Regression analysis issue

[Mirjam Medelli]

Hi!

We are writing a thesis about the relation between CSR (Corporate Social Responsibility) and Return on Assets and are performing regression analysis in STATA to be able to analyze the results.

Return on Assets is computed by multiplying the firm's Profit Margin with Asset Turnover and CSR is measured using an index over the years 2006-2014.

When performing a regression on CSR effect on Return on Assets, we get the results that higher CSR score has a negative effect on Return on Assets. But when performing a regression analysis on both the components (Profit Margin & Asset Turnover) we see that a CSR has a positive effect on both the components. This seems strange to us, as intuitively CSR effect on Return on Asset should have the same "sign" as CSR effect on the two components used to compute Return on Assets.

So our question is: does anyone know if this is even possible? Or if it's wrong? And if it is possible, how it is best explained? We have attached snapshots of our first regressions showing a negative sign between CSR and Return on Assets, but positive signs on Profit Margin and Asset Turnover.

[Carlo Lazzaro]

Mirjam:
if, as you stated, your research Group is performing what you're after with Stata (not STATA, please), you are kindly requested (per FAQ) to post what you typed and what Stata gave you back via CODE delimiters.
The reasons why screenshots or the like are not an efficent way to share what you have done with Others on the list are several; the first one in my personal rank is that they make commenting on whatsoever topic very uncomfortable.

[Phil Bromiley]

There is no necessary reason that the association of CSR with ROA has the same sign as the association of CSR with profit margins or asset turnover.

The normal way to calculate ROA is profit/total assets. By using profit margin and asset turnover, you're making it more complicated than it needs to be.

Something can have a negative association with ROA but not with profits or total assets. That is why we use ratios like ROA. For example, both profits and assets might go up with firm size but if profits go up more slowly than assets, ROA would go down with firm size.