Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and...

47
Using Microsimulation for Policy Evaluation Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP) President, International Microsimulation Association

Transcript of Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and...

Page 1: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Using Microsimulation for Policy Evaluation

Cathal O’Donoghue

Teagasc Rural Economy and Development Programme (REDP)

President, International Microsimulation Association

Page 2: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Focus of Lecture

Managing Complexity: Policy x Population Microsimulation Modelling Types of Models Implementation

Page 3: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Objectives of Policy Evaluation

Page 4: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Objectives of Policy Evaluation

How much will it cost? Who is affected? Are different parts of the population affected in different ways?

Distributional Impact? Who wins? Who loses? Where is the main impact?

Is there a behavioural response? What are the impact on Policy Objectives

Poverty Pollution Labour Supply Tax Revenue

What is the impact of alternative policy designs?

Page 5: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Complexity

Page 6: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Sources of Complexity

Challenge: Understand policy impact in a complex world

Page 7: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Sources of Complexity – Other

Page 8: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Type of Analysis

Page 9: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Ex Post Analysis Evaluate policy after program has been implemented Use of Treatment and Control Groups Randomised Control Experiment

Randomly Select Quasi-experimental design

E.g. Localities or groups randomly assigned Survey Before and After Program Implementation However

Difficult to have randomised experiment in public policy Expensive as one needs to implement policy first

Ex Post Analysis

Page 10: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Ex Ante Analysis Simulate impact on computer

Collect Data on Population Model Policy in Computer Assess Policy Impact before implementing policy

Less accurate Structure of behaviour may change in response to policy instrument

Simpler as experiment on computer Cheaper can be tested on computer before roll out

Ex Ante Analysis

Page 11: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Examples of Complexity

Page 12: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Unemployment Trap

-40

-20

0

20

40

60

80

100

120

T VAT SIC IT CB UA FIS Earns

• No Incentive to Work

Dis

po

sab

le In

com

e p

er W

eek

(£)

Hours Worked per Week

Means Tests Unemployment Assistance

Earnings

Child Benefit

Page 13: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Unemployment Trap – Introduce Family Income Supplement

-40

-20

0

20

40

60

80

100

120

T VAT SIC IT CB UA FIS Earns

• Introduce In-Work Benefit to provide an incentive to work 20 hours

Dis

po

sab

le In

com

e p

er W

eek

(£)

Hours Worked per Week

In-work Benefit (Family Income Supplement)

Page 14: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Combine Solutions Poverty Trap

-40

-20

0

20

40

60

80

100

120

T VAT SIC IT CB UA FIS Earns

• Combine • FIS with 60% withdrawal rate• Tax and Marginal Relief at 40% tax rate• PRSI – c. 8%• 108% Marginal Effective Tax Rate• Poverty Trap

Dis

po

sab

le In

com

e p

er W

eek

(£)

Hours Worked per Week

Taxation

Page 15: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Irish M2K Budget Constraint 1994

-40

-20

0

20

40

60

80

100

120

T VAT SIC IT CB UA FIS Earns

• Solution• Make FIS dependent upon after tax income• Exemption Limits and Marginal Relief eventually abolished

Dis

po

sab

le In

com

e p

er W

eek

(£)

Hours Worked per Week

Page 16: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Is there an Average Household?

Single Earner Couple with Children at Average Wage? Couple with Children – 33.5% Single Earner Couple with Children – 13.1% Single Earner Couple with Children at Average Wage –

1.7% Conclusion Almost no one at “average” Population Highly Complex

Page 17: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Micro Targets and Policy Objectives

Page 18: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Targets and Policy Levers

Increased focus on outcomes, targets and policy levers Millennium (Sustainable) Development Goals National Anti-Poverty Targets National Action Plans on Social Inclusion (EU) Labour Supply Targets

Pensions Policy Reform Multidimensional Financial sustainability of system Early retirement incentives Poverty risk Coverage of membership

