can show that this is more relevant and reliable because (a) market prices are not readily available and (b) fair values estimates have too much subjectivity.

Posted at 16:54 on Mon, 30 Jan 2006     View/Post Comments (0)     permanent link

Wed, 25 Jan 2006

Exchange Software Bugs Yet Again

Three instance of software glitches from Japan, United Staes and India during the last two months have convinced me that exchange software must go open source. This software is too important to be kept under wraps. The complete source code must be disclosed to the whole market to prevent recurrence of such problems.

Today’s Business Standard (N Mahalakshmi, “Sebi to audit NSE systems”, Business Standard, January 25, 2006) reports that the Securities and Exchange Board of India intends to conduct a systems audit of the National Stock Exchange (NSE) in response to the software bug in the computation of the index last week.

The NSE’s description of the error is as follows:

The special session for Reliance Industries Ltd was held from 8 a.m. to 9 a.m. so as to discover the price after the demerger. ... After the close of the special session the volume weighted average price for Reliance Industries Limited was Rs. 714.35. The adjustments to the base index value were suitably carried out to compute the index value so as to give effect to the demerger of Reliance Industries Ltd.

Trading was resumed as per normal market timings ... The market opened and the correct adjusted index value of NIFTY was also displayed to the market at the opening trade. The activity of NIFTY index computation was closely monitored after market opening and it was seen that the first few NIFTY index values were computed correctly taking into account the adjusted base index value. However once the first trade in Reliance Industries Ltd. was executed, it was observed that the NIFTY Index reflected incorrect value. The problem was analysed and found that due to memory initialization failure the last traded price being reckoned for index computation purpose was carrying an incorrect value. This resulted in a wrong NIFTY index value being displayed. The problem was identified and changes were carried out to reflect the correct value of the NIFTY index. The NIFTY index dissemination was stopped at 10.30 a.m and the correct display of NIFTY index value was made available to the market from 10.56 a.m onwards. The other indices remained unaffected.

This is the third serious exchange software bug that I have come across in the last two months. The other two errors happened in the two largest capital markets of the world:

  1. Last month, a software bug at the Tokyo Stock Exchange prevented a trader from cancelling a large erroneous order. I blogged about it here and here.
  2. Last week, a computer glitch at the Nasdaq casued closing prices of NYSE listed stocks to be misreported. Yahoo! News reported that “at approximately 5:50 p.m. Eastern time Wednesday, ... 16,669 transactions involving NYSE- and AMEX-listed stocks that had been made at 9:50 a.m. were reposted to the consolidated list. In many computer systems, those transactions overwrote the final closing price posted earlier that afternoon.

I am therefore completely convinced that exchange software must go open source. Alternatively, exchanges must take out large insurance policies to compensate any aggrieved party. By large, I mean something like 10% or 15% of the daily trading volume. For the NSE this may be therefore be in the range of a billion dollars.

Posted at 12:30 on Wed, 25 Jan 2006     View/Post Comments (3)     permanent link

Tue, 24 Jan 2006

UK Indexed Bond Bubble

John Plender has an interesting article in the FT (Risk aversion and panic buying, Financial Times, January 23, 2006) on the bubble in the UK inflation indexed bonds. Yields on the 50 year indexed bond have fallen to the extraordinarily low level of 0.38%. Plender argues that unlike other asset classes where bubbles arise from irrational exuberance, here it arises from panic or high risk aversion.

Compared to typical estimates of the historical average real long term interest rate of around 3%, the yield of 0.38% does appear ridiculously low. However, the situation is not so bad when we compare 0.38% with the historical average real short term interest rate of around 1%.

Morgan Stanley economists Richard Berner and David Miles discuss the issue of low long term yields in the US. They refer to the interesting FEDS paper by Don Kim and Jonathan Wright of the US Federal Reserve which decomposes the long horizon forward rate into four components. Recasting that analysis in terms of the real interest rate of a long term nominal bond we get three components:

The last of these is not present in an indexed bond and therefore the yield on an inflation indexed bond is likely to be lower than the real yield on a nominal bond. The interesting part is the real term structure premium. Kim and Wright show that this premium has collapsed from 2% in 1990 to 0.5% in 2005. From a theoretical point of view this premium can fall further and can in fact be negative. Only the liquidity preference theory of the term structure predicts a positive term structure premium. The expectations theory predicts a zero premium and the preferred habitat theory is agnostic about the sign of this premium.

