spb/wp21_uqo

1# RECORD 0462b0d87ab72b3c90375cd7f1b22600ca9530ef2# TITLE: Sorting or Steering: The Effects of Housing Discrimination on Neighborhood Choice3# YEAR: 2018.0  VENUE: nan  DOI: 10.3386/w248264# AUTHORS: P. Christensen; Christopher Timmins5# PAGES: 70  CANDIDATE TABLE PAGES (1-based): [39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 59, 60, 61, 63, 65, 66, 67, 69, 70]6# ABSTRACT: Housing discrimination is illegal. However, paired-tester audit experiments have revealed evidence of discrimination in the interactions between potential buyers and real estate agents, raising concern about whether certain groups are systematically excluded from the beneficial effects of healthy neighborhoods. Using data from HUD’s most recent Housing Discrimination Study and micro-level data on key attributes of neighborhoods in 28 US cities, we find strong evidence of discrimination in the characteristics of neighborhoods towards which individuals are steered. Conditional upon the characteristics of the house suggested by the audit tester, minorities are significantly more likely to be steered towards neighborhoods with less economic opportunity and greater exposures to crime and local pollutants. We find that holding locational preferences or income constant, discriminatory steering alone can contribute substantially to the disproportionate number of minority households found in high poverty neighborhoods in the United States. The steering effect is also large enough to fully explain the differential in proximity to Superfund sites among African American mothers. These results have important implications for studies of “neighborhood effects” and confirm an important mechanism underlying observed correlations between race and pollution in the environmental justice literature. Our results also suggest that the basic utility maximization assumptions underlying hedonic and residential sorting models may often be violated, resulting in an important distortion in the provision of local public goods.789===== PAGE 1 (score 2, numeric-paren 0, table-words 0, var-words 2) =====1011NBER WORKING PAPER SERIES12SORTING OR STEERING:13THE EFFECTS OF HOUSING DISCRIMINATION ON NEIGHBORHOOD CHOICE14Peter Christensen15Christopher Timmins16Working Paper 2482617http://www.nber.org/papers/w2482618NATIONAL BUREAU OF ECONOMIC RESEARCH191050 Massachusetts Avenue20Cambridge, MA 0213821July 2018, Revised June 202122Previously circulated as “Sorting or Steering: Experimental Evidence on the Economic Effects of 23Housing Discrimination.” We thank Spencer Banzhaf, Kelly Bishop, Don Fullerton, Ludovica 24Gazze, Lu Han, Kelly Maguire, David Molitor, Steve Ross, Kerry Smith, Matthew Turner, Randy 25Walsh, Ann Wolverton, John Yinger and seminar participants at the Wharton School, University 26of Arizona, University of Pittsburgh, Simon Fraser University, Washington University in St. 27Louis, 2019 meetings of the American Economic Association, the 2018 meetings of the 28Association of Urban Economics, the 2018 China Meetings of the Econometrics Society, and the 292018 meetings of the Southern Economic Association for excellent comments. This project 30makes use of data from an experiment conducted by the Department of Housing and Urban 31Development and the Urban Institute in 2012 (Turner et al., 2013). All HDS 2012 data and 32documentation 33are 34publicly 35available 36on 37https://www.huduser.gov/portal/datasets/hsg 38discrimination.html. All data and programs used in this paper will be made available on: https://39github.com/peterchristensen/Sorting-or-Steering. Thanks to research assistants in the University 40of Illinois Big Data and Environmental Economics and Policy (BDEEP) Group and the Duke 41Environmental Justice Lab for excellent work. We acknowledge generous support from the 42National Center for Supercomputing Applications. All errors are our own.43NBER working papers are circulated for discussion and comment purposes. They have not been 44peer-reviewed or been subject to the review by the NBER Board of Directors that accompanies 45official NBER publications.46© 2018 by Peter Christensen and Christopher Timmins. All rights reserved. Short sections of text, 47not to exceed two paragraphs, may be quoted without explicit permission provided that full 48credit, including © notice, is given to the source.495051===== PAGE 39 (score 23, numeric-paren 0, table-words 2, var-words 5) =====5253Table 1. Descriptive Statistics for Tester Dataset54Tester Characteristics55(True Actors, Not Assigned)56Variable57All58White59African60Hispanic61Asian62Groups63American64Age6540.9106641.4856741.4946842.0706936.39870Percent Male710.388720.415730.375740.284750.42076Percent Rented Home770.588780.580790.630800.516810.61482Percent Owned Home830.305840.296850.312860.421870.21788Personal Income89Under $10,000900.244910.275920.241930.125940.24995$10,000 - $19,999960.164970.130980.225990.1141000.235101$20,000 - $29,9991020.2371030.2931040.1361050.2491060.184107$30,000 - $39,9991080.1361090.1141100.1521110.1831120.152113$40,000 - $49,9991140.0961150.0861160.1051170.1171180.090119$50,000 - $74,9991200.0681210.0621220.0691230.1361240.022125$75,000 - $99,9991260.0091270.0131280.0081290.0001300.004131$100,000 or more1320.0031330.0021340.0041350.0111360.000137Education138Attended High School1390.0031400.0001410.0021420.0001430.018144GED1450.0121460.0211470.0041480.0001490.000150High School Diploma1510.0351520.0201530.0451540.0371550.079156Attended Vocational School1570.0051580.0051590.0001600.0181610.000162Vocational School Diploma1630.0221640.0321650.0061660.0071670.029168Attended College1690.1621700.1301710.1821720.3081730.108174Associate’s Degree1750.0781760.0591770.1011780.1361790.054180Bachelor’s Degree1810.3581820.4041830.2831840.1831850.484186Attended Graduate School1870.0521880.0741890.0201900.0481910.025192Graduate Degree1930.2431940.2301950.3081960.2531970.170198Assigned Characteristics199Monthly Rent2001, 3322011, 3342021, 3922031, 2772041, 285205Percent Tester Went First2060.5162070.5892080.4172090.4542100.462211Percent Appointment in AM2120.4162130.4102140.4252150.4282160.422217Percent Car Owner2180.8542190.8612200.7962210.9222220.861223Length of Employment (Years)2244.0912253.3452264.9362275.2032284.507229Years at Residence2303.6502312.9252324.4052334.7662344.101235Lease Type236Month-to-Month2370.5692380.5582390.5652400.5812410.620242Lease2430.4312440.4422450.4352460.4192470.380248N = 2,260249N = 1,161250N = 512251N = 286252N = 294253Note: Table reports the mean values of the innate (actor) and the assigned (scripted) characteristics of testers in the254study. Column 1 reports the mean for all testers in the study and columns 2-4 report the mean for testers in a given race255group. Testers were matched based on approximately similar actor (not assigned) characteristics and then tester pairs256were matched to assigned roles. Assigned incomes vary by trial and were calculated as a function of the listing price of an257advertised home.25837259260261===== PAGE 40 (score 23, numeric-paren 0, table-words 2, var-words 10) =====262263Table 2. Home and Neighborhood Characteristics264Characteristics of265Advertised Homes266Variable267All Groups268White269African American270Hispanic271Asian272Listing Price273378, 983274381, 064275344, 169276387, 648277408, 701278Building Type279Single family, detached2800.7392810.7442820.7382830.6992840.749285Duplex2860.0172870.0172880.0132890.0182900.018291Rowhouse or Townhouse2920.1342930.1282940.1182950.1582960.154297Multi-family structure2980.1012990.1013000.1253010.1123020.069303Mobile home3040.0013050.00130603070.0053080309Neighborhood Characteristics310School-Specific Test Scores311Elem. School Test Score3120.2783130.2793140.2063150.3443160.340317Midd. School Test Score3180.1733190.1643200.1853210.1163220.247323Housing Search Platform324Assaults32593.33632688.35232796.15032895.346329104.878330Elem. School Quality3316.2103326.2033336.2023346.1693356.260336ACS Block Group Characteristics337Poverty Rate3380.0893390.0883400.0923410.0903420.086343Percent High Skill3440.4673450.4673460.4673470.4633480.472349Percent College Graduate3500.5013510.5043520.5013530.4943540.501355Single-Parent HH3560.1463570.1463580.1413590.1553600.147361Ownership Rate3620.7313630.7353640.7313650.7153660.728367ACS Block Group Racial Composition368Percent White3690.6753700.6853710.6843720.6393730.664374Percent African American3750.0903760.0903770.1033780.0823790.084380Percent Hispanic3810.1483820.1403830.1343840.1853850.151386Percent Asian3870.0673880.0653890.0613900.0733910.079392Pollution Measurements393Superfund Sites3940.3273950.3113960.3373970.3533980.352399Risk-Screening Environmental Indicators (RSEI)4006, 076.84016, 252.54027, 865.34034, 283.24045, 131.2405PM 254069.2834079.3784089.2534099.0564109.293411N = 7,026412N = 3,607413N = 1,212414N = 1,028415N = 1,108416Note: Table reports the mean values of the advertised listings assigned to testers in the study, which are based on sampling417frames generated for each metropolitan market. The composition of trials within a given market is defined to match the418racial composition of the market, such that the characteristics of advertised listings can vary across the race groups (though419they will not vary within a trial). Column 1 reports the mean characteristics for all testers in the study and columns 2-4420report the mean for testers in a given race group.42138422423424===== PAGE 41 (score 111, numeric-paren 84, table-words 2, var-words 6) =====425426Table 3. Balance Statistics for Testers427Tester Characteristics428(True Actors, Not Assigned)429Variable430African431Hispanic432Asian433American434Age435-0.8744361.476437-2.060438(0.996)439(1.858)440(1.833)441Percent Male4420.0094430.0004440.000445(0.015)446(0.000)447(0.000)448Percent Rented Home4490.138450-0.0444510.083452(0.100)453(0.124)454(0.114)455Percent Owned Home456-0.153∗4570.033458-0.111459(0.093)460(0.113)461(0.110)462Personal Income463Under $10,000464-0.012465-0.107466-0.000467(0.097)468(0.116)469(0.125)470$10,000 - $19,9994710.1184720.0124730.235∗∗474(0.076)475(0.116)476( 0.102)477$20,000 - $29,999478-0.212∗∗479-0.066480-0.113481(0.086)482(0.151)483(0.107)484$30,000 - $39,9994850.1264860.147∗4870.008488(0.081)489(0.088)490(0.114)491$40,000 - $49,999492-0.0024930.011494-0.066495(0.063)496(0.079)497(0.077)498$50,000 - $74,999499-0.0235000.006501-0.090502(0.057)503(0.065)504(0.080)505$75,000 - $99,999506-0.028507-0.027508-0.012509(0.022)510(0.041)511(0.018)512$100,000 or more5130.0005140.0305150.000516(0.000)517(0.046)518(0.000)519Education520Attended High School5210.0005220.0005230.0205240.0005250.0005260.031527GED5280.011529-0.071530-0.028531(0.017)532(0.050)533(0.044)534High School Diploma5350.0365360.0415370.116∗538(0.035)539(0.037)540(0.068)541Attended Vocational School542-0.0075430.0155440.000545(0.011)546(0.034)547(0.000)548Vocational School Diploma549-0.014550-0.002551-0.012552(0.030)553(0.035)554(0.018)555Attended College5560.1005570.1095580.082559(0.080)560(0.138)561(0.103)562Associate’s Degree563-0.0175640.088565-0.016566(0.043)567(0.066)568(0.052)569Bachelor’s Degree570-0.161571-0.251∗572-0.032573(0.110)574(0.147)575(0.104)576Attended Graduate School577-0.062578-0.043579-0.036580(0.054)581(0.075)582(0.032)583Graduate Degree5840.1065850.114586-0.088587(0.107)588(0.120)589( 0.113)590Assigned Characteristics591Percent Tester Went First592-0.093593-0.231594-0.047595(0.165)596(0.200)597(0.209)598Percent Appointment in AM5990.002600-0.0146010.028602(0.041)603(0.038)604(0.038)605Percent Car Owner6060.0106070.000608-0.065609(0.015)610(0.000)611(0.058)612Length of Employment (Years)6131.559∗∗∗6142.050∗∗∗6151.444∗∗∗616(0.270)617(0.328)618(0.238)619Years at Residence6201.511∗∗∗6211.898∗∗∗6221.551∗∗∗623(0.288)624(0.301)625(0.203)626Lease Type627Month-to-Month6280.0126290.0726300.115631(0.047)632(0.091)633(0.076)634Lease635-0.012636-0.072637-0.115638(0.047)639(0.091)640(0.076)641N = 512642N = 286643N = 294644Note: Table reports the results of balance tests that compare tester characteristics from a given group in a trial to the645mean for the white (comparison) group. These tests are based on a simplified version of model 3 that includes a term646for race group and a trial fixed effect, but omits other controls (tester and advertised home attributes). The top panel647reports differences in innate (actor) characteristics, whereas the bottom panel reports differences in assigned (scripted)648characteristics. Significance levels are: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.64939650651652===== PAGE 42 (score 86, numeric-paren 59, table-words 2, var-words 11) =====653654Table 4. Balance Statistics for Homes655Characteristics of656Advertised Homes657Variable658African659Hispanic660Asian661American662Listing Price663-397.595664-388.358665-1, 866.292666(960.555)667(1,288.875)668(1,136.40)669Building Type670Single-family Detached6710.0096720.0066730.029***674(0.010)675(0.013)676(0.011)677Duplex678-0.001679-0.002680-0.005681(0.005)682(0.005)683(0.008)684Rowhouse or Townhouse6850.0056860.0076870.005688(0.010)689(0.011)690(0.015)691Multi-family Structure692-0.006693-0.006694-0.029**695(0.008)696(0.009)697(0.012)698Mobile Home699-0.001700-0.001701-0.000007020.0017030.0037040.00000705Neighborhood Characteristics706School-Specific Test Scores707Elem. School Test Score7080.012**7090.009710-0.006711(0.006)712(0.007)713(0.006)714Midd. School Test Score715-0.010716-0.007717-0.005718(0.006)719(0.005)720(0.005)721Housing Search Platform722Assault723-0.0407240.109725-0.153726(0.174)727(1.069)728(0.173)729Elem. School Quality7300.0017310.0197320.025733(0.021)734(0.022)735(0.034)736ACS Block Group Characteristics737Poverty Rate7380.00057390.001740-0.001741(0.002)742(0.002)743(0.001)744Percent High Skill745-0.005746-0.0027470.0004748(0.003)749(0.003)750(0.002)751Percent College Educated752-0.005**753-0.0037540.0002755(0.002)756(0.004)757(0.002)758Single-Parent HH7590.0027600.002761-0.001762(0.002)763(0.002)764(0.002)765ACS Block Group Racial Composition766Percent African American7670.0037680.004769-0.003770(0.002)771(0.003)772(0.002)773Percent Hispanic774-0.0027750.0037760.001777(0.001)778(0.002)779(0.001)780Percent Asian781-0.001782-0.0002783-0.001784(0.001)785(0.001)786(0.001)787Pollution Measurements788Superfund Sites789-0.00027900.0127910.001792(0.005)793(0.011)794(0.006)795Risk-Screening Environmental Indicators (RSEI)796-30.108797-108.29079889.371799(51.410)800(102.645)801(141.098)802PM 25803-0.0108040.00018050.015806(0.011)807(0.016)808(0.016)809Note: Table reports the results of balance tests that compare the mean values of home characteristics assigned to a given810race group to the mean for the white (comparison) group. These tests are based on a simplified version of model 3 that811includes a term for race group and a trial fixed effect, but omits other controls (tester and advertised home attributes).812The top panel reports differences in property characteristics within trials, which could occur if the advertised home was813removed from the market during a trial. The bottom panels report differences in neighborhood characteristics. Significance814levels are: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.81540816817818===== PAGE 43 (score 37, numeric-paren 16, table-words 5, var-words 8) =====819820Table 5. Differences in Recommendations and Availability of Advertised Properties821Dependent variable:822Number of Recommendations823Home Availability824Racial Minority825−0.1282826−0.14198270.00488280.0057829(0.1990)830(0.1987)831(0.0183)832(0.0183)833African American834−0.1608835−0.1690836−0.0097837−0.0087838(0.2713)839(0.2707)840(0.0220)841(0.0219)842Hispanic843−0.1340844−0.1304845−0.0090846−0.0077847(0.2465)848(0.2474)849(0.0255)850(0.0258)851Asian8520.12318530.08338540.01748550.0178856(0.2477)857(0.2465)858(0.0227)859(0.0227)860ln(Price) Advert Home861N862Y863N864Y865Racial Comp Advert Home866N867Y868N869Y870Observations8716,5808726,5558736,5888746,562875Adjusted R2876-0.2192877-0.2348878-0.1687879-0.1792880Note: Coefficients report differences in the number of homes recommended and number of available homes recommended881to minority testers relative to a white tester (the omitted category). Regressions are estimated at the tester level using882data for the testers from the total sample of 6,962 that report these variables. Racial minority encompasses all individual883minority groups. All regression specifications include trial fixed effects and control for the full set of actor characteristics,884assigned characteristics, and search characteristics. Actor characteristics: tester income, tester household income, gender of885tester, age of tester, month of test, educational attainment of tester; Assigned characteristics: household members, current886home ownership status, current lease type, car ownership status, reason for moving, years in current residence, length of887employment at current job, reason tester can afford down payment, current lease assigned to tester; Search Characteristics:888month of test, sequence of tester appointments, time of the appointment (am/pm), type of recommended building. Standard889errors are clustered by trial, which is the level of randomization. Significance levels are: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.89041891892893===== PAGE 44 (score 43, numeric-paren 20, table-words 5, var-words 8) =====894895Table 6. Discriminatory Steering and Neighborhood Racial Composition896Dependent variable: White Household Share897I898II899III900IV901V902Racial Minority903−0.0089904−0.0089905−0.0091906−0.0096907−0.0096908(0.0112)909(0.0112)910(0.0112)911(0.0112)912(0.0112)913African American914−0.0399∗∗∗915−0.0398∗∗∗916−0.0403∗∗∗917−0.0409∗∗∗918−0.0410∗∗∗919(0.0148)920(0.0148)921(0.0148)922(0.0147)923(0.0147)924Hispanic9250.00289260.00289270.00289280.00259290.0025930(0.0152)931(0.0152)932(0.0152)933(0.0152)934(0.0152)935Asian9360.01829370.01829380.01819390.01769400.0176941(0.0136)942(0.0136)943(0.0136)944(0.0136)945(0.0136)946Comparison Mean (White)9470.699480.699490.699500.699510.69952Share White Advert Home953N954Y955Y956Y957Y958ln(Price) Advert Home959N960N961Y962Y963Y964Racial Comp Advert Home965N966N967N968Y969Y970Poverty Share Advert Home971N972N973N974N975Y976Observations97721,90497821,89397921,87198021,87198121,871982Adjusted R29830.77729840.77729850.77669860.77669870.7765988Note: Coefficients report differences in the racial composition of neighborhoods recommended to minority testers relative989to a white tester (the omitted category).990Racial minority encompasses all individual minority groups.991All regression992specifications include trial fixed effects and control for the full set of actor characteristics, assigned characteristics, and993search characteristics. Actor characteristics: tester income, tester household income, gender of tester, age of tester, month994of test, educational attainment of tester; Assigned characteristics: household members, current home ownership status,995current lease type, car ownership status, reason for moving, years in current residence, length of employment at current996job, reason tester can afford down payment, current lease assigned to tester; Search characteristics: month of test, sequence997of tester appointments, and time of the appointment (am/pm). Standard errors are clustered by trial, which is the level of998randomization. Significance levels are: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.99942100010011002===== PAGE 45 (score 31, numeric-paren 12, table-words 4, var-words 8) =====10031004Table 7. Discriminatory Steering and Neighborhood Racial Composition by Income1005Dep. Variable: White Household Share by Income1006High Inc1007Mid Inc1008Low Inc1009Racial Minority1010−0.0265∗∗∗1011−0.003610120.0173∗∗∗1013(0.0081)1014(0.0076)1015(0.0062)1016African American1017−0.0338∗∗∗1018−0.0187∗10190.00991020(0.0104)1021(0.0100)1022(0.0081)1023Hispanic1024−0.01451025−0.003810260.0201∗∗1027(0.0108)1028(0.0100)1029(0.0084)1030Asian1031−0.0246∗∗∗10320.0155∗10330.0217∗∗∗1034(0.0095)1035(0.0092)1036(0.0077)1037Comparison Mean (White)10380.2110390.33910400.1461041Share White Advert Home1042N1043Y1044Y1045ln(Price) Advert Home1046N1047N1048Y1049Racial Comp Advert Home1050N1051N1052N1053Poverty Share Advert Home1054N1055N1056N1057Observations105821,470105921,470106021,4701061Adjusted