Market structure and decision-making in health insurance

(1)

Market Structure and Decision-Making in Health

Insurance

Field: Behavioural and Health Economics

Supervisor: Dhr. Dr. A.P. Kiss

Maud Castelijn

10565159

(2)

Abstract

In 2006 a new health insurance system has been introduced in the Netherlands. The market turned from a fully regulated form into a market in which the insurance companies must compete with each other to attract the insured. The demand side is important to reduce the health care costs and increase its quality. Do consumers search for and switch to alternative health insurances? Does the awareness of a competitive market structure influence the choice of insured to search or to switch? My research showed that insured are not searching and switching actively. People are switching and searching with a higher probability when they think the market is more competitive. Most of the people who do search or switch are influenced by the price of the health policy and are looking for the lowest expense. The consumers that do not search or switch to alternative insurances do so because of risk aversion and searching- and switching costs.

(3)

Table of Contents Introduction……….. 4 Literature Review……….5 The Netherlands………5 Some numbers……….. 5 Competition……….. 7 Consumer Behaviour……… 8 Research……….. 11 Research Method……… 11 Survey……… 11 Data……… 11

Searching and Switching……… 15

Competitive market and searching/switching behaviour………16

Competition and how much people think they can save when switching.. 19

Test the explanations for the consumers their behaviour………... 21

Logistic Regression……… 22

The effect of searching and switching on the premium amount………… 25

Conclusion……….. 28

References………... 30

(4)

Introduction

Several countries have issues with their health insurance market. Should it be fully regulated by the state and should health insurances be treated as a public good or should this market be privatized? The health insurance and care systems vary across countries. In the case of privatization the health insurance market would be based on competition and a market force will be present with only little intervention of the government. Insurers are then expected to be negotiators between consumers and providers of health care (van de Ven et al. 1994).

In the Netherlands, a system of competition between the health insurers has been introduced in 2006. One of the goals this market form induces is controlling and reducing the health care costs. However, this goal has thus far not been reached since the price for health insurance has only been increasing throughout the years. An important part of the well functioning of the competitive market is the behaviour of insured. They present the demand side of the health insurance market. Because of the negative results concerning the price of insurances, it is interesting to investigate the role of consumers in this. They can choose to search and to switch to alternative insurances each year. Although, if they choose not to do this, the insurance companies do not have an incentive to reduce their insurance premiums. Therefore the research question I have determined for this thesis is the following: do

consumers search for alternative health insurances, do they switch and does consumers’ belief in a competitive market structure influence their behaviour in the market?

I have designed and distributed a questionnaire to inhabitants of the Netherlands to find answers to these questions. This questionnaire includes questions about the thoughts, knowledge and behaviour of insured people concerning their health insurance and the competitiveness of the market.

(5)

Literature Review The Netherlands

The market for health insurances has been privatized in the Netherlands. The main reasons for this introduction were that medical technology innovates and social and economic

characteristics of the population change. Therefore a new health care system that is flexible, economically feasible and one that comes up with new ways of organizing and distributing health care in an efficient way was necessary.

Before 2006 the inhabitants of the Netherlands could choose between different kinds of basic insurances, such as the health insurance fund, the private insurance, the public insurance and the standard package policy. People with a low income had right to access the health insurance fund. The advantage of the health insurance fund was that an extended package of insurance was offered for a relatively low premium. Health care was therefore easier payable for the less rich people.

In the new health care scheme, which was introduced in 2006 and is still active nowadays, the pool of different basic insurances has been replaced by one general basic health insurance. Health insurance companies who aim to make a profit offer the basic health insurance. Next to this, it is compulsory to have such insurance for all the inhabitants of the Netherlands. They buy the basic insurance each year for a certain premium. The basic health coverage contract can be concluded individually or collectively (DNB, 2005). The employer often suggests collective insurance contracts.

In the Netherlands the total expenditures to health care of the insured are combined in the amount of premium they pay each year for their health insurance and the amount of own risk. Every year the government determines a certain amount of ‘own risk’ (deductible). Everyone has to pay the same amount of own risk before they get any reimbursement of health expenses.

Some numbers

Vektis, a Dutch company that collects and analyses data on the cost and quality of health care, has been analyzing the switching behaviour of the insured since the introduction of the new health scheme in 2006. From 2014 to 2015 approximately 6.8% of the total insured

population switched to another health insurance company. On average a minor growth in the percentage of switchers can be seen throughout the years (Vektis, 2015). Since the

(6)

health insurer until 2016 (Vektis, 2015).

In total there are 25 insurance companies that offer policies to the consumers in 2015. The number of health insurance firms slightly decreases, but the amount of health insurance policies increases rapidly (Dutch health authority, 2012). In 2010 there were 52 policies from which the insured could choose. In 2015 this amount was more than 70, so the choice between the different insurances becomes harder (Vektis, 2015).

In general, the health insurance premium increased from 2006 to 2015. This can be seen in figure 1.1.

Figure 1.1 The average health insurance premium in the Netherlands per year. Source: Vektis, april 2015

The average premium amounted in 2006 to sums between €950 and €1.060 (CBS, 2014). In 2015 this was already €1.218 (Vektis, 2015). This is an increase of at least 14.9%. The level of premium decreased compared to the levels in 2011 till 2013. This is because the costs of health care were significantly higher during that period than nowadays.

As said before, the government each year determines an amount of own risk. In 2010 the own risk implied an amount of €160, but in 2015 this amount is €375 (Vektis, 2015). The large increase in own risk makes the health insurance more costly for the insured.

The budget for the basic health insurance is financed for approximately one half by the income dependent premium, which is often paid by the employer. The insured pay one third of the budget by means of the premium and the own risk. The government finances the remaining portion. In the years before 2014, the income of the health insurance companies was higher than the damage or the expenditures on health. Because of this, health insurance firms were able to build up capital since 2006. In 2013 this capital had already grown to 8 billion Euros (CBS, 2014). The entire capital of all the insurers was in 2013 two and a half

Gemiddelde nominale jaarpremie per premiebetalende verzekerde (bron: Vektis, 2015) Figuur 6 € 1.300 € 1.250 € 1.200 € 1.150 € 1.100 € 1.050 € 1.000 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

3

Keuzes rondom premie

Na een daling over de afgelopen twee jaar, stijgt de nominale jaarpremie per 2015 naar gemiddeld € 1.218,- per jaar. Dit is een stijging met 5,3% ten opzichte van 2014. Voor het eerst sinds de invoering van de Zorgverzekeringswet daalt het percentage verzekerden dat collectief verzekerd is. Voor wat betreft het type zorgpolis dat verzekerden in 2015 hebben gekozen, valt de opmars van de polis met sterk selectief ingekocht zorgaanbod op.

