Price elasticities in the German Statutory Health Insurance market before and after the health care reform of 2009.

G Model

ARTICLE IN PRESS

HEAP-3365; No. of Pages 10

Health Policy xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Health Policy journal homepage: www.elsevier.com/locate/healthpol

Price elasticities in the German Statutory Health Insurance market before and after the health care reform of 2009 Jonas B. Pendzialek ∗ , Marion Danner, Dusan Simic, Stephanie Stock Institute of Health Economics and Clinical Epidemiology, University of Cologne, Gleueler Straße 176-178, 50935 Cologne, Germany

a r t i c l e

i n f o

Article history: Received 24 June 2014 Received in revised form 20 January 2015 Accepted 22 January 2015 Keywords: Health insurance Price elasticity Health plan choice Managed competition

a b s t r a c t This paper investigates the change in price elasticity of health insurance choice in Germany after a reform of health insurance contributions. Using a comprehensive data set of all sickness funds between 2004 and 2013, price elasticities are calculated both before and after the reform for the entire market. The general price elasticity is found to be increased more than 4-fold from −0.81 prior to the reform to −3.53 after the reform. By introducing a new kind of health insurance contribution the reform seemingly increased the price elasticity of insured individuals to a more appropriate level under the given market parameters. However, further unintended consequences of the new contribution scheme were massive losses of market share for the more expensive sickness funds and therefore an undivided focus on pricing as the primary competitive element to the detriment of quality. © 2015 Elsevier Ireland Ltd. All rights reserved.

1. Introduction A decade after the introduction of managed competition to the German Statutory Health Insurance a new law on strengthening competition (GKVWettbewerbsstärkungsgesetz) came into effect in 2009. Its aim was to foster competition between sickness funds, not only based on price, but also focusing on the improvement of quality and service [1]. One major change was the move away from price differences expressed by incomedependent contributions, deducted via payroll, towards price differences expressed by income-independent add-on premiums paid directly by the insured individual. Surveys conducted before the introduction of these add-on premiums, suggested extraordinary high switch-

∗ Corresponding author at: Schwalbenstrasse 59, 22305 Hamburg, Germany. Tel.: +49 173 9908479. E-mail addresses: [email protected] (J.B. Pendzialek), [email protected] (M. Danner), [email protected] (D. Simic), [email protected] (S. Stock).

ing rates. According to one survey 35% of all respondents were prepared to leave the sickness fund in case of an addon premium being charged [2]. Between 2009 and 2013 only a few sickness funds had to actually charge add-on premium. However, those that did suffered extraordinary losses in members. Table 3 gives an overview of all sickness funds with add-on premiums between 2009 and 2013 as well as their respective loss of members in the first year of add-on premium being charged. This development had a major impact on the deployed strategies of the sickness funds. They focused heavily on avoiding to charge add-on premiums, thereby dropping most measures on improving quality and the organization of care [3,4], especially when long-term investments would be necessary [5]. This study investigates the resulting research question on how much the price elasticity of sickness fund choice increased in the wake of the reform. The study also looks at whether this reform resulted in undesirably high price elasticity within the context of the German managed competition setting as well as in an international comparison. The significance of the price elasticity has to be balanced in the “value-for-money” competition of both price and

http://dx.doi.org/10.1016/j.healthpol.2015.01.014 0168-8510/© 2015 Elsevier Ireland Ltd. All rights reserved.

Please cite this article in press as: Pendzialek JB, et al. Price elasticities in the German Statutory Health Insurance market before and after the health care reform of 2009. Health Policy (2015), http://dx.doi.org/10.1016/j.healthpol.2015.01.014

G Model HEAP-3365; No. of Pages 10

ARTICLE IN PRESS J.B. Pendzialek et al. / Health Policy xxx (2015) xxx–xxx

2

Table 1 Overview of price elasticity estimations for Germany found in previous studies. Author

Data source

Year(s)

Mean price elasticities

Schut et al. [18] Schut et al. [18] Schwarze and Andersen [24] Tamm et al. [25] Schmitz and Ziebarth [26] Schmitz and Ziebarth [26] Schmitz and Ziebarth [26]

Data on sickness fund type level Data on sickness fund type level Survey for the German Socio-Economic Panel Data on sickness fund level Data on sickness fund level Data on sickness fund level Data on sickness fund level

1996–2001 1999–2001 1999–2000 2001–2004 2002–2009 2009–2010 2009–2010

−2.48 −4.31 −4.17a −1.06 −0.6 −1.8b /−0.9c −0.9

a b c

Calculation is an arc price elasticity based on the Schwarze and Andersen [24] finding, that 1% price increase will lead to a 4.17% change in membership. Mean price elasticity calculated for add-on premiums only. Mean price elasticity calculated for premium rebate only.

quality that represents the managed competition concept of Enthoven [6]. In the past decades price elasticity of health insurance choice has been subject to a variety of studies in different countries and substantial literature exists for international comparison. Price elasticities can either reflect the insurer’s perspective or the individual’s perspective. The perspective depends on whether the elasticity is calculated using the total premium charged by the insurance (total price elasticity) or using only the part paid by the individual (out-of-pocket price elasticity). Thus, total price elasticities are manifold higher then out-of-pocket price elasticities, depending on the proportion of the employer’s share of the premium. The following price elasticities are all out-of-pocket price elasticities if not stated otherwise. Most studies were conducted in the United States health insurance system. Scanlon et al. [7] and Buchmueller [8] give comprehensive reviews. In particular the later research during the past 20 years finds price elasticities ranging between −0.1 and −1.0 [9–14]. Besides the US, only few countries have similar systems with competition in the primary health insurance market [15,16]. Within this group of countries, research found higher price elasticities in the Netherlands (−0.5 to −7.0) [17–20], Switzerland (−1 to −2) [21–23] and Germany. In Germany an early study by Schut et al. [18] found increasing price elasticities between −2.45 and −4.31 in the first five years after the introduction of freedom of choice. These comparably higher elasticity values are in line with another analysis for approximately the same period by Schwarze and Andersen [24], finding a 4.17 percentage point decrease in market share when increasing the contribution rate by 1 percentage point. A different study found a lower short-term price elasticity of −1.06 for the years 2001–2004 [25]. A recent study by Schmitz and Ziebarth [26] using data on sickness funds estimated a considerably lower price elasticity of −0.6. In 2009 the way sickness funds charged for their coverage changed to a nominal value (for a more detailed explanation see Section 2). Since then, only the above mentioned study of Schmidt and Ziebarth investigated price elasticity. Using data on the sickness fund level of just five sickness funds (representing about one third of the market) until 2010, they calculated a price elasticity of −1.8 for the add-on premium and −0.9 for a premium rebate [26]. In particular the add-on premium price elasticity increased enormously, namely threefold. These findings are

