3. Seller's relinquished and buyer's replacement property in two separate transactions
4. Direct exchange (swap)
5. Seller's relinquished property in reverse exchange
6. Buyer's replacement property in reverse exchange
7. Reverse exchange (type not confirmed)
8. Exchange into Tenants-in-Common
9. Other tax-deferred exchange exchange which cannot be categorized in any of the
 above types or where CoStar was unable to confirm its type.

 To assure reliability of the data, CoStar requires agents to physically inspect the

site and record a variety of property characteristics and transaction details. I therefore

exclude sales not confirmed by CoStar. In addition, I exclude all transactions with

recorded sales price below $250,000. CoStar covers comprehensively only transactions

that are above this threshold, although in some cases brokers do report smaller

transactions. For the sake of consistency such smaller transactions are eliminated from

the sample. The final sample has 124,830 transactions which facilitates a comprehensive

empirical investigation of the Section 1031 exchange market, as well as other atypical

motivations, not possible with previous datasets. Of the 124,830 usable sales transactions,

23,989 (or 19 percent) involved the use of a Section 1031 exchange.

 Table 3 summarizes the number of transactions by year and property type. The year

1999 contains fewer transactions than year 2000. Also in 2005 there is only data for half

of the year through June 2005.

 A substantial increase in the use of exchanges in commercial real estate markets

has been discussed in the popular press (e.g., McLinden, 2004). However, the data

necessary to support a comprehensive analysis has not been available. Table 3 breaks

down the sample by exchange and non-exchange transactions. The table reveals that

exchanges as percentage of all sales are very stable over the 1999-2005 sample period.

 For example, between 1999 and 2000, the total number of commercial real estate