In 2013 en 2014 zagen we een daling van de nominale jaarpremie ten opzichte van het voorgaande jaar. In deze periode waren er onder andere meevallers in de zorguitgaven, die een neerwaarts effect hadden op de nominale premie. Voor 2015 komt de gemiddelde nominale jaarpremie uit op € 1.218,-, een stijging met 5,3% ten opzichte van 2014. Deze stijging wordt onder meer veroorzaakt door de overgang van een deel van de AWBZ-zorg naar de Zorgverzekeringswet en door de nominale stijging van de zorgkosten. Hierdoor neemt het kostenniveau in de Zorgverzekeringswet toe, wat een opwaarts effect heeft op de nominale premie. Figuur 6 toont de ontwikkeling van de gemiddelde nominale jaarpremie per premiebetalende verzekerde.

In tabel 4 staat een overzicht van de gemiddelde nominale jaarpremie vanaf 2006. Daarbij is onderscheid gemaakt tussen verzekerden die collectief verzekerd zijn en verzekerden die individueel verzekerd zijn.

In 2015 betalen collectief verzekerden gemiddeld € 44,- per jaar minder premie dan individueel verzekerden.

Tabel 4 Gemiddelde jaarpremie voor de basisverzekering (bron: Vektis, 2015)

Gemiddelde premie (exclusief betalingskorting) 2006-2015

2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

Gemiddelde basispremie

€ 1.060 € 1.147 € 1.094 € 1.110 € 1.145 € 1.262 € 1.287 € 1.280 € 1.157 € 1.218

Gemiddelde betaalde premie

Individueel € 1.053 € 1.135 € 1.081 € 1.088 € 1.127 € 1.226 € 1.241 € 1.230 € 1.111 € 1.164 Collectief _{€ 987 € 1.056 € 1.010 € 1.033 € 1.055 € 1.168 € 1.195 € 1.188 € 1.060 € 1.120} Totaal € 1.027 € 1.091 € 1.040 € 1.056 € 1.082 € 1.188 € 1.210 € 1.201 € 1.076 € 1.133 12 Av er ag e am ou nt o f pr em iu m

(7)

times as high as the required capital of The Dutch Bank. This amount of capital has been increasing mainly because of the profit-making firms. In 2013 the total profit was

approximately 1.4 billion Euros (CBS, 2014).

Competition

The new health insurance system in the Netherlands can be described as a competitive market, which is regulated by the government. In the new system three major groups play a central role. Firstly, the government regulates and places restrictions on the behavior of the health insurance companies to stimulate price competition (Enthoven, 1993). They are the entity that needs to secure the quality of the health care in this country. Secondly, insurance companies need to negotiate with the suppliers of health care and offer bundles with the best price-quality ratio as possible. These negotiations involve appointments on factors such as prices, benefits covered, enrolment procedures, etc. (Enthoven, 1993). Thirdly, the market functions through demand and supply. The demand for health care is dependent on the behaviour and choices of the insured. According to Culyer and Newhouse (2000) these choices are based on barriers to care seeking such as price, agency relationship, altruism and coverage.

To make sure that the health insurance companies have an incentive to offer their insurances for the lowest possible price, consumers should respond quite actively on a price increase or decrease by switching to alternative insurances. In this privatized market, the choices of the insured are therefore an important part of how the market functions. To achieve this, standardization of the good offered in the market should prevent product differentiation, simplify price comparisons, and counter market segmentation. Prevention of product

differentiation facilitates easier comparisons of the insured concerning the price-quality ratio of the health insurances. Because of standardization the insured are reassured that the lower-priced basic insurances did not mean savings by inventing secret gaps in health coverage (Enthoven, 1993).

According to a research of the CPB (2000), when competition is intensified, the health care costs, insurance premiums and the profits made by the insurance companies are reduced. This is an advantage for the consumers. The competition and the fact that insured people can switch to alternative health insurances each year would force insurers to attempt for lower prices of the insurance and high quality of care (DNB, 2005). However, as the previous years have shown, this does not happen. Therefore it is important to analyse the functioning of the market by investigating the behaviour of the demand side, the insured people.

(8)

Consumer behaviour

Until now there has not been much scientific research in Europe on the field of the consumer choices for health insurances and the effect of a privatized market mechanism on their choice. However there have been researchers who observed the switching behaviour of consumers whilst making their choice for their health insurance.

It is stated that the typical economic model of consumer choice and behaviour under uncertainty assumes that consumers maximize their expected utility. Frank and Lamiraud (2009) found that as the number of insurance packages offered to individuals grows, their willingness to switch insurance bundles given a certain price decreases. This allows large price differentiations for relatively similar insurances to exist. Particularly when a choice of a consumer includes the two factors ‘money’ and ‘health’, they often return to the risk aversion. They even do this if better options are available. There are two reasons that can explain this according to Frank and Lamiraud (2009). Firstly, people are overwhelmed by too much choice and therefore they can inhibit on processing information and on searching for alternative insurances. Secondly, people are concerned about making an incorrect decision and having suffering regrets in the future. In a situation of uncertainty and when the decision is complex it is hypothesized that there is a movement towards overstating the disadvantages of switching to alternative goods. The potential advantages are understated (Samuelson and Zeckhauser, 1988).

Next to this, the importance of health care to life and happiness, combined with the limited ability of consumers to make knowledgeable judgments about quality of care may be an explanation for inefficient consumer choices and can therefore have an effect on the health insurance market (Weisbrod, 1991). This agrees with the view of Diamond (1992), who said that there are all kinds of imperfections, on informational and market level that affect the way in which the privatized health insurance market functions.

Another research that has been performed indicates that competition in the health insurance market has not been effective so far, and reveals some inactivity among consumers who seem reluctant to switch to less expensive insurances (Dormont, Geoffard & Lamiraud, 2009).

Handel and Kolstad (2013) state that it is very important to analyse how consumers value several insurance factors, what their information about these factors is, and how these aspects explain their choices. It is for example important to know whether they value the quality of health care above their expenditure for it or the other way around. This is essential to market design and the regulation of the government. When these preferences and

(9)

information are well known, it can be decided how the different health policies should be presented and priced. Research shows that information asymmetries and observed costs because of the difficulty of the policies are important issues for consumer health insurance choices (Handel & Kolstad, 2013).

In contrast, Buchmueller and Feldstein (1996) find that consumers will respond to an increase in the premium of their initial health insurance by switching to an alternative cheaper

insurance. A rise of less than ten dollars on premium expenses per month was the reason that insured people were approximately five times as likely to switch to a policy with a premium that remained constant. An explanation for their findings is that the design of the program they used during their research makes it easy to compare different insurances based on their prices (Buchmueller & Feldstein, 1996).

Search costs are low in the health insurance market and the switching procedure is not complex. By use of the Internet, alternative basic insurances can be compared

straightforwardly. In a market with community-rated premiums for each policy, homogenous benefits, free admission for everyone and low switching costs, individuals would be expected to switch to the basic insurance with the lowest premium. Next to this a regulatory structure, which is present and carried out by the government, reduces many chances on service quality variation between health insurances (Frank & Lamiraud, 2009). While research shows that consumers who switch to alternative health insurances on average pay 15 to 16 percentages less on premiums for their policy per month, the switching rates are not substantial (Frank & Lamiraud, 2009). As results of the previous years show, the controlling of health costs by means of competitive health insurance companies did not work as well as intended.

