It’s a paper, way over my head as a layman, about the economic effects of unemployment insurance on job search behavior based on data from the French Pôle emploi (public employment services). I vaguely understand that fixed effects means none of the parameters used in their model are random (“we calibrate the model’s parameters with realistic values”, and explicit values for all parameters are given in the Appendix, Table A.1).
I have no idea what “spell” means in the context of a “spell fixed effect” or “time-of-spell fixed effects”. The tables also refer to samples of “spells”.
An “unemployment spell” would be a period during when someone is unemployed. So a time fixed effect would be the effect of a change in how long the person was unemployed. I don’t see that in Table A.1, but apparently it can be derived from their model (so they can predict how hard the person will look for a new job as time passes, etc)
FIxed effects models in econometrics are used in what are referred to as “panel data” sets. These are data sets where a group of entities are observed multiple times. The entities might be individual people, companies, or geographic units like states. A fixed effect model assigns a dummy variable to each entity, to isolate any idiosyncratic effect associated with a particular person, company or other unit.