Health Policy 102 (2011) 247–254
Contents lists available at ScienceDirect
Health Policy journal homepage: www.elsevier.com/locate/healthpol
Switching sickness funds in Israel: Adverse selection or risk selection? Some insights from the analysis of the relative costs of switchers Amir Shmueli ∗ Department of Health Management, The Hebrew University School of Public Health, POB 12272, Jerusalem 91120, Israel
a r t i c l e
i n f o
JEL classification: I1 Keywords: Switching Sickness funds Israel Adverse selection Risk selection
a b s t r a c t This paper uses medical care costs of joiners in their first year and of leavers in their last year prior the move, relative to the age–sex groups’ means, to examine the mechanisms behind the switching decisions. Since under the Israeli National Health Insurance Scheme no premiums are paid by the enrollees directly to the sickness funds, the paper focuses on the distinction between demand-side-adverse-selection type and supply-side-risk-selection type of reasons for switching. The latter is particularly important because of the incomplete Israeli age-based risk-adjustment scheme. The findings indicate that leavers are less costly than average, and thus their leaving cannot be attributed to dumping or restricted care. Joiners are more costly than average in younger ages and less costly than average in advanced age. A particular group of young joiners seems to consist of women looking for pre- and/or post-natal care. The current generous capitation rate for children provides future compensation for this first year loss. © 2011 Elsevier Ireland Ltd. All rights reserved.
1. Introduction On January 1st 1995, a new National Health Insurance Law was enacted in Israel. The Law defined a uniform package of benefits to be provided by the four competing sickness funds operating in Israel. It envisioned a “managed competition” scheme, where the sickness funds compete on the quality of the services included in the package of benefits as well as on some “extras” to attract consumers (e.g., on-line services, provision of medicines not included in the package of benefits, or participation in disease management programs). Unlike the Dutch or the German systems, Israelis do not pay member fees directly to their sickness funds (apart from co-payments), so competition on premiums is not relevant. Under the Law, healthcare is universal, and no (explicit) risk selection is allowed. Furthermore, open enrollment with a bi-annual switching option is specified.
∗ Tel.: +972 02 675 8514; fax: +972 02 643 5083. E-mail address:
[email protected] 0168-8510/$ – see front matter © 2011 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.healthpol.2011.07.008
Since 1998 the sickness funds are allowed to offer voluntary community-rated supplementary insurance policies to their members, which cover acute care services not covered by the package of benefits (e.g., alternative medicine, more physical therapy sessions, medicines not in the package, choice of surgeon etc.). These policies have become a major subject on which sickness funds compete. During the first three years of the Law, about 4% annually switched sickness funds (switching does not include newborns, deaths, and new immigrants). This relatively high (in Israeli standards, see below) figure reflected two factors: first, the sickness funds began aggressive marketing campaigns to sign members in (e.g., in malls or bus stations). A considerable fraction of those signed-in cancelled afterwards their commitment, so that the figures for those years are not reliable. Second, the Law cancelled effective restrictions on switching. Before 1995, two sickness funds exercised effective and explicit risk selection, and a third sickness fund was the only option available to certain parts of the population because of employers’ socio-political affiliation or geographical location. All those who had wished to switch but could not do that previously,
248
A. Shmueli / Health Policy 102 (2011) 247–254
materialized their wish during the first years after 1995. Since 1997, the intention to switch sickness fund must be registered in a Post Authority office, switching dates are 1/1 and 1/7, and the annual rate of switchers has been quite fixed around 1%. Since 2006, there are four switching dates and the annual switching rate reached 1.6%. The reasons for the low switching rates in Israel are discussed, in a five countries comparative context, by Laske-Aldershof et al. [11]. The conclusions are that, abstracting from reasons commonly associated with inertia and reservation from changes, the limited choice options reduce the net benefits of switching, leading to low mobility. And indeed, all four sickness funds operate as managed care organizations, but differ in the way the care is managed. The biggest sickness fund, with 55% market share in November 2009, operates like a staff-model HMO, owns eight general hospitals, and most of the primary and secondary care is provided by salaried physicians in the fund’s clinics. The small funds (market shares 24%, 12% and 10%) have no hospitals, and contract selectively with hospitals, with independent physicians and institutions for acute care. In the year 2009, the age-standardized mean cost across the sickness funds ranged between −5% and +5% the overall mean (the overall mean was 3276 NIS, or about 750 USD. No data is available on the mean cost by age groups). While it might be true that an annual rate of 1–2% switchers is low, an important issue is who are these switchers and under which circumstances these moves are done. The overall average of 1% mobility would have totally different interpretations and social implications if they were the