Cet ouvrage fait partie de la bibliothèque YouScribe
Obtenez un accès à la bibliothèque pour le lire en ligne
En savoir plus

Franchise sur les soins ambulatoires et équité sociale (version anglaise)

21 pages
L'objectif de maîtrise des dépenses de santé s'est davantage attaché depuis les années 90 à la régulation de l'offre qu'à celle de la demande, en maintenant inchangés le niveau et la structure de la participation financière des usagers. Leur responsabilisation reste cependant au coeur du débat : la mise en place de franchises sur le remboursement des soins pourrait constituer un facteur efficace de réduction du risque moral. Un modèle de microsimulation a été construit pour en évaluer les conséquences à la fois en terme d'équité et du point de vue de leur acceptabilité sociale. Un scénario de référence d'une franchise uniforme pour toutes les familles d'assurés sociaux (FAS), dont le montant serait déterminé par un objectif d'économies globales modéré, conduit pour le seuil de remboursement à un ordre de grandeur de 1 000 F par an et par famille. Par ailleurs, se trouve confirmée l'idée intuitive suivant laquelle un tel dispositif pénalise les familles disposant des revenus les plus modestes. Ce caractère régressif peut être atténué par une fixation individuelle du montant de la franchise, tenant compte de la taille et des ressources de la famille. Une modulation suivant le niveau de revenu par personne, notamment, permet d'aboutir à une situation plus équitable que la situation initiale sans franchise ; le seuil imposé aux familles appartenant à la tranche médiane de revenu s'avère une variable déterminante pour un tel objectif d'équité verticale.
Voir plus Voir moins

An Ambulatory Care
Deductible and Social Equity*
Microsimulation of a User Payment System
for Health Care Services
Claire Lachaud In the 1990s, curbing health care expenditure concentrated more on regulating
Fiume, Christine supply than demand by freezing the level and structure of user cost sharing. Yet
Largeron Leténo making users responsible for their own spending remains a core issue. An effective
and Lise Rochaix- way of reducing the moral hazard could be to introduce a deductible for health
care reimbursements like an excess in insurance policies. A microsimulation model
has been built to evaluate the consequences of this in terms of equity and social
The amount of the deductible would be the same for all covered families and
* Originally published as would be determined by a moderate total savings goal. It would be set at
“Franchise sur les soins
approximately the first 1,000 French francs of health care spending per family perambulatoires et équité
sociale,” Économie et year. Obviously, such a mechanism would penalise the families with the lowest
Statistique, n° 315,
incomes. This regressive aspect could be alleviated by setting individual amounts1998 5.
** Claire Lachaud Fiume by family size and resources. Adjustments, particularly individual income based
works for the Université
Lyon I, URA 934, Chris adjustments, could even produce a more equitable situation than before the
tine Largeron Leténo for introduction of a deductible. Under this hypothesis, the threshold for families in
the Université Jean
Monnet Saint Etienne - the median income bracket is a determinant variable for the vertical equity goal.
LASS Lyon I and Lise
Rochaix Ranson for the
Université de Bretagne
occidentale, Department
of Economy and ICI
ost health systems are based on user particularly good illustration of this. However,Laboratory (Information,
Co ordination and M cost sharing whether in the form of an this has not always been the case. J. P. Rey
Incentives). insurance mechanism as in the United States or (1993) notes that pre 1930 cost sharing was
The authors would like
incorporated into the tax system as in the more like a user charge designed mainly toto thank the Direction
1de la Prévision, the United Kingdom. Yet cost sharing is based on redeem patient registration charges. A long
Direction du Budget two separate types of reasoning. It is either a debate between the French Senate and the
and the Commissariat
sort of user charge that funds part of the totalChamber of Deputies finally led to theGénéral du Plan for
their financial backing spending or an incentive to reduce the moralintroduction of cost sharing proportional to
under a joint research 2hazard. The first is a set, limited sum thatconsumption in 1930. is
agreement and CREDES
the same for everybody. In the second case,and INSEE for providing
data. cost sharing is proportional to consumption. In
The names and dates in France, health system user cost sharing is
1brackets refer to the See Schneider (1998).
basically of the second type: the ticket 2bibliography at the end The moral hazard can be defined as the change in behaviour
of the article. modérateur (patient’s co payment) is a brought about by the presence of insurance.
INSEE Studies no. 38, September 1999 1Box 1
Many health insurance consolidation plans have - The addition of a 26th affliction to the list of afflictions,
been put forward over the last twenty years. They entitling beneficiaries to 100% reimbursement when
have generally contained proposals to increase the share of the cost they pay exceeds 80 FF/month
revenue and check the growth in expenditure. or 480 FF/month for six months;
- In December 1979, the creation of an extraordinary
The Durafour plan tax on the earnings of pharmaceutical dispensaries.
