This paper presents the first analysis in which the effects of both education and ageing on CHI demand in China are investigated across years while taking into account regional differences. Overall, results from the analysis support existing evidence which suggests that the effects of education and ageing on CHI demand are positive and statistically significant. The results also build on previous evidence, highlighting how regional disparities result in varying demand for CHI. These findings have wider implications for policy makers involved in health and health financing reform, insurance companies and other researchers. This section will expand on and interpret the results, as well as discuss key limitations of the study.
Findings from the national and regional analyses demonstrate that an overall improvement in education levels can increase CHI demand in China. Based on the literature, the main reasons for this are likely that higher education levels generally contribute to a better understanding of the function and need for CHI or insurance more widely, a better awareness of health risk mitigation and an improved ability to understand insurance claims processes and insurance products . Individuals with higher education levels are therefore more likely to purchase CHI, and may also be better able to afford CHI premiums. However, the increase in the education level of a province is mainly driven by the improved education of young people . In turn, younger populations with higher education level are more likely to purchase CHI for themselves as well as for their older relatives or dependents .
Taking into account regional differences, the positive effect of education level on CHI demand is significant in the Eastern region but not significant in the Central and Western regions. A possible reason for this is that there are more insurance companies in Eastern China, compared with other regions, which provide a number of CHI products to meet the various needs of consumers with different characteristics . Individuals living in the Eastern China that want to buy CHI can therefore find the insurance products and services they need more readily than other regions. In contrast, CHI is less developed in Central and Western regions where people are less likely to find CHI products and services suited to their needs and will instead resort to other financial protection mechanisms . Education level may therefore have a larger effect on CHI demand in Eastern China than other regions where there is less formal or service sector employment and less CHI products are available. A second reason that a significant positive relationship was only found between education and CHI demand in the Eastern region could be that there is more awareness of CHI in the Eastern region . Individuals that are better aware of CHI and understand CHI products as well, as how they can mitigate health risks, due to higher education levels are more likely to choose CHI as a risk protection tool . It is also important to note that the likely reason why a significant effect is observed for education levels in the national sample, despite no significant effect in Central and Western regions, is the growth of CHI markets in Eastern China. Eastern China, which is more developed than the national average across all of the variables considered in this analysis, leads to the development and growth of the CHI markets nationally [14, 43] – though this currently may not be sufficiently pronounced to result in a significant effect when regions are considered separately.
The national sample results on the effect of education on CHI demand presented in this paper are consistent with the wider literature [20, 30]. However, the regional sample results differ from a study carried out by Suo et al. (2015) . The latter report that the effect of education on CHI demand is more significant in the Central and Western regions than in Eastern China. This divergence of results may be due to the different samples used. Suo et al. (2015)  included both urban and rural populations in their sample, while the analyses in this paper focussed only on urban populations to minimise potential bias in the results given that current CHI market data are predominantly from urban areas. Central and Western regions are comprised of a higher rural population than Eastern China and overall education levels in rural areas are substantially lower than in urban areas [1, 44]. By including rural populations, the average education level therefore decreases, and the difference of education levels between the Central and Western regions compared with Eastern China would be larger. Subsequently, the marginal effect of education levels on CHI demand in Central and Western China would be amplified, which may result in a higher significance level for the effect of education than in the Eastern China.
In line with previous studies [19, 30] the analyses presented in this paper find that ageing has a significant positive effect on CHI demand. This can be explained by the fact that as individuals age, the probability of experiencing chronic disease increases as well as the likelihood of incurring associated medical expenses due to hospitalization and outpatient visits [4, 40]. Given that CHI can improve access to healthcare services, and in some cases financial risk protection from medical expenses, the demand for CHI increases accordingly with age [4, 40]. In addition, the increasing trend of ‘empty nesters’ in China, where children move out of parental households, also boosts CHI demand. This is because CHI can reduce the burden for carers associated with supporting dependents. As the number of empty nesters increases, CHI demand is therefore also expected to grow .
