Happily ever after... or until we run out of money. Effects of longevity on financial vulnerability

Cheerfully ever after ...
or until we run out of money. Effects of longevity on financial vulnerability.
Magda Malec, Krzysztof Makarski, Joanna Tyrowicz
ING Think Forward Innitative 2018
FAME|GRAPE

Living longer is good news, but brings about important challenges
We all think that we are cautious and take proper care of our relatives
and our own future. People make precautionary savings and employ
various strategies to make sure they are not out of money when the
situation becomes dire. Choices differ between people, depending on
financial situation, degree of optimism and attitude towards risk, but
we all like to think we came up with a smart way to secure our future.
However, a high fraction of elderly individuals cannot afford necessary
out-of-pocket health expenditure towards the end of their life. As ar-
gued by Marshal et al. (2011), the elderly from the bottom quartile
of income have out-of-pocket expenditure by roughly 30% lower than
the median. Since the out-of-pocket expenditures cover items such
as prescription drugs, home health care and helpers and physiother-
apy, all of which were shown to have substantial effect on longevity
(Hemmelgarn et al. 2007, Mohile et al. 2015, Lilley et al. 2016).
How to make the right decision? One way to think about it is to
take into account life expectancy: how much longer is my individual
birth cohort expected to live? However, with growing longevity, the life
expectancy as observed among current elderly individuals substantially
understates the life expectancy to be experienced by current young and
middle-aged individuals. Within the horizon of the coming 20 years,
an average 65 year old will expect to live for 5-7 years longer than a
person who turns 65 currently, often in good health. This impressive
increase in life expectancy at older ages comes also with improved
health and thus potentially brings about quality years of life.
Obviously, unless we are able to fully internalize the consequences of
increased longevity, we are all running the risk of saving too little. After
all, living 5 years longer than our grandparents is roughly 30% longer
retirement period than they had a chance to experience! Consequently,
longevity increases financial vulnerability of the elderly: if savings are
insufficient to supplement consumption after leaving the labor market,
as we age, we run a greater risk of old age poverty.
This risk will be more acute in the case of countries which implemented
a defined contribution pension system. Many Central and Eastern Eu-
ropean countries implemented in late 1990s or early 2000s a pension
system reform which yields substantially lower pension benefits, rela-
tive to those paid contemporaneously to the retirees. These reforms
were introduced with the objective to immunize the pension systems
against longevity, which makes their fiscal stance more viable in the
long run, but at the cost of large decline in pension benefits.
The currently working generations in those countries are then likely to
experience two negative surprises at once. First, their pension benefits
will be lower than currently observed. Second, their own savings will
have to last over a longer life expectancy. As a consequence one should
expect high poverty rates among the elderly and pensioners financial
vulnerability.
The lower than currently observed pension benefits should trigger ad-
justments in private voluntary saving behavior: being smart and for-
ward looking, we should take the necessary steps to account for future
financial needs. However, even with lower longevity individuals have
been demonstrated to save too little. We think we are going to save
more tomorrow, when indulging our today craving for current con-
sumption. A way to address this issue is to provide individuals with
smart incentives, instruments that can encourage responsible saving
strategies. Against this background, our report provides answers to
the following four questions:
□ What is the scope of old-age poverty rise due to the
increased longevity?
□ What are the consequences of longevity for inequality?
□ What can be done to contain the old age poverty?
□ How would it affect inequality?
FAME|GRAPE || ING Think Forward Initiative || 2018 1

Four questions about old age poverty and longevity
□ What is the scope of old age poverty rise due to the increased longevity?

Longevity matters for poverty more in defined contribution systems
Higher life expectancy, that we are all glad to observe in the future, is associated with two important changes. First, a smaller fraction of us will be
at risk of not making it to the old age. This improvement owes to healthier life styles, better diagnostics of cardiovascular diseases and more
effective treatment techniques for many diseases which in the past resulted in prime age mortality. Second, a larger fraction of us will enjoy more
than 20 years after 60th birthday. While in the 1960s only few percent of those who made it to the 60th birthday, blew 80 candles as well, in the
years to come this fraction will exceed 70%. In fact, of the generation born around the 2000s, 86% will reach the age of 65 and life expectancy at
65th birthday is projected to be roughly 20 more years for men and 23 years for women.
Longer life means that a larger number of years will be spent in retire-
ment. In pension systems which pay out a fixed share of pre-retirement
earnings (the so-called defined benefit systems, DB), longevity implies
higher fiscal costs: more pensions will be paid out to each retiree. In
pension systems which pay out what was contributed during the work-
ing years (the so-called defined contribution systems, DC), longevity
implies low pensions. How low they become depends on how the in-
crease in life expectancy is split between years of working longer (and
thus contributing more) and years spent in retirement (and thus col-
lecting pension benefits).
For example, an average 30 year old in 2015 expects to live for another
50 years. Of this time, an average representative of the Millennials
generation is likely to spend about 25-30 years working and another
25 years in retirement. In a DB pension system, a quarter of a century
in retirement means a huge fiscal cost, so perhaps our Millennial should
worry about tax raises, but the pension benefit would be the same as
we see with our grandparents today: roughly 60% of their average
monthly earnings.
Meanwhile, in a DC pension system, such proportion between working
years and years in retirement implies that contributing about 20% of
wages for old age benefits, one can expect a replacement rate of no
more than roughly 30% of their average monthly earnings. That is
pension benefits will be no more than a half of what we see currently,
relative to earnings.
In fact, for the generations who reach old age currently, private con-
sumption continues to be high, whereas with the aging increases the
role of public consumption of the health care services (Istenic et al.
2017). It is rare at all that the individual consumption would de-
cline substantially around the retirement age. In the future, decline in
consumption may come as a shock.
For high earning professionals, effectively holding a well paid job all
their lives, 30% of their average annual earnings can still be enough
to support a satisfactory life-style, even if it is then substantially lower
than during the years of active professional career. Naturally, the out-
of-pocket health care expenditure of several thousand euros per month
may be impossible, thus limiting their life expectancy, let alone quality
of life towards the old age. Yet, making the ends meet for a greater part
of old age seems feasible. However, for many individuals the 30 years
of career may comprise some periods of career interruptions (such as
job-seeking or inactivity) as well as periods of relatively lower earnings.
Given potentially lower labor market attachment and earnings, many
individuals will become at risk of poverty.
Once we account for that heterogeneity, we can simulate the future
evolution of poverty and old age poverty in the decades to come, as
longevity progresses. We do that for 6 Central European economies,
because they all replaced a DB system with a DC one at the turn
of the millennia, which means that all of the Millennials will collect
pensions from a defined contribution pension system.

