Marital Disruption and Economic Wellbeing: A Comparative Analysis

1

Marital Disruption and Economic Wellbeing:

A Comparative Analysis

Arnstein Aassve(ISER, University of Essex and CASE)

[email protected]

Gianni Betti(Department of Quantitative Methods, University of Siena)

[email protected]

Stefano Mazzuco(Department of Statistics, University of Padua)

[email protected]

Letizia Mencarini (Department of Statistics, University of Florence)

[email protected]

2

What do we know?

Large differences across European countries in terms of Divorce rates Poverty associated with marital dissolution

Marital dissolution is the most important cause for single motherhood

Poverty is particularly high among single mothers There is a strong gender bias in terms of economic wellbeing

Divorced women have a considerably higher likelihood of experiencing poverty than men

Divorced men do not have a higher poverty rate than non-divorced men.

These differences seem to be clustered – at least to some extent – by welfare regimes

4 typologies:Liberal, Social Democratic, Conservative, Mediterranean

3

Trends: Crude divorce rates around 2001 [Council of Europe (2002) ]

4

Trends: Divorces per 100 marriages in the selected countries, from 1994 to 2001 (Eurostat data)

0

10

20

30

40

50

60

70

80

United

King

dom

Irelan

d

Austri

a

Belgium

Franc

e

Germ

any

Denm

ark

Finlan

d

Nethe

rland

s

Greec

eIta

ly

Portu

gal

Spain

Pe

r 1

00

ma

rria

ge

s

1994 1995 1996 1997 1998 1999 2000 2001

5

Total Divorce Rate in the selected countries, from 1994 to 2001 (Eurostat data)

The mean number of divorces per marriage in a given year

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

United

King

dom

Irelan

d

Austri

a

Belgium

Franc

e

Germ

any

Denm

ark

Finlan

d

Nethe

rland

s

Greec

eIta

ly

Portu

gal

Spain

1994 1995 1996 1997 1998 1999 2000 2001

6

Mean duration of marriage at divorce in the selected countries, from 1994 to 2001 (Eurostat data)

0

2

4

6

8

10

12

14

16

18

20

United

King

dom

Irelan

d

Austri

a

Belgium

Franc

e

Germ

any

Denm

ark

Finlan

d

Nethe

rland

s

Greec

eIta

ly

Portu

gal

Spain

Ye

ars

1994 1995 1996 1997 1998 1999 2000 2001

7

What do we know?

Large differences across European countries in terms of Divorce rates Poverty associated with marital dissolution

Marital dissolution is the most important cause for single motherhood

Poverty is particularly high among single mothers There is a strong gender bias in terms of economic wellbeing

Divorced women have a considerably higher likelihood of experiencing poverty than men

Divorced men do not have a higher poverty rate than non-divorced men.

These differences seem to be clustered – at least to some extent – by welfare regimes

4 typologies:Liberal, Social Democratic, Conservative, Mediterranean

8

Research on divorce patterns in Europe

Increasing amount of research on marital dissolution and wellbeing of men and women in Europe. Recent studies:

Andres at al (2004) – comparative study of Belgium, Germany, Italy, UK, and Sweden:

Women with dependent children in the UK suffer most Sweden: hardly any gender difference due to high benefits and

generous provision of childcare Uunk (2004)

Generally women suffer from divorce, but at varying degree A country’s welfare state arrangements are important for the

economic consequences of divorce Median income declines weakest in Scandinavian countries

and largest in France, Austria, Luxembourg and UK

9

Research issues concerning the link between marital dissolution and economic wellbeing

Economic welfare very often measured in terms of income or poverty

BUT income and poverty are highly sensitive to the use of equivalence scales

The issue of equivalence scales becomes more precarious when considering divorce – as

this is normally associated with a significant change in household composition

They are also limited in measuring overall wellbeing Selection issues:

the prospect of a dramatic decline in wellbeing might itself affect the marital dissolution event.