R210620.717810630.706410640.66651065Note: Coefficients report differences in the racial composition (by income level) of neighborhoods recommended to minority1066testers relative to a white tester (the omitted category). Racial minority encompasses all individual minority groups. All1067regression specifications include trial fixed effects and control for the full set of actor characteristics, assigned characteristics,1068and search characteristics. Actor characteristics: tester income, tester household income, gender of tester, age of tester,1069month of test, educational attainment of tester; Assigned characteristics: household members, current home ownership1070status, current lease type, car ownership status, reason for moving, years in current residence, length of employment at1071current job, reason tester can afford down payment, current lease assigned to tester; Search characteristics: month of test,1072sequence of tester appointments, and time of the appointment (am/pm). Standard errors are clustered by trial, which is1073the level of randomization. Significance levels are: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.107443107510761077===== PAGE 46 (score 88, numeric-paren 54, table-words 7, var-words 10) =====10781079Table 8. Discriminatory Steering and Neighborhood Effects1080School-Specific Test Scores:1081Housing Search Platform:1082Elem School1083Middle School1084Assaults1085Elem School1086Racial Minority1087-0.0447**1088-0.0328*10898.35911090-0.3424*1091(0.0227)1092(0.0178)1093(6.4744)1094(0.1824)1095African American1096-0.0856**1097-0.0891***109823.1146**1099-0.4727**1100(0.0424)1101(0.0242)1102(9.8939)1103(0.2379)1104Hispanic1105-0.005611060.01001107-0.46431108-0.24621109(0.0349)1110(0.0294)1111(8.0628)1112(0.2423)1113Asian1114-0.04421115-0.028711165.15821117-0.3756*1118(0.0380)1119(0.0237)1120(7.6687)1121(0.2195)1122Comparison Mean (White)11230.2911240.18112580.8511266.431127ln(Price) Advert Home1128Y1129Y1130Y1131Y1132Racial Comp Advert Home1133Y1134Y1135Y1136Y1137Outcome Advertised Home1138Y1139Y1140Y1141Y1142Observations11439,36011449,713114511,161114611,0321147Adjusted R211480.762511490.912811500.797111510.74751152American Community Survey1153Poverty Rate1154High Skill1155College1156Single-Parent HH1157Ownership Rate1158Racial Minority11590.0116**1160-0.0254**1161-0.0182*11620.00771163-0.01701164(0.0054)1165(0.0099)1166(0.0101)1167(0.0062)1168(0.0142)1169African American11700.01201171-0.0302**1172-0.0340***11730.0167*1174-0.02631175(0.0075)1176(0.0130)1177(0.0127)1178(0.0087)1179(0.0184)1180Hispanic11810.0205***1182-0.0387***1183-0.0216*11840.0157*11850.00591186(0.0076)1187(0.0130)1188(0.0127)1189(0.0084)1190(0.0203)1191Asian1192-0.00091193-0.007211940.00291195-0.00611196-0.02321197(0.0069)1198(0.0128)1199(0.0142)1200(0.0085)1201(0.0176)1202Comparison Mean (White)12030.08212040.4912050.5312060.1412070.741208ln(Price) Advert Home1209Y1210Y1211Y1212Y1213Y1214Racial Comp Advert Home1215Y1216Y1217Y1218Y1219Y1220Outcome Advertised Home1221Y1222Y1223Y1224Y1225Y1226Observations122721,871122821,871122921,871123021,871123121,8711232Adjusted R212330.520512340.692512350.754812360.523712370.60241238p-values12390.0112400.03312410.07312420.2112430.231244hochberg q-values12450.05212460.1312470.2212480.2312490.231250Note: Coefficients report differences in the attributes of properties/neighborhoods recommended to minority testers relative1251to a white tester (the omitted category).1252Racial minority encompasses all individual minority groups.1253All regression1254specifications include trial fixed effects and control for the full set of actor characteristics, assigned characteristics, and1255search characteristics. Actor characteristics: tester income, tester household income, gender of tester, age of tester, month1256of test, educational attainment of tester; Assigned characteristics: household members, current home ownership status,1257current lease type, car ownership status, reason for moving, years in current residence, length of employment at current job,1258reason tester can afford down payment, current lease assigned to tester; Search Characteristics: month of test, sequence of1259tester appointments, and time of the appointment (am/pm). Standard errors are clustered by trial, which is the level of1260randomization. Significance levels are: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.126144126212631264===== PAGE 47 (score 52, numeric-paren 24, table-words 7, var-words 9) =====12651266Table 9. Discriminatory Steering and Local Pollution Exposures1267Pollution1268Superfund1269Toxics1270PM1271Racial Minority12720.1010∗∗∗1273765.9137∗∗1274−0.04381275(0.0325)1276(324.9800)1277(0.0504)1278African American12790.1256∗∗∗1280872.9819∗∗1281−0.02521282(0.0465)1283(403.4626)1284(0.0678)1285Hispanic12860.05911287459.728312880.01021289(0.0376)1290(544.8236)1291(0.0638)1292Asian12930.1046∗∗1294899.5148∗∗1295−0.1243∗∗1296(0.0484)1297(424.3961)1298(0.0617)1299Comparison Mean (White)13000.361301510013029.41303ln(Price) Advert Home1304Y1305Y1306Y1307Racial Comp Advert Home1308Y1309Y1310Y1311Outcome Advertised Home1312Y1313Y1314Y1315Observations131621,883131721,883131821,8831319Adjusted R213200.874413210.693913220.96341323Pollution1324Superfund1325Toxics1326PM1327Racial Minority13280.1444∗∗∗13291, 110.533013300.00821331(0.0496)1332(691.5594)1333(0.1091)1334African American13350.1978∗∗∗13362, 387.1960∗∗13370.08361338(0.0750)1339(992.5587)1340(0.1470)1341Hispanic13420.027613431, 361.698013440.20071345(0.0677)1346(853.3277)1347(0.1223)1348Asian13490.08081350509.97191351−0.03781352(0.0627)1353(858.7148)1354(0.1472)1355Comparison Mean (White)13560.381357630013589.31359ln(Price) Advert Home1360Y1361Y1362Y1363Racial Comp Advert Home1364Y1365Y1366Y1367Outcome Advertised Home1368Y1369Y1370Y1371Observations13728,01613738,01613748,0161375Adjusted R213760.850813770.596813780.96721379Note: Coefficients report differences in the pollution levels at the location of properties recommended to minority testers1380relative to a white tester (the omitted category). Racial minority encompasses all individual minority groups. All regression1381specifications include trial fixed effects and control for the full set of actor characteristics, assigned characteristics, and search1382characteristics. Actor characteristics: tester income, tester household income, gender of tester, age of tester, month of test,1383educational attainment of tester; Assigned characteristics: household members, current home ownership status, current1384lease type, car ownership status, reason for moving, years in current residence, length of employment at current job,1385reason tester can afford down payment, current lease assigned to tester; Search Characteristics: month of test, sequence of1386tester appointments, and time of the appointment (am/pm). Standard errors are clustered by trial, which is the level of1387randomization. Significance levels are: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.138845138913901391===== PAGE 48 (score 95, numeric-paren 63, table-words 7, var-words 9) =====13921393Table 10. Discriminatory Steering and Neighborhood Effects (Mothers)1394School-Specific Test Scores:1395Housing Search Platform:1396Elem School1397Middle School1398Assaults1399Elem School1400Racial Minority1401-0.1471***1402-0.0979***140339.4893***1404-3.0893***1405(0.0483)1406(0.0278)1407(14.0657)1408(0.9501)1409African American1410-0.2231**1411-0.2289***141279.7645***1413-3.0262***1414(0.0954)1415(0.0564)1416(12.7280)1417(0.7800)1418Hispanic1419-0.2374***1420-0.1214***142129.33931422-1.9212**1423(0.0729)1424(0.0364)1425(18.4929)1426(0.9598)1427Asian1428-0.05431429-0.1027**143022.20201431-3.8178***1432(0.1265)1433(0.0452)1434(25.4432)1435(0.9020)1436Comparison Mean (White)14370.3314380.2814398414406.51441ln(Price) Advert Home1442Y1443Y1444Y1445Y1446Racial Comp Advert Home1447Y1448Y1449Y1450Y1451Outcome Advertised Home1452Y1453Y1454Y1455Y1456Observations14573,80614583,81114594,58914604,5211461Adjusted R214620.762414630.892514640.834414650.69921466American Community Survey1467Poverty Rate1468High Skill1469College1470Single-Parent HH1471Ownership Rate1472Racial Minority14730.0453***1474-0.0347**1475-0.0424***14760.00011477-0.0950***1478(0.0102)1479(0.0144)1480(0.0163)1481(0.0125)1482(0.0242)1483African American14840.0510***1485-0.0622***1486-0.0769***14870.02641488-0.1659***1489(0.0147)1490(0.0210)1491(0.0234)1492(0.0164)1493(0.0329)1494Hispanic14950.0557***1496-0.0530***1497-0.0483***14980.01971499-0.1021***1500(0.0121)1501(0.0171)1502(0.0179)1503(0.0139)1504(0.0323)1505Asian15060.0356**1507-0.00811508-0.0432*1509-0.01631510-0.03891511(0.0155)1512(0.0206)1513(0.0232)1514(0.0164)1515(0.0313)1516Comparison Mean (White)15170.0815180.4815190.5215200.1415210.771522ln(Price) Advert Home1523Y1524Y1525Y1526Y1527Y1528Racial Comp Advert Home1529Y1530Y1531Y1532Y1533Y1534Outcome Advertised Home1535Y1536Y1537Y1538Y1539Y1540Observations15418,01515428,01515438,01515448,01515458,0151546Adjusted R215470.456115480.660515490.739515500.482615510.48631552Note: Coefficients report differences in the attributes of properties/neighborhoods recommended to minority testers that are1553assigned the role of a mother (female gender with children in household) relative to a white tester with the same assigned role1554(the omitted category). Racial minority encompasses all individual minority groups. All regression specifications include1555trial fixed effects and control for the full set of actor characteristics, assigned characteristics, and search characteristics.1556Actor characteristics: tester income, tester household income, gender of tester, age of tester, month of test, educational1557attainment of tester; Assigned characteristics: household members, current home ownership status, current lease type, car1558ownership status, reason for moving, years in current residence, length of employment at current job, reason tester can afford1559down payment, current lease assigned to tester; Search Characteristics: month of test, sequence of tester appointments, and1560time of the appointment (am/pm). Standard errors are clustered by trial, which is the level of randomization. Significance1561levels