Plender believes that indexed bond yields are depressed because pension funds are buying these assets for regulatory reasons and that the bubble could be pricked if either they turn to other assets or if the government could signal an intention to issue more long term indexed bonds. In the terminology of the preferred habitat theory, this merely states the truism that the term structure premium will change dramatically if some lenders or borrowers change their preferred habitat.

In the days when indexed bonds yielded say 3%, this yield would have decomposed into a expected short term real interest rate of say 1% and a term structure risk premium of say 2%. An yield of 0.38% would imply a term structure premium of -0.62% assuming that the short term real interest rate is unchanged. It is difficult to understand why a fall in the absolute value of the risk premium from 2% to 0.62% could be interpreted as a rise in risk aversion let alone as panic.

I share the view that there is a global asset market bubble and am quite sympathetic to the view that there is a bubble in UK indexed bonds as well. But I believe that Plender’s analysis is over simplified.

Posted at 12:44 on Tue, 24 Jan 2006     View/Post Comments (0)     permanent link

Wed, 18 Jan 2006

Best Price Rule in Takeovers

Two recent developments have brought into focus the right of all shareholders to receive the same price in a takeover (the “best price” rule). Many countries including India impose this requirement while the United States imposes it in a very narrow and almost meaningless way. One of the developments that I will talk about is that the United States is proposing to relax even further the already minimal best price rule that it has.

But I would like to begin with the United Kingdom. The Lex column in the Financial Times (“Lex: Virgin Mobile”, Financial Times, January 17, 2006) raises an interesting pricing issue in the proposed sale of Virgin Mobile to NTL in the United Kingdom. Lex calculates that if NTL rebrands its entire business as Virgin and pays the same royalty to Richard Branson as what Virgin Mobile pays currently, it would effectively add about 10% to what Richard Branson would get for selling his 72% stake in Virgin Mobile. Lex believes however that minority shareholders have no valid complaint:

Sir Richard does have more incentive than other shareholders to back the takeover, but under the City Code minorities do not have to sell out. They do not own the Virgin brand and have no independent entitlement to its value.

This is fair enough particularly because the Virgin brand is indeed separable from the cellular business. But in many other cases of this kind, there has been a problem in valuing the brand. Moreover, the brand is often inextricably intertwined with the business itself. There have been such instances in India as well.

The United States is proposing to cut right through this Gordian knot. Under its current regulations, the best price rule applies only to tender offers. If an acquirer takes a statutory merger route to an acquisition, the regulation does not apply at all. The two most important rules in tender offers are that:

  1. “The tender offer is open to all security holders of the class of securities subject to the tender offer ” (the all holders rule)
  2. “The consideration paid to any security holder pursuant to the tender offer is the highest consideration paid to any other security holder during such tender offer” (the best price rule)

Many (but not all) courts in the US have taken the view that the best-price rule applies to all integral elements of a tender offer, including employment compensation and other commercial arrangements that are deemed to be part of the tender offer, regardless of whether the arrangements are executed and performed outside of the time that the tender offer formally commences and expires. The US SEC believes that this interpretation has led many acquirers to disfavor tender offers in favor of statutory mergers where the best-price rule is inapplicable.

The SEC is therefore proposing amendments that establish that the best-price rule applies only to consideration “paid for securities tendered” instead of “during such tender offer” or “pursuant to such tender offer”. In addition, the SEC also proposes to introduce a blanket exemption for employment compensation, severance or other employee benefit arrangements.

The US regulations have always been fatally flawed because they provide almost no protection to minority shareholders in two step takeovers where large shareholders are bought at a high price and then other shareholders are bought out in a tender offer at a lower price. The best price rule looks only at price paid in the tender offer and does not look back to the price paid in transactions prior to the tender offer. Moreover, the ability to use the statutory mergers instead of tender offers has provided another loophole. It is strange that instead of plugging these glaring deficiencies in its regulations, the SEC is proposing to relax whatever little protections currently exist.

