Results
Where the jobs come from
Jobs by type of effect
Summary
Dollar figures in millions
Direct jobs by spending line
Industries gaining indirect & induced jobs
The top 15 industries by spillover jobs. Direct jobs are excluded.
How it works
This is a regional input-output (Leontief) model built from public data. National BEA tables show what each of 71 industries buys from every other industry. The model scales those purchases to each state with Flegg location quotients (FLQ), which estimate how much of each purchase stays in the state instead of going to suppliers elsewhere.
- Direct: the new spending, and the jobs, pay, and GDP in the industry that receives it.
- Indirect: the extra activity at in-state suppliers, their suppliers, and so on down the chain (Type I minus direct).
- Induced: the activity supported when workers in those industries spend their pay at in-state businesses (Type II minus Type I).
The federal government publishes each state's GDP by industry but not its total sales, so the model estimates sales from GDP. It then corrects those estimates with the Census Bureau's Economic Census, which counts actual sales by state and industry. For example, if a state's chemical industry has high profit margins, the same GDP means less in sales, and so more jobs per dollar of spending.
% in-state is the share of each line bought from businesses in the state. Equipment made elsewhere, out-of-state contractors, and imports create no jobs in the state, so leave them out. The build period spreads the total over time: a "job" is one job for one year, so 4,000 job-years over 4 years is about 1,000 jobs in an average year. The peak workforce is higher.
Spending means new final demand for an industry's output. For operating funds, like child care subsidies, use the service industry (Social assistance). To build something, use Construction.
Assumptions and limitations
- Payroll jobs come from BLS QCEW (jobs covered by unemployment insurance), plus railroads. With Count self-employed workers on, the model adds proprietors the way IMPLAN does. It uses BEA's state totals of self-employed jobs and income, split across industries with Census Nonemployer Statistics. These are counts of jobs, not full-time equivalents, and many self-employed jobs are part-time or side work, such as rideshare driving. Military jobs are not counted.
- Labor income is employee compensation (wages, salaries, and benefits), plus proprietors' income when self-employed workers are counted.
- The model assumes each industry's inputs, prices, and in-state purchase shares stay fixed, and that supply can always expand to meet demand. It measures the activity the spending supports, not net new jobs. It doesn't account for spending cut elsewhere to pay for the investment.
- How much spending leaks out of the state is set by the FLQ parameter δ = . Higher values produce smaller multipliers, especially in small states.
Data sources
- BEA Input-Output Accounts, summary Make and Use tables,
- BEA Regional Accounts: GDP by state (SAGDP2), compensation (SAGDP4), personal income (SAINC4), and personal consumption (SAPCE1)
- BEA earnings and compensation by industry (SAINC5N, SAINC6N) and proprietors' employment (SAINC4, CAEMP25N)
- BLS Quarterly Census of Employment and Wages; BLS Current Employment Statistics (rail)
- Census Bureau Nonemployer Statistics and 2022 Economic Census