Infectious Disease and Preventive Behavior in An

ARTICLE IN PRESS
Journal of Economic Dynamics & Control 29 (2005) 1673–1700

www.elsevier.com/locate/econbase
Infectious disease and preventive behavior in an

overlapping generations model
Akira Momotaa,, Ken Tabatab, Koichi Futagamic
a
College of Economics, Osaka Prefecture University, 1-1 Gakuen-cho, Sakai Osaka 599-8531, Japan
b
Kobe City University of Foreign Studies, 9-1, Gakuenhigashimachi, Nishi-ku, Kobe 651-2187, Japan
c
Faculty of Economics, Osaka University, 1-7 Machikaneyama, Toyonaka 560-0043, Japan
Received 15 January 2003; accepted 15 September 2004
Available online 24 December 2004
Abstract
This paper incorporates prevalence of disease and agent’s preventive behavior (i.e., health
investment) into a small open overlapping generations model and investigates the dynamic
behaviors of the competitive equilibrium allocation. Using the model, it is shown that agents’
‘prevalence-elastic behaviors’, which are supported from empirical studies, can be obtained,
and that such agents’ behaviors cause the cyclicity of the spread of disease. We also show that
although the agents’ preventive behavior in a competitive equilibrium may be insufficient
because of the existence of external effects, a one-shot medical aid from foreign countries does
not necessarily improve the agents’ welfare in the Pareto sense.
r 2004 Elsevier B.V. All rights reserved.
JEL classification: I10; I12; I18; O41
Keywords: Infectious disease; Health investment; Overlapping generations model; Prevalence-elastic

behavior
Corresponding author. Tel.: +81 72 254 9557; fax: +81 72 254 9594.
E-mail addresses: momotaro@eco.osakafu-u.ac.jp (A. Momota), tabataken@inst.kobe-cufs.ac.jp
(K. Tabata), futagami@econ.osaka-u.ac.jp (K. Futagami).
0165-1889/$ - see front matter r 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.jedc.2004.09.007
ARTICLE IN PRESS
1674 A. Momota et al. / Journal of Economic Dynamics & Control 29 (2005) 1673–1700
1. Introduction
According to the World Health Organization (WHO) (1999a), infectious diseases

were the primary cause of mortality in 1998, causing about one-fourth of all deaths
in the world. In Sub-Saharan Africa, in particular, the spread of HIV/AIDS and
malaria causes serious public health problems. Fig. 1 shows the United Nations
Population Division’s estimate of how the HIV/AIDS epidemic reduces average life
expectancy. In Botswana, for example, the average life expectancy decreases from
67.6 years to 44.4 years. Moreover, the spread of these diseases undermines
economic development. It decimates the workforce, creates a large number of
orphans, and puts tremendous pressures on health and social services. Over (1992)
estimates that HIV/AIDS reduces GDP growth rates in Sub-Saharan countries by
about 1.2 percent per year. In addition, Gallup and Sachs (2001) conclude
that malaria slows economic growth in African countries by 1.3 percent per
year. The World Bank (2001) estimates that, compounded over 35 years, this
amounts to a 32 percent reduction in the GDP of African countries. The fact that the
spread of the infectious diseases lowers the average life expectancy and reduces
economic growth implies that it worsens the welfare of economic agents, especially in
Sub-Saharan Africa.
Life Expectancy with AIDS

Life Expectancy without AIDS
80
70
60
50
Life Expectacy
40
30
20
10
0
Botsw
Mala
Moza
Nam
Rwan
South
Zamb
Zimb
ibia
wi
abwe
mbiq
da
Afric
ia
ana
ue
Fig. 1. The effect of HIV/AIDS on life expectancy. Source: World Population prospects the 2000 revision
highlights (United Nations Population Division).
ARTICLE IN PRESS
A. Momota et al. / Journal of Economic Dynamics & Control 29 (2005) 1673–1700 1675
Infectious diseases are transmitted by two major modes: first, transmission occurs
from person to person, as in the case of HIV/AIDS; and second, it occurs by vectors,
as in the case of malaria. People’s behavior can be significant in preventing
transmission by either mode. For example, the majority of HIV/AIDS infections
occur as a result of unprotected sexual intercourse and intravenous drug use. Thus,
their incidence can be reduced if people choose not to engage in such risky behavior.
The incidence of malaria infections, which occur as a result of human contact with
mosquitoes, can be reduced by wearing clothes that do not leave any skin bare, by
sleeping under insecticide-treated nets, and by not walking in areas where malaria
mosquitoes (e.g., anopheles) are breeding.1 These facts suggest that agents’
preventive behavior plays an important role in disease dynamics. Recent empirical
studies have focused on agents’ preventive behavior. For example, using both cross-
sectional and time-series data for the United States, Ahituv et al. (1996) show that
the propensity of young adults to use condoms increases by up to 50 percent in
response to a one percent increase in the prevalence of HIV/AIDS. Further,
Philipson (2000) concludes that the agents’ level of prevention is likely to increase as
the spread of infectious diseases (including, but not limited to, HIV/AIDS) becomes
more serious.2 Such agents’ behavior is often called ‘prevalence elastic behavior’.
Anderson and May (1991) provide a classical model of infectious diseases, explaining
the transmission mechanism in detail. However, until recently, the agents’ prevention
behavior described above was not incorporated into the classical models.3 Recent-
studies, such as Geoffard and Philipson (1997) and Gersovitz and Hammer (2004a),
have begun to incorporate private rational choices for prevention into the classical
model. Each analysis has its own unique contribution and provides insightful
suggestions for public policy.4 However, since these studies directly extend the
classical models, their analytical treatments are too complicated to analyze the
impact of infectious diseases on the macroeconomy. In recent years, recognition has
grown that investments in health are essential for economic growth and should
therefore be a key component of national development strategies. Consequently, it is
1
If it is not possible to prevent skin being bared, the most effective preventive behavior is to take
medicine like chloroquine. Mefloquine or doxycline is used in areas where a drug-resistant malaria
pathogen has emerged.
2
For example, Mullahy (2000) finds that people whose mortality risk due to influenza is high are more
likely to demand the influenza vaccine. Moreover, Philipson (1996) shows that the age (in months) at
which the first measles vaccination occurs is substantially reduced by the prevalence of measles. In other
words, ‘prevalence elastic behavior’ is observed in the case of the demand for the childhood combination
vaccine against measles, mumps, and rubella (MMR).
3
A frequent criticism is that, owing to the ignorance of ‘agents’ preventive behavior’, the traditional
epidemiological models cannot predict actual HIV/AIDS epidemics in Africa appropriately (e.g., United
Nations and the WHO, 1991).
4
For example, Gersovitz and Hammer (2004a) incorporate private rational choices for prevention and
therapy into a classical model of infectious diseases. They show that the optimal subsidy policy for
prevention and therapy varies dramatically according to differences in the modes of the transmission (i.e.,
from person to person or by vectors). For more comprehensive surveys, please refer to Philipson (2000)
and Gersovitz and Hammer (2004b).
ARTICLE IN PRESS
Honduras
Nicaragua
80000
70000
60000
The number of cases
50000
40000
30000
20000
10000
0
1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997
Year
Fig. 2. The number of Malaria cases. Source: Weekly epidemiological record 74 (WHO, 1999b).
important to investigate agents’ preventive behavior against infectious diseases in a

macroeconomic context.
This paper establishes a macroeconomic model that explicitly incorporates agents’
preventive behavior and the prevalence of infectious diseases. In particular, this
paper simplifies the description of disease transmission and examines in depth how
the spread of infectious diseases influences resource allocation and economic welfare,
using a small open overlapping generations model.
Furthermore, using this model, we present a potential mechanism that explains the
relationship between agents’ preventive behavior and the well-known cyclical nature
of infectious diseases such as malaria and syphilis. Fig. 2 shows prevalence patterns
of malaria in Nicaragua and Honduras. It can be seen that the patterns follow cycles
of two to seven years.5 The spread of malaria is considered to be the result of
complicated interactions among factors including malaria parasites (e.g., Plasmo-
dium falciparum), human hosts, vectors (e.g., anopheles), and the environments
surrounding these vectors (e.g., temperature, precipitation). As a result, various
hypotheses are proposed to explain the cyclical nature of malaria. For example,
Bouma et al. (1994) and Bouma and van der Kaay (1994) claim that changes in
precipitation due to the El Niño Southern Oscillation (ENSO) influence the breeding
environment of malaria vectors and induce the cyclical spread of malaria in the
Indian region of Punjab, Sri Lanka, and Sub-Saharan Africa. However, Hay et al.
(2002) insist that no statistically significant associations can be found between
5
Cycles of periodicities vary, depending on nations and areas (e.g., see Nájera et al., 1998).
ARTICLE IN PRESS
climate change and the resurgence of malaria in the East African highlands. In
addition to the environmental factors, it is worth noting that another major factor
that has increased the oscillation of malaria cycles from the 1960s is ‘the way of
agents’ anti-malarial activities’. In the late 1950s and early 1960s, a Malaria
Eradication Campaign began in most Asian and Latin American countries. Since
then, people’s decisions to undertake anti-malarial activities have greatly reduced
malaria morbidity. However, the Campaign has not altered the cyclical pattern of
malaria. Indeed, Nájera et al. (1998) argue that the oscillation of cycles has become
even greater. In this regard, Bouma and van der Kaay (1994) and Nájera et al. (1998)
argue that people tend to reduce their anti-malarial activities in low-risk (low-
precipitation) years, which increases the proportion of susceptible people and thus
induces a serious spread of malaria in the next high-risk (high-precipitation) years.
This paper focuses on such agents’ preventive behavior and presents the mechanism
by which the agents’ rational preventive behavior induces the cyclical spread of
infectious diseases.
Furthermore, our findings can explain the cyclical spread of syphilis. Syphilis has
cycles of 10 years in the United States and 20 years in Japan. As pointed out by
Gayle and Counts (2001), few studies have examined the cyclical nature of syphilis.
The mechanism of the spread is often explained as follows: the majority of syphilis
infections occur as a result of unprotected sexual intercourse, as in the case of HIV/
AIDS, so that the incidence of syphilis is strongly associated with an agent’s sexual
behavior. As the spread of syphilis becomes more serious, people come to recognize
the risk of infection, which induces people to choose safe sexual behavior. This
reduces the number of agents who are infected with syphilis. While this lowers
people’s risk of infection, it also lowers people’s awareness of the risk of syphilis.
Thus, the number of young new agents who do not actually perceive the danger of
syphilis increases. Consequently, there is an increase in the number of agents who
choose risky sexual behavior and thus syphilis spreads again. This is the mechanism
that we will examine using our model. Furthermore, it is worth noting that recent
empirical studies partly support this hypothesis. The HIV/AIDS epidemic has not
yet shown any cyclical properties. Since no effective cure exists, people’s attitudes
toward HIV/AIDS are very different from their attitudes to syphilis. However,
noting that HIV/AIDS, like syphilis, is a sexually transmitted disease, we cannot
deny the possibility that it may have a cyclical pattern of prevalence in the future.
Furthermore, this paper examines how medical aid from developed countries
influences the welfare level of agents in developing countries.6 The WHO’s
Commission on Macroeconomics and Health estimates that the minimum cost of
essential health intervention in the least developing countries (LDCs) is around US
$34 per person per year on average, in 1997 US dollars. If governments in the
6
According to the WHO (2001), total official development assistance (ODA) is around $53 billion, that
is, 0.2 percent of the GNP of the major donor nations. In addition, the annual flow of bilateral donor
support for health is $2.55 billion, which represents 0.01 percent of donor GNP. The World Bank (1996)
provides an extensive review of the existing case-study analyses of HIV/AIDS aid programs. It concludes
that the subsidies for the purchase of condoms and clean needles and for education about the prevention
and treatment of HIV/AIDS are effective in controlling the spread of the HIV/AIDS.
ARTICLE IN PRESS
developing countries established such essential health interventions today, future