supported by another study initiated by the German Federal Social Insurance Office [27]. Table 1 gives an overview of the German studies. Further international studies investigated supplementary health insurance [28–30] and duplicated private insurance in single payer systems [31,32]. This study aims to assess the expected increase in price elasticity after the reform by using recent data on the sickness fund level. This new data thus complements the research on price elasticity prior to the reform and tests the hypothesis whether the introduction of nominal add-on premiums increased the price elasticity massively. The use of this new data provides detailed figures for all sickness funds in Germany. This is the first study to investigate price elasticity after the 2009 health care reform based on complete data of the statutory health insurance market and not only a fraction of the market or survey based data. Furthermore, this study utilizes the newest relevant data up until 2013, thereby covering the entire period of add-on premiums. It therefore expands and complements knowledge on price elasticity after the reform that was covered by a previous study [26]. Changing to a nominal add-on premium, the German Statutory Health Insurance system became similar to other health insurance settings with managed competition. As a consequence the German price elasticity is now better comparable to US, Dutch and Swiss price elasticities, which enables the discussion in a different manner. The remainder of this paper is structured as follows: Section 2 explains important aspects of the German health insurance system and the 2009 reform to facilitate an understanding of the particular setting of this study. Section 3 describes the methods used, while the Section 4 characterizes the data the research is based upon and gives the results of the research. Finally, Section 5 concludes with a discussion of the findings and conclusions for policy makers.

2. Institutional background of the Statutory Health Insurance in Germany Health insurance is mandatory for all individuals in Germany. Almost 90% of Germans are covered by Statutory Health Insurance (SHI). Only few individuals who are self-employed or above a certain income threshold can opt out and get enrolled with private insurance companies. Private insurance covers about 11% of the population [33].




Dependents of SHI members are covered at no additional fee. Sickness funds within the Statutory Health Insurance have to enrol every individual and charge the same percentage contribution rate to all of its subscribed members. The benefit package of the SHI is predetermined by law and covers almost all medical needs [34]. The sickness funds have to contract with all certified providers. Therefore, switching between sickness funds is not expected to affect the individual’s relationship with providers [26]. As per 2013 a total of 134 individual sickness funds exist, which are historically grouped by sickness fund types depending on the historic clientele. The Betriebskrankenkassen (BKK) are company-based and used to be restricted to the employees of that company. Innungskrankenkassen (IKK) were guild-based and still others were open to farmers (Landwirtschaftskrankenkassen, LKK), sailors (Seekrankenkasse, not existing anymore) or mineworkers (Knappschaft, KBS). Ersatzkassen (EK) existed for white collar workers and the Allgemeine Ortskrankenkassen (AOK) insured the population not eligible for other sickness funds. With the introduction of freedom of choice in 1997 most of the sickness funds that used to have restricted access opened themselves to the general public. Sickness funds are generally financed by contributions.1 Before the 2009 reform, every sickness fund set its own contribution rate as a percentage of gross income of its members to meet its financial needs. For employed individuals, which represent the vast majority of the enrollees, the premium is almost co-shared between the employee and the employer and is automatically deducted by payroll. After the law on strengthening competition (GKVWettbewerbsstärkungsgesetz) came into effect in 2009 all contribution rates were fixed at an equal level (as of 2013 at 15.5% of the gross income). For additional funding, if needed, sickness funds were allowed to charge a nominal add-on premium (Zusatzbeitrag). Unlike the income-dependent contribution rate this add-on premium had to be paid entirely and directly by the member.2 Sickness funds with surplus distributions from the pool were allowed to grant a rebate on the individual’s premium for the past year. These rebates were also paid as an incomeindependent amount. The first sickness fund to charge an add-on premium did so in 2009. Several others followed in early 2010. By 2012 up to fourteen sickness funds were taking an add-on premium while ten sickness funds were paying rebates. In spite of the great awareness of these new add-on premiums, which markedly changed the rules of the game, little research was until now conducted with adequate quantitative data. This study aims to investigate the effects

1 Additional source of financing during the research period were subsidies by the government, a fixed copayment of a 10 Euro fee for doctor visits or hospital stay days as well as revenue from financial investments. 2 Originally the flat fee was limited to 8 Euro a month but a higher and then income-depended add-on premium of up to 1% of the net income was allowed. Later the sickness funds were allowed to charge more than 8 Euro add-on premium a month.

