Did the Working Families

1
Did the Working Families Tax Credit
work?Analysing the impact of in-work support on
labour supply and programme participation

• Mike Brewer, Alan Duncan, Andrew Shephard
• and María José Suárez

2
Outline

• The paper evaluating impact of changes to
  in-work benefits (WFTC) on labour supply
• Take account of all tax and benefit changes
  between 1997 and 2004
• Use structural ex ante evaluation, with
  validation from (internal and external) ex post
  evaluation results
• Focus is on initial 1999 WFTC reform more recent
  tax credit reforms in 2003 not covered in the
  paper
• Working Tax Credits for all low-wage workers
• Child tax credits combining non-work related
  child payments

3
Aims contributions of paper

• Use micro-data from before and after WFTC to
  estimate structural model of labour supply and
  programme participation
• Structural model needed to disentangle impact of
  WFTC from contemporaneous tax and benefit changes
• Data from before and after reform identifies
  changes in preferences for in-work benefits
  (stigma)
• Similar to earlier work (Blundell et al, 1999
  2000)
• Funded by UK Inland Revenue
• WFTC part of sustained assault on child poverty

4
More parents are working
5
The WFTC reform

• WFTC replaced Family Credit in October 1999
• Evolutionary reform
• Weekly, requires 16hrs/wk work
• Awards depend on hrs/wk, earnings of claimant
  partner, capital, family structure expenditure
  on formal, registered childcare
• Comparison with Family Credit
• Lower withdrawal (phase out) rate
• More generous
• New childcare credit
• Change in administration
• Aims relieve poverty, encourage work and reduce
  stigma

6
Budget constraints for lone parent (change in
in-work support only)
Assumes 2 children lt 11, hourly wage of 5/hour,
no childcare costs, no rent, no child support
7
Budget constraints for lone parent
Assumes 2 children lt 11, hourly wage of 5/hour,
no childcare costs, no rent, no child support
8
Budget constraints for a 2nd earner in a couple
with children
Assumes 2 children lt 11, hourly wage of 5/hour,
no childcare costs, no rent, no child support,
partner earns 300/wk
9
To what extent can policies explain changing
employment?

• Difference-in-differences/natural experiment
• Compares outcomes of eligibles and non-eligibles
• Difficult to isolate impact of specific reform
• Structural labour supply model
• Estimate utility function of income-hours
  trade-off
• Simulate effect of actual or hypothetical reforms

10
Specifying a structural labour supply model

• For lone parents, utility function defined over
  net income and hours

• Approximate function by

11
Methodology (continued)

• Model additionally allows for
• Unobserved work-related (fixed) costs
• Childcare costs
• Programme participation (hassle or stigma) costs

12
Methodology (cont)

• In couples, utility defined over total net income
  and individual hours choices

13
Estimation

• Data UK Family Resources Survey 19952003
• Sample includes both pre- and post-treatment data
  
• valuable both for identification and validation
• Missing wages childcare expenditures
  pre-estimated
• Structural likelihood integrated over rph and the
  estimated distributions of wages and childcare
  costs
• Use a simulated ML technique integrals replaced
  by averages over 10 random draws (independent
  errors)

14
Parameters (lone parents)

• Preferences for income
• Increase with number of children, age of youngest
• Decreasing in age and education attainment
• Distaste for work
• Increases with number of children
• Decreasing in age and education attainment

• Fixed costs of work
• Higher with young kids
• Vary by region
• Stigma costs
• Vary with age of youngest
• Increasing in age and education attainment
• Rise after WFTC, then fall

15
Simulating policy reforms

• Use parameter estimates to simulate the effect of
  moving between two systems.
• For given random draws, can calculate preferred
  choice of weekly hours and programme
  participation
• Averaging over many draws gives transition matrix
• One can calibrate transitions probabilities on
  observed outcomes by drawing from conditional
  distributions of stochastic terms

16
Transition matrix lone parents
Non-participation Part Time Full Time Total
Non-participation 49.56 2.36 2.75 54.68
Part Time 0.00 22.44 1.10 23.54
Full Time 0.00 0.41 21.37 21.77
Total 49.56 25.22 25.22 100.00
17
Transition matrix married women
Non-participation Part Time Full Time Total
Non-participation 27.79 0.21 0.05 28.05
Part Time 0.52 36.62 0.04 37.18
Full Time 0.31 0.11 34.35 34.77
Total 28.62 36.94 34.44 100.00
18
Ex ante evaluations All reforms, 1999-2002
Change in participation (ppt) WFTC All reforms
Lone Parents 5.11 3.66
Married Women -0.57 -0.49
Partner working -0.64 -0.52
Partner not working 0.06 -0.26
Married Men 0.75 -0.40
Partner working 0.19 -0.55
Partner not working 2.11 -0.01
19
How do ex ante evaluation results line up with
other ex post studies?
Effect on lone parents employment Effect on employment of parents in couples
Gregg and Harkness (2003) All reforms affecting lone parents between 1998 and 2002 5 ppts (lone parents) N/a
Francesconi and van der Klauw (2004) All reforms affecting lone mothers between 1998 and 2001 4 ppts in 1998 7 ppts by 2001 (lone mothers) N/a
Leigh (2004) All reforms affecting parents in autumn 1999 1 ppt 1 ppt
20
Conclusions

• Model suggests WFTC raised labour supply of lone
  parents by over 5ppt, but other reforms reduced
  labour supply
• Smaller effect for couples
• Decline in labour supply of women, increased
  labour supply from men in workless households
• Non-WFTC reforms reduced labour supply
• Natural experiment result broadly agree for
  lone parents less robust results for couples
• Recent reforms mean the incentive to work at all
  is
• stronger for most lone parents
• for adults in couples, more likely to be weaker
  than stronger
• couples with children face larger incentive to
  have 1 worker and 1 carer
• Part of sustained assault on relative child
  poverty