A survey on the consumer switching behaviour in the Dutch retail electricity market concludes that low switching rates are caused by a lack of interest (de Vries & Heijnen, 2006). Next to this they suggest that the general attitude towards competition has a strong effect on the switching behaviour of consumers. Switchers in the electricity market were also more likely to switch from health insurance (de Vries & Heijnen, 2006). They find that the low switching rate in the retail electricity market is not mainly the result of rational decisions.

Pomp et al. (2005) find that Dutch consumers are less likely to switch if the costs of switching are perceived to be high and the probability of switching is even lower when they do not know how to estimate the expected costs. However, Sturluson (2002) found that when consumers were confronted with information about switching, it stimulated them to actively search for more information afterwards. Next to this it was analysed that before rational choice, consumer behaviour is primarily determined by confusion and information overload

(10)

(Wilson & Waddams Price, 2005). Brennan (2006) believes that inertia is the factor in consumer behaviour that establishes the preference not to choose at all and stimulates the default effect. Brennan (2006) and Poiesz (2004) suggest that switching behaviour is not rational.

The market form in the current health insurance market in the Netherlands requires active participation, knowledge and therefore heavy searching and switching behaviour of the insured. The willingness of the consumers to switch has an influence on the achievement of the goals, such as lower health costs, that have been created whilst introducing the privatized market in 2006 (Buchmueller & Feldstein, 1996). Therefore their behaviour and choices concerning the care insurance is an important subject of attention. What choices do consumers make with respect to their health insurance? Which components influence these choices; does a competitive market structure have an effect? Do consumers change their health insurance at all each year?

(11)

Research

Research method

I collected the data for this research by means of designing and distributing a questionnaire. For every independent variable, approximately thirty data sets needed to be collected.

Because of the limited abilities to collect data, I will not control the independent variables for demographic factors. This is based on the fact that previous research has shown that switching behaviour did not seem to be influenced by demographic factors (de Vries & Heijnen, 2006). However, this information will be collected within the sample. During the writing of this thesis I will be aware of the small sample bias, small selectivity and endogeneity.

Survey

The questionnaire I distributed included sixteen questions regarding the choices, thoughts and knowledge about the health insurances of inhabitants of the Netherlands. The first questions investigated the demographics of the insured. I requested information about factors such as gender, age and family composition. After that, consumers were asked if they had searched or switched once for alternative health insurances in the past three years. I represented three standard reasons, which they could select as an explanation to why they did or did not search or switch. The three given arguments involved factors of risk aversion, searching and

switching costs or price awareness. If people had been searching for different policies I asked them the question by which means they searched. Next to this, people were questioned to provide numerical information about their expenditures and savings considering the health insurance. I asked the following questions:

-What amount of premium do you pay on yearly basis for your health insurance nowadays? -What is the maximum amount of money you think you can save if you switch policies? -What is the minimum amount of money you want to save to make you switch insurances? The last two questions examined the thoughts of the consumers about the market and about their own health expenses. The total questionnaire can be found in the appendix.

Data

In total 161 people responded to the questionnaire. The number of male respondents was 72 (44.72%) and there were 89 (55.28%) female persons who answered the survey. I divided demographics based on age in three groups. Next to that I asked about family composition. The results can be found in figure 1.2.

(12)

Figure 1.2 The age and partner distribution of the sample

In total 67 consumers answered that they had been searching for alternative insurances in the past three years. 94 people did not search. In the following table, the numbers and percentages of people per given explanation for their searching behaviour are presented. The reason before the slash represents the explanation why people did search and the argument behind the slash describes the reason why people did not search. The three major factors are in the end

characterized by one word, namely Risk Aversion, Price and Searching Costs.

Why searching yes or no? Yes No

I was not satisfied with my health insurance / I know what I can expect from my health insurance and I like certainty.

(Risk Aversion)

6 9% (6/67*100%)

50 53.2%

I was searching for the lowest possible price/ The lowest possible price is already being offered because of the competitive market

(Price)

42 62.7%

4 4.2% I have enough time and knowledge to search for alternative policies/ I do not

have enough time and knowledge to search for alternative policies (the process of searching is too difficult)

(Searching Costs) 8 11.9% 25 26.6% Other reason 11 16.4% 15 16%

Table 1.1, number of people who searched per reason why or why not

During the research, I also asked the insured by which means they searched for other health insurances. Most of the people make use of the Internet. They check the alternatives and

102 21

38

Age distribution sample

18-35 years old 35-50 years old 50-65 years old 84 77 53 108 0 20 40 60 80 100 120

Do you have a

partner? Do you have children?

Yes No

(13)

price-quality ratios on websites that compare the same policies at different health insurance companies. Other consumers search for more information on the website of the health insurers themselves and make a comparison for their own.

In the past three years, 50 people switched to alternative health insurances. A total of 111 people did not change to other policies. These results overall represent the Netherlands well, but the percentage of people switching in this research is a little bit higher than the examined percentage in the Netherlands. Table 1.2 represents the frequency per explanation that consumers have answered during the survey regarding their switching behaviour. Again the reason before the slash represents the explanation why people did switch and the argument behind the slash describes the reason why people did not change to another policy. The three major factors are in the end characterized by Risk Aversion, Price and Switching Costs.

Why switching yes or no? Yes No

I was not satisfied with my health insurance / I know what I can expect from my health insurance and I like certainty.

(Risk Aversion) 11 22% (11/50*100%) 71 64%

I was searching for the lowest possible price/ The lowest possible price is already being offered because of the competitive market

(Price)

28 56%

2 1.8%

I have enough time and knowledge to search for alternative policies/ I do not have enough time and knowledge to search for alternative policies (the process of searching is too difficult)

(Switching Costs) 2 4% 29 26.1% Other reason 9 18% 9 8.1%

Table 1.2, number of people who switched per reason why or why not

In the questionnaire I have also asked about numerical values regarding the expenditures and savings of the health insurance. I was able to determine the averages, standard deviations, minimum amount and the maximum amount of this data set. These values are represented in a table in the appendix. The distributions and the average of the responded values are given in the figures below.

(14)

Figure 1.3 Average amount of premium people pay on a yearly basis: €1,399.129

As can be seen from the largest premium amount, outliers do exist in this sample. Premiums larger than 3,000 euros are impossible. The outliers can probably be assigned to typos or to given amounts per family instead of per person.

From 161 observations, 39 people answered that they thought they could save zero euros in the case of switching to another health insurance. From the total sample, almost 75% of the people think they can save less or equal to 100 euros per year when they switch.

36.02% Of the sample answers that they find that the health insurance market is very competitive. Another 14.29% of the respondents think that the market is completely

competitive.

