that shorting Palm was incredibly expensive or that there was a large excess
demand for borrowing Palm shares, a demand that the market could not
meet for some institutional reasons.
Since the evidence from D’Avolio (2002) indicates a much lower, 35 per-
cent, shorting cost for Palm during this period, it is clear that there must be
other risks and costs associated with shorting Palm. First, there is the cost of
actually finding shares to borrow. Second, as discussed in Liu and Longstaff
(2000) and Mitchell et al. (2002), short-sellers are required to post addi-
tional collateral if the price of Palm rises. Third, as discussed in Mitchell et
al., there is “buy-in” risk, the fact that the Palm lender has the right to recall
his loan at any time. If the Palm lender decides to sell his shares after they
have risen in price, the short-sellers may be forced to close their position at a
loss if they are unable to find other shares to borrow. Fourth, even if the
loan is not recalled, the cost of shorting could increase if the rebate changes.
We now have three different market estimates of Palm’s value: the em-
bedded value reflected in 3Com’s share price, the value reflected in options
prices, and the actual share price. The options market and the shareholders
in 3Com seemed to agree: Palm was worth far less than its market price.
The direction of the deviation from the law of one price is consistent with
the difficulty of shorting Palm. To profit from the difference between the
synthetic security and the underlying security, one would need to short
Palm and buy the synthetic long. The price of the synthetic short reflects the
high demand for borrowing Palm stock and the low supply. Similarly,
Figlewski and Webb (1993) find that, in general, stocks with high short in-
terest have puts that are more expensive relative to calls (although they
look at implied volatilities instead of put-call parity).
Again, although the prices here are consistent with very high shorting
costs, one can turn the inequality around and ask why anyone would ever
buy Palm (without lending it). On March 17 one can create a synthetic long
Palm by buying a call and selling a put, and this synthetic long is 23 percent
cheaper than buying an actual share of Palm and holding it until Novem-
ber.^8 Arguments about the risk that the planned spin-off may not occur are
irrelevant to the synthetic long constructed using options. Why are in-
vestors who buy Palm shares directly willing to pay much more than they
could pay using the options market? One plausible explanation is that the
type of investor buying Palm is ignorant about the options market and un-
aware of the cheaper alternative.^9
156 LAMONT AND THALER
(^8) Of course, the put-call parity formula holds only for stocks paying no dividends. One ben-
efit of owning Palm is that it yields a “dividend” from lending it out to short-sellers. As before,
however, someone is holding all the Palm stock without lending it out; this owner would be
better off owning the synthetic short.
(^9) In the model of Duffie, Garleanu, and Pedersen (2001), all investors seek to lend the stock
out in order to reap lending income. Since the securities lending market works sluggishly, al-
though everyone is trying to lend, not all succeed, and at any one time 100 percent of the