The Census stopped using its state tax rankings. Why don’t we?
Now that the governor is in the middle of another stand-off with the Legislature over the budget, his press office has pounced on the latest story that purports to show that Vermont’s taxes are too high. The article appears this week in Vermont Business Magazine. Unfortunately, the magazine arrived at its conclusion using two pieces of flawed data: a self-selecting opinion survey and Census figures no longer published because they can be misleading.
Vermont Business Magazine hired a consulting firm to survey Vermont businesses. The consultants sent out 3,100 surveys and received 254 responses. The magazine conceded that the results were not statistically valid, but then went ahead to draw sweeping conclusions based on the results.
This kind of survey is a self-selecting opinion poll—sometimes know by the acronym SLOP—and shouldn’t be used the way Vermont Business Magazine is trying to use it. The results say something about the 254 businesses that chose to answer the survey questions. But they tell us nothing about the other 92 percent of the target businesses that didn’t respond. And they certainly say nothing about 20,000-plus private sector business in the state.
There are statistical methods for creating random samples, and only with a random sample can you draw conclusions about the opinions of a larger population. Vermont Business Magazine didn’t do that. Instead, knowing the results were flawed, it claimed in the opening sentence that “[a] statewide survey conducted for Vermont Business Magazine by a Stowe-based consulting firm identifies tax rates as the number one factor that could cause companies to leave Vermont.”
The article goes on to quote some survey respondents who complain about Vermont taxes, and then cites a 2005 U.S. Census report to back up those complaints. The magazine points out that state tax-ranking surveys are controversial and may be influenced by ideology. But it identifies the Census report as “one seemingly impartial finding.”
It’s too bad Vermont Business Magazine didn’t look closer at the report and the Census Bureau’s own conclusions about the data.
The Census Bureau produces two reports about revenues collected by the states. One focuses solely on state-level revenue. The other includes state revenues and revenue collected by local governments, like counties and municipalities.
The state revenue report, which is the one Vermont Business Magazine cited, distorts Vermont’s ranking because most of our property taxes are now state revenues. In other states, property taxes are primarily local taxes. This puts Vermont at the top of the state-revenue-only rankings.
To be sure, when state and local taxes are counted, Vermont still would be ranked higher than many other states on a per capita basis. But that’s why it’s important to read the fine print on the Census Bureau website.
After 2005, the Census Bureau stopped computing per capita tax rankings. The reason: “Analysis based on rankings or per capita statistics can be misleading and misinterpreted because of subtle yet important differences in state government organization and economic structure.” A big problem, the Census explains, is that the state or state and local revenue in its reports is all of the revenue collected within a state’s borders—not just the taxes paid by residents. Alaska and Florida are two cases in point. Alaska collects a severance tax from oil companies, but has no general sales tax or personal income tax. So it would rank high in per capita taxes even though much of the revenue doesn’t come from residents. Florida relies heavily on sales taxes, many of which come from non-Florida residents.
A better way to compare state taxes is to look at the taxes people actually pay. The Joint Fiscal Office did this in a study a couple of years ago. The study created a couple dozen hypothetical tax filers and then computed their taxes in each of 12 states, including Vermont, Florida, and other New England states. The Joint Fiscal Office study did not include property taxes.
The District of Columbia regularly analyzes the taxes paid in the largest city in each state and compares them to the district’s taxes. Like the Joint Fiscal Office study, the DC analysis creates hypothetical filers and calculates their taxes. The DC study does include property taxes. In the latest analysis released last fall, the DC study looked at families with incomes of $25,000, $50,000, $75,000, $100,000 and $150,000. Except for the lowest income bracket, where it ranked 44th, Burlington was in the middle of the pack.
The Vermont Business Magazine article was correct when it said: “Methods of calculating states’ comparative tax rates differ on the basis of who’s doing the calculating.” With credible information showing that Vermont’s taxes compare favorably to many other states, you have to wonder why the governor and others are so intent on seizing any information—even misleading information—that makes Vermont look bad. What would the public mood toward taxes be if the governor regularly cited the DC study and said families in Burlington with incomes of $75,000, $100,000, even $150,000 paid taxes that were lower than families in half the other states?
Thanks, Jack for a great article. It’s a relief to know there are some voices of reason to go against the ocean of misinformation. It’s a shame that Governor Douglas has to resort to such tactics, but I am not surprised. After all, what can you expect from a former two-time chair of the Bush campaign.