This paper presents the first analysis of CHI demand that considers the differences in ageing across regions in China. Placing these findings in the literature is therefore challenging. The regional regression results show that the effect of ageing on CHI demand is significant in the Eastern and Central regions but not in Western China. The reason for this could be due to the difference in the elderly dependency ratio between regions. As shown in the summary statistics in Table 4, the average elderly dependency ratio in Eastern and Central China is 13.8 which is higher than in the Western region (12.8). Individuals in the Eastern and Central regions are therefore more likely to experience chronic disease and associated expenses, and have developed a better understanding of the risks associated with ageing as well as how the operation and functions of CHI can mitigate these risks . As the elderly dependency ratio continues to rise, people living in Eastern and Central China are therefore more likely to respond to the changes by purchasing CHI products.
In terms of control variables, the significant positive effect of health expenditure per capita on CHI demand in the national sample likely reflects that the willingness to pay and prioritise health over other spending from the demand-side  is greater than the effect of risk aversion among insurance companies from the supply side . Findings are similar for the Central and Western regions, but no significant effect is observed for Eastern China. This may be because the average CHI premium is comparatively higher in Eastern China [43, 45], which may deter consumers that experience a substantial disease burden from purchasing CHI.
Disposable income per capita also has a significant positive effect on CHI demand in the national sample, with the largest coefficient, which suggests that increases in overall household income drive demand for CHI . However, disposable income per capita is not found to have a significant effect on CHI demand in any of the regions. This may be because, first, income levels in the Eastern and Central regions are already high, and an increase in disposable income per capita may therefore not translate into increased CHI demand as income is allocated for other consumption . Second, because a previous study has reported that the effect of income on CHI demand is greater in rural than in urban areas . Given that this study is based on urban populations the effect of income may be diminished, particularly in the Western region that has the lowest income and a high rural population. Overall, these results suggest that disparities in income levels, and not only in education levels, may be driving inequitable demand for CHI in China.
Unlike health expenditure and disposable income per capita, SHI participation is not found to have a significant effect on CHI demand in the national sample regression, which is likely because the promoting and substituting effects offset each other [15, 22]. By comparison, regional regression results show that SHI participation has a significantly positive effect on CHI demand in the Central region. This may be because the degree of ageing in the Central region is similar to the Eastern region, while CHI demand in Central China lags behind levels of demand in the Eastern region. Considering that an increase in SHI participation can spread knowledge of health insurance and also improve awareness of the availability of CHI, when coverage of SHI increases, understanding of CHI may also improve resulting in increased demand [15, 40].
Overall, however, some key limitations should be considered when interpreting the results presented. First, the concept of moral hazard, which is a key factor when studying health insurance, is considered only at the theoretical level and is not explicitly included in the empirical analysis. The implication of this is that the effect of health expenditure on CHI demand observed from the data may be underestimated. If people living in provinces with high health expenditures per capita due to higher health risks and disease burden are rejected by insurance companies, the results from the data may misrepresent the latter as lower demand for CHI. Second, although this paper focuses only on urban populations to minimise possible bias, this reduces the sample size which in turn could affect the power and generalisability of the analysis and results. A more representative sample should be used in future studies as additional CHI market data becomes available. Third, attitudes to risk, or risk preferences, have a significant effect on demand for CHI and insurance in general  but these are not considered in this analysis.
Nonetheless, the findings from this paper can help inform health policy and insurance reform, with implications for government and insurance companies. This study suggests that while CHI demand is increasing alongside ageing to supplement existing SHI coverage, the government must consider the education level of different regions to redress regional disparities and prevent further inequity as CHI coverage expands. There is a need to implement elderly focussed health insurance reform, for example through further expansion and consolidation of the existing SHI packages of services [46, 47] so that SHI better caters for an ageing population that currently has to resort to CHI. Alongside SHI reform, the significant effect of disposable income on CHI demand found at a national level calls for the government and insurance companies to work together to lower CHI premiums and make CHI available and affordable for households in regions where coverage is low. However, further research and policy discussions are needed to explore cost-effective and sustainable solutions to motivate CHI companies to extend complementary coverage among the elderly . Either way, these analyses support calls for education interventions for the elderly within a policy framework such as the World Health Organisation’s ‘healthy ageing’. The latter recommends making basic education, including health education and health literacy, available throughout the life-course  – which can also help progress toward life-long education that forms a part of the SDG on education . Such basic education could cover the health insurance system and different mechanisms available to protect against health risks, which may benefit the least educated most. The government could, for example, emulate aspects of Japan’s policy, which has with some success used life-long education as a tool to alleviate both the social and household burdens associated with ageing .