3

of the newly introduced add-on premiums and rebate payments on price elasticity of health insurance in Germany. Furthermore, this research seeks to explore differences in price elasticity by sickness funds that were not yet addressed in the German price elasticity studies presented in the first section. 3. Data and methods 3.1. Data collection The analysis is based on data at the sickness fund level. Statistics on their members are reported monthly by every sickness fund to the Federal Social Insurance Office (Bundesversicherungsamt). These figures were documented yearly in the Sickness Fund Factbook (Krankenkassenatlas) until 2009 [35] and since then quarterly in the specialist journal Dienst für Gesellschaftspolitik by the same editor and publishing house [36–38]. Based on these publications, yearly membership figures are obtainable from 2004 to 2013, being January the 1st the reporting date in each case. The documented data was, however, not entirely exhaustive since a few BKKs did not report their statistics. A comparison with data of the entire market reported by the Federal Ministry of Health shows, that in most years this affects only 0.5% or less of the relevant population and which is thus missing in this data set. An overview of the amount of missing data points is reported in Table 2. The data set had to be further cleaned of data points of non-public sickness funds. Presently, some few sickness funds have retained restricted accessibility and are only open to employees of a certain company or a certain profession. Since these sickness funds do not participate fully in the competition and are clearly limited in their growth potential, all data points of these funds were discarded. Until January 2007 some BKKs as well as two bigger special types of sickness funds, the Knappschaft and the Seekasse had restricted access. Thus, in these years almost 4% of the members in the data set were disregarded. From 2008 to 2013 only a few BKKs remained restricted, which then accounted for approximately 1.2% of the data set. Table 2 reports this in more detail. During the research period many sickness funds were actively merging with each other. In the case of merging funds having the same contribution rate, add-on premium or rebate a single case was plainly created from the funds by adding up the market shares of all merging partners of the year preceding the merger. Likewise, for merging funds with different but similar contribution rates, add-on premiums or rebates a single case was created the same way but using a contribution value weighted by the market shares of the year previous to the merger. A few mergers occurred between sickness funds with high pricing differentials involving some of the lowest and highest priced funds in the market. In these particular cases an average contribution value would be misleading. Therefore, the market share of this resulting fund in a given year was approximated by using the information on previous growth and the market share of the merged funds given in that year. Any study relevant effects resulting from the merging of funds were controlled for.


G Model

ARTICLE IN PRESS


J.B. Pendzialek et al. / Health Policy xxx (2015) xxx–xxx

52.43 (100.0%) 0.84 (1.6%) 0.61 (1.2%) 50.99 (98.8%)

3.2. Model Source: Federal Ministry of Health [39–44], Müller and Lange [35], Lange [36–38], Own calculations. a Not including the agricultural sickness funds (LKK).

2012

51.20 (100.0%) 0.00 (0.0%) 0.61 (1.2%) 50.57 (98.8%) 50.78 (100.0%) 0.03 (0.1%) 0.61 (1.2%) 49.44 (97.4%)

2011

50.50 (100.0%) 0.03 (0.1%) 0.62 (1.2%) 43.07 (85.3%) 50.25 (100.0%) 0.05 (0.1%) 0.64 (1.3%) 43.52 (86.6%)

2010 2009 2008

49.92 (100.0%) 0.11 (0.2%) 1.85 (3.7%) 44.55 (89.2%) 49.71 (100.0%) 0.18 (0.4%) 1.83 (3.7%) 44.72 (90.0%) 49.88 (100.0%) 0.26 (0.5%) 1.89 (3.8%) 46.04 (92.3%) 49.98 (100.0%) 0.74 (1.5%) 1.87 (3.7%) 43.26 (86.5%)

2007 2006 2005 2004

SHI totala No data available Restricted access Used data

Year

Table 2 Overall population and data used in the study, members in million/percentage of population.

The remaining data set represents between 85% and almost 100% of the SHI population.3 To the best of our knowledge this is the most comprehensive set of data at the individual sickness fund level used in research on price elasticity so far. In particular it contains discrete data for small IKKs and BKKs that were not examined in previous studies. Contribution rates until 2009 were also documented in the Sickness Fund Factbook Müller and Lange [35]. After 2009 the contribution rate was fixed by law as explained earlier. For the first half of 2009 the fixed contribution rate was 15.5%. In the second half of 2009 and 2010 the rate was lowered to 14.9% to stabilize the German economy. Since the beginning of 2011 it is again set at 15.5%. Since 2009 several sickness funds were taking an add-on premium or paying a rebate. Table 3 gives an overview. For comparison between nominal add-on premiums or rebates and income-dependent contribution rates, the contribution rates have to be converted into Euro values. To achieve that, mean income figures are used. These figures are calculated annually, based on statistical data of the Federal Statistical Office (Statistisches Bundesamt) and are published in the Social Insurance Code Book V, Appendix 1. The values for the relevant years between 2003 and 2011 can be found in Table 4. Table 5 shows descriptive statistics. For the yearly contribution CONTst we used only the share of the contribution rate paid by the individual, since the contribution is shared between employee and employer as explained earlier.

50.67 (100.0%) 0.03 (0.1%) 0.60 (1.2%) 50.04 (98.8%)

2013

4

The objective of this empirical analysis is to estimate price elasticities before and after reform of the sickness fund contribution system. The question was to be answered whether the introduction of nominal add-on premiums or rebates increased price elasticity. It is to be assumed that negative price elasticity causes a shift of market share from more expensive to cheaper sickness funds since the introduction of freedom of choice in 1997. Therefore the general model in Eq. (1) is used to estimate the relationship between price and change in market share from year t − 1 to t: Mst = ˇ0 + ˇ1 Pst + ˇ2 Ms,t−1 +

8

k=3

where Mst =

ˇk Tks +

18

ˇl Ylt + εst

(1)

l=9

Mst − Ms,t−1 Ms,t−1

Mst represents the market share of sickness funds on the 1st of January (record date) of year t. The value is calculated,

3 While the agricultural sickness funds (Landwirtschaftliche Krankenkassen, LKK) would normally be defined as part of the statutory health insurance, in this paper, they are excluded. The LKKs are open only to farmers, their families and their workers and differ in many ways from the other sickness funds. Their contribution calculation is different and the LKKs do not participate in the risk adjustment scheme but are highly subsidized by the Federal Ministry of Food, Agriculture and Consumer Protection.