Presented in December 1975, it included:
- A reduction in VAT on pharmaceuticals (from 20% to The Questiaux plan
7%); Presented in November 1981, it included the
- The removal of the ceiling on the payroll share of introduction of a 1% contribution for the unemployed
health insurance. (above the minimum wage) and the doubling of car
insurance tax paid to the health insurance.
The Barre plan
Presented in September 1976, it included: The Bérégovoy plans
-The exclusion of certain “luxury” drugs from 70% Presented in July and September 1982, they
reimbursement; comprised:
-An increase from 25% to 35% in the patient’s -The freezing of medical fees and pharmaceutical
co payment for medical auxiliary treatment (except product prices and earnings;
nurses); - No increase in the daily allowance rate provided for
- The introduction of a 30% patient’s co payment for after more than three months of treatment;
transport costs; - The creation of the fixed daily hospital fee for hotel
-The introduction of a government contribution to services;
social security funding in the form of a new road tax - The introduction of a total hospital budget, brought
disc. into general use in public hospitals in 1985;
- The reclassification of drugs to treat ailments that are
The Veil plans not normally serious: reimbursement reduced from
The first plan was presented in April 1977 and 70% to 40% for 1,258 drugs;
consisted of: - A 5% tax on pharmaceutical advertising.
-An increase in contribution rates for agricultural
wage earners and workers over 65 years old; The Delors plan
- The introduction of health insurance contributions for Presented in March September 1983, it created a
pensioners; 1% extraordinary levy on taxable incomes, which
-The increase from 30% to 60% of the patient’s was extended to unearned income in September.
co payment for “luxury” drugs; in return, the 90%
reimbursement rate for drugs was raised to 100%. The Dufoix plan
Presented in May June 1985, it comprised an
The second plan was presented in December 1978 increase in the patient’s co payment for certain care
and included: and the reclassification of 379 “luxury” drugs.
-The launch of a health supply control policy to
stabilise and even reduce the number of beds (in bothThe Seguin plan
the public and private sector), control the creation of Presented in November 1986, this plan was
major medical facilities and set an admission quota introduced step by step after a particularly poor year
for second year medical students; in 1986 (a deficit of 7.5 billion French francs for the
-A campaign to prevent social scourges such as CNAMTS). It included:
alcoholism and tobacco addiction; - The extension and revision of the list of 25 afflictions;
- The creation of the Commission des Comptes de la -Exemption from the patient’s co payment limited
Sécurité sociale (social security audit committee). solely to the exempting illness: the beneficiaries
entitled to 100% coverage for a long and costly
The Barrot plan affliction receive this coverage only for the treatment
Presented in July 1979, it consisted of: associated with the exempting affliction (5 May 1987);
-Freezing medical fees until the conclusion of an - Smoothing of the reference period over three months
agreement. This agreement was signed in 1980. In for the calculation of daily benefits;
return for not raising the fees, a new sector (Sector 2)- The end of official paid post;
was created for GPs and specialists wanting to - Actual 40% reimbursement of drugs subject to this
participate in it to freely set their fees. The difference rate (the “40 to 40” measure, 01/01/87). Drugs
between the statutory fee (defined on the basis of a bearing a blue price label are actually reimbursed at
Sector 1 consultation) and the price practised by the the rate of 40% regardless of the beneficiary’s
practitioner formed the balance to be paid by the situation in terms of exemption from the patient’s
patients (or their supplementary insurance where they co payment (formerly, patients exempt from the
have one and where it reimburses the cost); patient’s co payment for reasons of serious affliction
- No daily price increases (for clinics) and compliance were exempt even for blue label drugs used to treat
with initial budgets (public hospitals); ailments that are not normally serious);
2 INSEE Studies no. 38, September 1999Since, other types of cost sharing have of the costs involved. It is only applied to
gradually been introduced alongside the certain services (mainly ambulatory care,
patient’s co payment to satisfy other goals than hospital care paid directly by the insurers and
just the reduction of the moral hazard (see Box drugs sold in certain pharmacies).
