The practical file
Great schools by the numbers.
Around here, we track school test scores as passionately as sports fans track the stats of their favorite teams.
I go to greatschools.net for the latest scores, but I always take them with a grain of salt. I'm no education expert, but I'm skeptical that test scores are either the best or only way to judge a school's performance. But they're certainly the easy way. It's our habit to quantify things—"data don't lie"—that don't like being quantified, the same pounding of the square pegs of data into the round holes of reality that I see so often when Internet meets real estate.
Yes, it's easy to say "high scores, good school" and "low scores, bad school" and walk away feeling we've solved yet another of life's mysteries. Too easy, perhaps, since a school can improve its scores simply by teaching to the tests. If a few less-than-subtle nudges to a scoring system can make it light up like a pinball machine, we can rightly question how meaningful and rigorous that system is.
Another reason I don't take test scores to the bank is because high scores and high home prices have a strong positive correlation. Want high-scoring schools? It's simple: buy in an expensive neighborhood. The positive correlation exists because high home prices mean affluent homebuyers, and at a time when so much of a public school's resources come from parents, in the form of checks and volunteer time, the more affluent a school's parents are, the more resources that school will have. And when people buy into a neighborhood for its schools, and pay a stiff premium to do so, it makes sense that schools will remain a top priority to them.
There's another strong positive correlation, between low test scores and high-density neighborhoods—neighborhoods with a preponderance of apartment buildings. As a former property manager, I know that apartments generally, although not always, are rented by the most transient and least affluent of the community. "Least affluent" is relative, of course; the typical Menlo Park or Palo Alto renter may be as affluent as the typical Redwood City or Sunnyvale homeowner, and I've had the pleasure of helping some of the former transition to the later. But the correlation seems to hold true no matter how high a city's rents.
For example, upscale Burlingame's lowest-testing elementary school draws many of its students from the downtown apartment district. Pricey Palo Alto's lowest-testing elementary school draws many of its students from the Ventura neighborhood's relatively affordable apartments. The Mountain View Whisman school district's Web site makes a point of telling us that "Theuerkauf School [its lowest-testing school] draws its students from single family homes and rental apartments (italics mine)...The student body represents a wide variety of ethnic backgrounds."
Theuerkauf is an interesting case, because if you read the parent comments at greatschools.net you learn that until a few years ago the school didn't have a PTA (Parent Teacher Association). That's mind-boggling if you know how vital a PTA is to a school's fundraising, and how vital fundraising is to a school's performance. It gives you a glimpse of a parent base that in many cases may have neither the time, the money nor the stability to properly support its local school.
So how do you find a good school? Ask around? Then you might encounter the problem of parent perception. A client of mine considering buying in the San Mateo Village neighborhood recently forwarded me an email she received from a friend with children in the Village's elementary school, George Hall. Yes, it was a decent school, the parent wrote, but not nearly as good as the school where she worked, Meadow Heights. There are only four ironies here. First, George Hall has higher test scores than Meadow Heights. Second, other clients of mine were selling their home in the Meadow Heights attendance area because of dissatisfaction with the school. Third, the Village is more affordable than the neighborhoods surrounding Meadow Heights, yet its school has higher test scores. And fourth, one parent posting on the "parents' comments" section of greatschools.net claims Meadow Heights' test scores are "dragged down" by two special needs classes.
So who, or what, do you believe? How do you evaluate either school or any school? I'll answer those questions as soon as I get my PhD in education, but in the meantime I'm still convinced that there's more to a school than numbers, unless those numbers happen to have dollar signs in front of them. And speaking of dollars, let's address two more questions about the relationship between school quality and home prices.
First, you may have heard that homes in neighborhoods with good schools hold their value better. My historical outlook on this question is brief—ten years in real estate sales isn't long enough to be an expert on long-term trends—but I can tell you that the previous downturn, the dot-com bust of 2001, hit the middle-class communities with highly-regarded schools hardest. Why? Because middle-class communities with highly-regarded schools had seen the biggest run-ups in price. Why? Because the people driving the late 1990s boom, the beneficiaries of dot-com wealth, were generally highly educated and sought communities with highly-regarded schools.