are: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.156246156315641565===== PAGE 49 (score 38, numeric-paren 18, table-words 4, var-words 8) =====15661567Table 11. Discriminatory Steering: Low Poverty Neighborhoods1568Dependent variable:1569Low Poverty1570Low Poverty: Families1571Low Poverty: Moms1572Low Pov/High Dad1573Low Pov/High Dad: Families1574Low Pov/High Dad: Moms1575African American1576−0.1179∗∗1577−0.1730∗∗∗1578−0.4318∗∗∗1579−0.1175∗∗1580−0.1705∗∗∗1581−0.3958∗∗∗1582(0.0464)1583(0.0579)1584(0.1268)1585(0.0478)1586(0.0588)1587(0.1296)1588Hispanic1589−0.1585∗∗∗1590−0.1986∗∗∗1591−0.4283∗∗∗1592−0.1543∗∗∗1593−0.1657∗∗∗1594−0.3358∗∗1595(0.0425)1596(0.0534)1597(0.1114)1598(0.0429)1599(0.0518)1600(0.1308)1601Asian1602−0.04371603−0.07701604−0.3101∗∗1605−0.04191606−0.07571607−0.2618∗∗1608(0.0445)1609(0.0603)1610(0.1282)1611(0.0445)1612(0.0610)1613(0.1275)1614Observations16157,56616164,64316172,79416187,52416194,63916202,7901621Adjusted R216220.396716230.395916240.424816250.395416260.394116270.42051628Note: Coefficients report differences in the attributes of neighborhoods recommended to minority testers relative to a white tester (the omitted category). Racial minority encompasses all1629individual minority groups. All regression specifications include trial fixed effects and control for the full set of actor characteristics, assigned characteristics, and search characteristics. Actor1630characteristics: tester income, tester household income, gender of tester, age of tester, month of test, educational attainment of tester; Assigned characteristics: household members, current1631home ownership status, current lease type, car ownership status, reason for moving, years in current residence, length of employment at current job, reason tester can afford down payment,1632current lease assigned to tester; search Characteristics: month of test, sequence of tester appointments, and time of the appointment (am/pm). Standard errors are clustered by trial, which1633is the level of randomization. Significance levels are: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.163447163516361637===== PAGE 50 (score 30, numeric-paren 9, table-words 5, var-words 9) =====16381639Table 12. Discriminatory Steering: Median Income in Neighborhood1640Dependent variable: log(Median Income)1641All Testers1642Families1643Moms1644African American1645−0.0729∗∗1646−0.1085∗∗∗1647−0.2485∗∗∗1648(0.0335)1649(0.0401)1650(0.0738)1651Hispanic1652−0.04781653−0.03881654−0.07821655(0.0310)1656(0.0376)1657(0.0567)1658Asian1659−0.01261660−0.05111661−0.07731662(0.0271)1663(0.0367)1664(0.0583)1665Comparison Mean (White)166611.405166711.46166811.451669Observations167021,45716716,36916723,7721673Adjusted R216740.665716750.675616760.61041677Note: Coefficients report differences in the median income of households in neighborhoods recommended to minority1678testers relative to a white tester (the omitted category). Racial minority encompasses all individual minority groups. All1679regression specifications include trial fixed effects and control for the full set of actor characteristics, assigned characteristics,1680and search characteristics. Actor characteristics: tester income, tester household income, gender of tester, age of tester,1681month of test, educational attainment of tester; Assigned characteristics: household members, current home ownership1682status, current lease type, car ownership status, reason for moving, years in current residence, length of employment at1683current job, reason tester can afford down payment, current lease assigned to tester; Search characteristics: month of test,1684sequence of tester appointments, and time of the appointment (am/pm). Standard errors are clustered by trial, which is1685the level of randomization. Significance levels are: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.168648168716881689===== PAGE 51 (score 92, numeric-paren 59, table-words 7, var-words 10) =====16901691Table 13. Discriminatory Steering by Implied Preferences for Neighborhood Attributes1692School-Specific Test Scores:1693Housing Search Platform:1694Elem School1695Middle School1696Assaults1697Elem School1698Racial Minority1699-0.0618***1700-0.0326***17019.2013**1702-0.2583***1703(0.0131)1704(0.0092)1705(4.3081)1706(0.0618)1707African American1708-0.0810***1709-0.0210*17103.96751711-0.5920***1712(0.0177)1713(0.0122)1714(5.7978)1715(0.0821)1716Hispanic1717-0.0701***1718-0.0124171912.9738**1720-0.2014**1721(0.0204)1722(0.0138)1723(6.3158)1724(0.0922)1725Asian1726-0.0557***1727-0.0510***172817.3702***1729-0.02991730(0.0183)1731(0.0128)1732(5.9039)1733(0.0826)1734Comparison Mean (White)17350.0217360.0161737-7.417380.181739Observations17408,95117419,242174211,161174311,0321744Adjusted R217450.318717460.431317470.265717480.37511749American Community Survey1750Poverty Rate1751High Skill1752College1753Single-Parent HH1754Ownership Rate1755Racial Minority17560.00121757-0.0070**1758-0.001817590.0042*17600.00421761(0.0022)1762(0.0034)1763(0.0035)1764(0.0024)1765(0.0052)1766African American1767-0.0085***1768-0.0120***1769-0.006817700.0092***17710.00351772(0.0030)1773(0.0045)1774(0.0047)1775(0.0032)1776(0.0069)1777Hispanic17780.00521779-0.0158***17800.00421781-0.003217820.0146*1783(0.0033)1784(0.0049)1785(0.0051)1786(0.0035)1787(0.0075)1788Asian17890.003217900.0134***17910.002817920.0059*17930.0139*1794(0.0031)1795(0.0046)1796(0.0049)1797(0.0033)1798(0.0072)1799Comparison Mean (White)18007.6e-0518010.002518020.00681803-0.00251804-0.00631805Observations180621,871180721,871180821,871180921,871181021,8711811R218120.339618130.399218140.383818150.364318160.37271817Adjusted R218180.277818190.343018200.326218210.304818220.31401823Note: Coefficients report differences in the gap between implied preferences and homes recommended to minority testers1824relative to a white tester (the omitted category). Dependent variable is the difference between the level of each attribute1825in the advertised listing and the recommended property.1826Racial minority encompasses all individual minority groups.1827This specification omits trial fixed to allow identifying variation in the advertised home attribute, which is typically held1828fixed within a trial (and otherwise balanced). All regression specifications control for the full set of actor characteristics,1829assigned characteristics, and search characteristics. Actor characteristics: tester income, tester household income, gender of1830tester, age of tester, month of test, educational attainment of tester; Assigned characteristics: household members, current1831home ownership status, current lease type, car ownership status, reason for moving, years in current residence, length of1832employment at current job, reason tester can afford down payment, current lease assigned to tester; Search Characteristics:1833month of test, sequence of tester appointments, and time of the appointment (am/pm). Standard errors are clustered by1834trial, which is the level of randomization. Significance levels are: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.183549183618371838===== PAGE 52 (score 56, numeric-paren 27, table-words 7, var-words 9) =====18391840Table 14. Discriminatory Steering and Later Transactions1841Buyers Upon Sale1842White1843African American1844Hispanic1845Asian1846Differentiability Score184768%184815%184987%185091%1851Same Race Tester1852-0.00641853-0.001918540.0733*18550.1169**1856(0.0481)1857(0.0149)1858(0.0442)1859(0.0510)1860Buyers(n)18617,598186221718631,22418641,0141865Observations18669,42818679,42818689,42818699,4281870Adjusted R218710.365418720.203518730.388618740.41631875Dependent variable: Logarithm of Price1876I1877II1878III1879IV1880V1881Racial Minority1882-0.01171883-0.01161884-0.01071885-0.01061886-0.05401887(0.0336)1888(0.0336)1889(0.0336)1890(0.0336)1891(0.0371)1892African American18930.026318940.026218950.027518960.02761897-0.03631898(0.0404)1899(0.0404)1900(0.0404)1901(0.0404)1902(0.0445)1903Hispanic19040.030719050.030519060.036219070.03601908-0.00041909(0.0664)1910(0.0665)1911(0.0671)1912(0.0674)1913(0.0699)1914Asian1915-0.08991916-0.08971917-0.08961918-0.08921919-0.1101*1920(0.0561)1921(0.0561)1922(0.0560)1923(0.0561)1924(0.0592)1925Comparison Mean (White)19260.6619270.6619280.6619290.6619300.661931Share White Advert Home1932N1933Y1934Y1935Y1936Y1937ln(Price) Advert Home1938N1939N1940Y1941Y1942Y1943Racial Comp Advert Home1944N1945N1946N1947Y1948Y1949Poverty Share Advert Home1950N1951N1952N1953N1954Y1955Year1956Y1957Y1958Y1959Y1960Y1961Month of Year1962Y1963Y1964Y1965Y1966Y1967Observations196810,862196910,829197010,808197110,808197210,8081973Adjusted R219740.862419750.862119760.859919770.859919780.86891979Note: Coefficients report differences in the predicted race groups of home buyers and sale prices (recorded in ZTRAX1980transactions data) of properties recommended to minority testers relative to a white tester (the omitted category). Top1981panel reports estimates from a regression where the outcome is the predicted racial/ethnic identity associated with a1982first/last name of a ZTRAX home buyer. Differentiability score is calculated as the difference in predicted probability of1983the most likely buyer race group and the second most likely buyer race group. Buyers(n) reports the number of buyers from1984each predicted group that are observed in the ZTRAX transactions data. Bottom panel reports estimates from a regression1985where the outcome is the transaction price of the property. Racial minority encompasses all individual minority groups. All1986regression specifications include trial fixed effects and control for the full set of actor characteristics, assigned characteristics,1987and search characteristics. Actor characteristics: tester income, tester household income, gender of tester, age of tester,1988month of test, educational attainment of tester; Assigned characteristics: household members, current home ownership1989status, current lease type, car ownership status, reason for moving, years in current residence, length of employment at1990current job, reason tester can afford down payment, current lease assigned to tester; Search characteristics: month of test,1991sequence of tester