Stronger incentives contribute to income inequality in DC pension systems
An important consequence of replacing the defined benefit with de-
fined contribution pension systems relates to labor market incentives.
With defined benefit systems, there are only direct incentives to work:
the wages. The contributions paid on the earned income are effec-
tively a tax: our social selves know that this is social contribution and
inter-generation exchange, but our economic selves view social secu-
rity contribution as a pure tax, because pension benefits do not grow
in the contributions. However, with the defined contribution systems,
there are also indirect incentives to work, because the more we con-
tribute to the pension system, the higher the future pensions. These
stronger labor market incentives are one of the reasons why defined
contribution systems are believed to deliver higher economic efficiency.
.25
.3
.35
.4
.45
Gini
2000 2020 2040 2060 2080
year
Bulgaria Czech Hungary
Poland Romania Slovakia
Income inequality
Definitions in figure: Income inequality measured by Gini ratio. Income definition
includes earned income and pension benefits, but excludes capital gains. The
computed inequality measure includes weights for population structures.
The extent to which individuals respond to these indirect incentives
are heterogeneous. Those individuals, who are free to increase labor
supply will likely do so after the reform, in order to fully internalize
the benefits of higher pension benefits in the old age from higher con-
tributions during prime-age. Meanwhile, for some individuals higher
number of working hours may continue to be impossible despite the
favorable incentives. There can also be a group that does not spot
the stronger incentives and thus does not adjust behavior in any way
despite a change in pension system. Hence, reforming the pension
system increases income inequality via labor supply decisions.
The stronger labor market incentives, coupled with longevity, have
large effects on income inequality.Indeed, in absence of longevity, in-
come inequality could decline, as all individuals are subjected to the
stronger incentives of the defined contribution system. This appears
to be the case of Poland, Bulgaria and to a lesser extent also Hungary.
The heterogeneous reaction to the stronger incentives contributes to
larger income inequality in Czech Republic, Slovakia and Romania.
These diverse paths of income inequality reflect the fact that the abil-
ity to adjust labor supply across these countries differs.
Longevity leads universally to a substantial increase in income inequal-
ity. The Gini ratio for income inequality will increase by at least 4 per-
centage points and in some countries as much as 14 percentage points.
This rise is substantially bigger than experienced by those countries
when transitioning from central planning to a market economy. These
large income effects contribute to higher poverty for those individuals,
whose labor supply did not increase in reaction to the reform. Hence,
the overall rise in poverty rates after replacing DB with DC systems
stems from two sources: lowering pension benefits under longevity in
the DC scheme and growing income inequality.
The reason why the increase in inequality flattens out towards the
end of the simulated period is due to the fact that the extension in
expected lifetime duration is gradually slower.

A rational response to longevity and its consequences for wealth inequality
When expecting low effective replacement rates from the public pen-
sion system, accumulating private savings is a way to take precautions.
This private and voluntary pension wealth, can then be used to co-
finance old age consumption and thus cushion the decline of living
standards at retirement when low pension benefits replace earned in-
come.
The scope and direction of changes in savings patterns will depend on
the extent to which one is willing to give up current consumption in
exchange for a lower decline in consumption at old age. Some indi-
viduals particularly value smooth lifetime path of living standards and
are thus willing to accumulate more savings. Some other individu-
als prefer to take out substantial loans to smooth consumption over
lifetime. For others, low consumption in the future is not enough of
a driver for current decision making. This is portrayed by solid lines
in the upper graph. Expecting decline in pensions due to increase in
life expectancy, some households accumulate more private, voluntary
savings, but for others it is optimal to continue with some level of
indebtedness. This is portrayed by dashed lines in the upper graph.
The net effect of these changes depends on the population structure
and country-specific distribution of preferences. In all six countries
majority of social groups will substantially increase savings. However,
in Bulgaria and especially in Poland, debt taking intensifies as well.
Upper graph: Solid lines portray life cycle savings behavior of current generation,
dashed lines portray savings behavior of people born 60 years from now. Thin lines
represent individuals with limited ability to save for old age consumption, while thick
lines represent individuals with ability and preference to save for old age
consumption. The effective retirement age is country-specific, see p. 20.
Lower graph: Life cycle savings behavior of individuals with different levels of
abilities and preferences. The scatter plots savings behaviors of social groups within
current generation (horizontal axis) and future generation (vertical axis).
0
20 40 60 80 100
age
Bulgaria
0
20 40 60 80 100
age
Czech
0
20 40 60 80 100
age
Hungary
0
20 40 60 80 100
age
Poland
0
20 40 60 80 100
age
Romania
0
20 40 60 80 100
age
Slovakia
Evolution of wealth accumulation over the life−cycle
45 degree line
0
50
100
0 10 20 30 40 50
Bulgaria
45 degree line
0
20
40
60
80
100
0 10 20 30 40 50
Czech
45 degree line
0
20
40
60
80
100
0 20 40 60
Hungary
45 degree line
−10
0
10
20
30
0 5 10 15
Poland
45 degree line
0
10
20
30
−5 0 5 10 15
Romania
45 degree line
0
20
40
60
...amonggenerationborn60yearsfromnow
0 10 20 30
... among current generation
Slovakia
distribution of assets ...

