Suppressor Variables and Sales in Altman’s Bankruptcy Model

Summary

The note discusses why sales volume, measured as sales divided by total assets, can improve the predictive value of EBIT relative to total assets in Altman’s bankruptcy model. It describes sales volume as a suppressor variable: a predictor that may have little direct relationship with the outcome while correlating strongly with another predictor. The author highlights a reported negative correlation between sales volume and EBIT among bankrupt firms and asks how to interpret it.

The document provides no resolution to that question and offers no new analysis or trading method. It cites Altman’s 1968 paper as the source for the reported relationship, but does not reproduce the underlying sample, estimates, or tests. The observation should therefore be treated as a prompt about predictor interactions and interpretation, not as evidence that higher sales or profit margins cause bankruptcy. The post is relevant to financial distress modeling and statistical feature selection, though its scope is narrow.

Key ideas

• Sales divided by total assets is presented as a useful suppressor variable for EBIT divided by total assets in Altman’s model.
• A suppressor can improve another predictor’s usefulness through its correlation with that predictor.
• The note reports a negative association between sales volume and EBIT among bankrupt firms.
• It raises, but does not answer, how to explain this relationship.
• The reported association alone does not establish a causal effect of sales or margins on bankruptcy.

Tags

# negative correlation between EBIT and sales in Altman's Z


# negative correlation between EBIT and sales in Altman's Z












In his 1968 paper, Altman found that sales volume (i.e., sales divided by total assets) is a useful predictor not by itself, but as a suppressor variable to improve the predictive power of EBIT/TA.

Suppressor variables are such that they do not or not significantly correlate with the dependent variable, but they do have a high (positive or negative) correlation with another independent variable.

In this case, Altman found a high negative correlation between sales volume and EBIT in the bankrupt sample. All other things being equal, this would imply that companies go bankrupt when they sell their wares at a higher profit margin?

I've been racking my brain as to why that is.

Altman's 1968 paper: https://pdfs.semanticscholar.org/cab5/059bfc5bf4b70b106434e0cb665f3183fd4a.pdf

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.