April 10, 2018 What s Happened t Prperties with Expired Tax Abatements? Part II By Kevin C. Gillen, Ph.D. Huwzer Senir Ecnmic Advisr This is the secnd in a series examining previusly tax-abated prperties in Philadelphia. The first part f the series examined their transactins vlume in the pst-abatement perid. This secnd part examines what has happened t their market values fllwing the expiratin f their abatements. Since the current abatement prgram s inceptin in 2000, 10,404 residential prperties have seen their abatements expire. All f these were abated and purchased in the 2000-2009 perid. Because the key ratinale fr the prgram is that it is an incentive t imprve and expand Philadelphia s stck f real estate and thus grw its tax base it is reasnable t examine what has happened t the values f these prperties nce their favrable tax treatment has expired. If their values have drpped significantly, then this prvides supprt t thse critics f the abatement wh have asserted that the prgram s lng-term benefits are lw relative t its high shrt-term csts. Alternatively, if abated prperties have held their value r even grwn in value, then this prvides supprt t the prgram s supprters, wh have cntended that the prgram s shrt-term csts are mre than ffset by the lng-term benefits f a permanently expanded tax base that wuld nt have ccurred but fr the abatement. This paper will endeavr t prvide sme empirical analysis t infrm this debate. It is first necessary t identify thse pst-abated units that sld under reasnable market cnditins, and hence whse sales prices reflect reasnable market values. Of the riginal 10,404 dwellings that have since seen their abatements expire, nly 3,530 have since subsequently sld. Hwever, in rder fr this data t be useful in effecting an accurate analysis f hw the abatement may affect the value that buyers may place n it, the transactins f these abated dwellings had t meet all f the fllwing cnditins: 1) The initial purchase price f the abated unit had t ccur within ne year f it being granted an abatement, when the abatement s benefits were still large; and 2) The subsequent sales price f the abated unit had t ccur either in the year its abatement expired r after that; and 3) Bth the riginal purchase price and subsequent sales price had t ccur under arms-length cnditins 1 ; and 4) N unit culd transact in between its initial purchase and its subsequent pst-abatement sale. Of the riginal 3,530 units that transacted mre than nce, nly 1,175 met all f the abve cnditins fr further analysis. This cnstitutes nly 11% f all previusly abated units, which may seem like an 1 Sheriff sales, nminal sales, blanket sales, bank sales and inter-family transfers are nt cnsidered arms-length transactins. Such sales were drpped frm the dataset used in this analysis. 1
unexpectedly small sample t bth critics and prpnents f the abatement prgram. The fllwing map shws the lcatin f these 1,175 previusly abated dwellings: The Lcatin f Previusly Abated Residential Prperties in Philadelphia The greatest cncentratin f pst-abated prperties is in and arund the dwntwn area f Greater Center City. Hwever, (perhaps surprisingly t critics and skeptics f the abatement) there are als significant cncentratins f frmerly abated residences in University City, Nrthwest Philadelphia and Nrtheast Philadelphia. We begin ur analysis by prviding sme general summary statistics n the transactin prices f previusly abated prperties bth befre and after their abatements expired. The fllwing table gives sme summary statistics n bth the initial purchase prices and subsequent (pst-abatement) sales prices f these 1,175 dwellings: Summary Statistics n Prices f Previusly Abated Units Pst-Abatement Initial Purchase Sale %Change #Sales 1,175 1,175 N/A 25% Quartile Price $252,000 $255,000 1.2% 50% Median Price $347,000 $360,000 3.7% Mean Price $400,000 $425,000 6.3% 75% Quartile Price $471,000 $505,000 7.2% 2
In general, it can be bserved that the prices f frmerly abated units generally increased between their initial purchase and their subsequent pst-abatement sale: The lwest-priced 25% f previusly abated units were purchased at a price f $252,000 r less, but then sld fr a price f $255,000 r less. The median (50%) purchase price f an abated prperty between 2000 and 2009 was $347,000. But after their abatements expired, and they sld in the pst-2009 perid, their median sale price was $360,000. The average purchase price f an abated prperty was $400,000. But the average sales price f these same prperties in their pst-abatement perid was $425,000. The highest-priced 25% f previusly abated units were purchased at a price f $471,000 r mre, but they then later sld fr a