Year 12s: what will be the first way you research your future options? >>

Regression Analysis

Hi,

I am doing a statistical analysis for a module. I am studying whether the environmental performance of a firm affects the stock price. I have got a sample of 13 environmentally friendly companies and 87 non-environmentally friendly companies. I have the share prices of each company every quarter for the last 5 years, but how can I see whether there is a correlation between being environmentally friendly and having a positive effect on the share price?

Surely if I give the environmentally friendly companies a score of 1 and the non friendly companies a score of 0 and find each average stock price increase in percentages for the last 5 years this wouldn't really show anything?

Please help.. I am seriously struggling!

Thanks in advanced

Reply 1

Gregorius

Original post by mushpantsbc

Do you have any latitude in the the analysis that you perform? Or has this been set in this way as a specific question? I ask this, because if I was asked this question by a client, I would be straight back to them to get them to ask a better question! Two obvious problems are that:

(a) Being "environmentally friendly" is not a binary outcome; rather it should be assessed through a multitude of quite complex measures. Summarizing such a set of measures in a single binary variable is extremely poor statistical practice.

(b) The share price on its own, unadjusted for other factors, is a dangerous instrument to use for such an analysis. In the first place, you must at the very least be able to adjust for market sector. Also, a company whose share price is soaring may be in a much better position to spend money on environmental issues (especially on corporate window dressing!) than one whose share price is tanking. Reverse causation is a great danger.

Having said that, there are two simple analyses for the problem as stated:

(a) Linear regression with average share price in the last year as outcome with "environmentally friendly" as a binary covariate along with baseline average shareprice as a continuous covariate. For the latter you might take average share price in the first year of data that you have. Don't do an analysis based on change in average share price - that may suffer from regression to the mean.

(b) Do a regression (preferably with spline fit) with share price as a longitudinal outcome and with time and "environmentally friendly" as covariates. Use robust standard errors to deal with the outcome correlation.

(b) is more complex and harder to interpret, but uses all the data.

Quick Reply

Latest

Articles for you

Dealing with depression at university

Three foolproof tips for giving a brilliant presentation at uni

How to revise for GCSE Maths: AQA explains what to do

Disabled Students’ Allowance - Eligibility

Regression Analysis

Quick Reply

Related discussions

Latest

Trending

Trending

Articles for you