"The financial crisis has highlighted many shortcomings of conventional macro models. Agent-based (AB) models are considered a promising alternative and -thanks also to some important projects financed by the European Commission- Europe is on the frontier of research in the field. However, AB models are rarely taken to the data, but for some ad hoc calibration. Until estimation of AB models become a common practice, they will not get to the center stage of macroeconomics, and policy makers will not rely on them for guidance. The ultimate goal of LAB2DATA is to investigate the most suitable methods for estimating large-scale AB models. To reach this goal the scholar first needs to focus on the specificities of AB models which make estimation hard. These relate to the fact that the aggregate properties of an AB model remain hidden in the complexity of the relations among the different agents and layers (micro and macro) of the system and cannot be directly exploited for estimation.Then, the scholar needs to focus on the most promising approaches. One is simulated minimum distance (SMD). This has been applied to some small-scale AB models, with few parameters. Another one is sequential Monte Carlo (SMC) methods, also known as particle filtering, which is applied -in simplified settings- in DSGE models.A first research goal of the project is the application of SMD to larger scale models. A relevant issue here is the ergodicity of the simulation model, which has to be tested in the artificial data generated by the model. A second research goal is the application of SMC methods, which provide probabilistic assessments of the likely evolution of the system and do not rely on ergodicity. As AB models share in many respects the same mathematical structure and complexity of atmospheric models, and these are estimated by SMC methods, the scholar will look at this literature for guidance. A final research goal involves the comparison of the two approaches."