Factors that affect how much women and men invest

Sammanfattning: A popular type of investments are financial investments. Even though the Swedish society aspires for equality, there are still financial differences between the sexes. This thesis project focuses on what factors that affect the amount that women and men invest in stocks. The aim is to obtain a deeper knowledge about the investment market and how it appears for women and men. The reason for this is to raise awareness about the inequality issue when it comes to the low representation of females in the investment market.  The objective for this project is to produce two models where one of them is for women and the other one for men. Those models have equal regressors: inflation, GDP growth, OMX Stockholm Price Index, average income per year and eight different age groups. The response variable is the average portfolio value for a specific age group at respective model. The data is taken mainly from Swedish Statistical Central Bureau, but is also gathered from NASDAQ, as well as the World Bank. The data is collected between the years 2000 to 2020.  The models are firstly evaluated at their full model, meaning that all regressors are included. The women's model shows a considerably good fitting, since almost all regressors are significant with low p-values along with a Multiple R-squared at 0.773 and Adjusted R-squared at 0.757. However, improvements can be made since outliers need to be removed, and the regressor income hold multicollinearity. Men's full model has a poor performance with fewer significant regressors and lower Multiple and Adjusted R-squared. Both models are then transformed, with applied square root of inflation at both models, and the square of income at men's model.  To determine the reduced model, variable selection is implemented. With Best Subset Selection, women's model includes OMX, GDP, income, age group 1, age group 6, age group 7 and age group 8. On the other hand, men's model has the regressors GDP, income, age group 1, age group 2, age group 3, age group 4, age group 5 and age group 7. Both these models perform much better where all the regressors had significant p-values and satisfactory Model Assumptions plots. In addition, no multicollinearity exists in the models. In conclusion, both the reduced models are chosen for final models.  Proposed further research within this topic is to include more appropriate regressors that may affect the portfolio value as well as to compare with other countries and include real investment instead of just financial investments.