People hate the property tax more than other taxes. There are fairly regular “tax revolts” against the property tax, many of which are based on local or statewide referenda. Property tax limits, whether imposed by referenda or by state legislatures, often remain binding for a number of years–even decades. In contrast, successful revolts against other taxes, such as the income, sales, or corporate tax, are rare and often temporary. Although limits on other taxes are often considered and occasionally enacted, it appears that the public rarely has the will to insist upon their literal enforcement. In part because of tax revolts, tax limits, and general antipathy towards the property tax, property tax revenue has declined greatly as a share of all taxes collected in the U.S. It has also declined as a share of U.S. GDP.

So begins a recent NBER publication, “The Hated Property Tax: Salience, Tax Rates, and Tax Revolts,” NBER Working Paper No. 18514 by Marika Cabral and Caroline Hoxby. [Ungated version here.]

Cabral and Hoxby find that when property taxes are less salient, people tend to be less aware of the amount they pay and when people are less aware, property tax rates tend to be higher.

How do they get variation in salience? By looking at the percentage of people who pay their property taxes with tax escrow, that is, bundled with their mortgage payments. Check out their Figure 7 (page 55 in the ungated version) for a striking graph. The standard deviation of the difference between the property tax reported in surveys and actual tax paid was $2215 for people who used tax escrow and only $781 for people who did not.

Their bottom line on the effect of salience on property tax rates:

A standard deviation increase in tax escrow among mortgage holders generates a 0.46 mil increase in the property tax rate. A standard deviation increase in tax escrow among all owners or among all households generates about a 0.43 increase in the tax rate. These are not inconsiderable tax rate increases when we recall that there is only a difference of 1.1 mils between the 10th percentile tax rate and the average tax rate.

They also find on an effect on so-called property tax revolts, namely, measures, often undertaken first by citizens’ groups, to limit property taxes, Prop. 13 in 1978 in California being the most famous. Specifically:

The estimates indicate that an increase of one standard deviation in the tax escrow percentage generates a decrease of 0.93 to 1.24 in the number of limits in force. (The effect varies with whether the tax escrow percentage is measured among mortgage holders, home owners, or all households.) In other words, when tax escrow in a state rises by a standard deviation (suggesting that property tax salience falls), a state’s voters support about 1 fewer limits. The mean number of limits in the regression is 2.3 with a standard deviation of 1.2. Thus, the decrease in the enactment of limits is considerable.

This has obvious implications for other taxes. One of the most hated taxes in Canada, for example, is their highly visible GST, a tax on goods and services that began at 7% in 1991 and has now fallen to 5%, precisely because it is hated. And of the least visible taxes is the income tax, which, because of withholding, is not known well to most people.

Isn’t it interesting that shortly after this tax was introduced in Canada, there began the push by the Liberal government to rein in federal spending? I wrote about it here. Yes, it’s true, as some of my Canadian cognoscenti friends have pointed out to me off-line, that there were key bureaucrats in Canada’s Finance Department that pushed for these cuts. That’s a fact to which I didn’t give adequate attention. But it’s quite conceivable that Canadians did not resist the cuts and even themselves pushed for them because of the newly-visible cost of government. And Canadian politicians listen to Canadian voters as well as to Canadian bureaucrats.