top sickest 1% or the top healthiest 1%, and if these switches were initiated by the enrollees or caused by implicit selection exercised by the sickness funds. Implicit selection (dumping and skimping) is actually quite expected to be exercised by the sickness funds in light of the very incomplete risk-adjustment scheme which is based on age only. Because of the much skewed distributions of health and medical expenditure, a 1% of the population might concentrate much larger shares of health or medical care costs with significant implications. Shmueli et al. [17] identified persons with greater likelihood to switch sickness funds, based on their personal characteristics available in the National Insurance Institute’s (NII) data base: age, gender, employment status, wage, labor income, and the types of allowances received. The findings indicate that young persons are more likely to switch, and that switching is an inferior good, with persons with lower incomes, receiving income maintenance or unemployment benefits being more likely to switch sickness funds, controlling for age and gender. Since sickness funds compete neither on premiums nor on co-payments, which are expected to have a strong income effect, the interpretation of the findings focused on the effect on switching behavior of two personal correlates of income in Israel, number of children and health (see below for the effect of supplementary insurance ownership, however). Since the risk-adjusted payment to the sickness funds for children is particularly generous, children constitute a source of profit. Consequently, the sickness funds have tried to attract large families. In Israel, large families are typically families of Arabs or of Orthodox Jews, and
are also relatively poor. These families have switched sickness funds following massive marketing campaigns (such as “children are our baby”), contributing to the observed negative effect of income on switching. On the other hand, because of the correlation between low income and poor health, the higher switching likelihood of poor persons might have been actually caused by health-related factors, such as low satisfaction, low quality care, or other implicit risk-selection strategies of the sickness funds. This last argument could not be examined directly, since health data or medical care costs were not available for the study. The present paper focuses directly on that argument. In particular, the Israeli sickness funds’ data on the relative medical care costs of switchers are used to explore the health-based mobility directly. Two focal variables are examined: the relative mean cost of leavers with respect to the mean cost in their age–sex group, during the year prior to their leaving, and the relative mean cost of joiners with respect to the mean cost in their age–sex group during the first year after their joining the destination sickness fund. 2. Scientific background: switching health plans Most of the research on health plan switching focused on Americans, mostly Medicare enrollees, moving between generous but expensive indemnity insurance plans and less expensive but more restrictive PPOs or HMOs [3,12,8,9,13]. Altman et al. [1] found, for example, that employees moving from indemnity plans to HMOs spent 30–36% less, adjusting for age and sex, than people who remained in the indemnity plans, indicating the existence of adverse selection. They did not find a similar indication among joiners, whose spending was only about average for the enrollees in the new plan. A recent study based on the Medical Expenditure Panel Survey [10] found that switchers from HMO to non-HMO spend more on hospitalization and use less preventive measures. Switchers from non-HMO to HMO spend less on medicines and physician visits. Consequently, it appears that non-HMO private managed plans provide better coverage on hospitalization, physicians’ visits and prescribed medicine than the HMOs, attracting thus sick persons. Van Vliet [20] found that Dutch switchers are “good risks”, having medical expenditures which were around 40% below average. However, that difference was explained by their young age and good health, so that the risk-adjusted payments for them nearly equaled those made for stayers. If, however, only age and gender are used as risk-adjusters, joiners, if they maintain their pre-move behavior, will bring in around 18% of the mean expenditure as profit. Differences in coverage and costs translate into premium differentials, and are related to differences in quality of care. Some research has studied the price sensitivity of switching and the effect of quality information on health plan choice in the US and in Europe [15,18,2]. The price elasticity of switching is not relevant in the Israeli scene since enrollees do not pay direct premium to their sickness funds (see below for the effect of the supplementary insurance offered by the sickness funds). Formal and official comparative indicators of quality of care are not available, and most of the information needed by consumers to make switching
A. Shmueli / Health Policy 102 (2011) 247–254
decisions is provided by the sickness funds’ advertisements and informally through social networks. The possible reasons for switching in The Netherlands, Germany, Switzerland, Israel and Belgium are discussed by Laske-Aldershof et al. [11]. The conclusions are that, abstracting from reasons commonly associated with inertia and reservation from changes, the amount of choice options shapes the net benefits of switching and affects mobility. Recently, several authors showed the effect of the supplementary health insurance market on selection and switching in the basic insurance market [14,5].
