An Agent-Based Simulation of Urban Rental Markets

We simulate a rental housing market with search and matching frictions, in which both landlord and tenant agents are imperfectly informed. Homogeneous landlords set rents to maximise revenue, using information on the market to estimate the relationship between asking rent and time on the market (TOM). Tenants, heterogeneous in income, engage in undirected search accepting residences based on their idiosyncratic tastes for housing and a disagreement point derived from information on the distribution of offers. Landlords deduce hasard rates across asking rents through knowledge of the number of recent rentals and the total TOM of both these residences and of vacant residences in narrow intervals of rents. They fit an estimation of the relationship between TOM and rent, and using the expected duration of stay for a tenant and a discount rate, set their asking rent to maximise their revenue. Tenants see a percentage of offers and from the distribution of offers decide their disagreement utility for acceptation of a visited residence. The model is closed with equal numbers of tenants and landlords, who own one residence each. The simulation converges to a steady state with dispersed rents and positive vacancy rates and times on the market. The main results concern the asymmetric effects of increasing information on both sides of the market. When tenants see more of the distribution of offers: the population rises and tenants' utilities rise as does overall welfare, as tenants learn to refuse high rents. When landlords have greater information: landlords' utilities fall and overall welfare rises slightly, as increased competition between landlords causes rents to fall. When two types of landlords are present with different levels of information, the better informed landlords charge lower rents, and experience lower vacancy rates.