generations would receive great benefits in the form of improvements in life
expectancy and increases in per capita income. In other words, this implies that a
positive externality occurs between present and future generations. Theoretically, in
such a situation, it is possible to improve welfare in the Pareto sense by issuing a
national bond to subsidize agents’ health investments.
However, it seems too unrealistic to propose such a policy in relation to the LDCs.
In this regard, the WHO reports that LDCs can mobilize $15 per person at most,
even with more efficient resource allocation. Moreover, due to developments in
transportation systems and active international trade, infectious diseases occurring
in the developing countries can easily be transmitted to the developed countries.
Based on these facts, the WHO insists that the developed countries should increase
their medical aid to the developing nations from the current level of about $6 billion
per year to $27 billion per year. Although the influence of such medical aid policies
upon the welfare of agents in the developing countries has attracted much attention
from economists, to our knowledge, few theoretical studies investigate these issues.
This paper provides a positive analysis of a one-shot medical aid policy using a small
open overlapping generations model that explicitly incorporates the prevalence of
infectious diseases.
The rest of the paper is organized as follows. In Section 2, we present a basic
model and show that our theoretical finding is consistent with recent empirical
observations of ‘prevalence elastic behavior’. In Section 3, we derive a dynamical
competitive equilibrium path and shows that infectious disease may spread cyclically
when the degree of agents’ prevalence elastic behavior is sufficiently large. This
theoretical result provides new insights into the cyclical nature of infectious diseases.
Section 4 investigates the socially optimal allocation and compares it with a
competitive equilibrium. We argue, first, that there are two types of external effect in
the present model, a positive effect and a negative effect. Then, we show that an
agent’s health investment in a competitive equilibrium is insufficient in comparison
to the socially optimal allocation if and only if the positive external effect dominates
the negative one. In such a situation, we state that ‘a net positive externality exists’.
In Section 5, we assume that the net positive externality with regard to health
investment exists, a plausible assumption in low-income countries. We examine the
effect of a one-shot medical aid from a foreign country on the welfare of agents, and
show that it does not necessarily improve the agents’ welfare in the Pareto sense: that
is, we show that there exists a future generation that is damaged by the one-shot
medical aid. Finally, Section 6 concludes this paper.
2. Model
We use a two-period overlapping-generations model in which lifetime uncertainty

