Price Forecasting Using Financial Technology by (GAMLSS) Theory of NIST AI Risk Management Framework (AI RMF) for Sustainable Development Goals

Abstract

Introduction: This study explores the relationships between sustainability rating, price forecasting rating, financial technology, and the NIST AI risk management framework (AI RMF) in the digital age.
Objectives: The purpose is to crystallize an advanced price unified AI theory considering risk classifications to reinforce AI-RMF that facilitates the comprehensive transformation reshaping the competitive InsurTech industry from traditional Environmental, Social, and Governance (ESG) to Sustainable Development Goals (SDGs) Agenda 2030. However, “Risk Measurement” is the biggest challenge of AI-RMF application, including risks related to third-party software, hardware, methodology, and data, which are not uniform at the international or insurer levels.
Methods: the paper introduces GAMLSS machine learning for price forecasting in insurance companies, and compares it with traditional models, LMM, GLMM, and GAMM. GAMLSS is a flexible general framework for fitting semiparametric univariate regression-type models allowing adjustments through parametric/non-parametric additive smoothing functions or linear/nonlinear functions. Additionally, the empirical analysis examines a subset of actual heavy-tailed data observed from 2019 to 2023 from a major Egyptian “Non-life Insurance Company”. Finally, the programming language “gamlss” packages software was used in data science.
Results: As a technological and economic instantiation, this article provides recommendations for ministries, regulatory bodies, and insurance companies to use GAMLSS algorithm as a price -
reliable methodology enhancing risk assessment methodologies, standardizing sustainability metrics, including data accuracy, refining disclosure formats, and evaluating the influence of (SDGs) reporting on stakeholders, considering value-based AI principles according to the Economic Cooperation & Development (OECD).
Conclusions: In conclusion, this study examined the sustainable relationships of price forecasting, digital technology, the NIST AI Risk Management Framework (AI RMF) and Sustainable Development Goals (SDGs) in 2030. The traditional Environmental, Social, and Governance (ESG) dimensions are extended to include economic and technological considerations. However, “Risk Measurement” represents the most critical challenge to unify the methodologies at the international level. So, this paper reviewed the development of risk-based pricing strategies: LMM, GLMM, GAMM models. And this study suggested GAMLSS machine learning algorithm as a price-reliable AI methodology in insurance companies. Subsequently, this article reached several results, as follows: Firstly, GAMLSS achieved the lowest value AIC and GDEV test (568098.8 and 567569.2, respectively), compared to LMM (694230.8 and 694198.8, respectively), GLMM (600055.1 and 600091.6, respectively), and GAMM (597063.2 and 597031.2, respectively) for price forecasting. Secondly, GAMLSS as a semi-parametric model represents Box-Cox t (BCT) distribution as the most accurate distribution compared to about 100 distributions within “gamlss” software packages instead of the classic exponential distributions such as GAMMA within GLMMs, and GAMMs for claim forecasting. Thirdly, GAMLSS introduces the Cubic-splines as the most accurate compared to the P-splines smoother algorithms and other gamlss algorithms for auto-insurance price. Finally, GAMLSS developed RS algorithm reduces the computational burden of the Maximum Likelihood Estimation (MLE) of the model, and does not require chronological ordering, and can avoid overfitting problems. Additionally, the article recommends the importance of focusing future research on applying the NIST AI Risk Management Framework.