In previous sections we have briefly introduced the concept of multipliers. Let’s take a minute to explain these and how they are very important to IO models. Although economists talk about many kinds of multipliers, we are only concerned with regional multipliers in this guide.
For the purposes of regional IO, a multiplier is a number showing how changes (jobs, earnings, or sales) in one industry will propagate to other industries in a regional economy. For example, a jobs multiplier of 3 means that a change of 100 jobs in that industry would lead to a total change of 300 jobs (3 x 100 = 300) in the whole economy. Note that this 300 includes the original 100 jobs, meaning the additional change is 200. In the study of IO, the original jobs are called the “direct” effect, while the additional jobs are called “indirect” effects.*
* Strictly speaking, “indirect” effects involve the industries in the supply chain, while “induced” effects involve industry sales stimulated by added consumer spending. For convenience, in this guide we will refer to both kinds as simply “indirect” effects.
Let’s look at a real-life example to see how multipliers work in a region. Suppose a new company appears in a region, and begins making some product, the majority of which is exported outside the region. All these export sales result in “outside” money being brought into the region. This money enables the company to purchase local supplies and pay utilities, local taxes, local employee’s salaries, and so on, which means the total income in the region has increased, and local suppliers have more business which means they are likely to hire employees and spend more on their own inputs. As this new income passes from hand to hand, some is spent in the region and some “leaks out” at each stage. Adding up the amount spent inside the region at each stage of these transactions gives us the total effect of the original “outside income” of the new company. This effect can be expressed in either jobs or earnings. Suppose the new company’s payroll $2 mil-lion in earnings, but the total effect (including the original amount) is $2.8 million in earnings. The multiplier would then be 2.8 / 2 = 1.4.
Let’s take a simple example to illustrate the mechanics of multipliers. Suppose I’m part of the new company described above. For every $1,000 of total company sales, the company uses the money in the following ways:
$500 Purchasing materials and supplies used to produce its goods
$250 Payment of labor earnings in the form of salaries/wages, benefits, and other compensation to employees
$100 Profits (also called “operating surplus”)
The earnings, taxes, and profits all represent the total value that the company has added to its inputs. This represents the wealth it has created and is also called value-added, output, or productivity.
Now let’s look at the “indirect” effects of these sales. Of the $500 spent on the company’s “inputs,” $200 is spent on in-region suppliers. Those suppliers then spend $50 on other local suppliers (as well as $50 on employee compensation, $25 on taxes, and $25 goes to profits), and these other suppliers do the same to their suppliers in an ongoing cycle, until the “leakage” of money from the region has reduced the original amount to zero.
Similarly, there are “induced” effects from the $250 in earnings that goes to the company’s employees (we’ll assume they all live in the region). They spend these earnings on houses, food, utility bills, entertainment, and so on, with always a portion of the money staying in the region and a portion leaking out. These household-serving industries then spend a portion of this new income on their suppliers, employees, and so on, until “leakage” again has reduced the amount remaining in the region to zero.
The multiplier comes from adding up the total amount remaining in the region during each cycle of spending, until it finally disappears through “leaks.” So, in one sense we do “re-count” the same money as it benefits each recipient in turn. But in addition, this money is creating new wealth at every step (just like planting a seed or making an investment), since it allows each recipient to add more value by applying more labor to more materials. In the end, the original amount allows more goods to be produced and more services to be provided than otherwise, and the region as a whole is better off.
A. So where do multipliers come from?
Regional multipliers arise naturally out of regional IO models, so we need to review the process of creating a regional IO model.
First, because of the lack of comprehensive local data, all regional models are created by “regionalizing” national values calculated by the Bureau of Economic Analysis. So models primarily come from (a) the BEA’s national input-output accounts, and (b) regional purchase coefficients (RPCs).
Using the national input-output accounts we can create a table that quantifies, for each major industry, how much of the outputs from other industries is needed in order to produce its own output. There are additional non-industry entities in this table too, like household consumption, profits, and taxes, which are used to complete the balance sheet. Every major model uses these national IO accounts in some way. Before applying it to a specific region, we have to generalize it by converting total national dollar amounts to percentages of total output, as well as account for the region’s actual industry mix. Then regional purchase coefficients (RPCs) are applied to each transaction to translate each industry’s “total inputs” to “only inputs purchased locally.”
RPCs represent the percentage of local demand that is satisfied by local supply. If your local construction industry has a $500 million total demand for ready-mix concrete, and it purchases $450 million of it locally, then that particular RPC would be 90%. If your local agriculture industry has a $500 million demand for tractors, but satisfies none of it from local manufacturers (which is quite common since there are relatively few places where tractors are manufactured), then that particular RPC would be 0%. High RPCs will result in higher multiplier effects since money spent on input requirements is being retained locally. RPCs can be estimated in several ways by looking at each region’s industry mix (how much demand could possibly be supplied locally, and how much of that is actually likely to be supplied locally). While the details are beyond the scope of this document, suffice to say that EMSI uses a variation of the well-known Stevens technique* for estimating RPCs, which has been widely discussed in the academic literature of regional IO for more than 20 years.
* Stevens, B.H., G.I. Treyz, D.J. Ehrlich, and J.R. Bower, 1983. “A New Technique for the Construction of Non-Survey Regional Input-Output Models,” International Regional Science Review, 8(3), 271-286.