3. Data and methods Individual medical care cost data is private information of the sickness funds, and is not provided to researchers. This is particularly true for the medical care costs of those who switch sickness funds. For that reason, the data used in this analysis is quite limited. The sickness funds agreed to calculate, for each age and gender group, the ratio of the mean annual medical cost among leavers to the age–gender average during their last year of membership, and the ratio of the mean annual medical cost among joiners to the age–gender average during their first year of membership. These will be referred to as the relative cost of leavers and joiners, respectively. There were nine age groups, as defined by the national risk-adjustment scheme. The cost included the cost of hospitalizations and outpatient visits, medicines, laboratory tests and diagnostics, and visits to specialists outside hospitals. Primary care cost was not included due to administrative difficulties to monitor this service; however, this cost is relatively small. The purchasing cost was taken when the service was purchased. When the service was provided internally (e.g., when hospitals are owned by the sickness fund), the internal computed price used by the sickness fund was considered. Although the data consists of means only, some approximate statistical inference is still possible by assuming specific relationship between the mean and the standard deviation of the costs in any group of enrollees. We assume that s = cm, where m is the mean cost and s is the standard deviation, and c is a constant. The mostly cited value for c is c = 3 [19]. We will be interested in testing equality of mean cost between leavers (m1 ) and stayers (m0 ), and between joiners (m2 ) and veteran enrollees (m0 ) (for notation simplicity it is assumed that the benchmark groups are identical). It is quite straightforward to derive the appropriate t-tests and their critical values (see Appendix). Notice that aggregation of the relative cost ratios over the sickness funds is not possible without knowledge of the age–sex specific mean costs of the sickness funds (only if these mean costs are equal, the aggregate relative cost ratio is a weighted average of the relative cost ratios of the sickness funds, with weights being the age–sex specific relative share of each sickness fund switchers in total switchers). Since the cost information is not available, and the assumption of equal age–sex mean cost across sickness funds is unrealistic, the analysis will be conducted by sickness funds.
