Voluntary Private Health Insurance Demand by Older People in a National Health Service, the Case of Portugal

Abstract

The Portuguese health system is mainly described as a National Health Service (NHS). In parallel with the NHS, there are some Bismarkean features, like those arising from the existence of occupation-based health insurance. On top of these two layers of health insurance coverage, there is a market for private health insurance on a voluntary basis, which older people may not be able to access. The purpose of this work is to estimate the main determinants for older people in Portugal to buy private health insurance since no previous studies have been published. We use data collected by the National Health Survey of 2014 and estimate a multivariate probit. The main results are aligned with previous studies relating to income, education, and age. The role of the health status and behavior explaining the demand for private health insurance, in our results are mixed. People benefiting from parallel occupation-based insurance schemes are less likely to have a private voluntary health insurance policy. The results obtained in this work confirm that there is some inequality in health care access, to the detriment of older people.

1. Introduction

The Portuguese health system is mainly described as a National Health Service (NHS), following the Beveridge tradition, with universal coverage and mandatory participation. In parallel with the NHS, there are some Bismarkean features, such as those arising from the existence of occupation-based health insurance, which are also mandatory or quasi-mandatory. On top of these two layers of health insurance coverage, there is a market for private health insurance on a voluntary basis. This insurance is both supplementary and complementary to the NHS. People may be interested in buying a health insurance policy because it gives them faster access to health care, lets them choose the provider and enjoy a better experience when admitted to a hospital, and have access to services not included in the NHS, such as dental care [1].

Over the last 6 years, health care expenditure on private VHI in Portugal as a percentage of GDP has been less than 0.5% and it has been about 5% of current health expenditure [2]. Despite this trend, voluntary private health insurance in Portugal grew by about 3.5% between 2012 and 2015 [3].

The insurance market is characterized by asymmetric information [4, 5] expressed by moral hazard [6] and by adverse selection [7]. A moral hazard happens after the insurance contract has been signed and it refers to the situation where the insured person uses more health care services than they would need. On the other hand, adverse selection happens before the insurance contract has been signed, when the insurance company cannot assess the risk type of individual purchasing the insurance. This individual may be a low health risk, so they will be mainly healthy and generate low health care spending, or they may be high risk and will generate high health care spending. If the adverse selection does not take place, then low-risk individuals might prefer to buy the insurance and this is called advantageous or propitious selection [8], which is beneficial to insurance companies. This situation is often explained by a person’s risk preferences. Healthy people tend to be risk-averse and so they choose to buy VHI [9, 10].

Older people are often considered as high health risk individuals, who are very likely to have more health problems than younger individuals, have higher medical expenditures, and raise the claims paid by insurance companies [11]. Not only do these people have more health problems, but insurance companies face moral hazards and adverse selection issues that result in the overuse of health care and the payment of excessive claims. For these reasons, older people are not usually the customers desired by insurance companies.

The Portuguese health insurance market is not as highly regulated as it is in other countries, because health insurance is voluntary and the NHS has universal coverage. So, insurance companies can adopt several strategies to prevent or mitigate moral hazard and adverse selection [12] and to reduce paying out excessive claims. One strategy could involve the eligibility requirements excluding people older than 65. Even though this cream-skimming strategy discriminates against older people, there is a small health insurance market for older people in Portugal. In the last available National Health Survey, conducted in 2014, of the 5,701 people older than 65 interviewed, 5.8% of them stated they had a voluntary health insurance policy.

The purpose of this work is to estimate the main determinants for older people in Portugal to take out private health insurance since no previous studies have been published and there may be differences across European countries [13, 14]. We use data collected by the 2014 National Health Survey and we estimate a multivariate probit. We also compute the marginal effects associated with the most important variables. This analysis provides insights for designing health and social policies that will reduce the inequality in health care access that may be generated by differences in health insurance coverage.

