The Number Factory

In general, Lindall says, “you don’t see output multipliers over about 1.3 to 1.8. But employment and labor-income multipliers can bounce around a lot more, depending on production and wage-per-worker” data for a given industry and geographic region. “Multipliers are very [economic]-sector and region specific,” he emphasizes.

As for defining those sectors, Leontief identified just 10 back when he first started working on input-output modeling in the 1920s. Today, Minnesota Implan Group differentiates in its software and data sets between 509 sectors that are recognized by the U.S. Department of Commerce. Each sector has a unique output multiplier, because each industry sector has a different pattern of purchases from firms inside and outside of the regional economy. (The output multiplier is in turn used to calculate income and employment multipliers.)

But arriving at a multiplier is not the end goal of Implan users. They want to use multipliers to generate other numbers—answers to questions such as, How many jobs will this new firm produce? How much will the local economy be affected by this plant closing? What will the effects be of an increase in product demand?

To help clients get answers, Implan offers simple Windows-driven drop-down menus of commands for constructing and running economic input-output models. Users build a model by first choosing a geographic area of study—a state, county, city, or even a zip code. They can also combine several states, counties, et cetera, in a single study. They then select the county-by-county economic-data files that encompass their area of study—data sets constructed and sold by Minnesota Implan Group. Next, they specify the type of multiplier they want to use (output, labor income, or employment) depending on the type of change they’re looking to study.

Based on those user choices, Implan software constructs “social accounts,” which are mathematical representations of economic interactions—the flow of dollars from purchasers to producers within the region. The data in those social accounts will set up the precise equations needed to finally answer those questions users have—about the impact of a new company, a plant closing, or greater product demand—and yield the answers.

Too Easy to Use?

In theory, anyone could construct his own regional input-output model without benefit of Implan software just by surveying local businesses about things like which goods and services they purchase and from whom.

In fact, economists agree that that’s the most accurate way to do input-output analysis. But gathering data this way is too time consuming and expensive to be feasible in most cases. Instead, economists typically construct I-O models using aggregated production, employment, and trade data from local, regional, and national sources, such as the U.S. Census Bureau’s annual County Business Patterns report, and the U.S. Bureau of Labor Statistics’ annual report called Covered Employment and Wages.

(It’s noteworthy that a government agency, the U.S. Depart of Commerce’s Bureau of Economic Analysis, is one of Minnesota Implan Group’s few competitors. The bureau sells regional multipliers from its Regional Input-Output Modeling System, RIMS II. However, it doesn’t sell the data sets or software that would make using its multipliers in economic what-if scenarios easy. Users must do their own calculations in a spreadsheet.)

Even when aggregating data from other sources, though, constructing data sets county by county for each year is a complex process that takes about three months, Lindall says—which is why there’s not much competition to do it. Olson constructs data sets for Minnesota Implan Group, while Lindall specializes in software development these days. The company has data sets for the entire United States for the years 1990 through 2006. In addition to gathering enormous amounts of data from government sources, Olson must also estimate some data where they haven’t been reported at the level of detail needed (county-level production data, for instance), or where detail is omitted in government reports to protect the confidentiality of individual companies whose data would be easily recognized due to a sparse population of businesses in the area.