This whole accounting system tracks many links in the supply chain. As one dollar goes to one industry, portions of it are passed off to other local industries and another portion “leaks out” of the region completely. Then we look at all the other industries that got a piece of the original dollar and look at how much of those pieces go to other regional industries or leak out. We continue this indefinitely until the portion of the original dollar still remaining in the region approaches zero. At each step, we sum up the amount that stayed in the region during that step. The grand total (plus the original dollar) is the final multiplier. In EMSI’s IO models, we use an earnings multiplier as our foundational multiplier. This dollar-based earnings multiplier can also be converted to jobs or sales by using the jobs-to-earnings and sales-to-earnings ratios in each industry.
B. Types of Multipliers
There are many types of multipliers depending on the model you use and what you are trying to measure. Following are the three main types of multipliers used in EMSI’s regional IO models.
1. Sales Multipliers
Sales multipliers show a change in total sales for one industry will change total sales in other industries. EMSI advises caution when using sales multipliers, since there is some double-counting involved. Instead, we prefer to express impacts in terms of jobs or earnings.
Here’s an example that illustrates the problem with sales multipliers. Suppose a consulting firm lands a $100,000 contract. Deciding that it could do the work more efficiently by using outside help, it does a portion of the work, keeps $20,000, and passes off $80,000 to another consultant to complete the rest of the work. This second consultant keeps $40,000 and passes off the other $40,000 to a third consultant, who completes the work alone. In terms of sales, we have this situation:
This is misleading because we’re double-counting part of the sales at each step. However, in terms of earnings, the situation makes a lot more sense:
In this case, a sales multiplier would artificially inflate the impact, which is why we ad-vise caution when using sales multipliers and recommending using jobs or earnings multipliers instead.
2. Jobs (Employment) Multipliers
A jobs multiplier indicates how important an industry is in regional job creation. A jobs multiplier of 3, for example, would mean that for every new “direct” job in that industry, 2 more jobs would be created in other industries (for a total of 3 jobs). Typically, these additional jobs include many “fractions” of jobs spread over many industries.
Jobs multipliers are easily misinterpreted—jobs multipliers of 17 or higher are sometimes seen—but a high jobs multiplier for a set of one or more industries in an added-jobs scenario does not necessarily mean that attracting businesses in those industries to the region is the best or most viable option for regional economic growth.
Jobs multipliers are primarily tied to the type of industries in the scenario—industries with a high capital/labor ratio typically have a high jobs multiplier, and vice versa. For example, a nuclear power plant might have only 20 workers, but “behind” each of those workers there are millions of dollars of equipment costs and millions of dollars of electricity being generated. Thus, if we bring 20 more nuclear power jobs into a region, that would involve a huge amount of investment flooding into the region (to build another nuclear power plant or increase the size of a current one) and millions of dollars in new sales and profits—which means more jobs.
Some of that money would go to the employees’ salaries, some would go to local construction companies, real estate, janitorial services, etc. The overall jobs multiplier would be impressive—each new job in nuclear power might support 14 other jobs scattered throughout the rest of the economy (i.e., a jobs multiplier of 15). However, the effort it takes to attract 20 jobs in nuclear power (with all the necessary infrastructure) is substantially more than to attract dozens more jobs in an industry with a lower jobs multiplier.
3. Earnings Multipliers
Industry earnings are the total amount of employee compensation paid out by employers in the industry. An earnings multiplier of 1.5 means that for every dollar of compensation entered as a “direct effect” in a new scenario, an additional $1.50 is paid out in wages, salaries, and other compensation throughout your economy. This is important for understanding how a given scenario will affect not the number of jobs in your region, but the income-quality of those jobs. A scenario whose ripple effect brought two dozen lawyers and accountants into your region would have a much higher earnings multiplier than if that scenario brought the same number of indirect jobs mostly in food services and hotels.
Note that there is a tendency for industries with low earnings multipliers to have high jobs multipliers, and vice versa. In other words, it is a basic rule-of-thumb tradeoff that industries tend to either create a lot of jobs with lower than average earnings, or fewer jobs with higher than average earnings.
C. Can multipliers get out of date?
Yes, multipliers can get out of date, which is why you should use a regional IO model that uses the most recent data possible. Multipliers depend on the underlying IO model, which is always constructed using annual data for a single “base year.” Regional models in turn depend on the national input-output accounts and detailed information about a specific region’s industry mix, both of which change over time. The national accounts are released in detailed (“benchmark”) format every five years, along with a low-detail annual version. Regional data depend primarily on the BEA’s State and Local Personal Income reports, the BLS’s Quarterly Census of Employment and Wages, and the Census Bureau’s County Business Patterns. These data sources have lots of industry and geographic detail, but they also have a lag time of 6 months to 2 years. More current data sources (such as the Current Employment Survey) have a lag of only about 1 month, but have significantly less detail.
To produce our IO model, EMSI uses nearly 90 federal and state data sources and creates an integrated data set that balances accuracy with up-to-date relevance. We release new annual-average data on a quarterly schedule, constantly updating our model with new information.
D. Do multipliers depend on the size of the region?
Yes, multipliers are heavily influenced by region size because larger regions tend to satisfy more regional demand inside the region, simply because they are more economically diverse and transportation costs increase with region size. Small rural regions, on the other hand, tend to have low multipliers, because fewer goods and services can be provided locally and so money “leaks out” much more quickly.
Because of this, it is very important to define regional boundaries in a way that is appropriate to the analysis, and also never to use national or state-level multipliers for smaller regions. See part VI of this document, “Defining Appropriate Regions.”