1.1 Demand determinants for voluntary private health insurance

The factors explaining the demand for VHI are well discussed in the literature. While Outreville [15] focused on the demand for insurance in general, [16, 17] have looked at the demand for VHI and reviewed the determinants for buying it; [1] outlined these determinants for the EU countries. The explanatory factors include the demographic and economic determinants (also referred to as socioeconomic status), that is, gender, age, education, income, marital status, employment status, and other characteristics [11, 13, 16, 17, 18, 19]. In general, we can say that the likelihood of purchasing VPHI increases with age, income, education, being employed, and living in urban areas. The results are not conclusive for determinants, such as gender, family composition, and being a pensioner.

In Europe, where health systems tend to be mandatory and offer universal coverage, the determinants for holding a private VHI differ between countries [13, 14]. Regarding Portugal, an empirical study performed almost 20 years ago [20], concluded that the VHI buyers were most likely young, self-employed, living in urban areas, and receiving a middle to high income, thus leaving out older people.

The focus on the demand for private VHI by older people has been much less studied because it is known that as we age, the likelihood of having this type of insurance decreases [21]. Four empirical works should be mentioned that are concerned exclusively with older people and use the data collected by SHARE – Survey of Health, Aging, and Retirement in Europe [18, 13, 14, 19]. In general, being female, having had a good education, and receiving a higher income increases the demand for private VHI by older people.

Special attention is often given to the role of the health status and health behavior since there are proxies for the individual healthrisk type [22]. Health status can be measured by self-assessed health and the presence (or the number) of chronic diseases [18, 23], while health behavior can be proxied by body mass index or being overweight [24], and by smoking decisions [25].

The theory predicts that high-risk individuals are more likely to have health insurance, that is, adverse selection exists in health insurance. However, there are no conclusive empirical results regarding the relationship between individual risk variables and having a VHI policy [16]. This means that advantageous selection is a possibility, where low-risk type people choose to buy health insurance [23, 26].

The empirical results regarding indicators of health status and health-related behavior are mixed. Some studies have found that people reporting better health are more likely to have voluntary health insurance, supporting the hypothesis of advantageous selection [11, 12, 14, 18, 23, 27, 28, 29]; other authors found no significant correlation [27, 30]; and still, others found that healthy people tend to have VHI less often, as predicted by adverse selection hypothesis [11, 12, 14, 19].

In most studies, indicators of chronic diseases are found to be insignificant when explaining the demand for VHI [29, 31, 32]. Few works have reported a positive correlation between suffering from chronic diseases and having private health insurance [18, 23].

The relationship between health-related behaviors and private VHI coverage has been studied less and the results are mixed [14]. While some studies find that smokers are less likely to have VHI [27, 33], others find the opposite [18]. The results are similarly mixed for overweight people. It may be found that being overweight is associated with lower odds of taking out VHI or more likely [24].

1.2 Overview of the Portuguese health system

The Portuguese Health System, created in 1979, is defined as a National Health Service; it is mainly financed through taxes and is a universal coverage system. This means it covers all residents and most medical services.

In parallel with the NHS, there are occupation-based health insurance schemes. These include the public insurance schemes that cover civil servants (called ADSE), the armed forces (called ADM), and also private insurance that covers bank employees (called SAMS), among others. There are other, smaller, occupation-based health insurance plans. All the social contributions under these professional insurance policies are income based.

The Portuguese health system comprises conventional private health insurance and it is non-compulsory. Voluntary health insurance (VHI) provides faster access to appointments and treatments, which are also provided by the NHS, or provides access to services not covered by the NHS, such as dental care.

According to NHS rules, people should be registered with an NHS general practitioner, for primary care, and access to specialists in the NHS is controlled by general practitioner gatekeeping. However, people covered by insurance have direct access to specialists (provided by the private sector) according to the rules of the insurance, and private physicians can refer patients to NHS hospitals. So, having an insurance policy on top of NHS coverage has some advantages when it comes to accessing health care services [3]. These reasons for health care access and quality might justify the demand for private VHI by older people in Portugal.

2. Research design

2.1 Data and sample

We use data collected by the National Health Survey, which are representative of the Portuguese Population [34]. It is harmonized and regulated at the European level (EU regulation no 141/2013). It includes 18,204 individuals and our sample considers those aged over 65, that is, 5,701 individuals.

