Modeling the Health Insurance System of Germany Page 1
Modeling the Health Insurance System of Germany:
A System Dynamics Approach
Stefan Grosser, Dipl.-Kfm. Techn.
M.Phil. Candidate in System Dynamics at the University of Bergen!
Zeppelinstrasse 8/8, 73066 Uhingen, Germany
Phone: +49.173.7887771
Fax: +49.7161.352295
stefan.groesser@ web.de
www.stefan-groesser.de.vu
Abstract
The German Health Insurance System is balanced on the edge. Decision makers seem not
successful in developing and implementing sustainable health policies, which ensure at least
a balanced health insurance fund. Highly dynamic factors influence the health insurance fund
situation and complicate the decision making. The System Dynamics Methodology is used to
examine first possible causes of the enduring problem. In the formal simulation model, we
include among other variables the population dynamics, personal income, contribution frac-
tion and health expenses per capita as well as behavioral states of the agents. Second, the
model is used to conduct simulation-based policy testing to find improved decision rules. The
policy ‘expenses reduction pressure’ forces the government to reduce health insurance ex-
penses per request. It can improve the health insurance system situation best. The result will
be a reduction of the health insurance fund shortfall. Other policies worsen the problem sig-
nificantly due to increased oscillatory tendency in the health insurance system. As result of
the study, the different policies are discussed separately.
Keywords: — Soft System Dynamics, German Health Insurance, Sustainable Policy,
Co-Payment Policy
1. Introduction
1.1 Problem Context
In the 19" century, Industrialization gained momentum and automation forced mechanical
thinking. The microeconomic production theory evolved, which supported the idea of human
substitution by machine. Due to these facts, the employers considered employees as basic
production factors, similar to financial funds and raw materials. As a result, production com-
panies exploited their workers. This resulted in the deteriorating living circumstances, espe-
cially for the working class and most of these people lived under the poverty line. As a result,
they did not have access to health insurance and other social security systems.
To overcome main social problems, the German Emperor Karl Bismarck established the first
German State Social Security System in 1883. German citizens, with less income than a de-
fined amount of money, automatically became a member of the Social Security System. One
of Bismarck’s main ideas was the ‘Social Concept’, where the peoples’ contribution payments
depended on the individuals’ income situation rather than upon the individuals’ requests
''The author gives credit to Prof. Dr. P. I. Davidsen, Dr. M. Saleh and Ph.D. Student A. Siouxion, all members at the University of Ber-
gen/Norway, for their supervision.
Modeling the Health Insurance System of Germany Page 2
(Lindner 2003). This policy ensured that the poor and the sick got adequate financial and
medical support. The historical development in Germany shows the efficiency and success of
this system during the early 20" century. The 1883 introduced Social Security System con-
sisted of four pillars: Health Insurance, Pension Insurance, Unemployment Insurance, and
Accident Insurance (AOK-BV 2004b).
A modern definition of the Social Security System confirms, in general, the concept men-
tioned above.
"The Social Security System is a federal created, compulsive prevention system. The
main task is both to avoid the occurrence of specific risks and, in case of occurrence of
these risks, compensating unplanned expenditures and losses of income, under the
consideration of social goals." (G6rres-Gesellschaft 1989)
The general system structure of the Bismarck Social Security System still exists today. The
government, however, has introduced additional nursing care insurance, as the fifth pillar in
1994. An overwhelming demand for health insurance, especially by retired people, has
drained the health insurance fund drastically. Thus, this new pillar of the Social Security Sys-
tem was introduced to cover, in particular, nursing and nursing home expenses. Since then,
these insurance efforts are no longer covered by the state health insurance (Schnabel 2004).
Figure 1 and Figure 2 show the development of income and expenditure situation for the State
Health Insurance Fund, over the past eleven years. This situation is similar for the other insur-
ance pillars.
1404 —
= 1:2
© 130}
ME 2 =1—Contrbutions
= 120) st
2 =y=Health_expenses
1104
1.995 2.000 [Year]
Time
Figure 1: Development of Contributions and Health Expenses over the past eleven Y ears
The graph shows a continuous growth both in contributions and health expenses, with an av-
erage growth rate of 2.8 %/year. In Figure 2, the graph shows the development of the profit
and the contribution fraction. In general, both variables show a negative correlation. The
higher the health insurance fund profit per year, the lower the contribution fraction will be and
vice versa.
gi 60
Baye 144
a
8
Ea 06 1, ==? rofit
Wage 13.6 és =y—Contribution_ fraction
B-- 50) 172
== 13,1f
1.995 2.000
Time [Y ear]
Figure 2: Comparison of Profit and Contribution Faction over the past eleven Y ears
(Bundesministerium fiir Gesundheit und Soziale Sicherung 2003)
Modeling the Health Insurance System of Germany Page 3
1.2 Problem Statement
The Social Security Fund has in average a lack of 1.3 billion Euros every year (Bundesminis-
terium fiir Gesundheit und Soziale Sicherung 2003). A commission under the supervision of
Roman Herzog, former President of Germany, concluded that the German Social Security
System is balancing on the edge (dpa 2003). In order to preserve the social insurance principle
on the one hand and maintain high service quality simultaneously on the other hand, a drastic
intervention seems to be necessary. Several trials by the German Government to reduce the
lack of money have not been successful, even though, they have changed the system parame-
ters, mainly the contribution factor, in order to try to generate enough funds. With regards to
Figure 1 and Figure 2, this shows that the policy changes have not to been successful. The
question then arises, as to why it is not possible to stabilize the system and maintain zero
profit equilibrium. Is the reason for the failure of the responsible institution founded in the
system itself? Which intervention is then adequate to solve the lack of funding problem per-
manently? In order to find a solution to this question, this study was conducted.
We decided to work on this topic because problems in the Social Security Systems evoke se-
rious side effects in other sectors of the economy. Germany faces the burden of high indirect
labor costs, which reduce the global competitiveness of the German Industry. On a global
scale, Germany is one of the countries with the highest labor costs (dpa 2003). A possible
competition advantage shrinks remarkably. Less workforce demand, due to lower exports,
will result in higher unemployment ratios. Smaller economic growth or even economic stag-
nation is highly possible.
The Social Security System, in general, and the Health Insurance System, in particular, is par-
tially responsible for the higher indirect labor cost development. Because of the strong impact
on several sectors within the country, it is worthwhile to supply in-depth research to find
causes and improved policies. Each of the five sub-systems of the Social Security System has
its own structure and characteristics, which differs from the other systems. On account of this,
we decided to model one system in detail.
1.3 Adequacy of System Dynamics
Why is System Dynamics applicable to the problem? System Dynamics is applicable, because
it and System Thinking are methodologies that can be used for solving real world problems,
which highly depend upon feedback structure. System Dynamics includes concepts, which are
mainly known in control theory as the mentioned feedback structure, variable interdepen-
dency and dynamic systems with strongly changing effects. In addition, System Thinking
considers both long-term and short-term effects. System Dynamics is not applicable as a fore-
casting method, but it can be used to draw reasonable future behavior, depending upon the
structure of the system. Because the German Health Insurance System structure has not
changed significantly, the System Dynamics methodology was used firstly to replicate the
structure and secondly to find optimal policies for solving the funding problem mentioned
above. For replication of the system structure, feedback loops representing long and short-
term behavior were considered. In addition, System Dynamics is helpful, because of its simu-
lation-based theory. It is possible to evaluate simulation-based scenarios, in order to find bet-
ter policies.
On the other hand exist weaknesses of System Dynamics, which can effect the applicability of
the model outcome to the real world. The Social Security System, as well as the Health Insur-
ance System, is managed by humans, who use a rather limited rational (Sterman 2000). It is
Modeling the Health Insurance System of Germany Page 4
highly impossible to model irrational and sometimes arbitrary policies, because there is no
structural foundation existent for this behavior. In summary, we think that the insights a Sys-
tem Dynamics model gives prevails the uncertainty due to the model structure.
1.4 Time Horizon
The time horizon, the period of time over which the study is accomplished, is set to be 100
years. The point of departure is the year 2001, because the needed data to build the system
structure is available then. The study will range until 2101. The long time horizon period en-
ables us to picture possible future scenarios, which include long-term dynamic effects like
population development as well as short-term dynamics such as contribution adjustments.