is incorporated.7 Economy begins its operation in period 0, and a cohort born in
7
We can find a kind of this model in Pecchenino and Pollard (1997) and Chakraborty (2004).
ARTICLE IN PRESS
period t is called generation t. Agents live for at most two periods. In other words,
we assume that there exist only two generations in period t ðtX0Þ; that is, generations
t (as a young generation) and t 1 (as an old generation). Those who are old in
period 0 (i.e., initial period) are called the initial old. When agents are young, they
are assumed to be identical. We consider the existence and prevalence of diseases
explicitly, and due to the diseases, some young agents die at the end of the first
period of life. We denote the probability that agents of generation t can be alive in
the second period of life (i.e., period t þ 1) by qt : We denote the size of a newly born
cohort in period s by N s ; so that the size of generation t is expressed as N t in period t
(when young) and qt N t in period t þ 1 (when old). It can be immediately confirmed
that the number of total population in period t is N t1 qt1 þ N t : We assume that
parents bear their children in the first period of their life. This means that each young
N
agent has Ntþ1
t
children. We treat household’s decisions on fertility as exogenous and
assume N s ¼ 1 for all s, that is, the size of the newly born cohort in each period is the
same and normalized to unity.8
We express the degree of the prevalence of the diseases in period t as an index
Dt 2 Rþ ; and a higher level of Dt means that the degree of the spread of the diseases
is more serious. Let us designate Dt the ‘disease level in period t’ here. For example,
we can interpret Dt as the degree of insanity environment or the number of vectors
(e.g., anopheles) in case of malaria; and in case of HIV/AIDS, it is interpreted as
infection of AIDS/HIV from mothers to their fetus.9 Agents invest in themselves in
order to improve their health levels and prevent against diseases; ht denotes the level
of a representative agent’s health investment. Basically, we can interpret ht in a
broader sense. For example, it can be interpreted as toward the improvement of
water and sanitation systems, an agent’s choice of vaccination,10 receiving health
education, and a mixture of these behaviors, as have been described in Section 1.
According to the discussion above, we express the surviving probability qt as a
function with respect to ht and Dt ; qt ¼ qðht ; Dt Þ; and assume q1 40; q2 o0; q11 o0;
and q12 X0: q1 40 and q11 o0 imply that the agent’s health investment increases the
8
It is immediately confirmed that the qualitative property of results in this model remains unchanged
even if we assume N s ¼ nN s1 ; where n is an exogenous parameter.
9
As is well-known and has been described in Introduction, HIV/AIDS is a sexually transmitted disease,
and the majority of the infections occur as a result of unprotected sexual intercourse and intravenous drug
use. In this regard, if we model a process of the disease transmission by describing a matching process
between susceptible agents (i.e., those who have not been infected but susceptible) and infected agents, we
will be able to interpret Dt as the number of infected people. However, incorporating the matching process
into the model makes our analysis too complicated. Hence, for analytical simplicity, we do not treat this
issue explicitly in the present model, and we confine our interpretation of Dt to the infection from mothers
to their fetus.
10
There is an important issue with respect to vaccination; that is, R&D of new vaccines. We will be able
to incorporate the R&D activity by extending our model easily, for example, by defining effective units of
vaccines against diseases by xt At ht ; where At represents the quality of the vaccines and is assumed to be
an increasing function of the level of the R&D activity. However, once the R&D activity is modeled
explicitly, we will have to consider the issues that who should develop such vaccines and who should pay
for that type of R&D. These issues are of great importance, but they are beyond of the scope in the present
analysis.
ARTICLE IN PRESS
life expectancy, but the marginal effect decreases as the level of the health investment
becomes large. q2 o0 corresponds to the fact that severer prevalence of diseases
reduce the life expectancy of the agents, and q12 X0 implies that a marginal effect of
an agent’s health investment in improving their own health status is larger when the
disease level in a society is higher.
We express the aggregate level of ht as H t ; and consider that the disease level in the
next period is determined by the level of aggregate preventive behavior and disease
level in the present period:11
Dtþ1 ¼ zðH t ; Dt Þ; (1)
where we assume that z is twice continuously differentiable, and that as regards its
derivatives, z1 o0 and z2 40 hold, and that taking h as given, z is concave
(i.e., z22 p0) with respect to Dt : Furthermore, we impose the following boundary
conditions: limD!0 z2 ð0; DÞ41 and limD!þ1 z2 ð0; DÞ ¼ 0:12 We assume that the
aggregate health investment in period t affects the disease level in the next period.
For example, the assumption can be interpreted as that agent’s current contribution
to the improvement of water and sanitation systems improves a sanitary
environment in the next period, which reduces the number of young new agents
who are infected with diseases.13 However, each agent does not take this effect into
account when they choose the level of their health investment. That is, the agent’s
health investment has external effects.14
We consider a small-open economy and assume that capital can mobile freely
across the border. In this settings, the world interest rate r is treated as given. All
firms are assumed to be identical and act competitively. Each firm has a constant
returns to scale technology, and the aggregate production function is expressed as
Y t ¼ F ðLt ; K t Þ; where Y t ; Lt ; and K t denote aggregate levels of output, labor input,
and domestic capital stock, respectively. Representing the production function as a
per worker term, yt ¼ f ðkt Þ; where yt Y t =Lt and kt K t =Lt ; we can obtain the
optimal conditions for the representative firm as r ¼ f 0 ðkt Þ and wt ¼ f ðkt Þ f 0 ðkt Þkt ;
where wt denotes a wage rate in period t. In a small country model, we can confirm
that kt and wt are time invariant from the above equations, so we represent these
variables simply as k and w, respectively.
Agents are endowed with one unit of labor force and work inelastically in their
young period, and retire when old. They derive utility from their own consumption
when young and old. Thus, the lifetime expected utility of a representative agent in
11
Of course, we can consider that the aggregate preventive behavior also affects a health status of a
current generation as well as future generations. In this case, the functional form of q should be
represented as qðht ; H t ; Dt Þ; instead of qðht ; Dt Þ: However, such a change does not affect the qualitative
property of results of this paper significantly, while the analysis becomes more complicated. Thus, we do
not treat this intragenerational effect of H t explicitly in this paper.
12
These boundary conditions will be used to prove the existence of the non-trivial steady state
equilibrium.
13
We can also consider that each agent’s current choice of vaccination has a similar external effect. The
current choice of vaccination reduces the number of infected agents in a current generation, which leads to
a reduction of a infection risk of the new agents born in the next period.
14
We will consider this issue in detail in Section 4.
ARTICLE IN PRESS
generation t is expressed as
EUðc1t ; c2tþ1 ; ht ; Dt Þ ¼ uðc1t Þ þ bqðht ; Dt Þuðc2tþ1 Þ; (2)
where c1t and c2tþ1 denote consumption in periods t (i.e., when young) and t þ 1 (i.e.,
when old), respectively, and b is a time discount factor. We assume u0 40 and u00 o0:
Furthermore, in order to ensure that the level of EU increases as the probability of
survival q increases, we impose a condition that uðcÞX0 for all c 2 Rþ : It is useful to
note that uðcÞ u0 ðcÞc40 holds for any c 2 Rþ under the assumption concerning u.15
The agents allocate their labor income w between consumption c1t ; health investment
ht ; and savings st when young; when old, they consume all the principle and accrued
interest of the savings and have no bequest motive. We assume the existence of
actuarially fair insurance companies in this paper, as used in Yaari (1965) and
Blanchard (1985). The companies collect funds and invests them in firms or foreign
countries. Returns on the investment are repaid to the insured households still living.
In other words, the contract offered by the insurance company redistributes income
from people who have died to those who are still living.
Suppose that the insurance company collects at from each young agents (in other
words, At at N t in the aggregate) in period t. (Note that old agents have no
incentive to buy the annuity because they are not alive in the next period.) The
company invests them in firms or foreign countries and acquires the total proceeds
rAt in period t þ 1: Noting that only q̄t N t old people survive in period t þ 1; where q̄t
denotes the average level of qt ; each of them receives q̄r at from the insurance
t
company due to perfect competition among the companies. Thus, the rate of return
r
on the annuities is q̄ if they are alive and 0 if they die in the end of period t. We
t
consider that agents take the level of q̄t as given when they act to maximize their
lifetime expected utility. On the other hand, if the young agents in period t invest at
directly in firms or foreign countries, they will receive rat whether they are alive or
dead. (If the agents are not alive in the next period, their children will inherit it.)
Thus, the rate of return on the self-investment is r. As we assume that households
have no bequest motive, they accept the insurance contract, which yields a higher
interest rate than the self-investment. Thus, budget constraints of the agent of
generation t are
c1t þ ht þ st ¼ w; (3)
r
c2tþ1 ¼ st : (4)
q̄t
We can obtain a lifetime budget constraint from (3) and (4):
q̄t
c1t þ c2tþ1 þ ht ¼ w: (5)
r
Maximizing (2) subject to (5), we can derive the following first-order conditions:
u0 ðc1t Þ ¼ lt ; (6)
15
The proof for this is straightforward. When uðcÞ is strictly concave, uðcÞ uð0Þ4u0 ðcÞc for any c40:
Noting that uð0Þ is nonnegative, the proof is completed.
ARTICLE IN PRESS
q̄t
bqðht ; Dt Þu0 ðc2tþ1 Þ ¼ lt ; (7)
r
bq1 ðht ; Dt Þuðc2tþ1 Þ ¼ lt : (8)
where lt is a lagrangian multiplier with respect to the budget constraint (5). Using
^ ; Dt Þ;
(5)–(8), we obtain c1t ; c2tþ1 ; and ht as functions of Dt and q̄t ; that is, ht ¼ hðq̄ t
c1t ¼ c^1 ðq̄t ; Dt Þ; and c2tþ1 ¼ c^2 ðq̄t ; Dt Þ; respectively. Since agents are identical,
qðht ; Dt Þ ¼ q̄t holds in equilibrium, so that the levels of c1t ; c2tþ1 ; and ht in
equilibrium are expressed as function of Dt ; that is, c1 ðDt Þ; c2 ðDt Þ; and hðDt Þ;
respectively. Furthermore, using these, we can express the indirect utility of
generation t attained in equilibrium as
V ðDt Þ EUðc1 ðDt Þ; c2 ðDt Þ; hðDt Þ; Dt Þ
¼ uðc1 ðDt ÞÞ þ bqðhðDt Þ; Dt Þuðc2 ðDt ÞÞ:
With regard to the property of the indirect utility function V ðDt Þ; we can confirm
that dVdDðDt t Þ o0 holds.16 This outcome can be interpreted straightforwardly. It implies
that the welfare of generation t becomes worse as the disease level in period t
becomes severer.
Now we will present an important result concerning a comparative statics in the
following proposition:
dhðDt Þ
Proposition 1. dDt 40 holds when q12 X0 holds.
The proof for this proposition is shown in Appendix A. dhðD tÞ
dDt 40 shows that
agents increase their preventive behavior in equilibrium when the degree of
prevalence of disease in a society is severer, and importantly, this result is consistent
with empirical observations, which is called ‘prevalence-elastic behavior’. This
implies that our theoretical framework can be justified from an empirical viewpoint.
3. Dynamical equilibrium path
In this section, we examine the dynamical system of this economy. A dynamical

equilibrium path is obtained from (1) and H t ¼ ht ¼ hðDt Þ; and we express it as
16
Differentiating V ðDt Þ with respect to Dt ; we can obtain
dV ðDt Þ dc1 ðDt Þ dc2 ðDt Þ dhðDt Þ

¼ u0 ðc1 Þ þ bqu0 ðc2 Þ þ bq1 uðc2 Þ þ bq2 uðc2 Þ:
dDt dDt dDt dDt
Using (5)–(8) and qðht ; Dt Þ ¼ q̄t ; the above equation is rewritten as

dV ðDt Þ q dhðDt Þ lc2
¼ l 1 þ buðc2 Þ q :
dDt r dDt r 2
Furthermore, applying (7) and qðht ; Dt Þ ¼ q̄t to the above equation again, we can obtain
dV ðDt Þ dhðDt Þ
¼ bu0 ðc2 Þq1 þ bq2 fuðc2 Þ u0 ðc2 Þc2 go0:
dDt dDt
ARTICLE IN PRESS
Fig. 3. Competitive equilibrium path.
Dtþ1 ¼ zðhðDt Þ; Dt Þ FðDt Þ: This dynamical path shows that there exist two routes
through which a current disease level Dt affects the disease level in the next period.
One is the route in which Dt affects Dtþ1 directly (i.e., the second argument of z), and
the other is what we call an indirect route; first, Dt affects each agent’s health
investment ht directly, then the change of ht leads to a change in the aggregate health
investment H t ; and finally, the change in H t influences the level of Dtþ1 : We depict
the dynamical path in Fig. 3. To begin, fixing the level of h in zðh; DÞ at 0 and w,17 let
us define zð0; Dt Þ and zðw; Dt Þ as x̄ðDt Þ and xðDt Þ; respectively, and draw them in this
figure as auxiliary curves. The intersection of Dtþ1 ¼ x̄ðDt Þ (resp. Dtþ1 ¼ xðDt Þ)and
the 45 line is labeled A (resp. B).18 Noting that the curve Dtþ1 ¼ zðh; Dt Þ shifts
downward as the level of h increases, we can see that these two curves construct an
upper and a lower boundaries, respectively, of Dtþ1 ¼ FðDt Þ: In other words, since
xðDt ÞoFðDt Þox̄ðDt Þ holds, the phase diagram of Dtþ1 ¼ FðDt Þ lies between Dtþ1 ¼
xðDt Þ and Dtþ1 ¼ x̄ðDt Þ:
The steady state of this system is expressed as Dtþ1 ¼ Dt ; and we denote the level
of D in the steady state equilibrium by Dc ; that is, Dc ¼ FðDc Þ holds, labeled as point
E in Fig. 3. For notational convenience, let us denote dhðDÞ D qzðh;DÞ h
dD hðDÞ ; qh zðh;DÞ ; and
qzðh;DÞ D
by hD ; zh ; and zD ; respectively (chD ; czh ; and czD indicate that they are
qD zðh;DÞ
evaluated at the stationary competitive equilibrium). With regard to the existence,
uniqueness, and stability properties of the steady state equilibrium, we state the
following proposition:
17
It can be confirmed from (3) and (4) that the amount of resources which can be allocated to the health
investment in each period is at most w when a country does not receive any (medical) aids from foreign
countries.
18
Since limD!0 z2 ð0; DÞ41 holds, we can guarantee the existence of point A, but, on the other hand, it is
possible that point B does not exist. However, we do not have to modify our argument even if point B does
not exist.
ARTICLE IN PRESS
Proposition 2. (i) If limDt !0 hðDt Þ ¼ 0 holds, then a non-trivial steady state

equilibrium (i.e., Dc a0Þ exists.
(ii) Furthermore, when h0 ðDt Þ40 holds, the non-trivial steady state equilibrium exists
uniquely.
(iii) With regard to the stability of the non-trivial steady state equilibrium, the
following statements hold:
(a) The non-trivial steady state equilibrium is stable, and in the neighborhood of the
steady state, an equilibrium path is monotonic if and only if 19
0oczh chD oczD :
(b) The non-trivial steady state equilibrium is stable, and in the neighborhood of the
steady state, equilibrium paths fluctuate if and only if
czD oczh chD o1 þ czD :
(c) The non-trivial steady state equilibrium is unstable, and in the neighborhood of the
steady state, an equilibrium path fluctuates if and only if 20
1 þ czD oczh chD :
The proof of Proposition 2 is shown in Appendix B. Furthermore, for Proposition