Country-specific changes in wealth inequality patterns
Two effects are at play. First, savings behavior changes with longevity,
as we discussed above, displaying the life time patterns of saving and
dissaving. Second, due to aging, the population structure changes: a
higher share of living individuals are those who accumulated savings
for co-financing the old age consumption, whereas a smaller fraction
of populations are those individuals who are at the beginning of their
working career. The joint realization of these two effects determines
changes in aggregate inequality.
Indeed, the direction and path of change in wealth inequality is not
preconceived. While the preferences and abilities of individuals do not
change in our simulations, longevity translates differently to behaviors,
yielding disparity in terms of wealth accumulation. Some social groups
increase savings, whereas others do not, which, jointly with changing
population structure, drives changes in wealth inequality. The increase
in wealth accumulation occurs in many social groups in all 6 countries,
but the rate of increase and rate of aging jointly determine the speed
and scale of adjustment in wealth inequality.
Wealth inequality increases in Poland, remains stable in Bulgaria and
declines in the remaining 4 countries: Czech Republic, Hungary, Ro-
mania and Slovakia. Behind growth in inequality lies the persistent
indebtedness of some social groups in Poland and to a smaller extent
in Bulgaria. Noticeably, indebtedness in the 4 countries is low already
with current life expectancy and will further decline with longevity.
Meanwhile, in Poland it is high and will increase with longevity.
An increase in indebtedness and persistently low consumption path
show that with longevity, position of the vulnerable groups worsens
relative to what is observed currently. Further decline in voluntary
savings and high indebtedness raise even more the relevance of social
as well as macroprudential risks. Meanwhile, further increase in wealth
accumulation is bound to increase liquidity in the financial system,
ushering the risk of excessively lax regulation on credit.
The waves of changes in wealth inequality are associated with the de-
mographic waves: post-war baby boomers, early 1980s fertility boom,
etc. These waves result in unequal size of the subsequent cohorts and
thus translate to waves in inequality savings as observed in given point
in time. Note, that declining or increasing aggregate inequality need
not imply that within a given birth cohort wealth disparity widens.
Even if this channel was not operational, aging of the societies would
yield changes of wealth inequality.
.5
.55
.6
.65
.7
Gini
2000 2020 2040 2060 2080
year
Wealth inequality
Definitions in figures: Wealth inequality measured by Gini ratio. Wealth definition
includes private voluntary savings, i.e. the share of income not spent on
consumption, with accrued interest. Households are allowed to hold negative assets
(debt), but are constrained not to leave debt on subsequent generations. The initial
ratios were set to match the observed wealth inequality as observed prior to the
reform in these economies, weighted for population structures.

Longevity affects consumption inequality as well
Households which want to cushion a decline in consumption, observ-
ing longer life expectancy, need to increase savings. Even those with
low income levels will reduce consumption further. This process implies
that increased longevity will be associated with an increased consump-
tion inequality.
Moreover, the households that decide not to make precautions for
the low pension benefits, will experience low consumption during old
age. This phenomenon will increase the dispersion of consumption as
observed in the economy at any given point in time, thus contributing
to increasing measures of consumption inequality measures.
.2
.25
.3
.35
.4
Gini
2000 2020 2040 2060 2080
year
Consumption inequality
These two effects result in increased consumption inequality in addi-
tion to wealth inequality. Consumption inequality increases in the six
analyzed countries by roughly 3-5 percentage points when measured
by Gini ratio. This magnitude of the effect is substantially larger than
an increase in consumption inequality as observed by Spain or Greece
during the global financial crisis. Naturally, this increase is gradual, as
opposed to changes in inequality caused by business cycle fluctuations
and thus may be masked by short-run fluctuations.
Notably, growth in consumption inequality is smaller than the growth
in income inequality. This finding stems from the fact that households
smooth consumption over lifetime, hence avoiding extreme outcomes
such as high consumption in prime-age and low consumption in old
age. Even if this smoothing is not possible to the full extent for those
households who have constraints on the ability to save, the effects for
consumption inequality will be lower than for income or wealth.
Note that in our setup, households are free to reduce consumption as
much as they deem optimal, i.e. there is no threshold for subsistence
consumption. Neither do we model the actual minimum existence
costs, such as those captured by equivalence scales for absolute poverty
measures. Therefore, our measure of consumption inequality is the
lower bound. If it were the case that in a perfectly elastic world
households were choosing consumption levels below those that are
feasible in the real world, then the actual increase in consumption
inequality will be larger than forecast here.
Definitions in figure: Consumption inequality measured by Gini ratio. Consumption
definition includes all current household consumption. The initial values match
consumption inequality as observed in these countries prior to the introduction of the
pension system reform from a DB to a DC system. The computed inequality
measure includes weights for population structures.

People do not save enough for their retirement
Conventional economic models predict that people smooth consump-
tion over life cycle, hence rational old age poverty is only consistent
with extremely high preference for immediate consumption. However,
as has been demonstrated by survey and health studies, people do
not save “enough” for retirement. One important source of these in-
sufficient savings is the unforeseen and largely unexpected increase in
longevity. The other important source is of behavioral nature: setting
and carrying out the savings plan is difficult, requires self-control and
consistency over a long horizon (Benartzi and Thaler 2004). Typically,
a middle-class individual or a couple saves for retirement in two ways:
home ownership and compulsory state-run pension systems.
In the case of housing wealth, self-control is of secondary relevance.
Mortgage payments are claimed regularly by the banks who lent the
money for buying the property. Whether it is efficient to pay interest
on pension savings is naturally an open question, but one does not need
to demonstrate self-control and consistency to accumulate wealth that
can be used to co-finance consumption during old age.
In the case of the public pension systems self-control is not that rele-
vant either: every month the pay check that arrives from the employer
is already net of social security contributions, hence one does not even
“feel” accumulating pension wealth. Indeed, public pension schemes
are one of the so-called low-willpower savings techniques (Gustman
and Steinmeier 1998).
However, with longevity the savings in household and pension wealth
are likely to fall short of desirable savings in the old age. The first
hurdle occurs already at calculating ex ante, how much saving is de-
sirable. Financial literacy on average is too low to let individuals com-
pute the desirable level of wealth accumulation. Without these tools,
individuals go for either of the two extremes: minimum to satisfy the
eligibility criteria or maximum allowed by the legal regulations (Be-
nartzi and Thaler 2004). There is no reason for this value to be the
optimal level of savings.
The second hurdle concerns self-control and will-power. Choi et al.
(2004) show that even people who deem their savings too low and
declare a resolution, usually fail to do so: 86% of the individuals
failed to save more despite expressing a will to do so within a quarter.
Procrastination (sometimes referred to as hyperbolic discounting) is
one of the important reasons behind undersaving. People unjustifiably
assume that “tomorrow” they will do what they should have done
“today” – a mechanism that reflects in time-inconsistent decisions,
because, as “tomorrow” comes, it becomes “today”.
Another reason why sticking to an original saving plan may be difficult
is called status quo bias (Samuelson and Zeckhauser 1988). Retire-
ment plans choices reveal that people are biased to do nothing or
maintain current course of action. Therefore, they react too little to
information about expanding life expectancy or declining rates of re-
turn on accumulated savings. In the face of actuarial information, 75%
of people do not increase retirement saving even though they judge
their savings to be too low.
Saving more may also be unpleasant. Kahneman and Tversky (1992)
argued that people are loss averse – we have a tendency to regret
losses much more than we appreciate gains more than regret losses.
Since setting more disposable income aside for old age consumption
appears to be a loss of current consumption, people are reluctant to
save more.
Given this multiplicity of reasons, one should expect that a substantial
fraction of population saves too little relative to their preferred path
of life time consumption. Our objective here is to analyzewhat would
happen if suddenly everybody gained ability and taste for responsible
old age saving. We do not change the aggregate preferences in the
societies: those societies who are more patient and thus generally have
higher savings and investment rate remain so. What we do change,
though, is to equalize the access to the instruments which permit
smooth lifetime consumption.

