EXPLAINING SHORTTERM CHOICE THROUGH RANDOM UTILITY MODELS Ral

EXPLAINING SHORT-TERM CHOICE THROUGH RANDOM UTILITY MODELS Raúl Prellezo, AZTI-Tecnalia, Efimas 3/12/2021 1

INTRODUCTION Dynamics: Roughly speaking two levels of fleet dynamics can be found, short and long term: Behaviour of fleets, with or without (presumed) financial investment 3/12/2021 © AZTI-Tecnalia 2

INTRODUCTION Structure of the presentation • Two main approaches • Bioeconomic • RUM • Implementing RUM for Short Term Choice • IIA • Spatial Resolution • Identification of the explanatory variables • Selection of the Model • Case Studies • English Beam Trawlers • Danish Gillnetters • Basque Trawlers • Conclusions 3/12/2021 © AZTI-Tecnalia 3

MODELS • Bioeconomic models • Random utility models • Gravity models • “ad hoc” models 3/12/2021 © AZTI-Tecnalia 4

Bioeconomic Models • Homogeneity is the main assumption in these type of models: – Homogeneity in spatial stock distribution. – Homogeneity in fishing effort distribution. • Relevance of distance from ports to fishing grounds, in order to further understand short-run decision making of fishers in their allocation of fishing intensity. • Theory from ITQ: Minimize the cost of fishing the total quota. • … 3/12/2021 © AZTI-Tecnalia 5

RUM • Covers the heterogeneity of fishermen. • The benefit that an individual receives is observable with some degree of uncertainty. • No many approaches for integrating both (bioeconomic + RUM). • Both face problems of spatial data collection 3/12/2021 © AZTI-Tecnalia 6

RUM + FLR 3/12/2021 © AZTI-Tecnalia 7

Basque Trawlers (Area) • Low resolution model (three areas choices, based on ICES sub-areas). • Multinomial logit (individual specific) with a large set of significant variables, including economical, management, distance and behavioral variables. • Given that in the multinomial logit, one choice is normalized, marginal effects of each variable to each of the response were calculated. • For example a change in the expected turnover will decrease the possibility of selecting Bo. B increasing the probability of selecting Sub-area VII and maintaining the probability of selecting VI. 3/12/2021 © AZTI-Tecnalia 8

Basque Trawlers (Gear) • Short/mid term (no investment or small) • Binomial logit (individual specific) with a small set of significant variables, Production function, TAC (no hake), • Easy to implement. 3/12/2021 © AZTI-Tecnalia 9

CONCLUSIONS • RUM is the state of the art empirical analysis tool of short term discrete choice, not assuming homogeneity and providing the probability of a particular choice. • Importance of behavioral characteristic, such as past history is remarkable in all cases analyzed. Overall it has validated that past history is followed for locations fished. 3/12/2021 © AZTI-Tecnalia 10

CONCLUSIONS • It has been probed that this RUM approach is able to predict the tactical spatial behavior. • Given that short tem closures are getting more frequently used as a management instrument, the demand for this type of analysis is important (area). • Is it worthy to wait to January meeting? 3/12/2021 © AZTI-Tecnalia 11