We explore how trading networks among commercial real estate brokerage firms are formed, and the effects of network structure on centrality, assortativity and power in the six largest commercial real estate markets in the United State. Fitting the dynamic network formation model developed by Jackson and Rogers American Economic Review, 97(3), 890-915, (2007) to U.S. commercial real estate transaction data (2003–2016), we calculate the importance of network search relative to random search for commercial real estate brokerage firms. We observe that in general approximately 67% of trades of commercial properties are made through network searches. Network search is more important during the global financial crisis (2008–2010). The trading networks also show geographic heterogeneity as network search is most important in Chicago (79.7%), but relatively less important in Boston and Los Angeles (56.2% and 58.2%). Finally, brokerage firms with many trading partners are more likely to trade with those with fewer partners (negative assortativity) creating conditions for power asymmetry between firms.