External Validity of Hedonic Price Estimates: Heterogeneity in the - - PowerPoint PPT Presentation

▶
external validity of hedonic price estimates
SMART_READER_LITE
LIVE PREVIEW

External Validity of Hedonic Price Estimates: Heterogeneity in the - - PowerPoint PPT Presentation

Sep 12, 2022 156 likes •315 views

External Validity of Hedonic Price Estimates: Heterogeneity in the Price Discount Associated with having Black and Hispanic Neighbors Lei Zhang, Tammy Leonard North Dakota State University, University of Dallas lei.zhang.3@ndsu.edu,

slide-1
SLIDE 1

External Validity of Hedonic Price Estimates: Heterogeneity in the Price Discount Associated with having Black and Hispanic Neighbors

Lei Zhang, Tammy Leonard

North Dakota State University, University of Dallas lei.zhang.3@ndsu.edu, tleonard@udallas.edu

AREUEA; May 29, 2020

Lei Zhang, Tammy Leonard (NDSU, UD) Hedonic Price Estimates AREUEA; May 29, 2020 1 / 14

slide-2
SLIDE 2

Outline

1

Motivation

2

Methods

3

Data

4

Results

5

Conclusions

Lei Zhang, Tammy Leonard (NDSU, UD) Hedonic Price Estimates AREUEA; May 29, 2020 2 / 14

slide-3
SLIDE 3

Motivation

Housing markets & race Some authors have found price discounts associated with black neighborhoods (Brasington et al., 2015; Myers, 2004; Raymond et. al., 2016). There remained concerns that studies may suffer from omitted variable bias. Omitted variable bias associated with housing locational characteristics has been addressed using the boundary approach. Bayer et al. 2007

No effect of neighborhood race on home price on average. Study based on the San Francisco Bay Area housing market.

Lei Zhang, Tammy Leonard (NDSU, UD) Hedonic Price Estimates AREUEA; May 29, 2020 3 / 14

slide-4
SLIDE 4

Research Questions

We apply the boundary approach to a sample of 180 US housing markets to assess the hedonic price of neighborhood racial composition (proportion black and proportion Hispanic).

1 External validity: are the hedonic prices of neighborhood race

statistically significant and are results consistent with the omitted variable problem suggested in the literature?

2 To what extent are variations in the estimates across markets

explained by across market heterogeneity in supply constraints and consumer preferences?

Lei Zhang, Tammy Leonard (NDSU, UD) Hedonic Price Estimates AREUEA; May 29, 2020 4 / 14

slide-5
SLIDE 5

Hedonic Model

ln(Pisb) = βXisb + αNi + δs + λb + τi + ǫisb ln(Pisb): natural log of house price Xisb: housing characteristics Ni: neighborhood block group characteristics including average educational attainment, % of households below poverty, % black households, % Hispanic households δs: elementary school fixed effect λb: attendance zone boundary fixed effect τi: year and month of sale fixed effects

Lei Zhang, Tammy Leonard (NDSU, UD) Hedonic Price Estimates AREUEA; May 29, 2020 5 / 14

slide-6
SLIDE 6

Boundary Method

(Bayer et al., 2007; Black, 1999; Dhar & Ross, 2012) Lei Zhang, Tammy Leonard (NDSU, UD) Hedonic Price Estimates AREUEA; May 29, 2020 6 / 14

slide-7
SLIDE 7

Hedonic Model

1 Naive regression: All observations in housing market included in

model; δs = λb = 0

2 Boundary Method: Only observations along school boundary included

in model

Lei Zhang, Tammy Leonard (NDSU, UD) Hedonic Price Estimates AREUEA; May 29, 2020 7 / 14

slide-8
SLIDE 8

Data

Sale price and housing characteristics data from Zillow Transaction and Assessment Dataset (ZTRAX)

>8 million records, but only those in counties with enough observations to conduct boundary analysis were kept. 3.4 million in analysis sample (naive regression); 1.2 million in boundary sample.

