Finding the right IPO process

The traditional method of doing IPOs is the fixed price offering. Here, the issuer and the merchant banker agree on an "issue price" - e.g. Rs.100. Then you and I have the choice of filling in an application form at this price and subscribing to the issue.

Extensive research has revealed that the fixed price offering is a poor way of doing IPOs. Fixed price offerings, all over the world, suffer from `IPO underpricing'. In India, on average, the fixed-price seems to be around 50% below the price at first listing; i.e. the issuer obtains 50% lower issue proceeds as compared to what might have been the case. This average masks a steady stream of dubious IPOs who get an issue price which is much higher than the price at first listing. Hence fixed price offerings are weak in two directions: dubious issues get overpriced and good issues get underpriced, with a prevalence of underpricing on average.

What is needed is a way to engage in serious price discovery in setting the price at the IPO. No issuer knows the true price of his shares; no merchant banker knows the true price of the shares; it is only the market that knows this price. In that case, can we just ask the market to pick the price at the IPO?

Imagine a process where an issuer only releases a prospectus, announces the number of shares that are up for sale, with no price indicated. People from all over India would bid to buy shares in prices and quantities that they think fit. This would yield a price. Such a procedure should innately obtain an issue price which is very close to the price at first listing -- the hallmark of a healthy IPO market.

Recently, in India, we had an issue from Hughes Software Solutions which was a milestone in our growth from fixed price offerings to true price discovery IPOs. While the HSS issue has many positive and fascinating features, the design adopted was still riddled with flaws, and we can do much better.

How did the Hughes issue work?

  1. The Merchant Bankers selected "syndicate members" who helped in selling the issue.
  2. Orders were collected by the merchant bankers or syndicate members (only), and submitted using the computerised IPO system created by NSE. The NSE system only accepted these orders; it did not reveal any information to investors.
  3. Investors could place, modify or delete orders in the "book building period" -- however they were doing this in the blind since the system gave them no information.
  4. The NSE system revealed information to the merchant bankers. The full database of orders was passed on by NSE to the merchant bankers, who could then use this for discretionary allocation of shares.

We can point out numerous flaws in this:

  1. From a basic economic perspective, the IPO is a relationship between the issuer (Hughes) and investors. It is hard to justify the requirement that orders only go to merchant bankers or syndicate members; this greatly shrinks the extent to which the IPO harnesses NSE's remarkable distribution machinery. A superior IPO process would involve investors going to any NSE terminal and placing orders.
  2. The essence of modern market design is superior information display for superior price discovery. Investors should be able to continuously see how many shares have been bid for; the shape of the demand function; the cutoff price, etc. Using this information they would be able to think more effectively about order revisions, cancellations, fresh order placement, etc. The Hughes IPO, done through the space age VSAT technology of NSE, uses standards of transparency associated with a 19th century market design.
  3. In the Hughes IPO, the merchant bankers specified a band, from Rs.480 to Rs.630, in which orders had to fall. The merchant bankers proved to be wrong, and the IPO suffered from severe underpricing. If we take price discovery seriously, we should let the market set the price.
  4. The Hughes IPO process was too elongated in time. The IPO process should obtain price discovery in a short time-period where everyone interested in the issue is trading on the screen at the same time. An issue that lasts for more than an hour raises the cost for participants to constantly monitor the order book and revise orders.
  5. The sale of part of the issue at a fixed price (discovered at the auction) reduces the size of the auction and raises the probability of market manipulation at the auction.
  6. There should be no discretionary allocation of shares; instead shares should be allocated purely by price--time priority.

From this perspective, we design an idealised IPO process:

  1. On Monday morning, the newspaper should carry an advertisement which is the prospectus of the IPO, which only talks about the firm and is silent on valuation.
  2. The IPO should take place on Tuesday evening, from 4 PM to 5 PM. The auction should be a simple uniform-price auction with full transparency. A picture of the demand schedule, and the cutoff price, should update on the screen in realtime.
  3. Investors should be able to go to any NSE terminal and place orders into the auction. This harnesses 5,000 odd computer screens in 300 cities all over India in the auction process. From the issuers perspective, NSCC should perform the credit enhancement exactly as it does on the equity market. At a legal level, all orders on the screen should be placed by NSCC, thus shielding the issuer from the credit risk associated with anonymous order placement.
  4. There should be no fragmentation of the shares on offer. All shares to be sold should go through a single auction. If a retail investor wanted to "access the IPO at prices close to the offer price" she would just place non--competitive bids at the IPO, where she bids to buy (say) 100 shares at the IPO price, whatever it proves to be.
  5. Allocation of shares in the depository should take place on Tuesday itself. There should be no physical shares. Trading on NSE should start on Wednesday (the next day). This gives us a one--day lag between the IPO and the start of trading.
  6. Firms below a certain size should be barred from this IPO market. Using prospective valuations based on P/E, P/B and P/S ratios prevailing for the industry, a size of Rs.100 crore or so should be required. Smaller issuers should go to venture capitalists and private equity funds.

This proposed IPO process sounds startlingly effective. To put it in perspective, it is part of the same disintermediation process that we have seen in other areas of the financial markets. With anonymous, electronic trading on the equity market, the broker/dealer has been fundamentally disintermediated out of secondary market trading, which is now dominated by the actions of buyers and sellers (and not intermediaries). In that same fashion, the IPO process proposed here uses technology to link up the issuer and the investor with a transparent pricing mechanism, and eliminates the traditional overheads of intermediation.

Back up to Ajay Shah's media page