Proximal Parks DETERMINING THE EFFECTS OF PARKS ON

  • Slides: 33
Download presentation
Proximal Parks DETERMINING THE EFFECTS OF PARKS ON NEARBY RESIDENTIAL PROPERTY VALUES BY ERIC

Proximal Parks DETERMINING THE EFFECTS OF PARKS ON NEARBY RESIDENTIAL PROPERTY VALUES BY ERIC SVENSON GTECH 732

Sunset Park, Brooklyn http: //www. flickriver. com/photos/tomvu/28183 67744/

Sunset Park, Brooklyn http: //www. flickriver. com/photos/tomvu/28183 67744/

Sunset Park, Brooklyn Source: Google Maps

Sunset Park, Brooklyn Source: Google Maps

Parks Questions §How much are parks used per square meter? §How equally does the

Parks Questions §How much are parks used per square meter? §How equally does the Parks Department allocate funds? §Are parks patrolled by police equally? §How can we predict visitation?

Parks Questions No visitation data kept (except Central Park) Parks website search-based – no

Parks Questions No visitation data kept (except Central Park) Parks website search-based – no central or downloadable documents No police records kept with relation to parks

What do parks offer? Recreation areas Community Events Aesthetic benefits Environmental benefits Source: nycgovparks.

What do parks offer? Recreation areas Community Events Aesthetic benefits Environmental benefits Source: nycgovparks. org

Problem §Parks are a desirable amenity §Cities looking to expand green space, increase per

Problem §Parks are a desirable amenity §Cities looking to expand green space, increase per capita statistic §What effects could new, high-quality parks have on housing values? §What could be positive/negative effects of an increase in housing value?

Objectives §Determine whethere is a meaningful correlation between a residential lot’s value per unit

Objectives §Determine whethere is a meaningful correlation between a residential lot’s value per unit and its distance to the nearest quality park §Expected: a decrease in value with increasing distance from park §Create, use and evaluate two related proximity models to better understand the relationship §Produce plots and maps to gain a holistic view of the problem

Conditions §Several studies showing park presence to have an amplification effect on housing value

Conditions §Several studies showing park presence to have an amplification effect on housing value (Wen 2015, Lorenzo 2007, Donovan 2011) §Areas of study vary §Housing within 4, 000 feet of parks compared (More 1982) §Housing values compared between three buffer distances 300, 600, 1000 feet (Lorenzo 2007) §Housing type, green space type varies §Community gardens, street trees, greenbelts

Conditions (cont. ) §Many studies use hedonic pricing model §Sums all variables surrounding housing

Conditions (cont. ) §Many studies use hedonic pricing model §Sums all variables surrounding housing that people pay for (neighborhood, specific housing characteristics) §Costly – purchase data from housing sales §All use some form of proximity analysis, often with an index for park quality to eliminate noise from results

Conditions (cont. ) § 1500 feet – limit of study area § Park density

Conditions (cont. ) § 1500 feet – limit of study area § Park density §Buffers of 750 and 1500 feet § 750 feet – length of average long NYC block §Results: § % Change statistics with increasing distance from park based on mean property values per unit within buffers (Buffer-ring model) § Scatter plots based on distance of each lot to the nearest park, Rsquared statistics for regression lines to show correlation (+/-)

Assumptions §High quality parks have a stronger influence on increasing residential property values §Value

Assumptions §High quality parks have a stronger influence on increasing residential property values §Value per unit is directly related to how much tenants or owners pay for housing §Estimated total assessed value (Department of Finance) is accurate §At a certain distance, the influence of the park on property values is null. §Parks below 5 acres in size will have very little influence on nearby property values

Data § NYC Open Data – Parks Properties shapefile §“Flagship parks”, “neighborhood parks”, “community

Data § NYC Open Data – Parks Properties shapefile §“Flagship parks”, “neighborhood parks”, “community parks” § Bytes of the Big Apple – Map. PLUTO shapefile §Data on tax lot level: §Residential Lots (Land use codes 1 -4) §Total Assessed Value – land value * tax class §Number of units §Value per unit = Total Assessed Value / Number of units

Data (cont. ) § NYC Parks. gov – park inspection program data §Rated “acceptable”