249
The sickness funds included in their calculations members who left and new comers who joined during (sub-periods of) 1999–2001. For confidentiality reasons, the cell-sizes cannot be reported to prevent identification of the sickness fund. These sizes range from few thousands to several hundreds. The minimal cell size was 112. A further limitation of the data is that in general, because of different reporting periods among the sickness funds, there is no exact overlap between the joiners and the leavers; namely, the joiners in the data are not necessarily the same persons as the leavers. One sickness fund could not, for technical reasons, separate between leavers and deaths. For the same sickness fund, the data on joiners was only for the total population by age and not by gender. The conclusions from the examination of the age variation in relative costs among joiners are similar to those arrived at from the data of the other three sickness funds, so that sickness fund’s data was ignored, and the analysis is based on age and gender specific relative costs of joiners and leavers of three sickness funds. 4. Theoretical considerations Interpersonal variations in observed medical costs within age groups originate from three sources: health state and demand for medical care, availability of services (quantities) and unit costs of the services (prices). We are unable to distinguish between prices and quantities, neither across geographical districts within sickness funds, nor across sickness funds. Variation in costs, for a given health state and demand, will be interpreted as variation in intensity of care, which is a decision variable of the sickness funds. However, interpersonal variations in the cost ratio (relative to the age–sex group mean in the sickness fund) depend also on the mean health state of the group and the intensity of care provided to persons in different health states. Among leavers, a relative cost greater than one indicates that the leavers are more expensive than average. Such a finding is likely to indicate, in the Israeli context, dumping or implicit risk-selection, service level selection, or quality distortion exercised by the sickness fund [7,6,4]. Such a strategy is quite expected from the sickness funds, since the risk-adjustment scheme is based on only nine age group combined with five conditions risk sharing (fixed payment for persons suffering from HIV/AIDS, Hemophilia, Gaucher, Talesemia major and persons in need of dialysis). Such leavers will increase the sickness fund’s profit. Among joiners, a relative cost greater than one indicates that the joiners are more expensive than average, in their age–sex group. While it is true that switching physicians might be associated with an increase in the intensity of care for diagnostic and familiarization reasons, such a finding is likely to reflect the search for higher intensity of care by the joiners, namely, an expression of adverse selection on the part of the consumers. The joiners constitute, in such a case, a predictable loss. Within sickness funds, if the relative cost of joiners is above one, and that of leavers is below or equal one, the less-than-average inexpensive enrollees are replaced by
250
A. Shmueli / Health Policy 102 (2011) 247–254
Table 1 Percent leaving their sickness funds by age and gender. 1999
Total 0–4 5–14 15–24 25–34 35–44 45–54 55–64 65–74 75+
2000
Total
Men
Women
Total
Men
Women
1.01 1.39 0.85 1.88 1.37 0.73 0.49 0.45 0.37 0.23
0.99 1.37 0.86 1.73 1.40 0.73 0.44 0.41 0.35 0.22
1.02 1.41 0.85 2.02 1.34 0.74 0.53 0.49 0.38 0.23
1.04 1.46 0.91 1.71 1.45 0.82 0.57 0.52 0.42 0.28
1.03 1.44 0.92 1.53 1.52 0.86 0.54 0.45 0.40 0.25
1.05 1.47 0.91 1.86 1.39 0.79 0.60 0.57 0.43 0.30
more-than-average expensive newcomers, and the overall mean cost is rising. If the relative cost of joiners is lower than that of leavers, the mean cost drops. Under a constant age–sex capitation rates, in the first case the sickness fund’s margin will drop, while in the second case the sickness fund’ profit will increase. 5. Results By way of introducing the issue, Table 1 presents the percent switching sickness fund in the years 1999 and 2000 by age and gender among all four sickness funds. The overall rate is 1.1%, which has been stable during the period since 1998 (but has increased moderately to 1.6% in 2006). The mobility rates are quite similar among men and women in most age groups. Particularly high rates are found in age groups 15–34 and 0–4. The propensity to move declines steadily with advanced age. Among the 75+ elderly, only 0.3% switched sickness fund in 2000. Although the rates differ somewhat across the four sickness funds, that pattern is common to all of them. Figs. 1 and 2 present the cost ratios for joiners and leavers by sickness fund (A–C) for women and men, respectively. All of the cost ratios among men – leavers and joiners – are statistically different than one. Among women, all cost ratios among leavers are significantly different than