Policy Modelling used to design and evaluate policy

Page 19: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Microsimulation Overview

Page 20: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

What Microsimulation Models Are

Micro – Household Units Simulation – Policy, Social or Economic Change

Therefore: Powerful tools for Running scenarios to Understand the impact of economic, social or policy processes On the distributional characteristics of the population

Study Impact of Public Policy Effectiveness of Existing Policy Evaluate potential reform

Those who use models and their results want more Significant Impact Increasing spread and use is an example

Page 21: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

What Microsimulation Models are not

They are not forecasting models

Dynamic Microsimulation Modelling largely died off in Europe in the late 1980’s Perception of failure of earlier models to predict the future However

Expectations too high Predictive Capacity of Models weak

Models should be used for scenario analysis!

Page 22: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Types of Models

Page 23: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Model Types

Policy Only Hypothetical

Policy and Population Static

Policy, Population and Behaviour Labour Supply Consumption

Policy, Population and Behaviour and Inter-temporal Pension Reform

Policy, Population and Behaviour and Place Spatial Policy

Policy, Population and Cross Country Pension Reform

Page 24: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Abstracting from Population Complexity – Hypothetical Models

Simplest type of model Hypothetical Model Ignores Behaviour and Population Variability Focuses solely on one dimension of complexity: Policy

Use Communicating policy reforms OECD Jobs Study Comparative analysis of Work Incentives

Issues Easier to Build Easier to Communicate No Average Household

Page 25: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Hypothetical Model

Hypothetical Family

Model Framework

Analytical RoutinesPolicy Routine

Output Routine

Page 26: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Results

• Distribution of Costs of Child Care including fee, rebates, benefits, tax reductions to produce net cost for two earner couple, each with 67% of Average Wage• Net Costs and decomposition for lone parents• Useful for policy learning

Can Parents Afford to Work - Childcare Costs Tax-Benefit Policies and Work Incentives? (Immervoll and Barber)

Page 27: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Introducing Population Complexity – Static Models

Ignores Behaviour Variability Day after effect of Policy Reform Two dimensions of complexity: Policy and Population

Contains Fine Detail of Legislative Complexity Use

Costings Winners and Losers Poverty Effectiveness of Policy Reform Distributional Incidence Widespread use

Issues Can capture complex policy and population interactions Building block of more complicated models

Page 28: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

ResultsBenefit Changes

Welfare to Work Changes

• Children shift up income distribution• Reduction in Poverty due to employment at (i) PT, (ii) FT, (iii) PT/FT at (i) + (ii) MWl (iii) AW• Poverty effect of Employment change under 2 TB - systems

How Effective is the British Government’s Attempt to Reduce Child Poverty? (Piachaud and Sutherland)

Page 29: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Introducing Behaviour

Models Behavioural Impact of Policy Reform. E.g.: Labour Supply, Retirement, Child Labour Consumption Policy Participation (Benefit Take-Up, Tax Evasion, Education)

Links Static Microsimulation Model Contains Fine Detail of Legislative Complexity

Issues Change in Behaviour as a result of a policy reform Second – round impact on cost/distribution Often response small

Page 30: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Results

WORK INCENTIVES AND ‘IN-WORK’ BENEFIT REFORMS: A REVIEW (Blundell)

• New budget constraint• Increased pressure to participate• Increased pressure to move to 30 hours

• Change from Family Credit to Working Families Tax Credit• Increased Generosity• Reduced Taper• Child Care Component

• Change in simulated employment status before and after reform

Page 31: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Distributional Impact of a Carbon Tax

• Inequality increased by carbon tax

Carbon Taxation Prices and Inequality in Australia Cornwell and Creedy

Page 32: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Introducing Behaviour – Macro Shocks

Large Interest in micro level impact of a macro-economic change Trade Agreement Macro-Shock Financial Crisis/Deregulation