1.5 Goals of the Study
The overall goal of this study is to find a sustainable policy for the health insurance system,
considering specific factors for Germany, such as the unemployment ratio and the population
development. The creation of a feasible model for the German Health Insurance System is the
first step in achieving the goal. Secondly, the model should support decision makers in the
fields of fund management and social health politics. Therefore, several scenario analyses will
be conducted in order to find sustainable problem solving policies. A discussion of these poli-
cies and conclusions drawn from these discussions will conclude this study.
2. Dynamic Hypothesis
In the following, important feedback loops of the model will be represented as Causal Loop
Diagrams. The purpose of a causal loop diagram is to expose significant feedback loops
within the model boundary and plot chains of feedback (Wiener 1957). Compared with a
stock and flow diagram, the causal loop diagram has the ability to show the overall inter-
relationships between model variables and it can capture both short and long-term feedback
loops. Feedback can be described with Wiener as: “To avoid the dangers inherent in this con-
tingency, every effect, switch, or signal is attached to tell-tale back in the signal tower, which
conveys to the signalman its actual states and performance" (Wiener 1957).
Balancing Loop B1 ‘adjusting contribution fraction’
health ins uran :
heal fund goal
wo MW
contributions fund shortfall
adjusting
a contribution fraction
contribution +
fraction
Figure 3: Loop B1 ‘adjusting contribution fraction’
The point of departure is a fund shortfall, which is calculated between the health insurance
fund and the fund goal. A fund shortfall leads to an increase in the contribution fraction to
minimize the shortfall and close a possible gap between the health insurance fund and the
fund goal. The increase in the contribution fraction will consequently result in higher contri-
Modeling the Health Insurance System of Germany Page 5
butions per year. Furthermore, the health insurance fund will receive additional funds and
simultaneously the fund shortfall will be minimized (Figure 3).
Balancing Loop B2 ‘additional funds through co-payment’
health insurance fund goal
+ ~~ fund
©»
contributions fund shortfall
additional funds
a co-payment
co-payment.
fraction
Figure 4: Loop B2 ‘additional funds through co-payment’
The balancing loop B2 tries to increase the contributions when there is a fund shortfall, by the
means of a co-payment policy. A significant fund shortfall will cause an increase in the co-
payment fraction, which will lead to higher contributions. If the contributions are higher, the
health insurance fund will be supported with more financial means, which will in tum reduce
the fund shortfall (Figure 4).
Balancing Loop B3 ‘pressure on government reduction health expenses’
=
insurance efforts
health insurance
- fund fund goal
42)
health expenses per pressure on government z
reduce health expenses +
oa fund shortfall
+
expenses reducti
pressure
Figure 5: Loop B3 ‘pressure on government reduces health expenses’
A permanent fund shortfall creates over time pressure on the government to reduce the short-
fall. One possibility for the government is to reduce the health expenses by reducing the num-
ber of health requests. In Germany, we assume this alternative as not likely, because the direct
reduction and denial of health insurance requests would not be considered politically accept-
able. The pressure to reduce increased expenses will then result in a decrease of average
health expenses per request. Consequently, the insurance effort will decline, which will
shortly lead to an increase in the health insurance fund. The computed fund shortfall will de-
crease (Figure 5).
Balancing Loop B4 ‘co-payment reduces requests’
health insurance fund goal
fund
insurance efforts :
B4 i
* 4») fund shortfall
co-payment
reduves requests
requests. y
co-payment
fractio
Figure 6: Loop B4 ‘co-payment reduces | requests’
Modeling the Health Insurance System of Germany Page 6
The co-payment policy variable has a second effect within the system. Figure 6 shows this
effect. The already known causal relationship between the co-payment fraction and the fund
shortfall is that the co-payment fraction will rise accordingly to the fund shortfall. A higher
co-payment fraction will lead in turn to a decline in requests, because the co-payment fraction
is a specific amount that the state insured people will have to contribute, when they utilize the
health insurance system, i.e. when they visit the doctor or when they buy medicine using a
prescription. If the requests are lowered, the health insurance expenses will also be less. This
will immediately cause the health insurance fund to rise and the fund shortfall will be mini-
mized.
Balancing Loop B5 ‘avoid unemployment by working’
fund goal
2 -
_ health insurance jy
> fund fund shortfall
insurance efforts
wy, ®
request contribution
Ay 463) fraction
staff away sick
ratio avoid unemployment
‘through working discretionary
unemployment rate income
+ E
illegal employment+ _jjlegal employment
attitude
Figure 7: Loop B5 ‘avoid unemployment by working’
The fund management tries to cover a fund shortfall by increasing the contribution fraction.
Otherwise, an increase in the contribution fraction will decrease the people’s discretionary
income, i.e. the part of their income they can spend voluntarily. It is assumed that people
build an illegal employment attitude, when they have less discretionary income, which will in
tum result in higher illegal employment. Illegally working people will fill normal job offer-
ings, especially craftsmen activities. This will cause the unemployment ratio to rise. If the
unemployment ratio is growing, people fear that they will loose their jobs and that they will
have difficulties to find another job, due to the difficult economic situation. When jobs be-
come harder to find, people tend to avoid sickness and try not to miss work. Therefore, the
staff away sick work ratio, due to illness, will decrease and consequently the health requests
will decrease as well. In time, the insurance efforts will decrease which will cause an increase
in the health insurance fund. Finally, the fund shortfall will be reduced (Figure 7).
Reinforcing Loop R1 ‘shortfall creates shortfall’
RL
+
fund shortfall hort creates fund goal
short +
Figure 8: Loop R11 ‘shortfall creates shortfall’
Reinforcing loop B1 shows the logic behind a fund goal policy, which considers the actual
shortfall as well as the historic shortfalls. If there is a fund shortfall, the fund goal will be ad-
Modeling the Health Insurance System of Germany Page 7
justed according to the fund shortfall. The result is a higher fund shortfall (Figure 8). This
reinforcing loop destabilizes the system noticeably.
Reinforcing Loop R2 ‘shortfall drives unemployment’
_—s» health insurance fund goal
contributions > fund
‘
fund shortfall
woenpofen ne #2)
A shortfall drives
unemployment +
contribution
illegal employment fraction
illegal employment iscretionary
attitude ¥ income
Figure 9: Loop R2 ‘shortfall drives unemployment’
The fund management tries to lessen the fund shortfall by increasing the contribution fraction.
This will, on the other hand, increase the labor costs, more precisely the indirect labor costs
for the employers. As a result, the employer will have to pay more for each employee, which
means that the employer will reduce or stop hiring work force. Nevertheless, we assume the
referenced unemployment ratio includes the effect of rising indirect labor costs. It is therefore
not shown in the above causal loop diagram. Furthermore, we assume the effect of small
changes for indirect labor costs on the employment decision due to health insurance contribu-
tions as marginal. We consider more important the fact that people will decide to work ille-
gally, when the contribution fraction rises and their discretionary income decreases. The ille-
gal employment is building up, due to the decrease of the discretionary money. In the follow-
ing, the illegal employment will rise, which leads to a higher unemployment ratio. The more
people are unemployed, the fewer contributions the health insurance fund will receive. There-
fore, the causal relationship between unemployment ratio and contributions is drawn nega-
tively. Lower contributions will result in a lower health insurance fund and thus a larger fund
shortfall will be created (Figure 9).
Reinforcing Loop R3 ‘contributions make it expensive’
insurance efor’ health insurance fund goal
fund
i
health expenses per fund shortfall
request contributions make
+
labour costs + contribution
—+—_——_ fraction
Figure 10: Loop R3 ‘contributions make is expensive’
An increasing fund shortfall will drive the contribution fraction. Further, this will increase the
labor costs of the employer. Higher labor costs lead to higher expenses per health request,
because the salary for the employees will rise. The medical industry branch is thus directly
responsible for the rise in health expenses per request. This will increase the insurance efforts
and therefore drain the health insurance fund. Consequently, the fund shortfall will increase
(Figure 10).
Modeling the Health Insurance System of Germany Page 8
Reinforcing Loop R4 ‘exploitation mentality’
health insurance
fund
insurance efforts
Ans)
NP +
Tequests exploitation mentality fund shortfall
+
fund goal
i
+,
tribution
exploitation 5 com
attitude | fraction
Figure 11: Loop R4 ‘exploitation mentality’
An elevated contribution fraction will form over time in humans an exploitation attitude, be-
cause they assume they have paid too much compared to their requests. This exploitation
tends to result in a higher number of requests. A higher request number will lead to higher
insurance efforts and the consequence will be a drain of the health insurance fund. To close
the shortfall, the fund management will have to increase the contribution fraction even more
(Figure 11).
