Browsing by Subject "Stock return predictability"

Now showing 1 - 3 of 3

Open Access
Essays on investor attention
(2022-01) Bozok, İhsan
This thesis investigates the impact of firm centrality and macroeconomic uncertainty on attention allocation decisions of investors. First, we examine whether investor attention towards stocks is related to firm centrality in the input-output network. Using a data set of US firms’ principal customers, we find that stock prices do not promptly incorporate news about principal customers, generating return predictability which diminishes with the customer firm centrality levels. We show that this result is driven by limited investor attention. The evidence reveals that customer firms occupying more central positions in the network receive more investor attention. The results indicate that centrality effect is distinct from size effect. Our findings suggest that more central firms are associated with greater financial analyst coverage and greater institutional investor equity holdings. Second, we examine two competing theoretical models regarding the impact of macroeconomic uncertainty on investor attention to firm specific news. Kacperczyk et al. (2016) show that attention to firmspecific news decreases with the macroeconomic uncertainty, while Andrei et al. (2020) document that investor attention to firm-level news increases with the economic uncertainty. In line with the former hypothesis, we demonstrate that institutional investors raise their attention to customer news as economic uncertainty declines. However, there is no evidence of a meaningful association between retail attention to customer news and market-level uncertainty. We also find that stock prices incorporate customer news more quickly in times of low uncertainty, which is attributable to high institutional attention. During times of high uncertainty, stock prices underreact to customer news, generating return predictability which declines with an increase either in institutional attention or retail attention.
Open Access
Firm centrality and limited attention
(Elsevier BV, 2021-12-17) Bozok, İhsan; Özyıldırım, Süheyla
This study examines whether investor attention towards stocks is related to firm centrality in the input-output network. Using a data set of firms’ principal customers, we investigate the effect of firm centrality on the stock return predictability across economically linked firms. We find that stock prices do not promptly incorporate news about principal customers, generating return predictability which diminishes with the customer firm centrality levels. We show that this result is driven by limited investor attention. The evidence reveals that customer firms occupying more central positions in the network receive more investor attention. The results remain strong even when we make use of size-adjusted centrality values, indicating that the centrality effect is not driven by firm size. We find that more central firms are associated with greater financial analyst coverage and greater institutional investor equity holdings. Moreover, our analyses explicate that the source of information has an important role in attracting investor attention.
Open Access
Option-based variables and future stock returns in normal times and recessions
(Elsevier, 2024-10) Açıkalın, Özgür Şafak; Önder, Zeynep
We examine the prediction of future returns of optionable stocks trading in the US exchanges by several option-based variables for the period between 1996 and 2015. It is found that option-based variables are significant factors in estimating future stock returns in normal periods and during recessions. The spread between weighted averages of implied volatilities calculated with all call and put options of underlying stocks is found to have the highest effect on future stock returns. Although the mean squared errors of the option models are significantly higher during recessions than the expansion periods, the model with option-based variables outperforms the market model and the Fama-French Three Factor Model in both recessions and the whole sample period. The findings suggest that option-based models incorporate information about extreme events more than the traditional models.