The
Third Workshop of the RTN project on The Economics of Ageing in
Europe has been organized by
IFS
and has been held in London on October 2-4, 2003.
The
issues raised by ageing populations, and in particular the topics of
retirement, saving and pension provision, are some of the most important
policy issues currently facing the government. Increasingly, economists are
developing new techniques and collecting new data designed to understand such
issues. This conference brought together junior and senior economic
researchers from across Europe to present and discuss theoretical and applied
work on these topics. Issues in ageing were defined broadly, and included
pensions, retirement, and health, as well as papers related to the life-cycle
model of how individuals allocate resources across their life-time.
Thursday
2nd October: Afternoon session
The
conference started with the presentation Modelling
early retirement and labour supply, by Richard Blundell, which was
intended as a review of the main issues concerning the labour supply of older
individuals. Motivated by, and drawing on, the results of the NBER project on
Social Security and Retirement Around the World Richard presented an
international comparison of recent changes in life-expectancy, trends in
disability-free lifetimes, degree of labour market attachment and alternative
paths out of the labour force for those aged 50 and over. Next he reviewed the
importance of different retirement incentives in welfare systems, including
those from disability schemes, unemployment insurance, state provided social
security pensions, occupational pensions and individual pension accounts. The
final part of the talk dealt with the spikes in retirement hazards, the
prevalence of joint retirement within couples, the importance of liquidity
towards early and normal retirement and the modelling of labour market
opportunities for older individuals. A systematic way to explain or introduce
these issues in standard retirement models has yet to be found. The existing
approaches to dealing with them found in the literature were reviewed, and
suggestions were given for future work.
A
break for coffee was followed by the paper The
relationship between health and socio-economic status, by James P. Smith.
He highlighted the difficulty in determining the direction of causality
between health and socio-economic status (SES). One way of doing this has been
looking at samples of immigrants, whose SES relative to the rest of the
population varies greatly when migration takes place. Using a health
trajectory model James showed that the degree of health selectivity among
immigrants is very large and that widely held beliefs about the effects of the
receiving country on immigrants should be revisited. Until this is done, using
migration samples to solve the problems facing the relationship between SES
and health would be potentially misleading. He then considered the degree of
direct (SES on health) and reverse (health on SES) causation present in
different data sources, and found that they are very different when using data
from the RHS or the AHEAD samples. This calls for future research in certain
aspects of the relationship between SES and health, specifically on the roles
of income and education determining individual health.
The
last presentation of the day was Presenting
to policy makers, by Andrew Dilnot, who talked about how the economic
message from a technical paper should be presented to policy makers, who
generally do not have a specific training on the subject. In particular he
discussed the importance of trying to get a message across by keeping
information as simple as possible combined with the often conflicting, but
crucial, issue of ensuring that the context of findings were understood so
that they were not either deliberately or inadvertently misused.
Friday
3rd October: Morning session
The
morning session contained three papers looking at consumption behaviour. The
first paper was Is there a retirement consumption puzzle in Italy by
Raffaele Miniaci, Chiari Monfardini and Guglielmo Weber, with the presentation
being given by Guglielmo. This looked at changes in consumption around
retirement using Italian micro-data from 1985 to 1996. They find a one-off
drop in consumption at retirement – a phenomena that has also been found in
both US and UK studies. Within this they find that consumption of work-related
goods falls and home production of food increases. However once they control
for changes in demographics, leisure, wealth and income they find evidence
only for a gradual decline, rather than a one-off drop, in consumption. A
discussion of the paper was then given by Sarah Smith who pointed out, amongst
other things, that while Italians might have been able to predict their
initial retirement income with a degree of accuracy the finding of a gradual
decline in consumption could be consistent with the idea that households did
not anticipate the impact of price rather than earnings indexation of their
pension income throughout their retirement.
The
second paper was Revisiting the consumption boom by Orazio Attanasio,
Laura Blow, Rob Hamilton and Andrew Leicester with the presentation being
given by Orazio. This looked at the relationship between changes in house
prices and consumption using UK data from 1975 to 2001. The preliminary
findings of this work is that changes in house prices are found to impact on
consumption for owner-occupiers, with no effect found for renters. However the
large consumption boom in the late 1980s does not seem to be explained by
increases in house prices. Further work will examine whether the consumption
boom could alternatively be explained by changes in equity prices. A
discussion of the paper was then given by Martin Browning. In particular he
pointed out that using cross sectional surveys for this type of analysis was
difficult and that perhaps one of the findings of this and related work is the
potential benefits from information on the consumption of the same individuals
over time.
After
a break for coffee the third paper was Demand changes in an ageing economy
by Melanie Lührmann. This study uses German micro data on consumption from
1978 to 1998 to examine differences in household consumption shares by age.
These results are then applied to demographic projections from 1995 to 2050 to
estimate what the impact of an ageing population would be on the composition
of consumption under various different assumptions. The model estimated
predicts that the share of expenditure allocated to housing, health, education
and leisure goods will increase while the share allocated to food and clothing
is expected to fall. A discussion of the paper was then given by James Banks
who pointed out, amongst other things, that the impact of an ageing population
on consumption and therefore on the demand for both younger and older workers
was a key area for future research.
