Look around. Anyone can acquire the necessary data to do an economic impact analysis. It doesn’t matter at what level. We can buy RIMS-II coefficients from the U.S. Bureau of Economic Analysis. We can purchase subscriptions to utilize models from IMPLAN. We can lay out retail or services estimates on the basis of regional surplus-leakage analysis or gravity-model analysis. We can rely on rule-of-thumb multipliers from most anywhere. Some of these are valid. Some are not. We can, however, see all of them in action if we look at enough “Economic Impact” reports.
Even sticking to the Input-Output based analyses (RIMS-II or IMPLAN), simply being able to buy access to the data does not assure a good analysis. Getting a good analysis is very dependent on the modeler understanding what the models can and cannot do.
The Input-Output (I-O) Model
While the details of a working I-O model can be quite complex, conceptually, an I-O model is quite simple. An I-O model is basically a matrix of economic sectors. Sectors along one axis represent industrial inputs or suppliers to the industries on the other axis, which represent industrial users or demanders. Suppliers and demanders are connected by an interlocking set of mathematical relationships specifying how much of each input is required to make a unit of any output. When it is decided how much final output an industry will produce or how much labor an industry will employ, the model specifies how much of all necessary inputs are required and how those inputs are sourced from other industries.
It starts out looking like the large system of mileage charts (similar to those that you find in the back of a road atlas). Unlike the numbers in a mileage chart, however, each of the cells in an I-O model contains part of a system of production functions that is linked mathematically to all of the other cells in the model. The values of final goods produced or labor employed can be changed for any of the industries and these coefficients allow the modelled economy (the matrix) to be rebalanced, showing how the initial change affects all of the industries that supply inputs to or demand outputs from the industry altered.
This is the basis of the type of I-O-based impact analysis commonly used to estimate the effect of a given economic change. It is important to understand, however, some basic and very important constraints upon the model.
The Model is Static
While we can change the value of output or employment for an individual industry, the model itself is static and unchanging. The production coefficients in every cell remain unchanged. The changes we made to an individual industry simply changes the scale of overall output and employment. They do not change any of the production relationships in the model.
The Model is Linear
Fixed coefficients (production relationships) make the model linear. If we change the final output of industry “A” by $200, we will get precisely twice the impact we will get if we change the output by $100. If we change the output by $100 Trillion, we will get one trillion times the impact we got with a $100 change. Because all of the production coefficients are fixed, any individual output or employment shock can be scaled up or down.
The Model is Limitless
This means the model is limitless. No matter how big a change we submit, the fixed production coefficients will provide us with an economy scaled precisely to that change. We could take Franklin County, Iowa, for example. Franklin County has a population of just under 10,000 and a civilian labor force of about 5,500. An I-O model, however, would allow us to increase employment in an area industry by 20,000 and scale the economy to match that change. It doesn’t matter to the model that the initial change is somewhat ridiculous.
Similarly, Franklin County is heavily farmed. Nearly all the arable land within the county is cropped in corn and soybeans. There is no real excess capacity with respect to arable land, so production is pretty much fixed. The model, however, would allow us to shock the agricultural system in Franklin County by increasing the value of both corn output and soybean output by 20 times. It would dutifully scale the local economy to accommodate that change even if the change is impossible in the real world.
The static model will allow us to model ourselves into absurdity. It is important to understand the environment and economy in which we are modeling. It is important to define the area modeled such that its economy is somewhat commensurate with the change being investigated. The model knows no constraints of scale, so the modeler must be able to recognize, acknowledge, and accommodate them.
Prices in the Model Cannot Change
Constant production relationships require constant prices. If prices change, the value of production and input costs change. Because the model and all of its calculations are dollar-denominated, changing prices would violate the assumptions behind the structure of production relationships.
Prices cannot change. No matter how severely we shock the environment, the model assumes that the economy can provide limitless resources at a constant price. The model also assumes that the citizens in the modeled environment are ready and capable of purchasing limitless amounts of output at constant prices.
This is never the case in reality, but for small enough shocks it can be reasonably close. Going back to freshman economics, the I-O model is a Micro economic model. In Micro it is assumed that every participant in the economy is too small to affect the economy as a whole. As a result, prices are assumed to be fixed.
