ANALYSIS OF FACTORS AFFECTING INSURANCE CLAIM FREQUENCY IN NIGERIA

CHAPTER ONE

INTRODUCTION

1.1 Background to the Study

Claim frequency the expected number of claims arising from a policy per unit of exposure is one of the two multiplicative components of the pure premium, the other being claim severity. Because frequency is a count variable, it is modelled by discrete probability distributions, most commonly the Poisson when the mean and variance are approximately equal and the negative binomial when the data are overdispersed, as they typically are once heterogeneity among policyholders is present. Generalised Linear Models extended this framework by allowing observable characteristics of the insured person, the insured property, the geographic location and the policy itself to enter as covariates, so that frequency can be estimated conditionally rather than as a portfolio average (Klugman, Panjer, & Willmot, 2019).

Identifying which factors actually drive frequency is the substance of risk classification, and risk classification is what distinguishes actuarial pricing from flat-rate pricing. Gschlößl and Czado (2007) demonstrated that spatial location is a significant determinant of both claim frequency and claim size in non-life insurance. More recent work has expanded the factor set considerably: Meng, Gao and Huang (2022) incorporated driving-behaviour features from telematics devices into frequency models using boosted trees, and Gao, Huang and Meng (2023) showed that telematics-derived driving risk measures materially improve the interpretation and prediction of automobile claim frequency. Clemente, Guerreiro and Bravo (2023) applied gradient boosting to motor frequency and severity, finding that machine-learning estimators capture interactions among rating factors that GLMs with linear predictors miss.

In Nigeria, the factors bearing on claim frequency are distinctive and, in several respects, more severe than in the markets where this literature was developed. Road infrastructure quality has been shown to affect both the incidence of motor accidents and the value of the resulting claims in Nigerian urban settings. Only about a quarter of Nigerian vehicles carried insurance at the end of 2023, which creates a selection effect on the insured pool. Security incidents, flooding, fire safety deficiencies in commercial buildings and fraudulent claim submission all contribute to observed frequency in ways that have no close analogue in the datasets on which standard rating structures are built. The industry context reinforces the importance of getting this right: gross claims across the industry stood at ₦724.7 billion in 2025, with the non-life segment settling 75.5 per cent and the life segment 65.5 per cent of reported claims (National Insurance Commission [NAICOM], 2026).

Despite this, Nigerian insurance pricing remains heavily reliant on flat tariffs and broad class ratings that incorporate few of the factors the international literature has identified as significant. The consequence is adverse selection: low-risk policyholders subsidise high-risk ones, perceive the premium as poor value, and withdraw, leaving the insurer with a deteriorating risk pool. This study examines empirically which factors significantly affect insurance claim frequency in Nigeria and quantifies their effects.

1.2 Statement of the Problem

Nigerian insurers price largely without reference to the empirically established determinants of claim frequency, and the Nigerian literature offers little guidance because it has not systematically established what those determinants are in the local context.

First, risk classification is underdeveloped. Compulsory third-party motor insurance is sold at a prescribed flat premium regardless of the driver's age, experience, vehicle characteristics, annual mileage or location. A structure of this kind guarantees cross-subsidy between risk classes and provides no incentive for loss-reducing behaviour.

Second, the factor set has not been validated locally. Rating variables inherited from foreign practice age, sex, vehicle value, no-claims discount may be weak predictors in Nigeria relative to factors such as road condition, security environment, vehicle age and origin, driver training, and urban congestion, which the international datasets do not contain. Which factors matter, and how much, is an open empirical question in Nigeria.

Third, overdispersion and zero-inflation are unaddressed. Insurance portfolios generate a large proportion of claim-free policies and a small proportion of high-frequency policies. Where average-based pricing is used, this heterogeneity is ignored, and the variance of the claim count which drives the capital requirement is understated.

Fourth, claim fraud contaminates observed frequency. Fraudulent and exaggerated claims have been documented in Nigerian insurance, so observed claim counts may not be a clean measure of underlying risk, complicating both estimation and interpretation.

The problem, therefore, is that the determinants of insurance claim frequency in Nigeria have not been rigorously identified or quantified, leaving insurers unable to classify risk, price equitably, or hold capital commensurate with the variability of their claim experience.

1.3 Aim and Objectives of the Study

The aim of this study is to analyse the factors affecting insurance claim frequency in Nigeria.

The specific objectives are to:

1. examine the distribution and descriptive characteristics of claim frequency in the study portfolio;

2. determine the effect of policyholder characteristics (age, sex, occupation and claims history) on claim frequency;

3. determine the effect of risk-object characteristics (for motor business: vehicle age, type, usage and value) on claim frequency;

4. assess the effect of geographic and environmental factors (location, road condition, urban or rural classification) on claim frequency;

5. fit and compare competing count regression models (Poisson, negative binomial and zero-inflated variants) to the claim frequency data; and

6. rank the identified factors by the magnitude of their effect on expected claim frequency.

1.4 Research Questions

1. What are the distribution and descriptive characteristics of claim frequency in the study portfolio?

2. What effect do policyholder characteristics have on claim frequency?

3. What effect do risk-object characteristics have on claim frequency?

4. What effect do geographic and environmental factors have on claim frequency?

5. Which count regression model best describes the claim frequency data?

6. Which factors exert the greatest influence on expected claim frequency?

1.5 Research Hypotheses

The following null hypotheses will be tested at the 5% level of significance:

H₀₁: Policyholder characteristics have no significant effect on insurance claim frequency in Nigeria.

H₀₂: Risk-object characteristics have no significant effect on insurance claim frequency in Nigeria.

H₀₃: Geographic and environmental factors have no significant effect on insurance claim frequency in Nigeria.

