The Many Dimensions of Change in Boston: A Data Story Series on the Building Permits and Property Assessment Datasets


Day 1: “Seeing” Investment in Boston’s Neighborhood’s through Building Permits


Building permits offer a unique window into patterns of investment across Boston–literally capturing the physical work being done by property owners to expand, improve, and newly construct the buildings of the city. In recognition of this fact, BARI has created a series of ecometrics (i.e., descriptors of a neighborhood) that quantify the investment in every census tract in Boston on three different dimensions: renovations, additions, and new constructions. These categories are developed for each parcel in the city based on the types of permits that it pulled (see documentation for the Building Permits data set  for more information). We release these measures annually for census tracts in multiple forms, including total count of parcels undergoing each type, percentage of parcels undergoing each type, and total value generated by each type.


The 2017 measures of investment tell us where the “action” has been in the past year on two dimensions. First, we can combine additions and renovations because they both reflect alterations to existing structures, whereas new constructions do not. Further, they are highly correlated across tracts (i.e., tracts with lots of renovations also have lots of additions).[1] We will look at them in terms of percentage of parcels undergoing alterations in a neighborhood; across the city, 17% of parcels experienced a renovation or addition in 2017. New constructions, however, are far rarer–only 410 were undertaken in the year, less than .5%–for which reason we analyze them as raw counts in each neighborhood.


Using BostonMap, we can visualize the distribution of projects altering existing structures and constructing new buildings across the city. Laying them side-by-side, we see that they are highly dissimilar. Neighborhoods where property owners are improving their current property are not the same neighborhoods where larger investors are building new structures. In fact, they are in many cases opposites.[2] They clearly tell different stories, but what are those stories? The story of new constructions is probably a little simpler as such projects are isolated in certain regions. Some of these are oft discussed in the media, like the constant development of the Seaport District of South Boston. Others, like the Harvard III Commons in western Dorchester, are probably less well-known.



Figure 1. Investment in Boston tracts in 2017 measured in terms of (a) the proportion of parcels experiencing a renovation or addition and (b) the number of newly constructed buildings.


The map of additions and renovations shows a more concrete set of trends, however, with a strong concentration of such work in neighborhoods in and around the core of the city–Downtown, Back Bay and Beacon Hill, Fenway/Kenmore, and South End. This is not the whole story, however, as there are other hubs of activity, like eastern Charlestown. We can isolate these localized hubs by removing the impact of average neighborhood investment.[3] This gives us the additional map below, in which red census tracts are experiencing fewer additions and renovations than other parts of the neighborhood, and blue tracts are experiencing more. On this map, we see a few hubs of interest, including: the coasts of both east Charlestown and west East Boston along the Mystic River; the Dudley Square area of Roxbury; and, to a lesser extent, the west side of Dorchester Ave. near Codman Square and along Centre St. near the Roslindale-West Roxbury border.



Figure 2. Proportion of parcels experiencing renovations or additions above (blue) or below (red) average for each census tract’s planning district in Boston in 2017.


What accounts for these additional areas of activity? Interestingly, the answer is not what one might expect.[4] Property owners are not investing more in richer or poorer census tracts, nor in tracts with a particular racial composition (e.g., predominantly White or Black). There is however, one feature that seems to matter, hinting at the way investment in Boston is motivated right now: neighborhoods with higher levels of homeownership had fewer additions and renovations in 2017.[5] This resonates an ecological perspective known as invasion-succession that argues that change occurs in neighborhoods when there are opportunities for new elements to enter. Neighborhoods with fewer homeowners likely operate more as a pure market, with property owners buying, selling, and renovating as they see opportunities for business gain and have fewer emotional and personal considerations that might interfere. Similarly, local communities of renters may be less likely to push back against building efforts than homeowner neighbors would be.


This amounts to one short story of how investment has occurred across Boston in 2017, as well as an illustration of what can be done with BARI’s ecometrics of investment and growth. These annual ecometrics go back to 2010, and we encourage others to do additional tests of the geographic and longitudinal trends of investment in the city.


