Arlington 2020: the Cost of Low-Density Housing

This is the second in a series of “Arlington 2020” articles. The first article looked at the number of one-, two-, and three-family homes and condominiums in Arlington, and how that housing stock has changed over time. This article will examine changes in the value of those properties. We’re going to look at “value” through the lens of property assessments, so we should start with an explanation of what property assessments are and how they’re used.

A property assessment is simply the Town Assessor’s best estimate of what a property is worth, based on market values. The assessor’s office inspects properties every ten years; during intervening years, assessments are adjusted based on sale prices of similar homes in a given tax neighborhood. For all practical purposes, assessed values tend to trail market values by two years. In my neighborhood, property assessments are spot on — my house was assessed at $501,000 in 2020; during 2018, sales of similar homes in the neighborhood ranged from $495,000 to $520,000.

Condominiums have a single assessed value, which includes land and buildings. Otherwise, assessed values are broken down into land value, building value, and yard items (e.g., a garage or a shed).

Assessed values are used to determine the tax rate. The assessors page on the town website has calculations in worksheet form, but for all practical purposes, it’s just a division problem. One takes the total tax levy and divides by the sum of all property assessments (in thousands of dollars), and that’s the tax rate. An individual’s taxes are the assessed value of their property (in thousands of dollars) multiplied by the tax rate. If an individual owns (say) 1% of the assessed value in town, that individual will pay 1% of the property tax levy.

The main point is that assessed values are based on market values, but with a two-year lag. Consequently, we can use them as a way to see how home prices have changed over time.

With that background information out of the way, we can look at some numbers. Here’s a graph of the median assessed values for condominiums, one-family, two-family, and three-family homes from 2013 through 2020. (the “median” is a value such that half of the assessments are above, and half are below).

Graph of Median assesed values by year and housing type
yearCondominiumSingle FamilyTwo-familyThree-family

As one would expect, two-family homes are worth more than single-family, and three-family are worth more than two. Condominiums have a lot of variety; they could be half of a duplex, or a single unit in an apartment building. But a general upward trend is clearly evident.

These values are straight out of the assessor’s database, and not adjusted for inflation. The Bureau of Labor Statistic’s inflation calculator shows 12% inflation between 2013 and 2020; the %change is pretty considerable, even if one deducts 12% for inflation.

Next, I’d like to dig further into the 1–3 family assessments, by breaking them down into the value of land vs the value of buildings, and showing how that’s changed over time.

Single-family homes:

Graph of Land and building cost for single family homes, by year.
yearLand valueBuilding valueTotal assessed value

Two-family homes:

Median land and building cost for two-family homes, by year
yearLand valueBuilding valueTotal assessed value

Three-family homes:

Median land and building cost for three-family homes, by year
yearLand valueBuilding valueTotal assessed value

There are several things worth pointing out in these breakdowns.

First, note that the land and building values “jump” a bit between 2019–2020. 2020 was one of our full reassessment years, so I’m willing to attribute this to a periodic course correction. The total increase is generally linear, but the land/building composition has changed.

Second, the median land value for single-family homes is higher than the median building value, for all years between 2013–2020.

Third, most of the increases come from changes in land value. I believe this comes down to location, location, and location. Arlington has a well-respected public school system, and it’s close to universities and tech centers is Cambridge and Boston, and office parks in Lexington, Waltham, and Burlington. City amenities are close at hand.

So what does one do about our rising home prices, and in particular, the rising value of land? The first (and perhaps default) answer is to do nothing. Rising property values are a boon to homeowners who purchased a capital asset (i.e., a house) in the past, and have seen its value appreciate over time. The downside of doing nothing is that each year, increasing housing prices create an ever-increasing income threshold for new residents.

An alternative approach would be to allow more (and smaller) units to be built on each lot. This requires reconstruction or redevelopment, but it allows the cost of land to be amortized among several households. More units/lot means more people and more density, but it reduces the income threshold for buying in to Arlington. (Note that the per-unit cost for three-family homes is lower than the per-unit cost for two-family homes. Similarly, the per-unit cost for two-family homes is lower than the cost of a single-family home).

A third article will look at the distribution of housing prices in Arlington, and how the distribution varies by housing type.

Here is a spreadsheet of data shown in this post.

Arlington 2020: Low Density Housing

I’ve had an annual ritual for the past several years: obtain a spreadsheet of property assessments from the Town Assessor, load them in to a database, and run a series of R computations against the data. I started doing this for a number of reasons: to understand what was built where (our zoning laws have changed over time, and there are numerous non-conforming uses), the relationship between land and building values, the capital costs of different types of housing, and how these factors have changed over time.

