REMI Model
There are several types of models typically used to calculate economic impacts. Input-output models use financial flow data generated from businesses’ accounting data, and spending patterns for households of particular income levels, to describe the economic linkages that exist within a regional economy. These models begin with US government-generated county level data on business purchases and receipts in order to model the inputs that are used from across the many sectors of the economy in the production of particular goods and services. The level of geographic and commodity detail can vary from production of printing ink, to storage batteries, to banking services in a geographic area as small as a zip code or as large as the national economy. The most commonly reported and useful level of detail is county-level geography at the 1 to 6 digit North American Industry Classification System (NAICS) level of commodity detail, whereas the previous 2004 report was conducted with the 1, 2, and 3 Standard Industrial Classification (SIC) code level of commodity detail. Examples of these models include the RIMSII modeling system from the U.S. Department of Commerce and the IMPLAN modeling system from MIG, Inc. of St. Paul, MN.

Econometric simulation models combine the sector detail and geography detail of input/output models but provide for functioning economic linkages between sectors and regions over time. The current study uses REMI PI+, Version 1.1.9 (Regional Economic Models, Inc.), in a 67 region (one for each county) structural econometric model of the state of Florida to calculate the impact. It incorporates the basic input/output linkages, but also uses econometrically estimated county-specific parameters (for example, interregional migration in response to changes in economic opportunities) in generating impact results. Because of these between-sector linkages, the model incorporates general equilibrium tendencies as the economy responds to shocks over time. That is, changes in spending in a region affect not just conditions in that market, but also in other markets within the region (economists term this a “general equilibrium”) and outside the region (via trade and also via migration in response to changes in economic opportunities). This is in contrast to traditional input-output models that are both static (all effects are assumed to occur simultaneously, so there is no adjustment path over time) and partial equilibrium (e.g., changes in employment do not change wage rates) in nature. This describes the phenomenon whereby, for example, a new financial services back office call center opens in a county, and bank managers throughout the county find they have to give staff a raise in order to keep them from leaving to take a job at the new call center. A traditional input-output model description of the economic impact would have held everything else fixed (including bank wages across the county) and simply documented the employment and job creation effects resulting directly at the new call center and indirectly via businesses in its supply chain, as well as household spending induced by the new income flows.

A simulation model such as REMI captures not only the spending effects flowing from the call center and its local suppliers and employees and owners, but also the spillover effects into other markets as wages and prices change due to competition for the same employees and other resources. These are the general equilibrium (equilibrium across all markets simultaneously) tendencies of the model. It also simulates the adjustment path over time of these market responses, using historical parameters estimated specifically for that county (the dynamic component). A rule of thumb is that the smaller the spending change being considered, the more appropriate it is to use the traditional input/output model. However, the general equilibrium and dynamic characteristics of an economic simulation model are particularly important when considering “large” changes. The presence or absence of over $668 million in government spending on public library systems in Florida is a “large” change, because spending of this magnitude is likely to have spillover impacts in other markets not directly in the public library related supply chain.

One other benefit of using an economic simulation model is particularly important when considering large spending flows. In an input-output model, impacts are usually measured as gross impacts, or additions to the area’s economy without consideration of the extent to which, for example, a project’s use of labor force may make labor more expensive to other businesses, or require additional infrastructure investment. The use of REMI attenuates this problem and so comes closer to an estimate of net, rather than gross, economic impacts because of the feedback effects present in this simulation model.