2.2 Variables

2.2.1 Dependent variable

The dependent variable is given by the question if the individual has voluntary health insurance. This is a binary variable that takes value 1 if the respondent has private health insurance and 0 otherwise.

2.2.2 Independent variables

Independent variables are grouped into five categories—demographic, socioeconomic, marital status, health status, related behavior, and insurance status. These variables are described in Table 1.

Group of variables	Independent variables	Description
Demographic	Male	Dummy variable. Takes value 1 is male, 0 otherwise
	Age	Ordinal variable. Age is grouped into 15 classes. The first class takes value 1 and comprises ages 15–19; the last class takes value 15 and includes people older than 85. The variable is taken as continuous.
Socio-economic	Education	Ordinal variable. Education is grouped into five levels of education. First level 0 is those without schooling; fifth and last level is 5 and includes people with a college education. The variable is taken as continuous.
	Income	Ordinal variable. Income is grouped into five classes that represent the quantile of net monthly income per equivalent adult. The first value of income corresponds to the first quantile of income. The variable is taken as continuous.
	Urban	Dummy variable. Takes value 1 if the area is densely inhabited, 0 otherwise.
	Rural	Dummy variable. Takes value 1 if the area is sparsely inhabited, 0 otherwise.
	Moderate urban	Reference category.
Marital status	Single	Dummy variable. Takes value 1 if person is single; 0 otherwise.
	Married	Dummy variable. Takes value 1 if person is married; 0 otherwise.
	Divorced	Dummy variable. Takes value 1 if person is divorced; 0 otherwise.
	Widow	Reference category.
Health status and related behavior	SAH	Ordinal variable. Measures self-assessed health and ranges 1–5, where 1 means “very bad” and 5 “very good” health. The variable is taken as continuous.
	Chronic diseases	Dummy variable. Takes value 1 if person suffers from at least one chronic disease; 0 otherwise.
	Smoking	Dummy variable. Takes value 1 if person smokes; 0 otherwise.
	BMI	Body Mass Index.
Insurance status	ADSE	Dummy variable. Takes value 1 if person is covered through ADSE insurance; 0 otherwise.
	SAMS	Dummy variable. Takes value 1 if person is covered through SAMS insurance; 0 otherwise.
	Other insurance	Dummy variable. Takes value 1 if person has another occupational-based health insurance on top of NHS; 0 otherwise.
	NHS	Reference category. This is the case where respondents do not hold any occupational-based health insurance.

Table 1.

Description of independent variables.

2.3 Quantitative analysis

The model to be estimated in this analysis is written as follows:

$$VH{I}_i*=\mathit{constant}+{\beta }_i{X}_i+{\epsilon }_i\mathit{and}=\{\begin{array}{c}1\mathit{if}VH{I}_i^{*}\ge 0\\ 0\mathit{if}VH{I}_i^{*}<0\text{.}\end{array}$$E1

where VHI* is the latent dependent variable, VHI is the observable dependent variable, β_i’s are the coefficients to be estimated, X_i’s are the independent variables, and ε_i is the residual.

The dependent variable expresses whether the respondent has voluntary health insurance. The binary nature of this variable implies that the econometric method of estimation is a probit. The estimated coefficients provide the direction of the relation between independent and dependent variables. The computation of the marginal effects allows the comparison of the intensity of the estimated coefficients. The marginal effects provide information on how the probability of having VHI changes when there is a unit change in the independent variable.

The results are obtained using Stata 15 econometric software.

3. Results

3.1 Descriptive statistics

The sample comprises 5,701 respondents older than 65, of whom 85.4% are retired. Only 332 respondents say that they have private voluntary health insurance, the large majority (about 94%) do not. Most of them are not entitled to any other insurance coverage apart from the NHS, and only about 10% are covered by ADSE, the occupation-based insurance plan for public workers (Table 2).