one, and only in 3 out of 27 joiners’ cells (ages 5–14 in sickness fund B, and 15–24 and 55–64 in A) is the cost ratio not different from one (indicated in the figure by #). The results indicate that leavers, both men and women, and in all sickness funds, were less costly to care for during their last year of membership than average. This finding leads to the rejection of the hypothesis that the sickness funds exercise implicit risk-selection or dumping strategies. It is the healthier than average, in all but 3 (in sickness fund B) age–sex groups, who left their sickness funds. Better health is associated with lower transaction cost of switching providers, and if the benefit from it is not much lower than that among sick individuals, the healthy person will be more likely to leave. Naturally such a pattern cannot emerge from the sickness funds’ possible strategies to “exile” certain insurees, since these are the persons they would like to keep. In fact, to increase profits, the sickness funds should invest in preventing such switching patterns. This pattern originates from demand side considerations, where sick persons have
higher transaction costs of switching both providers and organizations, and therefore are more reluctant to switch sickness fund. Furthermore, at least among women, the ratio drops in advanced age, indicating that elderly persons need to be healthier than younger insurees relative to their age group in order to leave their sickness fund. Consequently, assuming that the capitation rates reflect mean national costs, among leavers, elderly women might constitute the biggest losses to their sickness funds. It is interesting to note that this pattern is similar across all three sickness funds, regardless of the level of the average costs in the age–sex groups. Among joiners, the picture is less clear cut and some variations exist between men and women, age groups and sickness funds. In general, up to a certain age which differs among men and women and across sickness funds, joiners are more expensive than average, and beyond that age – they are less expensive than average. In sickness fund B, young and middle-aged joiners are more expensive than average. In advanced age (65+ for men and 75+ for women), joiners are less costly than the average in the respective age–sex group. In sickness funds A and C, women joiners up to age 35 in A and 45 in B, are more expensive than average. Beyond that age, women joiners are less expensive than average. Among men, in sickness fund C, children up to 15 are 20% more expensive, and beyond that age – less expensive. In sickness fund A, men joiners in age groups of 0–4, 15–24 and 45–54 are between 2 and 2.4 – and women joiners aged 0–4 and 25–34 are 2–2.6 – times more expensive than the average in these age groups (while these particularly high ratios might result from a combination of relatively small cells and numerous very high relative costs, they confirm the general picture which emerges). In all sickness funds, it seems that women tend to join during their fertility age, and the reason their first-year cost is higher than average is that they join during pregnancy or just after giving birth. The relatively higher cost of joiners both men and women in the 0–4 age group might be explained by the joiners being newborns (aged 0–1). It is known that the mean medical care cost during the first year of life is about 60% more than that among children aged 1–4 [21]. Some of these switches might be technical – e.g., on the occasion of the (expected) newborn, the wife (and the newborn, if the switch was made after delivery) switches to the husband’s sickness fund. In sickness fund B, such switches are done mostly before giving birth (i.e., the baby is born to be a member), while in the other sickness funds these moves occur during the first year after delivery. Men joiners aged 45–54 in sickness fund A and aged 45–64 in sickness fund B are more costly than average as well. Fear or anticipation of (repeated) cardiac events, which peak at this age group, might lead these men to seek more intensive care in another sickness fund. In advanced age, joiners are less costly than average. As was seen with respect to elderly leavers, sick persons in advanced ages do not switch sickness funds, and those who do are less expensive than average, regardless possible differences in the mean cost across sickness funds.
A. Shmueli / Health Policy 102 (2011) 247–254
251
SF A 2.8 2.6 2.4 2.2 2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0
#
#
0-4
5-14
15-24
*
25-34
35-44
Women joiners
45-54
55-64
65-74
75+
*
Women leavers
SF B 1.6 1.4 1.2 1
#
0.8 0.6 0.4 0.2 0 0-4
5-14
15-24
25-34 35-44 45-54
Women joiners
55-64 65-74
75+
Women leavers
SF C 1.4 1.2 1 0.8 0.6 0.4 0.2 0 0-4
5-14
15-24 25-34 35-44 45-54 55-64 65-74
75+
Women joiners Women leavers Fig. 1. Relative cost of joiners and leavers by sickness fund (SF) – women. * Insignificant difference between joiners and leavers. # Insignificantly different from one.