Block group characteristics came from the American Community Survey 5-year estimates Housing market supply-side charcteristics:

land share of housing cost (Census) land supply elasticity (Saiz, 2010) home price appreciation (Annual Freddie Mac House Price Index)

Lei Zhang, Tammy Leonard (NDSU, UD) Hedonic Price Estimates AREUEA; May 29, 2020 8 / 14

slide-9
SLIDE 9

Estimated Price Discounts for % Black

Y Y Y Y Y Y Y Y YY YY Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y

Percent Black

Legend Y Result was not significant Coefficient Estimate for % Black

  • 5.095 - -0.523
  • 0.523 - -0.310
  • 0.310 - -0.167
  • 0.167 - 0.000

0.000 - 1.702

.

270 540 810 1,080 135 Miles

Lei Zhang, Tammy Leonard (NDSU, UD) Hedonic Price Estimates AREUEA; May 29, 2020 9 / 14

slide-10
SLIDE 10

Estimated Price Discounts for % Hispanic

Y Y Y Y Y Y Y Y YY Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y YY Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y

Percent Hispanic

Legend Y Result was not significant at 95% level Coefficient Estimate for % Hispanic

  • 2.411 - -0.527
  • 0.527 - -0.260
  • 0.260 - -0.167
  • 0.167 - 0.000

0.000 - 1.942

.

270 540 810 1,080 135 Miles

Lei Zhang, Tammy Leonard (NDSU, UD) Hedonic Price Estimates AREUEA; May 29, 2020 10 / 14

slide-11
SLIDE 11

Is the Omitted Variable Bias in the Naive Regression as Expected?

  • 2
  • 1

1 2 % Black (Boundary Sample)

  • 2
  • 1

1 2 % Black (Full Analysis Sample)

  • 2
  • 1

1 2 % Hispanic (Boundary Sample)

  • 2
  • 1

1 2 % Hispanic (Full Analysis Sample)

Lei Zhang, Tammy Leonard (NDSU, UD) Hedonic Price Estimates AREUEA; May 29, 2020 11 / 14

slide-12
SLIDE 12

Correlates of Price Discounts for Black Neighborhoods

Mean(sd) Variable Range Model 1 Model 2 Model 3 Land share of housing cost 29.32 (8.90) [16.09 , 53.45] 0.00810

  • 0.0379+
  • 0.0372*

(0.00660) (0.0202) (0.0167) Land supply elasticity 1.93(.77) [.82 , 3.71] 0.207* (0.0799) House Price Appreciation Growth 5.71 (3.08) [.4 , 12] 0.00768

  • 0.0122

0.0311 (0.0117) (0.0234) (0.0519) GDP Growth 2.73 (.52) [1.2 , 3.7]

  • 0.0614
  • 0.149
  • 0.270

(0.0695) (0.193) (0.177) Latitude 39.02 (4.64) [26.58 , 48.82 ] 0.0898+ (0.0333) Longitude

  • 86.46 (14.65)

[-122.04 , -79.10] 0.0160 (0.0277) Northeast Census Region .28 0.390 (0.401) South Census Region .33 0.333 (0.305) West Census Region .16 0.0841 (0.501) N 28 64 64

  • adj. R2

0.285 0.137 0.184 F 1.952 1.130 1.610

+ p<0.10, * p<0.05

Lei Zhang, Tammy Leonard (NDSU, UD) Hedonic Price Estimates AREUEA; May 29, 2020 12 / 14

slide-13
SLIDE 13

Conclusions and Discussion

External validity of analysis based on a single housing market?

Using the boundary method, about a third of housing markets examined produce statistically significant price discounts for minority neighborhoods.

Price discounts for black neighborhoods were greatest in housing markets with lower housing supply elasticity and in the south. There were no discernable patterns in the price discounts associated with Hispanic neighborhoods.

Possible endogeneity between Hispanic neighborhoods and preferences for Hispanics? Hispanic neighbors may look like white neighbors?

Lei Zhang, Tammy Leonard (NDSU, UD) Hedonic Price Estimates AREUEA; May 29, 2020 13 / 14

slide-14
SLIDE 14

Thank you!

Lei Zhang, Tammy Leonard (NDSU, UD) Hedonic Price Estimates AREUEA; May 29, 2020 14 / 14