Thus, there is stakeholder interest in both the private and the public sectors, for information in relation to the ex-ante impact of market and policy changes across the distribution of farms and across the dimensions discussed above, including spatial and temporal. Ex-ante assessments of all European Commission proposals are now required within the policy development process. These include agricultural sector behavioural models like FARMIS (Offermann et al. 2005) or FSSIM, see Ciaian et al. (2013) for a discussion of these models. This book focuses on the development of farm-level models and discusses their evolution and application over recent years.

Microsimulation modelling is a simulation-based tool with a micro unit of analysis that can be used for ex-ante analysis. It is a micro-based methodology, utilising micro units of analysis such as individuals, households, firms and farms, using surveys or administrative data sets. It is a simulation-based methodology that utilises computer programmes to simulate public policy, economic or social changes on the micro population of interest (O’Donoghue 2014). It is essentially a computer-based laboratory for running policy and market experiments, whose development has been facilitated by the advent of the personal computer in the 1980s and the availability of micro-data that has allowed the field to grow very rapidly.

For most of its history as a field, since Orcutt (1957, 1960) the focus has been on the household unit of analysis and focus on related policy such as tax, social policy, pensions etc. There is now a growing literature based on firms (Buslei et al. 2014) or farms (Richardson et al. 2014). Whether formally called microsimulation modelling or not, micro-based ex-ante simulation-based analysis is now used extensively around the world for policy analysis and design (Shrestha et al. 2016).

The field is multidisciplinary, reflecting the different policy focuses, but is bound together by researchers who utilise computer-based simulation models to simulate the impact of public policy and/or economic and social change on micro units such as households, firms and farms. Depending upon the policy area, the discipline has different names. For some, particularly those working in public finance, social policy and rural development, the field is called microsimulation; for others in the agricultural policy, it is farm-level modelling, while for others in labour economics, it is a branch of applied micro econometrics. However, methodologically, there is much in common and much that can be learned from the different fields. It is particularly appropriate in this time of economic crisis to focus on methodologies that can facilitate better policy design.

As a modelling framework, microsimulation modelling is a mechanism of abstracting from reality to help us understand complexity better. Figure 1.1 outlines potential sources of complexity in a static, single time period microsimulation model.

In the context of policy design and evaluation, complexity can take the form of

These levels of complexity themselves interact with each other, resulting in a degree of complexity that is difficult to disentangle without recourse to a model.

Consider first the dimension of population complexity. The first dimension of complexity in relation to population is whether an analysis takes place on a population with limited or extensive heterogeneity. The former equates to a typical farm model (Hemme et al. 2014), while the latter includes distributional farm models such as Louhichi et al. (2010). The next dimension of complexity is policy complexity. This relates to the range of different policy or socio-economic impacts and the degree of policy complexity as many microsimulation models try to replicate the fine detail of legislation in modelling policy to the different types of policy modelled as in implemented CAP policy or agri-environmental policy. The third dimension of complexity is behaviour. Models that abstract from behavioural response are known as static microsimulation models. However, many policies are explicitly aimed at influencing policy as in the case of attempts to improve environmental outcomes on farms (Hynes et al. 2008; Ramilan et al. 2011).

In the case of models that incorporate either spatial dimensions or inter-temporal dimensions, the level of complexity is increased further (Fig. 1.2). Land use and spatially targeted policy (Miller and Salvini 2001; Lau and Kam 2005) or spatially targeted socio-economic effects (van Leeuwen and Dekkers 2013) require spatial models. Policies which depend upon long-term horizons such as afforestation models (Ryan et al. 2015) utilise inter-temporal or dynamic models.