New tool forecasts unemployment

Analysts use year-over-year data to predict rates in Routt, Moffat


Predicted unemployment

Routt County

■ April: 10.8 percent

■ May: 11.8 percent

■ June: 10.9 percent

Moffat County

■ April: 9.3 percent

■ May: 8.9 percent

■ June: 9.1 percent

Source: Yampa Valley Partners Regional Economic Forecast

— Scott Ford and Yampa Valley Partners are trying to add perspective to the inexact science of unemployment figures.

The statistics are a challenging metric partly because of the way unemployment is calculated, said Ford, director of the Routt County Economic Development Co­­operative. Routt and Moffat counties are too small to get precise figures because the numbers are based only on active unemployment claims. Also, the counties’ smaller labor forces can result in manic swings in jobless rates.

“Following unemployment percentages is valuable, but it’s chaotic, especially in smaller places because small numbers have such a big impact,” Ford said.

But he and Kate Nowak, of Yampa Valley Partners, have created a quarterly newsletter to try to add context to the conversation.

The letter released last week includes predictions about retail sales, employment, real estate, transportation, construction and energy in Northwest Colo­rado.

The information includes a look at past unemployment and, more important, a forecast of the next three months’ numbers.

The team used a method developed by Robert Fountain, of Cali­­fornia State Uni­­ver­­sity, Sacra­mento, to create the predictions.

Here’s how it works: They take the number of people with jobs in, say, February 2010, and divide it by that same figure for February 2009. They then take the reported labor force in February 2010 and divide it by the reported labor force in February 2009.

They subtract the second number from the first. If the difference is less than 1, that signals jobs-related economic stress.

Each figure for each month — compared with the same month the previous year — becomes a data point plotted on a graph.

The comparisons allow Ford and Nowak, or whoever is looking at the data, to observe trends and make forecasts. They expect Routt County’s unemployment to be roughly 10.8 percent in April, 11.8 percent in May and 10.9 percent in June. It was 7 percent in February, according to figures released late last month by the Colorado Department of Labor and Employment.

The newsletter explains the calculations and points out the flaws of raw unemployment percentages, stating, “Although this is an important number, it does not tell the full story because it does not recognize the fluid nature of employment. To recognize the fluid nature, a useful assessment is to graphically compare the twelve-month percentage change in the number of employed persons with the twelve-month percentage change in the civilian labor force.”

The measurements still aren’t precise, Ford said. Employment statistics don’t differentiate between part- and full-time jobs and don’t reveal whether one person holds more than one job.

Also, the federal government has extended unemployment benefits, so some people are waiting longer before settling for a job that may be outside their field or pay less, Ford said.

“These programs have great value, but they will result in unemployment staying higher longer,” he said.

The good news is that the outlook is positive, Ford said. He expects the indicator to be above 1 in the fourth quarter of 2010 or the first quarter of 2011 for Routt County.

“It allows us to see the change in context,” Ford said. “It doesn’t mean we won’t have economic stress — we certainly are.”


Scott Wedel 7 years, 1 month ago

Personally, I would remove the size of workforce from the calculation for the region by using the number 1 for that part of the calculation. At the very least I would track employment via employed only, and the employed/workforce number.

The trouble I see with using size of workforce is that the point of using size of workforce is to adjust the final result for workforce trends such as an aging workforce that is shrinking due to retirement or for a young population that has many people entering and growing the workforce. Thus, the size of workforce adjustment is meant to adjust for demographic trends to show the true nature of the employment market.

But our labor market is not dominated by demographics. It is dominated by people moving here because there are jobs or leaving because there are no jobs. Thus, the size of our workforce is a very fluid number that is dictated by the number of jobs and is not an independent variable.

The flaw of using the workforce number can be seen from recent employment data. This area is still losing jobs, but because now so many people are leaving the area, we are now shrinking workforce faster that we are losing jobs. Thus, solely because of how quickly we are shrinking our workforce, this economic indicator is turning positive. That is messed up.

I think that it is entirely possible that economic indicators will show improvement in Q4 or Q1 2010 because employment numbers fell off a cliff during Q2 2009 so year over year comparisons become almost hard not to beat later this year.


Scott Ford 7 years, 1 month ago

Hi Scott – I was hoping to hear from you. The methodology used in this tool is from a University of California economist. It does not necessary measure the unemployment rate as much as it measures the relative economic stress that is occurring in a county as a result of the year over year change in the ratio of workforce to jobs. At looking at over 20 years of data – it does seem to foretell when we are emerging into or out of a period economic stress at the county level associated with the employment segment of the economy.

I like it because it is relatively inexpensive to calculate. In addition its value as an index is that it can be replicated month to month using the same methodology. By no means do I think this employment index is perfect. It is only my hope that we can expand the discussion about unemployment beyond focusing solely on the unemployment percentage. Because of our small size changes in either the numerator or denominator result in what is perceived as dramatic changes in the unemployment rate percentage.

The index is calculated using data from secondary sources that are updated monthly. Under a separate email I am sending you the MS-Excel workbook used in calculating the index. Since this index is still in the development phase - I am very open to change if another methodology can be used that produces relevant historical data and can be calculated monthly.


Scott Ford 7 years, 1 month ago

Good Morning Scott W -

I appreciate you disagreeing with me. I think debate with civility is welcomed in these blogs.

Although very tempting, focusing solely on the unemployment percentage it misses the bigger picture. I understand why this happens because unemployment is so heartfelt; however, the percentage rate is only one side of an issue. It is not the whole issue.

The Predictor Indicator you reference although it uses year-to-year changes in the number of jobs and labor force it does not predict unemployment percentages. It predicts the level of economic stress that is occurring in the county because of imbalance that between workforce and jobs. I know that this sounds on the surface counter intuitive. Let me if I can explain.

For example, let us assume we live in a county that is an isolated island and we make everything we consume ourselves. In addition, that everybody between the ages of 16 and 64 also works and they work full time jobs. That labor force in this island example totals 1,000. In our example, there are also 1,000 jobs. The calculated unemployment rate would be zero. Let us assume that it had been at zero for years. If 100 jobs were lost in a given month the calculated unemployment rate would spike to 10%. The folks that lost their job would reduce their spending, that would ripple through the economy of the island, and a few more folks would lose their jobs. The island economy would be under stress - its economy was contracting.

This ripple effect would eventually run its course. If there were no hope of these jobs, ever coming back the workforce without jobs would eventually leave the island. The economic stress would lessen and the economy itself would eventually rebalance and stop contracting. Although a smaller economy the unemployment rate would eventually return to zero.

In the real world nobody and particular no local/regional economy is an island. And of course zero unemployment does not exist. Things in Routt County are much more complex and influenced by a host of factors that boggles the mind. Some of these factors include but are not limited to that not all jobs are full time, not everybody counted the labor force works and personal financial resources including unemployment benefits allow folks to stay longer than they could in my silly island example.

The Predictor Indicator does not measure unemployment but attempts to measure the economic stress caused by the unemployment - it measures imbalance and identifies trends to determine if the imbalance is getting worse or better. Based on the March 2010 labor figures the Predictor Indicator for Moffat and Routt is essentially unchanged from February. Although not yet balanced It is not getting worse. Balance does not mean that the economy returns to some predetermined level of normal - I use other tools to do that.


Requires free registration

Posting comments requires a free account and verification.