The low end, on the other hand, virtually synonymous with low-testing schools, weathered the dot-bust quite well, since low-end buyers weren't as dependent on dot-com money. The most recent real estate cycle has been the reverse of 2001, as home prices in neighborhoods with low-testing schools first soared, pumped up by aggressive subprime lending, then nose-dived when subprime was snatched away, while the last year or so has seen local mid-priced communities experience only modest price reductions or, in a few cases, continued gains.
So "it's simple", as the self-proclaimed real estate experts say, because "it's all numbers". Sure it's simple. Good schools keep home values high, except when they don't.
The second question sounds similar to the first but is really completely different: how much value do good schools add, year in and year out, over the long term? The addition of long-term value is much easier to quantify than the retention of value during the peaks and valleys of a real estate cycle. Just measure the difference in average price between neighborhoods with different schools over a period of, say, ten years, long enough to get those neighborhoods through the unique characteristics of a real estate cycle or two.
Here we're comparing five Santa Clara County neighborhoods, all quite similar in appearance and housing stock—in fact, virtually indistinguishable—but with significant differences in the perceived quality of their school districts. You'll note that these five neighborhoods include two sets each located within the same city but within different school districts. The first set is Mountain View's MLS area 209 "Miramonte" with either Mountain View or Los Altos schools, the second the southwest corner of Santa Clara with highly-regarded Cupertino schools and the adjacent part of Santa Clara with less-regarded Santa Clara schools. The fifth neighborhood is MLS area 233 "South Palo Alto", which takes in much but not all of Palo Alto south of Oregon Expressway. From "South Palo Alto" I've excluded the small strip that's in Palo Alto but part of the Los Altos school district.
This mixing of cities and school districts brings me to an important point: city and school district boundaries aren't exact matches. In fact, they rarely match, because cities and districts are different entities with separate histories. Often the districts are the older entities, and reflect unofficial "spheres of influence" that unincorporated communities or embryonic cities had long before the annexation races of the post-World War II era. That's why you have, say, Campbell schools not just in Campbell but in San Jose, Saratoga, Los Gatos and Santa Clara, or Cupertino schools not only in Cupertino but in San Jose, Los Altos, Saratoga and Santa Clara.
The chart shows that the "right" school district can make a big difference in price. Bear in mind that the area in this group with what I think are the most consistent and attractive neighborhoods, the southern-most part of Santa Clara with Santa Clara schools, has the lowest average home price. That's good information, especially if you're a buyer who doesn't care much about local schools.
But does this chart accurately reflect the difference schools make? Not entirely, because the data hasn't been adjusted to reflect the often substantial differences in average home size between these areas. Which brings up another point: highly-regarded schools encourage new construction, which in this area always means tearing down a small house and replacing it with one far larger. Because of this, average home size in an area with highly-regarded schools will usually be larger. Sought-after schools also encourage homeowners to expand and extensively remodel their existing home.
So let's go back and adjust all five average home prices to the same square footage, 1473 sq.ft., which happens to be the average size of homes in one of these areas.
The premium that Mountain View area 209 homes in the Los Altos district command over area 209 homes in the Mountain View district, 11.43 percent, is a bit surprising when you compare the API test scores of the local Los Altos district elementary school, Springer, to its Mountain View district counterpart, Bubb: 948 and 866 respectively. Yes, that 9.47 percent difference is a nice tidy number that's close to the 11.43 percent difference in price, but if you're familiar with API scores, you'll recognize that Bubb is no slouch. In this case, 11.43 percent is the difference between very good and very very good.
The seemingly small difference between adjacent Santa Clara neighborhoods with Cupertino and Santa Clara schools is also surprising, given the typical homebuyer's perception of those districts. This is due partly to the fact that a 7.4 percent price difference appears more insignificant on a bar chart than it does to a prospective homebuyer calculating his or her monthly mortgage payments. But there's something else going on here, as there so often is in real estate: I mentioned that the neighborhoods in this part of the Santa Clara school district are usually of above average quality. Santa Clara with Cupertino schools, on the other hand, has a high percentage of the home many homebuyers love to hate, call it "Eichler" (incorrectly) or "midcentury modern" or "flattop". Subtract these lower-demand (and smaller) homes from the mix and the average Santa Clara with Cupertino schools home price goes up $5000 and the percentage difference between the two districts rises to 8.19 percent.