price f $505,000 r mre, fllwing expiratin f their abatements. Acrss all previusly abated prperties, the median price change was $8,800 and the mean price change was $25,600 2. Lastly, the higher a prperty s initial purchase price, the greater the typical price appreciatin it experienced, in bth dllar and percent terms. The bttm 25% f previusly abated prperties experienced a typical price gain f just $3,000 (r 1.2%) in the 10+ years between their riginal acquisitin and their pst-abatement sale. By cntrast, the tp 25% experienced a typical price gain f $34,000 (r 7.2%) during the same perid. Althugh these summary statistics indicate that these pst-abated prperties cllectively experienced psitive price appreciatin, the same may certainly nt be true f individual prperties. T examine t what extent this may r may nt be true, the dataset f 1,175 sld units was divided int Gainers and Lsers, depending n whether the difference between the riginal purchase price and pst-abatement sale price was psitive ( Gainers ) r negative ( Lsers ). The fllwing bar chart shws the number f sales in each categry during the pst-2009 perid, by year: 2 The price change fr each prperty was cmputed as the sale price minus the riginal purchase price. The median (r mean) price change is nt the same as change in the median (r mean price) because the difference in medians (r means) is nt the same thing as the median (r mean) difference). This is because the distributin f prices is nt perfectly symmetric, and skewness in the data causes these numbers t diverge. 3
The blue bars indicate gainers while the range bars indicate lsers. Fr example, in 2010 (the first year in which previusly abated prperties began t sell), 13 sld fr a psitive gain while there were 0 sales at a lss. In general, it can be nted that the gainers utnumber the lsers, but nt by especially large margins: In seven f the eight years since abated units have returned t the tax rlls at full value, the number f gainers has utnumbered the number f lsers. Acrss all eight years, 637 units have sld at prices higher than their riginal purchase price, while 538 have sld at less than their riginal purchase price. Hwever, the rati f gainers t lsers is nt especially large: 54% f sales were fr an abslute gain, while 46% were fr an abslute lss. Hence, the data indicate that the price f mst abated prperties increased between the perid f their riginal sale and in their subsequent sale after their abatement expired. Hwever, it shuld als be cnsidered that this increase in value is in abslute terms, and nt relative terms since it des nt adjust fr verall mvements in dwelling prices during that perid. First, it shuld be nted that there was enrmus vlatility in huse prices during the 2000-2017 perid since bth the largest husing bubble and then the deepest pstwar recessin ccurred during this perid. Simply cmputing the number f increases v. decreases withut adjusting fr this vlatility risks versimplificatin f the issue. Secnd, If the general level f huse and/r cnd prices in Philadelphia 4
increased by a larger margin than value f abated prperties during a given perid, then this culd reasnably be cnsidered a relative lss, despite being an abslute gain. Cnversely, if the value f abated prperties fell by less than verall huse prices during a given perid, then that culd be cnsidered a relative gain fr abated prperties, despite being an abslute lss. Just as equity fund managers evaluate the perfrmance f their particular prtfli by cmparing it t the verall perfrmance f the stck market, s t is it fair t cmpare the perfrmance f abated prperties t the verall perfrmance f the husing market. The fllwing chart cmpares the median prices f all previusly abated prperties t the median prices f nn-abated dwellings (huses+cnds) in Philadelphia in tw different perids: 2000-2008 (when the abatements were in effect), and 2009-2017 (after the abatements had expired). T ensure a clean applest-apples cmparisn, nly huses and cnds that met the same general criteria as pst-abated prperties were included this analysis 3. $400,000 $350,000 $300,000 $347,000 Median Dwelling Prices: Pre- and Pst-Abatement $360,000 Purchase Price (2000-2008) Sale Price (2009-2017) $250,000 $200,000 $150,000 $156,000 $100,000 $76,000 $50,000 $0 Pst-Abated Prperties Citywide In general, the price f abated prperties increased by less than husing prices citywide during this perid: Abated prperties were purchased fr a median price f $347,000. After their abatements expired, they sld fr a median price f $360,000; a 3.7% increase. 3 Only huses and cnds that sld twice in the 2000-2018 perid and under arms-length cnditins were included in the data sample fr this analysis. 5
