Leaving home and poverty among youth : A cross-European analysis

1
Leaving home and poverty among youth A
cross-European analysis

• Arnstein Aassve,
• Maria Iacovou
• Stefano Mazzuco
• Maria A. Davia

2
Outline

• Motivation
• What is youth?
• How do we define poverty?
• Literature what do we know?
• Data ECHP
• Descriptive analysis
• Welfare regimes
• Empirical approach PSM
• Results
• Conclusions

3
Motivation

• Leaving the parental home, are all crucial life
  events contributing to the vulnerability of youth
  .
• Large discrepancies across European countries in
  terms of the extent to which leaving home is
  associated with poverty
• We use data from the ECHP and propensity score
  matching
• and find that the event of leaving home does have
  an impact on entering poverty.
• The strongest effect for Scandinavian countries,
  weakest among Mediterranean countries.

4
What is youth?

• UN and EU 15-24 yrs
• ESRC 15-25, JRF 16-25
• Here we conceptualise youth as a process of
  transition to adulthood
• in many countries these transitions have hardly
  begun by mid-20s.
• Italy median age at leaving home for men is 30.
• Netherlands mean age at first birth for women is
  29.
• Define three subgroups 16-19, 20-24, 25-29.

5
How do we define poverty?
How we use the ECHP to compute household income
and poverty rates To compute household
equivalent income in year t, we use income data
pertaining to year t collected at year t 1,
summing this over all the individuals present in
the household at year t and using an equivalence
scale based on the numbers and ages of
individuals present at year t (Heuberger, 2003).
How we define poverty poor households net
equivalised household income income is below 60
of the median in the country
6
Literature what do we know?

• Eurostat (2002) the incomes of young people
  below age 24 are below national averages.
• Cantó-Sánchez and Mercader-Prats (1999) youth
  poverty due to temporary jobs.
• Pavis, Platt and Hubbard (2000) key role of
  education
• Smeeding et al (1999) and Berthoud and Robson
  (2003) single parenthood in Anglo-Saxon nations
• Magadi et al (2005) timing of first birth
• Aassve et al (2005) Employment, marriage, and
  cohabitation is associated VERSUS independent
  living having children and being without work are
  all associated with higher poverty risk.

7
The data ECHP

• Waves 1994 2001
• UK, Ireland, Denmark, Finland, Netherlands,
  Belgium, France, Germany, Austria, Italy, Greece,
  Spain and Portugal
• Several problems with the ECHP
• But is reasonably good for comparative analysis.

8
What the data initially tell us poverty rates
are much higher away from parents
9
Welfare regime typology (Esping-Andersen)

• Social-democratic
• (Denmark, Finland, Netherlands)
• Liberal
• (UK and Ireland)
• Corporatist (Conservative)
• France, Germany, Austria, Belgium
• Southern (Residual)
• Portugal, Italy, Spain, Greece

10
Empirical approach PSM

• WHY PROPENSITY SCORE MATCHING?
• the leaving home event cannot be considered
  exogenous with respect to household income and
  therefore poverty
• Average treatment effect on the treated (ATT)
  For those leaving home The effect on entering
  poverty from leaving home.
• Average treatment effect on the control (ATC)
  For those staying at home The effect on entering
  poverty if they did leave home instead of staying
  behind

11
Empirical approach PSM (II)

• matching methods yield an unbiased estimate of
  the average impact of leaving home on treated
• Algorithm Nearest Neighbour Matching with a
  caliper of 0.01 ( radius for consistency checks)
• psmatch2 module in STATA (Leuven and Sianesi
  2003)

12
Average Treatment Effect on Treated is defined
as ATTE(Y1i Di1)-E(Y0i Di1) (2) In (2)
Y1 is the potential outcome (entering poverty in
our case) in the case the individual i receive
treatment (here leaving home) and Y0 is the
potential outcome in the case the individual i
does not receive the treatment (stay in the
parental home). Thus E(Y1i Di1) is observable
whereas E(Y0 Di1) is not, but an approximation
is created through the matching. A naïve
estimator of (2) would be ATTE(Y1i
Di1)-E(Y0i Di0) But this assumes no
selection bias. In order to remove the selection
bias we implement the matching procedure. This is
based on the critical assumption (CIA) that Y0
independent of D X States that treatment status
is random conditional on some set of X
13
If CIA holds then bias depends only on observed
variables. Under this assumption, EX (Y0i Xi,
Di0) EX (Y0i Xi, Di1), thus the ATT can be
unbiasedly estimated by ATT EX (Y1i - Y0i
Xi, Di1) EX (Y1i Xi, Di1) -EX (Y0i Xi,
Di0). Average Treatment Effect on Controls
(ATC) defined as ATCE(Y1i Di0)-E(Y0i Di0)
(3) In (3) we observe E(Y0i Di0) but not
E(Y1i Di0), which has to be approximated via
matching. Under the assumption of homogeneous
treatment effect ATT and ATC should give the same
results. But they are rarely homogenous. In our
setting, for instance, we can hypothesise that
young adults staying at home do so because they
are aware of being highly at risk of entering
poverty in the case they leave.
14
Composition of the sample
15
Selection equation probability of leaving home

• Year of the interview
• Age in the moment of the interview
• Gender
• Number of siblings
• Household income
• Personal income from work
• Whether the individual is a student
• Whether the individual is inactive
• Education attainment
• Working status of mother and father
• Education attainment of mother and father
• Parental house is short of space

16
Results from matching 20-24
In the majority of countries we find ATT being
lower than ATC. Poverty risk is an important
reason for delaying the transition out of the
parental home. Youths tend to delay leaving home
because they their chances of entering poverty
are higher if they leave. But for those who
leave, their risk is smaller than for those who
decide to stay behind in the parental home if
they had left.
17
Results from matching 20-24 (II)
For people under 25 there is a negative link
between the propensity to become poor if youth
leaves and the propensity to leave home
(Belgium), except for Denmark and Finland, where
those who leave have actually a lower probability
of becoming poor if they leave than those who
stay (if they left)
18
Results from matching 25-29
In the majority of countries we find ATT being
lower than ATC. Poverty risk is an important
reason for delaying the transition out of the
parental home. Youths tend to delay leaving home
because they their chances of entering poverty
are higher if they leave. But for those who
leave, their risk is smaller than for those who
decide to stay behind in the parental home if
they had left.
19
Results from matching 25-29 (II)
This negative link between poverty risk and the
probability of leaving home is no longer visible
for those over 25 years old
20
Results from matching 30-34
The economic incentives to stay at home or to
leave home are less and less pronounced with age,
to the point that in some southern countries,
women over 30 would face a lower risk of poverty
if they left their parents than if they stay. By
staying they may be showing a decision of
supporting their parents, and not of being
supported.
21
Results from matching 30-34 (II)
There is no link between differential risk of
poverty and exit from parental home amongst adults
22
Conclusions

• The higher risk of poverty amongst those who have
  left home in Scandinavian countries are
  corroborated in this dynamic perspective
• The elder the youth, the lower is the risk of
  poverty deterring the decision to leave home.
• Youths, regardless the poverty rates they face if
  they leave, tend to be rational in those
  countries where youths who do not leave home
  would experience a higher poverty risk if they
  left than those who actually leave, the exit
  rates are really low, whereas in the Scandinavian
  countries this difference is negligible.
• It is rational to stay, and it is rational to
  leave those who leave face a higher risk of
  poverty but, at the same time, are better
  sheltered from poverty than those who stay.