Post on 16-May-2020
Infill and urban consolidation in Melbourne
Spillovers and adjustment effectsNational Housing Conference, Darwin 28-30 August
Dr Christian A. NygaardDr Stephen Glackin
Swinburne University of TechnologyCentre for Urban Transitions
Also thanks to Dr Liss Ralston for access to data.
Introduction
• Urban development challenge: • Containing the environmental footprint of cities: greenbelts and growth
boundaries.• Population growth: housing access, affordability and (transport) locations.
• This talk: focuses on urban redevelopment and densification in Melbourne.
• Have the nimbyist got it right? Equity issue: is there a negative externality?• Does densification appear to generate a supply effect? Thus reducing overall
price increases in well-located areas?
Overview1. Introduction2. Context/select
literature3. Data4. Method5. Results6. Conclusions
Context
• Infill/consolidation can generate supply and demand effectssimultaneously.
• Local property prices reflect demand for localities (live/work) and cost ofsupply.
• Infill/consolidation may relax supply constraints by providing additional locality (e.g.higher density).
• Effect depends on scale – difference between 1-for-1 replacement and higher densitydevelopments.
• Consolidation/redevelopment may raise price expectations – price push/gentrification(latent price-distance gradient).
Select literature• A literature on infill/residential redevelopment related to publicly subsidised housing.
• Tidying-up effect (increase demand) (Simon et al 1998, Ding et al 2000, Ellen et al. 2001); visualeffect esp from multi-dwelling buildings (Dye and McMillan 2006); partial price convergence(Schwartz et al 2006).
• Less on private sector infill/densification across the neighbourhood price spectrum.• Ooi and Le (2013) – Singapore: positive effect due to price exploration/future amenity value.• Zahirovich-Herbert and Gibler (2014) – Baton Rouge: increases prices of non-similar properties,
supply effect on similar properties.• Ahvenniemi et al (2018) – no positive or negative effect in 7 case study areas and 6000
transactions.• Redevelopment/knock-downs as indicator of obsolescence (Rosenthal and Helsley 1994,
Dye and McMillan 2006)• Redevelopment sites may thus be blighted sites.
• Mixed evidence on impact of urban containment:• Evidence of higher prices (Evans 2004, Quigley and Raphael 2005, Glaeser and Ward 2009, Kok et
al 2014), but also more neutral impacts where accompanied by land release and densification(Song and Knaap 2004, Nelson et al 2007, Buxton and Taylor 2011).
Geocode some 800k sales 2006-2015 (data issues)
Red= sold twice 2006-2015Green = sold once
Geocoded development data (ca 71.6k)Yellow= new development,
different typologies
Yield based typology Sales after dev.Knockdown & rebuild 131,128Lower density (2-4) 167,102Medium density (5-49) 37,687Higher density (50+) 6,496
100 meters
DiD approach: sales with development within 100-500m comparedto sales in same LGA, Zone & POA outside
Yellow= new development
500 meters
LGAZoningPostcode
Identification• Difference-in-difference hedonic identification of the impact of new residential investment in existing
neighbourhoods on nearby properties, Eq (1):
𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑖𝑖 = 𝛼𝛼 + 𝛽𝛽𝑖𝑖Χ′𝑖𝑖 + 𝛾𝛾𝛾𝛾 + 𝜗𝜗𝜗𝜗𝑖𝑖𝑖𝑖 + 𝜈𝜈𝛾𝛾𝜗𝜗𝜈𝜈1𝑖𝑖𝑖𝑖𝑖𝑖 + 𝛾𝛾𝜗𝜗𝐴𝐴𝑖𝑖𝑖𝑖𝑖𝑖𝑖,1 + 𝛾𝛾𝜗𝜗𝐴𝐴𝐶𝐶𝑖𝑖𝑖𝑖𝑖𝑖1,1𝑖 + 𝜎𝜎 + φ + ξ + 𝜏𝜏 + 𝜎𝜎 ∗ 𝜏𝜏 + 𝜀𝜀𝑖𝑖𝑖𝑖
• A= whether a property is sold after a development has started within 100-500m [0,1]• D=whether sold property is within 100-500 meter radius of any development site [0,1]• ADY1= whether property is sold after a development has started AND is within radius [0,1] – spillover Y1. • ADN=Whether property is sold after development starts Y2-10 AND is within radius – spillover over time.• ADNC= Whether property sold after development is completed Y2-10 AND is within radius.• σ, φ, ξ and τ are fixed effects for local authority (30), planning schemes (91), postcodes (254) and years,
respectively. ε is an error term. • X= additional factors that explaining property prices: property type/age, neighbourhood characteristics,
distance to amenities (CBD, trains, roads, trams, parks, hospitals, schools, flooding).
