BUSF_A01.qxd

Tests of capital market efficiency

Each of the dots on the graphs is one day’s price movement for a particular security,

plotted against that of the following day for the same security. Figure 9.2 reflects a

positive correlation, that is, it suggests that an increase in the security price on one day

will be followed by another increase on the following day. Figure 9.3 implies a

negative correlation so that an increase in price on one day would mean a fall on the

following day, and vice versa. Figure 9.4 shows what we should expect if the security

were traded in a weak-form efficient market: there appears to be randomness between

one day’s price movement and that of the next. Sometimes an increase is followed by

an increase, sometimes by a decrease, but with no patterns.

Many tests have sought to identify relationships between price movements on two

or more consecutive days or weeks, and found no such relationships, either positive

or negative, of significant size. This research shows that security price movements

closely resemble the sort of pattern that would emerge from a random number gener-

ator. Probably the most highly regarded of these serial correlation tests was conducted

by Fama (1965). Brealey (1970) and Cunningham (1973) conducted similar tests on

security prices in the LSE and found evidence of weak-form efficiency.

Most of the rules used by technical analysts have been tested. For example,

Alexander (1961) used a filter rule and found that abnormal returns could be made,

but as soon as dealing charges are considered the gains disappear. Dryden (1970),

using filter tests on UK security prices, came to similar conclusions.

Counter-evidence on weak-form efficiency

There is an increasingly large body of evidence of an apparent tendency for investors

to overreact to new information. There seems, for example, to be evidence that the

release of an item of news that reduces the price of a particular share tends to cause

the price to reduce more than is justified. This overreaction is subsequently corrected

Figure 9.4
Graph of a
security’s price on
one day (day t)
against that of the
following day (day
t++1) where the two
movements are
uncorrelated