Finally, 56 people from the 161 (36.65%) reacted that they think they pay the average price compared to the rest of the insured.

0 5 .0 e-0 4 .0 0 1 .0 0 15 D en sit y 0 2000 4000 6000 premium 0 .0 0 2 .0 0 4 .0 0 6 .0 0 8 D en sit y 0 200 400 600 wantsave

Figure 1.4 Average amount of money people think they can save when switching: €135.86.

0 .0 0 5 .0 1 .0 1 5 D en sit y 0 100 200 300 400 500 thinksave

Figure 1.5 Average amount of money people want to save when switching: €88.08.

(15)

Searching and Switching

An important aspect of the survey is the answer to the question if people search for better alternatives of their health insurance and if they switch their insurances actively. As can be seen from previous research, their behaviour and demand namely has an important effect on the functioning of the competitive market.

I calculated a 95% confidence interval for the percentage of people who search. The variable ‘n’ stands for the sample size and the variable ‘p’ stands for the percentage of people who search. The confidence interval can be estimated because the conditions; 𝑛 ∗ 𝑝 ≥ 5 and 𝑛 ∗ (1 − 𝑝) ≥ 5 are met. The observed average percentage of searchers is 41.61%. The sample size is 161. The 95% confidence interval1 is therefore 41.61% ± 7.64%. This means that the percentage of people who search in the population in the past three years is with 95% certainty between 33.97% and 49.25%.

Secondly, the 95% confidence interval of the switching behaviour has been investigated. The calculations have been performed on the exact same way. The average percentage of switchers in the sample of this research in the past three years is 31.06%. The 95% confidence interval implies that the percentage of people who switch to other health policies is with 95% certainty between 23.87% and 38.23% in the population. This percentage of switchers in the population is higher than what has so far been found in the Netherlands.

To conclude, it can be stated with a significance level of 5% that the percentage of people who search or switch is in any case lower than 50% of the population. Mainly the switching mobility is not stimulating the competitiveness of the market. This means that the demand side of the health insurance market is not as active as it should be and mainly has a passive attitude. To create more insight in the motivation behind this behaviour, more research has been done.

1_{This interval is calculated by means of this formula: 𝑝 ± 1.96 ∗} !∗ !""!!

!!! !.!

(16)

Competitive market and searching/switching behaviour

In the survey I have asked people what their thoughts were about the competitiveness of the health insurance market in the Netherlands. They were able to answer this question based on a scale from 1 to 5. One stood for a monopoly and five for a completely competitive market. In total there were 6 people who responded that they thought the health insurance market was a monopolistic market. These six consumers were in the minority. Most of the people, namely 58 in total, thought that the market was very competitive, but not completely. They answered four on the scale. In the following figures, the amount of searchers, non-searchers, switchers and non-switchers can be seen per thought about the level of competitiveness of the market.

Figure 1.6: competitive market thoughts and searching/switching behaviour

All the graphs are equal in shape, namely a concave one. Per graph it can be noticed that most of the people answered four, or in words, that the market is very competitive. More people who answered that they did not search or switch compared to the ones that did search or switch answer that they think the market is very competitive. However, there are also more people who did not search or switch. Therefore it is better to analyse the relative numbers. The graphs of percentages of people who switched and searched per scale level of

(17)

Figure 1.7: competitive market thoughts and relative searching/switching behaviour

In the tables 1.3 and 1.4 the results of the calculations and the survey can be found. The first number represents the absolute number of respondents per scale of competitiveness and the second number characterises the relative percentage of respondents per total amount of switchers, non-switchers, searchers or non-searchers.

Switch No (in total 111 people) Yes (in total 50 people)

1 (monopoly) 5, 4.50% 1, 2%

2 26, 23.42% 4, 8%

3 30, 27.03% 14, 28%

4 32, 28.83% 26, 52%

5 (completely competitive) 18, 16.22% 5, 10%

Search No (in total 94 people) Yes (in total 67 people)

1 (monopoly) 4, 4.26% 2, 2.99%

2 19, 20.21% 11, 16.42%

3 28, 29.79% 16, 23.88%

4 31, 32.98% 27, 40.30%

5 (completely competitive) 12, 12.77% 11, 16.42%

Table 1.3 and 1.4: absolute and relative numbers of respondents per competition scale in relation to switching and searching behaviour

Looking at the percentages of number of people who answered a certain scale number as percentage of the total amount of people who switched or the total amount of people that did not switch, it can be observed that relatively more people who answered that the market is very competitive did actually switch. However, when people answered that they thought the market was completely competitive, the percentage of people that did not switch is higher

(18)

than the percentage of people who did switch. Whilst analysing the searching behaviour in relation to the answer to how competitive the market is, again a relatively higher percentage of people who think that the market is very competitive (scale 4) do actually search. Though, when people think that the market is completely competitive (scale 5), also a higher

percentage of searchers instead of non-searchers can be noticed. This is in contrast to the findings of the switching behaviour. Amongst the people who answered that the market was a monopoly or only a little bit more competitive than a monopoly (scale 1 and 2), the

percentage of people that did not switch or search is higher than the relative number of people who were active.

When checking whether the previously found percentages agree with the probability that a consumer will not search or will not switch because of their thought about the market, I carried out two logistic regressions. The choice for a logistic regression is based on the fact that the dependent variable is a dummy variable and can only take the values zero and one. During the process of a logistic regression, the probability that the dependent variable is equal to one is being calculated.

The first logistic regression investigates the probability that someone does not switch as a function of the thoughts of the competitiveness of the market. In total 5 regressions have been performed, one per category of competitiveness of the market as an independent

variable. The regression function was as follows:

Pr 𝑌 = 𝑛𝑜𝑡 𝑠𝑤𝑖𝑡𝑐ℎ𝑖𝑛𝑔 = 𝛼 + 𝛽 ∗ 𝑡ℎ𝑜𝑢𝑔ℎ𝑡 𝑎𝑏𝑜𝑢𝑡 𝑠𝑐𝑎𝑙𝑒 + 𝜀. The thought about the scale ranged from 1 to 5 and the constant and error term are included to include the eventual errors that can occur.

The coefficients of the regression were subsequently transformed into probabilities that someone did not switch2. The probabilities can be found in the following table.

Competitiveness Pr 𝑌 = 𝑛𝑜𝑡 𝑠𝑤𝑖𝑡𝑐ℎ𝑖𝑛𝑔 𝑋_!, 𝑋_!, … , 𝑋_! 1 (monopoly) 69.80% 2 77.86% 3 48.78% 4 27.21% 5 (completely competitive) 63.53%

Table 1.5: probability of not switching with independent variable competitiveness market

2_Formula:_{Pr 𝑌 = 1 𝑋}

(19)

The discovered results agree with the earlier calculated relative numbers. The lowest probability that an insured person does not switch is when they think the market is very competitive (scale 4). People are most likely to not switch when they think the market form is monopolistic or somewhat more competitive than a monopoly (scale 2).