2(iii-c), we can derive the following corollary:
Corollary. When 1 þ czD oczh chD holds, there exists a two-period cycle equilibrium.
For the proof for this corollary, it is useful to see Azariadis’ (1993) Theorem 8.2.
Proposition 2(iii) states that the relative level between czh chD and czD is an important
factor determining the stability properties of the equilibrium path. As has described
in the beginning of this section, there exist a direct route and an indirect one through
which a current disease level affects the disease level in the next period, and by
definition, we can confirm that czD and czh chD represent the direct and indirect effects,
respectively. Thus, what Proposition 2(iii) argues is that when the indirect effect
dominates the direct one, the equilibrium path fluctuates, and furthermore, that
when the extent of the dominance is large, the equilibrium path fails to converge to
the steady state.
Now we interpret the outcome intuitively. chD represents how sensitive agents
respond to a small increase in a disease level by increasing their health investment.
On the other hand, czh represents what we call an obstinate feature of a disease. A
large level of czh means that although the disease level is easily suppressed by agents’
preventive behavior, the disease regains its power easily unless the agents continue to
19
Instead, if zh hD ozD holds for any Dt ; then the monotonicity of an equilibrium path is guaranteed
globally.
20
When we specify a functional form of zðh; DÞ ¼ cðhÞDat where 0oao1 and c0 ðhÞ40; zD ¼ a holds.
ARTICLE IN PRESS
be vigilant against the disease. Thus, we can interpret the product of chD and czh as
the extent that the agents’ preventive behavior can affect the disease level.
Here, let us consider a situation in which the levels of czh are chD are large, and
suppose that a current disease level Dt is large. Then agents increase their health
investment largely, and the increase in the health investment reduces the disease level
in the next period to a large extent as well. Thus, the level of Dtþ1 will be small.
However, the small level of Dtþ1 will lead agents to reduce their health investment in
period t þ 1; and such behavior will help the disease recover its power. Consequently,
the level of Dtþ2 will rise again. The discussion above indicates that agents’
preventive behavior is a source of a cyclical movement of a disease level.
On the other hand, we can interpret the direct effect czD as a potential infection
power of a disease, and it is considered to be monotonic (i.e., z2 40).
As a consequence, we can sum up the discussion above as follows: The direct effect
makes the evolution pattern of a disease level monotonic, and the indirect one
makes the pattern cyclic; when the direct effect overweighs the indirect one, the
equilibrium path of Dt is monotonic (Proposition 2(iii-a)); however, when the direct
effect is dominated by the indirect one, the equilibrium path turns to be cyclical
(Proposition 2(iii-b)); and furthermore, when the extent of the dominance is large,
the equilibrium path fails to converge to the steady state (Proposition 2(iii-c)) and
may be chaotic.21
4. Social optimum vs. competitive equilibrium
Each agent makes a health investment (i.e., ht ) in order to improve his (or her)
own health status; in other words, in order to raise the survival probability qt : As we
briefly suggested in Section 2, an agent’s health investment behavior has external
effects in the present model. Thus, the allocation attained in the competitive
equilibrium is expected to be different from the socially desirable allocation. Let us
start this section by arguing that the health investment has two external effects; one
is a positive intergenerational external effect, and the other is a negative
intragenerational one,22 and then, we derive the allocation chosen by a benevolent
central planner and compare it with the competitive equilibrium allocation.
As for the positive external effect, it is important to note that when each
representative agent increases his (or her) own health investment, the aggregate level
of ht ; H t ; also rises, and the disease level in period t þ 1; Dtþ1 ; is reduced. This effect
lies between a current generation and future generations, so that it is recognized as
an intergenerational external effect. As has been argued in Section 2, a reduction in
the level of Dtþ1 enhances the welfare of generation t þ 1; V ðDtþ1 Þ; but agents of
21
In this regard, it is useful to see Benhabib and Nishimura (1985), Nishimura and Yano (1995), Baierl
et al. (1998).
22
If we alter the functional form of q to qðht ; H t ; Dt Þ; another external effect will emerge. In this case, a
change in ht affects q not only directly (i.e., through qh ), but also indirectly through a change in H t (i.e.,
through qH ), which can be viewed as an intragenerational positive external effect. However, we do not
treat this effect explicitly in the present analysis (see also footnote 11).
ARTICLE IN PRESS
generation t do not consider this effect when they choose the level of ht :23 On the
other hand, there also exists a negative external effect. When individuals invest in
their preventive behavior, their average life span, 1 þ q̄; increases, which results in a
decline in the rate of return on annuity, q̄r ; but each agent does not take the decline
into account when he or she chooses the level of h.24
Now let us derive a social optimal allocation chosen by a benevolent central
planner. We denote a foreign asset that exists in the initial period (i.e., period 0) by
b0 : For the sake of brevity, we presume that a central planner makes the initial asset
b0 þ k belong to the initial old and maximizes a discounted sum of all agents’ utility
from generation 0 on.25 Thus, the social welfare function is expressed as
X
1
SW ¼ gi N i fuðc1i Þ þ bqðhi ; Di Þuðc2iþ1 Þg;
i¼0
where go1 represents a discount rate across generations. The term fuðc1t Þ þ
bqt uðc2tþ1 Þg represents the expected (or ex ante) utility of the representative agent of
generation t.26 As we assume that N i ¼ 1 for any i, it can be rewritten as
X
1
SW ¼ gi fuðc1i Þ þ bqðhi ; Di Þuðc2iþ1 Þg: (9)
i¼0
Taking the level of D0 as given, the central planner maximizes (9) subject to the
following resource constraints:
c10 þ h0 þ b1 þ k ¼ w; (10)
c1t þ qðht1 ; Dt1 Þc2t þ ht þ btþ1 þ k ¼ w þ rðk þ bt Þ; ðtX1Þ; (11)
Dt ¼ zðht1 ; Dt1 Þ; ðtX1Þ: (12)
23
Philipson (2000) stresses the intergenerational external effect of a disease eradication policy. He says,
‘‘there is a missing intergenerational market: future generations cannot pay vaccine producers for the
benefit of eradication, although collectively they benefit most from an eradication program.’’
24
As regards such a negative external effect, it is useful to refer Maeda and Akai (1998) and Philipson
and Becker (1998).
25
It is useful to note that such allocation satisfies the Pareto optimality. Moreover, even if we explicitly
incorporate the welfare of the initial old into the social welfare function (9), we obtain the same result as
what is shown in the present setting when we assume that there is no foreign asset in the initial period (i.e.,
b0 ¼ 0).
26
It is useful to note that the social welfare function can also be interpreted as a Benthamian type social
welfare function, that is, as the (weighted) sum of utilities over all existing agents in the economy (Eckstein
and Wolpin, 1985). To see this, let us rewrite SW as
X
1
SW ¼ g0 N 0 uðc10 Þ þ fgi1 N i1 qi1 buðc2i Þ þ gi N i uðc1i Þg:
i¼1
Noting that there exist N i young agents and N i1 qi1 old agents in period i, that the young and old agents
derive their utility uðc1i Þ and buðc2i Þ; respectively, and that the utility of generations i (i.e., young agents in
period i) and i 1 (i.e., old agents in period i) are evaluated by gi to gi1 from the perspective of social
welfare, it is confirmed that SW represents the (weighted) sum of utilities over all existing agents (except
the initial old) in the economy.
ARTICLE IN PRESS
The optimal solution is obtained by setting up the Lagrangian function:

L ¼ SW þ l0 ðw c10 h0 b1 kÞ
X
1
þ gi li fw þ ðr 1Þk þ rbi c1i qðhi1 ; Di1 Þc2i hi biþ1 g
i¼1
X
1
þ gi mi fzðhi1 ; Di1 Þ Di g:
i¼1
Thus we can obtain the optimal conditions:

