the sale is part of a replacement exchange. The coefficient on RELQ REPL also is

positive and significant in 13 out of the 15 markets. As discussed, this coefficient

represents the combined price effect of relinquished and replacement exchange

motivation. The magnitude of this coefficient tends to be larger than the coefficient on

replacement exchanges.

 The coefficient on the variable indicating that the buyer is out-of-state,

BUYEROUT, is statistically and economically significant in just two markets Phoenix

and Tucson. These are also markets of specific interest when analyzing the price

premiums associated with out-of-state buyers. In Phoenix 59 percent of sales were to out-

of-state buyers, while in Tucson their share was 44 percent.

 With condominium conversions there are only two markets of potential interest -

San Diego and Boston. I find that the coefficient on the variable indicating a purchase by

a condo converter, CONDOCONV, is economically and statistically significant in San

Diego. The coefficient is also significant in New York City and Oakland.

 The coefficient on the variable indicating a portfolio sale, PORTSALE, is

statistically significant in 3 out of the 5 markets of interest Chicago, New York City

and San Francisco. However, contrary to our intuition the coefficient on PORTSALE in

San Francisco is negative, rather than positive.

 The estimated year dummies, with 1999 as the omitted year, are generally positive

and significant. Moreover, the magnitude and significance of the time dummy

coefficients reveal substantial price appreciation over the seven year study period.

Finally, although submarket dummy variables are not reported in the regression tables,

the model fit is improved significantly by the use of the submarket location controls.