Posted at 13:46 on Wed, 18 Jan 2006     View/Post Comments (0)     permanent link

Mon, 16 Jan 2006

Financial Repression in China and India

Raghuram Rajan, Economic Counselor and Director of Research, International Monetary Fund says that “uniquely among fast-growing Asian economies, China has not raised its share of value added coming from high-skilled industries significantly, even as its per capita GDP has grown” and that the inadequacies of the Chinese financial system are to blame for this:

It is unlikely the chairman of a state owned corporation, cash rich because he no longer has to meet his social obligations to workers, will prefer to return cash to the state via dividends rather than retaining it in the firm, particularly when banks are under orders to restrain credit growth. And with financial investments returning so little, far better to reinvest cashflows in real assets. Indeed, liquidity plays a greater role than profits in determining real investments.

Similarly, the chairman of a private firm knows that financing from either the stock market or the state-owned banks is very uncertain. So he too will be unlikely to pay dividends, preferring instead to retain the capital for investment. Again, instead of storing this as financial assets and awaiting the right real investment opportunity, given the poor returns on financial assets, he has an incentive to invest right away.

These tendencies imply a lot of reinvestment in existing industries especially if cashflow in the industry is high, which inexorably drives down their profitability. And they imply relatively little investment in new industries. The inadequacies of the financial system would thus explain both the high correlation between savings and investment and the oft-heard claim that over 75 percent of China's industries are plagued by overcapacity. They also suggest why uniquely among fast-growing Asian economies, China has not raised its share of value added coming from high-skilled industries significantly, even as its per capita GDP has grown.

This is an interesting argument against financial repression. It is useful to remind ourselves once in a while that the occasional stock market scandal that we have seen in India since the beginning of economic reforms is a small price to pay for getting rid of financial repression. It is also necessary to recall that as a percentage of GPD, the annual losses to Indian households from financial repression were higher than the amount involved in even the biggest of the stock market frauds since 1991. For example, during 1980-81 to 1989-90, time deposits at commercial banks averaged over 25% of GDP. I have estimated that financial repression in the 1980s was about three percentage points. This implies that holders of time deposits at banks lost 0.75% of GDP annually. If we add the losses on other repressed financial assets (especially life insurance and provident funds), the total would certainly exceed 1% of GDP annually. By comparison, the total amount of fraud in the scam of 1991-92 (involving Harshad Mehta and others) was about 0.75% of GDP. The total amount of fraud in the scam of 2000-01 (involving Ketan Parekh) was less than 0.2% of GDP.

Posted at 14:26 on Mon, 16 Jan 2006     View/Post Comments (0)     permanent link

Mon, 09 Jan 2006

Some interesting titbits

The new year has brought with it a number of interesting titbits related to financial markets

Posted at 13:16 on Mon, 09 Jan 2006     View/Post Comments (4)     permanent link

Sat, 07 Jan 2006

Trading Error at BSE, India

Another new listing, another country, another trading error and a strange resolution. It is much smaller than the Mizuho trade that I have blogged about here and here. But its resolution is utterly strange.

On listing day a trader placed a sell order for about 400,000 shares of Tulip IT Services Limited at a price of Rs 0.25 at the BSE in Mumbai when the market price was around Rs 170. Since circuit filters do not apply on listing day in BSE, the trade executed causing a modest loss of about US $ 2.6 million.

What is interesting is the way that the exchange has resolved the issue. It says that:

  1. The buy orders at a price higher than Rs.96/- and matched against the said order, shall be deemed to have been transacted at such prices at which the trades were executed. The cut-off price of Rs.96/- has been arrived at by applying circuit filter limit of 20% on the issue price of Rs.120/- on the lower side on the lines of the existing practice of application of standard Circuit Filter of 20% in the regular market.
  2. All other existing orders below Rs.96/- and which got executed against the said sale order will be deemed to have been transacted at a price of Rs.171.15. The said price of Rs.171.15 has been arrived at by taking the weighted average price of the trades executed at or above Rs.96/- against the said sale order.
  3. All other orders placed subsequent to said sale order and which were matched against the said sell order will be deemed to have been transacted at a price of Rs.171.15.

