Browsing by Author "Hirmas Frisius, Alejandro Bernhard"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Heinrich's pyramid and occupational safety : a statistical validation methodology
    (2018) Marshall Rivera, Pablo; Hirmas Frisius, Alejandro Bernhard; Singer González, Marcos Johanan
  • Thumbnail Image
    Item
    Tell me the stakes and I will tell you how rational you will be model selection for entry games
    (2014) Hirmas Frisius, Alejandro Bernhard; Fosco, Constanza; Pontificia Universidad Católica de Chile. Instituto de Economía
    The following paper presents a different approach for selecting models of agents' behaviour. Specifically, this study is focused on Entry-Games, where agents choose whether to enter a risky market and compete or stay-out. Generally, models are assumed for the whole population and through simulations of these behaviours, the models are compared by the fit with the experimental data. This approach benefits the models which can represent partially different kinds of behaviour. Here instead, different specific strategies can fit part of the population, while others describe the rest. For that the models are compared jointly, allowing agents to choose one out of several models to act with. The models differ by the use of information and the likelihood for them to be used is conditional on the risk involved on the game (exogenous and related to agents' decisions). The estimations are performed by using a large database of Entry-Games, which differ in the risk involving the decisions. This paper shows, that when agents' decisions are more relevant, the use of information increases. Analogously, when there is a large exogenous risk, the agents seem to act using less information. It will also be proven that a positive skewness might reduce the information usage for the agents. Finally, the players are assumed to use heterogeneous strategies (i.e. have di fferent parameters). The results show that the agents seem to have significantly different distributions for the parameters. This paper shows a different approach for selecting models and gives useful insight for simulating agents' behaviour in Entry-Games. The agents seem to change their strategies, when the scenarios are different. Also, some models that are not preferred by the literature, seem to predict significantly well a great portion of the population.