The same logistic regression has been carried out to investigate the probability that a consumer does not search for alternatives as a function of their thought about the market. The following results have been found.

Competitiveness Pr 𝑌 = 𝑛𝑜𝑡 𝑠𝑒𝑎𝑟𝑐ℎ𝑖𝑛𝑔 𝑋_!, 𝑋_!, … , 𝑋_! 1 (monopoly) 58.45% 2 55.56% 3 59.90% 4 41.16% 5 (completely competitive) 41.96%

Table 1.6: probability of not searching with independent variable competitiveness market

The probability that an insured person does not search for alternative policies is highest when they think that the market is not so competitive (scale 1, 2 and 3). This partly agrees with the findings for the switching behaviour except for scale 3. The lowest probability that someone has a passive attitude can be found when people think the market is very competitive or completely competitive.

In conclusion it can be said that a positive relation between the thought of a

competitive market and a more active attitude towards searching and switching can be found. The fact that people choose to switch less when they think that the market is completely competitive (scale 5) can be explained by the idea that their own health insurance provider already offers the lowest possible price. This probability of not switching is still lower than when people see the market as a monopolistic one (scale 1 and 2). A notion has to be made about the significance of the coefficients determined by the logistic regression, because some of the coefficients are not significant. Though, the results do agree with the previously calculated relative numbers and there are no variables that can be left out in this research.

Competition and how much people think they can save when switching

The survey question about what people think they can save when they switch to another health insurance also provides insight in the overall thought of the competitiveness of the market. In total 39 people answered that they think they can save zero euros when switching.

(20)

This would mean that they think the market is completely competitive and that price differences do not exist or that a cartel exists (scale 1) and that therefore zero euros can be saved. However, only 23 people answer that they think the market is very competitive or completely competitive (scale 4 and 5). Above this, 17 people answered that they had been searching for alternative insurances. The remaining 22 people did not search at all. This results show that people are not really thinking rationally about the competitive market structure and their expenditures to health care. In the data part, the average amount of money that people thought they could save when switching was calculated. Using this average, I can calculate the probability that a consumer does not search or switch. To do this, I performed a logistic regression. The regression results can be found in figure 1.8. The probability that an insured person does not search as a function of the average amount of money people think they can save (€88.08) is 47.95%.

Figure 1.8 Logistic Regressions probability someone searches or switches on their saving thoughts

The probability that an insured person does not switch as a function of the average amount of money people think they can save (€88.08) is 46.79%.

.

_cons .447211 .2035675 2.20 0.028 .048226 .846196 thinksave -.0009325 .0013958 -0.67 0.504 -.0036683 .0018033 searchno Coef. Std. Err. z P>|z| [95% Conf. Interval] Log likelihood = -108.74817 Pseudo R2 = 0.0020 Prob > chi2 = 0.5048 LR chi2(1) = 0.44 Logistic regression Number of obs = 161 Iteration 3: log likelihood = -108.74817

Iteration 2: log likelihood = -108.74817 Iteration 1: log likelihood = -108.74821 Iteration 0: log likelihood = -108.97059 . logit searchno thinksave

_cons .9308845 .2184586 4.26 0.000 .5027135 1.359055 thinksave -.0014599 .0014436 -1.01 0.312 -.0042893 .0013694 switchno Coef. Std. Err. z P>|z| [95% Conf. Interval] Log likelihood = -99.244794 Pseudo R2 = 0.0050 Prob > chi2 = 0.3162 LR chi2(1) = 1.00 Logistic regression Number of obs = 161 Iteration 3: log likelihood = -99.244794

Iteration 2: log likelihood = -99.244794 Iteration 1: log likelihood = -99.246208 Iteration 0: log likelihood = -99.7471 . logit switchno thinksave

(21)

Test the explanations for consumers their behaviour

During the survey, I have been asking consumers about the reasons for their searching and switching behaviour. I have performed a Chi-squared test to examine if two variables cohere with each other. The higher the Chi-squared value is, the more coherence between the two variables is present. By means of the Chi-squared distribution it can be checked whether the test value is higher than what would have been expected based on coincidence. The number that can be found during the test has no causal meaning for itself.

In this part of the research the null hypothesis is that there is no statistical dependency between the arguments and the searching and switching behaviour. The alternative hypothesis tells that there is a statistical dependency. I have separated and summarised the arguments in words, namely searching- and switching costs, risk aversion and price. For all three arguments a specific Chi-squared test has been carried out. Firstly this has been done with respect to the variable about searching behaviour and secondly concerning the switching behaviour of insured people.

First of all I analysed the argument about searching- and switching costs. In this test I investigated whether people do not search because they do not want to take time and because they think it is too difficult to sort it all out. The Chi-squared test explores the coherence of those two variables. The examined value of the test is 5.1559. At a significance level of 5%, the link between the two variables is significant and the null hypothesis is therefore rejected.

For the other arguments, risk aversion and price, the same Chi-squared test has been performed. The results of the tests can be found in the following table.

Chi-square value p-value

Searching yes and Searching Costs 5.1559 0.023**

Searching yes and Risk Aversion 33.7449 0.0000***

Searching yes and Price 65.4430 0.0000***

Table 1.7 (significance level: 1% (***), 5% (**), 10% (*))

The p-values of the arguments about risk aversion and price are lower and it can be said that the link between these two variables and searching is more significant than the relation between searching costs and searching behaviour.

At a significance level of 5% all three the arguments would have a significant link to the searching behaviour of consumers. Price is the argument that has the most important effect on the searching behaviour.

(22)

Secondly, the coherence between switching behaviour and the three arguments has been analysed. The results can be found in the table below and in the appendix.

Chi-square value p-value

Switching yes and Switching Costs 10.8549 0.001***

Switching yes and Risk Aversion 24.2902 0.0000***

Switching yes and Price 66.7876 0.0000***

Table 1.8 (significance level: 1% (***), 5% (**), 10% (*))

At all three the tests with the specific arguments, the earlier stated null hypothesis is rejected. The argument switching costs again has the lowest value of the Chi-squared test. The highest Chi-squared value is again found at the argument about price. The relation between the

variable switching behaviour and the variable about price of the premium for the health policy is most significant.

Logistic Regression

Because of the fact that the Chi-squared test cannot tell anything about whether the relation between the variables is positive or negative, a logistic regression has been performed on searching behaviour as well as switching behaviour. The dependent variable of the regression is in the first model searching behaviour and in the second model it is replaced by switching behaviour. This variable is a dummy and can only take two values, namely 0 or 1. In this research 0 represents the fact that people did not search or did not switch and 1 represents the response that insured did search or switch to alternative policies. The logistic regression model is a nonlinear probability model. It therefore describes the effect of the independent variables on the probability that the dependent variable will take the value one. The logistic regression uses the logistic cumulative probability distribution functions.