2.2 Overview over the System Dynamics Model Structure
The model is divided into nine sectors, which are connected via several exchange variables. In
the following, we describe the overall rational of each sector.
The ‘contribution adjustment sector’ uses ‘health insurance fund’ and ‘actual fund goal’ to
compute the fund shortfall and adjust the contribution fraction, which will be fed into the
‘contribution calculation sector’.
In the ‘employees and retirees calculation sector’, every figure regarding population is calcu-
lated. ‘Total population’, ‘state insured employees’ and ‘state insured retirees’ are derived
from the population matrix. The last two mentioned variables, ‘contribution fraction’ and ‘co-
payment fraction’ will be used in the ‘contribution calculation sector’ to calculate the total
‘contributions’ for the main sector and ‘actual salary per employee’.
The ‘human behavior sector’ represents two human attitudes, which are depending on the
‘actual salary per employee’. The ‘illegal employment attitude’ will affect the employees and
retirees figures. The ‘exploitation attitude’ has an effect on the health insurance requests.
How many health insurance requests exist in Germany? This figure is calculated in the ‘health
insurance request calculation sector’, by mainly using the effects of the co-payment, the un-
employment ratio and the exploitation attitude.
In the ‘health insurance expenses sector’ the ‘health insurance expenses’ are calculated means
the ‘actual health requests’, the ‘actual salary per employee’ and the ‘expense reduction ef-
fect’, through pressure on the government.
Finally, the ‘government policy’ sector introduces three governmental policies trying to stabi-
lize the system. The first one is the ‘fund goal’ policy, which takes the ‘debt due to support’
into account. Another policy is introducing co-payment, which will increase contributions.
The third policy is the ‘expenses reduction pressure’ felt by the government due to a fund
shortfall. This will result in expense reduction.
Modeling the Health Insurance System of Germany Page 9
government
policies sector
health insurance fund
employees and
retirees calculation
sector
sector I
population sector
contribution
calculation sector
health insurance
expenses sector
human behavior
sector
ealth insurance
request calculation
sector
Figure 12: Overview of the Model Sectors
2.3 Model Assumptions
In order to create a functional and problem-oriented model, assumptions are needed, which
limit the complexity of reality in order to receive insights. According to Forrester, it is neces-
sary to draw a boundary chart defining endogenous, exogenous and not considered factors.
Moreover, "the closed boundary concept implies that the system behavior of interest is not
imposed from the outside but created within the boundary'( Forrester 1969). In order to pro-
duce the behavior being investigated, it is necessary to select relevant components. All other
components are considered irrelevant for the study and are therefore outside the model
boundary (Forrester 1969). Table 1 shows the model assumptions.
Endogenous variable
Exogenous variable
Excluded parameters
Policy variable
Health insurance fund
People leaving rate to
private health insurance
inflation rate
Fertility
Net immigration probability
Population decimation by
Normal net immigration
lon German government
Eepulation per age lexternal effects rate
Initial population (both Economic situation in
Employees cies teva scram (anP} Administration expenses
Retirees Death fraction Effect of EU health politics |) employment ratio
Contribution rate
iS ex ratio of newborn
[Technical and medical
innovations
co-payment fraction
Death rate
Birth probability over femal
lage
Effect of technical and
medical innovations on life
lexpectancy
Average expenses per
request
Staff away sick ratio
Reentry from private to
state health insurance
system
Federal support
Autonomy ratio
Health insurance fund goal
Direct labour costs
Health requests
Indirect labour costs
Table 1: Boundary Chart
It appears in the assumptions of the model that hard and soft system thinking is used to for-
mulate an insightful model. According to Checkland, hard system thinking is appropriate in
well defined technical problems. On the other hand, soft system thinking is more appropriate
in fuzzy ill defined situations, involving human beings and cultural considerations (Checkland
Modeling the Health Insurance System of Germany Page 10
1999). In this elaboration, both perspectives are used to formulate a quantified simulation
model (refer to chapter 3 and the simulation model in the appendix).
Exogenous Parameters
Leaving Rate to Private Health Insurance Company.
The leaving rate to private health insurance is depending on the salary per employee. In this
model, the average salary per employee, rather than the exact salary distribution, is used, be-
cause the data was not available. Therefore, the private health insurance leaving rate is con-
stant using the long-term average value.
Net Immigration Probability per Age
Net immigration probability per age is considered as exogenous parameter, as it seems highly
improbable that people decide to immigrate to Germany only because of the provided health
insurance, even if the health insurance in their country of origin is less comprehensive. We
use a reasonable probability distribution over age to include the migration dynamics.
Initial Population (both Male and Female), Death Fraction, Sex Ratio of Newbom, Birth
Probability of Females
These figures are according to actual statistical data from the German Federal Statistical Of-
fice and therefore represent the current situation in Germany. We do not consider the life ex-
pectancy explicitly, because it is inherent in the death fraction.
Staff Away Sick Ratio
We use the staff away sick ratio in the model as an exogenous, constant input, because we
assume only a unidirectional relationship between the reference staff away sick ratio and
health insurance requests. The higher the reference value, the more health requests there will
be. On the other hand, the amount of health requests or actual contribution fraction does not
affect the referenced staff away sick value.
Autonomy Ratio
The autonomy ratio is the ratio of self-employed people over employees. It is assumed con-
stant, because no relationship between contribution fraction and the decision to work self-
employed is seen.
Direct Labor Costs
We suppose no change of the direct labor costs due to the health insurance contribution frac-
tion. However, changes in both health insurance contribution fraction and co-payment frac-
tion will lead to higher indirect labor costs. This is explicitly modeled, contrary to the direct
labor costs, which are assumed constant.
Parameters Not C onsidered in the Model
Inflation Rate
Inflation is not considered in the model, because the transactions between employees, respec-
tive retirees and the health insurance fund management, as well as the expenses for health
insurance efforts all appear in the same year. The inflation rate would equally affect all
agents. Due to this, a change in the system behavior, because of considering the inflation rate,
is rather improbable. Only, if the "fund goal setting" policy were activated, would inflation
influence the accumulated “debt due to support”. For this case, we have considered the infla-
tion rate to be negligible.
Modeling the Health Insurance System of Germany Page 11
Population Decimation by External Effects
In the model, exceptional external effects, which could decimate or increase the population
significantly, i.e. wars, terrorist attacks, atomic accidents, harm by environmental change, i.e.
climate change, and deadly conterminous diseases, are excluded.
Economic Situation in Germany
We considered that the behavior of the system depended on the economic situation in Ger-
many. However, in the model only the unemployment ratio as an aggregated variable of the
economic situation in used. Other economic key values like GDP, exchange rate, import and
export of commodities and services are not included.
Effect of EU Health Politics on German Government
In the European Union (EU), different approaches are followed to harmonize the independent
national health insurance systems (Riesberg et al. 2003). Because these approaches are only
momentarily discussed, no serious activities seem to have been undertaken, therefore we de-
cided to exclude these probable effects on the German government and concentrate on the
health insurance fund management.
Technical and Medical Innovations
Technical and medical innovations are not considered, because it seems to be obvious that
innovative progresses, in both the technical and the medical field, can not cause the problem
of fund shortfall. This is because the innovations are not available to the public, during the
research stage, and are therefore relatively expensive. It is assumed that innovations will re-
place the existing technology, when they represent the same expense as the old technology.
Thus, the health expenses, due to medical and technical treatment, will be constant over the
time horizon.
Effect of Technical and Medical Innovations on Life Expectancy
It is supposed that the life expectancy is not changed by future technical and medical innova-
tions. This does not seem to represent reality but estimations about the possible effects of in-
novations on life expectancy increment are highly uncertain. We found no statistical data to
reasonably support any assumption.
Reentry from Private to State Health Insurance System
State insured people can leave the state system, if they earn more than a specific upper limit.
They can then go to a private health insurance system. The re-entering of people from the
private health insurance system to the state system is not modeled, because firstly, there is no
reliable data about this issue available and secondly we considered this number as marginal,
during the work life of the people. It is assumed that half of the employees will come back to
the state health insurance, after their work life.
Policy Variables
In contrast to the exogenous variables, there are the policy variables considered as not con-
stant over time. Furthermore, the chosen policy variables can be influenced directly by the
government and fund management, through different political programs.
Fertility
The implicated fertility rate for Germany is according to the statistical data Zahn (2004).
