Module 6.
2_Exercise 3_Group 1 (Angela, Nino, Oscar, Temmy, Lenja)
Recall from group work #3 task description:
Imagine, you are the teacher of this class and you need to design a final exam. Formulate
exam questions based on the literature stated under “required readings”. For each of the
sources of the required readings, formulate two exam questions. Also formulate a complete
answer for the respective question, such that you know what answer to your question you
expect from the students to give to successfully complete the question.
All your exam questions and their answers shall be formulated closely on basis of the text of
the literature. In brackets after the answer state the respective page number, the paragraph(s)
number on the page and the line(s) in the paragraph that is the basis for your exam question
and its answer.
Source 1
Salanauskaite, L. (2012), Tax-Benefit Microsimulation in European Transition Countries: A Review of
Issues and Applications. Chapter in ”Distributional Impacts of Public Policies: Essays in Ex-Ante and Ex-
Post Evaluation” (Dissertation, p. 87 -111)
Exam Question 1.1:
What is microsimulation?
Answer to Exam Question 1.1:
Microsimulation is a data analysis method based on the use of a sample of an entire population of
individual observational units (persons, households, firms, etc) to simulate micro-level effects of a
certain event of interest, like the effect on household income due to the implementation of a new
policy.
(Page 90, Paragraph 3)
Microsimulation analysis is based on a sample or the entire population of individual units
such as persons, households, vehicles or firms. Its purpose is to simulate micro-level effects
of a certain event.
Exam Question 1.2:
According to Salanauskaite (2012); the key challenges related to the proliferation of Micro
Simulation Models (MSMs) globally are: (a) accessibility of MSMs, (b) their complex technical
development and (c) difficulty of validation. For each one of the challenges, explain the main
arguments stated by the author to refer to them as challenges.
Answer to Exam Question 1.2:
1.2. (a) Accessibility of MSMs
Full access to the MSMs for third persons is rare due to property rights restrictions and data access
issues. Additionally, the large diversity in construction approaches makes it difficult to understand the
models without the help of the owner institution or proper detailed documentation that almost
always doesn't exist.
(Page 93, Paragraph 2)
First, access to MSMs is rarely (fully) feasible to external users often due to property rights
but also due to data access issues, etc. This impedes a more detailed understanding of the
MSM properties especially if external documentation on the MSM structure is limited or not
available. Second, there is a large diversity of MSM construction approaches. This implies that
substantial model specific knowledge is needed for its usage – a difficult barrier to overcome
without the support of the developing institution.
1.2. (b) Complex technical development
MSMs are costly in time and money, resources involved in a project to the construction of a MSM
come in several kinds: expert personnel, hardware, software and quality data. In addition, major
technical complexity emerges from integrating and parameterizing detailed policy rules into
large-scale models as well as combining various data sources and types
(Page 93, Paragraph 3)
Building an MSM implies working with complexities such as integrating and parameterizing
detailed policy rules into large-scale models as well as combining various data sources and
types. The construction costs accrue due to various types of demands: people, hardware,
software, data, etc. As a result, model setup, the development of its interface and the
detailed and clear documentation of the MSM all contribute to making MSM’s very costly
1.2. (c) Difficulty of validation
It is difficult to get the required micro and macro data with appropriate quality patterns to contrast
the outcomes of a MSM. In addition the large extent of assumptions and interrelations in the model
makes the validation task more difficult.
(Page 94, Paragraph 1)
In many cases the outputs of MSMs can diverge from reality because of data issues:
underlying data quality, biases with survey weighting, use of imputed data, etc. Other issues
such as the large number of interrelations, assumptions and conditions to account for, or
even difficulties in acquiring external aggregates against which the MSM outputs could be
compared o (i.e. Klevmarken, 2002) are of importance too.
Source 2
Tanton, R. and B. Phillips (2016), "Microsimulation modelling and the use of evidence" p. 187-203 in
book "Evidence-based policy making in the social science" edited by Gerry Stoker and Mark Evans
Exam Question 2.1:
What is an advantage of microsimulation models over other models used in economic and social
sciences, like Cluster analysis or CGE?
Answer to Exam Question 2.1:
Because microsimulation operates at the individual level, they can be used to investigate
distributional results. So, distributions of income can be calculated rather than just a summary
measure like mean. This was used by Tanton (2011) who was able to use a spatial microsimulation
model to compare the frequency distribution of incomes in two subcomponents in Australia, to
identify the reason for differences in poverty. This analysis identified a higher proportion of
pensioners in one sub district compared to others, leading to a higher poverty rate.
So, microsimulation allows the user to drill down into distribution of results, providing more
information to help explain particular aggregate results. One caution is that the number of records
that the results are being based on can rapidly reduce, meaning they can become meaningless.
microsimulation models can take into account behaviours in a way that other models cannot. For
instance, agent-based models are useful for studying human behaviour but they tend to be
experimental rather than empirical.
While microsimulation models have great advantages over other models, they can be a powerful tool
when combined with an aggregate model, for example, being able to get distributional outcomes
from CGE model or using regression models to impute new variables.
(pages 187-188).
Exam Question 2.2:
There are several types of microsimulation models mentioned in the Chapter "Microsimulation
modelling and the use of evidence". Please, name and explain these types.
Answer to Exam Question 2.2:
There are four types of microsimulation models explained in the chapter "Microsimulation modelling
and the use of evidence" by Tanton and Phillips.
● Static models,
● Behavioural models,
● Dynamic models,
● And spatial models.
Static models only estimate the monetary effect of a policy on households after their introduction
(the day-after effect). Static models do not consider possible behavioural changes of individuals,
initiated by the effect of the policy. Such behavioural changes are for example an increase in working
hours, if a policy change leads to lower overall income. (p.189, paragraph 4 and p.190 paragraph 1)
In behavioural models in contrast, behavioural responses of households to a policy change are
observed. The incorporation of possible behavioural changes, after a change in policy took place
enhances the microsimulation model, by fostering representation of effects due to including the
important aspect of societies reaction. Anyhow, modelling behavioural responses fully accurate is
challenging and comes along with several uncertainties. (p.190 paragraph 2)
Dynamic models include another very important factor within their microsimulation, the time effect.
That does mean, within dynamic models sample records are dynamically aged into the future to
observe demographic changes in society. Including the time effect into a model is quite complex.
Therefore, dynamic models are especially interesting for long term modelling and in areas where
strong cohort changes do exist, as in these two scenarios, great impacts of demographic change can
be determined. (p.191 paragraph 2f.)
In spatial models another interesting aspect but time is incorporated, the geographical aspect. Spatial
models do observe the geographical consequences of a policy change. For example, such models
involve the opportunity to compare estimates of poverty on a regional level before and after a policy
change. Thereby, one can check if there are greater effects of the policy change on people living in a
certain region. (p. 191 paragraph 4, p. 193 paragraph 2)