u0 ðc1t Þ ¼ lt ; (13)
bu0 ðc2t Þ ¼ glt ; (14)
lt ¼ rgltþ1 ; (15)
q1 ðht ; Dt Þfbuðc2tþ1 Þ gltþ1 c2tþ1 g lt þ gmtþ1 z1 ðht ; Dt Þ ¼ 0; (16)
q2 ðht ; Dt Þfbuðc2tþ1 Þ gltþ1 c2tþ1 g mt þ gmtþ1 z2 ðht ; Dt Þ ¼ 0: (17)
We can characterize the allocation chosen by the central planner using (10)–(17).
Here, by comparing (16) with (8) (it is useful to eliminate ltþ1 by using (15)), we can
see that the term gmtþ1 z1 ðht ; Dt Þ appearing in (16) represents the discounted value of a
positive intergenerational externality evaluated by the central planner.27 Taking
account of (14), we can see that the term q1 ðht ; Dt Þgltþ1 c2tþ1 represents the negative
intragenerational external effect. It can immediately be confirmed from (14) and (15)
that gr ¼ 1 must hold in order for this system to have a non-trivial steady state. Thus
we will confine our analysis to the case in which gr ¼ 1 holds. In this case, it can be
seen from (13)–(15) that the levels of consumption, c1t and c2t are time invariant, so
that we omit time subscripts from c1 and c2 : In this case, a dynamical system of the
central planner problem is expressed as follows:
gz2 ðht ; Dt Þmtþ1 mt þ bðuðc2 Þ u0 ðc2 Þc2 Þq2 ðht ; Dt Þ ¼ 0; (18)
bðuðc2 Þ u0 ðc2 Þc2 Þfq1 ðht ; Dt Þz2 ðht ; Dt Þ q2 ðht ; Dt Þz1 ðht ; Dt Þg
¼ bru0 ðc2 Þz2 ðht ; Dt Þ mt z1 ðht ; Dt Þ; ð19Þ
Dtþ1 ¼ zðht ; Dt Þ: (20)
27 qD
It is useful to note that z1 ðht ; Dt Þ ¼ qhtþ1
t
holds and mtþ1 represents a shadow price of Dtþ1 evaluated in
terms of generation t þ 1’s welfare. Thus, mtþ1 z1 ðht ; Dt Þ expresses the effect of generation t’s health
investment ht on the welfare of generation t þ 1; that is, the intergenerational externality. Noting that the
central planner discounts the future generation’s welfare at the rate of g; it can be confirmed that the term
gmtþ1 z1 ðht ; Dt Þ represents the level of an intergenerational externality evaluated from the perspective of
social welfare.
ARTICLE IN PRESS
Here, (18) is obtained from (14), (15), and (17). Next, (19) is derived using (14)–(17).
From (18)–(20), we obtain a two-dimensional dynamics with respect to fmt ; Dt g:
Moreover, using resource constraints (10) and (11) and a boundary condition
b
limt!1 tþ2 rt ¼ 0; we can also obtain
r X1
qðhi1 ; Di1 Þ X 1
hi r
c1 þ c2 i
þ i
¼ w: (21)
r1 i¼1
r i¼0
r r1
It is quite difficult to compare the allocation chosen by the central planner with the
competitive equilibrium allocation derived in Section 3, but we can investigate this
issue if we confine our analysis to the stationary allocation. Thus, let us limit our
attention to the steady state in the rest of this section.
Let us denote the allocation attained in the stationary competitive equilibrium by
ðcc1 ; cc2 ; hc ; Dc Þ: It can be confirmed that such allocation is a solution of the following
equations, which are derived from (1), (5)–(8), and qðht ; Dt Þ ¼ q̄t :
u0 ðc1 Þ ¼ bru0 ðc2 Þ; (22)
D ¼ zðh; DÞ; (23)
qðh; DÞ
c1 þ c2 þ h ¼ w; (24)
r
q1 ðh; DÞuðc2 Þ ru0 ðc2 Þ ¼ 0: (25)
On the other hand, let us denote the allocation attained in the presence of a central
planner by ðcs1 ; cs2 ; hs ; Ds ; ms Þ: As can be confirmed from (13)–(15), (18), (20), and (21),
such allocation satisfies (22)–(24), and28
g
q1 ðh; DÞuðc2 Þ ru0 ðc2 Þ ¼ z1 ðh; DÞm þ q1 ðh; DÞu0 ðc2 Þc2 : (26)
b
Noting that ms is negative,29 we can see that the first and second terms of the right-
hand side of (26) is negative and positive, respectively.The difference between the
competitive equilibrium allocation and the social optimal allocation is represented as
the difference between (25) and (26), and as has been mentioned earlier, this
difference represents that there exist two types of external effects, the former is a
positive intergenerational effect and the latter is a negative intragenerational one.30
Taking account of these two effects, we can state the following proposition:
28
To obtain ms explicitly, it is necessary to use (19).
29
Using (16) and (19), we can yield
bðuðcs2 Þ u0 ðcs2 Þcs2 Þq2 ðhs ; Ds Þ ¼ ð1 gszD Þms ;

where szD indicates the level of zD evaluated at the social optimum. To obtain the right-hand side of the
above equation, it is useful to note that z2 ¼ szD holds since D ¼ zðh; DÞ in the steady state. Furthermore,
noting that the left-hand side of the above equation is negative, and that szD p1 holds (using the same
argument as that referenced in footnote 15, we can confirm that 0ozD p1 for any D40:), we can confirm
that ms o0 holds.
30
If we formulate q as qðht ; H t ; Dt Þ; there exist a positive intragenerational external effect as well.
ARTICLE IN PRESS
Proposition 3. Comparing the allocation attained in the competitive equilibrium with

the social optimal allocation in the steady state, the following inequalities hold if and
only if a positive external effect dominates a negative one:
ðaÞ cc1 4cs1 ; ðbÞ cc2 4cs2 ; ðcÞ hc ohs ; ðdÞ Dc 4Ds :
The proof for Proposition 3 is shown in Appendix C. We can see from this
proposition that when the positive external effect with respect to agents’ preventive
behavior dominates the negative one (we will sometimes refer such a situation as
‘there is net positive externality’), in the competitive equilibrium allocation, the level
of consumption is larger than the command optimum allocation, while the level of
health investment is low and the disease level D is high, and vice versa. Using (14)
and the right-hand side of (26), it can immediately be confirmed that the positive
externality effect dominates the negative one if and only if lq1 c2 omz1 holds. If health
spending is less sufficient than its minimum requirement level, we will be able to
consider that jmj is large relative to jlj; and thus that there is net positive externality
with respect to the health spending in such a situation.
It is well known that because of the existence of the external effects, the
competitive equilibrium is not Pareto efficient. If there exists net positive externality
and thus the health investment is insufficient, there is scope for improving the agents’
welfare in the Pareto sense when the government implements a subsidy policy to
encourage the agent’s health investment behavior. Since the positive externality lies
between present and future generations, from a theoretical viewpoint, the subsidy
policy should be implemented by transferring resources from future to present
generations, that is, by issuing a national debt.31
Finally, let us refer to a realistic aspect of this issue briefly and discuss whether we can
consider that there is a net positive externality in reality, and whether the subsidy policy by
issuing a national bond can be applicable. As has mentioned in Introduction, the
prevalence of infectious diseases is in a severer situation in least developing countries.
Commission on Macroeconomic and Health in the WHO reports that the costs of the
essential health intervention are around $34 per person per year on average, but the least
developed countries can mobilize at most $15 per person,32 and insists that the developed
countries should increase their medical aid to the developing nations from the current level
of about $6 billion per year to $27 billion per year. Taking such a circumstance into
account, it can be justified that we presume a situation in which there is the positive
external effect. In addition, these figures suggest that in these countries, resource
constraints are too severe to deal with the infectious disease problems by themselves.
Indeed issuing a national bond to subsidy for agents’ health investment can improve their
31
In this regard, Geoffard and Philipson (1997) advocates this view; they state, ‘‘a deficit-financed
eradication program, which spends beyond tax revenues during its operation but recoups the deficit in
future generations, may improve welfare when discounting is at current market interest rates.’’
Following a procedure similar to the proof for Proposition 4 in the next section, we will be able to prove
that the Pareto improvement can be attained when the government issues a national bond to implement
the subsidy policy appropriately.
32
According to this report, the estimate that the essential health intervention are around $34 is very modest.
In the high-income countries, average per capita health spending amounts to more than $2000 per year.
ARTICLE IN PRESS
welfare in the Pareto sense theoretically, but it seems to unrealistic that we propose it to
these countries. Here we can find the significance of medical foreign aids to low-income
countries. Thus, let us investigate the effect of medical aids in the next section.
5. Effect of medical aid from a foreign country
As has been mentioned in the previous section, we presume that an agent’s health
investment in a competitive equilibrium is insufficient when we compare it with the
socially optimal allocation. In this section, we investigate the effect of a medical aid
from a foreign country on the welfare of the agents and draw an interesting result.33
Suppose that the government of a country receives a one-shot medical aid from a
foreign country in period t, whose level is denoted by m, and that by using the aid,
the government provides a subsidy for the health investment behavior of generation
t. In other words, thanks to the medical aid, the government need not impose any tax
on the domestic agents nor issue any national bonds in order to implement the
subsidy policy. Intuitively, we may imagine that such a medical aid improves
generation t’s welfare and raises the level of ht ; which leads to an increase in the level
of H t ; and that due to the increase in the level of H t ; all subsequent generations’
welfare is raised as well. In other words, we may expect that the foreign aid improves
the agents’ welfare in the Pareto sense. However, is such an intuition always true? We
examine this point and show that the answer is not always yes.
To begin, let us describe the behavior of agents of generation t. They maximize (2)
subject to the following budget constraint:
q̄t
c1t þ c2tþ1 þ ð1 tÞht ¼ w; (27)
r
where t represents a subsidy rate. In equilibrium, q̄t ¼ qðht ; Dt Þ holds and the
government faces the following budget constraint:
tht ¼ m: (28)
Since the government receives a one-shot medical aid from a foreign country in
period t, (27) and (28) are valid only in period t, and in period s for all sXt þ 1; the
equilibrium allocation can be derived in the same way as described in Section 2. Let
us represent the sequence of Dt in a competitive equilibrium by ðDt ; Dtþ1 ; Dtþ2 ; . . .Þ
when the country does not receive the medical aid in period t, and by
ðDt ; D
tþ1 ; Dtþ2 ; . . .Þ when the country does receive the aid.
33
In order to evaluate the welfare loss caused by infectious diseases, several measures, such as Quality
Adjusted Life Years (QALYs) and Disability Adjusted Life Years (DALYs), are often used in economics
and epidemiology. Such a measure of the ‘burden of disease’ (BOD) is generally defined as a product of the
prevalence and the (sometimes quality-adjusted) severity of a disease. However, as pointed out by
Philipson (2000), the BOD measure pays attention to the direct costs of infectious diseases only and
ignores indirect costs. That is, the welfare loss that stems from the infectious disease and distorts the
agents’ preventive health investment behavior is not considered in the measure of the BOD. On the other
hand, we consider lifetime utility as a welfare measure. In other words, both the direct and indirect costs
are considered in our notion of the welfare measure.
ARTICLE IN PRESS
Fig. 4. Case in which Dtþ2 oD