However, politicians have been discussing how to increase the birth rate per woman. Actual
studies show that their attempts to increase the birth rate have not been successful (Rasche
2004).
Modeling the Health Insurance System of Germany Page 12
Normal Net Immigration Rate
Current discussions in Germany concentrate on population development for the next century.
Some activities are considered in order to stop the population from declining. Besides an in-
crease in the birth rate per woman, immigration politics are getting more important. Several
altematives of the normal net immigration rate are discussed (Statistisches Bundesamt
Deutschland (2003).
Administration Expenses
Administration expenses are important for the setting of the contribution rate, because the
quota of administration expenses, regarding health expenses, reached 6% in the year 2001.
Therefore, the contribution fraction is driven by the administration as well. A sustainable ad-
ministration expense policy could help to decrease the contribution fraction permanently.
Unemployment Ratio
The unemployment ratio has an indirect effect of the contributions. Because the unemploy-
ment ratio is fluctuating in reality, with a small bandwidth, the contribution payment base is
changing constantly.
Co-payment Fraction
Co-payment fraction is an additional and fast responding means of the government. The co-
payment fraction will change both the contribution and expense side of the system very rap-
idly.
3. Model Validation and Analysis
The model validation process included both structural, behavioral and structural-behavioral
analyses. Extreme tests with the policy variables Fertility, Unemployment ratio, contribution
fraction, retirement age, labor market entrance age, and fund goal have been executed. Each
run created reasonable model behavior.
3.1 Base Run
The model settings for the base run are briefly described in the following paragraphs. For the
base run, all assumptions built in the model are activated. The administration expense fellows
a continuous linear expense growth path. Furthermore, the actual unemployment ratio affects
the staff away sick ratio and results in a decrease of health requests. In addition, adjustment of
contribution fraction has also effects on the illegal employment attitude and the exploitation
attitude. Besides, the employee and retiree number is computed from the population sector,
meaning that actual and dynamic figures are used. A growth rate for employees’ salary and
retirees’ pension is assumed. In the base run, no additional exogenous input is used.
The health insurance fund is the most important variable of the German Health Insurance Sys-
tem and is analyzed first. Compared to the reference mode, the base run shows fewer oscilla-
tions, but greater permanent fund shortfalls (Figure 13).
To understand this behavior, a detailed analysis is necessary. The population development can
be seen from Figure 14. With an initial value of more than 82 million inhabitants, the number
decreases, over time to around 58 million people. This is one of the major factors for the be-
havior of the total system. The amount of people, financing the health insurance fund, is erod-
ing every year.
Modeling the Health Insurance System of Germany Page 13
Health Insurance Fund ce
y tad
6
°
“
ry
e
we
®
6
&
ry
it
2.020 2.040 2.060 2.080 2.100
Time [Year]
Figure 13: Base Run: Health Insurance Fund
[Person]
80 000 000.
st
70 000 000
2 = —Reference_Mode_ Population
1 y= TOTAL_POPULATION
60 000 000.
st
wy
2020 2040 2 060 2080 2100
a
/|
Time [¥ ear]
Figure 14: Population Development Compared to the Population Reference Mode
In addition, administrative expenses increase every year with a constant slope of 3.92 [frac-
tion]. This second factor makes it, in combination with the population development, impossi-
ble to have a dynamic equilibrium, because the change in population figure causes a change in
contributions and health requests. On the other hand, the rise of administration expenses af-
fects only the expense side. Therefore, the inflow will not change according to the outflow.
Thus, a mismatch of inflow and outflow exists. Consequently, the health insurance fund can-
not be in dynamic equilibrium. Moreover, these dynamic effects cause the attached system to
adapt.
In the following, each side of the health insurance fund is analyzed separately, in order to ob-
tain an explanation for the health insurance fund behavior. With regards to the following
analyses, in chapter four and five, the expense and contribution rates have dimensions of
around 130 to 150 billion [euro]. The fund shortfall reference value is pushed to reality and
set to a value of one billion [euro], which is approximately 0.75 per hundred of the rates di-
mension. Small changes of this size can’t be seen in either the contribution graphs or in the
health expense graphs.
Contribution Side
At year 2001, the contribution inflow exceeds the health expenses outflow. The administra-
tion expenses cause a found shortfall, which will cause a positive shortfall ratio. As result, the
contribution rate will increase, whereby the contribution inflow rises. In addition, the actual
salary per year will increase after a time delay of four years. This also causes an increased
contribution inflow and the health insurance fund shortfall declines. Figure 15 shows parame-
ters, which primarily influence the contribution fraction. For the first five years, the variable
‘total retirees’ grows stronger than the ‘total employees’ variable due to the initial values of
Modeling the Health Insurance System of Germany Page 14
the population cohorts and depleted the employees. The fund shortfall rises, because the con-
tributions paid by employees are higher than those paid by retirees. This causes lower contri-
butions and leads to a greater fund shortfall. From year 2005 until year 2010, the total retiree
number falls. At the same time, the variable ‘total employees’ rises causing more contribu-
tions. From year 2010 until year 2050, the total amount of employees steadily decreases
whereas the amount of retirees increases until around year 2035. This leads to a reduced con-
tribution growth rate. The contribution fraction adjustment, in the beginning of year 2030, is
increased to compensate the reduction in employees. Furthermore, the contribution fraction
adjustment is steeper, when the number of retirees reaches its maximum value around the year
2035.
CONTRIBUTIONS
“=
=y-CONTRIBUTION_FRACTION
ACTUAL_SALARY_PER_EMPLO*
-
y= total_employees
total_retirees
=e
-gePension
2.020 2.040 2.060 2.080 2.100
Time [Year]
Figure 15: Base Run: Contribution Side
Health Expenses Side
Figure 16 is explained as follows. The fund shortfall causes a rising contribution fraction.
However, due to the rising contribution fraction, the exploitation attitude is building up. This
leads, in general, to more health requests (curve 3). After year 2020, this effect is significant.
It reaches the maximal value of 1.14 [dmnl] around the year 2060. This increases the health
requests by 14 percent. Contrary, due to rising contribution fraction, the illegal employment
attitude rises by 20 percent and leads to more unemployment (curve 4). The higher unem-
ployment ratio will reduce the staff away sick ratio by approximately nine percent (curve 1).
In consequence, the lower staff away sick ratio will reduce the request per capita by six per-
cent (curve 2), meaning that employees do not try to miss work. In general, both of the human
behavior effects have an asynchronous behavior. The result is an increase in health requests,
because the exploitation attitude is more effective than the illegal employment and staff away
sick effect on health requests. The result of both effects is the "multiplier request per capita"
(curve 5).
=; EFFECT_OF_ACT_UR_ON_SAS_RATIO
nynetfect_of_act_sas_ratio_on_request_per_canita
my etfect_of_exploitation_atitude_on_request_per capita
effect_of_illegal_employment_attitude_on_unemployment_
“4 ratio
ase Multiplierrequest_per_capita
2.020 2.040 2.060 2.080 2.100
Time [Year]
Figure 16: Base Run: Effects on Request per Capita
Modeling the Health Insurance System of Germany Page 15
The recently calculated ‘multiplier request per capita’ (curve 5, Figure 16 or curve 2, Figure
17) and the state insured population’ is used to calculate the actual health requests (curve 1,
Figure 17). Obviously, the amount of health requests depends strongly on the state insured
population. The request multiplier has only marginal effects. However, a net effect of seven
percent due to human behavior is considered significant. Because of the dependency of the
actual health requests and the population, the actual request number mainly follows the popu-
lation (curve 1 and curve 3).
SS
sxe 188000000
a 107 ,
mye ACTUAL, HEALTH. REQUESTS
eye 167200000 a :
ay a8 ° cam riper. request pr cnt
Se sse0n00, : mye TOTAL POPULATION
<= 146.000.000 ZA ay
”
-3- 83,000,000, 3
— 0.99
w= s4o00.000
nn
2.020 2.040 2.060 2.080 2.200
Time [Year]
Figure 17: Base Run: Calculation of Actual Health Requests
If the contribution rate is rising, so does the actual salary per employee. Figure 18 shows the
effect of the contribution fraction (curve 1) on the actual salary per employee (curve 3). From
year 2010 until year 2050, the actual employee salary is slightly less than the assumed con-
stant salary growth. From around year 2060 until year 2090, the actual salary is then slightly
higher, because of the growing contribution fraction. There is a lag time between both vari-
ables due to an information delay of four years.
