School Choice in Suburbia: Public School Testing and Private Real Estate Markets
description
Transcript of School Choice in Suburbia: Public School Testing and Private Real Estate Markets
Jack Dougherty, Jeffrey Harrelson, Laura Maloney, Drew Murphy,
Michael Snow, Russell Smith, and Diane Zannoni
Cities, Suburbs, and Schools Research ProjectTrinity College, Hartford CT
http://www.trincoll.edu/depts/educ/CSS
AERA meeting, April 2007
School Choice in Suburbia:Public School Testing
andPrivate Real Estate Markets
Question: How much more do homebuyers pay to choose a house on the higher-scoring side of a school boundary?
East School Attendance
Area
Public School Attendance Boundary
Study uses econometric analysis
West School Attendance
Area
A B
Question: How much more do homebuyers pay to choose a house on the higher-scoring side of a school boundary?
East School Attendance
Area
Public School Attendance Boundary
Study uses econometric analysis to control for- house characteristics
West School Attendance
Area
A B
Question: How much more do homebuyers pay to choose a house on the higher-scoring side of a school boundary?
East School Attendance
Area
Public School Attendance Boundary
Study uses econometric analysis to control for- house characteristics- school quality
West School Attendance
Area
A BTest Scores
Question: How much more do homebuyers pay to choose a house on the higher-scoring side of a school boundary?
Test Scores
East School Attendance
Area
Public School Attendance Boundary
Study uses econometric analysis to control for- house characteristics- school quality- neighborhood effects
West School Attendance
Area
A B
Question: How much more do homebuyers pay to choose a house on the higher-scoring side of a school boundary?
Test Scores
Test Scores
Context of the Study: West Hartford, CT
- one public school district with 11 elementary (K-5)
- 28 school boundaries
A boundary close-up:
A boundary close-up:
- all single-family homes sold during our study period (1996 - 2005)
A school attendance boundary close-up:
- all single-family homes sold during our study period (1996 - 2005)
- Neighborhood around a shared school attendance boundary
Our model builds on study by Sandra Black (1999):
House Price = function of (house characteristics,
school quality, neighborhood effects)
Our model builds on study by Sandra Black (1999):
House Price = function of (house characteristics,
school quality, neighborhood effects) - Logarithm of price of house, deflated to year 2000 dollars,
using price index of average sales in West Hartford
House Price = function of (house characteristics,
school quality, neighborhood effects) - Logarithm of price of house, deflated to year 2000 dollars,
using price index of average sales in West Hartford- Number of bedrooms, bathrooms, lot size, internal footage
House Price = function of (house characteristics,
school quality, neighborhood effects) - Logarithm of price of house, deflated to year 2000 dollars,
using price index of average sales in West Hartford- Number of bedrooms, bathrooms, lot size, internal footage- Percent of 4th graders at goal on CT Mastery Test (CMT),
as data appeared in newspaper graphics & internet
House Price = function of (house characteristics,
school quality, neighborhood effects) - Logarithm of price of house, deflated to year 2000 dollars,
using price index of average sales in West Hartford- Number of bedrooms, bathrooms, lot size, internal footage- Percent of 4th graders at goal on CT Mastery Test (CMT),
as data appeared in newspaper graphics & internetHartford Courant 1999
House Price = function of (house characteristics,
school quality, neighborhood effects) - Logarithm of price of house, deflated to year 2000 dollars,
using price index of average sales in West Hartford- Number of bedrooms, bathrooms, lot size, internal footage- Percent of 4th graders at goal on CT Mastery Test (CMT),
as data appeared in newspaper graphics & internet- Set of neighborhood dummy variables (rather than imperfect
census data), to account for unobservable neighborhood characteristics, and to avoid omitted variable bias
House Price = function of (house characteristics,
school quality, neighborhood effects) - Logarithm of price of house, deflated to year 2000 dollars,
using price index of average sales in West Hartford- Number of bedrooms, bathrooms, lot size, internal footage- Percent of 4th graders at goal on CT Mastery Test (CMT), as
data appeared in newspaper graphics & internet- Set of neighborhood dummy variables (rather than imperfect
census data), to account for unobservable neighborhood characteristics, and to avoid omitted variable bias
Sample = 8,736 single-family home sales, 1996-2005mean home price = $198,300 (in year 2000 dollars)mean test score = 73% (standard deviation 12 pct points)
Results: 1) Geographic Restriction & Variable Distance Analysis
2) Time Period Analysis (1996-2000 vs. 2001-2005)
3) Neighborhood Progression Analysis
Results: 1) Geographic Restriction & Variable Distance Analysis
What was test-price relationship when we used increasingly restrictive geography?
