NYC-WTC_000187725

NYC 9/11 Public Portal Document

45

exp (coin (Fj) +K j) Zj (20) STP,(z, (0) = j£exp(uln(F.) ' s

wheie Z^= Q^OjPji^.i---------- ------------- , V 1 ijkmn 9iji„

Hence,

Min^ '^STO^-STP^{z,(ji)]^ (21)

where STOj^ is the observed shopping trips arriving at 1. In the outer stage, K (in equation (4)) is chosen so that (with all = 0) it satisfies the constraint that the error in the predicted shopping trips is within 2% of the sum of observing trips:

-STP^ (K, w) I< 0.02XSTO^ (22)

If the above constraint is satisfied, then the desired z and u have been found. Otherwise, z is adjusted by a Newton-Raphson technique and the procedure goes again to the inner stage for the next iteration, given this new value of z. The calibrated values are: w = 0.483929 and z = $16.52. Next, the values of the shopping dummies, Kj^'s, are set for each 1 in such a way that observed shopping trips to that zone are replicated (one Kl is arbitrary and is set equal to zero). All other Kj^'s are then computed so that . STOj^ = STPj^ for each 1. (23) -coin STO^Zy

In the fifth and final step, the utility coefficients are calibrated. First, set 6^ = 6 for each i. Then, set c = 1. Hence, the problem reduces to estimating 5 and the coefficients of the housing subutility Wj)^. The attributes included in the subutility function are the logarithm of the average floor space per dwelling in j, the percentage of the housing stock built after I960, the percent black and Hispanic households, and dummy variables for Manhattan, Queens, Brooklyn, and the Bronx. The

I

CITY_21MC100_006662276

NYC-WTC_000187776