“- 020
== 770,00
== 24.700
== 13.900
=) —CONTRIBUTION_FRACTION
Sled 0,16 e 7 heatth .
average_expenses_per_health_request
== 704,00 agen SSIES -RMEETRES PSI EPELIE
== 73.380 =y—ACTUAL_SALARY_PER_EMPLOYEE
“y= 13.120 =y—Pension
aoe; bie
== 660,00
== 22500
== 12.600
2.020 2,040 2.060 2.080 2.100
Time [Year]
Figure 18: Base Run: Calculation of Average Expenses per Health Request
Pension payments are not effected in value by changing the contribution rate. Therefore, pen-
sions grow only with a constant slope, due to the salary and pension growth rate (curve 4).
In accordance with the actual salary per year, the average expenses per health request rise
significantly from 660 [euro] to 770 [euro] and saturate around this level (curve 2). The satu-
2-The total population is shown as curve 3, The state insured population is simply the 0.75 percent fraction of the total population, We used
the graph of the total population, because the state insured population is not available as a single value, only as an array. In addition, the
behaviour, over time, is the same for both variables.
Modeling the Health Insurance System of Germany Page 16
ration occurs less, because of a saturating contribution fraction, than the maximum increment
in health expenses per health request of an assumed 10 percent.
aS 1,38e11
== 199.000.000 x ‘i
a
we eet / \ =, -HEALTH_INSURANCE_EXPENSES
:
== 167.200.000 =p ACTUAL_HEALTH_REQUESTS
3° 704 = y—2verage_expenses_per_health request
3
= gett)
=2= 146.000.000 1
= 660
2020 2.080 © 2.060 2.080 2.100
Time [Year]
Figure 19: Base Run: Calculation of Health Insurance Expenses
As result for the health expense side, Figure 19 shows the development of health insurance
expenses (curve 1). Actual health request behavior (curve 2) depends mainly on the popula-
tion development, with slight adaptations from several effects. The average expenses per
health request (curve 3) have significance, especially during the period from year 2001 until
year 2060. After year 2060, the further rising of the contribution fraction is not reflected in the
average expenses per health request, due to the assumption of a maximum increase of 10 per-
cent.
Figure 20 shows the contribution fraction development (curve 2) compared to the reference
mode for the contribution fraction (curve 1). Obviously, the contribution fraction develop-
ment does not depend on the contribution fraction adjustment delay as assumed. Therefore,
the actual contribution fraction is smoothed. Nevertheless, the adjustment height of the frac-
tion is similar in both cases. Thus, the long-term development is well captured by the refer-
ence mode.
0,21
0,1!
0,18-
0,17
0,1
Contribution Fraction
0,15.
0,14-
2.020 2.040 2.060 2.080 2.100
Time [Y ear]
Figure 20: Base Run: Contribution Fraction
3.2 Sensitivity Analysis
Sensitivity analysis comprises, in contrast to the extreme value analysis, moderate changes in
specific variables, which are reasonable for the reality. Therefore, practical and graspable pa-
rameters are chosen and varied in feasible ranges. Policy improvements can also he derived
from sensitivity analyses.
Modeling the Health Insurance System of Germany Page 17
Fertility Rate
For this run, different fertility rates are used to analyze the sensitivity of the model to this
variable.
[Person
Time
Figure 21: Sensitivity Analysis: Fertility Rate
The variable ‘fertility’ is changed in the range from 1.0 to 2.0 [children/women]. The value of
1.4 [children/women] signifies the base run value, which is the actual statistical value in Ger-
many (2001). Figure 21 and Figure 22 show the results for the population development and
the health insurance fund. The fertility rate measured in children/woman for the different fer-
tility rate assumptions are: 1.0 (curve 1), 1.3 (curve 2), 1.4 (curve 3), 1.5 (curve 4), 1.8 (curve
5) and 2.0 (curve 6.)
Figure 21 shows that a fertility of 1.8 is needed to stabilize the population development
around the starting value of 82 million people. A fertility rate of 2.0 would lead to an increase
in population near to 100 million people by the year 2101. Any value below the equilibrium
value of 1.8 leads consequently to a decline of the population.
The result for each fertility assumption is shown in Figure 22. Apparently, a higher fertility
ratio worsens the situation at the very beginning of the simulation. Curve 6 (2.0 chil-
dren/woman), during the first 20 years, is permanently and significantly lower than curve 1
(1.0 children/woman). The following reasoning is behind the contra intuitive behavior: The
average age entering the labor market is set to ‘18 [year]’. Therefore, the newborns can first
enter the labor market after this 18-year period. Until then, they are consumers of health in-
surance efforts. This causes a stronger negative shortfall in the health insurance fund. After
the year 2045, the situation is tumed. The higher the fertility rate, the smaller the fund short-
fall, because the higher number of employees and retirees contributes more contribution pay-
ments.
(Wear)
Figure 22: Sensitivity analysis: fertility
Modeling the Health Insurance System of Germany Page 18
This sensitivity analyses gives important insights. If the government is able to increase the
fertility rate abruptly, it has to face the first 20 years of higher health expenses. The situation
will be improved after a time period of around 45 years.
Net Immigration Rate
In the following, the sensitivity of the model regarding the net immigration rate is analyzed.
The used values of 100.000 people/year (curve 1), 200.000 people/year (curve 2; base run)
and 300.000 people/year (curve 3) are derived from a study of the German Population Devel-
opment Office (Statistisches Bundesamt Deutschland (2003)).
Figure 23 shows the effect of the different net immigration rates on the total population. Evi-
dently, the following relationship is valid: The higher the net immigration rate, the higher the
total population. A change in the net immigration rate of 100.000 people/year will result in a
15 million people higher population after 100 years. In order to stabilize the actual population
of Germany, a net immigration rate of around 400.000 people/year is needed.
~ ~ = ee
Figure 23: Sensitivity Analysis: Net Immigration Rate
In Figure 24, the effect of increasing the immigration rate in contrast to the birth rate is seen.
The higher the net immigration rate, the lower the fund shortfall on a permanent basis. This is
contrary to the birth rate effect, which is shown in Figure 22.
This behavior is reasonable, because most of the immigrating people are older than 18 years
of age. If they immigrate to Germany, they will immediately enter the labor market and pay
contributions to health insurance fund. There is no childhood time delay, which causes health
requests and expenses, but no contributions.
This analysis gives additional insights. By increasing the net immigration rate, it is possible to
minimize the fund shortfall, because mainly adult people will immigrate, which can instanta-
neously start to work. Nevertheless, there is a very strong side effect, which decision makers
must consider. By increasing the number of inhabitants only through net immigration, the
national ability to sustain the population will be reduced. Consequently, the country would be
depending on the immigration.
(eUR] oo
oe
veg 7
Figure 24: Sensitivity Analysis: Net Immigration Rate
Modeling the Health Insurance System of Germany Page 19
Reference C ontribution Fraction
As a third sensitivity variable, the reference contribution fraction is selected, because we as-
sume a strong sensitivity of this variable on the health insurance fund. To test the sensitivity
five different values are used. Figure 25 shows the different developments of the contribution
fraction for the runs (starting value of 0.12 percent, curve 1; 0.13 percent, curve 2; 0.1358
percent, curve 3; 0.14 percent, curve 4; 0.15 percent, curve 5).
The following development can be seen: If the reference contribution fraction is higher at the
beginning than the actual reference contribution fraction (0.1358 [fraction]), an adjustment
process will reduce the fraction (curve 4 and 5).
oa
\
\
2.020 2040 2.060 2.080 2.00
Time
Figure 25: Sensitivity Analysis: Reference Contribution Fraction
This leads to a permanent lower contribution fraction due to the first correction. For this case,
the reference contribution fraction is set lower than the actual reference value. In the long run,
the fraction will increase steadily with higher values for the contribution fraction (curve 1 and
2).
The effect of the different reference contribution fractions on the health insurance fund is
shown in Figure 26. If the reference value is either higher or lower than the reference contri-
bution fraction for the base run (0.1358 [fraction]), a positive or negative health insurance
fund will appear, which will lead to the contribution fraction adjustments. Curve 4 and curve
5 show positive, whereas curve 1 and curve 2 shows negative, adaptation processes. In the
end, the health insurance fund reaches, in either case, a steady state value in a bandwidth of
one million Euros.