Links Macro-CGE to Microsimulation Model

Page 33: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Introducing Time: Dynamic Models

Many Policies have a time dimension Pensions Policy Higher Education Loans Retirement Long-term care

Dynamic Microsimulation Model Takes Population Simulates transition of population over time

Issues Can use statistical based models or Behavioural models where behaviour like retirement choice responds to

policy

Page 34: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Poverty Impacts of Alternative Pension Proposals

• Different Indexation

• Reduction in poverty rate due to elimination of coverage gap

Poverty Impact of State Pension Reform on the Elderly: an Analysis of Reform Proposals in the 2007 Irish Green Paper (Baroni and O’Donoghue)

Page 35: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Introducing Place: Spatial Models

Models Locational Impact of Policy Reform. E.g.: Spatial Incidence of Policy Spatial Incidence of Outcomes like Poverty Behavioural impacts Commuting

Requires Spatially references micro-data However typically not available Developed data enhancement tools to link different sources

Page 36: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Results

• Travel to Work Area

• Location of Economically active with low skills in Leeds, UK• At different levels of spatial aggregation

GIS and microsimulation for local labour market analysis (Ballas and Clarke)

Page 37: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Learning from Other Countries

Policy experiments may be politically difficult or expensive

However comparative analysis or modelling Can help to learn lessons

Increasing comparative research EUROMOD Latin America Africa

Issues Comparability of Data Comparability of Policy

Page 38: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Different Units of Analysis: Farm Level Models

Models Behavioural Impact of Policy Reform. E.g.: Labour Supply, Retirement, Child Labour Consumption Policy Participation (Benefit Take-Up, Tax Evasion, Education)

Links Static Microsimulation Model Contains Fine Detail of Legislative Complexity

Issues Change in Behaviour as a result of a policy reform Second – round impact on cost/distribution Often response small

Page 39: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Winners and Losers Analysis from post 2014 CAP analysis

WinnersLosers

Winners in peripheral areas and in North EastLosers in the East and South EastHowever overlaps

Page 40: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Microsimulation in Practice – UK Department of Work and Pensions

Page 41: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Context – Objectives of DWP

PSA Targets PSA 1 Reduce the number of children in low-income households PSA 3 Demonstrate progress on increasing the employment rate PSA 4 & 7 Increase employment rates for disadvantaged areas and

groups (lone parents; ethnic minorities; people aged 50 and over; lowest qualifications; poorest local authority districts; disabled )

PSA 5 Reduce the proportion of children in households with no one in work

Other: Change in the ratio of spending on pensions by the State to

spending on pensions by the private sector from around 60:40 to 50:50 by 2025

Page 42: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Existing Models

Treasury Sets Targets DWP uses microsimulation models to design policies to achieve targets

within the budget allowed Microsimulation Models on desktop of policy analysts ~ used like

EXCEL Models used:

PSM – Tax-Benefit Analysis EHM – Labour Supply Model & Employment and Hours Model Pensions – GENESI/Pensim2 Dynamic Microsimulation Model Benefit Forecasting – Other Genesis Based Models Family Change Model

Page 43: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Don’t Re-invent the wheel

Page 44: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Learn from Others

www.microsimulation.org New Handbook

Page 45: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Learn from Others

Journal

Page 46: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Model Frameworks

Microsimulation Model Framework – Model Engine Expensive to create – 1 to 2 person years (or more) Because of cost, more effort spend on computing

environment than policy question Specific not general, so models die after initial use

General Modelling Frameworks LIAM(2) – Dynamic and Spatial Microsimulation (C++) URBANSIM – Land Use and Transport Policy (planned link

to LIAM) EUROMOD – Cross-country microsimulation (C++) XLSIM – Easy to use EXCEL based framework for national

models

Page 47: Using Microsimulation for Policy Evaluation / Cathal O’Donoghue Teagasc Rural Economy and Development Programme (REDP)

Thank You