Comments on William Fischel on Property Taxes and School Finance

Comment on William Fischel, “The Median Voter and School-Finance Reform:
How Tax-Base Sharing Undermines the Efficiency of the Property Tax”
Richard K. Green
University of Southern California
February 12, 2009

In the course of 18 years as an academic, I don’t think I have ever had an assignment as intimidating as discussing a paper by William Fischel on the property tax. The only thing worse would be to discuss a paper by Professor Fischel on zoning.

So let me begin by agreeing with many of the points Professor Fischel argues in his paper:

o The median voter in most instances makes pretty smart decisions. The anecdote about the general correctness of majority answers to his multiple-choice exams illustrates the point quite vividly.

o A tax price of unity for school funding produces good outcomes. Districts with tax prices of less than unity will spend more on schools, sometimes for good, and sometimes not. Districts with tax prices of greater than unity might well underspend on schools. Certainly, when tax prices are higher for schools, households are willing to spend less on schools.

o Redistributive funding mechanisms that raise the tax price of schools in some districts can be counter-productive.

In the end, though, this is not entirely satisfying. The fact is that differences in property values, and in tax prices (before redistribution) produce unequal outcomes for school children. As Oates (1970) showed, in a regime where schools are funded locally, higher school spending produces higher property values. The inference we may draw from this is that the net benefits of schools were greater than the net costs of funding them. At the same time, the inequality may be self-reinforcing, as I will discuss below.

To illustrate the problem, let me use as examples two places where I have lived: Wisconsin, and Metropolitan Washington. Wisconsin nicely illustrates two dilemmas about using the property tax to finance schools. First, the distribution of property values per pupil is both highly dispersed and skewed (Figure 1). The average school district in Wisconsin has taxable property per pupil of $664,000, while the standard deviation of property values is $773,000. The dispersion is not driven just by outliers: at the top quartile of the property value distribution, property value per pupil is roughly double the value at the lowest quartile of the distribution.

At the same time, the tax price of schools varies dramatically. In the Town of Brookfield, more than 50 percent of property value comes from commercial property, so the tax price is quite low. In the city of Wisconsin Rapids, on the other hand, substantial chunks of manufacturing property are exempt from the property tax, and farm land is taxed at use value. Because of this, residential property makes of a disproportionately large share of the tax base, and the tax price for schools is higher there than elsewhere.

Turning to Metropolitan Washington, we see the correlation between school quality and house prices, when we look at four suburban counties: Montgomery and Prince Georges Counties in Maryland, and Arlington and Fairfax Counties in Virginia. Average SATs in three of the four counties (Montgomery, Arlington and Fairfax) were above 1600, while the SAT in Prince George’s County was 1283 in 2007. At the same time, the median price of a house in Montgomery County was $475,000, and in Fairfax and Arlington Counties was above $500,000 in 2007, while in Prince George’s County it was $340,000. While house prices in Washington, DC are high (the median price was $450,000), the city has a shockingly small number of married couple families with children. Moreover, the parts of the city with the worst schools—the area east of the Anacostia River—have median house prices in the $250,000 range.

I don’t want to push this too far: Montgomery County is closer to the job centers of metropolitan Washington than Prince George’s County, but Montgomery’s network of roads is actually not as well developed. We of course cannot draw any statistical inferences about capitalization in the DC area, but we can certainly have suspicions.

If school quality gets capitalized into prices, we get both a current and intergenerational dilemma. Because house prices are so high in the areas around DC except for Prince George’s County, the best public schools in the region are not accessible to low-income students. This places these students, already at a disadvantage because of the circumstances of their households, at a disadvantage in accumulating human capital, leading to increasing income inequality across generations.

Because Professor Fischel related personal anecdotes, I will relate one too. My kids went to a magnet high school in Montgomery County, Maryland. Both the students with whom they went to school and their teachers were extraordinary. I often thought it must be both a pleasure and pain to be a teacher or principal in Montgomery County: a pleasure because the students are generally so serious; a pain because the parents are heavily involved with the schools, sometimes to the point of annoyance. The upshot was that my kids got a lot out of their school, and their parents were content with the education they received.

But I couldn’t help but think about the unfairness of it all. The schools (particularly the high schools) in Prince George’s County and Washington DC were dysfunctional, and the kids who were stuck there had much less promising futures than the kids who got to go to schools in one of the strong districts. While it may be a coincidence that recent data show increasing persistence in intergenerational wealth inequality—who your parents are seems to matter more now than it did a generation ago—my prior is that it is not.

Of course, there is a more direct method than redistributing resources across districts for putting children on a level playing field—vouchers that do not tie children to their local schools. Tom Nechyba argues that it makes no policy sense for geography to determine child outcomes. I tend to like vouchers myself. And yet it is the nexus of geography and schools that leads to the positive outcomes Professor Fischel attributes to property tax based school funding. At the same time, as a practical matter, it would be difficult for parents in Anacostia to transport their children to Rockville, Maryland or Falls Church Virginia, for school.

Compounding the dilemma is the fact that the evidence, much of it cited by Professor Fischel, suggests that central government funding of education does not work very well. Public school systems in California used to be jewels of the state. Over the years in which most school funding has flowed through Sacramento, public schools in California have deteriorated. So where does this leave us?

Perhaps an answer arises from a simple insight of microeconomics: that marginal things matter more than average things. The best policy (or perhaps I should say second best policy) might be one in which all schoolchildren had access to the minimum level of resources necessary to receive an adequate education. I think there might be a great deal of consensus about what constitutes this minimum: proficient reading and math test scores at the grade school level; sufficient numbers of classes to prepare students for college at the high school level.
Each school district would receive the funding necessary to provide the minimum level of education necessary. This might not include such things as AP courses.

From an efficiency standpoint, the ideal tax would be a lump-sum tax leveled at the state—or perhaps even federal—level. Such a tax would, of course, be politically infeasible and regressive. The least distortionary tax I can think or is a sales tax of a value added tax. Any spending a community did beyond the bare minimum would be determined and financed by the community. By doing this, the marginal tax price of marginal improvements in education would be close to unity. Such plans exist, and are known as foundation plans. Many of the people in this room have worked on such plans.

One final point. While the median voter model works well for Dartmouth economics students and New Hampshire Villages, it is not entirely clear that it works well everywhere. Alienation is a serious problem in some places. According to the Los Angeles County register’s web site, only 23.6 percent of registered voters in Compton voted in the last school board elections there. Perhaps more problematic for Los Angeles County is that a very large share of its residents are not even eligible to vote.