A Modified Lee-Carter Model for Age-Specific Fertility Rate Forecasting: Modelling Ethnic-Based Fertility Change in Malaysia

Abstract

Accurate fertility prediction is crucial for long-term policy planning, particularly in health economics and demographics. This article proposes a modified version of the Lee-Carter model, which incorporates the skew-logistic probability density function to accurately capture the unimodal curves of age-specific fertility rates (ASFRs) in Malaysia. The model was fitted into Malaysian age-specific fertility data between 1958 and 2005, categorised by three major ethnic groups: Malay, Chinese and Indian. The forecast performances of both the modified and original Lee-Carter models were evaluated by estimating out-of-sample errors between 2006 and 2021. The model that demonstrated the highest accuracy was used to forecast ASFRs between 2022 and 2041. The analyses revealed a notable decline in overall fertility trends over the years, with particularly pronounced decreases observed among the Chinese and Indian populations. Furthermore, there has been a shift towards delayed childbirth, with the highest number of births occurring at older ages in recent years. The results indicate that the modified Lee-Carter model outperforms the original version for the Chinese and Indian populations, suggesting its ability to capture recent significant changes in fertility patterns and enhancing predictive accuracy.