1). These are the fixed daily hospital fee, which
separates the hotel part from the health careIn France, user cost sharing is made up of a
part, and balance billing by Sector 2 doctorswide range of methods and forms a
4(GPs and specialists entitled to freely set theinrot inconsiderable share of the total funding
fees) whereby the user pays the difference of health insurance expenditure (CREDES,
between the statutory fee negotiated with the1989). The growing weight of these direct
3 3 This corresponds to producer and the price billed by this producer.
the American concept of Although the fixed daily fee is similar to a user4extra billing or balance In July 1997, the Commission des Comptes de la Santé
charge, balance billing refocuses the collectivebilling. reported that the share of social security in the funding of
responsibility on a basket of services (namelyambulatory care (doctors, dentists, medical auxiliaries,
laboratories, dispensaries and spas) fell from 66.7% in 1980those of Sector 1 doctors) and has the user pay
to 57.5% in 1996. Households and their private supplementary
the difference for Sector 2 doctor services. Theinsurance policies (especially the mutual insurance
initial patient payment (paying the entire costcompanies) picked up the bill for most of the remainder with
their joint contribution rising from 32.4% to 41.8%. The sharewith subsequent health insurance
of local and central government showed little change from
reimbursement) is another form of cost sharing.0.9% to 0.7%. Also in 1996, the costs paid by households were
The purpose of this is to make users more awareevaluated at 22.6%.
Box 1 (end)
-The definitive withdrawal of the 26th affliction The Durieux plan
mechanism (30/06/87); Presented in July 1991, it established for the first
- An end to exemption from the patient’s co payment time the concept of regulating the supply of medical
for sick leave lasting over three months; care, without any demand measures. It included a
- The raising of the fixed daily hospital fee to 25 Frenchdrugs plan, a control totals project for the
francs; ambulatory sector and the renewal of medical
-The non reimbursement of vitamins with the classifications, especially in biology.
exception of vitamins D and B12;
- Extraordinary 0.5% income tax from 1985 to 1986; The 1993 Veil plan
- A 2% increase on tobacco. Presented in June 1993, it consisted of a series of
measures for social security beneficiaries including:
This plan produced substantial savings and even a - A 5% increase in the patient’s co payment for medical
decrease in health insurance reimbursements in goods and services;
constant French francs in 1987 ( 0.8% compared - An increase of 5 French francs in the fixed daily fee.
with 1986). Total consumption growth slowed in
1987 compared with the previous years. This Other measures concerned the medical, dispensary
phenomenon is not easy to interpret, but could be and pharmaceutical sectors.
due to doctors becoming more aware of properly
prescribing treatment, beneficiaries tempering their In the medical sector:
consumption and the absence of an influenza - A 5 FF increase in fees;
epidemic. The CNAMTS reports that this plan - The introduction of a “medical control” for expenditure
produced general social security system savings of based on a list of 24 reference points applicable to
10 billion French francs. doctors and the coding of treatment, afflictions and
prescriptions to check compliance with these
The Evin plan references;
Presented in September 1988, it took up some of the- The setting of a 3.4% growth rate in 1994;
measures in the previous plan and included: - Ongoing GP medical files on patients (over 70 years
- Exemption from the patient’s co payment for “off list” old with double afflictions).
long term ailments (LTA); In the dispensary sector, the tax on the turnover of
- Abolition of the six month lead time between the startwholesalers distributors was reinstated (rate
of treatment and the start of exemption for exempting adjusted according to the distributor’s growth in
afflictions; turnover).
- Exemption from the patient’s co payment for patients
suffering from multiple afflictions for all care In the pharmaceutical industry, a framework
associated with the exempting affliction; agreement was concluded to restrict the volume of
-Exemption from the patient’s co payment for drugs prescribed while authorising an adjustment in
blue label drugs prescribed for a patient with an the price of certain traditional specialities so as not
“on list” LTA, an “off list” LTA or a polypathology and to affect the laboratories’ research and innovative
for care associated with the exempting affliction. capabilities.
INSEE Studies no. 38, September 1999 3payments has prompted the development of a plan was to introduce a deductible for health
supplementary insurance market where suppliers care reimbursements modeled on car insurance
compete mainly on the extent of coverage. Some policy terms. The idea is that users pay for all
insurance policies pay both the patient’s their health care spending up to a certain ceiling
co-payment and the balance billing. Contrary toand the health insurance pays thereafter. This
public economic theory findings that show a need proposal requires a detailed assessment of the
to keep some costs non reinsurable to reduce the consequences of such reforms, the savings they
moral hazard (Laffont, 1985), most of the current would engender, their social acceptability, and
mechanisms thus give rise to a form of especially the equity implications. This last
reinsurance of the patient’s co payment. consideration is part of ongoing European work
Econometric studies made by Caussat and on international comparisons of equity in
7Glaude (1993) in France confirm the existence of health reforms.
the moral hazard and adverse selection (high risk
individuals are more likely to take out
5 5 Static microsimulation: However, this exercise insurance).
