If you use an EMSI product or service, chances are you’ve seen some of our labor market projections. We provide detailed job estimates, by industry and occupation, for each of the next 10 years from the current year. A common question that we get from clients who see these figures is where do EMSI projections come from? The short answer is that we use a combination of government-published sources at the national, state, and regional levels and a combination of the short-, mid-, and long-term trend lines. But if you’re after something a little more detailed, read on.
Industry & Staffing Pattern Projections
The first step in both our industry and occupation projections is our robust industry data (which we’ve explained in more detail here). We first create simple short-, mid-, and long-term trend lines for every industry and county based on the historical data. Those lines are then averaged and damped to prevent extreme change. We then incorporate national industry projections (BLS national employment projections) as well as state and regional projections provided by state labor market organizations to adjust these numbers again. Finally, we take aggregate geography and industry group figures for 10-year percentage change and make sure it matches the corresponding published geography and industry numbers.
One question we occasionally receive in response to our projections is whether they come with confidence or error estimates. The answer is that since EMSI does not use an econometric or regression-based projection model, we do not include such estimates. The primary reason for this is that our projection methodology relies heavily on projections produced by the BLS and state LMI entities that don’t come with such confidence or error estimates. This in turn makes it impossible for us to determine composite confidence intervals based on those sources.
Why Not ARIMA?
We did a study of our methodology compared to the autoregressive integrated moving average (ARIMA) model looking at the short-, mid-, and long-term forecasts. We attached the model to QCEW and compared those results with our projections. At each level, our projections showed a smaller percentage of error from the actual figures.
EMSI Projections vs. Other Sources
As we have outlined, EMSI projections will not match the BLS or any other individual state projections at the most detailed level. This can be explained in a variety of ways.
- Differences in Coverage
EMSI has broken out our labor market data into four different classes of worker that allow our clients to dig into specific elements of their economy. In most states, the first two classes of workers (QCEW Employees and Non-QCEW Employees) are most closely comparable to the industry projections from the published sources. We don’t expect our self-employment and proprietors projections to match BLS and state-level projections.
- We Don’t Use BLS/State Occupation Projections
Our occupation projections are based off of our industry numbers with our projected, regionalized staffing pattern applied to the figures. As a result, the occupation projections will necessarily differ from BLS occupation numbers.
- Differences in Methodology
EMSI uses a combination of BLS, state, and internal projection methods. This will naturally differ from the BLS and state projections alone.
- Release Lag
EMSI projections are based on the most recent available data and published projections, but the BLS and states release only long-term projections on a two-year basis. EMSI’s internal projections are more heavily weighted as state/national projections age.
In summary, EMSI has developed a proprietary methodology for determining projections that incorporates national-(BLS), state-, and regional-published projections and that outperforms traditional projection models.
If you have a question you’d like us to address in the FAQ series, please leave a comment.