Conversely, in the context of the Macro economy, prices change as more or less resources are demanded. These changes cause people to adjust their purchases and producers to adjust their inputs in order to maximize their purchasing power.
The I-O model, however, cannot accommodate price changes and the resulting adjustments. This means that the model overestimates impacts for any event, shock, or change modeled. It only overestimates by a little if the shock or change is small relative to the economy (say, adding a pool hall in Des Moines, Iowa). The overestimation grows, however, as the size of the shock or change grows larger relative to the area economy (removing the entire farm and construction machinery manufacturing sector from Pella, Iowa). The larger the shock is relative to the economy the larger the I-O model overestimate of the economic impact will be.
In all cases, if the expected result of an event is stated to be a change in related prices, the modeler needs to be very circumspect in evaluating impact model results. Building an ethanol facility in Iowa, for example, is nearly always promoted as a means to raise the price of the surrounding area’s corn production. Because the surrounding area’s corn production is relatively fixed, this violates two of the basic constraints of the model. First, constant prices cannot be assumed. Second, the area cannot be assumed capable of increasing the necessary inputs to support the event within a fixed area subject to fixed production relationships.
The modeler must be able to place the event within an economy that can reasonably handle the resulting impact in a fixed-price environment. The modeler must be cognizant of where the proposed event violates the underlying constraints of the model. In all cases, these issues must be dealt with transparently in the presentation of model results.
In-area Substitutions
The model cannot distinguish between new economic activity (economic impacts) and changes in existing economic activity (substitution). The modeler must be sufficiently aware of the local economy and the event being analyzed to make these distinctions.
For example, Joe buys a factory from Bill. The factory doesn’t alter its operations or output. It just changes hands. There is no economic impact. We just substituted Joe for Bill.
On the other hand, Bob opens a new grocery store in a town of 8,000 that is already served by two existing grocery stores. An economic impact model evaluates the initial investment involved in opening the store as well as the annual impact of store operations. Within a short time, however, the existing stores begin to struggle from decreased sales, and one of them closes. In the short run, there appeared to be economic impact. In the long run, it revealed itself as a substitution.
It can get murky. If the local opera house draws sixty percent (60%) of its audience from the local area and forty percent (40%) from outside the area, is 60% of its modeled impact merely substitution, because local residents would have spent their money at some other local venue? Or is it reasonable to assume that if the local opera house did not exist, local patrons would travel to nonlocal opera houses? After all, it is reasonable to assume that nonlocal opera aficionados travel into the local area to enjoy the existing opera house.
Similarly, suppose Bill was going to close the factory in the first example. Then is Joe’ purchase and continuance equivalent to opening a factory after the previous one closed? Quite often, “Jobs saved,” is presented as a justifiable impact, even if the property transfer appears to be substitution. Similarly, if our local opera patrons would have traveled out of town, “Entertainment expenditures saved,” might be presented as a justifiable impact.
There are no hard-and-fast rules in these situations. It is the function of the modeler to define the event with respect to the origin of the effects modeled and to present modeled results in a way that make the implications of those origins clear.
A Basic Conundrum of the I-O Model and Economic Impact Analysis
The I-O model lives in somewhat uncomfortable territory. It is a Microeconomic model, so the players are assumed to have no effect on size and price relationships. The goal of economic impact analysis, however, is nearly always to show that an event will have significant effects on the overall area economy. The larger these effects, the more likely it is that we are violating the fixed-price assumption of the model and overestimating the resulting impact.
One way to mitigate this size problem is to define a larger area. Doing so reduces the relative size of the modeled event with respect to the overall area economy. Increasing the size of the modeled area also increases the size of the resulting economic impact, because more expenditures happen within the area before the shock begins to dissipate as transactions go beyond the area.
Increasing the size of the modeled area, however, also exacerbates the issue of in-area substitution. As the area expands, the chance that the modeled event simply replaces existing activity grows.
This basic conundrum is why it is imperative that the economic impact modeler thoroughly understand the modeled event, the modeled area, the modeled economy, and their interrelationships.
Good Luck!
Regional Strategic, Ltd. is always available to assist your community, business, or development group with economic impact modeling and other development needs. As time goes on, we will be posting additional thoughts and information regarding the types of services we are engaged in.