Table 14. Differences in Mean Prices of Control Sample and Identified
 for Office Properties
 Mean Value Standard
 Office Observations a e Err
 of Sales Price Error
 Control Group 5,780 3,472,731 96,485

 EXREPL 856 3,842,241 202,277

 EXRELQ 454 3,027,999 261,805

 RELQ REPL 240 2,824,935 201,544

 BUYEROUT 1,044 16,200,000 478,294

 SALELEASEBACK 239 3,768,159 476,300

 PORTSALE 79 12,500,000 1,585,380


Groups of Interest


 T-test Value



 -1.41

 1.26

 1.36

 -41.92

 -0.61

 -10.69


 I perform differences of means tests to examine differences in selling prices for

variables of interest at the aggregate level. Table 14 presents the results. The null

hypothesis is that the average observed price of the properties in the comparison group,

which contains sales that have no conditions and are typically motivated, is equal to the

average price of the sample composed of replacement property sales (EXREPL group).

Similar hypotheses are formed with respect to the relinquished properties (EXRELQ), the

sample containing sales that are part of two separate exchanges (RELQ REPL),

purchases by out-of-state buyers (BUYEROUT), sale-leasebacks (SALELEASEBACK)

and portfolio sales (PORTSALE). Table 14 shows that replacement office exchanges are

associated, on average, with a 10.6% price premium, but the price difference is not

statistically significant. Relinquished exchanges and properties that are part of two

different exchanges have lower average prices than the control sample, but the

differences in the mean prices are not significant. Out-of-state buyer purchases and

portfolio sales are associated with very high prices, which are also statistically