Managerial sentiment, investor sentiment and stock returns
Abstract
It is well established that investor sentiment plays a vital role in global financial markets. However, the sentiment of other economic agents has received less
attention in the behavioural finance literature. This thesis aims to address the
impact of managerial sentiment on the UK stock market. It investigates the performance of managerial sentiment in predicting stock returns relative to investor
and consumer sentiments. In addition, it examines how sentiment is transmitted from managers to investors and whether the response of investor sentiment is
asymmetric towards positive versus negative managerial sentiment. Finally, this
thesis provides a comparative study of traditional and sentiment-augmented asset
pricing models.
Using monthly data from January 1985 to December 2014 and a sample of
consumer and business confidence indicators provided by the European Commission, the first chapter provides novel evidence on how managerial and consumer
sentiment indicators affect stock returns. The findings show no support for consumer confidence as a predictor of stock returns. However, managerial sentiment
shows a significant impact on aggregate market and sector return indices. Furthermore, results indicate that parameter estimates for sector groupings are not
consistent, implying that the sentiment-return relationship differs across sectors
and that parameters are sensitive to industry characteristics.
In the second chapter, the investigation extends to assess the long and short
run dynamics of the sentiment transmission from managers to investors. Using
threshold autoregressive (TAR), momentum threshold autoregressive (MTAR),
and asymmetric threshold vector error correction (ATVECM) models, the findings
provide evidence on the impact of managerial sentiment on investor sentiment in
support of the Catering Theory. Results show that investors’ sentiments converge
with long-run equilibrium relationships in response to positive rather than negative shocks in managerial sentiment. Furthermore, findings indicate that investor
sentiment reacts negatively to positive managerial sentiment with a delay of four months, suggesting an over-confidence in managers’ expectations of their future
business outcomes.
Finally, the third chapter examines the ability of managerial and investor sentiment to explain cross-sectional variation in stock returns. It compares the performance of CAPM, Fama & French (1993) Three factor model and Carhart (1997)
four factor model to sentiment-augmented asset pricing models, which incorporate
measures of both managerial and investor sentiment. The findings indicate that
inclusion of sentiment factors significantly adds to the power of traditional asset
pricing models to explain the cross-sectional variation in stock returns. In addition, results show that managerial sentiment outperforms investor sentiment in explaining three out of four test portfolios formed on size, book-to-market, volatility
and size/momentum factors. Moreover, findings show that managerial sentiment
exhibits stronger prediction power for size premium over short (1-3 months) forecasting horizon relative to investor sentiment. However, value premiums respond
to changes in managerial and investor sentiment over the relatively longer time of
12 months. In addition, the investigation failed to find any significant relationship
between sentiment indices and momentum premium.
This study has several implications for empirical researchers, practitioners
and policy makers. It provides academics who are concerned with the empirical
tests of asset pricing models with new insights on how the inclusion of managerial
sentiment impacts the performance of longer term investigated models. For practitioners, our findings suggest that managerial sentiment and its impact on sector
returns provide new opportunities for enhancing trading as well as asset allocation
strategies. In developing investment strategies, practitioners may consider sectors
that are more or less prone to sentiment in response to investor risk preferences. In
addition, results on sentiment-augmented asset pricing models may be of interest
to regulators who are concerned with the estimation of businesses’ cost of capital
when pricing public services.