Group of variables	Independent variables	Descriptive statistics
		Number	%
Demographic	Male
	Male	2,215	38.85
	Female	3,486	61.15
	Age
	65–69	1,533	26.89
	70–74	1,319	23.14
	75–79	1,259	22.08
	80–84	979	17.17
	+85	611	10.72
Socio-economic status	Education (years)
	0	1,942	34.06
	6	2,938	51.53
	9	353	6.19
	12	189	3.32
	15	14	0.25
	17	265	4.65
	Income (quantile)
	Q1	1,368	24.00
	Q2	1.534	26.91
	Q3	1.174	20.59
	Q4	867	15.21
	Q5	758	13.30
	Level of urbanization
	Urban	1,563	27.42
	Median	1,689	29.63
	Rural	2,449	42.96
	Retired
	Yes	4,869	85.4
	Other status	832	14.6
Marital status	Marital status
	Single	363	6.37
	Married	2,913	51.10
	Widow(er)	2,132	37.40
	Divorce	293	5.14
Health status and related behavior	SAH
	1. Very bad	542	9.51
	2. Bad	1,443	25.33
	3. Fair	2,817	49.45
	4. Good	787	13.81
	5. Very good	108	1.90
	Chronic diseases
	None	778	13.65
	At least one	4,992	86.35
	Smoker
	Yes	266	4.7
	No	5,433	95.3
	BMI
	Average	27.0
Insurance status	Insurance
	None (only NHS)	4,836	84.83
	ADSE	566	9.93
	SAMS	83	1.46
	Other insurance	216	3.78

Table 2.

Descriptive statistics.

The remaining descriptive statistics for the independent variables are also shown in Table 2. The majority of the people in the sample are women (circa 61%) and most of them are aged between 65 and 75. Portuguese older people have very low levels of education, with more than 80% having less than 6 years of schooling. They have very low levels of income, about 50% receive an income valued in the first and second quintile of per capita household income. A large share of the people in the sample live in rural areas (circa 43%), and most of them are married or used to be married.

Finally, regarding their health status, the majority of older Portuguese assess their health status below the median level and about 86% of them report suffering from at least one chronic disease.

Other descriptive statistics regarding the distribution of respondents with private health insurance across income, education, and self-assessed health are shown in Table 3. Considering those individuals who said they had private health insurance (332 people), their distribution across income shows that a larger share of respondents has a high-income level. The distribution of people with health insurance across levels of education has two peaks, one at 6 years of schooling and the other at 17 years of schooling. The distribution of the health status of people having an insurance policy shows that most people with health insurance report a health status better than fair.

Income
Quantile	Q1	Q2	Q3	Q4	Q5
number	21	31	50	66	164
%	6.3	9.3	15.1	19.9	49.4
Education
years	0	6	9	12	15	17
number	30	120	63	42	3	74
%	9.1	36.1	19.0	12.7	0.0	22.3
SAH
levels	1	2	3	4	5
number	10	43	172	90	17
%	3.0	13.0	51.8	27.1	5.1

Table 3.

Distribution of respondents with voluntary health insurance.

To finish the description of the variables, we now report the correlation between health risk indicators. The pairwise correlation between SAH and suffering from chronic diseases is equal to −0.364 and between SAH and smoking it is equal to 0.107, both for a statistical significance of less than 0.001. The tetrachoric correlation between smoking and suffering from a chronic disease is equal to −0.257 for an identical statistically significant level. So, there is no strong correlation that could prevent the joint utilization of these variables in a regression analysis.

3.2 Model results

The results obtained with the estimation of the probit for having voluntary health insurance are presented in Table 4. The statistically significant coefficients at 5% are marked with *.

