Congestion, risk aversion and the value of information

Assumptions 1 and 2 are illustrated in Figure 1. The number of drivers taking route R, NR, is measured in the usual way to the right from the left-hand vertical axis, while the number taking route S, NS , is measured to the left from the right-hand axis. [...] For each information regime, all drivers are assumed to know all the parameters of the model (values of p and t−R, congestion functions tS (.) and t+ (.), distribution of θ, price of information π), and to be able to compute the equilibrium solution.3 With Private information the equilibrium is the same as with No information since the route choice of one (atomless) driver does not affect traffic [...] For the second step the sets of drivers choosing each strategy and thus the equilibrium values of nR, nS and nI are determined given the individual strategy choices derived in step 1. 5.1 Driver strategy choice with exogenous traffic The first step is to derive an individual driver’s choice between strategies R, S and I as a function of θ, the price of information, π, and traffic conditions (n , n [...] Transitivity of preferences implies that, when R Â S and I ≈ S, then R Â I and the curve R ≈ I is located at a lower π. Similarly, when S Â R and I ≈ S, then R Â I and the curve R ≈ I is located at a higher π. Lemma 1 confirms that the decision whether to become informed involves a comparison between I and R for the least risk-averse drivers, and a comparison between I and S for the most risk-aver [...] As explained in Section 2, the Costly information equilibrium approaches the equilibrium with No information and the compensating variation for Costly information (gross of the cost) approaches the compensating variation for Private information.