For decades, federal tax code had allowed employers to apply a portion of workers’ pre-tax salary to pay for parking at work—$250 per month before the recent adjustment. According to a 2014 analysis by TransitCenter, that tax subsidy added over 800,000 daily car trips to U.S. roads, while costing the U.S. Treasury $7.3 billion in revenue each year.

Lawmakers eventually responded to critiques of the parking subsidy by adding a parallel benefit for commuting by train or bus, but it topped out at only $130 a month as of last year. Then, in a legislative victory for transit advocates, last December’s omnibus budget and tax extension bill (known inside the beltway as the “omnibender”) set both benefits to $255 a month.

Get Data Sheet, Fortune’s technology newsletter.

But TransitCenter, using data on transportation spending and behavior, concluded that this apparent equalization still leaves people more likely to drive than they would be in a world with no transportation exemptions at all. The effects varied by locale, with dense urban centers like D.C. and San Francisco seeing measurable reductions in the incentive to drive—but even there, the matched subsidies still encouraged driving over transit use. And less dense areas, such as Miami and Seattle, will see practically no change in commuter behavior with the new scheme.

It’s a puzzling result. According to Steven Higashide, senior program analyst for TransitCenter, one reason equalizing the tax benefit will have limited impact is that in-city bus or subway transit is generally much more affordable than urban parking. That means that raising the transit exemption from $130 to $255 doesn’t provide much marginal benefit. The exception, says Higashide, is for those using pricier monthly commuter rail passes, mostly in the Northeast.

Another factor making the driving incentive ‘sticky’ is that far fewer employers offered the benefit for transit—only 12%, compared to 87% providing on-site parking. Higashide also acknowledges that, at least at face value, driving is simply easier than using transit.

“When you’re giving people free parking, it just appears to be so convenient that it’s very hard to overcome this.”

For more on traffic solutions, watch:

But that immediate sense of convenience is deceiving. By encouraging more traffic overall, the parking benefit is actually making things worse for those who choose to drive—on top of increasing greenhouse gases and other pollutants.

“We really don’t see why the parking subsidy should exist,” says Higashide. “It’s a really perverse incentive.”

Both tax credits also disproportionately benefit the wealthy, because the exempted portion of their income would otherwise have been subject to a higher rate. TransitCenter found that in a dense area like Philadelphia, the parking benefit was worth $459 to household with an income of $50,000 a year, but $1,010 to a household with an income of $300,000.

The final puzzling question is whether it makes sense to have two equal subsidies, encouraging competing transportation models.

“You’re subsidizing people in one direction on the one hand,” asks Higashide, “And in the other direction with the other hand. Why not be clear about what you’re trying to do as a policymaker?”