Join our community of smart investors

On contrast effects

Investors don't judge companies solely on their merits.
October 17, 2017

When the American kitchenware company Williams Sonoma launched a breadmaking machine it initially sold badly – until it introduced a more expensive model after which the cheaper one did well. Speed-daters are less likely to want to meet a partner again if their previous speed-date was especially attractive. Estate agents sometimes show clients a grotesquely overpriced property with the intention of making their other properties seem more attractive. And watching TV makes us unhappy by reminding us of the mundanity of our own lives.

All these are examples of the contrast effect: we judge things not just on their own merits but by comparing them to other things – sometimes even irrelevant ones. Recent research shows that the same thing happens in stock markets.

Samuel Hartzmark at the University of Chicago and Yale University’s Kelly Shue have discovered a pattern in investors’ reactions to earnings news. They show that if prominent big companies announce good results one day then the market reacts poorly to earnings announcements the following day by other companies even if these results are reasonably good. Conversely, the market reacts well to such announcements if the previous day’s news has been bad. These are examples of the contrast effect. Good news one day makes news the following day seem less good, while bad news makes average news the next day seem good.

Hartzmark and Shue estimate that a strategy of going short of shares due to announce results tomorrow if today’s results were good, and long of such shares if today’s results were bad, would have generated a return of 15 per cent per year in the US between 1984 and 2013.

There are good reasons why you might not want to follow such a strategy yourself. Nevertheless, there are important messages here.

One is that this is (yet more) evidence against the efficient market hypothesis. This theory says that existing information should be embedded into share prices. But the fact that yesterday’s price moves affect some of today’s moves is inconsistent with that idea.

Secondly, this might help explain some of the success of momentum investing and the phenomenon of post-earnings announcement drift, the process whereby prices underreact to good news and drift up in the following days. If good news comes a day after other companies have announced good news, that news will not look so good and so shares will underreact. Only later, as investors wise up to their mistake, will shares rise to reflect the good news. Hartzmark and Shue show that mis-reactions to earnings announcements caused by the contrast effect are corrected in the following two months.

Thirdly, this reminds us that financial professionals are not wholly rational. Despite their qualifications, experience and strong incentives, they are prone to systematic error just like the rest of us. What Lee Ross and Richard Nisbett wrote in 1980 in Human Inference (one of the first books on cognitive biases) remains true – that trained experts are prone to inferential error just as lay people are.

There might, though, be a broader implication here. Ever since the work of Richard Easterlin at the University of Southern California in the early 1970s, there has been a puzzle of why economic growth doesn’t seem to make us much happier. One reason for this – emphasised by Cornell University’s Robert Frank – could be that comparisons with other people’s high incomes makes those on decent incomes feel dissatisfied. That's another example of the contrast effect. In this sense, it has deep psychological and social effects.