BUSF_A01.qxd

(Darren Dugan) #1

Chapter 9 • The secondary capital market and its efficiency


It is interesting to note that the consensus view outperforms all individuals over the
three years and indeed outperforms all but one or two in any particular year (it tied
with two individuals in 1966 and was beaten by one individual in 1967 and 1968). It is
clear from the table that the performance of the successful individuals is inconsistent,
suggesting some element of luck in their successful year. Luck may not be the only
reason for the success, since Biondo performed better than average in all three years.
Yet despite the possible presence of skill in one individual, over the three years the
consensus easily beat them all.
It would appear that some forecasters are more skilled than others. It also seems
that the consensus performs even better than the best individual. This is despite the
fact that the consensus combines the forecasts of all the individuals, skilled and not so
skilled. Far from having the effect of dragging down the quality of the forecasts of the
best individuals, combining the forecasts, to find the consensus, actually improves the
quality of forecasting. Beaver suggests that this is because idiosyncratic factors (such
as personal loyalties to a particular football team), which might influence the fore-
casts of even the best forecaster, tend to cancel out when a reasonably large number
of different individuals is involved in forming the consensus. Thus the consensus
represents a rather more clear-sighted and objective forecast than any individual can
provide, on a consistent basis. This is rather like the portfolio effect of equity investing
that we met in Chapter 7. Random factors (specific risk, in the case of investing) can-
cel out, leaving only the common factor (systematic risk).
Beaver (1998) also refers to a similar effect with predictions of UK gross domestic
product by 30 economists. Here the consensus beat 29 and tied with one of the
economists.
Another example of this seems to arise in the context of the TV quiz show Who
Wants To Be A Millionaire? According to Surowiecki (2004), the ‘friends’ (who are
selected by the contestant for their expertise) answer correctly 65 per cent of the time,
whereas the audiences (each one a random selection of quite a large number of
people) supply the correct answers to 91 per cent of the questions asked of them. This

Table 9.1Performance of forecasters of American football games

1966 1967 1968

Number of forecasters (including the consensus) 15 15 16
Number of forecasts per forecaster 180 220 219
Rank of leading forecasters:
J. Carmichael 1 (tie) 8 16
D. Nightingale 1 (tie) 11 5
A. Biondo 7 1 6
H. Duck 8 10 1
Rank of the consensus 1 (tie) 2 2

Source: Chicago Daily News

The Chicago Daily News, on each Friday over the period 1966 to 1968, reported the pre-
dictions of each of its 14 or 15 sports staff of the outcome of the games to be played
over the forthcoming weekend. The newspaper also published the consensus view of
the sports staff, that is, the single most popular view on each game. When the success
of the predictions was summarised for the three years, the results were as shown in
Table 9.1.
Free download pdf