The results for the coefficients of a logistic regression model do not have a meaning for itself. The values of the Betas can be integrated into the following function:

Pr 𝑌 = 1 𝑋_!, 𝑋_!, … , 𝑋_! = 𝐹 𝛽_!+ 𝛽_!𝑋_!+ ⋯ + 𝛽_!𝑋_! = 1

1 + 𝑒!(!!!!!!!!⋯! !!!!) When these values are filled into the equation, the probability that the dummy about searching and switching behavior is one is calculated. The independent variables searching- and

switching costs, price and risk aversion are all binary variables as well.

In this research the effect of the different arguments on the searching and switching behaviour is analyzed. Therefore, by filling in the previously mentioned function, only one

(23)

independent variable is examined and the other dummies are put equal to zero. The

coefficients of the independent variables that are determined by means of a logistic regression are all significant at a significance level of 5%.

I have calculated the probabilities that searching or switching takes place for all the separate behavioural arguments. This can be seen in the table below.

Pr 𝑆𝑒𝑎𝑟𝑐ℎ

= 𝑌𝑒𝑠 𝑋_!, 𝑋_!, … , 𝑋_! Pr 𝑆𝑤𝑖𝑡𝑐ℎ = 𝑌𝑒𝑠 𝑋!, 𝑋!, … , 𝑋! Searching/Switching cost 0.189820583 = 18.98%3 0.06451615 = 6.45%

Price 0.923130654 = 92.31% 0.949745154 = 94.97%

Risk Aversion 0.10714287 = 10.71% 0.119910782 = 11.99%

Table 1.9 the probabilities that a consumer will search or switch concerning the different behavioural reasons.

These results can be interpreted, as the chance that an insured person searches for other health insurances is approximately 18.98% when this people’s behaviour is dominated by factors of searching/switching costs. The same calculation can be done to calculate the probability that a consumer switches to an alternative policy when aspects of searching- or switching costs lead his or her behaviour. The chance that someone switches when this is the case is 6.45%.

In conclusion, it can be seen that a consumer with behaviour that is mainly influenced by factors of pricing is most likely to search or switch to alternative health insurances.

To check the effects of behaviour on the non-searching or switching behaviour, a logistic regression on the probability that someone does not search or switch has also been carried out. These results are tabled below.

Pr 𝑆𝑒𝑎𝑟𝑐ℎ

= 𝑁𝑜 𝑋_!, 𝑋_!, … , 𝑋_! Pr 𝑆𝑤𝑖𝑡𝑐ℎ = 𝑁𝑜 𝑋!, 𝑋!, … , 𝑋! Searching/Switching cost 0.834006879 = 83.40% 0.93548385 = 93.55%

Price 0.093493211 = 9.35% 0.050254846 = 5.03%

Risk Aversion 0.89285712 = 89.29% 0.880089218 = 88.01%

Table 1.10 the probabilities that a consumer will not search or switch concerning the different behavioural reasons.

For the argument about risk aversion, a probability on not searching of 89.29% can be found. The chance that a consumer, who lets its behaviour effect a lot by risk aversion, does not switch to other policies is 88.01%. These percentages are calculated on the same way as before.

The price has a very low probability effect on not searching or switching. Therefore it can be concluded that insured consumers do not think that the competitive health insurance

3_{Pr 𝑆𝑒𝑎𝑟𝑐ℎ = 𝑌𝑒𝑠 𝑋}

!, 𝑋!, … , 𝑋! = _!!!!(!!.!!"#$%$! !.!"#$!%∗!!!.!"#$%&∗!!!.!""##$∗!)! =

(24)

market offers the lowest price possible for the policies. In the case of the belief that the lowest possible price would already been there, consumers did not have an incentive to search or switch when they cared a lot about the expense. Though, in the previous logistic regressions it can be seen that when someone’s behaviour is mainly influenced by the price of the health insurance, that same person is most likely to search or switch.

In table 1.2 the absolute numbers show that risk aversion behaviour has the largest effect on consumers making the choice not to switch. However in this regression the complete behaviour of consumers considering different factors is being checked, so not only the

explanations of people who do not switch or search are studied. Therefore the lower

probability on not switching at the factor of risk aversion compared to switching costs can be explained by the fact that the chance that a consumer does switch is higher when his or her behaviour is influenced by risk aversion than by switching costs.

(25)

The effect of searching and switching on the premium amount

To analyse the effect of searching and switching behaviour on the amount of premium the insured pay, I did another OLS-regression. In this case the dependent variable was the amount of premium and the independent binary variables concerned the searching behaviour, the switching behaviour or both.

First of all the regression of premium on the searching behaviour has been performed. The model of this regression would look like the following:

𝑃𝑟𝑒𝑚𝑖𝑢𝑚 = 𝛼 + 𝛽 𝐷𝑢𝑚𝑚𝑦𝑆𝑒𝑎𝑟𝑐ℎ𝑖𝑛𝑔

A coefficient for Beta of -105.1151 is found. This means that the effect of searching has a negative effect on the dependent variable, so that the amount of yearly premium decreases when someone does search for alternative health insurances. This coefficient cannot be defined as significantly different from zero. The constant of the model has a value of 1,442.87, which is approximately the average amount of premium, the respondents in the sample pay. The regression results can be seen in figure 1.9.

Figure 1.9 The regression of premium on active searching behaviour

Secondly the regression of premium on the switching behaviour has been carried out. The model for this regression exactly looks the same as the one before except for the fact that the independent variable is now about switching instead of searching. Again the value for Beta is negative, namely -73.58 and so the amount of premium consumers have to pay each year decreases when they switch to other health policies. Nevertheless, this effect cannot be stated as significantly different from zero, because of the high p-value. The regression results can be found in figure 1.10.

_cons 1442.872 60.69757 23.77 0.000 1322.995 1562.75 searchyes -105.1151 94.09074 -1.12 0.266 -290.944 80.71373 premium Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 55496206.5 160 346851.291 Root MSE = 588.48 Adj R-squared = 0.0015 Residual 55063984.3 159 346314.366 R-squared = 0.0078 Model 432222.244 1 432222.244 Prob > F = 0.2656 F( 1, 159) = 1.25 Source SS df MS Number of obs = 161 . regress premium searchyes

(26)

Figure 1.10 The regression of premium on active switching behaviour

The fact that both the coefficients are not significant can probably be assigned to the high standard deviations. This can be remedied if the sample size would be greater. Next to this, because it is a small multiple regression model, the chance to have omitted variable bias is present. When there are omitted factors that are correlated to the other regressors and which are a determinant of the dependent variable too, it can imply that the OLS-estimators are inconsistent. Next to this, the premium and the fact of switching are jointly determined. During this research I only wanted to check whether there is a correlation between the two factors. However I do not believe that the regression results show a causal relationship. An OLS-estimator should be unbiased, consistent and asymptotically normally distributed, so when inconsistency arises the estimator cannot be interpreted optimally. Another explanation for the insignificant coefficients can be found in the fact that the premiums are quite high relative to the amount of money that can be saved when searching or switching. This is partly because of the regulated market force, which makes sure that the price differences between the different health insurers do not become enormous. It is therefore very hard to find a significant effect on the comparatively high amount of premium.