That's closer to the 9.47 difference we saw in Mountain View, but still surprisingly low, so maybe there's another "something else" going on. And there is: Cupertino schools are Cupertino schools, right? Not according to the API test. The elementary school serving Santa Clara, Eisenhower, is one of the lower-testing Cupertino district schools, although a score of 894 is far from shabby. And Santa Clara schools are Santa Clara schools, right? Wrong again. Our sample neighborhood includes the attendance areas of two Santa Clara elementary schools, and one of them, Sutter, weighs in with a hefty 865 API score. The other, Westwood, has a more typical 733 score although, again, there's something else going on: 33 percent of Westwood students are English language learners, and 39 percent participate in a free or reduced-price lunch program. I'll bet these two facts have a lot more to do with those lower scores than a lousy principal or uninspired teachers.
I've included area 233 "South Palo Alto" in this group, even though there's no contiguous area to compare it to, because of its international reputation as the Holy Grail of relatively affordable outstanding schools. The scores of the two elementary schools that take in most of area 233, Palo Verde (919) and Fairmeadow (923) would seem to confirm this perception.
Let's wrap up this section with a quick overview of how these five areas perform in the real estate marketplace in ways other than price.
This chart shows two market characteristics, the average number of days on market (DOM, on the left axis) a home takes to sell, and "bid", the average amount over (or under) list price buyers have offered since 1999. Since the two characteristics combined are a homebuyer "excitement-o-meter", a measure of how jazzed buyers are by the prospect of living in each of these five areas, you'd expect to see a strong negative correlation between how much over list buyers are willing to pay ("bid") and the length of time a home sits on the market unsold. In other words, the higher the perception of neighborhood school quality, the less time homes in that neighborhood take to sell and the more over list price they sell for. And the chart shows that's true, for the most part. Mountain View with Los Altos schools sells quicker and for more over list price, as does Santa Clara with Cupertino schools.
Note that not only does South Palo Alto sell quickly, but that it's in a class by itself when it comes to overbid (as any South Palo Alto buyer can tell you) at a whopping 4.88 percent over list price since 1999. And in fact, adjust South Palo Alto from its 1670 sq.ft. average home size down to 1473 sq.ft. and the overbid jumps to 6.01 percent, a sign that over the past ten years the hottest competition in South Palo Alto has been for smaller, more affordable "starter" homes. But note how seemingly un-jazzed homebuyers have been about living in Santa Clara with Cupertino schools, even with its great test scores. Yes, there's another "something else" behind this as well: South Bay homebuyers really really want Cupertino schools. However, 1) they'd rather have a Cupertino than a Santa Clara address, and 2) South Bay homebuyers are far more conservative in their overbids than Mountain View or Palo Alto homebuyers. In fact, the road leaving Palo Alto for the South Bay is cluttered with buyers who couldn't bring themselves to make that one no-fear all-out offer it takes to bag a home in Palo Alto.
Finally, let's look at absorption, or what might be called the "success rate" of a marketplace: the percentage of listings in each area for sale that actually sold. We'll also look at a measurement that seems even more arcane (and may be), the ratio of homes taken off the market unsold to homes sold, or what could be called a marketplace's "failure rate". Both measurements will tell us in different ways how hot real estate has been in each area since 1999.
Again, we're looking for a strong positive correlation between success rate and highly-regarded schools, and a strong negative correlation between failure rate and those same schools. And, again, we find it—sort of. Mountain View with Los Altos schools does indeed have higher absorption and a lower failure rate. But see how Mountain View with Mountain View schools is virtually neck-and-neck with much more highly-regarded South Palo Alto, a sign that many buyers have been willing (or forced) to trade a few percentage points of test score for a few points off the price. And check out Santa Clara: virtually no difference, most likely an insight into the fact that many low-midrange neighborhoods in off-brand school districts have turned in a far more consistent performance over the booms and busts of the past ten years than the more volatile upper-midrange neighborhoods with name-brand schools.
And now for the grand finale, with spectacular fireworks and dancing elephants. As my stirring salute to the indomitable spirit of data-don't-lie, I'd like to make a questionable contribution to this equally-questionable cause with an anything-but-infallible measurement tool I call School IQ®. (If I was thirty years younger and had better marketing instincts I'd call it something really hep and geezer-puzzling like Flamingporpoise® or Zillow®.) I've played around with this idea for years: quantify the ostensible bang a local elementary, middle or high school gives you for the very real bucks you spend on a home in its attendance area.
