June 15, 2018

What’s Happened to Properties with Expired Tax Abatements?
Part III
By Kevin C. Gillen, Ph.D.
Houwzer Senior Economic Advisor
This is the third and final paper in a series examining previously tax-abated properties in Philadelphia. The
first part of the series examined their turnover rate in their post-abatement period. The second part
examined what has happened to their values following the expiration of their abatements. This final part
examines the fiscal implications of the abatement, focusing on how much the abatement is actually worth
versus how much the market is actually pricing it.
As noted in previous installments to this series, 10,404 residential properties have seen their abatements
expire since the program’s original inception in 2000. All of these were abated and purchased in the 20002009 period. Of the original 10,404 dwellings that have since seen their abatements expire, only 3,530
have since subsequently sold. And, of these 3,530 units, only 1,175 met the conditions1 listed in the
second installment to this series to be used for further empirical analysis. This constitutes only 11% of all
previously abated units.
According to the program’s proponents, the general rationale for Philadelphia’s tax abatement is that it
helps promote development that wouldn’t happen otherwise by increasing the final value (e.g. price) of
the finished project by a margin large enough to cover Philadelphia’s generally high cost of construction2.
It does this by conferring a significant tax advantage to the property (e.g. very low real estate taxes for 10
years) that increases the market value of the property.
But just how large is this advantage? And, how accurately do buyers and investors value it? This is
important in order to promote the efficient administration and successful continuation of the program
for several reasons. First, if the value that the abatement confers is “excessively” large—in that it results
in above-market returns to real estate investment—then the program is unnecessarily expensive to the
taxpaying public and can be curtailed. Second, if developers, investors and buyers are not accurately
pricing the value of the abatement into the sale and purchase of the finished product, then the program’s
efficiency can be improved via either modifications to the program or better educating the public on its
benefits.
This paper will endeavor to empirically calculate both the value and the price of the abatement for
individual units, and then draw some larger policy-oriented inferences about the program based upon the
results. We begin by defining some terminology:

1

These conditions were: 1) The initial purchase price of the abated unit had to occur within one year of it being granted an
abatement, when the abatement’s benefits were still large; and 2) The subsequent sales price of the abated unit had to occur
either in the year its abatement expired or after that; and 3) Both the original purchase price and subsequent sales price had to
occur under arms-length conditions1; and 4) No unit could transact in between its initial purchase and its subsequent postabatement sale.
2

According to ENR.com, Philadelphia has the fourth-highest cost of development of any city in the U.S.

1

 Let:
AV=the value of the abatement. This is the total dollar value of the foregone property taxes,
adjusted for time.
AP=the price of the abatement. This is the dollar premium that individual investors and buyers
actually paid for newly abated properties. Since it is impossible for anyone to forecast their
property tax bills ten years out with complete accuracy, this is their estimate of AV.
Now, conceptually, the total value of an abated property (TV) is simply the value it would have without
the abatement (P), plus the value of the abatement (AV):
TV = P + AV
In a perfectly efficient world, buyers of abated properties would have a 100% accurate estimate of AV,
and they would pay TV for an abated property. But, they must estimate or guess AV, based upon their
current expectations of future property tax bills. As a result, AP can (and almost always does) differ from
AV. So the actual price they pay for an abated property (represented by TV*) is calculated as follows:
TV* = P + AP
Since P is the same in both equations, then when the recorded purchase price TV* differs from the
property’s equilibrium price of TV, it can only be because the buyer’s estimated value of the abatement
AP differs from the true value of the abatement AV.
The values of each of the above variables are computed for each previously abated property in
Philadelphia. AV is computed as the present discounted value of each property’s tax bill over the ten
years it has an abatement by applying the city’s taxation formula to the property’s assessed and exempt
values in its purchase year and sale year3. AP is then computed as the difference between an abated
property’s actual purchase price (TV*) and its counterfactual purchase price (P), where P is determined
by discounting the property’s post-abated sales price back to the year of its original purchase utilizing a
price index for residential properties in Philadelphia.
Since the formal math behind this exercise may seem somewhat complicated, we show how this is done
for just one of the 1,175 properties that meet the necessary conditions for this analysis: consider 503
Governor’s Court in the exclusive and historic townhome community of Naval Square. This is a Toll
Brothers project located in the revitalized neighborhood of Gray’s Ferry at the edge of Center City. The
dwelling is a 3.5 story brick townhome that has the legal status of being a condo, due to its location in a
gated HOA community.