is complicated by the fact a measurement of the mechanical
that the population
This complex structure ultimately lays the effects of policy choiceswithout supplementary
insurance is limited and cost sharing burden on a minority of users:
highly specific (young those with resources above the social security This study uses the methodological approach of
people and senior
benefits entitlement ceiling (the social securitysimulation. A model is built usingcitizens account for a
large proportion). system pays for the others), who have no microeconomic data to provide an estimated
long term ailments (LTA for which exemption breakdown of total health expenditure into two
applies) and no supplementary insurance (i.e.groups (limited to ambulatory medicine):
6 6 See Boisselot and approximately 17% of the population)
Rémond (1994). (Rochaix, 1995a and 1995b). Moreover, Outlays paid by the social security system;
INSEE’s Health surveys find consumption
spillover effects on services affording smaller- Remaining outlays to be paid by the social
contributions or exemption from initial security beneficiaries and/or their
payment (hospital services in particular) supplementary insurance, in the form of the
(Mormiche, 1986). patient’s co payment and the balance billing in
certain cases.
The increase in cost sharing by an average of
approximately 5% with each new cost cuttingThe simulation consists of studying how this
mechanism has increased the burden borne bymodel behaves and alters under decision
this minority. Yet despite much criticism of the variables reflecting changes in the cost sharing
equity and efficiency of the current user methods applied to social security
payment schemes, structural reforms are a longbeneficiaries. The potential consequences of
time coming. The Health 2010 group’s the government’s adoption of a given measure
workshop on “Producer Remuneration and are thus deduced. To be more precise, an
User Cost-sharing” (Commissariat Général du analysis is made of the change in the
Plan, 1993) found that this cost sharing neededbreakdown of the proportions of spending
to be clarified or the entire system overhauled covered and not covered by the social security
to take account of income, among others, above system as brought about by applying one of the
and beyond the current exemptions for proposed methods for a deductible on
long term ailments (LTA). The 10th Plan Social ambulatory care consumption. The measure’s
Security Committee had already recommended impact is also evaluated by means of financial
aligning user cost sharing with income with the return, equity and social acceptance indicators.
introduction of user co payment proportional
to income (Commission Général du Plan, Note that this study’s simulations project the
1989). In addition to these equity consequences of the hypotheses based on
considerations, some form of user payment microeconomic data, but not their feasibility.
system would inevitably be demanded by theHence, this is not a statistical model. The
health professions and drugs manufacturers, purpose is not to forecast the savings produced
themselves subject to more restrictive
regulations regarding their volume of business.
7 “Equity Project”, COMAC HSR, fourth EEC medical research
One of the courses of action recommended by programme (see Van Doorslaer et al, 1993; Lachaud and
the health insurance financial consolidation Rochaix, 1995).
4 INSEE Studies no. 38, September 1999by a measure, but to provide indications to reconstitute household expenditure after
better evaluate the consequences of reimbursement by the social security system,
implementing a measure compared with the but before payment by supplementary
current situation or even to compare differentinsurance policies (see Appendix III).
measures. The microsimulation technique Appendix IV presents some significant features
more accurately documents the distribution of of medical service consumption by BFs in the
the financial load of residual payments and survey (tables A, B and C).
compares the impact of different deductible
measures on this distribution. This technique is
used since actually testing this type of measure A range of deductible scenarios
would be extremely expensive.
The impact of a user payment measure is
This kind of model measures the sensitivity of closely linked with the way it is introduced.
the results to the different payment setting
methods. Although it is essentially static, it The simulated measure chosen here from all the
does not include feedback elements and givespossible forms of user payment is the annual
no indication of the impact of such decisions on deductible. The households would pay for all
individual behaviour. Individual behaviour is their outlays below a certain threshold.
thus assumed to be constant for both the Reimbursements by the compulsory health
beneficiaries and the health care producers. Theinsurance system would only be made for
reasoning is more budgetary than economic, spending above the ceiling. This mechanism
mainly to measure the spending transfers would apply to everybody, contrary to private
among income deciles based on their health insurance policy excesses whose selectivity
care consumption. An analysis of the potentialallows tailor-made policies for high risks
effects of reducing the moral hazard would compared with the others. Such a measure
require including the economic behavioural reflects the intention to increase the financial
change hypotheses (price and income load of users and reduce the moral hazard.
elasticities for the beneficiaries and the impact
of the health care price increase on doctors’However, the equilibrium that would
prescription behaviour). However, the problemredistribute the reduced social security
8with this is that the few estimates available are contribution gains derived from the deductible
not individual, but concern large population is not simulated.