[1] r = .66, p < .001 for tracts with greater than 75 parcels (limited to avoid undue influence of outliers).

[2] r = -.31, p < .001.

[3] Calculated by extracting tract-level residuals from a linear model with Boston Planning and Development Agency’s Planning Districts as the sole set of predictors.

[4] Results that follow based on a multiple regression.

[5] 𝛃 = -.30, p < .01 in linear model controlling for planning district, income, and racial composition.

Day 2: Residential Growth in Boston: Evidence from Building Permits


The City of Boston’s building permits dataset allows us to see both how much development is happening and what types of development this is. Overall, building permits are needed for construction of new buildings, demolition of and addition to existing structures, and renovation. The occupancy type of the buildings for which these permits are granted can tell us more about the character of development in Boston. Over time, the overall number of building permits granted by the city has increased – by about 25% since 2011. The breakdown of occupancy type shows yet another dynamic: while the majority of these permits are for residential buildings, the trend in this breakdown over time shows that residential development has not kept pace with other categories of occupancy for building permits – though some of this may be driven by the increasing number of uncategorized permits in the data since 2015.



What kind of residential development is happening, though? Some residential development is for single-family houses, while other development may be for large apartment complexes. The building permits data also allow us to unpack this aspect of development in Boston. Since 2015, the majority of permits for residential development have been for single-family and two-family housing, as shown below.



One of the things that might be limiting this residential development, however, is the lack of vacant land or space for larger building (such as those that might fit more than a single family). The dataset can give us a hint as to why that might be happening – and why it also restricts where new residential construction can occur. The building permits data show what type of work they are for – some of them are for new construction, but some are for additions and demolition of existing structures or renovations. Overall, most of the building permits in these data are for renovation, but there is still some percentage of the permits that are for new construction. BARI’s ecometrics, posted on the BARI Data Portal, indicate these type of development in the geographic aggregate tabular files and shapefiles.


The below maps demonstrate where the new construction is occurring in the city during the years of 2016 and 2017.  Census tracts with a high proportion of all parcels being permitted for new construction are plotted in blue, while those with fewer of their parcels being developed with new construction plotted in rec. As these maps show, the development in Boston is fairly mixed across geographies, with some census tracts with higher amounts of new development immediately next to census tracts with lower amounts of development. One such area is Boston’s Seaport, where in 2016 there was a high proportion of parcels with new construction in the tract immediately next to downtown, but a low proportion in the tracts towards the harbor and Southie. In 2017, however, this changes – the areas of Southie next to the harbor have more permits issued in them for new construction, and the area of the Seaport immediately next to downtown and the Fort Point Channel continues to have a high degree of new construction.



You can access the building permits dataset on the Boston Data Library, and explore these metrics visually on BostonMap.


Day 3: A Neighborhood’s Contribution to Property Assessment


How much does your neighborhood affect your property’s value? Every year, local governments assess the value of a property for tax purposes [1][2]. The typical tax assessment determines a property’s value based on key physical characteristics of the property and building that are used for comparison against other similar properties. Yet property values for similar property types can differ across neighborhoods (see Table 1). The BARI team set out to identify the unique effect Boston’s neighborhoods have on assessed property values using the recently released 2018 Property Assessment dataset from the City of Boston.



We used census tracts to approximate neighborhoods to run a multilevel (or mixed effects) model. Multilevel models allow us to analyze nested or grouped data (like properties in a neighborhood) and disentangle the effects at each of these levels. We were then able to estimate the value of property in each neighborhood while accounting for critical differences between properties, including the lot size, the area of the building, the amount of “living space” in the building, the number of floors in the building, the age of the building, when the building was last renovated, and the type of land use [3].


We plotted the effects of each census tract on property value (see Table 2). This plot shows how much the neighborhoods’ estimated value deviates from the city’s estimated mean property value when we account for all pertinent property-level variables . The farther the point is from zero (whether negative or positive), the greater effect the neighborhood has on the assessed property value. This plot is another way of showing how different neighborhoods have different effects on property values.