3

We recognize that the city went through several re-assessments, changes in its assessment practice and changes in its
statutory real estate tax rate from between 2000 and 2018. The appropriate assessed value, taxation formula and tax rate were
applied for the respective years. The author is grateful to Marisa Waxman of Philadelphia’s Revenue Department for providing
some guidance in identifying the correct taxation formula for past years.

2

 Naval Square

This unit was purchased in 2007 for $679,975. Since it was 100% new construction, the value of the
structure was fully abated and the owner paid taxes only on the land. According to OPA’s property file
from 2008, the total assessed value of the property was $540,000, with $15,900 of this being allocated to
the value of its land. Because the structure was fully exempt, a whopping 97% of the property’s value was
spared from taxation due to its abatement. The property’s tax bill in its first year was a mere $4204. In
the absence of the abatement, it would have been $14,2805.
In 2016, the last year of the property’s abatement, its taxable value had risen to $564,300, with $56,430
(10%) being allocated to land. Its tax bill in this year was $790. Minus the abatement, it would have been
$7,8996.
The abatement’s value is computed by taking the present discounted value7 of the abated portion of the
property’s annual tax bill in each year of the ten years it was abated. In year one of the property’s
abatement, the tax savings to the owner were $14,280-$420=$13,860. In year ten of the property’s
abatement, the tax savings to the owner were $7,899-$790=$7,109. The total value of the abated income
stream is then computed as follows8:
10

𝐴𝑉 = ∑
𝑡=1

$13,860
$7,109
+ ⋯+
= $87,124
𝑡
(1 + 0.05)
(1 + 0.05)10

4

$420=$15,900 x 0.32 x 0.08264. This was the city’s real estate taxation formula in 2008, before the implementation of the
Actual Value Initiative that eliminated fractional assessment and also decreased the tax rate.
5 All revenue numbers are rounded to the nearest dollar.
6 $564,300 x .013998 = $7,899. This was the city’s real estate taxation formula in 2016.
7 A 5% discount rate was used.
8 The terms representing the years between the first and last year of the property’s abatement are omitted from the formula
for brevity’s sake, but they are included in the calculation.

3

 Adjusted for the time value of money, the owner of the unit was spared a total of just over $87,000 in
taxes due to the property’s abatement status. So, in a perfectly rational world with complete foresight,
the dwelling would have commanded an $87,000 premium in its first year of purchase, due to the
abatement; i.e. AV=$87,000. Conversely, absent the abatement, the property could be expected to sell
for $87,000 less than what it actually sold for. This implies that the tax advantages conferred by the
abatement constituted 13% of the property’s total market value9. Note that this is very close to the
average of 12% that was estimated by the City Controller’s recent report analyzing the value of
Philadelphia’s abatements10.
While it is reasonable to expect a buyer to pay a premium for a property with favorable tax treatment, it
is equally unrealistic to expect them to be able to predict their future tax bills with complete accuracy,
which would be required to accurately arrive at the $87,124 value. However, it is possible to be able to
estimate what they thought their future tax bills would be by computing what the owner actually paid for
the abatement; i.e. AP. To do this, a counterfactual purchase price was computed by discounting the
property’s post-abatement sales price back to its original date of purchase, and then taking the difference
between the actual purchase price and the counterfactual price as the abatement’s price AP. Since the
property’s post-abatement sales price represents its market value once the abatement has expired, then
adjusting this price for the general fluctuations in local property prices over time yields a price that the
property would have originally sold for if it did not have an abatement.
The previous chapter in this report estimated a repeat-sales property price index for Philadelphia using all
paired sales between 2000 and 201711. The index is analogous to a “Dow Jones” index for Philadelphia’s
housing: the percent change in the level of the index between any two periods reflects the average
percent change in the value of local residential property values. A chart showing the index is below:

9

13%=$87,000/$679,975.
An Analysis of Tax Abatements in Philadelphia, Office of the Controller, City of Philadelphia, April 23, 2018.
11 See: https://houwzer.com/blog/what-happens-when-philadelphia-tax-abatement-expires-part-2 for details.
10

4

 Philadelphia Residential Property Price Index 2000-2017
2000=100
250
Abatements in Effect

Abatements Expired

225

Price Index Value

200
An unabated dwelling
that sold for $775,000
in 2017...

175

150
...would likely have sold
for $545,511 in 2007.

