Cluster analysis identifies geographic concentrations of firms in a service area. Cluster numbers and sizes are based on information about firms gathered at a subregional geographic level, usually county-level. Cluster analysis can be based on the numbers of establishments or on the concentrations of core industries; the latter requires some preliminary analysis before the cluster analysis can be performed. One useful measure derived from this approach is cluster density, which is calculated by dividing the number of establishments in each county by the number of square miles. Another useful measure is time/travel distance.

This information is often combined with county coordinate data and displayed in map form. The numbers and boundaries of clusters may be determined through simple methods such as eye-balling these maps and delineating regions around metropolitan statistical areas (MSA). More systematic statistical analyses based on Euclidean distances may also be used. The distinctive characteristics of each cluster are presented so that service strategies may be tailored accordingly.

Use

Nearly all manufacturing assistance programs use some sort of cluster analysis in their planning stages to define their service regions. Generally, these analyses lean more toward the "eye-balling" end of the scale and away from more sophisticated, rigorous analyses.

Cluster analysis addresses such questions as "into how many service regions should I divide my state?" "Which counties should go into which regions?" "Where should I locate the field office to serve the region efficiently and effectively?" Once service regions have been established, additional needs assessment analyses are often conducted at a smaller geographic level to depict the characteristics and needs of firms in a particular service region.

Case Example

The Cleveland Advanced Manufacturing Program's Great Lakes Manufacturing Technology Center (CAMP/GLMTC) used cluster analysis to extend its service delivery beyond greater Cleveland. (See Figure 2.) Researchers conducted disjoint cluster analysis to aggregate manufacturing establishments into geographic clusters; they incorporated density, manufacturing share and driving time in their calculations. This information was paired with county coordinate data for mapping. Twelve clusters were identified, including some MSAs (Cleveland, Akron and Toledo). To date, seven of these clusters have been organized into satellite areas to be served primarily by local providers (for example, consultants, community colleges).

CLUSTER: CLEVELAND/LORAIN/NORTHEASTERN OHIO

County          Number of         Number of        Primary SIC's and  
                Manufacturing     Manufacturing    Percentages for 60% or 
                Establishments    Employees        more of all Cluster
                                                   Establishments
-------------------------------------------------------------------------
Cuyahoga        4,647             373,783          35-Ind. mach--29%     
Lake            889               33,768           34-Fab. metal--18%    
Lorain          561               56,793           27-Printing--9%       
Ashtabula       198               11,099           30-Rubber--6%         
Geauga          193               11,254                                 
Erie            149               11,608                                 
Huron           126               11,047                                 


Totals          6,763             509,352          62%                   

NUMBER OF MANUFACTURERS WITHIN A CLUSTER

--------------------------------------------
Cleveland/Lorain/Northeastern Ohio     6,763            
Dayton/Cincinnati/Northern Kentucky    5,638            
Southwestern Pennsylvania              3,686
Columbus/Mansfield                     2,651
Toledo/Lima                            2,764            
Akron and surrounding areas            2,173            
Northwestern Pennsylvania              1,899            
West Virginia                          1,811            
Canton and surrounding areas           1,450            
Youngstown/Mahoning Valley             1,422            
Northeastern Indiana                   1,273            
Southeastern Ohio                      718              

Total                                  32,107           

Industry sector share analysis of the firms in each cluster was performed, highlighting those primary SICs that accounted for 60 percent or more of all cluster establishments. Industrial machinery/equipment and fabricated metal products had a strong presence in most clusters; furniture and printing and publishing were the leading industries in two clusters.

Strengths

Cluster analysis lets a manufacturing assistance program serve its area more efficiently. More firms can be serviced with less travel time, and service delivery can be tailored based on knowledge about the characteristics of the firms in the cluster.

Weaknesses

1. Cluster analysis does not indicate whether firms in a particular cluster are likely to need, be ready for or desire assistance services.
2. As with the other approaches described, the input data can be dated and inaccurate. For example, to protect the anonymity of firms in counties with very small business bases, the Census Bureau's County Business Patterns reports cite results in ranges. The midpoint of the range can be used to estimate the numbers of employees and firms in a county, but these are only approximations.

For More Information

About the case example:

1. Fogarty, Michael S., Stephen J. Gage, and Jar-Chi Lee. "Expanding the MTC Program: Economic and Design Considerations." Paper presented at the Technology Transfer Conference, Ann Arbor, MI. June 28-29, 1993.

   Stephen J. Gage, President
   CAMP/GMTC
   4600 Prospect Avenue
   Cleveland, OH 44103
   216-432-5300

Back to Regional Needs Assessment Guide home page