We also mapped the neighborhoods’ effects on property values in Boston (see map). Similar to Table 2, this map uses a color scale to show the degree of deviation among the census tracts. Red and Blue demonstrate the extreme effects the neighborhood has on property values while the yellow represents deviations that are close to zero or no effect. Unsurprisingly, the Financial District, Back Bay, Fenway/Kenmore, and the South End have positive effects on the value of their properties, while areas of Roxbury, Dorchester, and Mattapan are seeing more negative effects to their property values. Interestingly, just about the entirety of Allston/Brighton is close to or at the city’s estimated average value, while two islands of purple in Jamaica Plain and East Boston are seeing disproportionately higher effects on their property values than their surrounding neighbors. It is also interesting to note, that many of the orange census tracts correspond with areas reporting recent cases of gentrification: Roslindale, Ashmont, Savin Hill, Roxbury Crossing, and East Boston. This map is available on the Boston Data Portal where it can be combined with other layers detailing a wealth of information about the City of Boston.




[1] Fisher, Glenn. “History of Property Taxes in the United States”. EH.Net Encyclopedia, edited by Robert Whaples. September 30, 2002. URL

[2] Oates, Wallace E. “Local Property Taxation: An Assessment”. Land Lines, May 1999. URL

[3] It should be noted that when compared with the intercept model, our model is statistically significant at the 0.000 level (χ2(11)=95714, p<2.2×1016)


Day 4: Using Property Assessment Data to Calculate Changes in Properties’ Land Value and Floor Area Ratios


Boston recently released its annual Property Assessment data, which provides detailed information on properties in Boston. The data set can be downloaded here, and documentation describing what is in the data set is available

With the release of this dataset, there are many different questions that researchers and local practitioners can pursue at different geographic scales. Here, we focus on changes in properties’ land value and floor area ratios aggregated at the census tract level.


Improvement to Land Value Ratio (ILVR):
To calculate this ratio, we divide the building value by the land value for each
parcel. Values lower than one indicate that the building value is larger than the land value, and thus, the land parcel is underutilized. We found these values for all properties in 2017 and 2018. We then calculated the percent change between the two years for each property. We aggregated these values to the census tract level by taking the mean of the changes for all properties within a census tract. The Improvement to Land Value Ratios are shown in Figure 1 below. The darker blue census tracts had at least a 10% increase in its average value; conversely, census tracts colored white indicate a decrease in its value. As we can see, there has been an increase in the ILVR over the last year in areas that are developing more rapidly (such as, South Boston, East Boston, and Hyde Park), while there was a reduction in ILVR in areas such as Roxbury and Back Bay and North End.


Figure 1: Changes in Improvement to Land Value Ratio between 2017 and 2018 in Boston


Floor Area Ratio (FAR):
The Floor Area Ratio is the total floor area of the property divided by the land area of the parcel. A FAR equal to one indicates that total floor area (including multiple levels in a building) is equal to the lot area. For example, a FAR equal to five indicates that the total floor area is five times larger than the lot area. Higher values are more common for large commercial or residential buildings that have multiple floors. We similarly calculate change in FAR as we did for ILVR. These results are shown in Figure 2 below, where darker green census tracts had larger changes in their average floor area rations between 2017 and 2018. While most of the census tracts in the city experience a no change in the Floor Area Ratio, there was an increase in several areas, such as Mission Hill, East Boston and Allston, which are areas that do not have many tall buildings. However, it is interesting to note the drop in FAR in Downtown and South Boston Area: this is a consequence of the high building density and the fact that few new buildings were added.


Figure 2: Changes in Floor Area Ratio between 2017 and 2018 in Boston


These examples illuminate different patterns of development throughout Boston. Moreover, these values can be calculate at finer geographic granularity and then linked with additional data sources available on the Boston Area Research Initiative’s Data Library. Doing so enables researchers and community leaders to pursue their specific research questions related to Boston.

Published On: April 19, 2018 |
Facebook Twitter Google Print Friendly and PDF