G Model

ARTICLE IN PRESS



5

Table 3 Sickness funds with add-on premiums or rebate payments between 2009 and 2013. Sickness fund

Begin of add-on/rebate

Monthly add-on premium (in Euro)

Deutsche Angestellten Kasse (DAK) Kaufmännische Krankenkasse (KKH) BKK Gesundheit Deutsche BKK NOVITAS BKK CITYBKK BKK für Heilberufe

1-Feb-2010 1-Mar-2010 1-Feb-2010 1-Feb-2010 1-Jun-2010 1-Apr-2010 1-Jan-2010 1-Jan-2011 1-Jan-2011 1-Jan-2010 1-Jul-2011 1-Jul-2009 1-Jan-2010 1-Apr-2010 1-Apr-2010 1-Jan-2010 1-Jan-2010 1-Mar-2010 1-Jan-2011 1-Jan-2009 1-Jan-2009 1-Jan-2009 1-Jan-2011 1-Jan-2010 1-Jan-2011 1-Jan-2011 1-Jan-2011 1-Jan-2012 1-Jan-2011 1-Jan-2009 1-Jan-2011 1-Jan-2011

8.00 8.00 8.00 8.00 8.00 8.00 1%b 10.00 15.00 8.00 6.50 8.00 1%b 8.00 8.00 8.00 8.00 8.00 8.00

BKK Hoesch advitaBKK GBK Köln BKK Merck ESSO BKK BKK Axel Springer BKK PHOENIX E.ON BKK BKK Publik G&V BKK IKK Südwest hkk BKK Gildemeister Seidensticker BKK ALP plus BKK A.T.U. BKK Verbund Plus BKK Wirtschaft & Finanzen BKK Schwarzwald-Baar-Heuberg BKK Würth BKK Textilgruppe Hof BKK Groz-Beckert

Yearly rebate (in Euro)

End of add-on/rebate

Gain/loss of memberships in first year of add-on premium/rebate

31-Mar-2012 28-Feb-2012 31-Mar-2012c 30-Sep-2012 31-Dec-2010 30-Jun-2011a

−6.8% −9.0% −21.7% −12.4% 32.92%d −7.4% −45.6%

31-Dec-2011a 30-Sep-2012

−1.5% −10.2%

31-Mar-2012 −6.9% 31-Dec-2010 31-May-2012 31-Dec-2010 31-Mar-2012c 31-Mar-2012 30-May-2011 31-Dec-2011 72.00 100.00 60.00 60.00 50.00 30.00 60.00 60.00 72.00 60.00 120.00 60.00 85.00

31-Dec-2009 31-Dec-2011 31-Dec-2010 31-Dec-2011

31-Dec-2011 31-Dec-2011

1.0% −12.9% −2.4% −8.4% 1.1% −11.1% −2.3% 0.7% 0.9% 0.7% 8.1% 4.3% 0.8% 61.5% n.s. 0.5% 2.0% 3.0%

Source: Sickness funds’ statements. a Sickness fund was closed. b Until 2011 sickness fund had to charge an income dependent of max. 1% if they needed more funds then an 8 Euro monthly charge could raised. Since then higher absolute add-on premiums of more than 8 Euro a month are allowed. c Continued as DAK after January, 1st 2012. d Sickness fund with merger in the relevant year.

Table 4 Yearly gross income average in Germany (2003–2012), in Euro. Year

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

Average income

28,938

29,060

29,202

29,494

29,951

30,625

30,506

31,144

32,100

33,002

Source: Social Insurance Code Book V, Appendix 1.

Table 5 Descriptive statistics of sickness funds included in the calculation.

2004–2009 Mst , relative change of market share Mst , market share CONTst , yearly contribution, in Euro N 2010–2013 Mst , relative change of market share Mst , market share ADDONst , yearly sum of add-on premiums, in Euro REBATEst , yearly rebate payments, in Euro N

Mean

Standard deviation

Min.

Max.

−0.0003% 0.598% 2096.07

0.069% 1.518% 136.97

−0.875% 0.001% 1678.40

0.945% 11.595% 2587.81

0.076% 2.101% 23.72 11.94

−0.654% 0.001% 0.00 −100.00

0.683% 13.133% 311.44 0.00

970 0.006% 0.920% 4.52 −2.21 430


G Model

ARTICLE IN PRESS



6

based on all members in the SHI in t except members of the agricultural sickness funds (see Section 4.1 for more details). Pst is the contribution paid by the member in the time between record date of t − 1 to record date of t. Tks are dummy variables for the five types of sickness funds while Ylt are the year dummies. Finally εst is the error term. The data set is split into a part before the reform covering the years 2004–2009 and after the reform covering the years 2010–2013. The models and the subsequent elasticity estimates are calculated separately. Since the representation of price Pst varies in both data splits the general model was adapted for both data splits. For 2004–2009 the following model is used:

8

ˇk Tks +

k=3

15

ˇl Ylt + εst

(2)

l=9

The price is only represented by CONTst the sickness funds’ contribution rate (as reported on the record date in t − 1) expressed in its Euro equivalent. Mst , Tks , Ylt , and εst are defined as above. For 2010–2013 the general model is modified as follows: Mst = ˇ0 + ˇ1 ADDONst + ˇ2 REBATEst + ˇ3 Ms,t−1 +

9

ˇk Tks +

k=4

12

ˇl Ylt + εst

(3)

l=10

In this model the price is represented by the sickness funds’ add-on premium ADDONst and its rebate payment REBATEst . These values are expressed in Euro or are zero in case no add-on premium is charged or rebate is paid. Again Mst , Tks , Ylt , and εst are defined as above. The models in Eqs. (2) and (3) are calculated by means of OLS for the years before and after the reform respectively. Several variations of the above model were considered. For instance different representations for the dependent variable, such as the simple market share Mst, a natural logarithm of the market share, a simple change in market share Mst − Ms , t − 1 or a relative increase of the market share were used alternatively. As expected, these alternatives did not deliver a statistically improved result. Thus Mst as defined for Eq. (1) was used in the preferred model. This is moreover also well in line with Douven et al. [20] who found similar results when testing different models against each other within the Dutch health insurance market. In addition, different variations of the explanatory variable Pst were tested, namely the difference between the fund’s price and the mean price or the price change to the preceding year. Again, these alternatives did not deliver better statistical results. Price elasticities were estimated by using the following equation in all cases: PM =

P ım P = ˇ M ıp M

4. Results 4.1. Regression results

Mst = ˇ0 + ˇ1 CONTst + ˇ2 Ms,t−1 +

where PM is the price elasticity, P is price at which a sickness fund has the market share M. ˇ represents the correlation coefficient calculated in the above models. Obviously the price elasticity is different for every P-M-combination. For the estimation of the mean price elasticity, the mean price and market share for the corresponding data split was used. These calculations are based solely on the contribution paid and rebate received by the individual. Thus the estimated price elasticity represents the individualperspective.

