Download presentation
Presentation is loading. Please wait.
1
IMA Meeting Budapest 21-22 September 2016
The Australian Commonwealth’s labour supply microsimulation model CAPITA-B IMA Meeting Budapest September 2016 Joseph Mercante Australian Department of Employment The findings and views reported are those of the CAPITA-B development team and do not necessarily reflect those of the Department of Employment or the Australian Government. The simulation results presented are preliminary and should not be quoted nor used to inform policy development.
2
Outline Background to the project Model Results so far Future work
3
Why behavioural microsimulation?
Model changes to labour supply from changes in financial incentives arising from tax, welfare & employment policies Help develop policies to enhance labour force participation Understand responses to increase/decrease participation (extensive margin) and increase/decrease working hours (intensive margin) Behavioural impact may make policies less expensive
4
History of behavioural microsimulation by the Australian Government
MITTS-B (Melbourne Institute) STINMOD-B (Hybrid model, Treasury’s Participation Modelling Project, ) Used for policy over Need for a new model CAPITA replacement model (Commonwealth, led by Treasury, released 2015) CAPITA-B prototype (Productivity Commission, 2015, experimental and limited for policy applications) Department of Employment enhancements: CAPITA-B Comparative Analysis of Personal Income tax and Transfers of Australia – Behavioural
5
CAPITA-B Project Started end 2015 Phase 1
Basic model to replace STINMOD-B functionality Improve Productivity Commission’s prototype Completed July 2016 (currently undergoing QA) Phase 2 Re-estimate wages & labour supply models Childcare policy (non-behavioural & behavioural) Started July 2016 completion by beginning 2017
6
Main elements of CAPITA-B model
Micro data for the simulations (SIH) The tax/transfer model CAPITA (calculates disposable incomes) Preference equations to estimate utilities Wage equations to impute wage rates for non-workers Rules for benefit take-up A Monte-Carlo process to simulate labour supply Calibration to observed labour supply Modules to report results of simulations.
7
CAPITA-B design and architecture
Wage Equations CAPITA-B CAPITA Policy Modules Basefiles Output Data Behavioural Components Calibration Results Policy Parameters Preference Equations
8
CAPITA arithmetic model
Analyses of the impact of changes to tax and transfer payment policies on individual’s tax liability, transfer payment entitlement, and their final income Survey of Income and Housing (2011/12) Future policy years (5 years) Benchmarked Broad tax and welfare system coverage Most benefits & family payments No superannuation No consumption taxes
9
Labour supply modelling in CAPITA-B
Classical utility maximising labour supply assumption Singles and single parents Joint utility for couples Dependents’ labour supply ignored Discrete choice framework 11 hour choices for singles 66 combined hour choices for couples Borrowed from MITTS Kalb (2002), Van Soest (1995) Imputation of wage rates for non-workers Heckman sample selection bias correction model Borrowed from MITTS (Kalb & Scutella 2002, Heckman 1979)
10
Modelled population Exclusion category Age Pension age 3,502,000 *
Age Pension age 3,502,000 * Disabled, Carers & Veteran 1,361,000* Self-employed 1,781,000* Full-time students 2,531,000* Other 2,227,000* Non Private Dwellings 1,192,000 Excluded adults 8,871,000 Dependents 15 and over 1,215,000 Total excluded 10,086,000 Not excluded 8,179,000 Total persons 18,622,000 45% * Not mutually exclusive
11
Benefit take-up Simulating people at non-observed hours choices.
At 0 hours, people don’t usually have zero income because they get transfer payments. CAPITA-B applies a set of rules to non-beneficiaries determine what payments to put people on. Assumes 100% take-up. Use rule to determine principal carer in couple income units. Re-allocation of benefits for some persons Take-up for dependents 15 years and over
12
Other features Treatment of tax deductions
Treatment of negative incomes Parental income test
13
Testing and benchmarking
How do we know the model produces valid results? Predicted vs observed hours distributions Wage distributions Impact of benefit take-up rules Policy simulations Do non-behavioural results make sense? Do fiscal results (without behaviour) match expectations? Do labour supply results make sense? Elasticities Comparing models Sensitivity analysis / confidence intervals
14
Observed and predicted working hour distributions
15
Observed and predicted working hour distributions (cont.)