During 2000-2008, the median purchase price f all nn-abated dwellings in Philadelphia was $76,000. After 2008, the median sales price f these same prperties was $156,000; a 105% increase. Althugh previusly abated prperties are generally much higher-priced than nn-abated prperties, the value f abated prperties appreciated by much less than the value f nn-abated prperties fllwing their riginal purchase in bth dllar terms ($13,000 v. $80,000) and in percent terms (3.7% v. 105%). While these numbers may indicate an aggregate relative lss in value, they d nt prvide any insight int the number f individual gainers and lsers. T d this, it is necessary t mark t market each previusly abated prperty by cmparing the percent change in its riginal purchase price and subsequent pstabatement sale price t the verall percent change in huse prices during the same perid. Hwever, since huse prices were exceptinally vlatile during this perid, and the price data is cntaminated with nn-ecnmic nise such as seasnality and hetergeneus differences in the types f dwellings that sld, using simple median r average huse prices wuld be bth incrrect and incnclusive. Instead, a regressin-based methdlgy was deplyed t cmpute a huse price index fr Philadelphia. This prduces a much smther huse price index whse fluctuatins ver time reflect secular market mvements that are free f statistical nise and nn-market idisyncracies 4. The fllwing chart plts the huse price indices fr bth previusly abated prperties and all nn-abated prperties frm 2000 thrugh 2017: 4 The technical term fr this type f huse price index is a weighted repeat sales price index. It was cmputed using the exact same methdlgy as Case-Shiller, which prduces huse prices indices, updated mnthly, fr mst majr U.S. cities. See the appendix fr details. 6
250 Philadelphia Huse Price Indices 2000-2017 2000=100 225 Abatements in Effect Abatements Expired 200 175 150 125 100 Pst-Abated Prperties Citywide Nn-Abated Prperties 75 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 The blue line in the index fr previusly abated dwellings, while the range line represents the index fr nn-abated dwellings. The percent change in either index between any tw years reflect the general price appreciatin (r depreciatin) rate f prperties in each index. The vertical dashed line in the middle f the chart represents when abatements that were previusly granted began t expire after 2008. Hence, the mvements in the blue line prir t 2008 represent hw abated prperties changed in value when their abatement was in effect, while mvements in the same line after 2008 represent hw these same abated prperties changed in value after their abatements expired. Price appreciatin fr bth abated and nn-abated prperties was bth strng and very similar while abatements were in effect during the 2000-2008 perid. The index fr abated dwellings increased frm 100 t 209, reflecting a 109% increase in the general price level f abated dwelling during this perid. The index fr nn-abated dwellings increased frm 100 t 203, indicating a 103% increase in the general price level f nn-abated dwellings during the same perid. Hence, bth abated and nn-abated dwellings rughly dubled in value during the 2000-2008 husing bm years. The pst-bm recessinary years f 2008-2012 saw the value f previusly abated prperties fall by significantly mre than nn-abated prperties. 7
Frm peak t trugh, the price index fr abated dwellings fell frm 209 t 136; a 35% decline. Frm peak t trugh, the price index fr nn-abated dwellings fell frm 203 t 176; a 13% decline. Hence, the relative value f previusly abated prperties fell by almst three times the value f nn-abated prperties during the pst-bm recessin. The greater magnitude f the decline in pst-abated prperties is likely due t their abatements expiring at the same time the verall husing market and ecnmy underwent a significant cntractinary phase. While the decline in value fr previusly abated prperties may be large, it shuld be nted that it was largely a paper decline. Of the ttal 1,175 prperties that sld after their abatements expired in the pst-2008 perid, nly 59 f them (barely 5% f the ttal sample) sld between 2009 and 2012 when prices were generally falling. Hence, althugh the fall in prices was significant, mst f these lsses went unrealized by their wners as the vast majrity f them (95%) waited until prices began t recver after 2012 t sell their previusly abated unit. During the pst-2012 recvery perid, previusly abated prperties increased in value by mre than nn-abated prperties. Frm trugh t their current peak, the index fr abated prperties increased frm 136 t 187; a 38% increase. Frm trugh t their