Life cycle stages defined based on cash flow patterns as a risk factor in the asset pricing model
Name: RENATO LOUREIRO FALLER
Publication date: 27/04/2017
Advisor:
Name![]() |
Role |
---|---|
PATRÍCIA MARIA BORTOLON | Advisor * |
Examining board:
Name![]() |
Role |
---|---|
CLÁUDIO MÁRCIO PEREIRA DA CUNHA | Internal Examiner * |
PATRÍCIA MARIA BORTOLON | Advisor * |
Summary: This study investigates the capacity of firms' life cycle stages to describe the return of stocks,
specifically, combined in the three-factor model of Fama and French. The approach proposes
the use of a factor constructed from the stages of the life cycle, called MMG - Maturity Minus
Growth -, which is the difference between the returns of the portfolios composed of mature
firms stocks and the returns of the portfolios formed by growth firms stocks as an alternative
to HML - High Minus Low - due to possible distortions arising from the book-to-market in
order to test whether there is gain in the capacity of model to capture the return of the stocks.
The sample is composed of non-financial companies listed on the BM&FBOVESPA in the
period from 2008 to 2016. For the classification of companies in stages of the life cycle is
employed the method of Dickinson (2011), in which combinations of the cash flow signals
are used to determine the stage the company is. In determining of the market factor, the
Ibovespa is used as benchmark and the T-Bond nominal rate (issued by the Treasury of the
United States of America) plus Brazil country risk is used as risk-free rate. Three regression
models are estimated: the first is the three-factor model in its traditional form; The second is a
four-factor model, in which there is the addition of factor derived from the life cycle stage -
called MMG; And the third is the modified three-factor model in which the HML factor is
replaced by the factor constructed from the stages of the life cycle. First, the regressions are
estimated for the period from 05/2009 to 10/2011, in an in-sample procedure. The results of
this approach indicate that the MMG factor is positively related to the returns of the study
stocks. Then the out-of-sample analysis is performed for the period from 11/2011 to 04/2016,
testing which of the models provides better forecasts for the portfolios. In the comparison of
the predictions related to the effective returns, the test of Diebold and Mariano (1995) is used
to verify which of the models presents precision in the forecast statistically superior to the
others. It is observed that the models 1 and 3 show balanced performance in the predictions of
returns. Further, the results, both in the in-sample approach and in the out-of-sample
approach, indicate a complementary relationship between HML and MMG factors. That is,
the MMG factor worked well for the portfolios WHERE HML did not work, and the opposite
also occurred.