(4)

Table 6 shows the regression results for data splits before and after the reform. Column 1 gives the results for all sickness funds in the years 2004–2009 before the reform came into effect, while column 2 shows the results for years 2010–2013, after the reform. As expected, a negative correlation was found between the price and the change in market share in all cases and which proved to be significant for the contribution rate (2004–2009) and the add-on premium (2010–2013) settings. The rebate (2010–2013) is also negatively correlated with change in market share,4 although not significant. The previous year’s market share Mt − 1 is, likewise, significantly correlated with change in market share; whilst for pre-reform years the relationship is negative with increasing market share and positive for post-reform years. In 2004–2010 all sickness fund dummies are significant with a positive effect. For 2010–2013, however, no sickness fund dummy has a significant effect. Calculations were tested for multicollinearity, autocorrelation and heteroscedasticity. Results can be found in the appendix. The assumptions for OLS were met with some expected minor multi correlation between market share and sickness fund type dummies. It was found, that the distribution of the resulting errors was not normal, when performing the Shapiro–Wilk test. This assumption is important to justify tests on statistical significance when working with small sample sizes. For larger sample sizes, statistical theory has shown that the assumption is of less concern [45]. Therefore working with (almost) the entire sickness fund population, biased results are particularly unlikely. 4.2. Elasticity estimates The correlation coefficients allow deriving price elasticities for the two data splits using Eq. (4). The results are presented in Table 7. The price elasticities stated here are all calculated from the individual’s perspective taking into account the individual’s share of the contribution rate, the add-on premium and the rebate. Since the contribution rate

4 To ensure the same direction in all price variables, negative premium rates were used for the regression when rebates were paid by the sickness funds.


G Model

ARTICLE IN PRESS



7

Table 6 Regression results. 2004–2009 Coefficient CONT ADDON REBATE Mt−1 Sickness fund type dummies AOK EK IKK BKK KBS

Coefficient

Standard error

−12.548 × 10−6 * −0.746 × 10−6 0.014*

(1.383 × 10−6 ) (2.812 × 10−6 ) (0.002)

(0.082 × 10−3 ) (0.105 × 10−3 ) (0.067 × 10−3 )

−0.177 × 10−3 −0.105 × 10−3 0.006 × 10−3

(0.119 × 10−3 ) (0.203 × 10−3 ) (0.133 × 10−3 )

1.09 × 10−3 *

(0.366 × 10−3 )

−0.050 × 10−3

(0.340 × 10−3 )

−0.309 × 10−3 * −0.206 × 10−3 *

(0.073 × 10−3 ) (0.069 × 10−3 )

−0.137 × 10−3 −0.062 × 10−3

(0.091 × 10−3 ) (0.092 × 10−3 )

(0.204 × 10−6 )

−0.012*

(0.002)

a

a

a

0.037 × 10−3 0.224 × 10−3 * 0.3830 × 10−3 *

a

−0.093 × 10−3

Adj. R2 N

0.186 970

(0.070 × 10−3 )

−0.263 × 10−3 *

(0.113 × 10−3 )

0.224 430

Statistically significant at the p < 0.05 level. Excluded during the calculation.

Table 7 Price elasticity estimations.

P M ˇCONT ˇADDON ˇREBATE PM , for CONTRIBUTION PM , for ADD-ON PM , for REBATE *

(0.069 × 10−3 ) (0.071 × 10−3 ) (0.076 × 10−3 )

−0.128 × 10−3

Other dummies Sickness fund involved in merger

*

Standard error

−2.296 × 10−6 *

0.268 × 10−3 * 0.540 × 10−3 * 0.6395 × 10−3 *

Year dummies 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

a

2010–2013

2004–2009

2010–2012

2096.07 0.598% −2.296 × 10−6

2587.92 0.920% −12.548 × 10−6 * −0.746 × 10−6

−0.81 −3.53 −0.21

Statistically significant at the p < 0.05 level.

is deducted directly from payroll, the price elasticities are not out-of-pocket. In the pre-reform years the estimation for the overall price elasticity is −0.81. As expected, a distinct higher addon premium price elasticity for the post-reform years was calculated. The estimated elasticity is −3.53, which is more than four times higher than before the introduction of addon premiums. 5. Discussion and conclusion The German health care reform of 2009 reflects the recent trend of increasing state regulation of healthcare financing in Germany [46]. It aimed to improve fair and managed competition between sickness funds in the SHI. By introducing nominal add-on premiums and rebates the government expected an increase in competition through

quality and service. Until today, however, the competition focuses solely on price. Quality and innovation fall behind in order to avoid or minimize an add-on premium. This study measures the change in price elasticity for all sickness funds in Germany between 2004 and 2013. The data, covering almost all sickness funds, suggests a more than four times higher price elasticity after the reform compared to before the reform. Rebate payments, as the factual opposite of add-on premium, in contrast have a low negative effect that is not significant. A change of the direction of correlation of the previous year’s market share occurred from pre- to post-reform years. This could be explained by a change of preferences of the individuals. While in pre-reform years the newly opened and smaller sickness funds where more attractive, in post-reform years the bigger sickness funds, who may supposedly be providing more security in a changed scenario, were preferred. Using the membership data on the sickness fund level takes advantage of including the complete population in the price elasticity calculations, allowing for the above new findings. However, a number of limitations have to be considered when drawing conclusions from this data. First, the data does not provide any information on socio-economic aspects and their influence on price elasticity. Several studies found evidence that age and health status clearly influence price elasticity (e.g. [9,47–49]). Other less demographic and more behavioural explanations are possible. Individuals that might have experienced fund switching in the past and might therefore be more



