Determinants of debt level target

This paper examines the choice of debt level target. A survey was sent to CFOs of all Swedish
listed companies in 2005 and 2008. The main question of the survey concerned the debt level
target of the company. The answer from the survey was matched with accounting information,
obtained from the companies’ annual reports. There was missing data in both the survey and in
the accounting information. The missing data was treated with CART-based MICE imputation.
To find out the correct model specification, six penalizing models were used: grouped
multinomial lasso, ungrouped multinomial lasso, parallel ELMO, semi-parallel ELMO,
nonparallel ELMO, and cumulative GMIFS. One variable was found to have a significant effect
on debt level target: quick asset ratio. An ordered logit regression model was used to find out
how the quick asset ratio affects the debt level target. The results showed that an increase in
quick asset ratio reduces the prospected debt level target from a strict to a flexible or no target
debt level at all.