ABSTRACT
This chapter focuses on the ex post alignment methods, as they are the most common form of alignments in microsimulation. In most cases, a microsimulation model applies this prediction process to all observations individually without constraint at aggregate level. Microsimulation models typically simulate behavioural processes such as demographic, labour market and income characteristics. The method uses statistical estimates of these systems of equations and then applies Monte Carlo simulation techniques to generate the new populations, typically over time, both into the future and when creating histories with partial data, into the past. Multiplicative scaling, which was described in Neufeld, involves undertaking an unaligned simulation using Monte Carlo techniques and then comparing the proportion of transitions with the external control total. The chapter also discusses the purpose of alignment in a microsimulation model and the common practise of their statistical implementation.