125

100

75
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

Year

The value of the index in the year that the post-abated property sold (2017) is 239.5812. The value of the
index in the year that the same property was originally purchased (2007) is 168.64. To obtain the
counterfactual price (P) of the property, the 2017 sale price is deflated using the value of the index as
follows:
P = $775,000 x (168.64/239.58)
= $545,511
This price is ~30% less than its subsequent sales price, which is exactly the same as the percent change in
the level of the index between 2007 and 2017; i.e. average residential property prices in Philadelphia rose
30% from 2007 to 2017. So, to predict the likely purchase price of a property in 2007 when only its 2017
sale price is available, the 2017 sale price is simply deflated by 30%.
The $545,511 price represents our best counterfactual estimate of what this property would have been
purchased for in 2007 if it didn’t have an abatement. However, the actual purchase price of the property
in 2007 (with an abatement) was $679,975. The difference between these two numbers, $134,464,
represents the actual price (AP) that the buyer “paid” for the abatement.
Since the actual value of the abatement was $87,124, these numbers imply that the buyer overpaid for
the abatement by $47,340, which added a 20% premium (rather than a 13% premium) to the base price
of the property. Such a higher premium could be rationalized if the buyer expected to see their assessed
12

The actual index values were computed to 7 decimal places. So, for brevity’s sake, some rounding is involved in reporting
these numbers here.

5

 value (and hence, tax bill) increase by a faster and higher amount than it actually did. The 13% premium
that represents the actual value of the abatement reflects the actual appreciation rate of its assessed
value, which was an average of 1.4% per year. Based upon the 20% premium that the buyer paid, they
would have expected the property’s assessed value to increase by significantly more than 1.4% per year.
The above exercise was carried out for all 1,175 abated properties that have since seen their abatements
expire and have also since sold under arms-length conditions13. For each property, four variables of
interest were computed:
Abate_Value (AV) = the $ value of the abatement
Abate_Price (AP) = the $ price premium that the buyer paid for the abated property
Pct_Abate_Value = the % premium that the abatement should add to the original purchase price
of the property. It is computed as: Abate_Value/Purchase Price
Pct_Abate_Price = the % premium that the abatement actually added to the original purchase
price of the property. It is computed as: Abate_Price/Purchase Price
The following chart compares the distribution of Abate_Value to Abate_Price across all post-abated
properties, with red representing Abate_Value and blue representing Abate_Price:

Summary Statistics
Statistic Abate_Value Abate_Price
Min.
$2,674
-$80,142
Pct. 1
$4,658
-$40,557
Pct. 25
$23,895
$32,582
Median
$34,729
$67,632
Mean
$38,144
$80,148
Std. Dev.
$21,781
$70,749
Pct. 75
$48,870
$116,145
Pct. 99
$124,804
$333,180
Max.
$157,592
$420,567

13

This latter condition is necessary in order to obtain a post-abatement market price, and hence a counterfactual (non-abated)
price. As noted in a previous installment to this series, most abated properties remain in the hands of their original buyers,
even after their abatements expire.

6

 The following inferences can be made from the data presented in the chart:


The price that buyers paid for the abatement typically exceeded the actual value of the
abatement:
o The price premium that buyers paid for their abatement was greater than the actual value
of the abatement for 70% of these post-abated properties.
o The median abatement price is $67,632, while the median abatement value is $34,729.
o The average abatement price is $80,148, while the average abatement value is $38,144.
o The 25% lowest-priced abated properties had an abatement price of $32,582 or less, but
an abatement value of only $23,895 or less.
o The 25% highest-priced abated properties saw their buyers pay a price premium of
$116,145 or higher for their abatement. But, the actual value of their abatement started
at just $48,870.



