untransformed prices. I also divide square footage of improvement and land square

footage by 1000 to keep size ranges consistent.

 Following Weirick and Ingram (1990), I use a semi-log model with quadratic

transforms for square footage in thousands (SQFT), and land square footage in thousands

(LANDSQFT). With this semi-log form, unit price per unit change in the characteristic is

given by simply multiplying the estimated coefficient by the observed selling price.

 To quantify the effect of Section 1031 exchanges, condominium conversions, sale-

leasebacks, portfolio sales, and out-of-state buyers on sale prices, I use a stepwise

estimation technique and estimate the following model separately for each of the

identified 15 markets in the apartment sales sample


LNPRIC, = a, +a1EXREPL+ a2EXRELQ+ aRELQ_ REPL+ aA GE+ aAGE2 + aSQFT
+ a7SQFT2 + a8LANDSQFT+ a9LANDSQF72 + PARKING+ a, FLOORS+ a, UNITS

+ f, CONDITION, + a13BUYEROUT+ a F .0JR+ a5 SUBSIDIZED+ a1l6CONDO
 2005 P<43
+aC17CONDOCONV+a ,PORTSALE+ z,zYRn,+ 7, \\i)(U \ (17)
 n=2000 s=2

 I use a similar model to estimate the effect of tax-deferred exchanges,

condominium conversions, sale-leasebacks, portfolio sales, and out-of-state buyers on

property sale prices in offices and retail transactions

LNPRICE = ao +aEXREPL+ aEXRELQ+a3RELQ_ REPL+a4AGE+a AGE2 + 6SQFT

+ aSQFT2 + a LANDSQFT+ a LANDSQF72 + a,0PARKING+ a FLOORS+ E P, CONDITION
 2005 P<43
+al2BUYEROUT+al3SALELEASEBICK+a14PORTSALE+ XnYR, + YS \\l)( u (18)
 n=2000 s=2

 In the estimation of the hedonic pricing equation, I include a dummy variable

(EXRELQ) that quantifies the extent to which transaction prices are lower (higher), all

else equal, if the seller of the property is a taxpayer initiating the "downleg" portion of a