The biggest drawback to School IQ is that it's best used to evaluate elementary schools only. Middle and, especially, high schools draw from such large and often demographically diverse areas that their test scores are less a meaningful indication of the quality of education they deliver and more a judgment on the quality of their feeder schools. For example, Menlo-Atherton High School has low average test scores, yet many of its graduates go on to top-flight colleges and universities. Why? Because M-A, like, I suspect, many other lower-scoring high schools, is a two-track school, one track coming from poorly-performing elementary school districts and ill-prepared, the other from affluent neighborhoods with highly-regarded schools and on the fast track to success.
So let's keep things simple and stick with elementary schools. We'll spot-check three, one that I know will seem to offer lots of bang, one that will seem to offer little more than a muffled thud, and finally that ever-popular local and even international gold standard, South Palo Alto. To keep things even more simple, and so that those of you at home can play along, I'll use 2007 sales data from the Multiple Listing Service area that corresponds most closely to each school's attendance area. Those of you with more spare time and a penchant for spreadsheets can track sales prices by street address within each attendance area. (As always, be sure to confirm school availability and attendance area boundaries before buying a home.)
First, the neighborhood with the mostest, the extremely affordable northeast corner of Redwood City called either Lenolt or "the letter streets". Through a twist of fate and attendance map drawing, this humble area is part of the high-scoring (compared to typical Redwood City district scores) Clifford attendance area. Let's divide Lenolt's 2007 average single-family sales price of $698,060, cheap by mid-Peninsula standards, with Clifford's healthy 804 API score, to get a School IQ score of 868.
Woo-hoo! Now we need another School IQ to compare this one to. Let's stay in the Redwood City district but move way across the district to one of the most expensive neighborhoods on the mid-Peninsula and in the universe, central Atherton. In 2007 the average sales price for central Atherton with Redwood City schools was a scintillating $2,985,250. The local Redwood City district elementary school, Selby Lane, scored 719 on the API test. Divide sales price by API score and you have a School IQ of 4152. So when it comes to School IQ, apparently high isn't good. And no, big bucks don't always get big test scores. And yes, even rich kids go to low-scoring schools. Except that they don't, of course. If I remember correctly, a few years ago only three Atherton kids attended Redwood City schools. I happen to know who one-third of them are.
But Selby Lane brings up one last point I'd like to make: schools should be evaluated against their context. Selby Lane's API score gets a dismal 3 of 10 rating when compared to statewide schools. But compare it to similar schools—schools with, as greatschools.net says, "similar demographic profiles" based on "parent education level, poverty level, student mobility, student ethnicity and other data"—and Selby Lane's rank jumps to a very solid 7 of 10. The lesson here for parents poring over test scores is that the school you might dismiss for its low test scores is almost certainly in an affordable area, perhaps one that you can comfortably afford, and it may be doing a good job with the hand it's been dealt. Atherton isn't affordable, of course, but the neighborhoods that Selby Lane actually draws from are some of the most affordable on the mid-Peninsula, and that's a nice place to live.
Let's wind things up with MLS area 233 South Palo Alto and see how it stands up to School IQ's unblinking scrutiny. Average 2007 sales price: $1,372,409. And 36 cents. 2007 API test score, averaged between the two local schools, Palo Verde and Fairmeadow: 921. School IQ: 1490.
So according to the numbers—and we know data don't lie—Atherton with Redwood City schools is a huge rip-off, Palo Alto less a rip-off but still an overpriced fad while Lenolt offers the rational homebuyer unswayed by fad or emotion a clear advantage. But drive Lenolt and then South Palo Alto and, more important, compare the infrastructure and quality of life each offers, and if you've got an extra $700,000 or so rattling around, I think I know which neighborhood you'll buy. Then drive Atherton's quiet shady streets and admire its elegant homes on acre-plus lots and if you can find an extra $1,600,000 or so in your sofa cushions, I think you might at least briefly consider Atherton over South Palo Alto.
But that would imply that data do lie, or maybe get confused, or maybe confuse the people who love them, which would imply that the strength-through-numbers data-is-bliss approach to self-empowerment so popular these days is really pretty darn silly, which would set back western civilization 150 years and then we'd be back to reading Jane Austen by flickering candlelight and gosh where's the empowerment, not to mention the bucks and eyeballs, in that?
Why don't we find out?