Because the previous Chi-squared tests showed that the argument concerning price had the most important coherence with the searching and switching behaviour, another two least squares regressions have been carried out. The dependent variable is the premium amount and the independent variable is the dummy if people search or switch because of the price.

When insured people search for alternatives because they want to pay a lower price for their health policy, they eventually pay a lower premium. The coefficient for the Beta

(-168.331) is negative. Though, the coefficient cannot be seen as significant because of the high p-value. It is however noteworthy to emphasise that this coefficient is more significant

_cons 1421.979 55.98093 25.40 0.000 1311.417 1532.541 switchyes -73.57928 100.4542 -0.73 0.465 -271.9759 124.8174 premium Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 55496206.5 160 346851.291 Root MSE = 589.8 Adj R-squared = -0.0029 Residual 55309577.9 159 347858.981 R-squared = 0.0034 Model 186628.59 1 186628.59 Prob > F = 0.4650 F( 1, 159) = 0.54 Source SS df MS Number of obs = 161 . regress premium switchyes

(27)

than the coefficient of searching behaviour in general. Therefore it is probable that the

influential force of price has a larger effect on a decrease in the yearly premium than the point of searching alone. This can be explained due to the fact that searching because of

dissatisfaction with the current health insurance often leads to eventually paying a higher premium amount.

The regression of premium on the variable that people switch because they want to pay a lower price for their health insurance offers the same negative coefficient. With a Beta of -207.1989, people approximately have to pay 207 euros less for their premium when they switch to alternative policies. In the case of a significance level of 5%, this coefficient is not significant.

(28)

Conclusion

I have carried out this survey to provide information about the behaviour of consumers with respect to their choices for health insurances.

The central question of this research was: do consumers search for alternative health insurances, do they switch and does their belief about a competitive market structure

influence the searching or switching behaviour of consumers concerning their health insurance? This question can be answered by the following findings.

Based on a confidence interval of 95% it can be stated that the behaviour of insured consumers can be described as passive. Enough competition is essential for the efficient functioning of the new market system in the Netherlands and for reassuring that the premium decreases. This has so far not been happening and the very low mobility percentage of the insured can be comprehended as an explaining feature of this. Health insurance companies namely do not have an incentive to offer their insurances for the lowest possible price if the demand is not elastic and moving.

A relation between the thought of a competitive market and the searching and switching behaviour can be noticed. The probability that a consumer does not search or switch is namely lowest when people think that the market is very competitive or completely competitive. Insured are mostly passive when they consider the market to be monopolistic or almost monopolistic. The thoughts of people about the amount of money they think they can save when switching to another policy in relation to their thought about the competitiveness of the market is not totally in agreement. Therefore it cannot be said that the behaviour of consumers in the health insurance market is completely rational.

The searching and switching behaviour is mainly dominated by the fact that people want to pay the lowest price. However, the high probability that consumers do not search can be explained by factors of behaviour concerning risk aversion. People are satisfied with the policy that they currently have and they are afraid of an unknown health insurance or making a mistake whilst choosing a new one. The probability that someone does not switch is highest when the behaviour of insured people is dominated by switching costs. However, aspects of risk aversion are in absolute numbers also a main reason of not switching to other health policies. The findings of this research correspond to previously done investigations of for example Frank & Lamiraud and Enthoven.

(29)

yearly premium amount the person has to pay. However, this effect is not significant, so no real judgements can be made about that.

In order to increase the number of people who search and switch each year and

therefore to increase the activity on the demand side of the market, the procedure of searching and switching should be easier and not so time-consuming as nowadays. This will reduce the effect of the searching- and switching costs. All offered health insurances should be compared effortlessly and the provision of information on each health policy should be completely transparent and optimal. This, according to me, will probably take away a large part of the uncertainty and fear of new health insurances.

In follow-up studies it is necessary to increase the amount of observations and expand the sample size. Next to this is would be useful and interesting to know whether demographic factors have an influence on the outcomes of searching and switching behaviour. An example of this would be the question if younger people are more actively searching and switching than the older generation.

(30)

References

Brennan, T. J., 2007. “Consumer preference not to choose: Methodological and policy implications”, Energy Policy, Vol. 35, No. 3, pp. 1616-1627

Buchmueller, T. C. & Feldstein, P. J., 1996. “The effect of price on switching among health plans”, Journal of Health Economics, 16, pp. 231-247

CBS, 2014. “CBS: Premie basisverzekering bleef lange tijd gelijk, vermogens van zorgverzekeraars stegen”, Webmagazine. Consulted on the 1st of Decemember 2015, of

http://www.cbs.nl/nl-NL/menu/themas/gezondheid-welzijn/publicaties/artikelen/archief/2014/2014-4199-wm.htm Culyer, A. J. & Newhouse, J. P. (2000) Handbook of Health Economics. Vol. 1a. Elsevier

De Nederlandsche Bank, 2005. “Het nieuwe zorgstelsel internationaal vergeleken”, Kwartaalbericht December 2005.

De Vries, L. & Heijnen, P., 2006. “An empirical analysis of consumer switching behavior in the Dutch

electricity market”, Delft University of Technology, Faculty of Technology, Policy and Management

Diamond, P., 1992. “Organizing the Health Insurance Market”, Econometrica, vol. 60, No. 6., pp. 1233-1254 Dormont, B., Geoffard, P. and Lamiraud, K.,2009, “The influence of supplementary health insurance on

switching behaviour: evidence from Swiss data”, Health Economics

Enthoven, A. C., 1993. “The History And Principles of Managed Competition”, Health Affairs.

Frank, Richard G. & Lamiraud, Karine, 2009. "Choice, price competition and complexity in markets for

health insurance," Journal of Economic Behavior & Organization, Elsevier, vol. 71(2), pp. 550-562, August

Handel, B. R. & Kolstad, J. T., 2013. “Health insurance for ‘’Humans’’: Information frictions, Plan Choice,

and Consumer Welfare”, NBER Working Paper No. 19373

Miller, R. H. & Luft, H. S., 2002. “Private Health Insurance in OECD Countries”, Health Affairs, vol. 21 (4), July/August

Poiesz, Th. B. C., 2004. “The Free Market Illusion Psychological Limitations of Consumer Choice”, Tijdschrift voor Economie en Management. Vol. XLIX (2)

Pomp, M., Shestalova, V. and Rangel, L., 2005. “Switch on the competition; Causes, consequences and policy

implications of consumer switching costs”, The Hague: CPB

Rothschild, M., Stiglitz, J., 1976. “Equilibrium in Competitive Insurance Markets: An Essay on the Economics

of Imperfect Information”, The Quarterly Journal of Economics, Vol. 90, No. 4, pp. 629-649

Schoen, C., Osborn, R., Squires, D., Doty, M. M., Pierson, R. and Applebaum, S., 2010. “How Health

Insurance Design Affects Access To Care And Costs, By Income, In Eleven Countries”, Health Affairs, 29, no.