tþ2 occurs.
For the level of ht ; we can obtain a natural result that dh t

dm 40 holds. (The process of
the derivation is shown in Appendix D.) Thus, we can see that Dtþ1 4D tþ1 holds,
and that Dtþ1 ¼ FðDt Þ determines the levels of D v from period t þ 2 on. If the
competitive equilibrium path is monotonic (i.e., zh hD ozD holds for all D), we can
confirm that Dv 4D v holds for all vXt þ 2: On the other hand, when the
competitive equilibrium path is not monotonic around the steady state, it is not true
that Dv XD v holds for all vXt þ 2 for almost all cases; that is, there exists at least
one period, say period s, in which Ds oD s : To begin, using Fig. 4, let us explain this
point briefly, and then, we argue it rigorously in Proposition 4. Let us designate D^ tþ1 ;
which is seen in Fig. 4, the level of Dtþ1 which satisfies both FðD^ tþ1 Þ ¼ Dtþ2 and
D^ tþ1 oDtþ1 ;34and we designate m ^ the level of m which satisfies D ^
tþ1 ¼ Dtþ1 :
First, suppose that the level of a foreign medical aid is smaller than m; ^ that is,
mom: ^ Then D^ tþ1 oD tþ1 oDtþ1 holds, as seen in Fig. 4. In such a case, it can be

confirmed that D

tþ2 is larger than Dtþ2 : This implies that the level of generation
t þ 2’s welfare is reduced by the medical aid compared to the welfare when the
medical aid is not implemented.35 Secondly, let us consider the case in which m ¼ m: ^
In this case, D ^ ^
tþ1 is equal to Dtþ1 ; and thus, smaller than Dtþ1 : By definition of Dtþ1 ;
we can confirm that D
tþ2 ¼ Dtþ2 holds, and therefore, Dv ¼ Dv holds for all vXt þ
3: This indicates that when the level of foreign medical aid is equal to m; ^ this aid
improves the welfare of the recipient country in the Pareto sense; in this example,
generations t and t þ 1 are better off, and the generations who are born after period
t þ 2 are neither better off nor worse off by the medical aid. Finally, let us consider
the case in which m4m: ^ In this case, the medical aid improves the welfare of
generations t, t þ 1 and t þ 2: However, it cannot be guaranteed that D
v 4Dv holds
34
For simplicity, we consider the case in which D^ tþ1 exists uniquely in Fig. 4. In this regard, we will
argue rigorously in Proposition 4.
35
Note that dVdDðDt t Þ o0 holds.
ARTICLE IN PRESS
for all vXt þ 3; so that we need a more rigorous investigation. Investigating the
effect of the medical aid on the welfare of present and future generations rigorously,
we obtain the following results:
Proposition 4. Suppose that the government receives a one-shot medical aid from
foreign countries, and that by using the aid, the government provides a subsidy for the
health investment behavior of generation t. Then, the following can be stated:
(i) If the competitive equilibrium path is monotonic (i.e., zh hD ozD holds for all DÞ;
such an aid improves the welfare of agents in the Pareto sense: In other words, the
welfare of all generations from generation t on is raised by this aid.
(ii) If the competitive equilibrium path is cyclical around the steady state, the case in
which this aid improves the welfare of agents in the Pareto sense is measure 0: In
other words, there exists at least a generation whose welfare is reduced at almost
everywhere mt 2 Rþ :
The sketch of the proof for Proposition 4 is shown in Appendix E. Proposition 4
offers us an important implication concerning what medical aids should be. It
maintains that when the equilibrium path is cyclical, a counterintuitive result occurs;
that is, there exists a future generation that is damaged by the medical aid from
foreign countries in almost all cases. By receiving a one shot medical aid in period t,
the disease level will be suppressed in the next period. However, Proposition 4 insists
that we must continue to be vigilant against the disease when the pattern of the
prevalence of the disease is cyclical, and especially, that although the level of the aid
is large, a ‘one-shot’ aid is not desirable.36
6. Concluding remarks
We considered the prevalence of diseases and the agent’s preventive behavior

explicitly, and investigated a dynamical behavior of the economy, by using a small
open overlapping-generations model. Using the model, we derived an agent’s
‘prevalence-elastic behavior’, which is supported by empirical studies, and showed
that the agent’s prevalence-elastic behavior may cause the cyclicity of the spread of
disease. We also compared the competitive equilibrium allocation with the social
optimum allocation, and showed that the agent’s health investment in a competitive
equilibrium is insufficient when there exists net positive externality with respect to
the preventive behavior. Furthermore, we examined the effect of a one-shot medical
aid from a foreign country on the welfare of agents, and obtained an important
implication: when the pattern of the spread of disease is cyclical, there exists a future
generation that is damaged by the medical aid.
36
There exists preceding literature that shows that the foreign aid does not necessarily increase and even
may reduce the welfare of the recipient country, and this phenomenon is called as a transfer paradox. (In
this regard, see Haaparanta (1989) and Yanagihara (1998).) Although the result obtained in the present
analysis seems to resemble the transfer paradox, the mechanism by which this paradoxical result occurs is
quite different.
ARTICLE IN PRESS
The model of this paper is stylized and could be extended in many ways. First, it is
interesting to analyze extensively how the extent of disease spread influences
economic development. Recognition has grown in recent years of the fact that no
country can secure sustainable growth or poverty reduction without healthy, well-
nourished, and well-educated people. For example, the 2002 World Bank’s Health,
Nutrition, and Population Sector Strategy paper states, ‘Good health, nutrition, and
reproductive policies, and effective health services, are critical links in the chain of
events that allow countries to break out of the vicious circle of poverty, high fertility,
poor health, and low economic growth, replacing this with a vicious circle of greater
productivity, low fertility, better health, and rising incomes.’ Secondly, in order to
discuss the design of realistic and effective policies on public health and health
promotion, the incorporation of the explicit asymmetric information structure
between the government and individuals is necessary and significant. Hence, this
issue also represents a promising direction for future research.
Acknowledgements
We would like to thank two anonymous referees, Shingo Ishiguro, Yasushi

Iwamoto, Kazuo Mino, Chikashi Moriguchi, Katsuya Takii, and seminar
participants at Osaka University and Tezukayama University.
Appendix A. Proof for Proposition 1

qht
In this appendix, we show that qD t
40 holds. We can obtain the following
equations from (5)–(8) by differentiating totally with respect to c1 ; c2 ; h, D, and l and
using q̄t ¼ qðht ; Dt Þ:
0 1
qc1
B qD C 0 0
1
B C
B qc2 C B C
B C 0
B qD C B B
C
C
DB C
B qh C ¼ B bq uðc Þ C;
B C @ q B 12 2 C
B qD C 2 A
B C c 2
@ ql A r
qD
where D is defined as
0 00 1
u ðc1t Þ; 0; 0; 1
B qC
B 0; 00
bqu ðc2tþ1 Þ; 0; C
B rC
DB B 0;
C:
B bq1 u0 ðc2tþ1 Þ; bq11 uðc2tþ1 Þ; 1 CC
@ q q A
1; ; 1 þ 1 c2 ; 0
r r
ARTICLE IN PRESS
qht
Using the Cramer’s law, we can obtain qDt as follows:
qh jDh j
¼ ;
qDt jDj
where Dh is a 4 4 matrix which we can obtain by replacing the third column vector
of D; t ð0; 0; bq11 uðc2tþ1 Þ; ð1 þ qr1 c2 ÞÞ by t ð0; 0; bq12 uðc2 Þ; qr2 c2 Þ: jDj is calculated as
q q q 2
jDj ¼ u00 ðc1t Þ bq1 u0 ðc2tþ1 Þ 1 c2tþ1 þ 1 bq11 uðc2tþ1 Þ
r r r
q
bqu00 ðc2tþ1 Þ 1 c2tþ1 þ 1 b2 qq11 uðc2tþ1 Þu00 ðc2tþ1 Þ:
r
Therefore, jDjo0 holds. On the other hand, jDh j is calculated as
q q q 2
jDh j ¼ u00 ðc1t Þ bq1 u0 ðc2tþ1 Þ 2 c2tþ1 þ bq12 uðc2tþ1 Þ
r r r
q
þ 2 c2tþ1 bqu00 ðc2tþ1 Þ þ b2 qq12 uðc2tþ1 Þu00 ðc2tþ1 Þ:
r
Thus, when q12 40 holds, we can confirm that the sign of the numerator is negative,
dht
and thus that dD t
40 holds.
Appendix B. Proof for Proposition 2
To prove the existence of a non-trivial steady state equilibrium (Proposition 2(i)),

let us show that there exists a D40 which satisfies
D ¼ FðDÞ:
Showing that the following (1)–(3) hold true, we can confirm the existence of the
non-trivial steady state equilibrium:
(1) FðDt Þ is continuous.
(2) There exists a D which satisfies that FðD tÞ
Dt is less than unity.
FðDt Þ
(3) limD!0 Dt 41 holds.
To begin, let us examine that (1) holds. Since FðDÞ ¼ zðhðDÞ; DÞ holds by
definition, we can prove (1) by showing that both zðh; DÞ and hðDÞ are continuous. In
this regard, zðh; DÞ is assumed to be continuous, and as regards hðDÞ; the continuity
can be immediately confirmed by applying the Berge’s Maximum Theorem. Next,
FðDÞ x̄ðDÞ
we prove (2). Since FðDÞpx̄ðDÞ holds, D o D holds as well. Noting that
0
limD!þ1 x̄ ðDÞ ¼ 0 holds on the assumption of limD!þ1 z2 ðh; DÞ ¼ 0; it can be
confirmed that for a sufficient large value of D, x̄ðDÞ FðDÞ
D o1 holds, and so is D :
Finally, let us prove (3). When limD!0 hðDÞ ¼ 0 holds and hðDÞ is continuous, for
any ; there exists D which satisfies 0ohðDÞo: Noting that z1 ðh; DÞo0;
zð; DÞ FðDÞ zð0; DÞ
o o
D D D
ARTICLE IN PRESS
holds, and thus,