- Gradually eliminated school boundaries that followed rivers, parks, and major 4-lane roads, leaving behind only those drawn through residential areas (Set D)
Results: 1) Geographic Restriction & Variable Distance Analysis
What was test-price relationship when we varied the sample by distance to the school boundary?
Results: 1) Geographic Restriction & Variable Distance Analysis
What was test-price relationship when we varied the sample by distance to the school boundary?
Results: 1) Geographic Restriction & Variable Distance Analysis
What was test-price relationship when we varied the sample by distance to the school boundary?
Results: 1) Geographic Restriction & Variable Distance AnalysisRegression Results for Most Restrictive Geography (Set D),Variable Distance of 0.15 miles, for all time periods (1996-05)(Heteroskedastic-adjusted standard errors; t-statistics in parentheses)Dependent Variable = ln (house price)
House distance from attendance area
boundary
.15 miles
Test score .0022 (4.862)
Bedrooms .01307 (1.596)
Bathrooms .2439 (7.464)
Bathrooms2 -.0349(-5.183)
Lot size (sq ft) .000006 (4.976)
Internal size (sq ft) .0003 (15.066)
Boundary fixed effects Yes
N 1822
R2(adjusted) .7619
Interpretation:
A 12 percentage point increase in test scores* is associated with a $5,065 increase in average home price***One standard deviation
**In year 2000 dollars
Results: 2) Time Period Analysis (1996 - 2005)
Expanded on S. Black’s analysis by using a 10-year sample, which allowed us to ask:
How did test-price relationship change from pre-2000 to post-2000 period?
Pre-2000: Test data availability limited; reported annually in local newspaper
Post-2000: Test data more widely and instantly available on various websites
Results: 2) Time Period Analysis (1996 - 2005)
Regression Results for most restrictive geography (Set D), for houses located within 0.15 mile from boundary(Heteroskedastic-adjusted standard errors; t-statistics in parentheses)Dependent Variable = ln (house price)Time Period (1)
All Periods(1996-2005)
(2)First Period(1996-2000)
(3)Second Period(2001-2005)
Test score .0022 (4.862)
.0010(1.748)
.0035(3.844)
Bedrooms .01307 (1.596)
.0072(.565)
.0175(1.707)
Bathrooms .2439 (7.464)
.2657(6.143)
.1998(6.906)
Bathrooms2 -.0349(-5.183)
-.0375(-4.288)
-.0273(-4.427)
Lot size (sq ft) .000006 (4.976)
.000006(3.672)
.000006(3.487)
Internal size (sq ft) .0003 (15.066)
.0003(10.969)
.0003(10.997)
Boundary fixed effects Yes Yes Yes
N 1822 850 972
R2(adjusted) .7619 .7480 .7827
Interpretation:
A 12 percentage point increase in test scores* is associated with an average increase in home price**
Pre-2000: $2,244
Post-2000: $8,060*One standard deviation
**In year 2000 dollars
Results: 3) Neighborhood Progression AnalysisWhat if test scores are capturing not the effect of school quality on
home prices, but rather some quality change in neighborhood, from worse to better, which is captured in the prices of homes?
Results: 3) Neighborhood Progression AnalysisWhat if test scores are capturing not the effect of school quality on
home prices, but rather some quality change in neighborhood, from worse to better, which is captured in the prices of homes?
Test by comparing actual versus artificial school attendance areas:
Actual attendance area(0.15 mile from boundary)
(1) (2)
Artificial attendance area(0.15-0.45 miles from
boundary)(3)
Test Score .0022(4.862)
Higher-scoring side dummy variable
.0307(3.342)
Artificial attendance area dummy variable
- .0559(- 4.797)
House characteristics Yes Yes Yes
Boundary Fixed Effects Yes Yes Yes
N 1822 1821 1544
Adjusted R2 .7638 .7603 .737
Regression Results for most restrictive geography (Set D), for all time periods(Heteroskedastic-adjusted standard errors; t-statistics in parentheses)Dependent Variable = ln (house price)
Results: 3) Neighborhood Progression Analysis
Interpretation
Test scores do matter at elementary school attendance boundaries
Discussion:- In West Hartford, homebuyers grew more sensitive to test scores as data became more readily available over time
- See parallel qualitative study (Ramsay 2006), interviewed 89 homebuyers on social construction of school quality
- Further research on school racial composition, influence of middle & high school zones, and role of real estate agents
Discussion:- In West Hartford, homebuyers grew more sensitive to test scores as data became more readily available over time
- See parallel qualitative study (Ramsay 2006), interviewed 89 homebuyers on social construction of school quality
- Further research on school racial composition, influence of middle & high school zones, and role of real estate agents
This paper and others available at
www.trincoll.edu/depts/educ/CSS