I’d typically compile these analyses into a fact-book of sorts, and email it around to people that I thought might be interested. This year, I’m going to post the analyses here as a series of articles. This first installment contains basic information about Arlington’s low-density housing: single-, two-, and three-family homes, as well as condominiums. Condominiums are something of an oddball in this category — a condominium can be half of a two-family structure, part of a larger residential building, or somewhere in between. There’s a lot of variety.

Here’s a table showing how the number of units has changed over time, since 2013.

land use20132014201520162017201820192020
Single Family79847983799180007994799479987999

Arlington’s predominant form of housing — the single family home — has stayed relatively static; we’ve added 15 over the last seven years. The number of condominiums has increased significantly: +585 over seven years. That, coupled with the reduction of two-family homes (-213) and three-family homes (-25) leads me to believe that a fair number of rental units have been removed from the market.

Next, I’d like to look at how these homes are spread across our various zoning districts. (The “Notes” section at the bottom of the post explains what the zoning district codes mean).


A few points to note:

  • R0 is our newest district, which was established in 1991. It consists only of conforming single-family homes.
  • R1 is Arlington’s original (per 1975 zoning) single-family district. It’s predominantly single-family homes, but there are a fair number of two-family homes, and even a few three-families. The presence of condominiums suggests additional multi-family homes (that consist of two or more condominiums)
  • R2 is predominantly two-, and three-family homes. Although three-family homes are no longer allowed in this district, R2 has the largest number of three-families in town.
  • Residential uses are no longer allowed in the industrial (I) districts, but the I districts contain 34 homes. These buildings pre-date the current zoning laws (aka “pre-existing non-conforming”). A good portion of the Dudley street industrial district is a residential neighborhood.

I’m pointing out these conformities (and non-conformities) for a reason. The zoning map (and use tables) dictate what is allowed today, along with specifying a vision for the future. Our zoning bylaw happens to contain a strong statement to this effect: “It is the purpose of this Bylaw to discourage the perpetuity of nonconforming uses and structures whenever possible” (section 8.1.1(A)). Despite the strong statement of intent, it can take decades (if not generations) for a built environment to catch up with the bylaw’s prescriptions.

I’ll finish this post with a breakdown of how condominiums are distributed across the various zoning districts:


The last column (“delta”) shows the difference between 2013 and 2020. The largest increase occurred in the R2 (two-family) district, followed by R6 (medium-density apartments, where most of the increase took place in 2014) and R1 (single-family).

That it will do it for the first installation. In the next post, we’ll look at how the cost (assessed values, actually) of Arlington’s low density housing has changed over the last seven years.

Here is a spreadsheet, containing the various tables shown in this article.


Arlington’s zoning map divides the town into a set of districts, and each district has regulations about what kinds of buildings and uses are allowed (or not allowed). The districts mentioned in this article are:

  • B1 (Neighborhood Office district)
  • B2 (Neighborhood Business distrct)
  • B2A (Major Business District)
  • B3 (Village Business District)
  • B4 (Vehicular-Oriented Business District)
  • I (Industrial District)
  • R0 (Single-Family, large-lot district)
  • R1 (Single-Family Distict)
  • R2 (Two-Family District)
  • R3 (Three-Family District)
  • R4 (Townhouse District)
  • R5 (Low-Density Apartment District)
  • R6 (Medium-Density Apartment District)
  • R7 (High-Density Apartment District)

Arlington’s Zoning Bylaw describes each district in detail (see sections 5.4.2, 5.5.2, and 5.6.2)

When was Arlington’s Housing Built?

The material in this post came from my efforts to learn about when Arlington’s housing was built. The data comes from the town’s 2019 property tax assessments, where I took our nineteen-thousand-and-some-odd homes and apartments and broke them down by housing type and decade built. It’s not exactly a history housing of production, though it is a close approximation. In this analysis, a single-family home built in the 1912 and rebuilt as a two-family in 1976 would show up as two units built in the 1970s. Similarly, a three-family home that was built in the 1924 and later converted to condominiums would show up as three condominiums built in the 1920s.

Here’s the visual summary:

Bar chart showing Arlington housing by decade built.

And here’s a small spreadsheet with the underlying numbers.

My first surprise was at how much we built in the 1920s: just under five thousand units. This was our biggest decade for housing production, and nearly double our second biggest (the 1950s). Another surprise was the 1990s; 132 of our homes were constructed during that decade, which is the smallest number since the 1870s.

What about homes constructed before 1850? There are only 117 of them, and they’re omitted from the data set. I’ve also omitted residential units in mixed-use buildings, since my copy of the assessors data doesn’t break mixed-use buildings into residential and non-residential units.