zð; DÞ FðDÞ zð0; DÞ
lim p lim p lim :
D!0 D D!0 D D!0 D
If limD!0 zðh; DÞ40 holds, it can immediately be confirmed that limD!0 FðDÞ D ¼ þ1;
and thus (3) holds trivially. On the other hand, when limD!0 zðh; DÞ ¼ 0 holds. we
0
can confirm that limD!0 zð0;DÞ D ¼ limD!0 x̄ ðDÞ holds by applying the L’opital’s
Theorem, and thus that we can confirm that limD!0 zð0;DÞ D 41: Furthermore, since
zðh; DÞ is continuous, zðh;DÞ
D is also continuous. Thus, we can guarantee the existence
of ¯ which satisfies that limD!0 zð¯D;DÞ 41 holds. Thus, limD!0 FðD tÞ
Dt 41 is proved.
Now let us prove the uniqueness of the non-trivial steady state equilibrium (i.e.,
Proposition 2(ii)). To prove it, we have only to show
(4) FðDÞ
D is a decreasing function with respect to D.
We can immediately obtain that this condition holds if and only if DF0 ðDÞ
FðDÞo0: From the definition of FðDÞ; we can obtain the following equation:
DF0 ðDÞ FðDÞ ¼ Dz1 h0 ðDÞ þ ½Dz2 z:
Here, noting that Dz2 zp0 hold (see footnote 29 on this point), we can confirm
that DF0 ðDÞ FðDÞo0 holds when h0 ðDÞ holds.
Now let us prove (iii). We can obtain from Dtþ1 ¼ FðDt Þ zðhðDt Þ; Dt Þ that
dDtþ1 zðhðDt Þ; Dt Þ
¼ ½zh hD þ zD :
dDt Dt
dD
Noting that zðhðDc Þ; Dc Þ ¼ Dc holds at steady state Dt ¼ Dtþ1 ¼ Dc ; the level of dDtþ1
t
evaluated at the steady state is expressed as follows:

dDtþ1
¼ czh chD þ czD : (29)
dDt Dt ¼Dc
Using (29), we can complete the proof for Proposition 2(iii).
Appendix C. Proof for Proposition 3
To begin, using (22) and (23), we express c2 as a function of c1 ; and express h as a

function of D, and denote them by c1 ¼ zðc2 Þ and h ¼ kðDÞ; respectively. It can be
seen that z0 ðÞ40 and k0 ðÞo0 hold. For k0 ðÞo0; it is useful to note that z2 ðh; DÞo1
holds in the steady state (see footnote 29 on this point).
Now let us define Gðc2 ; DÞ and Hðc2 ; DÞ as follows:
qðkðDÞ; DÞ
Gðc2 ; DÞ w hðDÞ zðc2 Þ c2 ;
r
Hðc2 ; DÞ uðc2 Þq1 ðkðDÞ; DÞ ru0 ðc2 Þ:
Here we note that Gðcc2 ; Dc Þ ¼ Gðcs2 ; Ds Þ ¼ 0 holds from (24); Hðcc2 ; Dc Þ ¼ 0 from
(25); and from (26) Hðcs2 ; Ds Þo0 if and only if a positive intergenerational external
ARTICLE IN PRESS
effect, bg z1 m; dominates a negative intragenerational one, q1 u0 ðc2 Þc2 : Furthermore,

noting that
qG qðkðDÞ; DÞ
¼ z0 ðc2 Þ o0;
qc2 r
qG c2
¼ k0 ðDÞ ½q1 k0 ðDÞ þ q2 40;
qD r
qH
¼ u0 ðc2 Þq1 ru00 ðc2 Þ40;
qc2
qH
¼ uðc2 Þ½q11 k0 ðDÞ þ q12 40;
qD
we can confirm that cc2 4cs2 and Dc 4Ds is the only consistent case if and only if
Hðcs2 ; Ds Þo0 holds. Since z0 ðÞ40 and k0 ðÞo0 hold, it can immediately be seen that
cc1 4cs1 and hc ohs hold in this case.
dht
Appendix D. Derivation of dm 40
By maximizing (2) subject to (27), we obtain the following first-order conditions:
u0 ðc1t Þ ¼ nt ; (30)
q̄t
bqðht ; Dt Þu0 ðc2tþ1 Þ ¼ nt ; (31)
r
bq1 ðht ; Dt Þuðc2tþ1 Þ ¼ ð1 tÞnt ; (32)
where nt denotes costate variables concerning (27). From (27), (28), and (30)–(32),
and using q̄t ¼ qðht ; Dt Þ hold in equilibrium we obtain the following equation:
0 1
dc1t
B dm C
B C
B dc2tþ1 C 0 0 1
B C
B dm C B C
B C 0C
B dh C B B C
B t C
GB C ¼ 0C B
C; (33)
B dm C B
B C B 0
C
B dnt C @ A
B C
B dm C 1
B C
@ dt A
dm
ARTICLE IN PRESS
where G is defined as
0 00 1
u ðc1t Þ; 0; 0; 1; 0
B q C
B 0; 00
bqu ðc2tþ1 Þ; 0; ; 0 C
B r C
B C
GB B 0; bq1 u0 ðc2tþ1 Þ; bq11 uðc2tþ1 Þ; ð1 tÞ; u0 ðc1t Þ C
C:
B q q1 C
B 1; ; c2tþ1 þ ð1 tÞ ; 0; ht C
@ r r A
0; 0; t; 0; ht
dht
Applying the Cramer’s law to (33), we can obtain dm as
dht jGh j
¼ ;
dm jGj
where Gh is a 5 5 matrix which we can obtain by replacing the third column vector
of G; t ð0; 0; bq11 uðc2tþ1 Þ; ðqr1 c2tþ1 þ ð1 tÞÞ; tÞ by t ð0; 0; 0; 0; 1Þ: We can confirm that
the signs of jGh j and jGj are both negative, and thus that dh t
dm 40 holds.
Appendix E. Sketch of Proof for Proposition 4
As regards Proposition 4(i), the proof is immediate and obvious, so that we

describe only the proof for Proposition 4(ii); in other words, we consider only the
case in which the equilibrium path is cyclical around the steady state.
To begin, let us consider the case in which the steady state equilibrium is stable.
Then, it is guaranteed that there exists a period s such that for all period jXs ; the
equilibrium path is cyclical (that is, zh hD 4zD holds at Dj ). Similarly, as far as
tþ1 a0 holds,
37
D it can also be guaranteed the existence of period s such that for all
period jXs ; the equilibrium path is cyclical. Here, let us define s as maxfs ; s g and

compare the level of D with D : In this regard, we can consider the following three
s ; (2) Ds oDs ; and (3) Ds ¼ Ds :
cases: (1) Ds 4D
When (1) occurs, generation s’s welfare is reduced by a foreign medical aid, and
when (2) occurs, generation s þ 1’s welfare is reduced. Therefore, it can be seen that
a necessary condition that the foreign medical aid improves the present and future
generations’ welfare in the Pareto sense is that case (3) takes place.
tþ1 a0
37
Assuming that zðh; DÞ ¼ 0 holds only if D ¼ 0 or h tends to infinity, we can guarantee that D
holds when mo þ 1:
In this regard, Geoffard and Philipson (1997) argued, ‘‘A major technology to limit the infectious
diseases is vaccine. Although the introduction of a vaccine usually produces a sharp drop in the occurrence
of a disease, the eradication of vaccine-preventable diseases predicted by many at that time of these
interventions has never been achieved. Of the roughly 40 vaccines on the market, only the smallpox
vaccine has been successful in eradication.’’ Taking this into account, it can be said that such assumption is
justified from a realistic viewpoint.
ARTICLE IN PRESS
Now we define a set LðsÞ

s1 as follow:
LðsÞ
s1 fDs1 jFðDs1 Þ ¼ Ds ; Ds1 oDs1 g:

Hereafter, let us limit our analysis in the case in which the number of elements in set
G which is defined as G fDj jF0 ðDj Þ ¼ 0g is finite.38 In this case, it is guaranteed that
the cardinality of LðsÞ
s1 is finite and not empty. Furthermore, based on Ls1 ; we
ðsÞ
ðsÞ
define a set Lsj for 2pjps ðt þ 1Þ as the following manner:
LðsÞ ðsÞ
sj fDsj jFðDsj Þ 2 Lsjþ1 ; Dsj pDsj g:

It can be confirmed that the cardinality of each set LðsÞ sj is finite.
ðsÞ
Now suppose that D tþ1 2 L tþ1 holds as a consequence of receiving a foreign
medical aid in period t. (Note that sXt þ 2 because Dtþ1 4D tþ1 necessarily holds
ðsÞ
when mt 40:) It can be confirmed that only if D tþ1 2 L tþ1 holds, the foreign medical
aid can improve the welfare of recipient country in the Pareto sense. That is,
comparing a sequence ðDt ; D
tþ1 ; Dtþ2 ; . . .Þ with ðDt ; Dtþ1 ; Dtþ2 ; . . .Þ; we can confirm
that the welfare of generations from t (the present generation) to s 1 is improved,
and that as regards after generation s; they are not better off nor worse off by the
foreign medical aid.
T1
Using LðsÞ
tþ1 ; let us define a set Ltþ1 as Ltþ1
ðsÞ
s¼tþ2 Ltþ1 : Then, it can be
confirmed that the cardinality of Ltþ1 is at most countable finite. Finally, we define
qD
Dt Rþ as a set of m 2 Rþ which satisfies D tþ1 2 Ltþ1 : Noting that qm o0 holds,
tþ1
it can be concluded that the cardinality of Dt is at most countable infinity, which

