The article considers developing the algorithm of local economic communities’ identification on the municipal union territory. The description of authors’ algorithm is presented. The algorithm was tested on the real object of research. Searching of local entrepreneurs’ communities was conducted in the city with population of 59,000 citizens. The analyses were made according to sampling of 12, 000 transactions between 2 933 organizations during one month in 4 bank branches. There were found 17 closed networks with 102 organization-participants. The largest network of agents consists of 59 organizations-participants. The comparison of the largest network with random graphs (Bernoulli, Multinomial) showed that the structure is formed not randomly