Although such a reduction may be desirable to
The data are taken from the CREDES 1998 give the beneficiaries a better image of the link
Health and Social Security (ESPS) survey (see between benefits and contributions, it is not
Appendix I). The information includes really feasible in practice due to the health
sociodemographic data on sickness, social insurance system’s cumulative deficit.
security and medical consumption over the 21Moreover, the structure of health insurance
days of the survey. The survey’s main coveragecontributions (extremely low, especially for
is households with at least one beneficiary of pensioners) and the differences in consumption
the social security system for wage earners by beneficiary category mean that such a
(system covering over 80% of the population in reduction would necessarily entail transfers
France). A total of 3,817 beneficiary families among beneficiary categories. Otherwise, the
(BF) were reconstituted using the ESPS decrease in contributions might no longer
survey’s precise identification of the benefit those who reduce their immoderate
connections between respondents. The Social consumption.
Security Code defines a BF as being made up of
a beneficiary and his/her dependents such that a The services covered by the deductible exclude
household can be made up of more than onehospital care, because these outlays are
beneficiary family. essentially the responsibility of the doctors and
The purpose of the first set of data processing is
to transform the initial discrete information
(income brackets) into a continuous income
8 The study by the RAND Corporation in the United States is thevariable (see Appendix II). A second set of
only example of social security testing in health. It provides price
processing uses the outlays stated in the survey,elasticities of health care demand, but only at an aggregate level
an inherently inaccurate concept (see Box 2), to(Manning et al, 1987; Newhouse, 1993).
INSEE Studies no. 38, September 1999 5Box 2
The information collected for the INSEE Health patient’s co payment or the entire sum. This outlay
surveys is based on household spending during the may be subsequently reimbursed by supplementary
survey period. Yet this “outlays” variable does not health insurance policies. Although the large number
correspond exactly to the residual payment, similar and variety of supplementary insurance policies
to what is measured in most other health systems make it difficult to analyze this second aspect, the
(out of pocket expenditure ). The fact that it is difficult outlays notion can be transcended to move towards
to define what the household ultimately pays is the residual payment notion.
explained by the complex nature of the French user
payment system. The system works on the basis of A first attempt at this was made in the European
initial patient payment for medical services (payment comparative study on the distributive effects of health
of the entire sum with subsequent partial health insurance (Van Doorslaer et al, 1993). It consisted of
insurance reimbursement) coupled with the patient’s a valuation of reported consumption volumes. The
co-payment, whereas hospital services are paid Health and Social Security (ESPS) survey made by
directly by the insurers and thus entail no initial the Health Economics Research and Documentation
patient payment. Certain pharmacies also apply Centre (CREDES) refined this initial work by including
direct payment by the insurers. Here again, the user an additional question (“Debtar” variable) to define the
makes not initial patient payment. nature of the payment made. The distribution of
patients’ co payment and balance b illing covered by
Consequently, the outlays notion included in the this study was obtained using this question (see
Health surveys is hybrid and the information Appendix III for a presentation of the direct payment
collected may either correspond solely to the reconstitution method).
not the users. Moreover, it is difficult to include are below the deductible ceiling. Once this
hospital care since the data provided in the ceiling has been reached, the beneficiary’s
survey declarations for this service category are exemption conditions are reinstated. Scenario 3
9 9 These inaccuracies not always accurate (CREDES, 1989). Lastly, tests providing care over and above the
can be seen from the most of these outlays are paid directly by the deductible free of charge, which is tantamount
fact that the beneficiaries
insurers (users make no initial payment). Thisto discontinuing the patient’s co payment. Thisfind hospital outlays
obscure and have raises collection and accounting problems. scenario evaluates the financial cost of such an
problems stating their adjustment. Scenario 4 measures the impact of
total and breakdown.
The taxable unit used for the microsimulationsexcluding pharmaceutical and biological
is the beneficiary family rather than the outlays from deductible coverage.
individual or the household. This definition is
closer to the reality, since it takes into account Other scenarios were simulated in addition to
10 the specific nature of the French system these main variants to answer the questions of
whereby entitlement to benefits is linked eitherequity raised by a deductible system. The
to professional activity or the personal deductibles were aligned with the BFs’
relationship with the main beneficiary. It alsoeconomic and sociodemographic profile. Their
means that the health insurance’s usual benefitseffects were studied separately and then
processing unit can be used. combined as shown by Table 1, which covers
all the scenarios used.
