NYC-WTC_000187725

NYC 9/11 Public Portal Document

43

Table 2.2 Aggregation of Census Travel Modes into NYREG Modes

1. Auto Drive Alone

— Shared Ride 2 person SR ( Car pool ) 3 person SR 4 person SR I— Taxi H 2. Bus ALL 3. Rail 4. Subway/Elevated 5. Walk + — Walk — Bicycle — Motor Cycle -- Other ■ — Worked at Home

The model's coefficients were calibrated by means of a sequential procedure. The first step is to estimate a regression which is the log form of the shopping intensity function:

In(Sj) =ln(B) +pln(rj) (16)

The shopping density is observed as Sj^ = /Fi, where STjo^j^ is the matrix of shopping trips from each^j to each shopping zone 1 by each mode n. This estimation gives p=0.49866. The slope coefficient is then adjusted for each 1 so that the observed shopping density in each 1 is perfectly predicted. This defines the coefficients Bj^ = Sj^/(rj^*)P. In the second step, the rents of housing are estimated. Because census data do not give rents by housing type, sales data obtained from the tax assessor were used. Then, the average sales prices for each (j,k) were used to estimate the housing rent in that zone. For this, it was assumed that rents, Rjj^, are related to housing values, Wjj^, as follows: Rjj^ = aj^(Wj)^)^. The value of b was set somewhere between zero and one, and each aj^ was calibrated by minimizing a least squares function so that "on average" households spend 25% of their disposable income after commuting on housing. This assumption, in turn, means that the utility coefficient B = 0.25. In doing this, it is also assumed

CITY_21MC100 Jd 066622 7 4

NYC-WTC_000187774