The general observation that emerges with respect to young joiners in their first year of membership is that they consume relatively more intensive care, and constitute a possible financial loss to the sickness funds. If capitation rates are set at mean cost, joiners up to middle age (up to age 64 in sickness fund B) constitute predictable losses. These are demand-side initiatives originating from persons
searching for more and/or better care, or from technical circumstances. Sickness fund B is exceptional in attracting persons in most age groups who become high cost joiners. The sickness funds might find it profitable in the short run to invest in preventing these persons from joining (see below, however, for the long run considerations). Elderly joiners (joiners over middle age in sickness funds A and
252
A. Shmueli / Health Policy 102 (2011) 247–254
SF A 2.4 2.2 2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0
0-4
5-14
15-24
25-34
35-44 45-54
55-64 65-74 * * Men leavers
Men joiners
75+
SF B 1.4 1.2 1 0.8 0.6 0.4 0.2 0 0-4
5-14
15-24
25-34
35-44
Men joiners
45-54
55-64
65-74
75+
Men leavers SF C
1.4 1.2 1 0.8 0.6 0.4 0.2 0 0-4
5-14
15-24 25-34 35-44 45-54 55-64 65-74 Men joiners
75+
Men leavers
Fig. 2. Relative cost of joiners and leavers by sickness fund (SF) – men. * Insignificant difference between joiners and leavers.
C), on the other hand, are likely to constitute predictable profits, and sickness funds might find it profitable to encourage them to join (or are actually doing so in some implicit ways). In all but 4 cells (in sickness fund A, men aged 55–64 and 65–74, and women aged 15–24 and 65–74), within sickness funds, the mean cost among joiners is significantly different than that among leavers. Among men aged
up to 64 and women up to age 74, in all sickness funds and for most age groups, the costs of the switchers in the year-prior-switching are lower than the costs in the yearafter-switching. Among men aged 65+ and women aged 75+, leavers in their last year are more costly than joiners in their first year. Both mean costs are lower, however, than the age–sex specific mean cost of care. While among elderly switchers, both relative costs are below one and the per-
A. Shmueli / Health Policy 102 (2011) 247–254
cent switching is low so that this relation between joiners’ and leavers’ cost does not bear a significant effect on mean cost, in younger ages, this pattern causes, all other determinants held constant, an increase over time in age–sex specific mean costs. This conclusion is independent of to where or from where the switch has been made. A separate interpretation of the different relative costs of joiners and leavers, which cannot be advanced further since we do not follow (from–to) individual switchers, would follow from associating the differences in relative costs to the different age–sex mean costs across sickness funds. Under this interpretation, the higher relative costs of joiners do not originate from higher consumption of care during the first year of membership, but from joiners switching to sickness funds with lower age–sex specific mean cost, possibly keeping their previous (absolute) intensity of care. A person with low relative cost in a sickness fund where the members in the same age–sex group are sick might find it beneficial to switch to a sickness fund where the members in the same age–sex group are healthy. Even keeping the same level of care as in the previous sickness fund, this person will now have a higher level of relative cost as a joiner. To illustrate the benefits from such a move, consider waiting times for doctors’ appointment and in their offices. If total physicians’ office hours are incompletely adjusted to the populations’ sizes and needs, in a sick population, average waiting times will be higher than in a healthy population. If waiting times are common knowledge, a healthy person might benefit from shorter waiting times if switches from a sickness fund with sick members (high mean cost) to one with a healthy population (low mean cost). Since we cannot identify the sickness funds of origin or destination even on the aggregate level because of different data collection periods, this interpretation remains as a hypothesis to be examined in further research. 6. Conclusions The results indicate that while only about 1% of the population switch sickness fund in Israel yearly, the switchers are different from the rest of the population in their medical care costs. In that, Israeli switchers are not different from switchers in the US or The Netherlands, although their switches are not related to different premiums. In general, the findings do not confirm the existence of switching due to dumping or skimping actions by the Israeli sickness funds. This is quite an unexpected conclusion in light of the very incomplete Israeli risk adjustment scheme which is based on age only. We note however, that dumping or skimping actions are only one type of selection. Some evidence suggests, for example, that low availability of community services in poor and sick towns might be used as a risk selection tool [16]. Persons leaving their sickness funds are, in all age groups, relatively healthy, whom the sickness funds would like to keep, let alone making them leave. The (demand-side) reason people leave their sickness fund cannot be inferred from the present data, but it is clearly related to higher transaction costs – and lower benefit net of possible dissatisfaction – of switching born by high users of the sickness fund’s services.