The variation in what buyers paid for the abatement is significantly greater than the variation
in the actual value of the abatement:
o It is easily discernible that the red bars (Abate_Value) exhibit a significantly tighter cluster
than the blue bars (Abate_Price).
o The standard deviation of abatement values is $21,781, while it is a much larger $70,749
for abatement prices.
o This result shouldn’t be surprising, since abatement values can be calculated with the
benefit of perfect hindsight (i.e. with perfect precision), while abatement prices are
computed with highly imperfect foresight (i.e. the buyer must guess at what their future
assessed values and tax bills will be).



Abatement prices can have negative values, while abatement values are always positive:
o Mathematically, this occurs when the property’s (post-abatement) sales price is
discounted back to the year of its original purchase, and its resulting imputed value is less
than the property’s actual purchase price.
o However, abatement values are always positive since they are simply the present
discounted value of a positive cash flow; i.e. tax payments.
o An empirical examination of these properties with negative abatement prices revealed
that most of them were purchased in the 2000-2003 period, before the significant
inflation of the housing boom/bubble. But, they were also disproportionately sold in the
2009-2012 period, after the bubble had burst and prices were in a significant deflationary
cycle. Hence, when the post-bubble low sales prices were discounted even further to
obtain pre-bubble purchase prices, a negative difference between the implied purchase
price and actual purchase price resulted.
o Thus, rather than characterizing the owner of these properties as “shrewd buyers”
(negotiating a purchase price that was less than the abatement’s value), it would be more
appropriate to characterize them as “panicky sellers” (they sold their unit at a significant
discount at a time when the market was also in a downturn).

7

 To examine the percent premium that the abatement contributes to property prices the next chart
compares the distribution of Pct_Abate_Value to Pct_Abate_Price across all post-abated properties.
These numbers were computed by simply dividing Abate_Value and Abate_Price into the original
purchase price of each property:

Summary Statistics
Statistic Pct_Abate_Value Pct_Abate_Price
Min.
0.0%
-24.0%
Pct. 1
2.0%
-24.0%
Pct. 25
9.0%
10.0%
Median
11.0%
20.5%
Mean
10.6%
21.5%
Std. Dev.
4.9%
17.1%
Pct. 75
12.0%
33.0%
Pct. 99
26.0%
62.0%
Max.
87.0%
72.0%

The information in the chart gives the following results:


The price premium that buyers paid for the abatement was typically greater than its actual
value:
o The typical (i.e. median) abated property carried a price premium of 20.5%, while the
actual value of the abatement should only add 11% to the value of the property.
o Similarly, the average price premium was 21.5% while the average value was only 10.6%.
o The overwhelming majority of abated properties had abatement values ranging from 1%
to 26%, but the actual price premiums that these properties were purchased for ranged
as high as 72%.

To empirically quantify the relationship between the prices that buyers paid for the abatement versus the
abatement’s actual value, a regression of the former against the latter was computed. The next chart
shows a scatterplot of abated prices against abated values, with the fitted regression line showing the
explicit mathematical equation between these two variables fitted to the data:

8

 The regression yields the following findings:


Despite abatement prices generally exceeding abatement values, the correlation between the
two is still very strong. The R-squared of the regression (a measure of the correlation’s strength)
is nearly 52%, out of a possible maximum of 100%. And the t-value of the regression line’s slope
(a measure of the validity of the correlation) is 32.28, which easily exceeds the threshold of 1.96
to be considered “statistically significant”. These results would be considered by most
statisticians to be quite strong for a simple bivariate regression.



For every $1 in a property’s abate values, buyers paid an average of $1.75 in a higher price for
that property. This can be inferred from the slope of the regression line. Its value of 1.75438
implies that every $1 increase in an abatement’s intrinsic value is correlated with an increase of
$1.75 (rounded) in abatement prices.

Lastly, using both the historical and current assessed values of these properties, we can calculate what
revenue they have and are generating in real estate-related taxes14:

14

This assumes a 100% collection rate; i.e. no delinquencies.

9

 

While their abatements were in effect, these approximately 10,000 properties have collectively
generated $103.7m in tax revenues to the city15, most of which is from their (unabated) land
values.



In 2017, when all of their abatements had expired, these properties generated $44.8m in tax
revenues.



With their 2018 assessed values, they are projected to generate $50.2m in tax revenues.



