Abstract: Accurate estimation of claims is fundamental in the insurance industry for maintaining financial stability and ensuring effective risk management. Traditionally, deterministic approaches, including the chain ladder and Bornhuetter-Ferguson models, have been used to estimate reserves, considering their ease of implementation. However, these models fail to capture the inherent uncertainty associated with unpredictable future variations since they provide point estimates, and hence may result in improper reserve allocation (over-reserving and under-reserving). In contrast, the proposed stochastic model, specifically the bootstrapping technique, introduces a probabilistic framework to quantify reserve variability and provide a distribution of possible outcomes. The goal of this study is to evaluate the effectiveness of the stochastic model as compared to deterministic approaches in quantifying uncertainty, a subject that is largely underexplored in the Kenyan market. In particular, modeling is done for the Incurred but not Reported (IBNR) reserve using real data obtained from a local and already established general insurance company in Kenya (CIC Insurance). Both the deterministic and stochastic approaches are applied on the data, and the model performance is assessed based on accuracy in reserve prediction, mean square errors, confidence intervals, and volatility. The findings demonstrate the advantages of integrating stochastic models in the claim reserving process since they provide a detailed view of uncertainty. The insights support actuarial decision-making and enhance assessments of capital adequacy, hence protecting insurance companies against solvency risks. The study highlights the necessity of integrating stochastic approaches into reserving procedures to enhance robustness of actuarial valuation practices.

Keywords: Uncertainty, Claim Reserve, stochastic, Deterministic.

| DOI: 10.17148/IARJSET.2025.12641