Executive Compensation and Firm Performance: Analysis of US Firms

Instructions:
You have to submit the Stata do file as well as the final report. The do file should be self-sufficient in terms of explaining which lines display the outputs for a particular question. These can be added in the following format in the do file.

Your tables need to be incorporated using images, directly obtained through Stata. Each image (table) should show the Stata command and its output. These tables are not included in your word count. Each table should be followed by a discussion / interpretation of the results. Do not exceed the word limit.

You should use Times New Roman 11 point font on A4 pages with 2.5 margins from each side.

Each student has been assigned a specific industry (SIC). You have to do the analysis only for your industry. You can find the distribution by tabulating SIC in the Stata data file. 

Complete the tasks below:

Scenario: There is widespread media and investor scrutiny into CEO behaviour and its implications for firm policies. Some argue that CEOs are paid too much due to which they tend to implement very risky firm policies; others argue that such risky policies are justifiable as they lead to +ve NPV projects. That is, there is an improvement in firm performance. You are an external compensation consultant hired to look into the performance of US firms.

Download the Stata file “Executive Compensation and Firm Performance.dta”. It contains information on executive compensation of the CEO and various other firm financials. A list of key variables is shown in Appendix A.  

Answer the following questions for your industries (refer to point (4) of Instructions). You can find the SIC codes under the variable name (SIC): 

Descriptive Statistics: Present a table of summary statistics for all the variables included in the project. Amongst other statistics, include mean, standard deviation, 25th percentile, median, 75th percentile, number of observations. Interpret the statistics to inform the reader of sample characteristics. Label this table: Table1: Summary Statistics 
 

Winsorising: Winsorise the variables at 1%. Using the winsorised data report a new table of summary statistics (as above) and interpret any changes in the summary statistics as compared to Table 1. Label this table: Table 2: Summary Stats after winsorising     

where, CEO compensation could be measured as Total Direct Compensation (tdc1). Suggestions regarding the construction of both dependent and independent variables are given in Appendix A. You don’t have to include all of the variables but feel free to add relevant variables based on your reading of the literature. A list of some basic articles is shown in Appendix B, but it is not necessary that you include all the variables that have been analysed in these research papers. Note that this is just an indicative list of research papers and you are welcome to read more articles. Each student has to run BOTH regression models for their assigned Industry.

3.Multiple Regression: Run a multiple regression with the key determinants of firm performance/risk based on your reading of key articles in the field of executive compensation. Run this regression model for the assigned industry/industries over two separate time periods. For instance, you could run a cross-sectional regression for year=2016 and a second one for year=2015. A different student may run a regression for year=2007 and year=2010. Please justify your choice of the two time periods. Label this table Table 3: Multiple Linear Regression .
    
a.Carry out a test to detect and correct for heteroscedasticity and report your main results making sure to conclude whether there is homoscedasticity or not. 

b.Interpret the impact of Pay on Performance using model (1) taking care to include all relevant control variables.

c.Interpret the impact of Pay on Risk using model (2) taking care to include all relevant control variables.

Discuss the impact of other independent variables on firm performance (model (1)) and on firm risk (model (2)).

Are your results likely to be unbiased? Discuss