Erasmus University Rotterdam Erasmus School of Economics Master Accounting, Auditing and Control Master's Thesis Accounting, Auditing & Control Successful-Efforts
Book value model for the successful-efforts method
Download 0.83 Mb.
|
- Navigate this page:
- 6.2.2 Regression model
Book value model for the successful-efforts methodThe second paragraph provides the test results the book values for the successful-efforts method. First the assumption will be tested and afterwards the regression model. 6.2.1 Meeting assumptionsThe linearity assumption isn’t met by al variables, but according to Moore et al (2003, 447) a sample between 15 and 40 observation may have some skewness. This criterion is met by all variables because the sample is 38 observations and there is no extreme skewness. In appendix three the scatterplots show normality and homoscedasticity. In the first scatterplot there is one outlier. The second plot has several outliers. This is because it are absolute values. The scatterplots do not indicate that the variables do not meet the assumption homoscedasticity and linearity. 6.2.2 Regression modelThis section provides the scores from the regression model for the book values with the successful-efforts. This model maintains several variables that are: total assets before capitalized development expenditures; capitalized development expenditures; new capitalized development expenditures; amortization; and the mutations per variable. This is the model that scaled the variables with the market value of the company. The explanatory power is 0,206. The variables used in this model are not significant. Only the change in assets before capitalized development expenditures is significant. Perhaps the absolute models provide more significant variables.
The regression analysis with the same scaled variables and the control variables: common law / code law; and R&D intensity provided an explanatory power of 0,178. This is lower than the model without the control variables, but like the model above the variables are not significant. The controlled model doesn’t show a different relation than the uncontrolled model. Scaled controlled
The following two regression analyses are based on the absolute book values and returns. The first model is without control variables. This model scored an 0,902 explanatory power. This is very large and means that the variables explain approximately 90% of the returns. The variables of this model are significant. Only the three mutation variables development cost capitalized at year end; development cost amortized; and new development cost capitalized are not to be significant. The only limitation that should be kept in mind is the multicollinearity. This is explainable, because the variables are calculated from each other. The capitalized development costs are part of the assets. The new capitalized development costs are part of the capitalized cost at year end. These combinations explain the multicollinearity. Absolute value
This model is also checked with control variables. The control variables are common law / code law and revenue. The explanatory power is 0,932. This means the variables explain 90% of the returns. The relations in the above model do not change. The multicollinearity factors (VIF) are again above 10. This means that the variables do correlate with each other. This limitation should be kept in mind in analyzing the results. Absolute observations with control variables
Paragraphs one and two give the test results for the book values for both the cash expense method and the successful-efforts method. The results are not yet compared with each other in this chapter. This chapter only provides the test results. The comparison and conclusion will be discussed in chapter seven “Analysis”.
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||