A shift-share analysis of a region's employment growth (or decline) breaks down the change in employment that is attributable to three different sources of growth. The three sources of growth (or decline) in employment are:

  • change in employment due to the growth of the overall economy of the state or nation (the growth of Michigan employment was used in our analysis);
  • change in employment due to the growth of the overall industry in the state relative to the growth of the overall economy of the state (also referred to as "industrial mix"); and
  • change in employment due to characteristics of the region or county (also referred to as "competitive share").[*]

Mathematically, this relationship is expressed by the identity:

click to enlarge

The second term on the right side of the equation is often referred to as the "industrial mix," while the final term on the right side of the equation is often referred to as the local area's "competitive share." This language is adopted in Graphic 16.

The impact of MEGA credits cannot be directly measured from this shift-share analysis. However, MEGA credits are meant to have a positive influence on an area's economy independent of overall trends in state employment or in the shift of the industrial mix in manufacturing. Consequently, MEGA credits in recent years should be strongly positively correlated with a local area's manufacturing employment growth. That can be tested statistically.

From 1995 through 2000, there were 107 MEGA deals, and the life of the credits ranged from five years to 20 years into the future. For the purpose of this analysis, only deals that resulted in the start of operations prior to 2001 were included. Hence, if a deal was approved by MEGA in 2000, but operations did not begin until 2003, that case was excluded from the study.

The shift-share analysis was performed for each of Michigan's counties from 2001 through 2007 using the state of Michigan as the region of comparison. The results of that analysis are presented below for manufacturing employment by county. The "Michigan's growth" column represents the change in employment or "jobs" from 2001 to 2007 attributable to Michigan's overall economy. The "industrial mix" column represents the change in jobs attributable to the growth (or decline, in this case) of statewide manufacturing. Finally, the "competitive share" column represents the change in the county's manufacturing jobs due to characteristics of that particular county.

Graphic 16: Shift-Share Analysis of Manufacturing Employment Changes From 2001 to 2007 With the State of Michigan as the Region of Comparison

Graphic 16: Shift-Share Analysis of Manufacturing Employment Changes From 2001 to 2007 With the State of Michigan as the Region of Comparison - click to enlarge

Graphic 16: Shift-Share Analysis of Manufacturing Employment Changes From 2001 to 2007 With the State of Michigan as the Region of Comparison - click to enlarge

Graphic 16: Shift-Share Analysis of Manufacturing Employment Changes From 2001 to 2007 With the State of Michigan as the Region of Comparison - click to enlarge

Source: U.S. Bureau of Labor Statistics, Michigan Economic Growth Authority, and calculations by Michael Hicks. The MEGA tax credit data was filtered to separate manufacturing facilities from nonmanufacturing facilities. The MEDC included Standard Industrial Classification codes for most credits, and only the manufacturing SIC codes were included. There were seven instances where SIC codes were absent. In these cases, the credits were sorted by scanning the credit's Briefing Memo for a description of the facility. Only facilities primarily engaged in the creation of goods were included, meaning that some companies with SIC manufacturing codes were excluded.

This shift-share analysis is useful in further testing the effect MEGA grants have actually had on manufacturing employment in Michigan counties. The first test performed by Hicks determined whether a MEGA grant provided to a business in each of the 83 Michigan counties prior to 2001 was associated with subsequent employment change attributable to county-specific competitive-share job growth in manufacturing. In this statistical test, Hicks found that the MEGA grant was correlated with a reduction in manufacturing employment. Every $1 million in MEGA credits awarded was associated with a reduction of 95 manufacturing jobs. The decline in manufacturing employment related to MEGA tax credits was strongly statistically significant. A t-test also indicated it was extremely probable that the relationship between MEGA credits and changes in the county's manufacturing employment is negative.

A scatter plot of this relationship is telling. In Graphic 17, county-level changes in manufacturing employment attributable to county-specific conditions appear on the vertical axis, while the total MEGA credits from 1995 through 2000 appear on the horizontal axis. There is no clear linkage between the two, and the fact that the statistical relationship is negative strongly discounts the argument that the MEGA program is improving employment in manufacturing.

