Maatschappij Beleid en Evaluatie Tim Goedem tim goedemeuantwerpen
Maatschappij, Beleid en Evaluatie Tim Goedemé tim. goedeme@uantwerpen. be Les 8 – 17/11/2015
Follow up brief evaluation Resultaat tussentijdse evaluatie Maatschappij, Beleid & Evaluatie 14 12 10 8 Gemiddelde Snelheid: 6, 2 Hoeveelheid: 7, 10 Moeilijkheid: 6, 48 6 4 2 0 1 2 3 4 snelheid n = 29 5 hoeveelheid 6 7 moeilijkheid 8 9 10
Lecture 8 Microsimulation as a tool for policy evaluation 1. Detailed example 2. Relevant indicators & assignment 3. Microsimulation questions 4. Characteristics of microsimulation 5. Assignment
Main questions this lecture • What is microsimulation? • How does it work in practice? • What types of policy evaluation & preparation can be carried out with it? • What are its strengths and weaknesses? • How should results be interpreted?
Background texts • Van Mechelen, N. , and Verbist, G. (2005), 'Simulatiemodellen: instrumenten voor sociaal-economisch onderzoek en beleid' in Tijdschrift voor sociologie, 26(1/2): 137 -153. [compulsory] • Figari, F. , Paulus, A. , and Sutherland, H. (2015), 'Microsimulation and Policy Analysis', in Anthony, B. A. and François, B. (eds. ), Handbook of Income Distribution, Oxford: Elsevier, pp. 2141 -2221. [background]
1. Detailed example • Experiments are powerful tool, but require specific data & specific assumptions => important limitations: – Average causal effect & constant effect – Stable-unit-treatment value assumption (SUTVA) often not realistic for policies – Singular events • Important alternative (for some questions): microsimulation
1. Detailed example • Based on EU-SILC data • See separate excel file
Relevant indicators • See output MEFISTO (http: //flemosi. be). • Budgetary effects • Winners and losers • Inequality & poverty • … • Assignment (in groups of 2 -3 students): reforming child benefits in Flanders: to whose benefit?
Microsimulation questions • What is the (re)distributive impact of a (set of) policies? • What is the budgetary and (re)distributive impact of tax-benefit reforms? (new or implemented in the past) • What can we learn from systems in other countries? Are they transferable to the local context? • What is the cost / benefit of particular policies (or reforms)?
Crucial questions • What is the public good / ‘public interest’? [Bart DW 02: 30] – Who is taken into account (voters, current generation, future, …? ) – Which interests are taken on board? – What is the time horizon? • Who gains, who loses?
Public Policy: definition Who gains, who loses?
Crucial questions • What is the public good? • Who gains, who loses? • What is the purpose / objective? – The direct purpose – The underlying purpose – The ultimate purpose – How does the current measure relate to this?
Public Policy: definition Why?
Microsimulation questions • Who gains, who loses from particular policies or changes in policies? • But also: – – Forecasting & nowcasting; Imputation missing values in data Calculation of specific indicators (e. g. work incentives) …
Characteristics of microsimulation • Simulations at the micro-level: take full account of heterogeneity in the population • Three types (though boundaries are blurred): – Static: deterministic (+ reweighting) – Dynamic: allow characteristics of population to change – Behavioural: estimate policy effects on behaviour (e. g. labour supply) • Also in other fields of study (e. g. schools, firms, transportation, …) • Typically ‘what if’ questions => constructing counterfactuals
Characteristics of microsimulation Advantages: – Full heterogeneity of population (<-> macro / representative agent models) – Analysis of singular cases – Counterfactual for same persons (cf. ‘fundamental problem of causality’) – Decomposition of effects and identification of driving factors (cf. Lecture 9)
Characteristics of microsimulation Advantages (continued): – Interactions between policies + interactions between individuals taken into account: no SUTVA assumption required (to be nuanced) – Not only average effect, but complete distribution of effect – With representative data (in contrast to ‘model family simulations’) – Both ex-ante and ex-post evaluations are possible (causes of effects and effects of causes)
Characteristics of microsimulation ÞImportant tool in the toolkit of those evaluating macro-level policies!!! ÞBut limitations should not be overlooked, combination with other research techniques is always warranted
Characteristics of microsimulation Limitations • Data quality – measurement error, missing variables (e. g. wealth, contribution record), representativeness, sample size… – Sampling & non-sampling errors are important! – Mismatch with existing administrative aggregate data • Quality simulation model: – Not all policies simulated (lack of data & complexity, e. g. pensions) – Non-take up and tax evasion not adequately modelled – Discretionary and local measures not adequately modelled
Characteristics of microsimulation Limitations • First-order effects: direct effect (“morning-after effect”) – Fixed vs. floating poverty line • Second-order effects: behavioural changes – – Labour supply models Models of consumption (e. g. estimate effect on VAT) Demographic effects, … (Reweighting vs. simulation) • ‘Third-order’: general equilibrium (inflation, labour demand, …) => Second, and third order effect may substantially change conclusion! Caution is necessary!
Budget neutrality • Adding benefits without taxes or other shifts in policies captures very partial redistributive effect • Second-order assumptions even stronger • But not all taxes simulated, budget neutrality not always helpful requirement
Assignment • Reforming child benefits in Belgium: an exercise with MEFISTO • Groups 2 -3 persons • Next Lecture: discussion
References • Van Mechelen, N. , and Verbist, G. (2005), 'Simulatiemodellen: instrumenten voor sociaal-economisch onderzoek en beleid' in Tijdschrift voor sociologie, 26(1/2): 137 -153. [compulsory] • Figari, F. , Paulus, A. , and Sutherland, H. (2015), 'Microsimulation and Policy Analysis', in Anthony, B. A. and François, B. (eds. ), Handbook of Income Distribution, Oxford: Elsevier, pp. 2141 -2221. [background]
- Slides: 23