In response my first Mizuho blog, Piyush Mishra commented that an erroneous trade is due to “the lack of oversight on the part of the broker/dealer” I agreed with that and wrote that “By nullifying erroneous trades, exchanges may actually be reducing the incentives for traders to install and use such software checks.”

I have therefore little sympathy for the BSE bailing out the offending trader by changing the traded price at all. But putting that objection aside for the moment, the solution adopted is still perverse. A trader who bought at Rs 97 pays Rs 97 while another who bought at Rs 95, is asked to pay Rs 171.15. That two traders in very similar situations are treated so differently is manifestly absurd and unfair. That the person who bid a lower price pays higher makes the solution even more ridiculous.

I recall a similar situation in the US in 2001. A hedge fund offered to buy Axcelsis Technolgies at $95 instead of $9.50 on the Nasdaq. Nasdaq cancelled all trades at prices below an arbitrary threshold of $22 and let the other trades stand. Floyd Norris wrote (“At the Nasdaq Casino, the Winners Get Stiffed”, New York Times March 2, 2001) about a day trader who sold into the erroneous trade and then covered his short position at a profit of $145,908. When the Nasdaq cancelled the trades selectively, this trader found that his share sales had been cancelled while his purchases stood producing a loss of $130,065 instead of the profit of $145,908. That the offending hedge fund was bailed out while an innocent day trader was penalized was clearly absurd. In a scathing comment, Floyd Norris wrote ”Nasdaq looks a lot like a casino that values a customer's business only until he starts winning.”

Clearly exchanges can not be trusted with the discretion that is vested in them. The rule should be very simple. Traders should bear the responsibility (and the losses) of their erroneous trades.

Posted at 17:17 on Sat, 07 Jan 2006     View/Post Comments (6)     permanent link

Wed, 04 Jan 2006

Tokyo Stock Exchange Trading System: Why Not Open Source?

The Financial Times reports that “The Tokyo Stock Exchange is considering replacing its trading system, even though it is merely a year old, following computer problems that have shattered the exchange's reputation and damaged Tokyo's status as a financial centre.”

Clearly the Mizuho trading error that I blogged about last month has been the main driving force behind this move. I would argue that open source is the better way to go if the goal is to make the trading system more robust.

I have been reading the official explanation that the Tokyo Stock Exchange (TSE) put out on the Mizuho incident. As I understand it the sequence of events was as follows.

Summing up the nature of the problem, the Tokyo Stock Exchange states:

“This is an incident that occurs when an issue is newly listed on the TSE directly and, as in this case, while a special bid quote is displayed, such a large amount of orders is placed that the net amount exceeds the number of the special quote order, and many sell orders still remain after the initial price is determined, to which deemed processing is then applied and then orders are placed at that price. As such, we are committed to strengthening supervision of newly listed issues in the near future and conduct extensive, detailed investigations of our system while considering the possibility of this and all other cases in the future, in ensuring irregularities such as this do not occur again. Also, the TSE will conduct a prompt, thorough analysis of the details of the cause of this recent irregularity in cooperation with the trading system developer, Fujitsu, Ltd.”

It appears that the irregularity that was observed would have occured only under very special circumstances that may never be repeated in future. It is also evident that in a complex trading system, the number of eventualities to be considered while testing the trading software is quite large. It is very likely that even a reasonable testing effort might not detect all bugs in the system.

Given the large externalities involved in bugs in such core systems, a better approach is needed. The open source model provides such an alternative. By exposing the source code to a large number of people, the chances of discovering any bugs increase significantly. Since there are many software developers building software that interacts with the exchange software, there would be a large developer community with the skill, incentive and knowledge required to analyse the trading software and verify its integrity. In my view, regulators and self regulatory organizations have not yet understood the full power of the open source methodology in furthering the key regulatory goals of market integrity.

Also, there is a case for simplifying the trading system. The trading system at TSE is unnecessarily complex because of the existence of price limits and the complex combination of call auction (Itayose method) and continuous auction (Zaraba method). TSE needs to question the very need for special quotes.

Posted at 14:37 on Wed, 04 Jan 2006     View/Post Comments (2)     permanent link