8


likely to switch again. They might have lower transaction costs because they have experience or they are less bound to their sickness fund. Second, in the research period 2010–2013 only a few sickness funds had to charge add-on premiums or paid rebates. Furthermore, the payment values were very homogenous on the same level (e.g. 8 Euro a month as the typical add-on premium). This somewhat limits generalization of price elasticity found in this study for a health insurance market with add-on premiums charged by all sickness funds in greater variety of prices. Third, until 2009 price differences were only expressed as differences of income-dependent contributions. Therefore absolute price differences faced by the individual depended on the individual’s income. The latter is not observed in the data and had to be derived from the average income, which in itself may bias the price elasticity estimations, since considerable differences in income levels can be observed between sickness funds [50]. The general price elasticity of −0.81 before the reform, found in this study,5 is on a comparable level with price elasticities found in the US [9,10] and the Netherlands, prior to the local reform [17–19], but somewhat lower compared to Swiss price elasticities [21–23]. That seems unexpected given several factors that would lead to a higher expectation on price elasticity in Germany. Unlike for instance in the US, the coverage is universal and almost unified. Unlike the Netherlands no relevant supplementary insurance exists that interferes with insurance switching. In addition, in Germany switching the insurance never involves switching the physician or hospital, since all sickness funds are obligated to contract with all (licensed) providers and patients are free in their choice of these. Imaginable explanations are a possible unawareness of the magnitude of price differences between sickness funds since the price was expressed as percentage of income, long established relationships between members and sickness funds or an overwhelming choice set.6 The substantially higher price elasticities after the reform is well in line with earlier studies on German price elasticity that found similar high price elasticities of about −4 [18,24] In light of this comparison, price elasticities in the early years of sickness fund choice and most recent price elasticities of add-on premiums seem to reflect a more adequate level of price competition given the above described characteristics of the German SHI system. The newly introduced free choice might have increased price awareness, which had decreased over time. Certainly the add-on premiums were contributive to again create the awareness and therefore providing a desirable stimulus for adequate price competition. Schmitz and Ziebarth stress the importance of contributions expressed in nominal terms [26]. The effects of the approximately same price elasticities in early years of sickness fund choice (1996–2000)

5 And as matter of fact the price elasticity found by Schmitz and Ziehbarth [26] too. 6 Research finds that too much choice decreases the decision quality [51,52].

and in years following the reform were, however, largely different. Between 1996 and 2000 contribution rates differed between the entire range of sickness funds whereas add-on premiums were only charged by very few sickness funds. Therefore the add-on premium price elasticity originates only from a fraction of the market. Even though the price elasticities were similar, the consequences for the affected sickness funds were highly different. As a result of these consequences sickness fund management focused on avoiding add-on premiums at all cost [4,53]. Investment in (optional) selective contracts, that would have allowed some degree of quality competition, ceased when the initial financing ran out [54]. Even so the price elasticity seems adequate in the above comparison, price competition is now in fact overrated in relation to quality competition. Thus true “value-formoney” competition in Enthovens terms is inhibited. This results in exactly the opposite of the original intention of lawmakers to foster quality competition. Furthermore the use of price elasticity as an indicator of the price intensity witnessed in a market might not be a sufficient tool here. Only few sickness funds with add-on premiums explained the price elasticity effects for the entire market between 2009 and 2013 in contrast to the time before 2009 when contributes rates were more broadly distributed between sickness funds. This might limited the comparability of the results of this study with difference health insurance markets. Furthermore, the great number of individuals leaving sickness funds that have add-on premiums, overemphasis the exit-option of the Hirschmann terminology of exit, voice and loyalty [55,56]. Especially the two sickness funds that were forced to close rapidly after massive losses in enrollees could not improve through means of voice of the sickness funds self-administration in the short period of time. This has to be seen critically in the context of already existing negligence of the voice option by policymakers [10]. In conclusion, the positive effects of the introduction of nominal add-on premiums, namely increased price awareness and a more adequate price elasticity for the German SHI market, are counteracted by non-desired effects due to a limitation of add-on premiums to a small group of sickness funds. To overcome these negative effects add-on premiums should be introduced to all or almost all sickness funds. This could be achieved by lowering the fixed contribution rate. Presumably, the likelihood of rebates to occur would disappear that way, which should be of no further problem since this study found no significant effect of rebates on market share changes. The proposed change would probably lead to reasonable price awareness and price competition whilst providing the possibility of more quality competition when add-on premiums are common and not “in the spotlight” with extreme effects on market share. As a result of the good economic climate and a rather high fixed contribution rate, no sickness fund currently has to charge an add-on premium. Price differentiation is therefore limited to paying a rebate, which was found to have no significant impact on sickness fund switching in this study. However, current government already abolished




the nominal add-on premium and rebate and passed new legislation allowing the sickness funds to instead charge an income-dependent add-on premium that is again automatically deducted via payroll. If the assumptions of this discussion prove right, it is likely that price awareness will decrease soon and a lower than adequate price elasticity will return. Acknowledgements We wish to thank Christina Samel for help, comments and suggestions on the statistical methods (and their documentation) used in this paper. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.healthpol.2015.01.014. References [1] Federal Ministry of Health. Die Gesundheitsreform ist in Kraft getreten; 2007. Available at: http://bmg.bund.de/ministerium/ aufgaben-und-organisation/das-ministerium-in-der-gesetzgebung. html [accessed 11.08.14]. [2] Paquet R, Stein M. GKV-Mitglieder: preissensibel und leistungsbewusst. Die BKK 2008;(5):274–80. [3] Igel C, Schaufler TM. Der neue Zusatzbeitrag als Wettbewerbsparameter: Strategische Wettbewerbsoptionen der Krankenkassen nach der Gesundheitsreform 2006 – eine positive Analyse. Beiträge zur Gesellschaftspolitik 2006;(9):2–8. [4] Staeck F. Kein Feiertag für die Patienten, Standpunkte zu Kassenfinanzen; 2012. Available at: http://www.aerztezeitung.de/ politik gesellschaft/krankenkassen/article/807204/standpunktkassenfinanzen-kein-feiertag-patienten.html [accessed 11.08.14]. [5] Laschet H. Flaute bei der Integrations-Versorgung 2010;(10):1–8. [6] Enthoven AC. The history and principles of managed competition. Health Affairs 1993;12(Suppl. 1):24–48. [7] Scanlon DP, Chernew M, Lave JR. Consumer health plan choice: current knowledge and future directions. Annual Review of Public Health 1997;18(1):507–28. [8] Buchmueller TC. Consumer-oriented health care reform strategies: a review of the evidence on managed competition and consumerdirected health insurance. Milbank Quarterly 2009;87(4):820–41. [9] Royalty AB, Solomon N. Health plan choice: price elasticities in a managed competition setting. Journal of Human Resources 1999;34(1):1–41. [10] Haarmann A, Klenk T, Weyrauch P. Exit, choice – and what about voice? Public Management Review 2010;12(2):213–31. [11] Carlin C, Town RJ. Adverse selection, welfare and optimal pricing of employer-sponsored health plans. University of Minnesota Working Paper; 2008. [12] Dowd BE, Feldman R. Premium elasticities of health plan choice. Inquiry 1994/1995;31(4):438–44. [13] Scanlon DP, Chernew M, McLaughlin CG, Solon G. The impact of health plan report cards on managed care enrollment. Journal of Health Economics 2002;21(1):19–41. [14] Wille E. Privatrechtliche Organisation der gesetzlichen Krankenkassen: Reformperspektiven für Deutschland, Erfahrungen aus den Niederlanden. 1. Aufl. Baden-Baden: Nomos, vol. 37. Beiträge zum Gesundheitsmanagement; 2012. [15] Paris V, Wei L, Devaux M. Health systems institutional characteristics: a survey of 29 OECD countries. OECD Health Working Papers 50; 2010. [16] Wendt C. Changing healthcare system types. Social Policy & Administration 2014;48(7):864–82. [17] Schut FT, Hassink WHJ. Managed competition and consumer price sensitivity in social health insurance. Journal of Health Economics 2002;21(6):1009–29. [18] Schut FT, Greß S, Wasem J. Consumer price sensitivity and social health insurer choice in Germany and the Netherlands. International Journal of Health Care Finance and Economics 2003;3(2):117–38.