Fostering equal access to saving instruments reduces (old age) poverty
In the six analyzed economies, individuals have heterogeneous ability
save – unequal access to financial instruments. Prior to reforming the
pension system, we assume that some individuals were able to con-
sistently put aside a higher fraction of their income. Pension system
reform provides incentives to both increase hours worked and raise the
share of earned income set aside for old age consumption. These in-
centives interact with longevity, changing the optimal set of decisions.
We run two scenarios. In the first scenario, all individuals live longer
and receive pensions from a new pension system with unequal access
to financial instruments. In the second scenario, we additionally make
access to financial instruments universal for everyone.
Fostering equal access to saving instruments reduces (old age) poverty
in all six analyzed countries. Solid lines portray the evolution of old
age poverty if the ability to save remains as unequal as we find in the
data in late 1990s in the six analyzed economies. The dashed lines
show what would happen if suddenly access to financial services was
perfectly equal across all the social groups in these six countries. Equal
access does not need to imply that everybody saves the same amount,
quite the opposite, savings continue to reflect preferences for smooth
lifetime consumption as well as expected longevity and macroeconomic
tendencies. Effectively, the dashed lines portray the evolution of old
age poverty if there are no frictions, limitations, constraints or privi-
leges in accessing savings instruments.
Reduction in poverty is large enough to counterweight otherwise domi-
nant trends related to longevity. The fact that equal access to financial
instruments outweighs the effects of demographic processes shows the
scope for potential policy to mitigate old age poverty.
Definitions in figure: Consumption poverty measured as a share of households with
consumption below 60% of median consumption. Old age poverty measured as share
of old age poor among poor. The initial values match poverty levels as observed in
the six countries prior to the introduction of the pension system reform from a DB to
a DC system. The inequality measure includes weights for population.
.05
.1
.15
.2
2000 2040 2080
year
unequal access (solid)
equal access (dash)
Bulgaria
.05
.1
.15
.2
2000 2040 2080
year
equal access (dash)
Czech
.05
.1
.15
.2
2000 2040 2080
year
equal access (dash)
Hungary
.05
.1
.15
.2
2000 2040 2080
year
equal access (dash)
Poland
.05
.1
.15
.2
2000 2040 2080
year
equal access (dash)
Romania
.05
.1
.15
.2
2000 2040 2080
year
equal access (dash)
Slovakia
Consumption poverty
.1
.2
.3
.4
2000 2040 2080
year
equal access (dash)
Bulgaria
.1
.2
.3
.4
2000 2040 2080
year
equal access (dash)
Czech
.1
.2
.3
.4
2000 2040 2080
year
equal access (dash)
Hungary
.1
.2
.3
.4
2000 2040 2080
year
equal access (dash)
Poland
.1
.2
.3
.4
2000 2040 2080
year
equal access (dash)
Romania
.1
.2
.3
.4
2000 2040 2080
year
equal access (dash)
Slovakia
Consumption old−age poverty

Are there effective ways to nudge people into saving more?
Our scenarios show the outcomes equivalent to effectively equalizing
ability to save for old age consumption. Are there effective ways to
do that? Evidence from the Save More Tomorrow (SMT) experiment
are positive. Benartzi and Thaler (2007) formed savings plans based
on better understanding of decision-making process. The plan is as
simple as it gets and assumes commitment to increasing the savings
in the future (hyperbolic discounting) and at the moment of the pay
raise (loss aversion). The program is administered by employer (self-
control) and individuals are automatically enrolled with the opt-out
option (procrastination and status quo bias). The effects of SMT are
spectacular: 3 years after introducing the plan, the savings rate of the
participants was 370% higher than in other plans.
Another way to address the behavioral challenges are the Prize Linked
Savings (PLS), that is savings accounts which on top of “boring”
interest offer also a lottery-like feature. In each period, one participant
gets to accrue a prize equivalent in value to the premium accrual from
all the participating accounts. Linking the boring (i.e. old age saving)
with the pleasurable and exciting (i.e. lottery with a high prize) helps
to overcome the procrastination and status quo bias, whereas the value
of the prize helps to sustain commitment (Kearney et al. 2010). Well-
established PLS include government-run Premium Bonds in the United
Kingdom and privately-run Million a Month Account (MaMa) in South
Africa. Holders of PLS increase total savings rate by 1% of disposable
income, a 38% growth at the mean (Cole et al. 2017).
Also financial engineering comes in handy. Multiplicity of options dis-
courages people from taking up savings decisions (status quo bias)
whereas the prospect of readjusting their decisions delays taking up
any action (procrastination). Targeted maturity funds may match ma-
turity of the financial portfolio with the expected timing of retirement,
thus embodying the long term commitment (facilitating self-control).
They also replace the prospect of repeated decision-making with a
prospect of high payout at retirement (Sethi et al. 2004).
The advantage of both SMT and PLS is that they are virtually costless
to introduce and administer in a sense that even if some administrative
costs are higher, they do not entail state subsidies to encourage partic-
ipation. Naturally, financial incentives prove useful as well, especially
when combined with the ability to nudge via modern technologies.
In a voluntary savings program in Kenya, participants received reg-
ularly SMS from the government informing them that any deposit
made on that given day will be matched with an additional deposit
by the government to a private account of each participant. In a
control group the same subsidy was offered, but information was dis-
tributed through traditional communication channels. Regular texting
increased propensity to deposit as well the amount of the deposit even
though the subsidy did not depend on the size of the deposit.
Indeed, it appears that many of the savings decisions are actually fast
in the language of Daniel Kahneman (2011). In a field study, he finds
that 75% of people does not readjust portfolio structure for as long as
10 years even if the original decision was instinctive and unreflective. In
fact, financial investment decisions are rarely slow and deliberate, i.e.
when taking up financial obligations people are much more reflective
than when deciding about the allocation of their assets.
This brings about another potential for technology use: given the
ability of social media to design psychological profiles of the users,
they are likely to offer better matched portfolio choices. Subject to
diligent regulation and supervision, offering portfolios based on profiles
could lead to a substantial improvement in assets allocation at virtually
no cost, simply by improving the matching in consumer finance.
These and other methods may be used to equalize access to financial
instruments across social groups, types of households and for individu-
als at all income levels. In order to achieve reduction in old age poverty,
the instruments need to encourage saving, align actual savings path
with an optimal one and facilitate the execution of the savings plan.