appointments, and time of the appointment (am/pm). Standard errors are clustered by trial, which is1992the level of randomization. Significance levels are: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.199350199419951996===== PAGE 53 (score 0, numeric-paren 0, table-words 0, var-words 0) =====19971998Figure 1. Markets in 2012 HUD Buyer Experiment1999Note: Red points identify the location of 28 markets utilized in 2012 HDS study. See Turner et al. (2013) for detailed2000discussion of sampling design and methodology.200151200220032004===== PAGE 59 (score 66, numeric-paren 32, table-words 9, var-words 8) =====20052006C2007Steering and Neighborhood Racial Composition2008This section replicates estimates of steering by neighborhood racial composition reported2009as white shares in Tables 6 and 7 using shares of each minority group.2010Table C.1. Differential Steering and Neighborhood Racial Composition:2011African American Share2012Dependent variable: African American Household Share2013I2014II2015III2016IV2017V2018Racial Minority20190.0155∗∗20200.0156∗∗20210.0156∗∗20220.0157∗∗20230.0156∗∗2024(0.0073)2025(0.0073)2026(0.0073)2027(0.0073)2028(0.0072)2029African American20300.0388∗∗∗20310.0389∗∗∗20320.0390∗∗∗20330.0393∗∗∗20340.0396∗∗∗2035(0.0107)2036(0.0107)2037(0.0107)2038(0.0107)2039(0.0107)2040Hispanic20410.012620420.012620430.012620440.012620450.01262046(0.0086)2047(0.0086)2048(0.0086)2049(0.0086)2050(0.0086)2051Asian2052−0.00682053−0.00682054−0.00682055−0.00672056−0.00672057(0.0095)2058(0.0095)2059(0.0095)2060(0.0095)2061(0.0094)2062Comparison Mean20630.0920640.0920650.0920660.0920670.092068Share Black Advert Home2069N2070Y2071Y2072Y2073Y2074ln(Price) Advert Home2075N2076N2077Y2078Y2079Y2080Racial Comp Advert Home2081N2082N2083N2084Y2085Y2086Poverty Share Advert Home2087N2088N2089N2090N2091Y2092Observations209321,904209421,893209521,871209621,871209721,8712098Adjusted R220990.771921000.772021010.769321020.769321030.76932104Table C.2. Differential Steering and Neighborhood Racial Composition by Income:2105African American Share2106Dep. Variable: African American Household Share by Income2107High Inc2108Mid Inc2109Low Inc2110Racial Minority21110.0019∗21120.0097∗∗∗21130.00502114(0.0011)2115(0.0036)2116(0.0041)2117African American21180.0051∗∗∗21190.0175∗∗∗21200.0176∗∗∗2121(0.0016)2122(0.0061)2123(0.0058)2124Hispanic21250.000521260.0068∗21270.00552128(0.0015)2129(0.0041)2130(0.0049)2131Asian2132−0.000021330.00302134−0.0095∗2135(0.0015)2136(0.0042)2137(0.0056)2138Comparison Mean21390.0121400.0421410.042142Share White Advert Home2143N2144Y2145Y2146ln(Price) Advert Home2147N2148N2149Y2150Racial Comp Advert Home2151N2152N2153N2154Poverty Share Advert Home2155N2156N2157N2158Observations215921,470216021,470216121,4702162Adjusted R221630.729921640.762521650.67062166Note: Coefficients report differences in the racial composition (by income) of neighborhoods recommended to minority2167testers relative to a white tester (the omitted category). Racial minority encompasses all individual minority groups. All2168regression specifications control for the full set of actor characteristics, assigned characteristics, and search Characteristics.2169Actor characteristics: tester income, tester household income, gender of tester, age of tester, month of test, educational2170attainment of tester; Assigned Characteristics: household members, current home ownership status, current lease type, car2171ownership status, reason for moving, years in current residence, length of employment at current job, reason tester can afford2172down payment, current lease assigned to tester; Search Characteristics: month of test, sequence of tester appointments, time2173of the appointment (am/pm), type of recommended building, total number of homes recommended to tester, availability2174of advertised home as stated by agent. Standard errors are clustered by market for consistency with sampling design.2175Significance levels are: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.2176Appendix–5217721782179===== PAGE 60 (score 65, numeric-paren 32, table-words 9, var-words 7) =====21802181Table C.3. Differential Steering and Neighborhood Racial Composition:2182Hispanic Share2183Dependent variable: Hispanic Household Share2184I2185II2186III2187IV2188V2189Racial Minority21900.003221910.003321920.003421930.003421940.00342195(0.0070)2196(0.0070)2197(0.0070)2198(0.0070)2199(0.0070)2200African American22010.015922020.016022030.016222040.016122050.01602206(0.0103)2207(0.0103)2208(0.0103)2209(0.0103)2210(0.0103)2211Hispanic2212−0.00162213−0.00142214−0.00132215−0.00132216−0.00132217(0.0091)2218(0.0091)2219(0.0091)2220(0.0091)2221(0.0091)2222Asian2223−0.00942224−0.00922225−0.00932226−0.00932227−0.00932228(0.0088)2229(0.0088)2230(0.0088)2231(0.0088)2232(0.0088)2233Comparison Mean22340.1322350.1322360.1322370.1322380.132239Share Hispanic Advert Home2240N2241Y2242Y2243Y2244Y2245ln(Price) Advert Home2246N2247N2248Y2249Y2250Y2251Racial Comp Advert Home2252N2253N2254N2255Y2256Y2257Poverty Share Advert Home2258N2259N2260N2261N2262Y2263Observations226421,904226521,893226621,871226721,871226821,8712269Adjusted R222700.843722710.843722720.843722730.843722740.84372275Table C.4. Differential Steering and Neighborhood Racial Composition by Income:2276Hispanic Share2277Dep. Variable: Hispanic Household Share by Income2278High Inc2279Mid Inc2280Low Inc2281Racial Minority22820.00152283−0.001022840.00302285(0.0017)2286(0.0039)2287(0.0043)2288African American22890.0053∗∗22900.004422910.00542292(0.0026)2293(0.0048)2294(0.0072)2295Hispanic2296−0.00162297−0.004822980.00502299(0.0023)2300(0.0056)2301(0.0056)2302Asian2303−0.00122304−0.00422305−0.00232306(0.0024)2307(0.0050)2308(0.0047)2309Comparison Mean23100.0323110.0723120.032313Share White Advert Home2314N2315Y2316Y2317ln(Price) Advert Home2318N2319N2320Y2321Racial Comp Advert Home2322N2323N2324N2325Poverty Share Advert Home2326N2327N2328N2329Observations233021,470233121,470233221,4702333Adjusted R223340.652123350.832823360.71152337Note: Coefficients report differences in the racial composition (by income) of neighborhoods recommended to minority2338testers relative to a white tester (the omitted category). Racial minority encompasses all individual minority groups. All2339regression specifications control for the full set of actor characteristics, assigned characteristics, and search Characteristics.2340Actor characteristics: tester income, tester household income, gender of tester, age of tester, month of test, educational2341attainment of tester; Assigned Characteristics: household members, current home ownership status, current lease type, car2342ownership status, reason for moving, years in current residence, length of employment at current job, reason tester can afford2343down payment, current lease assigned to tester; Search Characteristics: month of test, sequence of tester appointments, time2344of the appointment (am/pm), type of recommended building, total number of homes recommended to tester, availability2345of advertised home as stated by agent. Standard errors are clustered by market for consistency with sampling design.2346Significance levels are: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.2347Appendix–6234823492350===== PAGE 61 (score 65, numeric-paren 32, table-words 9, var-words 7) =====23512352Table C.5. Differential Steering and Neighborhood Racial Composition:2353Asian Share2354Dependent variable: Asian Household Share2355I2356II2357III2358IV2359V2360Racial Minority2361−0.0097∗2362−0.0098∗2363−0.0098∗2364−0.0095∗2365−0.0097∗2366(0.0053)2367(0.0053)2368(0.0053)2369(0.0053)2370(0.0053)2371African American2372−0.0150∗∗2373−0.0149∗∗2374−0.0149∗∗2375−0.0147∗∗2376−0.0148∗∗2377(0.0069)2378(0.0068)2379(0.0068)2380(0.0068)2381(0.0068)2382Hispanic2383−0.0137∗2384−0.0140∗2385−0.0140∗2386−0.0136∗2387−0.0136∗2388(0.0079)2389(0.0078)2390(0.0079)2391(0.0078)2392(0.0078)2393Asian2394−0.00142395−0.00132396−0.00132397−0.00112398−0.00112399(0.0065)2400(0.0065)2401(0.0065)2402(0.0064)2403(0.0064)2404Comparison Mean24050.0724060.0724070.0724080.0724090.072410Share Asian Advert Home2411N2412Y2413Y2414Y2415Y2416ln(Price) Advert Home2417N2418N2419Y2420Y2421Y2422Racial Comp Advert Home2423N2424N2425N2426Y2427Y2428Poverty Share Advert Home2429N2430N2431N2432N2433Y2434Observations243521,904243621,893243721,871243821,871243921,8712440Adjusted R224410.743324420.743324430.743324440.743324450.74322446Table C.6. Differential Steering and Neighborhood Racial Composition by Income:2447Asian Share2448Dep. Variable: Asian Household Share by Income2449High Inc2450Mid Inc2451Low Inc2452Racial Minority2453−0.00202454−0.00362455−0.00222456(0.0019)2457(0.0027)2458(0.0022)2459African American2460−0.00332461−0.0067∗∗2462−0.00392463(0.0023)2464(0.0033)2465(0.0029)2466Hispanic2467−0.00342468−0.00462469−0.0055∗2470(0.0030)2471(0.0041)2472(0.0028)2473Asian24740.00132475−0.000224760.00112477(0.0023)2478(0.0033)2479(0.0027)2480Comparison Mean24810.0124820.0324830.022484Share White Advert Home2485N2486Y2487Y2488ln(Price) Advert Home2489N2490N2491Y2492Racial Comp Advert Home2493N2494N2495N2496Poverty Share Advert Home2497N2498N2499N2500Observations250121,470250221,470250321,4702504Adjusted R225050.687025060.723525070.62582508Note: Coefficients report differences in the racial composition (by income) of neighborhoods recommended to minority2509testers relative to a white tester (the omitted category). Racial minority encompasses all individual minority groups. All2510regression specifications control for the full set of actor characteristics, assigned characteristics, and search Characteristics.2511Actor characteristics: tester income, tester household income, gender of tester, age of tester, month of test, educational2512attainment of tester; Assigned Characteristics: household members, current home ownership status, current lease type, car2513ownership status, reason for moving, years in current residence, length of employment at current job, reason tester can afford2514down payment, current lease assigned to tester; Search characteristics: month of test, sequence of tester appointments, time2515of the appointment (am/pm), type of recommended building, total number