low returns, and repurchases by high returns. The models we have dis-
cussed so far do not, however, shed light on the size anomaly, nor on the
dividend announcement event study.
- Application: Closed-end Funds and Comovement
6.1. Closed-end Funds
Closed-end funds differ from more familiar open-end funds in that they
only issue a fixed number of shares. These shares are then traded on ex-
changes: an investor who wants to buy a share of a closed-end fund must
go to the exchange and buy it from another investor at the prevailing price.
By contrast, should he want to buy a share of an open-end fund, the fund
would create a new share and sell it to him at its net asset value, or NAV,
the per share market value of its asset holdings.
The central puzzle about closed-end funds is that fund share prices differ
from NAV. The typical fund trades at a discount to NAV of about 10 per-
cent on average, although the difference between price and NAV varies sub-
stantially over time. When closed-end funds are created, the share price is
typically above NAV; when they are terminated, either through liquidation
or open-ending, the gap between price and NAV closes.
A number of rational explanations for the average closed-end fund dis-
count have been proposed. These include expenses, expectations about fu-
ture fund manager performance, and tax liabilities. These factors can go
some way to explaining certain aspects of the closed-end fund puzzle. How-
ever, none of them can satisfactorily explain all aspects of the evidence. For
example, management fees can explain why funds usually sell at discounts,
but not why they typically initially sell at a premium, nor why discounts
tend to vary from week to week.
Lee, Shleifer, and Thaler (1991), LST henceforth, propose a simple be-
havioral view of these closed-end fund puzzles. They argue that some of
the individual investors who are the primary owners of closed-end funds
are noise traders, exhibiting irrational swings in their expectations about
future fund returns. Sometimes they are too optimistic, while at other times,
they are too pessimistic. Changes in their sentiment affect fund share prices
and hence also the difference between prices and net asset values.^30
This view provides a clean explanation of all aspects of the closed-end
fund puzzle. Owners of closed-end funds have to contend with two sources
of risk: fluctuations in the value of the funds’ assets, and fluctuations in
noise trader sentiment. If this second risk is systematic—we return to this
A SURVEY OF BEHAVIORAL FINANCE 47(^30) For the noise traders to affect the differencebetween price and NAV rather than just
price, it must be that they are more active traders of closed-end fund shares than they are of
assets owned by the funds. As evidence for this, LST point out that while funds are primarily
owned by individual investors, the funds’ assets are not.