9

[19] van Dijk M, Pomp M, Douven RC, Laske-Aldershof T, Schut E, Boer Wde, et al. Consumer price sensitivity in Dutch health insurance. International Journal of Health Care Finance and Economics 2008;8(4):225–44. [20] Douven RC, Ligthart M, Lieverdink H, Vermeulen I. Measuring annual price elasticities in Dutch health insurance: a new method. CPB Netherlands Bureau for Economic Policy Analysis; 2007. [21] Beck K. Risiko Krankenversicherung: Risikomanagement in einem regulierten Krankenversicherungsmarkt. Bern: Haupt; 2004. [22] Rütschi C. Health and health insurance in Switzerland: an empirical investigation. Berlin: dissertation.de; 2008. [23] Diserens D. Die Prämienbudgetierung – ein Kernprozess von Krankenversicherungen; 2002. Available at: http://www.dimpro.ch/ pdf/sozialversicherungen/SV-4-Pr%C3%A4mienkalkulation-2002.pdf [accessed 11.08.14]. [24] Schwarze J, Andersen HH. Kassenwechsel in der Gesetzlichen Krankenversicherung: Welche Rolle spielt der Beitragssatz; 2001 [Diskussionspapier Nr. 267]. [25] Tamm M, Tauchmann H, Wasem J, Greß S. Elasticities of market shares and social health insurance choice in Germany: a dynamic panel data approach. Health Economics 2007;16(3):243–56. [26] Schmitz H, Ziebarth NR. In absolute or relative terms? How framing prices affects the consumer price sensitivity of health plan choice, vol. 304. Essen: RWI; 2011 (Ruhr economic papers). Available at: http:// www.rwi-essen.de/publikationen/ruhr-economic-papers/425/ [27] Gaßner M, Otto F, Ludwig E. Analyse der Mitgliederentwicklung sowie weiterer Kennzahlen der bundesunmittelbaren Krankenkassen ohne und mit Zusatzbeitrag bzw, Prämie. Gesundheits- und Sozialpolitik 2012;5:52–6. [28] Ellis RP, Savage E. Run for cover now or later? The impact of premiums, threats and deadlines on private health insurance in Australia. International Journal of Health Care Finance and Economics 2008;8(4):257–77. [29] Finkelstein A. The effect of tax subsidies to employer-provided supplementary health insurance: evidence from Canada. Journal of Public Economics 2002;84(3):305–39. [30] Gerrits KP. Estimating demand for private health insurance: lessons for Alberta. ProQuest; 2005. [31] Costa J, García J. Demand for private health insurance: how important is the quality gap? Health Economics 2003;12(7):587–99. [32] Fernandez JG. Three essays on competition and health insurance markets. Diss. Boston University; 2012. [33] Verband der Privaten Krankenversicherung e.V. Zahlenbericht der Privaten Krankenversicherung 2011/2012; 2012. Available https://bestellungen.pkv.de/w/files/shop zahlenberichte/ at: zahlenbericht 2011 2012.pdf [accessed 11.08.14]. [34] Greß S, Niebuhr D, Rothgang H, Wasem J. Benefit Decisions in German social health insurance. In: Jost TS, editor. Health care coverage determinations: an international comparative study. Maidenhead: McGraw Hill; 2005. p. 115–31. [35] Müller A, Lange WG. DfG-Krankenkassenatlas [Grundwerk]. Berlin: MC. B, Verlag für Gesellschaftspolitik; 2008. [36] Lange WG. DfG-Ranking: Liste der deutschen Krankenkassen [Ranking nach Mitgliederzahlen, Stichtag 1. Oktober 2010]. Beiträge zur Gesellschaftspolitik 2010;9:1–7. [37] Lange WG. DfG-Ranking: Liste der deutschen Krankenkassen [Ranking nach Mitgliederzahlen, Stichtag 1. Juli 2011]. Beiträge zur Gesellschaftspolitik 2011;8:1–7. [38] Lange WG. DfG-Ranking: Liste der deutschen Krankenkassen [Ranking nach Mitgliederzahlen, Stichtag 1. Oktober 2012]. Beiträge zur Gesellschaftspolitik 2012;5:1–6. [39] Federal Ministry of Health. Gesetzliche Krankenversicherung. Mitglieder, mitversicherte Angehörige und Krankenstand Jahresdurchschnitte 1998 bis 2008; 2009. Available at: http://bmg. bund.de/fileadmin/redaktion/pdf statistiken/krankenversicherung/ KM1JD2004-pdf-5112.pdf [accessed 11.08.14]. [40] Federal Ministry of Health. Gesetzliche Krankenversicherung. Mitglieder, mitversicherte Angehörige und Krankenstand Jahresdurchschnitt 2009; 2010. Available at: http://bmg.bund.de/ fileadmin/redaktion/pdf statistiken/krankenversicherung/ Mitglieder mitversicherte Angehoerige und Krankenstand 2009.pdf [accessed 11.08.14]. [41] Federal Ministry of Health. Gesetzliche Krankenversicherung. Mitglieder, mitversicherte Angehörige und Krankenstand Jahresdurchschnitt 2010; 2011. Available at: http://bmg.bund.de/ fileadmin/dateien/Downloads/Statistiken/GKV/Mitglieder Versicherte/KM1 Jahresdurchschnitt 2010.pdf [accessed 11.08.14]. [42] Federal Ministry of Health. Gesetzliche Krankenversicherung. Mitglieder, mitversicherte Angehörige und Krankenstand