□ Are there private financial instruments to contain the old age poverty?

Fostering equal access to saving instruments reduces wealth inequality
.3
.4
.5
.6
.7
Gini
2000 2040 2080
year
equal access (dash)
Bulgaria
.3
.4
.5
.6
.7
2000 2040 2080
year
equal access (dash)
Czech
.3
.4
.5
.6
.7
2000 2040 2080
year
equal access (dash)
Hungary
.3
.4
.5
.6
.7
2000 2040 2080
year
equal access (dash)
Poland
.3
.4
.5
.6
.7
2000 2040 2080
year
equal access (dash)
Romania
.3
.4
.5
.6
.7
2000 2040 2080
year
equal access (dash)
Slovakia
Wealth inequality
.2
.3
.4
.5
Gini
2000 2040 2080
year
equal access (dash)
Bulgaria
.2
.3
.4
.5
2000 2040 2080
year
equal access (dash)
Czech
.2
.3
.4
.5
2000 2040 2080
year
equal access (dash)
Hungary
.2
.3
.4
.5
2000 2040 2080
year
equal access (dash)
Poland
.2
.3
.4
.5
2000 2040 2080
year
equal access (dash)
Romania
.2
.3
.4
.5
2000 2040 2080
year
equal access (dash)
Slovakia
Income inequality
In addition to a lower risk of the old age poverty, equal access to
financial instruments reduces also substantially wealth inequality. By
removing barriers to saving – as well as the privileges – we let some
of the households to change they behavior pattern. In fact, instead
of consuming their entire income contemporaneously or even being
in-debt, some of the households in our simulation scenarios gain both
access and taste for saving. Meanwhile, those households who saved
a high fraction of income, when no longer in a privileged position
of unique access to saving instruments, somewhat increase contem-
poraneous consumption at the expense of wealth accumulation. As
a consequence, the saving behavior becomes more similar between
the household groups, hence wealth distributions become more com-
pressed.
Substantially lower levels of wealth inequality in our setup are consis-
tent with prior economic research. First, using the case of Sweden,
Domei et al. (2002) and Hurd et al. (2012) show that generosity of
the pension system virtually eliminates the need to use voluntary sav-
ings as a mechanism to smooth consumption over lifetime. Notably,
in countries with generous pension systems, long-term savings is very
low even among households with high earnings in both relative and
absolute terms. Analogously, low pensions – as in our setup – should
foster the need for private voluntary savings, as shown in our model.
The need to save is then mediated by the ability to do so: level of
income, ability to adjust labor supply and access to financial instru-
ments. In our setup, once frictions in access to financial instruments
are removed, households adjust both labor supply and saving behavior,
consistent with their preferences. Reduced wealth inequality implies
that these adjustment result in assets more evenly spread across social
groups and age groups.
Definitions in figure: Wealth inequality measured by Gini ratio. The initial values
match consumption inequality as observed in these countries prior to the
introduction of the pension system reform from a DB to a DC system. The
computed inequality measure includes weights for population structures.

