Henrik Herbst and Mikael Ingelsson

Non-Parametric regression of the relationship between company performance and capital structure 

ABSTRACT:
The discussion of capital structure and its connection to performance has for long 
been a hot topic, and presently it is a very discussed subject. Theories by Hossein 
Asgharian implies that the relationship between lagged leverage and perfor- 
mance has a sigmoidal shape. In this thesis we further examine this theory on 
Swedish companies included in Affärsvärldens Generalindex (AFGX). 
We will do this with non-parametric models in order to find the inflection point 
in which the level of lagged leverage is optimal to increase the performance mea- 
sure we investigate, sales growth and change in profitability. To achieve this we will 
foremost use unimodal regression with Minumum Lower Sets Algorithm (MLSA) 
where we will assume the set below the inflection point to be increasing and the set 
after the inflection point to be decreasing. A difficult part of the investigation is to 
find a consistent estimator of the mode.