Friday
3rd October: Afternoon session
The
three papers presented on the Friday afternoon shared an emphasis on issues
that affect the portfolio of assets that people choose to hold. Antoine
Bommier (Université de Toulouse) opened the discussion by presenting his
paper on Why older people seem to be less risk averse, that
is co-authored with Jean-Charles Rochet. The paper is motivated by the
puzzling fact that, contrary to the received wisdom from financial advice and
economic theory, elderly people seem hold more risky portfolios than their
younger counterparts. The authors argue that this puzzle could be resolved by
relaxing the standard theoretical assumption that the amount that an
individual consumes in any particular period does not affect the pleasure that
they derive from consuming at other points in time. In particular, they show
that if people view consumption at any point in time as a substitute for (rather
than a complement to) consumption at other dates, then they will tend to
become more tolerant of risks as they get older. The discussion of this paper
by Jerome Adda (UCL and IFS) focussed on the issue of whether or not older
people actually do hold more risky portfolios, and was illustrated with U.K.
data.
The
second paper was Testing Portfolio Choice with a Correlated Background Risk,
written by Luc Arrondel, Hector Calvo-Pardo and Xisco Olivier (DELTA –
Paris), and presented by the latter of these three. This paper identified
another apparent dichotomy between economic theory and evidence – namely
that evidence does not unambiguously verify the prediction that households
with risky earnings should limit their overall exposure to risks through low
participation in financial assets. These authors seek an empirical resolution
to their puzzle. They look at whether stronger empirical results can be
obtained if the analysis controls for whether or not agents perceive that
risks in asset income and earnings are offsetting so that increasing both
types of risk could reduce the overall riskiness of an income stream. The
authors find some evidence that those who think that such offsetting is
possible do take on more financial risks, but little evidence that those who
think that the two risks are reinforcing opt not to hold risky assets when
their labour income is risky. Discussant Carl Emmerson (IFS) suggested that
these weak results could be due to the fact that the survey questions that are
exploited focus attention on hypothetical or aggregate risks, rather than
asking about the specific circumstances that are relevant to each
individual’s portfolio decision.
After
the coffee break Michael Haliassos and Yannis Bilias shared out the
presentational duties for their paper on The Distribution of Gains from
Access to Stocks. This ambitious paper combines a theoretical analysis of
the possible determinants of portfolio choice with an empirical analysis that
identifies the mechanisms that have actually been important in shaping
people’s portfolios. The authors use empirical techniques that are designed
to identify whether large and small investors have been affected differently
by factors influencing portfolio decisions. Their conclusions include that, in
the US in the late 1990s, changes in stock market participation costs and the
behaviour of stock returns have mainly affected the participation decisions of
small and potential investors, rather than the behaviour of established
stockholders. Tullio Jappelli (University of Salerno) led the discussion and
pointed out that the presence of income risk in the theoretical model leads to
several counter-intuitive results. He also suggested to simulate theoretical
models with participation costs, rather than concentrating on the two polar
cases of absence of transaction costs or transaction costs that are
effectively infinite.
Saturday
4th October: Morning session
Saturday’s
session focussed on retirement. The first paper was Retirement and private
savings by Grzegorz Kula and Coen N. Teulings, with the presentation given
by Grzegorz (Tinbergen Institute). This paper examines, from a theoretical
perspective, retirement behaviour when there is no retirement insurance,
incomes are risky and people finance their retirement through private savings.
It develops a dynamic programming model to find the threshold ratio of
expected incomes to accumulated savings at which it is optimal to retire, and
to relate this threshold to the process driving income in order to predict the
probability of retirement as a function of time. The model is solved using
numerical simulations, and these are carried out for people with different
attitudes towards risk and different tastes for leisure. Some of the main
findings are that the more people dislike working, the earlier they retire,
and that the higher their risk aversion, the later they retire. A discussion
of the paper was given by Agar Brugiavini. She pointed out that the assumption
of irreversibility of retirement is perhaps too strong, particularly in a
context where there is no retirement insurance, and that it is unclear why
borrowing is not allowed in the model.
The
second paper was Retirement expectations and pension reforms by Tullio
Jappelli, Mario Padula and Renata Bottazzi, with the presentation given by
Mario (University of Salerno). It uses Italian micro-data from 1989 to 2000 to
look at whether individuals’ expectations of retirement outcomes (retirement
age and replacement rate) and wealth accumulation decisions are updated in the
direction suggested by pension reforms that take place during the observation
period. The authors find a substantial, although partial, offset between
private wealth and perceived pension wealth. They also find that people do not
fully internalise the implications of the reforms into their expectations of
social security pensions or their retirement plans. These two findings
together imply that the pension reforms of the 1990’s did not have a large
impact on the household private wealth. A discussion of the paper was given by
Elsa Fornero (University of Turin and CeRP). She pointed out that the
hypothesis of continuous working careers could lead to overestimates of
pension wealth, particularly for women, and that a potential selection issue
could emerge if those that expected the reform decided to retire before it
became effective.