Discussion of the results
Access to old age saving instrument is a concern in many advanced
countries. Even in those countries, where a large fraction of the pop-
ulation has a banking account and uses instruments such as credit
cards and deposits, old age savings instrument may be inaccessible to
the population at large. For example, a deposit in a bank account
usually does not entail incentives for old age savings. Moreover, the
interest rate to be earned on a deposit is substantially below the econ-
omy interest rate, yielding low return to savings and thus reducing
savings. Investing in financial markets in order to construct a portfolio
requires skills and effort inaccessible to population at large. Hedge
funds, investment funds and ETFs, which could potentially substitute
for individual financial portfolios have none of the incentives character-
izing SaveMoreTomorrow or prize-linked savings. Instead they charge
fees, reducing the effective rate of return on savings and thus disin-
centivizing old age saving.
Given these constraints, it is realistic to assume that a fraction of indi-
viduals effectively does not have access to old age savings instruments
and only those who pay particular attention to the future actually ac-
cumulate wealth to smooth consumption. Our simulations show (a)
substantial rise in old age poverty if we continue with this inequality in
access to financial instruments and (b) large potential for improvement
in these frictions are removed and access to old age saving becomes
universal. Our results should be viewed as a range for potential out-
comes if at least a part of currently underprivileged individuals gained
access to old age saving.
Additional gains could be achieved if the saving instruments offered
an annuity. Indeed, a perfect instrument offers convenient saving for
old age consumption and an annuity on these savings, i.e. insurance
against outliving one’s own savings.
Indeed, dissaving decision is no easier than saving decision. People
need to take into account a multiplicity of factors: how fast they want
to draw assets, how much they want to leave as bequests, probability
of negative health shocks, life expectancy, etc. Given how difficult
it is to optimize individually, given the unpredictable nature of all
these factors, ability to acquire insurance against making a mistake
in judgment is of immediate value (Davidoff et al. 2005). Annuity
hedges us against the longevity risk, provides a survivor premium and
raises life time consumption for all.
The problem of old age saving and dissaving in the old age are inti-
mately related. Without accumulated wealth, individuals cannot pur-
chase an annuity. Even with positive but low wealth, and with the
expectation of pension benefits, many individuals may prefer to hold
a liquid asset (e.g. in case of family emergency) than convert the
accumulated savings to annuity.
Despite the benefits of annuity, evidence shows that even affluent in-
dividuals tend to acquire too little of it (Schaus 2005), with reasons
similar to those which underly insufficient old age savings. It is not
about preferences or willingness to take the risk of outliving one’s own
saving, quite the contrary. As discussed by Benartzi et al. (2011), par-
ticularly with a defined contribution pension system the actual choices
depend on institutional factors. Framing effect, poor financial literacy
and loss aversion greatly contribute to disregarding the annuities. For
individuals, transforming pension wealth into a stream of payments
until the end of lifetime is viewed as a loss: large sum of money is
replaced by a small sum of money (even if spread over many periods).
Moreover, annuity reduces risk of outliving one’s own savings, but in-
dividuals tend to worry about dying sooner: from a perspective of an
average retiree, dying sooner than expected is not only a bad luck, but
also a bad investment.

□ Summary

Summary
Assuring universal access to savings instruments will not only reduce
poverty and old age poverty, but will also contribute to less unequal
distribution of assets across social groups and generations.
Our objective in this study was to analyze the effects of longevity
on poverty and inequality in six countries Bulgaria, Czech Republic,
Hungary, Poland, Romania and Slovakia. These six countries changed
a defined benefit pension system characteristic for most welfare states
in the post-war era to a defined contribution pension system. Defined
contribution pension system is believed to be more efficient, because
it improves the incentives in the economy: benefits will be equivalent
to what we contribute to the pension system, hence social security
contributions can be viewed as delayed income rather than a tax.
An additional feature of the defined contribution pension systems is
that in response to longevity, the actual per period pension benefits
will be low. The longer the life expectancy at retirement, the larger the
number of periods which need to be financed from the accumulated
pension contributions. Although this effect can be partly alleviated by
extending the working career beyond the current retirement age, pen-
sion benefits from a defined contribution system will be substantially
lower than observed currently from a defined benefit system.
□ What is the scope of old age poverty rise due to the
increased longevity?
Despite the intensifying need to co-finance old age consumption from
private savings, one does not observe increased savings in the six
economies which implemented the reform. We replicate the distribu-
tion of assets across the types of the households from the pre-reform
period and show that the poverty rates will double as the generation of
Millennials retires. In fact, old age poverty will be behind this increase
in poverty in all the six countries.
With the incentives inherent in the defined contribution pension sys-
tem, income and consumption inequality will grow as well. The growth
in inequality is larger than these six countries experienced when tran-
sitioning from centrally planned to market economies.
Comparing the model in which many individuals cannot access ad-
equate savings instruments with a model where access to financial
instruments becomes universal, we show that both poverty and in-
equality can be substantially reduced. We show that this result is
universal, i.e. holds for all six countries in the sample, regardless of
their demographic structure and macroeconomic features. Our key re-
sult does not depend on the behavioral reaction to longevity: among
the six analyzed countries longevity raises wealth inequality in Poland,
leaves wealth inequality essentially unaffected in Bulgaria and reduces
wealth inequality in four remaining countries. These differentiated
patterns owe to the differences in deep economic preferences embed-
ded in these countries, such as preference for leisure and aggregate
discounting of the future. Universal access to savings instruments re-
duces wealth inequality to a comparable extent across all these six
countries despite their inherent differences.
The mechanisms behind these processes are both economic and be-
havioral. Longevity implies lower pensions in the defined contribution
system, hence yielding pure economic incentives to raise private volun-
tary savings. In the case of some social groups the lifetime savings will
effectively double. However, behavioral mechanisms at play imply that
many individuals will not make an accurate judgment by how much
should they increase the savings. Moreover, many individuals will be
unable to execute this plan due to a variety of mechanisms such as
status quo bias, procrastination, loss aversion and self-control issues.