16
Predicted (non-workers)
Wage estimation Observed Predicted (workers) Predicted (non-workers) Partnered men 34.9 30.4 26.2 Partnered women 28.4 26.9 18.8 Single men 25.8 25.4 22.8 Single women 25.3 23.3 21.0 Single parents 25.1 24.2
17
Benefit take-up (non-dependents, at observed working hours)
Benefit type CAPITA CAPITA-B Difference Non-modelled population 7,151 7,146 -5 (0%) Sole Parents with child ≤ 8 (PPS) 407 513 106 (26%) Unemployment benefit (NSA) 591 1138 547 (93%) Partner principal carer with child ≤ 6 (PPP) 85 178 67 (60%) Youth Allowance (non-student) 189 104 (122%) None 19,645 18,826 -819 (-5%) Total 27,990 Unweighted sample
18
Simulated elasticities
Partnered men Partnered women Single men Single women Single parents Own wage 0.18 0.36 0.16 0.20 % change 3 2 1 Cross wage 0.00 -0.02 Based on aggregate labour supply change to 10% increase in wage rate Partnered women highest Expect single parents much higher Participation response (extensive margin) around 70% of elasticity Partnered women higher response to their partner’s wage
19
Simulation example Increase tax threshold to deliver small tax cut
$37,000 increased to $38,000 income rate $18,200 $18, $37, $38,000 19% $38, $80,000 32.5% $80, $180,000 37% $180,000+ 45% Notes: Uses tax system as at 2015 (that is what model is calibrated to)
20
Non-behavioural Impact – Simulation example
Partnered men Partnered women Single men Single women Single parents All Impacted (%) 60.6 40.0 47.0 35.7 32.0 43.8 Winners (%) Average win ($ pa) $134.4 $133.8 $134.2 $134.1 Losers (%) Average loss ($ pa) $0.00 All persons Overall 43% winners (partnered men largest group) Average $134 per annum to winners
21
Behavioural Impact – Simulation example
Partnered men Partnered women Single men Single women Single parents All Persons responding (%) 0.13% 0.14% 0.05% 0.10% 0.04% 0.11% Net employment change (persons) 1,374 1,476 683 526 130 4,187 Δ FTE (persons) 1,818 2,338 956 713 157 5,982 Total hours change (hours) 63,646 81,838 33,453 24,951 5,491 209,380 Participation effect (hours) 59,370 52,773 28,932 20,567 4,678 166,321 Hours effect (hours) 4,276 29,065 4,521 4,384 813 43,059 Participation component (%) 93% 64% 86% 82% 85% 79% Low number of responders Modest changes Modelled persons
22
Fiscal Impact – Simulation example
Millions $ Change (non-behav.) Change (behav.) Revenue -1,154 -1,093.8 Expenditure -49.6 Net expenditure 1,154 1,044.2 All persons
23
Future work Wage and utility equations re-estimated
May experiment with improving wage imputations Child care policy in non-behavioural & behavioural model Confidence intervals Policy toolkit to aid interpreting results
24
Limitations of CAPITA-B model
Some policies not currently modelled in CAPITA Some population groups not modelled, yet may be a policy focus (eg. people close to retirement age) Supply side (desire to work), no labour demand & business cycle considerations 100% take-up may not be realistic Currently cannot model employment policies that involve obligations/activity tests or the impact of work-ready programs on participation Single time period (no savings, human capital accumulation) Exclude institutional factors / barriers to working additional hours
25
Lessons Modelling is assumption-laden.
Many factors can make results sensitive Need extensive testing to check if the results make sense Very low number of people responding to changes in incentives One of a number of useful tools for policy analysis/development
26
A team effort! Acknowledge the contribution of all team members in Quantitative Analysis Section Other members of the team Quoc Ngu Vu (Director) Wei Ying Soh Nathaniel Leonard
27
Questions
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.