What causes the underpricing of auctioned IPOs? Experimental Evidence

1
What causes the underpricing of auctioned IPOs?
Experimental Evidence

• Panos Patatoukas, October 2007

2
IPO mechanisms in the US

• An IPO is the first sale of shares in a company
  to the public.
• Once an IPO occurs, the company will be listed on
  a stock exchange, and shares begin to trade.
• Currently there are two mechanisms for IPOs in
  the US the book-building and the auction-based
  method.
• In both mechanisms, the main role of the
  underwriter is to price and allocate the shares
  of the issuer.
• The key difference of auctioned from book-built
  offerings is the mechanism under which the
  underwriter carries out his role
• In the book-building mechanism the underwriter
  prices and allocates the new issue with complete
  discretion.
• In the auction mechanism the underwriter has
  limited discretion, both the pricing and
  allocations are market driven since both are
  determined in a uniform price sealed bid auction
  where all winning bidders pay the clearing price.
  

3
The underpicing phenomenon

• An issue is under (over) priced if the price
  received by the issuer in the primary market is
  lower (higher) than the price of the same
  securities in the secondary market.
• The underpricing phenomenon traditionally refers
  to the short-term performance of book-built IPOs
  and is typically measured by the pop in the new
  issuers stock price within the first few days of
  trading.
• The auction mechanism offers the promise of
  limiting underpricing. The argument is that the
  clearing price of a well-designed auction is
  virtually equal to the market price so the first
  day return should be zero .
• The question that surfaces is whether auctioned
  IPOs are, in fact, underpricing-free.

4
Auctioned IPOs are not underpricing-free

• In the US, over the period 1999-2007, the mean
  first day returns of completed auction-IPOs are
  14.27.
• In France, over the period 1992-1998, auction
  IPOs experience on average 9.70 returns on the
  first day of trading.
• The average underpricing level is 4.5 for
  auction-IPOs conducted on the Tel Aviv Stock
  Exchange from 1993-1994.
• Auctioned deals completed from 1993 to 2001 in
  Japan experience on average 11.40 first day pop.

5
US Treasury auctions

• The auction IPO mechanism is based on a model
  similar to that used by the US Treasury.
• Uniform price Treasury auctions also appear to be
  associated with underpricing since the average
  price received by the Treasury is less than the
  price of the same securities in the secondary
  market.

6
Motivation

• Theories to explain the underpricing of
  traditional IPOs usually examine the relations
  between the three actors of an IPO the issuer,
  the underwriter and the investors.
• Explanations which rely on the underpricing
  benefits to the issuer and/or the underwriter are
  not relevant for auctioned IPOs since the issues
  price and allocation is determined by investors
  in an auction and not by either the issuer or the
  underwriter.
• The empirical finding that auctioned IPOs are
  underpriced poses a puzzle.

7
Research Question

• Empirically investigate the causes of the
  underpricing of auctioned issues.
• My research question cannot be empirically
  evaluated with field data (at least for the time
  being).
• In order to investigate the causes of the
  underpricing of auctioned IPOs I develop a
  computerized laboratory market that enables me to
  test hypotheses grounded on economic theory.

8
The demand reduction effect (1)

• Vickrey (1961) demonstrates that in single-unit,
  uniform-price, sealed-bid auctions the bidders
  dominant strategy is to truthfully bid his
  valuation and thus uniform price auctions are
  allocatively efficient and maximize the sellers
  revenue.
• Vickrey also points out that the uniform-price
  auction is not expected to be demand-revealing if
  bidders have multi-unit demand, as in the case of
  securities auctions.
• When a bidder desires multiple units there is a
  positive probability that his bid on the second,
  or later units will be pivotal in determining the
  price that he pays on all units that he wins.
• So the bidder has an incentive to bid less than
  his true valuation on later units in order to
  reduce the price he pays on the earlier units.

9
The demand reduction effect (2)

• The demand reduction effect predicts that the
  issue will be underpriced even if the signals
  generating process is common across investors
  (information homogeneity).
• Signals noise
• As the noise in the signals generating process
  increases bidders have an incentive to behave
  more strategically (Kagel, 1995) and potentially
  exhibit more demand reduction.
• Number of Bidders
• As the number (N) of investors increases the
  probability that any given investor will be the
  pivotal bidder decreases and the demand reduction
  incentives decline.

10
The marginal bidder hypothesis

• (H3) In a market setting where all investors are
  symmetrically informed about the pure common
  value of the securities for sale, auctioned IPOs
  are underpriced and the underpricing is ceteris
  paribus (i) increasing in the noise of the
  signals generating process and (ii) decreasing in
  the number of bidders

11
Pilot Results (1)
Noise True Price Clearing price Closing Price Deviation of CP from TP
Round 1 1 18.7 17.6 18.6 -5.88
Round 2 2 13.2 12.3 13.8 -6.82
12
Pilot Results (2)
13
Pilot Results (3)
14
Concluding Thoughts

• Consistent with the marginal bidder hypothesis
• 1. the issue can be underpriced even if all
  investors are symmetrically informed.
• 2. the issue is more (less) underpriced when the
  noise in the common signals generating process is
  higher (lower).
• What about ?
• Experience Learning effects.
• Asymmetric Information Structure
• Price bubbles or false equilibria in the
  aftermarket.
• Speculative bidding in the primary market
• Merits of running pilot experiments