□ Methodology of the overlapping generations model behind this study

Mechanics of our model
We develop an overlapping generations general equilibrium model, in line with Makarski
et al. (2017). The model uses demographic forecasts for evolution of longevity and
fertility for each of the analyzed six countries. The model follows also European Com-
mission in the country-specific assumptions about the future evolution of the rate of
technological progress.
Households in our model choose how much time to work (the so-called intra-temporal
choice) and how much of the current income to save (the so-called inter-temporal
choice). These choices are made in order to maximize lifetime well-being, which in-
creases with consumption, but declines with labor. Consumers in our model generally
like their life-time path of consumption to be smooth: high consumption in working
period and the extremely low consumption in the old age makes them less happy than a
comparable path of consumption which is spread more equally throughout the life time.
Households in our model are heterogeneous in a sense that households may differ in
their preference for leisure. Hence, in some cases the labor supply will be such that we
assume effectively someone worked full time their whole professional life, while in other
cases the actual share of life worked will be lower. The actual amount of time work
and the share of population with such preference were calibrated closely to replicate the
distribution of working time in the labor market as observed in the Labor Force Survey
data in each of the six analyzed countries.
Households differ also in how productive an hour of their work is, which reflects the
differences in wages as observed in the labor market. We use linked employer-employee
data from the EU Structure of Earnings Survey for each of the analyzed six countries
to replicate the distribution of wages as observed in reality.
Our replications are successful in a sense that for every country our model reproduces
consumption inequality as observed in the data. Namely, our calibration permits to
capture the fact that some households are more affluent than others, in ways that
are similar to actual data. Note that in the data one has both prime age and old age
households, which means that the population structure in the survey data that underlies
the real world estimates of the consumption inequality is similar to what is produced
in our model. We use report by World Bank (2005) as source for information about
consumption inequality.
Finally, households differ in access to financial instruments. In our model this is equiv-
alent to households having different time preference in a sense that some households
actually save a higher share of their income than others. We calibrate this heterogeneity
in such a way that we replicate wealth inequality data as observed in the six analyzed
countries prior to the reform, see Table 1. We follow Davies et al. (2011) as source for
information about wealth inequality.
The key feature of our analysis consists of equalizing the preferences for saving across
the households. In the baseline scenario we keep the heterogeneity in saving behavior
as calibrated to the data. In the reform scenario we check what would happen if the
ability and preference to save became the same among households. The households
continue to differ in productivity and preference for leisure, but faced with the same
inter-temporal options, all households of a given birth cohort will choose to save the
same proportion of their income. Note that because of longevity and general economic
changes, households of different cohorts may still choose different savings as optimal,
because their life expectancy and economic conditions will differ.
The model begins with an economy which has a defined benefit pay-as-you-go system,
as all the six analyzed countries had a system like that. We then implement a change
to a defined contribution pay-as-you-go system, replicating the features of the reforms
in all the six countries. All of these countries honored the pension obligations towards
individuals already in retirement and those who were close to retirement, so also in our
model the implementation of the reform is gradual.
When they reformed their pension systems, the six countries introduced also a capital
pillar to the pension system, but within roughly a decade some of the six analyzed
countries dismantled the capital pillars, while others reduced its size. For simplicity and
due to the low relevance of compulsory capital pillar in the general pension system for
the question at hand, we do not replicate this feature in our model. Namely, the new
pension system is a defined contribution pay-as-you-go system.
Table 1. Wealth inequality (Gini ratio) - data vs model
Country Data Model
Bulgaria 0.65 0.65
Czech 0.63 0.62
Hungary 0.65 0.65
Poland 0.66 0.65
Romania 0.65 0.66
Slovakia 0.63 0.64

Mechanics of our model - continued
The model is calibrated to replicate the macroeconomic features of the six analyzed
economies prior to the introduction of the pension system reform. Using data from
national accounts we know the investment rate in these economies, which helps us to
calibrate the depreciation rate. Using data from the labor force survey, we know the
aggregate labor force participation, which helps us calibrate the aggregate preference
for leisure. We also use the taxation data from the OECD to adequately calibrate the
tax rates for capital, labor and consumption in all six countries to match in the model
the share of those tax revenues in GDP to what was actually observed.
Since these countries were undergoing economic transition, the economic situation was
quite volatile year on year. To avoid the situation that a specific year affects the original
calibrations, we compute 10 year averages for the whole decade prior to the pension
system reform in each country. The interest rate was set at 6.5 %, if capable by others
parameters.
Our model is a general equilibrium model, which means that all of the macroeconomic
variables adjust to changing demographics and behavior of households. For example, as
longevity increases, individual households will decide to save a higher fraction of their
income during the prime age, which will result in change of the capital stock in the
economy, thus affecting labor productivity and interest rates. Indeed, these variables
adjust endogenously in the economy, in response to a new pension system, secular
changes and behavior of the agents.
Also fiscal variables adjust endogenously. We calibrated the taxation in the period
prior to the reform, but once the reform from defined benefit to defined contribution is
implemented in our simulations, the consumption tax adjusts endogenously to satisfy
the government budget constraint. In order to mitigate the potential effects of change
in fiscal policy on the analyzed processes, we assume that the governments continue
with the debt-to-GDP ratios as in the periods prior to the reform, which in most of the
six analyzed countries was close to roughly 45%.
The government needs to adjust taxes for two main reasons. First, it balances the
pension system. We calibrate the pensions from the defined benefit pension system
and the replacement rate to match the share of pensions in GDP as observed in the
six analyzed countries. We calibrate the contribution rate to the compulsory pension
system to match the deficit observed in the six analyzed countries prior to the reform.
Second, the government balances the government expenditure with the revenues such
that the public debt in relation to GDP does not change. The government expenditure
was set to match the data from national accounts on government expenditure share in
GDP. Throughout the simulations this share does not change, because the interest of
the analysis is to isolate the effect of longevity on inequality rather than analyze the
potential scenarios of government expenditure.
Table 2. Calibration of parameters
Calibration Target
Bulgaria Czech Hungary Poland Romania Slovakia
Data Model Data Model Data Model Data Model Data Model Data Model
depreciation rate investment rate 23.9% 23.8% 30.4% 30.7% 24% 23.7% 20.5% 20.5% 20.9% 20.9% 29.8% 30.3%
labor tax revenue as % of GDP 7.6% 7% 9.4% 9.4% 15.3% 15.4 5.5% 5.5% 6.5% 6.5% 7.5% 7.7%
consumption tax revenue as % of GDP 12.5% 12.4% 10% 10.5% 15.3% 15.4% 12.2% 12.2% 10.6% 10.5% 11.6% 11.4%
social security contribution benefits as % of GDP 5.4% 5.4% 5.7% 5.6% 5.9% 5.9% 6.9% 6.9% 6.8% 6.7% 5.3% 5.3%
preference for leisure average hours 56.7% 58.2% 52.6% 52.6% 57.6% 58% 56.8% 56.6% 58.8% 58.1% 53.5% 53.4%
reforming from DB to DC 2002 1996 1998 1999 2004 2003
effective retirement age (OECD) 60 61 59 61 66 58