3
HEALTH_INSURANCE_FUND
ee Se
————
v
2.020 2.080 2.060 2.080 2.100
Time [Year]
Figure 26: Sensitivity Analysis: Reference Contribution Fraction
Recapitulate, the initial value of the reference contribution fraction is the point of departure
for the behavior of the total system. The health insurance fund will reach in the end nearly the
Modeling the Health Insurance System of Germany Page 20
same value for different starting values, even thought, the single contribution fractions are
different for each starting value.
Contribution Fraction Adjustment Effect
For this sensitivity analysis, the nonlinear relationship ‘effect of shortfall ratio on contribution
adjustment fraction’ is slightly changed. This relationship is assumed to be strong sensitive
parameter. Small changes in the contribution fraction lead to larger changes in the contribu-
tions due to the large total income figures.
Four different runs are conducted. The first run is the base run, with the nonlinear relationship
seen in. Figure 27 shows the impact of the different relationships on the contribution fraction,
whereas the curve represents the base run, curve 2 to curve 4 represents the different varia-
tions of the nonlinear relationship. The chosen assumptions about the relationships cause
slightly different contribution fractions, especially during the years between 2040 and 2070.
Moreover, the lower the slope of the nonlinear relationship is, the fewer the adjustments of
the contribution fraction (curve 2 compared to curve 1).
Figure 28 shows the effects of the accomplished changes on the health insurance fund.
Changes with only 0.01 - 0.02 [fraction] in magnitude, causes an additional fund shortfall of
one million at its maximum difference around the year 2055.
ono _—
aE
oas|
—_
re}
one} Ta
oa] f
as]
ll
2020 2.080 2.060 2.080 2100
{Year}
CONTRIBUTION. FRACTION
Figure 27: Sensitivity Analysis: Contribution Fraction Adjustment Effect
{EUR}
HEALTH_INSURANCE_FUND
2020 2060 7280 200
Figure 28: Sensitivity Analysis: Contribution Fraction Adjustment Effect
In conclusion, the health insurance fund, as one of the most important entities of the model, is
highly dependant on the effect of the shortfall on the contribution adjustment. This effect
represents the policy of the fund management and how to adjust the contribution according to
a fund shortfall. In the sensitivity analysis, a higher slope of the contribution adjustment effect
seems to be preferable in order to close the fund shortfall. However, a higher slope means
Modeling the Health Insurance System of Germany Page 21
stronger adjustment effects, if there are greater shortfalls. The system tendency to oscillate is
strengthened by stronger contribution adjustment effects.
4. Policy Testing
Policy testing can be done by two different approaches. The first approach is by changing
parameters that already exist in the model. We consider this as soft policy testing, because the
model remains the same structure while it is tested.
The second approach is changing the structure of the system, by introducing new feedback
loops, by means of a new system structure. This is considered structural policy testing.
4.1 Soft Policy Testing
In chapter 3.4, ‘sensitivity analysis’, we have already analyzed the effects of several changes
of model parameters, such as the fertility rate, net immigration rate and contribution fraction.
The main insights are summarized in this chapter. Further soft policy testing runs are possible,
ie. for retirement age and average age entering the labor market.
The fertility rate policies face a time delay of about 20 years until the effects can be seen. This
means that the situation during the first 20 years will be worsened rather than improved. In
terms of the legislation period, it is impossible to build party politics only on this parameter,
because almost five legislation periods are needed to see the results. However, after this pol-
icy delay period a sustained improved situation is present.
The net immigration rate policies show results in the short and long-term perspective. How-
ever, it is necessary to challenge one underlying assumption. The preferred net immigration
rate is a demanded amount. In this model, it is assumed that the demand will be fulfilled per-
manently. Over a time horizon of one century, it is highly improbable that in reality the de-
mand will be fulfilled constantly.
Contribution fraction changes are the easiest way of changing the system behavior. Therefore,
the health insurance fund management mostly uses this type of adjustment. But, as the sensi-
tivity analyses shows (Figure 25 till Figure 28), the contribution fraction is a highly effective
leverage in this system. It is very easy to cause oscillatory behavior, due to over adjustments.
4.2 Structural Policy Testing
‘Health Insurance Fund Goal Adjustment’ Policy
The ‘health insurance fund goal adjustment’-policy considers the fund debt, which is created
due to federal support. In the stock and flow diagram of the main sector, the stock ‘debt due to
support’ represents the accumulated federal support.
This policy uses the created debt to formulate the actual fund goal for the health insurance
fund. The stock and flow diagram for this policy is included in the ‘government activities’
sector).
Figure 29 represents the system behavior, with and without the described policy. If the policy
is inactive, the actual fund goal (curve 1) is zero due to the fact that the support (curve 3) rises
according to the fund shortfall (curve 1, Figure 30).
Modeling the Health Insurance System of Germany Page 22
If the policy is switched on, the actual fund goal (curve 2) is a time-delayed blueprint of the
‘debt due to support’ variable.
[EUR] 3e11-
2ell- —_—
—_—_—
2ell- 3 my ACTUAL_FUND_GOAL
a faye ACTUAL_FUND_GOAL
dell. get 5 @
, 2g DEBT_DUE_TO_SUPPOF
510 N y= DEBT_DUE_TO_SUPPOF
of ) 4
2
0 se 4: 4 4204 1a
2.020 2.040 2.060 2.080 2.100
Time [Year]
1,3 base run
2,4 goal formulation policy
Figure 29: Health Insurance Fund Goal A djustment Policy: Debt and Actual Fund Goal
Figure 30 shows the result for the health insurance fund. The health insurance fund, without
the ‘goal setting’-policy shows small negative values (curve 1, figure 64 or more detailed:
curve 2, Figure 13). The health insurance fund with active policy, shows, especially during
the years from 2065 to 2092, strong overreactions due to strong contribution fraction adjust-
ments.
(EURI
UND
se104
HEALTH_INSURANCE_F &
2.020 2.040 2.060 2.080 2.100
Time [Year]
base run
2 goal formulation policy
Figure 30: Health Insurance Fund Goal Adjustment Policy: Health Insurance Fund
This policy is not adequate to improve the health insurance system, because a possible fund
shortfall will increase the "debt to due support" variable, which causes a larger fund shortfall
and thus stronger adjustments in the contribution fraction. This reinforcing loop leads to over-
reactions in the contribution fraction adjustment, due to the delayed adaptation process and
the repayment condition. Only if the health insurance fund has a positive balance, can repay-
ments be made. Because the system is highly sensitive to the contribution fraction, the over
adjustments occur and will destabilize the system.
‘Corner Cutting Pressure on Government’ Policy
If the prices for health products grow either to fast or to high, the government must take ac-
tions to contain these expenses by increasing the effects on the health insurance system
(Bundesministerium fiir Gesundheit und Soziale Sicherheit 2004). This assumption underlies
the ‘comer cutting pressure on government’ policy. The pressure on the government occurs,
because of a shortfall in the health insurance fund and the government has to support the
health insurance fund. The governmental budget is limited, meaning that a reallocation of the
financial means, is needed in order to support the fund. This means that budget reductions
will be necessary in other sectors of the national economy. These other sectors will create
pressure on the government to reduce the lack of money that the health insurance fund needs,
in order to avoid budget reallocations.
Modeling the Health Insurance System of Germany Page 23
[EUR]
average_expenses,
per_health_request
2020 2.040 2.060 2.080 2100
Time [Year]
1 base run
2 expenses reduction pressure
Figure 31: Comer Cutting Pressure Policy: Average Expenses per Health Request
Figure 31 shows the successful reduction of the average health expenses per health request
(curve 2) compared to the average expenses per health requests during the base run (curve 1).
If the policy is activated the fund shortfall will create expense reduction pressure, which will
result in a health expense reduction effect. As a result, the health insurance expenses will de-
cline (curve 4 compared with curve 3). Because of lower health expenses, the contribution
fraction can be lower than in the base run, where the policy deactivated. Figure 33 represents
the described behavior of the contribution fraction. Lower contributions will follow in se-
quence with the lower contribution fraction (Figure 32, compared curve 1 and curve 2).