H₀₄: Claim frequency data in Nigeria are not significantly overdispersed relative to the Poisson assumption.

H₀₅: There is no significant difference in goodness of fit between the Poisson and negative binomial regression models for Nigerian claim frequency data.

1.6 Significance of the Study

For insurance companies, the identification and ranking of significant frequency factors supplies the empirical basis for a risk classification and rating structure, permitting differential pricing that rewards low-risk policyholders and prices high-risk ones adequately. For NAICOM, the findings bear on whether prescribed flat tariffs most notably in compulsory third-party motor insurance remain defensible, and on the calibration of underwriting risk within the risk-based capital framework introduced by NIIRA 2025. For underwriters and claims managers, significant factors serve as underwriting screens and as indicators for the triage of claims requiring closer investigation. For the Federal Road Safety Corps and state road agencies, evidence that road condition and location significantly affect claim frequency strengthens the economic case for infrastructure investment. For policyholders, risk-based pricing produces greater equity in premium allocation and creates incentives for loss prevention. For scholarship, the study extends a frequency-modelling literature developed largely on European and North American portfolios into a Nigerian setting with materially different risk drivers.

1.7 Scope of the Study

The study covers insurance claim data obtained from one or more insurance companies licensed by NAICOM in Nigeria, over a period of not less than five consecutive years. The primary class of business examined is motor insurance, chosen because it generates the largest volume of individual claims and because the candidate rating factors are observable; where data permit, fire and general accident business will be examined comparatively. The explanatory variables comprise policyholder characteristics, risk-object characteristics, policy characteristics (sum insured, deductible, cover type) and geographic or environmental variables. Analysis is confined to count regression modelling of claim frequency; claim severity and aggregate loss are outside the scope except where referenced for context. Secondary data on road conditions, crash statistics and location characteristics are drawn from the FRSC Statistical Digest and National Bureau of Statistics publications.

1.8 Limitations of the Study

(i) Covariate availability Nigerian policy administration systems often capture only a minimal set of fields, so rating factors known to be significant elsewhere (annual mileage, driver training, telematics behaviour) may simply be unrecorded. (ii) Exposure measurement accurate frequency modelling requires exposure measured in policy-years, and mid-term cancellations, lapses and renewals are not always recorded in a way that permits exact exposure calculation. (iii) Under-reporting of small claims policyholders frequently absorb minor losses rather than claim, in order to preserve no-claims status or avoid settlement delay, so observed frequency understates true incident frequency. (iv) Fraud contamination fraudulent claims inflate observed frequency and are not separately flagged in most datasets. (v) Selection bias with roughly three-quarters of Nigerian vehicles uninsured, the insured pool is not representative of the vehicle population, limiting the generalisability of estimated factor effects. (vi) Data confidentiality insurers may supply only anonymised extracts, restricting geographic granularity.

1.9 Operational Definition of Terms

Claim frequency: The number of claims arising from a policy per unit of exposure, usually expressed per policy-year.

Exposure: The amount of risk carried, measured in policy-years or vehicle-years, used as the offset in count regression models.

Risk classification: The grouping of policyholders into classes expected to exhibit similar claim experience, for the purpose of differential pricing.

Rating factor: An observable characteristic of the policyholder, the insured object or the policy used to determine the premium.

Overdispersion: The condition in count data where the variance exceeds the mean, violating the equidispersion property of the Poisson distribution.

Zero-inflation: The presence of more claim-free policies than the assumed count distribution predicts.

Generalised Linear Model (GLM): A regression framework in which a transformation of the mean of a response from the exponential family is modelled as a linear function of covariates.

Adverse selection: The tendency for higher-risk individuals to purchase insurance disproportionately when premiums do not differentiate by risk.

No-claims discount (NCD): A premium reduction granted to policyholders with a claim-free record, functioning as an experience-rating mechanism.

References

Clemente, C., Guerreiro, G. R., & Bravo, J. M. (2023). Modelling motor insurance claim frequency and severity using gradient boosting. Risks, 11(9), 163. https://doi.org/10.3390/risks11090163

Federal Republic of Nigeria. (2025). Nigerian Insurance Industry Reform Act, 2025. Federal Government Press.

Federal Road Safety Corps. (2024). FRSC statistical digest, first quarter 2024. Federal Road Safety Corps.

Gao, Y., Huang, Y., & Meng, S. (2023). Evaluation and interpretation of driving risks: Automobile claim frequency modeling with telematics data. Statistical Analysis and Data Mining, 16(2), 97–119. https://doi.org/10.1002/sam.11599

Gschlößl, S., & Czado, C. (2007). Spatial modelling of claim frequency and claim size in non-life insurance. Scandinavian Actuarial Journal, 2007(3), 202–225. https://doi.org/10.1080/03461230701414764

Klugman, S. A., Panjer, H. H., & Willmot, G. E. (2019). Loss models: From data to decisions (5th ed.). John Wiley & Sons.

Meng, S., Gao, Y., & Huang, Y. (2022). Actuarial intelligence in auto insurance: Claim frequency modeling with driving behavior features and improved boosted trees. Insurance: Mathematics and Economics, 106, 115–127. https://doi.org/10.1016/j.insmatheco.2022.06.001

National Insurance Commission. (2026). Bulletin of the insurance market performance: Fourth quarter 2025. NAICOM.

Nigerian Insurers Association. (2024). Nigeria insurance digest 2023. Nigerian Insurers Association.

Shi, P., Feng, X., & Ivantsova, A. (2015). Dependent frequency–severity modeling of insurance claims. Insurance: Mathematics and Economics, 64, 417–428. https://doi.org/10.1016/j.insmatheco.2015.07.006