12

Smith, V. L., 1962. “An experimental study of competitive market behaviour”, Journal of Political Economy, vol. 70, no. 2, pp. 111-137

Stock, J. H. & Watson, M. W. (2015).“Introduction to Econometrics”(Updated Third Edition). Harlow: Pearson Education Limited

(31)

Sturluson, 2002. “The Importance of Consumer Search- and Switching Costs for Competition in Electric Power

Retailing”, Stockholm: Stockholm School of Economics and Institute of Economic Studies.

Van de Ven, W. P., van Vliet, R. C., van Barneveld, E. M. and Lamers, L. M.,1994. “Risk-adjusted

capitation: recent experiences in The Netherlands”, Health Affairs, 13, no.5 pp. 120-136

Romp, M. & Merkx, P., 2015. “Verzekerden in beeld 2015”, Vektis Zorgthermometer.

Weisbrod, B., 1991. “The Health Care Quadrilemma: An Essay on Technological Change, Insurance, Quality

of Care, and Cost Containment”, Journal of Economic Literature, vol. 28, pp. 523-552

Wilson, C. M. and Waddams Price, C., 2005. “Irrationality in Consumers’ Switching Decisions: When More

(32)

Appendix Questionnaire 1. What is your gender?

• Male

• Female

2. What is your age? • 18-35 years old • 35-50 years old • 50-65 years old 3. Do you have a partner?

• Yes • No

4. Do you have children? • Yes

• No

5. Did you search for alternative health insurances in the past three years? • Yes

• No

6. In case of answering yes in question 5, why did you search?

• I wasn’t satisfied with my current health insurance of health insurer • I was looking for the lowest possible price for my health insurance • I had enough time and knowledge to compare different health insurances

• Other, namely:

7. In case of answering yes in question 5, by which means did you search? • By means of a comparing website on the Internet

• By means of the websites of each health insurance company • By means of calling to the health insurers

• By means of word-of-mouth advertising

8. In case of answering no in question 5, why didn’t you search?

• I know what I can expect from my health insurance and I like certainty

• The lowest possible price is already being offered because of the competitive market • I do not have enough time and knowledge to search for alternative policies (the process of

searching is too difficult)

9. Did you switch to an alternative health insurance in the past three years? • Yes

• No

(33)

• I wasn’t satisfied with my current health insurance of health insurer • I was looking for the lowest possible price for my health insurance • I had enough time and knowledge to compare different health insurances

11. In case of answering no in question 9, why didn’t you switch?

• I know what I can expect from my health insurance and I like certainty

• The lowest possible price is already being offered because of the competitive market • I do not have enough time and knowledge to search for alternative policies (the process of

searching is too difficult)

12. What do you currently pay (premium) for your health insurance on a yearly basis? 13. How much money do you think you can save when switching to another health insurance? 14. How much money do you want to save to make you switch to another health insurance?

15. On a scale from 1 to 5 (1=monopoly, 5=completely competitive), how competitive do you think the market for health insurance is in the Netherlands?

16. On a scale from 1 to 5 (1=lowest possible price, 5=highest possible price), how high do you think is the current premium price that you pay compared to the rest of the Netherlands?

Data

Statistics on the numerical questions about expenditures and savings

Average Standard Deviation Minimum amount Maximum amount Premium Amount €1,399.129 588.9408 €624 €6,000

How much they

think they can

save by switching

€88.08 114.16 €0 €500

How much they

want to save to

make them switch

(34)

Competitive market and searching/switching behaviour

Logistic Regressions searching/switching behaviour on thought competitiveness market No Searching:

_cons .3519764 .163138 2.16 0.031 .0322318 .6717211 comp1 .3411708 .8812571 0.39 0.699 -1.386061 2.068403 searchno Coef. Std. Err. z P>|z| [95% Conf. Interval] Log likelihood = -108.89325 Pseudo R2 = 0.0007 Prob > chi2 = 0.6941 LR chi2(1) = 0.15 Logistic regression Number of obs = 161 Iteration 3: log likelihood = -108.89325

Iteration 2: log likelihood = -108.89325 Iteration 1: log likelihood = -108.89335 Iteration 0: log likelihood = -108.97059 . logit searchno comp1

_cons .3234002 .1770303 1.83 0.068 -.0235728 .6703731 comp2 .2231436 .418187 0.53 0.594 -.5964879 1.042775 searchno Coef. Std. Err. z P>|z| [95% Conf. Interval] Log likelihood = -108.82652 Pseudo R2 = 0.0013 Prob > chi2 = 0.5914 LR chi2(1) = 0.29 Logistic regression Number of obs = 161 Iteration 3: log likelihood = -108.82652

_cons .2578291 .1864386 1.38 0.167 -.1075839 .6232421 comp3 .4014165 .3686581 1.09 0.276 -.32114 1.123973 searchno Coef. Std. Err. z P>|z| [95% Conf. Interval] Log likelihood = -108.36604 Pseudo R2 = 0.0055 Prob > chi2 = 0.2715 LR chi2(1) = 1.21 Logistic regression Number of obs = 161 Iteration 3: log likelihood = -108.36604

(35)

No Switching:

_cons .4953214 .2031404 2.44 0.015 .0971736 .8934693 comp4 -.3571711 .3325073 -1.07 0.283 -1.008873 .2945313 searchno Coef. Std. Err. z P>|z| [95% Conf. Interval] Log likelihood = -108.39416 Pseudo R2 = 0.0053 Prob > chi2 = 0.2830 LR chi2(1) = 1.15 Logistic regression Number of obs = 161 Iteration 3: log likelihood = -108.39416

.

_cons .4115074 .1738678 2.37 0.018 .0707328 .752282 comp5 -.324496 .4521863 -0.72 0.473 -1.210765 .5617729 searchno Coef. Std. Err. z P>|z| [95% Conf. Interval] Log likelihood = -108.71455 Pseudo R2 = 0.0023 Prob > chi2 = 0.4742 LR chi2(1) = 0.51 Logistic regression Number of obs = 161 Iteration 3: log likelihood = -108.71455

_cons .7716188 .1727487 4.47 0.000 .4330375 1.1102 comp1 .8378188 1.108982 0.76 0.450 -1.335747 3.011384 switchno Coef. Std. Err. z P>|z| [95% Conf. Interval] Log likelihood = -99.410522 Pseudo R2 = 0.0034 Prob > chi2 = 0.4120 LR chi2(1) = 0.67 Logistic regression Number of obs = 161 Iteration 3: log likelihood = -99.410522

Iteration 2: log likelihood = -99.410522 Iteration 1: log likelihood = -99.41203 Iteration 0: log likelihood = -99.7471 . logit switchno comp1