Abstract: This paper proposes a measure of investors’ expectation error by following the basic idea of Piotroski & So (2012) and discusses how expectation error premiums could exist in the market in the long run based on novel research by Lakonishok et al. (1994). Particularly, this paper investigates why naïve investors repeatedly make expectation errors in stock markets, while contrarian investors fail to eliminate mispricing to make a profit by effectively constructing risk-free arbitrage trading strategies. It is found that the existence of arbitrage risk makes it difficult to eliminate mispricing when stocks’ prices are higher than their fundamental value. This generates expectation error premiums as stocks’ prices converge to their fundamentals in price discovery and market trading. By investigating return characteristics on both sides of expectation error factor ERROR, this paper finds that the effectiveness of ERROR factors ultimately comes from significantly negative returns on the short sides caused by arbitrage risk. In contrast, the returns on the long sides have no significance. This paper conducts a difference-in-difference model under the quasi-natural experiment setting when China’s margin trading and Shanghai Stock Exchange 50ETF options were implemented and finds that the margin trading and options’ short-selling effectively reduces arbitrage risk of their underlying assets and then weakens the expectation error premium as securities’ prices are higher than their fundamental value. This research argues that expectation error premium is caused by stocks’ arbitrage risk that cannot be hedged under short-selling regulation.

Key words: arbitrage risk, expectation error, margin trading, stock index options