The first step in the simulations consisted of a
reference scenario (scenario 1), which provedTaking household income into account should
to be the simplest: the introduction of an annual mitigate the regressivity inherent in scenario
11deductible identical for all BFs regardless of 1’s uniform deductible. In particular, it
their size and income. In the second step, the should ensure that the coverage rate for the
deductible mechanism in the reference scenariopoorest households is maintained with part of
was made more complex by the introduction of
10 See Lachaud Fiume, Largeron Leténo and Rochaix (1997) forA preliminary set of scenarios set down the
a detailed presentation of these four scenarios.main precepts likely to govern such a reform. 11 A regressive redistributive mechanism (as opposed to a
Hence, scenario 2 consists of revoking LTA progressive mechanism) tends to put the poorest households,
families and individuals at a disadvantage.scheme exemptions when the BF’s total outlays
6 INSEE Studies no. 38, September 1999the effort to curb health insurance spending Lastly, a variant of this mechanism is studied: a
being transferred to the wealthiest households. definition by income brackets with increasing
Such an arrangement could be a good deductible levels for each bracket defined
compromise between the two apparently (scenario IND/CU Brackets) (see Table 1).
antagonistic objectives of guaranteeing
continued government payment of The first calculation method (perfect
health related outlays and maintaining proportionality) represents an ideal mechanism
incentives to reduce the moral hazard. In this that minimises the standard of living inequity
scenario 1-R %, taking income into account (vertical inequity). The second, which is easier
makes the deductible proportional to householdto apply, is obviously farther from a perfect
income. vertical equity situation.
There are two additional factors of inequality:To make the scenarios comparable, the savings
the existence of one or more BFs in the made by introducing the deductible were kept
household (depending on the number of mainconstant and the deductible level was adjusted
beneficiaries) and the fact that the more up or down depending on the case. The
individuals there are in a BF, the quicker the hypothesis used for the annual value of these
deductible ceiling could be reached (see Boxsavings is six billion French francs, which
3). Scenario 1 IND hence adopts a deductiblecorresponds to approximately 0.2 social
sum per BF equal to the product of an security contribution points in a full year for
individual deductible multiplied by the number1988. This is the level expected from the
of people associated with it. consolidation plans launched by the
government in the 1990s. Although realistic,
Nevertheless, this scenario does not considerthis sum might appear to be arbitrary. In this
the fact that large families have a lower case, the introduction of a deductible with a
standard of living, given identical resources. zero rate of return could be simulated by
This shortcoming can be overcome by basingkeeping the level of cost sharing constant.
the individual deductible on the concept of However, the gains generated in the form of
equivalent income. It is hence defined as theprice decreases (e.g. for primary care) would
product of an individual sum multiplied by thheave to be redistributed, since the simulation
number of people making up the BF, taking intomodel cannot attain an equilibrium by reducing
account the equivalent income of the household contributions.
to which the BF belongs to set the individual
deductible sum to be applied. This comes down With the scenarios comparable in terms of
to making the deductible proportional to the savings, their respective performances were
household’s income (scenario IND/CU %). compared using two additional criteria.
Table 1
Eight deductible scenarios
Income excluded Income included
? Identical deductible for all ? Aligned deductible
Exclusion of the effects of beneficiary Scenario 1 (reference) Scenario 1 R %
family (BF) size and the composition per (General deductible defined as a
BF in the household Scenario 2 (No exemptions below the percentage of income)
? General deductible/BF deductible ceiling with reinstatement of
exemptions above)
Scenario 3 (free care)
Scenario 4 (reduction of coverage)
Inclusion of these two effects Scenario 1 IND Scenario 1 IND R/CU %
? Deductible/BF = Individual deductible Individual deductible (Individual deductible proportional to
x number of members in the BF income per consumer unit percentage)
Scenario 1 IND R/CU Brackets
income per consumer unit brackets)
INSEE Studies no. 38, September 1999 7Box 3
There is a clash of thinking over whether or not to BFs, there are still situations of upstream inequality
consider the structure and size of the beneficiary associated with composition per BF in the
family for BFs with comparable incomes. households. The following example illustrates this
BFs with a large number of dependents will reach
the user payment ceiling more quickly (particularly if Consider two households each made up of four
they include young children who often have to visit adults, with the first containing two beneficiaries and
the paediatrician) and quickly attain their previous the second only one. We therefore have:
reimbursement level. More fundamentally, the extra
financial burden per beneficiary as a result of the - a “bi BF” household made up of a BF 1 reduced to
deductible is lighter for these BFs than for those one beneficiary and a BF 2 containing a main
containing only one beneficiary. It would thus seem beneficiary and two adult dependents;
important to take the BF structure into account. -A “mono BF” household made up of one main
beneficiary and three adul.
However, large BFs could also be deemed to have a
lower income per individual in general.