[EUR]
a a 4 aye CONTRIBUTIONS
7 ze : CONTRIBUTIONS
3 ss .
irre 3 ee oe
lel1- Te jy =ge HEALTH_INSURANCE_EXPENSE
Jel1. —— my HEALTH_INSURANCE_EXPENSE
2.020 2.040 2.060 2.080 2.100
Time [Year]
1,3 base run
2,4 expenses reduction pressure
Figure 32: Comer Cutting Pressure Policy: Contributions and Health Insurance Expenses
CONTRIBUTION_FRAC
TION
oc os
> oe
Po
™
\
\
N
2020 2.040 2.060 2.080 2100
Time
1 base run
2 expenses reduction pressure
Figure 33: Comer Cutting Pressure Policy: Contribution Fraction
The resulting effect of the ‘expense reduction pressure’ policy on the health insurance fund is
shown in Figure 34. Due to lower health expenses, the fund shortfall is lower until year 2060.
Because of the permanent lower contribution fraction, the fund shortfall cannot be reduced as
quickly as without the policy.
Modeling the Health Insurance System of Germany Page 24
UND
\
|
2020 2040 2060 2.080 2100
Time [Year]
base run
2 expenses reduction pressure
Figure 34: Comer Cutting Pressure Policy: Health Insurance Fund
‘Co-Payment’ Policy
Another answer from the government on a fund shortfall could be the introduction of a co-
payment policy. People have to pay additional contributions to the fund, which are independ-
ent from the original contribution fraction, when there is a fund shortfall. The stock and flow
representation of this policy is in the ‘government activities’ sector. The reference co-
payment fraction is set to 0.01 [fraction].
If the policy is active, the original contribution fraction (Figure 35, curve 1) is reduced (curve
2), because the co-payment fraction, as an additional inflow to contributions, partially substi-
tutes the contribution fraction. The variable ‘sum of contribution fraction’ (curve 3) represents
both inflows. Due to the higher adjustment rate, of the co-payment fraction, the contribution
fraction of the system begins to oscillate, because even a small shortfall ratio leads to quick
and over adjusting responses. After the year 2040, the damped oscillations find a steady state
around a small negative shortfall with oscillations small in amplitude and long in frequency
(Figure 36).
0,20 _——
a
0,18 w
Ea CONTRIBUTION. FRACTION
2
ad ‘ —_—— —,—sum of contribution traction
—<—_
a a per
a
2.020 2.040 2.060 2,080 2.100
Time [Year]
Lbase run
2,3 co-payment policy
Figure 35: Co-Payment Policy: Contribution Fraction
wig UR
Zz
w
2
QO
2 5ed-
q
«
2
g |
* ee ,
ze y as
F -5e9.
z
a
=
2020 2040 2060 2080 2100
Time [Year]
base run
2 co-payment policy
Figure 36: Co-payment Policy: Health Insurance Fund
Modeling the Health Insurance System of Germany Page 25
By introduction of a co-payment fraction, the oscillatory tendency of the system is amplified.
If the system is exposed to an external shock, the probability to oscillate is higher with the co-
payment fraction policy than without it. Therefore, the policy is not considered as useful to
stabilize the system and strengthen its shortfall behavior.
‘Fixed Contribution’ Policy
The fixed contribution policy is yet to be discussed between the German Government and the
opposition. The major idea is not to determine the contribution payments based on the fund
shortfall, since the introduction of the health insurance system by Bismarck (AOK-BV
2004a). The parties discuss whether the contributions should be a constant amount of money
per employed capita, therefore the name ‘Capitation’ is used. To model this, we use the exist-
ing circumstances in the model that represent an average employee salary income. A constant
contribution fraction combined with an average constant salary is equal to a constant amount
of money for each employee, i.e. ‘The Capitation’. However, to maintain a constant salary, it
is necessary to change the model structure. The "salary and pension growth rate" is no longer
directly connected to the actual salary per employee. In order to obtain a constant salary
value, this link has to be erased. The switch "sw_contribution_policy" can activate the "fixed
contribution"-policy. A constant contribution fraction of 0.15 [fraction] is assumed.
Figure 37 shows the result of the "fixed contribution"-policy for the contributions and the
health insurance expenses. The contributions, with the fixed contribution fraction (curve 2),
are from 2001 until 2040 and are higher than the contributions collected in the base run with
the flexible contribution fraction (curve 1). This is simply because the fixed contribution frac-
tion is set to 0.15 [fraction], whereas the base run contribution fraction starting value is set to
0.1358 [fraction]. After the year 2035, the “flexible contribution” policy causes more contri-
butions, because the contribution fraction increases to values higher than 0.15 [fraction]
(Figure 38, curve 1 compared with curve 2). Instead, contributions due to the fixed contribu-
tion fraction decline due to the population development.
A comparison of average expenses per health request shows that at the beginning a nearly
identical development. After the year 2040, the difference becomes significant. The health
insurance expenses produced by the base run (with flexible contribution fraction, curve 3),
exceed the health expenses for the system with a fixed contribution fraction (curve 4).
(EUR) 2e11
ee
nar a ——____.,
1°
: a ae =) — CONTRIBUTIONS
zs
Jell- att a Se =p CONTRIBUTIONS
3 My HEALTH_INSURANCE _EXPENSES
HEALTH_INSURANCE_EXPENSES
an
2.020 2.040 2.060 2.080 2.100
Time [Year]
1,3 base run
2.4 fixed contibution policy
Figure 37: Fixed Contribution Policy: Contributions and Health Insurance Expenses
v2
32 on _ ee
ra i
2
3 ol ———
B le
rd
E o1 7 6
3 Co
2.020 2.040 2.060 2.080 2.100
Time {Year}
Lbase run
2 fixed contribution policy
Figure 38: Fixed Contribution Policy: Contribution Fraction
Modeling the Health Insurance System of Germany Page 26
The reason for fewer health expenses is the reduced effect of the contribution fraction on the
average expenses per health request (Figure 39). However, in both cases, the average ex-
penses per request reach its maximum value of around 770 Euros. Nevertheless, a difference
exists in the total health expenses between the two policies. This is due to the lower health
requests with a constant contribution fraction, because the lower contribution fraction causes
a lower exploitation attitude and fewer health requests. Therefore, the total amount of health
request and health expenses is reduced, when the ‘fixed contribution’ -policy is activated.
{EUR}
‘ : 15 <= ™ ™
2a 2!
ti ie
o£ 70 Ae
e
g
a 650.
2.020 2.040 2.060 2,080 2.100
Time [Year]
1 base run
2 fixed contribution policy
Figure 39: Fixed contribution policy: average expenses per health request
Figure 40 shows the effect of the discussed policy regarding the health insurance fund. Sev-
eral effects cause abrupt change. Firstly, the population development curve shows the time of
its inflection point (Figure 14, curve 1). Secondly, the fraction of retirees declines faster than
the fraction of employees due to the initial population values. Consequently, the amount of
requests is reduced and the contribution inflow is increased.
[EUR]
HEALTH_INSURANCE_|
UN
iN
Time [Year]
base run
2 fixed contribution policy
Figure 40: Fixed Contribution Policy: Health Insurance Fund
The result of this policy is simply the cancellation of the balancing feedback loop B1 "adjust-
ing contribution fraction". By this means, the oscillatory tendency of the system can be re-
duced. However, any chosen fixed contribution fraction will not be lasting or correct of ever.
Figure 40 shows that the incipiently contribution fraction is higher than the actual necessary
value. Circumstances change and make contribution adjustments necessary, especially be-
cause of the population development. It is highly uncertain that with incomplete information
the chosen fixed contribution fraction can fulfill the demand.
‘Per Capita Administration Expenses’ Policy
The fifth discussed policy stresses the administration expense development. The administra-
tion costs rise with an almost constant slope every year. In the base run model, it is assumed
that a permanent development of the expenses is used in the same manner. A more realistic
assumption takes the amount of state insured people in account. The amount of administrated
people acts as cost driver.’ The more people that are insured, the more expenses are generated
* This is an application of the activity-based-costing theory.
Modeling the Health Insurance System of Germany Page 27
and vice versa. Administration expenses per capita are calculated and multiplied with the ac-
tual population figure. *
[EUR]
2ell. ent 1:
a:
4 a =) — CONTRIBUTIONS
Vs *
— =~ CONTRIBUTIONS
tel =,
my HEALTH_INSURANCE_EXPENSES
my—HEALTH_INSURANCE_EXPENSES
sel as administration expenses
—_——" ‘administration_expenses
—— —' x
— oo
2020 2040 2.060 2080 2100
Time {Year}
1.3.5 base run
2,4,6 administration expenses policy
Figure 41: Per Capita A dministration Expense Policy: Contributions and Expenses
As seen in Figure 41, the administration expenses in the base run grow constantly with a fixed
slope (curve 5). Contrary, the population dependant administration expenses (curve 6) follow
the population development. The lower administration expenses lead to a lower shortfall and
therefore to lower contribution fractions (Figure 42, curve 2) than the contribution fraction in
the base run (curve 1). Consequently, the health expenses will decline and in sequence the
total health requests will decline, too (Figure 41, curve 4 compared with curve 3).