implies Dt is measure 0 on Rþ : Therefore, when the steady state equilibrium is stable,
Proposition 4(ii) holds true.
As regards the case in which the steady state equilibrium is unstable, we cannot
guarantee the existence of s and s in general. However, strictly speaking, the
essential part of the logic of the proof mentioned above can be valid if only there
exists two successive periods, say periods v and v þ 1; at which both equilibrium
paths ðDt ; Dtþ1 ; Dtþ2 ; . . .Þ and ðDt ; D
tþ1 ; Dtþ2 ; . . .Þ are in cyclical phases simulta-
neously. (If so, we can apply the argument that ‘(3) Ds ¼ D s is a necessary
condition in order that a foreign medical aid may attain Pareto improvement’ in this
case as well.) It is useful to note that the number of emergence of cyclical phases is
39
countably infinite in both ðDt ; Dtþ1 ; Dtþ2 ; . . .Þ and ðDt ; D
tþ1 ; Dtþ2 ; . . .Þ: Taking this
into account, the probability that the there exists no two successive periods at which
both equilibrium paths ðDt ; Dtþ1 ; Dtþ2 ; . . .Þ and ðDt ; D
tþ1 ; Dtþ2 ; . . .Þ are in cyclical
phases simultaneously can be regarded as measure 0. Therefore, (the sketch of) proof
of Proposition 4(ii) is completed.
38
In other words, we preclude the case in which the level of FðDÞ is constant at an interval. Taking
account of the probability that such a case occurs can be regarded as measure 0, we can consider that the
preclusion is appropriate when we prove Proposition 4(ii).
39
When we suppose that the number of emergence of cyclical phases is finite in ðDt ; Dtþ1 ; Dtþ2 ; . . .Þ
(or ðDt ; D
tþ1 ; Dtþ2 ; . . .Þ), we can easily confirm that the supposition leads to contradiction.
ARTICLE IN PRESS
References
Ahituv, A., Hotz, V., Philipson, T., 1996. The responsiveness of the demand for condoms to the local
prevalence of AIDS. Journal of Human Resources 31, 869–898.
Anderson, R.M., May, R.M., 1991. Infectious Diseases of Humans. Oxford University Press.
Azariadis, C., 1993. Intertemporal Macroeconomics. Brackwell Publishers, Cambridge, Massachuetts.
Baierl, G., Nishimura, K., Yano, M., 1998. The role of capital depreciation in multi-sectoral models.
Journal of Economic Behavior and Organization 33, 467–479.
Benhabib, J., Nishimura, K., 1985. Competitive equilibrium cycles. Journal of Economic Theory 35,
284–306.
Blanchard, O.J., 1985. Debt, deficits, and finite horizons. Journal of Political Economy 93, 223–247.
Bouma, M.J., van der Kaay, H.J., 1994. Epidemic malaria in India and the EL Niño southern oscillation.
Lancet 344, 1638–1639.
Bouma, M.J., Sondorp, H.E., van der Kaay, H.J., 1994. Climate change and periodic epidemic malaria.
Lancet 343, 1440.
Chakraborty, S., 2004. Endogenous lifetime and economic growth. Journal of Economic Theory 116,
119–137.
Eckstein, Z., Wolpin, K.I., 1985. Endogenous fertility and optimal population size. Journal of Public
Economics 27, 93–106.
Gallup, J.L., Sachs, J.D., 2001. The economic burden of malaria. American Journal of Tropical Medicine
and Hygiene 64, 85–96.
Gayle, H.D., Counts, G.W., 2001. Syphilis elimination: a unique time in history. Journal of the American
Medical Women’s Association 56, 2–3.
Geoffard, P., Philipson, T., 1997. Disease eradication: public and private vaccination. American Economic
Review 87, 221–231.
Gersovitz, M., Hammer, J.S., 2004a. The economical control of infectious diseases. Economic Journal 114,
1–27.
Gersovitz, M., Hammer, J.S., 2004b. Infectious diseases, public policy, and the marriage of economics and
epidemiology. World Bank Research Observer 18, 129–157.
Haaparanta, P.J., 1989. The intertemporal effects of international transfers. Journal of International
Hay, S.I., Cox, J., Rogers, D.J., Randolph, S.E., Stern, D.I., Shanks, G.D., Myers, M.F., Snow, R.W.,
2002. Climate change and the resurgence of malaria in the East African highlands. Nature 415,
905–909.
Maeda, Y., Akai, N., 1998. An optimal tax scheme to finance social security in a model with endogenous
longevity. Journal of Economic Research 3, 51–67.
Mullahy, J., 2000. It’ll only hurt a second?: microeconomic determinants of who gets flu shots. Health
Nájera, J.A., Kouznetsov, R.L., Delacollette, C., 1998. Malaria epidemics detection and control
forecasting and prevention. World Health Organization, Geneva.
Nishimura, K., Yano, M., 1995. Nonlinear dynamics and chaos in optimal growth. Econometrica 63,
981–1001.
Over, M., 1992. The Macroeconomic Impact of AIDS in Sub-Saharan Africa. Washington, DC, World
Bank, Africa Technical Department, Population, Health and Nutrition Division, 1992. Technical
Working Paper No.3.
Pecchenino, R.A., Pollard, P.S., 1997. The effects of annuities, bequests, and aging in an overlapping
generations model of endogenous growth. Economic Journal 107, 26–46.
Philipson, T., 1996. Private vaccination and public health: an empirical examination for U.S. measles.
Journal of Human Resources 31, 611–630.
Philipson, T., 2000. Economic epidemiology and infectious diseases. In: Culyer, A.J., Newhouse, J.P.
(Eds.), Handbook of Health Economics 1B. Elsevier Science, Amsterdam.
Philipson, T., Becker, G.S., 1998. Old-age longevity and mortality-contingent claims. Journal of Political
Economy 106, 551–573.
ARTICLE IN PRESS
United Nations and World Health Organization, 1991. The AIDS epidemic and its demographic
consequences. United Nations and World Health Organization, New York and Geneva.
United Nations Population Division, 2001. World Population Prospects: The 2000 Revision Highlights.
Population Division Department of Economics and Social Affairs, United Nations, New York.
World Bank, 1996. Confronting AIDS: Public Priorities in a Global Epidemic. Oxford University Press,
Oxford.
World Bank, 2001. Malaria at a glance report. World Bank, New York.
World Bank, 2002. Health, nutrition, and population sector strategy paper. World Bank, New York.
World Health Organization, 1999a. Removing Obstacles to Healthy Development. World Health
Organization, Geneva.
World Health Organization, 1999b. Malaria 1982–1997. Weekly Epidemiological Record 74, 265–270.
World Health Organization, 2001. Report of the commission on macroeconomics and health. World
Health Organization, Geneva.
Yaari, M.E., 1965. Uncertain lifetime, life insurance, and the theory of the consumer. Review of Economic
Studies 32, 137–150.
Yanagihara, M., 1998. Public goods and the transfer paradox in an overlapping generations model.
Journal of International Trade and Economic Development 7, 175–205.

Infectious Disease and Preventive Behavior in An

Uploaded by

Copyright:

Available Formats

You might also like

Infectious Disease and Preventive Behavior in An

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Infectious Disease and Preventive Behavior in An

Uploaded by

Copyright:

Available Formats

ARTICLE IN PRESS

Journal of Economic Dynamics & Control 29 (2005) 1673–1700

Infectious disease and preventive behavior in an

JEL classification: I10; I12; I18; O41

Keywords: Infectious disease; Health investment; Overlapping generations model; Prevalence-elastic

According to the World Health Organization (WHO) (1999a), infectious diseases

Life Expectancy with AIDS

important to investigate agents’ preventive behavior against infectious diseases in a

developing countries established such essential health interventions today, future

We use a two-period overlapping-generations model in which lifetime uncertainty

3. Dynamical equilibrium path

In this section, we examine the dynamical system of this economy. A dynamical

dV ðDt Þ dc1 ðDt Þ dc2 ðDt Þ dhðDt Þ

Fig. 3. Competitive equilibrium path.

Proposition 2. (i) If limDt !0 hðDt Þ ¼ 0 holds, then a non-trivial steady state

The proof of Proposition 2 is shown in Appendix B. Furthermore, for Proposition

4. Social optimum vs. competitive equilibrium

c1t þ qðht1 ; Dt1 Þc2t þ ht þ btþ1 þ k ¼ w þ rðk þ bt Þ; ðtX1Þ; (11)

Dt ¼ zðht1 ; Dt1 Þ; ðtX1Þ: (12)

The optimal solution is obtained by setting up the Lagrangian function:

Thus we can obtain the optimal conditions:

bu0 ðc2t Þ ¼ glt ; (14)

q1 ðht ; Dt Þfbuðc2tþ1 Þ gltþ1 c2tþ1 g lt þ gmtþ1 z1 ðht ; Dt Þ ¼ 0; (16)

q2 ðht ; Dt Þfbuðc2tþ1 Þ gltþ1 c2tþ1 g mt þ gmtþ1 z2 ðht ; Dt Þ ¼ 0: (17)

Dtþ1 ¼ zðht ; Dt Þ: (20)

D ¼ zðh; DÞ; (23)

bðuðcs2 Þ u0 ðcs2 Þcs2 Þq2 ðhs ; Ds Þ ¼ ð1 gszD Þms ;

Proposition 3. Comparing the allocation attained in the competitive equilibrium with

5. Effect of medical aid from a foreign country

Fig. 4. Case in which Dtþ2 oD

For the level of ht ; we can obtain a natural result that dh t

conﬁrmed that D 

We considered the prevalence of diseases and the agent’s preventive behavior

We would like to thank two anonymous referees, Shingo Ishiguro, Yasushi

Appendix A. Proof for Proposition 1

Appendix B. Proof for Proposition 2

To prove the existence of a non-trivial steady state equilibrium (Proposition 2(i)),

holds, and thus,

Appendix C. Proof for Proposition 3

To begin, using (22) and (23), we express c2 as a function of c1 ; and express h as a

effect, bg z1 m; dominates a negative intragenerational one, q1 u0 ðc2 Þc2 : Furthermore,

By maximizing (2) subject to (27), we obtain the following ﬁrst-order conditions:

bq1 ðht ; Dt Þuðc2tþ1 Þ ¼ ð1 tÞnt ; (32)

Appendix E. Sketch of Proof for Proposition 4

As regards Proposition 4(i), the proof is immediate and obvious, so that we

Now we deﬁne a set LðsÞ

it can be concluded that the cardinality of Dt is at most countable inﬁnity, which

You might also like

Fig. 4. Case in which Dtþ2 oD

conﬁrmed that D