of homes recommended to tester, availability2516of advertised home as stated by agent. Standard errors are clustered by market for consistency with sampling design.2517Significance levels are: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.2518Appendix–7251925202521===== PAGE 62 (score 6, numeric-paren 0, table-words 2, var-words 2) =====25222523D2524Comparison with Specifications Used in HDS 20122525Report2526This section examines differences in steering into high poverty neighborhoods reported2527in the present study versus differences documented by Turner et al. (2013) in the initial2528HDS 2012 report. Replication code is not available for the steering-into-poverty tests or2529racial composition test from the HDS 2012 report. We therefore construct tests that are2530as similar as possible to the description provided in Turner et al. (2013) using publicly2531available data files that contain census tract measures of poverty rates and white shares2532used by the authors.2533Table D.1 reports estimates of differences for African Americans relative to white2534testers using Census tract-level data (Columns 1-2) and tester averages (Columns 3-4)2535from the prior report. We compare results on white shares and poverty rates using the2536primary specification with block group measures used in the present study (Table 8) to2537comparable specifications with/without controls that use census tract measures (Columns25381-2) and specifications that aggregate recommendations by tester and also use census tract2539measures (Columns 3-4). The tests reported in the final two columns are the most similar2540to those reported in Turner et al. (2013) though there are differences in the samples and2541we stress that this is an approximate and not an exact replication. We note a substantial2542number of null values in RHGEO dataset, which contains the census tract data used in2543Turner et al. (2013). This reduces the sample size relative to the tester dataset that is2544merged with block group measures. Both sets of results indicate that the power and the2545magnitude of effects diminishes as a result of aggregation to the trial level and to the2546census tract. The choice of controls also affects estimate magnitudes, but less so.2547In both cases, estimates of differences become substantially smaller and tests less2548powered when we measure outcomes at the census tract level and aggregate to the level of2549a tester recommendation set (comparing differences in tester means). We find a difference2550of 2.8% [0.87] for differences in white shares when the test is constructed in this way,2551which is larger but not statistically different from the the 1.8% [0.8] difference reported2552in Turner et al. (2013). Differences in poverty rates fall by an order of magnitude and2553become statistically non-significant.2554Using this specification, we estimate that homes2555recommended to African American testers have a .27% [0.29] higher poverty share than2556those recommended to a white tester. This is very similar to the difference of .2% [0.2]2557that is documented in the HDS report. Variation in poverty rates between block groups2558appears to be particularly diminished when using Census tract measures. For instance,2559while the tract-level poverty rate (8.5%) is similar to the block group-level poverty rate2560(8.6%) for the average home in the HDS sample, the standard deviation of neighborhood2561poverty rates for the set of homes within a given trial is substantially smaller when2562using the tract-level measure of poverty: 3.79% at the tract level vs. 5.02% at the block2563group level. The standard deviation of block group-level poverty rates within the average2564census tract where recommendations are made in the study also appears to be quite2565large: 3%. When considering that the average neighborhood level poverty rate is 8.6%49,2566it becomes clear that using tract-level measures may mask considerable heterogeneity in2567the neighborhoods that testers are steered into.256849The average neighborhood level poverty rate is 8.5% when measured at the tract level.2569Appendix–8257025712572===== PAGE 63 (score 31, numeric-paren 8, table-words 7, var-words 7) =====25732574Table D.2. Differential Steering and White Share2575Dep. Variable: White Share2576Tract2577Tract2578Tester Mean2579Tester Mean2580African American2581−0.0315∗2582−0.0358∗∗2583−0.0280∗∗∗2584−0.0294∗∗2585(0.0166)2586(0.0164)2587(0.0087)2588(0.0135)2589Poverty Rate Advert Home2590N2591Y2592N2593Y2594Racial Comp Advert Home2595N2596Y2597N2598Y2599ln(Price) Advert Home2600N2601Y2602N2603Y2604Observations260515,270260614,342260749826084802609Adjusted R226100.832026110.829326120.759826130.74922614Dep. Variable: Poverty Rate2615Tract2616Tract2617Tester-Tract2618Tester-Tract2619African American26200.009326210.007726220.002726230.00272624(0.0076)2625(0.0075)2626(0.0032)2627(0.0029)2628Poverty Rate Advert Home2629N2630Y2631N2632Y2633Racial Comp Advert Home2634N2635Y2636N2637Y2638ln(Price) Advert Home2639N2640Y2641N2642Y2643Observations264415,269264514,341264649826474802648Adjusted R226490.587726500.592626510.672126520.70712653Note: Coefficients report differences in the poverty rates in neighborhoods recommended to minority testers relative to a2654white tester (the omitted category). Columns 1-2 report estimates from tests that measure poverty rates at the census2655Tract level whereas columns 3-4 report estimates using tester averages (data come from (Turner et al., 2013). All regression2656specifications control for the full set of actor characteristics, assigned characteristics, and search Characteristics. Actor2657characteristics: tester income, tester household income, gender of tester, age of tester, month of test, educational attainment2658of tester; Assigned Characteristics: household members, current home ownership status, current lease type, car ownership2659status, reason for moving, years in current residence, length of employment at current job, reason tester can afford down2660payment, current lease assigned to tester; Search characteristics: month of test, sequence of tester appointments, time2661of the appointment (am/pm), type of recommended building, total number of homes recommended to tester, availability2662of advertised home as stated by agent. Standard errors are clustered by market for consistency with sampling design.2663Significance levels are: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.2664Appendix–9266526662667===== PAGE 64 (score 3, numeric-paren 0, table-words 1, var-words 1) =====26682669E2670Steering and Neighborhood Effects: Controls2671This section compares three alternate specifications for the main tests reported in Table26728. The first column under each attribute reports estimates from specifications that drop2673controls for the attributes of advertised homes (buyer preferences). Column 2 reports2674estimates from the preferred specification in the paper. Estimates do not change when2675the implied preferences of buyers are added as controls, indicating that these do not2676appear to affect differences in recommendations made to minority buyers relative to a2677white counterpart. Column 3 adds a control for the average level of the particular neigh-2678borhood attribute being considered among recommended homes within a tight interval2679of price (within intervals of $20,000). Reported estimates of differences do not change2680with the addition of this control, suggesting that minorities are steered into homes with2681worse attributes than others at a similar price point and that the neighborhood steer-2682ing behavior documented in this study is not explained by differential steering of buyers2683into neighborhoods that are simply more affordable. This evidence is consistent with the2684findings reported in Appendix F.2685Appendix–10268626872688===== PAGE 65 (score 283, numeric-paren 211, table-words 13, var-words 11) =====26892690Table E.1. Discriminatory Steering and Neighborhood Effects: Robustness2691Elem School Test Scores2692Middle School Test Scores2693Assaults2694Racial Minority2695-0.0358*2696-0.044**2697-0.049**2698-0.043**2699-0.033*2700-0.036**27017.798327028.359127033.49362704(0.0210)2705(0.0227)2706(0.0222)2707(0.0179)2708(0.0178)2709(0.0176)2710(6.0232)2711(6.4744)2712(6.5735)2713African American2714-0.06252715-0.086**2716-0.095**2717-0.01***2718-0.09***2719-0.09***272019.213**272123.115**272217.600*2723(0.0385)2724(0.0424)2725(0.0401)2726(0.0239)2727(0.0242)2728(0.0238)2729(7.7954)2730(9.8939)2731(10.0867)2732Hispanic27330.00132734-0.00562735-0.01212736-0.001027370.010027380.009627390.38262740-0.46432741-4.70862742(0.0338)2743(0.0349)2744(0.0353)2745(0.0291)2746(0.0294)2747(0.0291)2748(8.0997)2749(8.0628)2750(8.3707)2751Asian2752-0.04092753-0.04422754-0.04482755-0.03122756-0.02872757-0.033627582.348027595.158227600.26062761(0.0374)2762(0.0380)2763(0.0387)2764(0.0240)2765(0.0237)2766(0.0236)2767(8.3593)2768(7.6687)2769(7.9009)2770Comparison Mean (White)27710.2927720.2927730.2927740.1827750.1827760.18277780.85277880.85277980.852780Outcome Advert Home2781N2782Y2783Y2784N2785Y2786Y2787N2788Y2789Y2790Racial Comp Advert Home2791N2792Y2793Y2794N2795Y2796Y2797N2798Y2799Y2800ln(Price) Advert Home2801N2802Y2803Y2804N2805Y2806Y2807N2808Y2809Y2810Outcome by Price Bin2811N2812N2813Y2814N2815N2816Y2817N2818N2819Y2820Observations282110,00428229,36028239,297282410,25828259,71328269,652282714,385282811,161282911,1012830Adjusted R228310.769028320.762528330.764728340.913928350.912828360.914928370.776328380.797128390.80002840Elem School2841Poverty Rate2842High Skill2843Racial Minority2844-0.22662845-0.3424*2846-0.3465*28470.0115**28480.0116**28490.0125**2850-0.025**2851-0.025**2852-0.02**2853(0.1622)2854(0.1824)2855(0.1802)2856(0.0055)2857(0.0054)2858(0.0054)2859(0.0100)2860(0.0099)2861(0.0094)2862African American2863-0.43**2864-0.47**2865-0.54**28660.011428670.012028680.01232869-0.03**2870-0.03**2871-0.025**2872(0.2035)2873(0.2379)2874(0.2322)2875(0.0076)2876(0.0075)2877(0.0075)2878(0.0131)2879(0.0130)2880(0.0124)2881Hispanic2882-0.10122883-0.24622884-0.228228850.02***28860.02***28870.02***2888-0.04***2889-0.04***2890-0.03**2891(0.2246)2892(0.2423)2893(0.2462)2894(0.0076)2895(0.0076)2896(0.0076)2897(0.0132)2898(0.0130)2899(0.0127)2900Asian2901-0.06792902-0.38*2903-0.37072904-0.00052905-0.000929060.00062907-0.00702908-0.00722909-0.00422910(0.1962)2911(0.2195)2912(0.2255)2913(0.0070)2914(0.0069)2915(0.0069)2916(0.0129)2917(0.0128)2918(0.0120)2919Comparison