		Coef.	Std. Err.	P > z
Demographic	Male	0,061	0,068	0,370
	Age group
	70–74	−0.132	0.077	0.085
	75–79	−0.226*	0.089	0.011
	80–84	−0.497*	0.120	0.000
	+85	−0.469*	0.148	0.002
Socio-Economic	Education	0.073*	0.009	0.000
	Income
	Q2	0.051	0.121	0.672
	Q3	0.353*	0.116	0.002
	Q4	0.542*	0.117	0.000
	Q5	0.988*	0.122	0.000
	Urban	0.112	0.077	0.146
	Rural	0.042	0.077	0.588
	Retired	−0.073	0.087	0.399
Marital status	Single	−0.367*	0.181	0.043
	Married	−0.039	0.118	0.738
	Widow(er)	−0.193	0.131	0.141
Insurance	ADSE	−0.647*	0.108	0.000
	SAMS	−0.200	0.179	0.265
	Other insurance	−0.410*	0.149	0.006
Health status and health behavior	SAH
	2	0.128	0.160	0.424
	3	0.212	0.150	0.159
	4	0.330*	0.164	0.044
	5	0.392	0.226	0.083
	Chronic diseases	0.180*	0.091	0.047
	BMI	−0.006	0.008	0.405
	Smoking	−0.277*	0.135	0.041
	_cons	−2.285	0.318	0.000
	Number of obs5,509
	LR chi2(26)518.92
	Prob > chi20.000
	Pseudo R20.207
	Log likelihood−994.916

Table 4.

Probit results.

Note: * Significant at less than 5%.

The estimated coefficients show that as someone gets old or is single, the probability of having private health insurance decreases, while for higher income or education levels that probability increases.

Regarding the insurance status of people, being a beneficiary of ADSE or another occupation-based insurance decreases the odds of having private VHI. Lastly, the results for health status and health-related behavior are mixed. On the one hand, higher levels of SAH may be related to having VHI, but on the other hand, suffering from a chronic disease is also positively related to having VHI; additionally, the observable behavior of smoking results in a lower likelihood of benefiting from VHI coverage.

The marginal effects associated with the most important and statistically significant coefficients are presented in Table 5.

	dy/dx	Std.Err.	P > z
Age group
70–74	−0.015	0.008	0.083
75–79	−0.024	0.009	0.009
80–84	−0.044	0.009	0.000
+85	−0.042	0.011	0.000
Income
Q2	0.003	0.006	0.670
Q3	0.025	0.008	0.002
Q4	0.045	0.010	0.000
Q5	0.117	0.016	0.000
Education	0.007	0.001	0.000
ADSE	−0.063	0.011	0.000
SAMS	−0.019	0.017	0.265
SAH
2	0.010	0.012	0.402
3	0.018	0.011	0.112
4	0.030	0.014	0.026
5	0.038	0.024	0.109
Chronic diseases	0.017	0.009	0.047
Smoking	−0.027	0.013	0.041

Table 5.

Marginal effects.

These effects represent the change in the probability of having a VHI policy after the discrete change from the base level of the independent variable. In this way, the change to the oldest age groups implies a decrease of about 4% in the probability of having VHI, while the change from the lowest income quintile to the highest expresses an increase of 11% in the probability of being covered by VHI. Being a member of ADSE results in a 6% less chance of having VHI, and finally, the change from poor health status to a better one increases the likelihood of benefiting from a VHI; for instance, it increases almost 4% for people reporting very good health.

4. Discussion

In Europe, health systems tend to be mandatory and offer universal coverage. Despite this major trend, there is a market for voluntary private health insurance. Portugal is characterized by having a National Health Service of universal coverage with distinctive features of mandatory occupation-based insurance. Because the health insurance market suffers from asymmetric information, insurance companies adopt cream-skimming strategies to minimize adverse selection and moral hazards. One such strategy is to set the eligibility requirement for buying an insurance health policy is having to be under 65. In this way, most older people are unable to buy a health insurance policy. However, there is a small market and about 5% of Portuguese older people report having voluntary private health insurance of some kind.

Our aim in this work was to find the main determinants of the demand for private health insurance by older people in Portugal and contribute to the literature on voluntary health insurance schemes in different European countries, as there is no study for Portugal. We used data collected by the 2014 National Health Survey and estimated a probit for people over 65 having private health insurance.

The main results are aligned with previous studies concerning the importance of income and education [11, 16, 17, 18, 19]. The higher the income and the better educated the individuals, the greater the probability of having private health insurance.

