How We Did the Math on Our Unemployment Insurance Tracker
We’ve just launched our Unemployment Tracker, an application that follows the current balance of each state’s unemployment trust fund and predicts future solvency. As of January 2010, 25 states have run out of funds and been forced to borrow from federal government, raise taxes or cut benefits. We project another nine heading into the red within six months.
For each state, we gathered the freshest information on unemployment programs, including how much the state paid out in benefits and how much it earned in unemployment tax revenue. Then we plugged the numbers into a formula we created that projects the trust fund’s trend out for six months, and another formula that gauges the fund’s current health.
Here’s what our data includes for each state:
- Unemployment trust fund balances from Treasury
- Benefits paid out by the Department of Labor
- Revenues collected from unemployment taxes on employers
- Interest earned by state trust funds
- Funds distributed to states for unemployment system administration
Where does this data come from?
The Department of Labor releases unemployment revenues and benefits for each state quarterly, but often that data is two quarters old. Instead of relying on stale data, we found up-to-date information on the Web sites of the Department of Treasury and the Department of Labor.
Neither site offers a standard downloadable format for the data, so we pulled the data from searchable databases for each of the three monthly values we needed:
- Benefits are at the Monthly Program and Financial Data tool at Department of Labor’s Employment and Training Administration Web site. For benefits reporting we included the values logged in the Benefits Paid column in these reports.
- Revenues are at the Treasury Department’s “Unemployment Trust Fund Report” page under Transaction Statement (sample report) Each state collects revenue from two main sources, tax income and balance transfers from other states. When an eligible person moves out of state, his home state pays the new state his remaining balance from the fund. For our purposes, a given month’s revenue is the sum of each these two. We’ve also included other amounts of income such as interest payments or administrative disbursements even though they are relatively small.
- Trust Fund Balances - The Treasury Department also tracks each trust fund balance and presents from the same interface report under “Account Statement.” These present a summary of all transactions—revenue and benefit payments—for the month. They also provide the month’s ending balance on the very last line which is the number we’re looking for.
You can download this data for each state from the state pages on the Unemployment Tracker.
What does this data tell us:
- Trust fund balance - This is simply a monthly snapshot of the amount of money in each state’s trust fund. Trust fund balances tick up and down as revenues come in and benefits payments go out. Once a state is borrowing, the trust fund balance becomes a much more static amount. Many borrowing states allow their balance to approach zero; others will maintain a small balance while the bulk of benefits payments are actually financed by federal loans.
- Revenues - This represents the money coming into the state’s trust fund from taxes collected, stimulus payments, interest earned and administration funds. Plotting this data tells us several important things about a state fund’s health. First, in nearly all states a significant drop in revenue since the start of the recession can be observed. Second, the peaks in a state’s revenue curve give an indication of its tax policies. States that tax only the first few thousand dollars of each worker’s income (the federal minimum is $7,000) will see nearly all of their revenue by the second quarter of the year. This is particularly relevant now, because those states can anticipate very limited revenue for the next five months, while their benefits obligations remain high.
- Benefits - With the recession, every state has seen an uptick in the amount of benefits paid out. In states where the increase in unemployment has been modest, the benefits paid out have increased only slightly. In other states, a many-fold increase in unemployment has led to a dramatic spike in benefits paid out. Where benefits exceed revenues, shown in the graph by a benefit curve that is higher than the revenue curve, the trust fund balance will decrease. Many states show a precipitous decline in trust fund balance in recent months.
- Net income - This is just a measure of the number of dollars a state’s trust fund changed last month. For November, all states had a negative net income, sometimes very significant. On average, states that were not already borrowing saw their trust funds lose 13 percent of their value. In some states the decrease was as high as 50 percent.
To add context to the graphs, we are also including the state’s most recent unemployment rate, level of benefits and the percent of unemployed workers receiving benefits. For further context, these values are also ranked relative to other states. Taken together, these indicators paint a picture of state policies. In particular, some states have maintained solvency by paying low benefits or tightly restricting worker eligibility for benefits.
Predicting the future
We used our database of monthly revenues, benefits and trust fund balances to analyze each trust fund’s future potential.
Each state’s unemployment trust fund is an account that is bolstered by tax revenue and depleted by unemployment benefits. In any given month, the trust fund balance will be the balance of the month before plus that month’s net income (revenues minus benefits), or as a formula:
Where m is the current month. In order to project trust fund balances, we can use the current trust fund balance and sum the differences of projections for both revenues and benefits:
where n is the number of months in the future we’re trying to predict and BENavg is a simple average of the benefits collected in the most recent quarter:
To calculate REVadj we summed the revenue of the most recent quarter and divided that by the sum of the revenue of the corresponding quarter last year. This gave us a correction factor that allowed us to adjust last year’s revenue for the month we’re trying to predict. Then we multiplied that factor by the corresponding month we tried to predict from last year:
Predicting benefits proved relatively easy: they tend to slowly migrate up and down, making a rolling average a fairly accurate predictor. Predicting revenues was the hard part - they fluctuate wildly from month to month. However, using data from last year, adjusted for the revenue drop-off that most states saw in 2009, allowed us to predict revenues with a reasonable degree of accuracy. This chart comparing California’s 2008 revenues to our predictions, for example, shows our estimated curve has the right shape and similar values:
Please note that when a state starts borrowing from the federal government we can’t project a future trust fund balance because some or all revenue is set aside to pay down the outstanding loan.
Gauging the Past
The Unemployment Tracker also flags the periods a state’s fund is in trouble. Because we can rely on real data from the past the formula is simpler than our projection formula:
This formula essentially flags any state where benefits paid out exceed revenues to a point that cannot be sustained by the trust fund for more than four months.
Our Hottest Stories
- Segregation Now
- Beyond Ratings: More Tools Coming to Pick Your Doctor
- Rocky Mountain High or Reefer Madness? Legal Pot in Colorado Comes with Risks
- Even After Doctors Are Sanctioned or Arrested, Medicare Keeps Paying
- Long After Sandy, Red Cross Post-Storm Spending Still a Black Box
- Shake-Up Inside Forensic Credentialing Org
- Brooklyn DA Moves to Free Man after Long-Buried Evidence Surfaces
- The U.S. Government: Paying to Undermine Internet Security, Not to Fix It
- How the Labor Department Has Let Companies Off the Hook for Unpaid Internships
- Brooklyn Man Walks Out of Court, Cleared of Murder After 24 Years in Prison