Data (cont. ) § NYC Parks. gov – park inspection program data §Rated “acceptable” or “unacceptable” based on a park’s ability to satisfy minimum standards for cleanliness, structure, and landscaping § Google Trends – park search data §Based on “search interest topics”

Park Quality Index Inspection Data (includes Cleanliness, Landscape Features and Structural Features): Percent of

Park Quality Index Inspection Data (includes Cleanliness, Landscape Features and Structural Features): Percent of inspections rated “Acceptable” versus “Unacceptable” Google Trends Data: ([Park Name] + nyc) entered into search bar % Acceptable 0 -49 Score 0 Methods 50 -69 1 70 -84 2 85 -100 3 Score 0 No “search interest topic” on Google Trends Search interest topic exists, little or no 1 information Search interest topic exists, either 2 searched for consistently from 20042015 OR > 1 country searching for topic Search interest topic exists, searched for consistently from 2004 -2015 AND > 1 country searching for topic 3

Park Quality Index Score High (2. 5 -3) Low Medium-High Medium-Low (Low 0 -0.

Park Quality Index Score High (2. 5 -3) Low Medium-High Medium-Low (Low 0 -0. 5) (1. 5 -2) (1) Number of Parks 8 42 37 31 Percent of Total Parks 6. 8 35. 6 31. 4 26. 2 Notable Parks Battery Park, Central Park, Riverside Park, Pelham Bay Park, Flushing Meadows Van Cortlandt -Corona Park, Prospect Park Bronx Park, Queensbridge Highland Park, Randall’s Morningside Park, Island Park Inwood Hill Park

Methodology

Methodology

Methodology - 750 and 1500 foot buffers - Erase features – park and 750

Methodology - 750 and 1500 foot buffers - Erase features – park and 750 foot buffer respectively, create “rings” - Identity – buffers take attributes of residential lots - Summary Statistics – calculate mean property value per unit within “inner ring” and “outer ring” - Join fields with buffer, Append to empty shapefile - Join field from shapefile to parks dataset, calculate percent change from inner to outer ring

Buffer-Ring results §Results vary widely between parks – 50% split between negative and positive

Buffer-Ring results §Results vary widely between parks – 50% split between negative and positive % change §Park quality has little bearing on the relationship §Average number of lots within each buffer ring is 1245. 2 for parks with negative % change, 667. 5 for parks with positive % change §Parks with high levels of change (+/-) looked at more closely with Near Analysis in next step

Percent Change by Quality

Percent Change by Quality

Methodology (cont. ) §Select parks with highest change statistic from buffer-ring model §Select lots

Methodology (cont. ) §Select parks with highest change statistic from buffer-ring model §Select lots within 1500 feet (Select Layer by Location) §Near analysis for each lot to park dataset §Scatter plots (SPSS)

Central Park Medium-High Quality -35% Change with distance R-squared: . 002

Central Park Medium-High Quality -35% Change with distance R-squared: . 002

Riverside Park High Quality -34% Change with Distance R-squared: . 0006

Riverside Park High Quality -34% Change with Distance R-squared: . 0006

City Hall Park High Quality -34% Change with Distance R-squared: . 046

City Hall Park High Quality -34% Change with Distance R-squared: . 046

St. Mary’s Park Low Quality +118% Change with Distance R-squared: . 002

St. Mary’s Park Low Quality +118% Change with Distance R-squared: . 002

Fort Greene Park Medium-High Quality +79% Change with Distance R-squared: . 0004

Fort Greene Park Medium-High Quality +79% Change with Distance R-squared: . 0004

Bryant Park High Quality +67% Change with Distance R-squared: . 001

Bryant Park High Quality +67% Change with Distance R-squared: . 001

Issues §Park density makes the use of buffers iffy §R-squared extremely low (1 =

Issues §Park density makes the use of buffers iffy §R-squared extremely low (1 = plot 100% determined by model variables) §R-squared statistic will decrease with more data points

Future Improvements §Minimum threshold for number of lots within 1500 feet §Buffer-Ring Model: §

Future Improvements §Minimum threshold for number of lots within 1500 feet §Buffer-Ring Model: § Select fewer parks to avoid buffer collisions § Three thinner buffers to capture closest edge to park §Incorporate hedonic pricing model to produce results more related to target variables §Buffer-ring good first step – necessary? §More conclusive results needed to understand possible effects of new parks on housing prices

Sources

Sources