253
Young joiners, in particular women, are relatively more expensive to care for during their first year of membership, so that sickness funds are hardly expected to attract these joiners, unless the capitation rates are significantly higher than the mean cost in all three sickness funds (Israeli capitation rates are not determined explicitly by the sickness funds’ mean costs). Young women joiners join the sickness fund in order to receive pre- and/or post-natal care, which is relatively (to the age group) expensive. Care of newborns is also more expensive than the average in the 1–4 age groups. Recent changes (2005) in the capitation rates divided the 0–4 age group into two groups – a more expensive 0–1 and 2–4. This change will reduce the predictable loss associated with (mothers and) newborns joining a sickness fund. While young pregnant women and young mothers joining with their newborns are bad risks during their first year of membership, later on the children become, under the present capitation rates, good risks due to the generous rate for children aged 1–15 in the risk-adjustment scheme. This long-term consideration provides incentives for the sickness funds to attract high-fertility families, which may constitute a waste of social resources as well. It might explain the finding that the rate of switchers among the Orthodox Jews and Arab families, which have high fertility, was 4 times the switching rate in the general population in 2005 [17]. It is likely that these longer-term considerations help the sickness funds overcome “its own transaction costs of switching” associated with the first-year high relative costs of young joiners. Elderly joiners are less costly than their age–sex counterparts, and constitute good risk for the sickness funds, and should have been targeted by the sickness funds. However, under the present age-based Israeli risk-adjustment scheme, the sickness funds have no incentive to attract chronic elderly whatsoever, and indeed chronic elderly do not switch sickness funds, which in turn assures that there is no actual threat imposed by sick elderly switchers on the sickness funds. Another factor leading to higher transaction costs of switching among sick persons is a waiting period imposed on persons joining the sickness funds’ supplementary health insurance (supplementary health insurance is community-rated, and ownership rate reached 70% in 2006). Healthy individuals (who do not care too much on supplementary insurance coverage) and poor individuals (who do not own such insurance) are more likely to switch, from that view point. Unfortunately, while the data at hand is unique, it is quite limited. It does not allow for a complete follow up (before–after) of individual switchers across the sickness funds of origin and destination. Nor does it explore why the sicker than average persons do not switch, beyond facing higher transaction costs of switching. It focuses on costs during one year prior and one year after the move, which might be quite transient. The conclusion that most switching originates from demand-side adverse selection motives does not rule out other forms of implicit risk selection, and therefore does not alleviate the need for improving the Israeli risk adjustment scheme, in particular, considering the inclusion of health
254
A. Shmueli / Health Policy 102 (2011) 247–254
status. It does call up for further examination of the Israeli switching behavior in particular and the nature of the competition among the Israeli sickness funds, in general. Acknowledgements The cooperation of the sickness funds, and in particular of Tuvia Horev, Orly Ouri, Francis Wood, and Irit Zmora, and the very helpful comments of Roger Feldman, Kobi Glazer, Rene van Vliet and Francis Wood on earlier drafts are greatly appreciated. Appendix A. Appendix Let the respective standard deviations be s1 , s2 and s0 , and the group sizes – n1 , n2 and n0 , respectively. The tstatistic for testing equality of the means between leavers and stayers is: t = [m1 − m0 ]/[(s1 2 /n1 ) + (s0 2 /n0 )]1/2 = [(m1 /m0 ) − 1]/c{[(m1 /m0 )2 /n1 ] − (1/n0 )}1/2 . The statistic for testing equality of the means between joiners and veterans is similar, with m2 and n2 replacing m1 and n1 , respectively. Since n0 is relatively large, 1/n0 is relatively small, and the t-statistic can be approximated by t = [(m1 /m0 ) − 1]/{(m1 /m0 )[c/(n1 )1/2 ]}. The difference in the means is significant (5%) if |t| > 2, namely, when (m1 /m0 ) > 1/[1 − 2c/(n1 )1/2 ] > 1 (for n1 > 4c2 ), or when (m1 /m0 ) < 1/[1 + 2c/(n1 )1/2 ] < 1. For example, for c = 3 and n1 = 100 (m1 /m0 ) is significantly greater than one if it is greater than 2.5, and is significantly smaller than one if it is smaller than 0.625. If c = 4 and n1 = 1000, the critical values for (m1 /m0 ) are 1.33 and 0.8. Since the proportion of switchers is small in all age–sex groups (see Table 1 below), in what follows (m1 /m0 ) will be referred to as the (age–sex specific) relative cost of leavers, and (m2 /m0 ) – as the relative cost of joiners. For each sickness fund, ignoring the effect of the mean cost of joiners on the mean cost of stayers and of the mean cost of (next year) leavers on the mean cost of veterans due to their small proportions, it is also possible to test the equality of the mean cost of joiners and leavers within sickness funds. Following a similar derivation as above, the approximate test statistic is [(m1 /m2 ) − 1]/c{[(m1 /m2 )2 /n1 ] − (1/n2 )}1/2 , and (m1 /m2 ) = (m1 /m0 )/(m2 /m0 ), using the above notation.