Mean (White)29206.4329216.4329226.4329230.08229240.08229250.08229260.4929270.4929280.492929Outcome Advert Home2930N2931Y2932Y2933N2934Y2935Y2936N2937Y2938Y2939Racial Comp Advert Home2940N2941Y2942Y2943N2944Y2945Y2946N2947Y2948Y2949ln(Price) Advert Home2950N2951Y2952Y2953N2954Y2955Y2956N2957Y2958Y2959Outcome by Price Bin2960N2961N2962Y2963N2964N2965Y2966N2967N2968Y2969Observations297014,401297111,032297210,984297321,904297421,871297521,744297621,904297721,871297821,7442979Adjusted R229800.740929810.747529820.749529830.520029840.520529850.530229860.692729870.692529880.70132989College2990Single-Parent HH2991Ownership Ratel2992Racial Minority2993-0.02*2994-0.02*2995-0.02*29960.00729970.007729980.00862999-0.01593000-0.01703001-0.01863002(0.0101)3003(0.0101)3004(0.0099)3005(0.0062)3006(0.0062)3007(0.0062)3008(0.0142)3009(0.0142)3010(0.0139)3011African American3012-0.04***3013-0.03***3014-0.04***30150.015*30160.017*30170.017*3018-0.02553019-0.02633020-0.02513021(0.0127)3022(0.0127)3023(0.0126)3024(0.0088)3025(0.0087)3026(0.0088)3027(0.0184)3028(0.0184)3029(0.0178)3030Hispanic3031-0.02073032-0.0212*3033-0.023*30340.014130350.016*30360.018**30370.008530380.005930390.00393040(0.0126)3041(0.0127)3042(0.0124)3043(0.0086)3044(0.0084)3045(0.0087)3046(0.0202)3047(0.0203)3048(0.0202)3049Asian30500.002730510.002930520.00243053-0.00653054-0.00613055-0.00563056-0.02343057-0.02323058-0.02593059(0.0142)3060(0.0142)3061(0.0138)3062(0.0086)3063(0.0085)3064(0.0085)3065(0.0177)3066(0.0176)3067(0.0174)3068Comparison Mean (White)30690.5330700.5330710.5330720.1430730.1430740.1430750.7430760.7430770.743078Outcome Advert Home3079N3080Y3081Y3082N3083Y3084Y3085N3086Y3087Y3088Racial Comp Advert Home3089N3090Y3091Y3092N3093Y3094Y3095N3096Y3097Y3098ln(Price) Advert Home3099N3100Y3101Y3102N3103Y3104Y3105N3106Y3107Y3108Outcome by Price Bin3109N3110N3111Y3112N3113N3114Y3115N3116N3117Y3118Observations311921,904312021,871312121,744312221,904312321,871312421,744312521,904312621,871312721,7443128Adjusted R231290.755031300.754831310.761031320.523631330.523731340.531031350.602231360.602431370.60773138Superfund3139Toxics3140PM3141Racial Minority31420.10***31430.10***31440.11***3145779.06**3146765.91**3147682.7*3148-0.04383149-0.04383150-0.02643151(0.0326)3152(0.0325)3153(0.0332)3154(326.3205)3155(324.9800)3156(359.3615)3157(0.0504)3158(0.0504)3159(0.0531)3160African American31610.13***31620.13***31630.13***3164891.8**3165872.9**3166690.973167-0.0253168-0.0253169-0.0223170(0.0467)3171(0.0465)3172(0.0457)3173(409.0238)3174(403.4626)3175(468.2560)3176(0.0678)3177(0.0678)3178(0.0690)3179Hispanic31800.057831810.059131820.06033183471.32703184459.72833185456.424531860.006331870.010231880.04573189(0.0381)3190(0.0376)3191(0.0392)3192(546.7939)3193(544.8235)3194(559.8852)3195(0.0639)3196(0.0638)3197(0.0684)3198Asian31990.10**32000.11**32010.09*3202909.50**3203899.51**3204885.51*3205-0.12**3206-0.12**3207-0.113208(0.0484)3209(0.0484)3210(0.0480)3211(429.1031)3212(424.3961)3213(473.3210)3214(0.0618)3215(0.0617)3216(0.0649)3217Comparison Mean (White)32180.3632190.3632200.3632215100322251003223510032249.432259.432269.43227Outcome Advert Home3228N3229Y3230Y3231N3232Y3233Y3234N3235Y3236Y3237Racial Comp Advert Home3238N3239Y3240Y3241N3242Y3243Y3244N3245Y3246Y3247ln(Price) Advert Home3248N3249Y3250Y3251N3252Y3253Y3254N3255Y3256Y3257Outcome by Price Bin3258N3259N3260Y3261N3262N3263Y3264N3265N3266Y3267Observations326821,916326921,883327021,756327121,916327221,883327321,756327421,916327521,883327621,7563277Adjusted R232780.874532790.874432800.877732810.694132820.693932830.694532840.963532850.963432860.96413287Note: Coefficients report differences in the attributes of neighborhoods recommended to minority testers relative to a3288white tester (the omitted category) using different sets of controls.3289The 3rd specification for each variable contains a3290fixed effect using tight bins of list price ($20,000) for the recommended home. Racial minority encompasses all individual3291minority groups.3292All regression specifications control for the full set of actor characteristics, assigned characteristics,3293and search Characteristics. Actor characteristics: tester income, tester household income, gender of tester, age of tester,3294month of test, educational attainment of tester; Assigned Characteristics: household members, current home ownership3295status, current lease type, car ownership status, reason for moving, years in current residence, length of employment at3296current job, reason tester can afford down payment, current lease assigned to tester; Search Characteristics: month of test,3297sequence of tester appointments, time of the appointment (am/pm), type of recommended building, total number of homes3298recommended to tester, availability of advertised home as stated by agent. Standard errors are clustered by market for3299consistency with sampling design. Significance levels are: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.3300Appendix–11330133023303===== PAGE 66 (score 39, numeric-paren 12, table-words 8, var-words 8) =====33043305F3306Discriminatory Steering and Home Price3307This section provides a test of differences in the prices of homes recommended to minority3308testers relative to a white counterpart. We do not find strong evidence that minority3309testers receive recommendations for lower priced homes relative to their white paired3310testers, though we do find a marginally significant price reduction for Hispanic/LatinX3311testers. Looking at point estimates alone, it appears that homes recommended to African3312American and Asian buyers may have somewhat higher list prices on average. Along with3313the tests of differences in transaction prices reported in Table 14 and within tight intervals3314of list price reported in Column 3 of Table E.1, we take these findings as evidence that3315the primary results on steering that are documented in this paper are not explained by3316steering of minority buyers into more affordable neighborhoods. There is no evidence3317that this is the case for African American testers in particular, who are subject to the3318strongest steering effects.3319Table F.1. Discriminatory Steering and Home Price3320Dependent variable: log(Price)3321133222332333324Racial Minority33250.002433260.003433270.01203328(0.1047)3329(0.1055)3330(0.0994)3331African American33320.077633330.076433340.09373335(0.1633)3336(0.1634)3337(0.1583)3338Hispanic3339−0.15623340−0.15263341−0.1504∗3342(0.0988)3343(0.0996)3344(0.0908)3345Asian33460.064233470.064233480.06653349(0.0966)3350(0.0970)3351(0.0904)3352Comparison Mean (White)335312.40335412.40335512.403356Racial Comp Advert Home3357N3358Y3359Y3360ln(Price) Advert Home3361N3362N3363Y3364Observations336521,787336621,776336721,7633368Adjusted R233690.722333700.722233710.72413372Note: Coefficients report differences in the list prices of homes recommended to minority testers relative to a white tester3373(the omitted category). Racial minority encompasses all individual minority groups. All regression specifications control3374for the full set of actor characteristics, assigned characteristics, and search Characteristics. Actor characteristics: tester3375income, tester household income, gender of tester, age of tester, month of test, educational attainment of tester; Assigned3376Characteristics: household members, current home ownership status, current lease type, car ownership status, reason for3377moving, years in current residence, length of employment at current job, reason tester can afford down payment, current3378lease assigned to tester; Search Characteristics: month of test, sequence of tester appointments, time of the appointment3379(am/pm), type of recommended building, total number of homes recommended to tester, availability of advertised home3380as stated by agent. Standard errors are clustered by market for consistency with sampling design. Significance levels are:3381∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.3382Appendix–12338333843385===== PAGE 67 (score 96, numeric-paren 57, table-words 8, var-words 14) =====33863387G3388Discriminatory Steering and Implied Preferences3389This section explores interactions between the implied preferences of buyers and dis-3390criminatory constraints facing buyers. Table G.1 tests for differences within the sample3391of buyers who present an advertised listing that falls in a census block group with an3392above-median white share. The results suggest stronger constraints for African American3393testers, in particular, that search in these neighborhoods. Homes recommended to these3394buyers are located in attendance zones with considerably lower elementary school test3395scores and neighborhoods with greater numbers of assaults, higher poverty rates, lower3396skill levels, lower college attendance rates, higher shares of single-parent households, lower3397home ownership rates, and higher exposures to toxic pollution.3398Table G.1. Discriminatory Steering: White Neighborhoods3399School-Specific Test Scores:3400Housing Search Platform:3401Elem School3402Middle School3403Assaults3404Elem School3405African American3406-0.4317***3407-0.0670**340847.1246***3409-2.06783410(0.0418)3411(0.0304)3412(9.2629)3413(1.7155)3414Hispanic3415-0.1327*34160.01973417-45.7471***34181.05443419(0.0697)3420(0.0309)3421(11.7732)3422(0.9374)3423Asian3424-0.1533**3425-0.0081342618.5930***34271.03083428(0.0685)3429(0.0274)3430(6.4357)3431(1.3990)3432Comparison Mean (White)34330.3734340.2834355134366.703437Observations34385,02134395,21734405,57434415,4153442Adjusted R234430.841934440.924034450.910634460.75693447American Community Survey3448Poverty Rate3449High Skill3450College3451Single-Parent HH3452Ownership Rate3453African