¥8
EF 020 —_—_—
2! —_—
Oo os. ve” a
—
a
Zz (Olé. 1!
5 sae —_——
8 ————
c 014 lt a
2020 2040 2060 2080 2100
Time [Year]
‘1 base run
2 administration expense policy
Figure 42: Per Capita Administration Expense Policy: Contribution Fraction
Figure 43 shows the effect of these developments on the health insurance fund. From year
2001 until approximately year 2072, the behavior is satisfying from a fund management per-
spective. In the following year, the total population decline causes less administration ex-
penses, which lead to an increase in the health insurance fund. Consequently, the contribution
fraction decreases. Because of the falling contribution fraction, the effect of exploitation atti-
tude on requests per capita declines, which reduces the number of requests. Therefore, a sur-
plus in the health insurance fund is created, which will be reduced by the contribution fraction
adjustment (curve 2).
“Calculated expenses per capita per year by mean of data (2001): 7.640.000.000 [euro/year]/ 70.994.000 [insured people] = 107.61
[euro/year/people].
Modeling the Health Insurance System of Germany Page 28
[EUR]
UND
HEALTH_INSURANCE_F
i .
2 i ‘i 4 <i.
ee atl
2020 2040 2060 2080 2100
Time Tear]
base un
2 administration expense policy
Figure 43: Per Capita Administration Expense Policy: Health Insurance Fund
6. Conclusion
The German Health Insurance System seems to be, on one hand, a very rigid and apathetic
system (Lindner 2003). On the other hand, it responds to shortfalls with large changes in the
contribution fraction, which causes instabilities and oscillations. We have analyzed the health
insurance structure by means of different analysis methods, with the goal to identify system
behavior and find sustainable policies.
Sensitivity testing or soft policy testing has shown that the system is sensitive to changes in
the fertility rate and the net immigration rate. However, changes of these entities, correspond-
ing to reality, are long-term projects. Especially a change of the fertility rate depends highly
on other politic sectors and is not easy accomplished.
In contrast, changing of either the reference contribution fraction or the contribution fraction
adjustment effect is easily done, because both variables are policy variables. However, as the
sensitivity analyses show, these variables are at the same time high leverage points of the sys-
tem. Drastic changes of these variables will lead to unpredictable outcomes. The sensitivity
analysis shows improvement potential for the contribution adjustment effect. Therefore, find-
ing an optimal contribution adjustment policy remains an important task.
Besides sensitivity analysis, several policy analyses have been performed in order to firstly
relate the model more to reality and secondly to improve the model outcome, according to a
permanent zero health insurance fund perspective. The introduced ‘goal adjustment’ structure
has not improved the behavior of the health insurance fund, because of additional shortfall,
which is caused by the debt due to support, which causes large contribution fraction adjust-
ments. As mentioned in the previous paragraph, the contribution fraction adjustment effect is
the strongest leverage in the system. Therefore, a policy, which strengthens the adjustment
effect, is not applicable to stabilize the system.
The ‘comer cutting pressure on government’ policy seems to be effective in stabilizing the
system. The effect occurs, if there is a shortfall of the health insurance fund, and reduces the
average expenses per health request. This policy is suitable to lower the system oscillatory
tendency, due to its damping effect on health requests.
The ‘co-payment’ policy affects the health insurance fund as the ‘goal adjustment’ policy. In
general, the adjustment effect due to a shortfall is increased. Consequently, the system oscilla-
tory tendency is increased as well. However, if the adjustments are within certain boundaries,
the system is able to find a sustained equilibrium.
Modeling the Health Insurance System of Germany Page 29
In Germany, they have discussed the possibility to change the contribution charging method
from the old Bismarck's system to a fixed amount of money per capita and year and this is
modeled by the ‘fixed contribution’ policy. First, the fixation of the contribution fraction is a
means to stabilize the system. However, the process of choosing the constant contribution
fraction must be uncertain, because the future development of both the total population and
the health requests are unknown. Therefore, is it impossible to choose an adequate contribu-
tion fraction, which ensures at least a zero health insurance fund situation over a long time
period.
Another policy, which makes the model more related to the reality, is the change of the ad-
ministration expense assumption. The ‘per capita administration expenses’ policy denies the
permanent growth of the administration expenses and substitutes this assumption by a ‘fixed
amount per capita’ assumption. Then, the administration expenses behave in accordance with
the population development. This will lead to a more stabilized system, because the unrealis-
tic assumption regarding the administration growth causes high changes in the system.
In conclusion, the model regarding the German Health Insurance System represents the gen-
eral structure of the system. The most difficult task for the fund management is to maintain a
contribution fraction policy, which is suitable to the demands of the population development.
At the same time, the fund management has to consider the effects of the contribution fraction
on the human behavior.
7. Further Research
For further research, the model can be improved to include more accurate figures, especially
regarding the contribution adjustment policy that is actually used by the fund management.
This would give additional insights and would strengthen the explanatory power of the model.
Furthermore, a healthiness measurement ratio could be introduced. In our model, we have this
figure calculated (average health state of inhabitant), but it is not shown and discussed in the
report, because we consider the performed calculation as too simple and therefore too inaccu-
rate to represent reality. The explanations derived from the variable are not very meaningful.
Despite this, future research could improve the model by introducing a reliable health meas-
urement system. In addition, as Lindner has pointed out, the existing German Health Insur-
ance System is rather concentrated in curing diseases than in preventing them (Lindner 2003).
This issue is a worthwhile subject to be studied. The existing model can be improved to cover
this topic.
Moreover, the model should be integrated in a comprehensive model to cover more side ef-
fects and gain more insights. A possible model is the Threshold 21 (cf.
www.threshold21.com). We have discovered following possible interfaces to the T21-model:
population, income distribution, employment, government debt, household and life expec-
tancy.
To conclude, we hypothesize that the topic of the Social Security System, especially the
Health Insurance System, will become more important in Germany, as with the rest of the
world, over the next decades. Research in this field is therefore needed and worthwhile to
accomplish.
Modeling the Health Insurance System of Germany Page 30
Bibliography
AOK 2004a. Gesamtausgaben der gesetzlichen Krankenversicherung, V ol. 2004
AOK 2004b. Gesetzliche Krankenversicherung (GKV), Vol. 2004:
Checkland P. 1999. System Thinking, System Practice: Chichester, New Y ork, Wein-
heim, Brisbane, Singapore and Toronto
dpa. 2003. Umbau des Sozialstaats, Siiddeutsche Zeitung, Nr. 283 ed.: 11: Munich
Forrester JW. 1969. Urban Dynamics: Cambridge and London
Gorres-Gesellschaft. 1989. Staatslexikon - Recht - Wirtschaft - Gesellschaft: Freiburg
and Basel
Lindner U. 2003. Chronische Gesundheitsprobleme. Aus Politik und Zeitgeschichte
33(34): 21-28
Rasche U. 2004. Die Initiative "Lokale Biindnisse fiir Familie", Frankfurter
Allgemeine Zeitung, No. 273 ed.: 4: Frankfurt
Riesberg A, Weinbremner S, Buse R. 2003. Gesundheitspolitik im europaischen
Vergleich. Aus Politik und Zeitgeschichte 33(34): 29-38
Schnabel M. 2004. Leistungen der Pflegeversicherung, V ol. 2004:
Bundesamt fiir Gesundheit und Soziale Sicherung. 2004. Festhetrage senken
Arzneimittelausgaben und starken A rzneimittelinnovationen, Vol. 2004:
Bundesamt fiir Gesundheit und Soziale Sicherung 2003. Kennzahlen und
Faustformeln der gesetzlichen Krankenversicherung, Vol. 2004: Bonn
Statistisches Bundesamt Deutschland. 2003. Bevolkerung Deutschlands bis 2050 - 10.
koordinierte Bevélkerungsvorausberechnung: Wiesbaden
Sterman J. 2000. Business Dynamics - System Thinking and Modelling for a Complex
World: Boston
Wiener N. 1957. Cybernetics or control and communication in the animal and the
machine: New Y ork and Paris
CC BY-NC-SA 4.0