current peak, the index fr nn-abated prperties increased frm 176 t 224; a 27% increase. Despite differential appreciatin rates, previusly abated prperties still have significantly higher values than nn-abated prperties. Because the current value f the index fr previusly abated prperties is less than that f the index fr nn-abated prperties (187 v. 224, respectively), it wuld be tempting t interpret this as meaning that the general price level f pst-abated prperties is belw that f nn-abated prperties. This is emphatically nt true. The fact that the frmer index is less than the latter implies that the general price appreciatin f pst-abated prperties has been less than that f never-abated prperties frm 2000 thrugh 2017. Of curse, this statement des nt apply t sme sub-perids within thse 18 years, when abated dwelling did utpace nnabated dwellings. As mentined earlier in this paper, the median price f previusly abated dwellings in the pst-2008 perid is $360,000, which is substantially higher than the median price f $156,000 fr all nn-abated dwellings in Philadelphia during the same perid. 8
Using these results, a cunterfactual sales price was cmputed fr each prperty by applying the citywide huse price index t the dwelling s riginal purchase price. Essentially, the riginal purchase price was grwn by the percent change in the index t the dwelling s actual time f sale 5. This price represents what each pst-abated dwelling wuld have sld fr if it had appreciated at the same rate as nn-abated prperties. The fllwing table cmpares summary statistics n actual v. cunterfactual sales prices: Actual Sales Prices v. Cunterfactual Sales Prices fr Prperties with Expired Abatements Actual Sale Price Cunterfactual Sale Price %Difference #Sales 1,175 1,175 N/A 25% Quartile Price $255,000 $323,000-21% 50% Median Price $360,000 $440,000-18% Mean Price $425,000 $506,000-16% 75% Quartile Price $505,000 $592,000-15% The results indicate that abated prperties generally appreciated at a slwer verall rate than that f the general husing market: The 25% lwest-priced sales f pst-abated prperties had sales prices f $255,000 r less. Had they appreciated at the citywide rate, they wuld have had a cunterfactual price f $323,000 r less; a -21% difference. The median sales price f a previusly abated prperty was $360,000. Their cunterfactual median price is $440,000; a -18% difference. The tp 25% highest-priced sales f pst-abated prperties had sales prices f $505,000 r mre. Had these dwellings appreciated at the citywide rate, they wuld have had sales prices f $592,000 r mre; a -15% difference. Using these results, all frmerly abated prperties were then re-classified as Relative Gainers r Relative Lsers based upn whether r nt their percentage change in price pre- and pst-abatement either exceeded r lagged the verall percentage change in the husing market s price index during that same perid. The fllwing chart cmpares the number f relative gainers and lsers in each year as abatements expired: 5 Example: a new cnd unit in Center City is cmpleted in 2005 and immediately sells fr $430,000. Ten years later, it sells fr $475,000; a 10% gain. But, during this same perid, the citywide price index increased by 18%. Had this cnd appreciated at the citywide (nn-abated) rate, it wuld have sld fr $514,798, which is its cunterfactual price. 9
300 250 # Sld fr $Gain # Sld fr $Lss Frmerly Abated Units Selling fr Relative $Lss v. $Gain by Year 251 277 200 177 150 100 50 0 97 86 75 62 52 24 29 10 14 3 5 10 0 2010 2011 2012 2013 2014 2015 2016 2017 These results indicate that the number f relative lsers exceeded the number f relative gainers in each year fllwing the expiratin f these dwelling s abatement: Of the 1,175 prperties which sld after their abatements expired, 282 sld fr relative gains, while the remaining 893 sld fr relative lsses. Als, the dllar amunt f relative gains was generally smaller than the dllar amunt f relative lsses: Of thse that sld fr a relative gain, the median gain was $56,000. Of thse that sld fr a relative lss, the median lss was -$108,000. Lastly, this difference in gains v. lsses is nt distributed unifrmly acrss prperties. Lwerpriced dwellings typically experienced small gains and large lsses, while high-priced dwellings generally experienced large gains but nly small lsses: Of the 25% lwest-priced prperties (riginally purchased fr $255,000 r less), thse that sld fr a gain experienced an average gain f $22,000 while thse that sld fr a lss experienced an average lss f $169,000. Of the 25% highest-priced prperties (riginally purchased fr $505,000 r mre), thse that sld fr a gain experienced an average gain f $124,000 while thse that sld fr a lss sld fr an average lss f nly $64,000. 10