□ References

References
1. Benartzi, S., Previtero A. and Thaler R.: 2011, Annuitization puzzles. Journal of Economic Perspectives 25(4), 143-164.
2. Benartzi, S. and Thaler R.: 2004, Save more tomorrow™: Using behavioral economics to increase employee saving, Journal of Political Economy 112(1), 164-187.
3. Benartzi, S. and Thaler R.: 2007, Heuristics and biases in retirement savings behavior, Journal of Economic perspectives 21(3), 81–104.
4. Brown, J., Kling, J., Mullainathan, S. and Wrobel, M.: 2008, Why don’t people insure late-life consumption? A framing explanation of the under-annuitization
puzzle, American Economic Review 98(2), 304–309.
5. Choi, J., Laibson, D., Madrian, B. and Metrick, A.: 2004, For better or for worse: Default effects and 401 (k) savings behavior, in: Perspectives on the Economics
of Aging, 81–126, University of Chicago Press.
6. Choi, J., Haisley, E., Kurkoski, J. and Massey, C.: 2017, Small cues change savings choices, Journal of Economic Behavior & Organization 142, 378–395.
7. Cole, S., Iverson, B. and Tufano P.: 2017, Can gambling increase savings? Empirical evidence on prize-linked savings, Saïd Business School Working Papers.
8. Davidoff, T., Brown, J. and Diamond P.: 2005, Annuities and individual welfare, American Economic Review 95(5), 1573-1590.
9. Davies, J., Sandstrom, S., Shorrocks, A. and Wolff, E.: 2011, The level and distribution of global household wealth, The Economic Journal 121(221), 223-254.
10. Domeij, D. and Klein, P.: 2002, Public pensions: To what extent do they account for Swedish wealth inequality?, Review of Economic Dynamics 5(3), 503-534.
11. Gustman, A. and Steinmeier, T.: 1998, Effects of Pensions on Saving: Analysis with Data from the Health and Retirement Study, NBER Working Paper No. 6681.
12. Hemmelgarn, B., Manns, B., Zhang, J., Tonelli, M., Klarenbach, S., Walsh, M., Culleton, B., Alberta Kidney Disease Network and others: 2007, Association
between multidisciplinary care and survival for elderly patients with chronic kidney disease, Journal of the American Society of Nephrology 18(3), 993-999.
13. Hurd, M., Michaud, P. and Rohwedder, S.: 2012, The displacement effect of public pensions on the accumulation of financial assets, Fiscal Studies 33(1), 107-128.
14. Istenic, T., Hammer, B., Seme, A., Dolinar, A. and Sambt, J.: 2017, European National Transfer Accounts, Deliverable 1.4: The European NTA Manual.
15. Kahneman, D. and Tversky, A.: 1992, Advances in prospect theory: Cumulative representation of uncertainty, Journal of Risk and Uncertainty 5(4), 297-323.
16. Kahneman, D,: 2011, Thinking, fast and slow, Macmillan.
17. Kearney, M., Tufano, P., Guryan, J. and Hurst, E.: 2010, Making savers winners: An overview of prize-linked savings products, NBER Working Paper No. 16433.
18. Lilley, E., Khan, K., Johnston, F., Berlin, A., Bader, A., Mosenthal, A. and Cooper, Z.: 2016, Palliative care interventions for surgical patients: a systematic review,
JAMA surgery 151(2), 172-183.
19. Makarski, K., Tyrowicz, J., Bielecki, M. and others: 2017, Inequality in an OLG economy with heterogeneous cohorts and pension systems, Society for Economic
Dynamics, 2017 Meeting Papers, 958.
20. Marshall, S., McGarry, K. and Skinner J.: 2011, The risk of out-of-pocket health care expenditure at the end of life, in: Explorations in the Economics of Aging,
University of Chicago Press.
21. Mohile, S., Velarde, C., Hurria, A., Magnuson, A., Lowenstein, L., Pandya, C., O’Donovan, A., Gorawara-Bhat, R. and Dale, W.: 2015, Geriatric assessment-guided
care processes for older adults: a Delphi consensus of geriatric oncology experts, Journal of the National Comprehensive Cancer Network 13(9), 1120-1130.
22. Samuelson, W. and Zeckhauser, R.: 1988, Status quo bias in decision making, Journal of Risk and Uncertainty 1(1), 7-59.
23. Schaus, S.: 2005, Annuities make a comeback, Journal of Pension Benefits: Issues in Administration 12(4), 34-38.
24. Sethi-Iyengar, S., Huberman, G. and Jiang, W.: 2004, How much choice is too much? Contributions to 401 (k) retirement plans, in: Pension design and structure:
New lessons from behavioral finance 83, 84-87, Oxford University Press.
25. The World Bank: 2006, World Development Report, Equity and development. Oxford University Press.

About us
Group for Research in Applied Economics (GRAPE) is an independent,
non-governmental research center operated by Foundation of Admirers
and Mavens of Economics (FAME), based in Warsaw, Poland. We house
a cheerful bunch of accomplished academics and brilliant adepts and focus
on producing high quality academic research. Since foundation in 2011,
FAME|GRAPE develops research programs in economics of inequality,
pension systems, efficiency and misallocation, applied macroeconomics
and cultural economics. Our key goal is to to facilitate academic ex-
change in economics in Poland and co-develop scientific projects interna-
tionally. In the interest of academic integrity and transparency, we strive
to openly share data and methodology.
In addition to scientific research, people at GRAPE translate their aca-
demic experience to addressing social policy issues in Poland. Our pro
bono policy projects are always based on rigorous research and draw on
earlier experience from specific research projects. So far, we’ve published
several policy reports on the subject of Polish pension system, labour
market and school feeding programs. Mainstreaming of research results
and promoting evidence-based-policy stands behind our presence in me-
dia: we write a weekly op-ed for one of the biggest newspaper in Poland
and participate in many policy-oriented events. In policy projects, we
cooperate with The World Bank, OECD, EY and Polish NGOs.
web grape.org.pl tt GRAPE_ORG
mail grape@grape.org.pl fb GRAPE.ORG
Magda Malec
Warsaw School of Economics
Research Assistant at FAME|GRAPE
Krzysztof Makarski, PhD
Warsaw School of Economics | NBP
Co-founder of FAME|GRAPE
Joanna Tyrowicz, PhD
University of Warsaw | IAAEU | IZA
President & Co-founder of FAME|GRAPE

Happily ever after... or until we run out of money. Effects of longevity on financial vulnerability

More Related Content

What's hot

Similar to Happily ever after... or until we run out of money. Effects of longevity on financial vulnerability

More from GRAPE

Recently uploaded

Happily ever after... or until we run out of money. Effects of longevity on financial vulnerability