American34540.0293***3455-0.0550***3456-0.0434***34570.0203*3458-0.0479**3459(0.0083)3460(0.0149)3461(0.0149)3462(0.0106)3463(0.0216)3464Hispanic34650.0329***3466-0.02153467-0.010834680.0154*3469-0.03723470(0.0077)3471(0.0173)3472(0.0186)3473(0.0087)3474(0.0235)3475Asian34760.00403477-0.01493478-0.0621***34790.0278*3480-0.0639**3481(0.0108)3482(0.0199)3483(0.0213)3484(0.0146)3485(0.0298)3486Comparison Mean (White)34870.0734880.5034890.5434900.1234910.773492Observations349310,975349410,975349510,975349610,975349710,9753498Adjusted R234990.463135000.711135010.780635020.479735030.56853504Pollution3505Superfund3506Toxics3507PM3508African American35090.031535101,537.3840***35110.05003512(0.0400)3513(480.2213)3514(0.0727)3515Hispanic35160.1098**3517779.818735180.08283519(0.0437)3520(524.6936)3521(0.0875)3522Asian35230.1709***3524328.77713525-0.2565**3526(0.0622)3527(519.5311)3528(0.1008)3529Comparison Mean (White)35300.303531590035329.803533Observations353410,977353510,977353610,9773537Adjusted R235380.864235390.591635400.94383541Note: Coefficients report differences in the attributes of properties/neighborhoods recommended to minority testers relative3542to a white tester (the omitted category) for the subset of trials where the advertised home falls above the 50th percentile3543in the block group share of white households. Racial minority encompasses all individual minority groups. All regression3544specifications include trial fixed effects and control for the full set of actor characteristics, assigned characteristics, and3545search characteristics. Actor characteristics: tester income, tester household income, gender of tester, age of tester, month3546of test, educational attainment of tester; Assigned characteristics: household members, current home ownership status,3547current lease type, car ownership status, reason for moving, years in current residence, length of employment at current job,3548reason tester can afford down payment, current lease assigned to tester; Search Characteristics: month of test, sequence of3549tester appointments, and time of the appointment (am/pm). Standard errors are clustered by trial, which is the level of3550randomization. Significance levels are: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.3551Appendix–13355235533554===== PAGE 68 (score 4, numeric-paren 0, table-words 1, var-words 2) =====35553556Table G.2 tests for differences within the sample of buyers who present an advertised3557listing that falls above the median price in the audit ($245,000). These results suggest3558that the differences in elementary and middle school test scores, poverty rates, skill,3559college attainment, single-parent households, and exposure to toxics may be amplified for3560African Americans who present these preferences.3561Appendix–14356235633564===== PAGE 69 (score 111, numeric-paren 67, table-words 10, var-words 13) =====35653566Table G.2. Discriminatory Steering and Neighborhood Effects: Higher-Priced Homes3567School-Specific Test Scores:3568Housing Search Platform:3569Elem School3570Middle School3571Assaults3572Elem School3573Racial Minority3574-0.028535750.1056***3576-10.9046*3577-0.01343578(0.0320)3579(0.0222)3580(6.3792)3581(0.2832)3582African American3583-0.1731***3584-0.02193585-22.2033**35860.14973587(0.0668)3588(0.0285)3589(10.8133)3590(0.3350)3591Hispanic35920.042035930.1119***3594-49.4148***3595-0.09653596(0.0378)3597(0.0190)3598(10.2786)3599(0.5100)3600Asian3601-0.2061***36020.0704**36032.27843604-0.45173605(0.0622)3606(0.0357)3607(7.7459)3608(0.5991)3609Comparison Mean (White)36100.436110.36361280.936136.943614ln(Price) Advert Home3615Y3616Y3617Y3618Y3619Racial Comp Advert Home3620Y3621Y3622Y3623Y3624Outcome Advertised Home3625Y3626Y3627Y3628Y3629Observations36303,79936313,83136325,29336334,9063634Adjusted R236350.837936360.928336370.839336380.74403639American Community Survey3640Poverty Rate3641High Skill3642College3643Single-Parent HH3644Ownership Rate3645Racial Minority36460.0146**3647-0.0208*3648-0.0332***36490.01063650-0.01183651(0.0062)3652(0.0118)3653(0.0111)3654(0.0068)3655(0.0169)3656African American36570.0254***3658-0.0390**3659-0.0599***36600.0157*3661-0.02233662(0.0078)3663(0.0171)3664(0.0150)3665(0.0095)3666(0.0239)3667Hispanic36680.01043669-0.02403670-0.02243671-0.002436720.01823673(0.0083)3674(0.0172)3675(0.0156)3676(0.0086)3677(0.0233)3678Asian36790.00633680-0.00033681-0.015636820.01333683-0.01413684(0.0080)3685(0.0158)3686(0.0151)3687(0.0084)3688(0.0207)3689Comparison Mean (White)36900.07136910.5436920.636930.1236940.733695ln(Price) Advert Home3696Y3697Y3698Y3699Y3700Y3701Racial Comp Advert Home3702Y3703Y3704Y3705Y3706Y3707Outcome Advertised Home3708Y3709Y3710Y3711Y3712Y3713Observations371410,850371510,850371610,850371710,850371810,8503719Adjusted R237200.558937210.682937220.723637230.545337240.66983725Pollution3726Superfund3727Toxics3728PM3729Racial Minority37300.065137311,360.2900**37320.01793733(0.0554)3734(589.0348)3735(0.0670)3736African American37370.07583738994.0474*37390.07083740(0.0831)3741(578.0719)3742(0.0868)3743Hispanic37440.024737452,239.0900**37460.11333747(0.0716)3748(907.8452)3749(0.0889)3750Asian37510.057837521,272.7980*3753-0.08963754(0.0648)3755(744.7305)3756(0.0939)3757Comparison Mean (White)37580.453759490037609.53761ln(Price) Advert Home3762Y3763Y3764Y3765Racial Comp Advert Home3766Y3767Y3768Y3769Outcome Advertised Home3770Y3771Y3772Y3773Observations377410,862377510,862377610,8623777Adjusted R237780.875837790.606337800.95993781Note: Coefficients report differences in the attributes of properties/neighborhoods recommended to minority testers relative3782to a white tester (the omitted category) for the subset of trials where the advertised home falls above the 50th percentile3783in the block group share of the price distribution (above $245,000). All regression specifications include trial fixed effects3784and control for the full set of actor characteristics, assigned characteristics, and search characteristics. Actor character-3785istics: tester income, tester household income, gender of tester, age of tester, month of test, educational attainment of3786tester; Assigned characteristics: household members, current home ownership status, current lease type, car ownership3787status, reason for moving, years in current residence, length of employment at current job, reason tester can afford down3788payment, current lease assigned to tester; Search Characteristics: month of test, sequence of tester appointments, time of3789the appointment (am/pm), type of recommended building. Standard errors are clustered by trial, which is the level of3790randomization. Standard errors are clustered by trial, which is the level of randomization. Significance levels are: ∗p<0.1;3791∗∗p<0.05; ∗∗∗p<0.01.3792Appendix–15379337943795===== PAGE 70 (score 59, numeric-paren 28, table-words 6, var-words 12) =====37963797H3798Sensitivity Analysis: Superfund Proximity3799This section explores the sensitivity of results to different distance-based measures of3800proximity to Superfund sites.3801We rely on proximity as our measure of exposure and3802evaluate the trade-offbetween different distances and statistical power. Table H.1 reports3803estimates using a range of distances between 3-6km, which is slightly wider than the 2-38043 mile range used in prior work (Gamper-Rabindran and Timmins, 2011, Greenstone3805and Gallagher, 2008). The number of sites for the comparison mean recommendation3806gets mechanically smaller with a tighter radius and so do our estimates of differences3807(in levels) for the minority group as a whole. Estimates do not decrease monotonically3808within each tester group, but are consistent in magnitude across the range. We do not3809have sufficient power to detect differences within 3km.3810Table H.1. Discriminatory Steering and Superfund Proximity by Distance3811Dependent variable: Superfund38123 Km38133.5 Km38144 Km38154.5 Km38165 Km38175.5 Km38186 Km3819Racial Minority38200.039238210.0782∗∗38220.0773∗∗38230.0813∗∗∗38240.1010∗∗∗38250.1010∗∗∗38260.1031∗∗∗3827(0.0280)3828(0.0316)3829(0.0313)3830(0.0306)3831(0.0325)3832(0.0365)3833(0.0387)3834African American38350.059538360.0789∗38370.0805∗38380.1041∗∗38390.1256∗∗∗38400.1118∗∗38410.1008∗∗3842(0.0364)3843(0.0406)3844(0.0431)3845(0.0435)3846(0.0465)3847(0.0491)3848(0.0488)3849Hispanic3850−0.021738510.042538520.059338530.051638540.059138550.034338560.03823857(0.0378)3858(0.0406)3859(0.0405)3860(0.0380)3861(0.0376)3862(0.0446)3863(0.0490)3864Asian38650.0820∗∗38660.1231∗∗∗38670.0963∗∗38680.0849∗38690.1046∗∗38700.1339∗∗38710.1528∗∗∗3872(0.0385)3873(0.0427)3874(0.0425)3875(0.0443)3876(0.0484)3877(0.0527)3878(0.0578)3879Comparison Mean (White)38800.1338810.1738820.2238830.2938840.3538850.4138860.483887ln(Price) Advert Home3888N3889Y3890N3891Y3892Y3893Y3894Y3895Racial Comp Advert Home3896N3897Y3898N3899Y3900Y3901Y3902Y3903Observations390421,883390521,883390621,883390721,883390821,883390921,883391021,8833911Adjusted R239120.802839130.841239140.843939150.859739160.874439170.886839180.90153919Note:3920∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.013921Note:3922Coefficients report differences in the number of Superfund sites within close proximity (3-6km) to proper-3923ties/neighborhoods recommended to minority testers relative to a white tester (the omitted category).3924All regression3925specifications include trial fixed effects and control for the full set of actor characteristics, assigned characteristics, and3926search characteristics. Actor characteristics: tester income, tester household income, gender of tester, age of tester, month3927of test, educational attainment of tester; Assigned characteristics: household members, current home ownership status,3928current lease type, car ownership status, reason for moving, years in current residence, length of employment at current job,3929reason tester can afford down payment, current lease assigned to tester; Search Characteristics: month of test, sequence3930of tester appointments, time of the appointment (am/pm), type of recommended building. Standard errors are clustered3931by trial, which is the level of randomization. Standard errors are clustered by trial, which is the level of randomization.3932Significance levels are: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.3933Appendix–163934