Nte that this result is als true fr changes in abslute prices: dwellings that were higherpriced t begin with generally experienced larger gains in bth dllar and percentage terms than lwer-priced dwellings. In summary: Mst abated prperties generally sld fr prices that were higher than their riginal purchase price nce their abatements expired. Hwever, their general appreciatin rate has lagged that f nn-abated prperties. The results wuld seem t suggest a number f brader implicatins abut the Philadelphia s abatement prgram: First, any assertins r cncerns that the expiratin f the abatement wuld lead t a massive liquidatin f previusly abated prperties at steep discunts is strngly refuted and rejected by the data. As mentined in the first paper in this series, apprximately tw-thirds f previusly abated dwellings still remain in the hands f their riginal wner-ccupants. Of the remaining third that have sld, mst have sld fr prices higher than their riginal purchase price. Secnd, the data indicate that the value that buyers/investrs place n the abatement is bth very large and very real: adding apprximately 15-20% t a prperty s purchase price. After adjusting fr general price fluctuatins in the market, mst previusly abated prperties sld fr a relative discunt f 15-20% cmpared t nn-abated prperties 6. Since abated prperties are either new cnstructin r had recently undergne significant imprvements, it is unlikely that this decline can be explained by deteriratin in their physical quality. It is als unlikely that this culd be explained by declines in the lcatin value f these dwellings (e.g. deterirating neighbrhd quality-f-life) since the majrity f them are either in the prime neighbrhd f Center City r in the revitalizing neighbrhds surrunding Center City. That nly leaves the nly ther thing that can affect the prperty s value: its tax treatment; i.e. the expiring abatement. Third, interpreting the financial benefit that the abatement s tax treatment cnfers t a dwelling is likely t depend upn ne s persnal pinin abut the abatement. T critics and skeptics f the prgram, the 15-20% premium that the abatement adds t a prperty s initial value but then dissipates afterwards will likely be viewed as an unnecessary and wasteful tax giveaway that simply pads the bttm line f develpers wh build these prperties. T prpnents f the prgram, the 15-20% premium is prf that the prgram is giving a needed bst t huse prices in rder t help cver Philadelphia s very high cst f cnstructin 7, thus making new develpment happen that wuldn t therwise 8. Finally, preserving and increasing the lcatin value f abated dwellings is the best way t mitigate against any future (pst-abatement) lsses in value, whether abslute r relative. 6 Astute readers may pint ut that the true value f an abatement is the present discunted value f the fregne tax payments ver ten years. This will be cmputed and examined in the final installment f this series, which will examine the fiscal impact f past-abated prperties. 7 Philadelphia s cst f cnstructin is 4 th -highest in the U.S., after New Yrk, San Francisc and Bstn. Surce: enr.cm. 8 Full disclsure: this authr has previusly published research supprting this latter view f the abatement. 11
Once an abatement is granted t a prperty, the prperty s future (pst-abatement) value is determined by a metaphrical hrse race between its lcatin value and its structure value. The lcatin value is reflected in the land value f the prperty, which is nt cvered by the abatement and als generally increases ver time due t bth general inflatin and the cntinued revitalizatin f thse neighbrhds that attract abated dwellings (which bth reflect and imprve the desirability f thse neighbrhds). Cnversely, the structure value f an abated prperty, which is cvered by the abatement, generally declines ver time due t bth physical depreciatin f the prperty and the dwnward amrtizing f the abatement. The mre that city fficials and cmmunity grups can d t increase the desirability f these neighbrhds, the greater the increase in the land value f abated prperties will be, which will ultimately sften and ptentially cunteract declines in the structure value f these prperties nce their abatement expires. The next and final installment in this series will examine the fiscal impact that pst-abated prperties have had in Philadelphia. Email: Kevin.Gillen@huwzer.cm 12
Appendix Fr an intuitive tutrial n hw repeat-sale huse price indices are cmputed, here is an excellent vide: https://www.yutube.cm/watch?v=jnbesm7saja 13
Price Index Regressin Output fr Abated Prperties 14
Price Index Regressin Output fr Citywide Nn-Abated Prperties 15