Association for Public Policy Analysis & Management

First, we construct a set of spatial weight variables to define the proximity of REO to NSP properties. Assume there is a market with two NSP properties j and k. For each REO property i in month t we define

S_it^d = 1/min(d_ij,d_ik)

S_it^N = ((1/d_ij)*N_j + (1/d_ik)*N_k) / (1/d_ij + 1/d_ik)

S_it^W = ((1/d_ij)*W_j + (1/d_ik)*W_k) / (1/d_ij + 1/d_ik)

where min(d_ij,d_ik) identifies the nearest NSP property to REO property i by distance in miles; N_j and N_k are the number of NSP properties within a 2-mile radius of NSP property j and k, respectively; and W_j and W_k are measures (not specified here) of the concentration of NSP properties within a 2-mile radius of NSP property j and k, respectively.

To examine how the allocation of NSP funds affects the overall stock of REO properties in an area, a logistic regression is estimated for

Y_it = β₁S^d_it + β₂S_it^N+β₃S_it^W + β₅NSP2_t + β₇S_it^dNSP2_t + β₈X_b+ β₉UR_ct + α_st+u_it

where Y_it is estimated for both listings and sales, which, in the first case, gives the likelihood of a bank listing an REO property for sale and, in the second case, gives the likelihood that the listed REO property is sold. The interaction terms, S_it^dNSP2_t , is the variable of particular interest, which we would expect to be positive if NSP effectively stabilizes an area.

Preliminary results, only for Cuyahoga County, show that banks are more likely to list and sell their REO properties as a result of NSP2 investment. Additional analysis provides some evidence that investor purchases increased and bank liquidity costs decreased—however, these coefficients are not statistically significant.