Editorial

With the Spring 2017 issue, the International Journal of Microsimulation enters its 10th year of existence. We will elaborate on this important milestone in the forthcoming World Congress of the International Microsimulation Association, which will be held in Moncalieri (Italy) on June 2123, 2017 and more specifically in the winter issue of the journal later this year.

This issue contains 6 research articles and a book review, covering a broad spectrum of methods and topics. The first two articles, by Maitino & co-authors and Gastaldi & co-authors respectively, develop tax-benefit models for Italy, with a specific focus on indirect taxation (and in-kind transfers for the first article). While Maitino et al. offer an assessment of the distributional effects of three recent policy changes in Italy, Gastaldi et al. adopt a more normative perspective and discuss the well-known issue of regressivity of value-added tax (VAT) (the tax rate is independent of income, but poorer households have a higher marginal propensity to consume, hence to pay VAT); the paper shows that a proper modulation of different rates for different goods can mitigate the problem.

The third paper, by Veronica Amarante, models demographic change in uruguay. The paper shows that the trends in household formation and fertility have exacerbated inequality and poverty, as household size of poorer households has increased relative to household size of richer households.

Changes in labour market behaviour have only partially offset this effect.

The last three papers are more methodological in nature. The paper by Richiardi and Richardson introduces JAS-mine, an open-source platform for discrete-event simulations, including microsimulations and agent-based models. The platform offers tools that speed up model development and increase model transparency, and contributes to the trend of convergence between the two modelling techniques.

The paper by Justin van de Ven presents a calibration exercise of the parameters of a structural overlapping generation model, based on hierarchical minimisation of the distance between moments computed on the simulated and the real data. Because (i) stationarity of the moments is not considered (and is not even possible, given the policy changes that occur during the observation period), (ii) some of the parameters are exogenously fixed (i.e. observed), and (iii) distance minimisation is achieved by manual adjustment of model parameters, the approach belongs to the class of mixed calibration-estimation methods, which cannot ensure consistency of the model estimates, but is nevertheless a significant advance with respect to simpler methods still used in the literature.

Finally, the contribution by Philips & co-authors deals with the problem to generate a synthetic population at a fine spatial resolution when not all attributes are available in the micro-data. The paper proposes to use simulated annealing to simulate the majority of the required attributes, and Monte-Carlo sampling to add the missing ones.

This brings us to Trond Husby’s review of the book SpatialMicrosimulation with R by Robin Lovelace and Morgane Dumont (2016), a practical manual on how to write spatial microsimulation models using the popular open-source R programming language and software environment, which will be of interest to many of the readers of the journal.