References [1] Altman D, Cutler DM, Zeckhauser RJ. Adverse selection and adverse retention. American Economic Review 1998;88:122–6. [2] Beaulieu ND. Quality information and consumer health plan choices. Journal of Health Economics 2002;21:43–63. [3] Call KT, Dowd B, Feldman R, Maciejewski M. Selection experiences in Medicare HMOs: pre-enrollment expenditures. Health Care Financing Review 1999;20:197–209. [4] Cao Z, McGuire TG. Service-level selection by HMOs in medicare. Journal of Health Economics 2003;22:915–31. [5] Dormond B, Geoffard P-Y, Lamiraud K. The influence of supplementary health insurance on switching behavior: evidence from Swiss data. Health Economics 2009;18:1339–56. [6] Feldman R, Dowd B. Simulation of a health insurance market with adverse selection. Operation Research 1982;30:1027–42. [7] Glazer J, McGuire TG. Optimal risk adjustment of health insurance premiums: an application to managed care. American Economic Review 2000;90:1055–71. [8] Glide SA. Managed care. In: Culyer AJ, Newhouse JP, editors. Handbook of health economics. NH: Amsterdam; 2000. [9] Hellinger FJ. Selection bias in HMOs and PPOs: a review of the evidence. Inquiry 1995;32:135–42. [10] Ji L, Liu F. HMO versus non-HMO private managed care plans: an investigation on pre-switch consumption. Health Care Management Science 2007;10:67–80. [11] Laske-Aldershof T, Schut F, Beck K, Greß S, Shmueli A, Van de Voorde C. Consumer mobility in social health insurance markets: a fivecountry comparison. Applied Health Economics and Health Policy 2005. [12] Luft HS, Miller RH. Patient selection in a competitive health care system. Health Affairs 1988;7:97–119. [13] Nicholson S, Bundorf K, Stein R, Polsky D. The magnitude and nature of risk selection in employer-sponsored health plans. Health Services Research 2004;39:1817–39. [14] Roos A-F, Schut FT. Spillover effects of supplementary on basic health insurance: evidence from the Netherlands. European Journal of Health Economics 2010 [Online First, 23 September 2010]. [15] 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;3:117– 39. [16] Shmueli A, Engelcin-Nissan E. Local availability of physicians’ services as a tool for implicit risk selection by health plans in Israel, unpublished 2010. [17] Shmueli A, Achdut L, Bendelac J. Who switches sickness funds in Israel? Health Economics, Policy and Law 2007;2:251–65. [18] Strombom BA, Buchmueller TC, Feldstein P. Switching costs, price sensitivity and health plan choice. Journal of Health Economics 2002;21:89–116. [19] Van Vliet RCJA. Predictability of individual health care expenditures. Journal of Risk and Insurance 1992;56:443–60. [20] Van Vliet RCJA. Free choice of health plan combined with riskadjusted capitation payments: are switchers and new enrollees good risks? Health Economics 2006;15:763–74. [21] Zmora I. The capitation formula of Clalit Health Services. Paper Presented at the RAN Meeting. Amsterdam, November 2004.