The Influence of Weather on Emergency Homeless Shelter Use

Abstract

We specifiy, estimate, and evaluate logistic smooth transition regression (LSTR) models to investigate how weather conditions influence the movements of people into and out of homeless shelters. What distinguishes this paper from most studies in the smooth transition literature is its use of exogenous transition variables, in our case precipitation and windchill, in addition to the standard autoregressive lags of the dependent variable. Our data describes variations in weather conditions and homeless shelter use in Calgary, Alberta, Canada. Our results highlight how severe weather conditions influence the choice of those whose experience with homelessness involves sleeping outdoors, people who are often referred to as rough sleepers. An implication of our results is that changes in weather conditions have the potential to challenge the capacity of emergency shelters to provide protection from the elements. The notable increase in homeless encampments and the consequent increase in the number of people exposed to dangers to health and life of extreme weather raises the importance of this consideration. Our estimates provide shelter operators with information helpful in maintaining the capacity required to avoid potentially catastrophic outcomes.

Keywords: Nonlinearity, Smooth Transition Regression, Homelessness, Weather conditions

JEL Classification: C01, C53, I32