Graphic 17: Michigan Counties' 1995-2000 MEGA Manufacturing Tax Credits and 2001-2007 County "Competitive Share" of Manufacturing Job Changes Against the State as a Region

Graphic 17: Michigan Counties' 1995-2000 MEGA Manufacturing Tax Credits and 2001-2007 County

Source: Calculations by Michael Hicks.

Statistical tests point to a strong negative relationship between MEGA and manufacturing employment changes attributable to county specific competitiveness. While this model does not assert causation - i.e., the model does not imply that MEGA grants are destroying manufacturing employment in the counties the authority targets - it is clear that MEGA's efforts to promote manufacturing employment have not been successful. This relationship is surprisingly strong in statistical terms and suggests MEGA has no positive effects on manufacturing job growth in Michigan (see Graphic 18).

Graphic 18: Results of Statistical Tests

Graphic 18: Results of Statistical Tests - click to enlarge

Source: Calculations by Michael Hicks.

To check that the county competitive shares were representative of changes unique to the counties themselves, rather than an accidental product of the choice of the state of Michigan as the region of comparison, we recalculated the shift-share analysis using the United States as the region.

The results of that analysis appear in Graphic 19.

Graphic 19: Shift-Share Analysis of Manufacturing Employment Changes From 2001 to 2007 With the Nation as the Region of Comparison

Graphic 19: Shift-Share Analysis of Manufacturing Employment Changes From 2001 to 2007 With the Nation as the Region of Comparison - click to enlarge

Graphic 19: Shift-Share Analysis of Manufacturing Employment Changes From 2001 to 2007 With the Nation as the Region of Comparison - click to enlarge

Graphic 19: Shift-Share Analysis of Manufacturing Employment Changes From 2001 to 2007 With the Nation as the Region of Comparison - click to enlarge

Source: U.S. Bureau of Labor Statistics, Michigan Economic Growth Authority, and calculations by Michael Hicks.

Graphic 20 shows a scatter plot of the relationship between the county's "competitive share" from 2001 through 2007 with the nation as the region and the claimed MEGA manufacturing credits in the years 1995 through 2000. Note that the negative relationship is roughly the same, indicating that the earlier analysis of the county competitive share was not an artificial result of using the state of Michigan as the region for comparison.

Graphic 20: Michigan Counties' 1995-2000 MEGA Manufacturing Tax Credits and 2001-2007 County "Competitive Share" of Manufacturing Job Changes Against the Nation as a Region

Graphic 20: Michigan Counties' 1995-2000 MEGA Manufacturing Tax Credits and 2001-2007 County

Source: Calculations by Michael Hicks.


[*] For a concrete example, suppose manufacturing employment in County X rose from 1,000 jobs to 1,050 jobs over one year — an increase of 5 percent. Is this rate of growth in manufacturing "good"? It depends on how fast the larger economy was growing and how fast the larger manufacturing sector was growing. A 5 percent increase in jobs might not be very good if manufacturing in the state increased by 8 percent over that year. This is where a shift-share analysis can help shed some light. Assume that annual state employment growth was 2 percent and that annual state manufacturing employment growth was 3 percent. A shift-share analysis of County X's change in manufacturing employment would say that of the 50 new local manufacturing jobs:

  • Overall growth in state employment was responsible for 20 of the jobs (2 percent divided by 5 percent — 40 percent — of the 50 jobs);
  • Relative growth in state manufacturing employment was responsible for 10 of the jobs (3 percent minus 2 percent — 1 percent — divided by 5 percent, yielding 20 percent of the 50 jobs); and
  • Characteristics specific to County X's manufacturing environment were responsible for the remaining 20 jobs, which are 40 percent of the 50-job total.

In this case, the 50-job increase in County X's manufacturing sector was pretty good, since 20 of the jobs, or 40 percent, are due to purely local trends that were independent of statewide employment and manufacturing growth.