Results: baseline Y1 (n=460,236)
• Some evidence that properties close to denser areas systematically are cheaper thanother properties in same LGA/POA/Zone.
• However, little evidence that development reduces prices further in Y1.• Exemption high density development with a negative effect within 200m (ca 5%)
-1
0
1
2
100m 200m 300m 400m 500m
Near knockdown-rebuild
-4
-2
0100m 200m 300m 400m 500m
Near low density
-3
-2
-1
0100m 200m 300m 400m 500m
Near medium density
-10
-5
0
5
100m 200m 300m 400m 500m
Near high density
-1
-0.5
0
0.5
1
100m 200m 300m 400m 500m
After knockdown-rebuild, Y1
-2
-1
0
1
100m 200m 300m 400m 500m
After low density, Y1
-2
-1
0
1
100m 200m 300m 400m 500m
After medium density, Y1
-10
-5
0
5
100m 200m 300m 400m 500m
After high density, Y1
Results: baseline Y2-10
• Positive effect on property prices following knockdown & rebuilds and medium density development. Effect strongest in 100-200mradius.
• Low density development largely flat over time (small positive effect in 100m ring, not shown)• High density development has negative effect on properties very close by, but effect weakens with distance and turns positive.• Small additional positive effect within 100-200m following completion of developments.
0
0.5
1
1.5
2
2.5
Y2 Y3 Y4 Y5 Y6 Y7 Y8 Y9 Y10
After knockdown and rebuild Y2-Y10
100m 200m 300m 400m
0
1
2
3
4
5
Y2 Y3 Y4 Y5 Y6 Y7 Y8 Y9 Y10
After medium density Y2-Y10
100m 200m 300m 400m
-15
-10
-5
0
5
10
Y2 Y3 Y4 Y5 Y6 Y7 Y8 Y9 Y10
After high density Y2-10
100m 200m 300m 400m -0.05
0
0.05
0.1
0.15
0.2
100m 200m 300m 400m
Per annum change after completion of development
Results: houses only Y2-Y10
• Melbourne housing stock 67% houses.• Densification a separate submarket?• Continued positive effect of nearby developments. • Negative effect within 200m disappears for higher density development.
0
0.5
1
1.5
2
2.5
3
3.5
Y2 Y3 Y4 Y5 Y6 Y7 Y8 Y9 Y10
After knockdown & rebuild Y2-Y10
100m 200m 300m 400m
00.5
11.5
22.5
33.5
44.5
Y2 Y3 Y4 Y5 Y6 Y7 Y8 Y9 Y10
After medium density Y2-Y10
100m 200m 300m 400m
-5
0
5
10
15
20
Y2 Y3 Y4 Y5 Y6 Y7 Y8 Y9 Y10
After high density Y2-Y10
100m 200m 300m 400m -0.02
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
100m 200m 300m 400m
Per annum additional change after completion
Controls
• Prices fall:• Distance to CBD• Greater area density• Distance from train and tram stops• Flooding zone
• Price increase:• Distance from arterial roads• Higher SEIFA status• Heritage (11.5%)• Size of property lot
Conclusion• Price pushing effects/externality:
• Removal of obsolete/neglected properties.• Properties subject to development generate price
premium: 10% average increase.• Densification may be accompanied by additional
area amenities (parks, design, personal services)that generate positive neighbourhood effects.
• Income effect:• Densification raises local price expectations?• Income effect: total and/or average income
increases that sustains additional personalservices/gentrification.
• Multiple infills/ area or precinct developmentadditional positive effect: 3%.
• Cannot separate price-push or income effect,but closely connected.
• Little evidence of supply effect.• Even large developments typically only have very
contained impacts.• Impact seems to disappear when considering
houses only. Suggests densification is onlylimited substitute for detached housing.
• Considerable period of under-investment inhousing stock. Insufficient scale to generatesupply effect?
• Low-density development (2-4) are typicallyfound in lower price areas.
• Redevelopment of low-density generate littlewider effect. Low scale and no additional areabenefits?
The absence of a clear density-supply effect is positive for those who favour densification, but problematic if density is an argument for affordability in the housing system more widely.