10

[43]

[44]

[45]

[46]

[47]

[48]


Jahresdurchschnitt 2011; 2012. Available at: http://bmg.bund.de/ fileadmin/dateien/Downloads/Statistiken/GKV/Mitglieder Versicherte/KM1 Jahresdurchschnitt 2011.pdf [accessed 11.08.14]. Federal Ministry of Health. Gesetzliche Krankenversicherung. Mitglieder, mitversicherte Angehörige und Krankenstand Jahresdurchschnitt 2012; 2013. Available at: http://bmg.bund.de/ fileadmin/dateien/Downloads/Statistiken/GKV/Mitglieder Versicherte/KM1 JD 2012.pdf [accessed 11.08.14]. Federal Ministry of Health. Gesetzliche Krankenversicherung. Mitglieder, mitversicherte Angehörige und Krankenstand Jahresdurchschnitt 2013; 2014. Available at: http://www.bmg.bund.de/ fileadmin/dateien/Downloads/Statistiken/GKV/Mitglieder Versicherte/KM1 JD 2013.pdf [accessed 11.08.14]. Berry WD. Understanding regression assumptions. Quantitative applications in the social sciences. [Nachdr.], vol. 92. Newbury Park, CA: Sage Publication; 2006. Wendt C, Agartan TI, Kaminska ME. Social health insurance without corporate actors: changes in self-regulation in Germany, Poland and Turkey. Social Science & Medicine 2013;86:88–95. Strombom BA, Buchmueller TC, Feldstein PJ. Switching costs, price sensitivity and health plan choice. Journal of Health Economics 2002;21(1):89–116. van den Berg B, van Dommelen P, Stam P, Laske-Aldershof T, Buchmueller TC, Schut FT. Preferences and choices for care and health insurance. Social Science & Medicine 2008;66(12):2448–59.

[49] Andersen HH, Schwarze J. GKV ’97: Kommt Bewegung in die Landschaft? Eine empirische Analyse der Kassenwahlentscheidungen. Arbeit und Sozialpolitik 1998;(9/10):11–23. [50] Schnee M. Sozioökonomische Strukturen und Morbidität in den gesetzlichen Krankenkassen; 2008. [51] DiCenzo J, Fronstin P. Lessons from the evolution of 401(k) retirement plans for increased consumerism in health care: an application of behavioral research. EBRI Issue Brief 320; 2008. [52] Schram A, Sonnemans J. How individuals choose health insurance: an experimental analysis. European Economic Review 2011;55(6):799–819. Available at: http://linkinghub.elsevier. com/retrieve/pii/S0014292111000031 [53] Jacobs K. Vorfahrt für Selektivverträge: Erforderlich ist ein konsistenter Ordnungsrahmen – und mehr Mut. IMPLICONplus 2012;11:1–12. [54] Amelung V, Hildebrandt H, Wolf S. Integrated care in Germany – a stony but necessary road! International Journal of Integrated Care 2012;12:e16. [55] Hirschman AO. Exit, voice, and loyalty: responses to decline in firms, organizations, and states. Cambridge, MA: Harvard University Press; 1970. [56] Braun B, Greß S, Rothgang H, Wasem J, editors. Einfluss nehmen oder aussteigen: Theorie und Praxis von Kassenwechsel und Selbstverwaltung in der Gesetzlichen Krankenversicherung. Berlin: Edition Sigma; 2008.


The political economics of health insurance market reform.

Use of health care services by people with mental illness: secondary data from three statutory health insurers and the German Statutory Pension Insurance Scheme.

[Work place health promotion programmes of the statutory German Pension Insurance].

Insurance cancellations in context: stability of coverage in the nongroup market prior to health reform.

Rates of insurance for injured patients before and after health care reform in Massachusetts: a possible case of double jeopardy.

Health care reform after the ACA.

The prospects for national health insurance reform.

The impact of health insurance reform on insurance instability.

Problems with employer-provided health insurance--the Employee Retirement Income Security Act and health care reform.

The Children's Health Insurance Program Lessons for Health Reform.

Health reform and changes in health insurance coverage in 2014.

Affordability of health care under publicly subsidized insurance after Massachusetts health care reform: a qualitative study of safety net patients.

Mandated Health Insurance and Provider Reimbursement via Private Insurance: Evidence From the Massachusetts Health Reform.

Social inequalities in the utilization of outpatient psychotherapy: analyses of registry data from German statutory health insurance.

Improvements in health status after Massachusetts health care reform.

Switching gains and health plan price elasticities: 20 years of managed competition reforms in The Netherlands.

Comparison of Chinese inpatients with different types of medical insurance before and after the 2009 healthcare reform.

Experiences applying for and understanding health insurance under Massachusetts health care reform.

Health insurance and the American public sector labor market.

Patient turnover and nursing employment in Massachusetts hospitals before and after health insurance reform: implications for the Patient Protection and Affordable Care Act.

Mandate-based health reform and the labor market: Evidence from the Massachusetts reform.

Health care reform and the transcultural nurse.

Health Care Reform and the American Congress.

Trauma Patients: Health Insurance Reform Is Only the Beginning.