Concerning age and health insurance, we found that as they get older, they are less likely to have private voluntary health insurance [23, 35, 36]. The results show that only a minority of individuals, about 332 people, have a voluntary private health insurance policy. These people tend to have a high income and a high level of education, which is uncommon among people older than 65. Most older people in Portugal receive small pensions and have a low level of formal education, which deters them from taking out health insurance. The lack of schooling is the origin of illiteracy, both financial and health-related, which precludes people from making wiser choices on how to make better use of their savings and reduce future out-of-pocket expenditures. One major concern relates to dental care and the (high) associated cost. This aspect of health care is usually neglected by older people because it is not covered by the NHS and because they do not have a complementary private health insurance policy to cover it [37, 38].

Regarding the role of health status and behavior in explaining the demand for private health insurance, our results are mixed. On the one hand, there is some evidence of advantageous selection because better health status and no smoking are associated with taking out health insurance. On the other, reported suffering from chronic diseases is also associated with health insurance, this time reflecting adverse selection.

It could be that insurance companies are discriminating based on observable traits, such as smoking. Or, related to high health risks, such as suffering from a chronic disease, it may be the case that people fail to report them. Perhaps insurance companies do not “cream skim” based on these conditions, either because they lack sufficient reliable information, or because they may calculate the probability that a person suffers from a certain disease at a certain age, or even because the insurance company can control claims associated with those health conditions by cost-sharing.

Another explanation of the mixed results found when relating health risk to health insurance is based on the demand side of the market. Maybe there is heterogeneity in the risk preferences of older people. In some countries, healthier individuals might be more risk-averse [14, 16] and so they are more prone to take out voluntary health insurance. Maybe this is the case with Portugal as it was with the UK [39]. On the other hand, people suffering from chronic diseases have a default health status that they consider to be a reference status in the sense proposed by the prospect theory [40]. These people may thus tend to be risk-averse with reference to their health status, and consequently, they are also more prone to have a private health insurance policy.

Finally, regarding the existence of parallel occupation-based insurance plans, our results indicate that people benefiting from ADSE, the largest occupation-based insurance for public servants, or from any other form of private or public health insurance (public health insurance is for the armed forces; private insurance includes bank workers, Portugal – Telecom workers, and postal CTT workers) are less likely to have VHI. This is expected to happen because occupation-based insurances provide a second layer of health coverage on top of the universal provided by the NHS. People benefiting from occupation-based insurance policies pay taxes to finance the NHS and pay a percentage of their income to finance occupation-based insurance. Therefore, this double financing by people deters them from looking for additional private health insurance coverage. In fact, these people do not need private health insurance because their health care needs are covered either by the NHS or by their occupation-based insurance.

The organization of the Portuguese health system creates inequity in access to health care. In the first place, people with double coverage have easier access to health care, and then people with high incomes can afford to buy private health insurance coverage. On top of this, inequality is aggravated by a tax system that gives some benefits to wealthier people for buying private health insurance or for spending on private health care [1]. The findings reported in this work confirm the existence of this sort of inequality, especially among older people.

One limitation of this work is that it is not possible to analyze the type of coverage provided to older people by voluntary private health insurance. This sort of information would show us what health care services older people want, and what could be lacking in the supply of NHS.

The results found in our analysis provide some insights into what makes older people decide to take out voluntary private health insurance. We have concluded that income is a determinant factor for taking out private health insurance, but it is also a factor for generating inequality in health care access. Older people can find it hard to access dental care or simple eye care because it is not covered by the NHS, or because the NHS waiting lists are too long. But the difficulty of complementing NHS coverage with private health insurance increases health care access inequity. Health and social policies may aim to narrow the gap either by providing health care in the NHS or by subsidizing the purchase of private health insurance for low-income older people. The first approach to this has already been put into place. The instrument called “dentist-check” for older people, created by the Ministry of Health attempts to mitigate the inequality in access to dental care, but it needs to be assessed.

Highlights

• small share of older people buy voluntary private health insurance

• higher income and higher education increase the likelihood of holding voluntary private health insurance

• benefiting from occupation-based insurance schemes reduce that likelihood

• voluntary private health insurance reflects inequity in healthcare access

Conflict of interest

The author declares no conflict of interest.

Author funding

The author received no financial support for the research, authorship, and/or publication of this article.

Declaration

The author declares this work does not require any human/animal subjects to acquire ethical approval.

JEL classification

I13; D81; C3