10

Focus of this paper

Measurement: Here we construct a range of wellbeing indices and compare

these with poverty status Wellbeing indices are less sensitive to large shifts in the

household composition (and hence the equivalence scale) Q: Do they produce similar results? Q: Do differences between welfare regimes persist with

alternative measures of wellbeing? Selection:

Since the prospect of a dramatic decline in wellbeing might itself affect the marital dissolution event, dissolution is potentially endogenous with respect to wellbeing

We would like to identify the net effect of marital dissolution on wellbeing

We do so by applying propensity score matching (PSM) techniques

11

Measuring wellbeing: poverty

Poverty status i.e. a dichotomous version of household income[Poverty threshold defined as 40-50-60-70 percent of median net

equalised household income] Equivalence scales used

OECD (modified) They do have an impact on computed poverty rates but no

impact on the ranking of countries

Relative income Poverty is treated as a matter of degree (1 for poorest

and 0 for the richest) The level is determined by the rank in the income

distribution and the individual’s share in the total income received by the population

12

Measuring wellbeing: Deprivation index

Defining wellbeing in terms of a dichotomous poverty status variable is

an over-simplification only monetary

A Deprivation index, in contrast is multidimensional, and may include subjective measures, conditions of the

dwelling, possession of consumables, affordability etc. We built a total d.p. and then one for each of 5

“dimensions”

13

Measuring wellbeing: Deprivation index

24 items grouped:

Basic non-monetary deprivation (affordability) Secondary non-monetary deprivation (lack of

common durables) Lacking of basic housing facilities Housing deterioration Environmental and neighborhood problems

14

Deprivation index and weights

The imposed weight used in calculating the deprivation index is important

Two factors influence the weight The item’s power to describe deprivation

depends on the frequency distribution of the item. That is, lack of an item should be assigned a large weight if most people possess this item

Correlation between deprivation items Smaller weight should be assigned if the correlation is very

high

Also, not possessing an item (durables) counts toward deprivation only if the household could not afford it

15

Estimating the impact of marital disruption on wellbeing

Approach: compare wellbeing of divorced and non-divorced women

Ideally …. we would like to compare a divorced individual with him

or herself in the case where she/he is NOT divorced This would give us the effect of the marital dissolution

Problem (of course!): the two cases are mutually exclusive: A woman cannot

be divorced and not divorced at the same time. In other words – the counterfactual is non-existent

16

Propensity Score Matching (PSM)

so instead of the counterfactual, we construct an approximation to the counterfactual

Dividing individuals into a control group and a treatment group.

Treatment refers to a divorce event Control group: those who did not experience a

divorce event Then “match” or “pair” treated (divorced) with

individuals from the control group (not divorced) - which are similar in their characteristics

17

Propensity Score Matching (PSM) - STEPS

1. Matching on background variables is based on the propensity score.

2. Approach is to estimate a probit model of divorce events on background variables this gives the propensity of experiencing a treatment

3. Treated (had a divorce) and untreated (did not have any) are then matched based on this estimated propensity

4. The (average) effect of divorce on wellbeing is computed by comparing differences between matched individuals

But note that we are estimating the effect of a CHANGECHANGE in divorce status on a CHANGECHANGE in well-being status

18

Results (PSM): The effect of marital dissolution on entering poverty

Conservative countries

0,0000,0500,1000,1500,2000,2500,3000,3500,4000,450

40% 50% 60% 70%

Poverty thresholds

Men

Women

Liberal countries

0,0000,0500,1000,1500,2000,2500,3000,3500,4000,450

40% 50% 60% 70%

Poverty thresholds

Men

Women

Mediterranean countries

0,0000,0500,1000,1500,2000,2500,3000,3500,4000,450

40% 50% 60% 70%

Poverty thresholds

Men

Women

Scandinavian countries

0,0000,0500,1000,1500,200

0,2500,3000,3500,4000,450

40% 50% 60% 70%

Poverty thresholds

Men

Women

19

Results (PSM): Average Treatment Effect of marital dissolution on relative income