Suppose these households have an identical income
Alignment with income per consumer unit R. The equivalent income R/CU( ) is also identical for
both households:
The fact that large families have a lower relative
income can be taken into account using the Oxford R/CU = R/( 1+3 x 0,7) ) = R/3,1
equivalence scale to convert the income of the
household to which the BF belongs into an Let ‘F’ be the amount of the general annual
equivalent income or income per consumer unit. deductible (calculated for all the BFs) and ‘f’ the
This can then be used as the ba sis for calculating amount of the annual deductible calculated on an
the deductible. Equivalent income is denoted R/CU individual basis. The table below compares the
to differentiate it from standard income denoted R in situations of these two BFs for the cases of
the scenarios. If R is the total income for the deductible calculation per BF (case 1) and
household considered and if the p are the deductible calculation per individual (case 2).j
equivalence scale weights associated with each
individual j in the household, then the equivalent Although the calculation of a general deductible
income (using the Oxford scale) is: (case 1) is defined on the basis of equivalent
income, it does not correct the inequality associated
R/ pj, where pj = 1 with the different composition per BF of households
j 1 and 2. The first has to pay two deductibles even
if j is the beneiciaryf , pj = 0.7 for any other adult though it is the same size as household 2 (two
(whether a dependent or not) and pj = 0.5 for a child. adults and two adult dependents). Only the definition
of the deductible on an individual basis (but still
Alignment with composition according to equivalent income case 2) handles
these two households on an equal footing. This
Although such an alignment can be introduced by latter case was the one used in scenario
taking into account the equivalent income of the 1 IND R/CU.
Case 1 Case 2
BF structure Deductible per BF Individual deductible
per BF in the household
based on R/CU based on R/CU
Bi BF household BF 1 1 beneficiary F = F F = f
BF 2 1 beneficiary + 2 dependents F = F F = 3 f
Bi BF household total F = 2 F F = 4 f
Mono BF household BF 1 beneficiary + 3 dependents F = F F = 4 f
8 INSEE Studies no. 38, September 1999
Sthe consuming population as a share of the totalMeasuring vertical equity
population. Both of these proportions were
The impact is measured using the vertical calculated for high incomes (H), median
equity concept (different contributions to theincomes (M) and low incomes (B).
financing system in view of different
contributive capabilities) based on a A second type of indicator covers the
methodology developed for the European distribution among users of the average
12COMAC HSR project. This method was additional burden resulting from the measure.
extended to switch from a cross sectional It uses two categories of ratio: the distribution
analysis to a longitudinal analysis (tracking theof the average additional annual burden per BF
consequences of a change). This compares thefor equivalent income deciles (R.1) and this
distributions of patients’ co payments with thesame distribution as a percentage of annual
contributive capacity of the beneficiaries average income per consumer unit (R.2).
before and after the application of a deductible.
An Ekj indicator is thus defined to gauge the
Measures that would be detrimental tooverall impact of the measure considered (see
Box 4). The impact’ negatives or positive sign poor families
indicates animpr ovement or deterioration in
the situation of the lowest income brackets. The The results of these microsimulations depend
quantitative impact of the measure is assessedstrictly on the sum of savings chosen (six
from its absolute value. billion French francs). They should also be
viewed as rough estimates, given the
inaccuracy inherent in any exercise of this
14Assessing social acceptability type.
Whether aligned with the BF’s income and/orThe results of scenarios 1 to 4 show that the
size, the deductible still results in an increase in expected savings (six billion French francs)
the share of health spending not covered by therequire an annual ceiling of approximately
social security. Compared with a past reference1,000 French francs per BF. The case of free
situation, it necessarily increases the care above the deductible (scenario 3) stands
household’s burden. In addition to this overallapart from the others, since this provision
effect, measured by the equity indicator, therewould require a much higher level (5,800
is a risk of a bias against the fraction of theFrench francs) (see Table 2). The positive
population that consumes the least and wouldvalues of the Ekj indicators confirm the
therefore not be reimbursed at all. The intuition that a deductible mechanism would be
deductible consequently risks strengthening detrimental to the lowest income population.
arguments in favour of the insurer’s free This development is specified by the values of
choice. It is important to identify these the Kakwani indices before and after the
potential subgroups to find whether the introduction of a deductible (columns 7 and 8
implementation of the scenario considered is of Table 2).
liable to meet with sufficient social acceptance.
