Economic Base Analysis:  Introduction

One of the most common questions the council and community will ask a city staff is “What will happen if we get (lose) this industry?”  While it’s always good to gain industry (and bad to lose)—at least from an economic point of view—some industries will have a larger impact than others. 

It would be nice if there were a single, constant answer for every industry and every place in the country.  But that would be too easy.  In fact, the answer will be different for every industry in almost any place in the country.  The impact of any one industry depends in part on the role that industry plays in the national economy, in part on the role that industry plays in the local region, and in part on the local region’s share of that industry. 

Enter “Economic Base Analysis.”  The overall idea is that every economic activity in a location (city, county, region, state, nation) has some impact on all the other economic activities (call them “industries”—keeping in mind that households and one-person newsstands are considered as “industries”).  It is possible to construct a “chart of accounts” which spreads the impact of any industrial grouping (any “row”) across all of the other industrial groupings, including itself (the “columns”), which can be added across the row to get the total economic impact of that industry in the region.  A complete economic base matrix would have a row for every industrial grouping in the region.  The “rows” need not be the identical with the columns—it is possible to run a more detailed analysis of impacts (“columns”) against a different classification of industries tailored to the region’s major activities, for example (so long as all industries are accounted for somewhere in each set of categories).  There are about 500 major industrial groupings (the NAICS Manual runs to 1400 pages; a listing of the 6-digit codes is available at ).  A commonly used model (“RIMS II”) combines these into 38 categories.  Even using those few categories, the amount of data required (and generated) is immense (a 38 x 38 table has 1,444 cells).

The table of regional industrial activities and their impacts on all other industrial activities is called an “Input/Output Table” (or I/O Table, for short).  Generally, the data in the table are presented as decimal fractions (sometimes less than, sometimes more than 1.00) of the impact of the row on the column.  In other words, for “$1 change in output from industry X” you can expect “x.xx% change in input from industry Y.”  The tables usually show $1 change in output or earnings or, in the case of employment, number of jobs per $1 million of output.  The I/O table is usually created based on national data—in order to get reliable and stable estimates, a large amount of data are needed.

But national trends could easily misrepresent the local situation (for example, in 2005 unemployment in the Bellingham, WA area was running around 7.5%, while the Mankato, MN region was experiencing around 3.5%).  So the national I/O table must be adjusted to the local situation.  Usually, this is done at a scale no smaller than a single county, and preferably for a region of counties.  If the region is too small, an economic base analysis will either be highly unreliable or else irrelevant.

Suppose, for example, that one attempted an economic base analysis for Prairie, MN (pop. 2300).  Most of the residents of Blooming Prairie work in other cities, and the largest employer in the town is Arkema, a chemical manufacturing plant that employes 56 people.  An economic base analysis would show that, in Blooming Prairie, households are the main economic engine (money somehow just “appears” in the community—few people work there), or else it would show that most of the jobs are indirectly tied to the Arkema plant.  But it wouldn’t take a fancy analysis to tell the town that it would have major problems if Arkema were to close!

So, the next step is to scale the national Input/Output Table to an appropriate regional scale.  This is called a “shift/share” analysis.  Different industries are experiencing different “shifts” in their economic activity (some industries, like biotechnology, are growing; others, like furniture manufacturing, are declining).  Different regions have different “shares” of those industries (Minnesota, for example, does not do very much with orange-growing, but it has more than its share of medical technology devices).  So, one multiplies the regional share times the national shift in the industry and adjusts the cells of the Input/Output matrix accordingly. 

The final step is to allocate these effects to “direct effect” and “final demand.”  Direct effect is the change in demand from other industries that are directly attributed to a single industry.  To return to the Blooming Prairie example, if Arkema were to leave the community would lose the 56 jobs, and whatever supplies, materials, services, and resources the company purchased, from other businesses in the community (trucking, printing, water & sewer, etc.).  The businesses supplying those resources would also experience a loss.  The total of these losses are the “direct effect” of Arkema.  But that is not the whole story.  Those 56 employees and those businesses that supplied Arkema also purchased goods and services in the community, and that business would now be lost.  If the impact is large enough, it could lead some of those businesses to cut back or close, which would affect still other businesses, and so on.  It is possible (using matrix multiplication) to work out all of this indirect effects, to arrive at “final demand,” or the total of the direct and indirect effects of a change in any industry.

Fortunately, you do not need to gather all the data and do all the calculations by hand.  The federal Bureau of Economic Accounts has software and data which makes the analysis fairly straightforward.  It is called RIMS II (Regional Input/output Modeling System II).  The handbook for the model is available at , as is ordering information.

For Further Information:




© 1996 A.J.Filipovitch
Revised 11 October 2005