If both the the recently released assessments and proposed real estate tax rate of 1.4572% are
adopted, then they will generate $52.3m in tax revenues in 2019.



Due to the overpayment for the abatement, the city is estimated to have reaped a $15m windfall
in additional transfer taxes during the 2000-2009 period when these properties were first
purchased16. If this same condition also applied to currently abated properties, this number
would be approximately double that.

The numbers clearly show that abatement has both a very real and very significant value in the
marketplace, typically adding 11-20% to a property’s value. And, the improvement/construction of these
properties has contributed significantly to the city, with over $100m to date and approximately half of
that in each year since their abatements have expired.
However, what are the larger implications of these results for the abatement program and its fiscal
costs/benefits to Philadelphia? In particular, what is to be made of the result that buyers have typically
“overpaid” for the abatement?
From the short-term and narrow perspective of city government, having the abatement priced at a
number greater than its intrinsic value may not necessarily seem like a bad thing. It results in both real
estate transfer tax revenues and real estate tax revenues being higher than they would be otherwise,
since both are indexed to market values. However, this myopic view ignores several key facts:


First, every unnecessarily additional dollar spent on the abatement by Philadelphia households
is a dollar that has been wasted and can’t be spent elsewhere. This results in both a lower
standard of living for these households due to foregone purchases on other items, as well as lower
tax revenues to the city if these purchases would be subject to city taxes; e.g. the sales tax. So,
additional revenues from real estate taxes are offset at least partially by lower revenues from
other taxes.



Second, paying an excessive premium for the abatement can have distortionary effects on both
the overall real estate market and assessed values. In areas with high concentrations of abated
properties, such as Center City and the revitalized neighborhoods surrounding it, the general level
of real estate values will be higher than what they should be since the premium placed upon

15

This number covers the years 2000-2016, and is not an annual number.
This was computed by summing the difference between Abate_Price and Abate_Value across all 1,017 properties and
multiplying the resulting number times the city’s (then) 3.0% transfer tax rate. This number was then multiplied times ten to
adjust for the fact that these 1,017 sales represent only ~10% of all properties which received their abatements in the 20002009 period.
16

10

 abated properties will skew average prices upwards. This can mislead households and investors
who are buying non-abated properties to also pay a higher price than they would otherwise,
which results in the same loss in both personal welfare and city tax revenues that the previous
paragraph pointed out. Longer-term, it can also result in a more inequitable and inefficient
distribution of property taxes, since both abated and non-abated households in these
neighborhoods will have higher assessed values than they should.


Thirdly—and importantly—it should be pointed out that all of the analysis in this paper used
property sales from 2000-2009, when property prices were rising very rapidly, and that this may
explain why most abatement prices surpassed their values. The primary economic justification
for paying a relatively high price for the abatement is that it is a shield against future property tax
increases. If the buyers of these properties expected their assessments—and hence, tax bills—to
continue to rise quite rapidly in the years following their purchase, then this could lead them to
value the benefits of the abatenment greater than its intrinsic worth. Since the majority of these
abated properties were built and constructed near or at the peak of the housing bubble in 20052008, it seems reasonable that many of these buyers were subject to the “irrational exuberance”
that marked the spirit of the times. If so, then the fact that abatement prices exceeded abatement
values may be just a transitory and outlying artifact of that particular period, with minimal
implications for the current debate over the abatement program’s future.

Lastly, it could be quite tempting—but also quite possibly incorrect--to interpret these results as
evidence that the abatement program could be scaled back with minimal adverse effects on housing
production and sales in Philadelphia. After all, if the abatement is generally overpriced, then curtailing
its benefits would seem to intuitively work to bring its market price back into line with its intrinsic value.
But, any reduction in the program’s scope or duration will also result in a decline in the abatement’s
intrinsic value in addition to a decline in its market price. So, the gap between price and value would still
persist. Moreover, this reduction would drop the overall return on housing development in Philadelphia.
If this resulting return is lower than a developer’s typical cost of capital (i.e. the return they must offer
lenders and investors to raise financing for new projects), then this would result in a dramatic decline in
new housing production and investment in Philadelphia, even if the abatement still had a positive market
value. Hence, these results should be understood and applied with caution.

11