The Kakwani index ( kak) measures the
Two types of indicators were defined to deviation from proportionality of a
appraise this notion of social acceptability. contribution system compared with taxpayers’
Firstly, the relative weight of BFs deprived of
See Lachaud (1992) for a more detailed description of thereimbursement by the deductible was measured
equity measurement method and Lachaud-Fiume,13by two proportions: the first P1( ) covers the
Largeron Leténo and Rochaix (1996), report to the
consuming BFs below the deductible ceiling as Commissariat Général du Plan, for a presentation of the context
surrounding these measures.a proportion of the total population; the second
Let N be the population of BFs in the sample made up of a
(P2) covers the consuming BFs below the
number Nnc of non consuming BFs and a number Nc of
deductible ceiling as a proportion of consuming consuming BFs; Nc = Ncnf + Ncf, where Ncnf is the number of
consuming BFs not surpassing the deductible (F) and Ncf is theBFs. Proportion P1 directly measures the
number of consuming BFs surpassing the deductible. The
unpopularity of a measure that burdens a large
following proportions can thus be defined: P1 = Ncnf/N and P2
fraction of the population with an outlay = Ncnf/Nc.
Due in particular to the uncertain income allocation (seepreviously paid by the government. Proportion
Appendix II), the survey related biases (see Appendix IV) and
P2 is especially useful when the expenditure
the systematic error due to the non inclusion of behavioural
field is varied, which brings about a variation in responses.
INSEE Studies no. 38, September 1999 9
PBox 4
The equity concept used in this study is derived from - A concentration curve showing direct payments
concepts formulated in terms of equality, themselves before the application of the measure considered
the product of an ancient legacy. Aristotle, in his (Gtax1);
1research on the foundations of distributive justice, - And a concentration curve showing direct payments
was the first to define the principles later called after application of the measure (Gtax2), since a
vertical equity (different individuals should be treated deductible changes the interindividual distribution of
differently) and horizontal equity (identical payments compared with the income brackets.
individuals should be treated identically). This is the
principle of vertical equity applied here, i.e. If Cinc, Ctax1 and Ctax2 are the concentration indices
“different” individuals (defined here in terms of associated with each of these three concentration
contributive capacity) should be treated differently curves, the effect of the deductible considered is
(the “treatment” being defined by personal monetary measured by calculating the two following Kakwani
contributions to the financing system). This entails indices:
comparing, in terms of vertical equity, a distribution
of payments not covered by the social security kak1 = Ctax1 – Cinc
(likened to private contributions to the financing
system) based on the contributive capacity of the kak2 = Ctax2 – Cinc
beneficiaries before and after simulating the
application of a deductible. The difference between these two Kakwani indices
provides a numerical indicator of the effect of the
The advantage of this equity definition is that it deductible considered. If Ekj is the overall effect of a
respects the criterion of operationality, i.e. it can be measure j, then:
used in an empirical analysis. Equity can thus be
measured using concentration indices constructed Ekj = kak1 – kak2 = C tax1- Ctax2
by analogy with the Kakwani index (1977), which
measures the deviation of a tax system from Ekj represents twice the area between the two
proportionality. By extension, this index was used to payments concentration curves. The sign of Ekj
measure the deviation of a health care financing gives a preliminary indication of the overall effect of
system from proportionality (Lambert, 1985; Van the measure j considered. If Ekj < 0, then the measure
Doorslaer et al, 1993). The Gtax concentration curve improves the initial situation in terms of equity.
is thus the result of all the contributions to the healthConversely, Ekj > 0 reflects a worsening of the initial
system based on the population’s accumulated situation in terms of equity (see graphs A and B).
Tables B and C show that the improvement ( Ekj < 0)
By definition, the Kakwani index varies from - 2 to + may reflect the increase in the initial progrssivity ofe
1. This index can be used to summarise the the direct payments system (favourable case 1), the
situations corresponding to the different values transition from a regressive system to a progressive
taken by this index (see Table A). system (favourable case 2), or a reduction in initial
regressivity (favourable case 3). Likewise, the
The reforms can be analysed by comparing the worsening (Ekj > 0) may reflect a reduction in the
Kakwani indices before and after introducing the initial progressivity of the direct payments system
measure. Three concentration curve profiles are (adverse case 1), the transition from a progressive
therefore required to evaluate a deductible system’s system to a regressive system (adverse case 2) or an
impact on vertical equity: increase in the initial regressivity (adverse case 3) .
- The income concentration curve or the Lorenz curve
1 Aristote, Nicomachean Ethics, Book V.9, 4th century BC.for the population ( Ginc);
Table A
Kakwani index 2 0 + 1
Characteristics of values Total income belongs to one The contribution is proportional Income is equally distributed
2, 0 and +1 person. to income and the contribution and financing is provided
(extreme theoretical values) Financing is provided by the rate is constant. entirely by the wealthiest
goods of people with no income person.
Kakwani index 2 < < 0 0 0 < < + 1
Intermediate value Regressive system with no Vertically equitable Non vertically equitable
characteristics vertical equity, in favour of the proportional system (CASE II) progressive system, in favour
wealthiest (CASE I) of the poorest (CASE III)
10 INSEE Studies no. 38, September 1999