•No significant effects for men

•Significant effects for women, with important differences according to presence of children or not

MALES FEMALES

All Couples Couples with

children All Couples Couples with

children

Att t-value Att t-value Att t-value Att t-value

Liberal 0.014 0.624 -0.011 -0.508 0.298 9.039 0.346 8.268 Social Democratic 0.022 1.178 0.035 1.716 0.141 6.106 0.077 3.918 Conservative -0.000 -0.031 -0.001 -0.541 0.144 9.464 0.152 7.528

Mediterranean 0.011 0.472 -0.015 -0.696 0.205 9.561 0.199 7.697

20

Results (PSM): Average Treatment Effect of marital dissolution on change in total deprivation

MALES FEMALES

All Couples Couples with

children All Couples Couples with

children

Att t-value Att t-value Att t-value Att t-value

Liberal 0.124 3.100 0.093 2.203 0.138 4.166 0.123 3.094

Social Democratic 0.023 0.723 0.073 2.054 0.106 3.646 0.097 2.736

Conservative 0.041 2.688 0.044 2.341 0.058 3.682 0.075 4.795

Mediterranean 0.034 1.137 0.036 1.112 0.115 4.860 0.105 3.831

Different picture:

• Significant effects for men as well (see Soc.Dem. Men with children)

• For women: Liberal still highest, but the effects are closer

21

Results (PSM): Average Treatment Effect of marital dissolution on change in basic lifestyle deprivation (affordability)

MALES FEMALES All Couples Couples with children All Couples Couples with children

Att t-value Att t-value Att t-value Att t-value

Liberal 0.114 2.785

0.136 2.178 0.224 4.541 0.194 3.303

Social Democratic 0.033 0.850 0.100 2.251 0.166 3.646 0.104 2.173

Conservative 0.086 4.840 0.089 3.904 0.127 6.010 0.145 5.881

Mediterranean 0.025 0.809 0.024 0.613 0.126 4.374 0.118 3.988

•Effects are consistent with the results for total d.i.

•This time effects for women are higher than total d.i., weaker in case of children

•For men, now, higher and more signficant in case of children

22

Results (PSM): Average Treatment Effect of marital dissolution on change on secondary lifestyle deprivation

MALES FEMALES

All Couples Couples with

children All Couples Couples with

children

Att t-value Att t-value Att t-value Att t-value

Liberal 0.149 3.311

0.148 2.734 0.147 3.067 0.077 1.647

Social Democratic 0.069 2.147 0.119 2.625 0.157 4.976 0.129 3.179 Conservative 0.052 2.750 0.046 1.987 0.086 4.840 0.109 4.938

Mediterranean 0.049 1.578 0.042 1.008 0.134 4.393 0.149 4.826

•The gender gap is reduced

•For women Social Dem. have quite strong effects

23

Concluding remarks and caveats

Gender gap confirmed When monetary measures are used women suffer

disproportionately larger effects than men Findings consistent with welfare regime theory Effects differ depending on poverty threshold chosen The effects are somewhat different when wellbeing is

measured in terms of deprivation Gender gap not so clear any longer for social democratic and

liberal welfare regimes Gender difference is smaller when children are not present Separate estimates for couples with children and no

children Divorced women with children suffer considerbly more

compared to women without children – as long as economic wellbeing is measured in terms of poverty

Much less difference when economic wellbeing is measured in terms of deprivation or relative income

24

Extensions (not included in paper)

At the moment we analyse economic wellbeing before and after a marital disruption

Of interest is to analyse economic wellbeing two, three and four periods/year after divorce.

Will answer to what extent a marital disruption may lead to long term disadvantage

Estimates country by country, clustering the results

to see whether the welfare regimes grouping is confirmed