Developing and validating regression models for predicting household consumption to introduce an equitable and sustainable health insurance system in Cambodia

Background Financial protection is a challenge for low- and middle-income countries, where the fiscal space is limited, and majority of the population is engaged in the informal economy. This study developed and validated household consumption predictive models for Cambodia to collect contributions according to one's ability to pay. Methods This study used nationally representative survey data collected annually between 2010 and 2017, involving 38,472 households. We developed four alternative models: the manually selected linear model, the linear model with stepwise technique, the mixed effects linear model, and the model with regularisation technique. Subsequently, we performed out-of-sample cross-validation for each model, and evaluated the model prediction performance. Results Overall, observed and predicted household consumptions were linearly related in all four models. While the prediction performance of the models did not substantially differ, the stepwise linear model showed the best performance. The regularisation and the mixed effects were not particularly effective in these regressions. The household consumption was better predicted for those with lower consumption, and the predictivity declined as the consumption level increased. Conclusions This study suggests the possibility of predicting household consumption at a reasonable level. This would maximise the contribution revenue, optimise the government subsidy, and ensure equity in healthcare access.