The aim of this thesis consist of three parts. Firstly, the aim was to developan accurate historical inflation index suitable for the insurance business, usinginformation about insurance matters. The calculated inflation index was compared to an in-house benchmark at the insurance company Gjensidige, it wasfound to be a good match. Secondly, to determine the best model for explainingthe Swedish CPI inflation shocks, the thesis employed Multi Linear, RandomForest and XGBoost Regression. Thirdly, feature importance estimation wasconducted to identify which macroeconomic variables that were the most important in explaining inflation. Also, a time lag analysis was implemented tobetter understand with what delay these features best explain the inflation. Theresults revealed that Random Forest and Multi Linear Regression were the mostsuitable model candidates in terms of performance and transparency based onthe available dataset. Furthermore, the study found that unemployment rate,interest rate, and energy were the most crucial features in explaining inflation.It was also found that features with a low time lag entailed a high importance.The belief is that the study’s findings can assist insurance companies in developing a more agile product pricing process and sharpen their awareness towardsimportant macroeconomic variables. Overall, this study can be a valuable resource for insurance companies seeking to avoid underwriting risk and gain abetter understanding of the inflation’s underlying drivers.
Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:umu-209101 |
Date | January 2023 |
Creators | Liljestrand, Jacob, Nyberg, Fredrik |
Publisher | Umeå universitet, Institutionen för matematik och matematisk statistik |
Source Sets | DiVA Archive at Upsalla University |
Language | English |
Detected Language | English |
Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
Format | application/pdf |
Rights | info:eu-repo/semantics/openAccess |
Page generated in 0.0417 seconds