Presentation on theme: "Policies for Mixed Communities: Still Looking for Evidence? Paul Cheshire 5 th Feb 2010 Neighbourhood Effects: Theory & Evidence University of St Andrews."— Presentation transcript:
Policies for Mixed Communities: Still Looking for Evidence? Paul Cheshire 5 th Feb 2010 Neighbourhood Effects: Theory & Evidence University of St Andrews
Policies for Mixed Communities Rather like IRSR Aug 2009 – sorry: about whether significance of any neighbourhood effects justifies policy What do you do when you are confronted with new facts? Here the reverse: new emerging evidence re-enforces the reality of capitalisation: nicer neighbourhoods really do cost more: and the poor can’t afford them More emerging evidence about extent and persistence of inequalities: a real problem deserving effective action Two rigorous new studies showing neighbourhood effects are minor or undectable And stronger evidence that ‘mixed neighbourhoods’ cost: and cost the poor
Policies for Mixed Communities ‘Mixed Communities’ an explicit aim of government policy in Britain – and in other OECD countries ; In USA an aspiration of New Urbanism: (Imbroscio, 2008 – ‘the Dispersal Community’) In fact aspiration far older: Howard & Garden City Movement: early examples e.g. Bedford Park, London, 1871; Hampstead Garden Suburb, 1910 May be suspiciously old? – retro-fitting the facts to justify the solution? Costs real - if opaque resources – visible expenditures but in England mainly via ‘Section 106 Agreements’
And Real problem… Inequalities deeply rooted and persistent (National Equality Panel 2010)
And Real problem…
But ‘Evidence’ still just Circumstantial Mixed Communities a ‘solution’ but not tested against the evidence E.g. ODPM( in poorest neighbourhoods (most deprived 10%) Life expectancy 2 years less than mean One third of inhabitants – no formal qualifications Crime higher…etc Not in dispute…But not evidence of causation Evidence to support policy requires: 1.Demonstration of direction of causation: concentrated poverty ‘worse’ than diffused – positive externalities for poor of living close to affluent 2.If causation – size of negative ‘neighbourhood effect’ 3.Evaluate potential foregone gains from ‘specialised’ neighbourhoods relative to any negative neighbourhood effects 4.Then – is forcing communities to be ‘mixed’ a cost effective means of addressing problem?
Nice Neighbourhoods Cost More…. Because overwhelming evidence of causation from income to neighbourhood choice (subject to income constraint) Nice neighbourhoods cost more & the nicest cost much more - hedonic studies on capitalisation Green spaces, river frontage, less noise, less crime, better golf: Consumption/Quality of life factors…. Better schools, better access to jobs: Production/life chance factors All capitalised into house prices
Nice Neighbourhoods Cost More…. Much More Wide range of such ‘goods’ only consumable given housing location: and – in quasi-fixed supply Hedonic analysis: real ‘scientific’ progress – theoretical understanding, data, statistical techniques, experience & computing power My position shifted housing markets pretty efficient, with complex and sophisticated search processes, reasonably modelled as in equilibrium Attributes – if appropriately defined – have uniform prices within given housing market Interaction structural attributes & value of expected school quality; interaction local densities, incomes, distance from edge of city & demographic structure for price of open space Brett Day – noise: Gibbons – crime; Troy & Grove – open space local crime rate; Hilber – social capital; & toxins
Nice Neighbourhoods Cost More…. Much More And estimated price functions highly non-linear ‘quantity’ Logic of hedonics is attributes have separate supply & demand characteristics Not identify formally? but can usefully think about them Supply of some – fixed e.g. frontage on Thames, Hampstead Heath, St Andrews golf…catchment area of best state school in community Of others highly elastic – e.g. produced by industrial process But UK planning renders supply of land almost fixed So: ability to buy attributes in inelastic supply not mainly function of income level but of income relative to others in housing market competing to buy: truly ‘positional’ goods! Implies - incomes rise - attribute prices rise differentially – income elasticity of demand + supply elasticity: & if Income distribution more unequal - most desirable neighbourhoods become relatively more expensive incidence of segregation more intense In more unequal societies more neighbourhood segregation: Sweden - UK
Nice Neighbourhoods Cost More…. Much More Much more…e.g. Primary school quality 10 th to 90 th decile => +10.4% 90 th to Max observed => +16.9% Similar patterns with access to CBD & space attributes (& access to Thames) And high income neighbourhoods have facilities for the affluent ….a short drive to an upmarket deli, gastropub…. Poor neighbourhoods have facilities for demand of low income households ….walk to discount store, cheap take-out So we really do know poor people live in poor neighbourhoods because they can’t afford rich ones
Quantifying ‘Neighbourhood Effects’ The (partial) sorting of rich & poor into separate neighbourhoods – almost inevitable: and not necessarily a bad thing. Spatial articulation of societal income inequality. Question: does living in a poor neighbourhood make the poor poorer - independently of factors making them poor in first place? Damage life chances? Methodologically difficult problem – unobserved characteristics; self-selection of neighbourhoods Two main approaches 1.Observe impact on moving individuals from deprived to affluent neighbourhoods [ or now vice versa – Weinhardt 2009] 2.Track individuals over time Best – or still best known- example of 1. Moving to Opportunity Program (MTO) set up 1992
MTO Programme/Experiment Quasi-experimental: offered chance to move from very poor neighbourhood(= Census Tract 40%+ below poverty line) to affluent one (<10% below poverty line) 5 cities: participants randomly allocated to 3 groups Group 1 – financial & professional help to move to affluent neighbourhood Group 2 – vouchers to get new housing of their choice Group 3 – no help though can move if able Self-selection – only 25% of eligible volunteered 13% of volunteers rejected as unsuitable (would not pass 1 st base for testing new drug…)
MTO Results: Long Term Follow-up But Kling et al, 2005; 2007 4-7 years: focus on adolescents Results complex & quite negative No economic gains for adults in Gp 1 Adolescents Gp1 : Gp 2 – small non-significant behavioural improvements overall Girls showed non-significant improvements Boys showed significant deterioration especially - property crime, behaviour in school & relationships Kling et al, 2007 – Confirmed no economic gains for adults: differential impacts girls – boys: some health improvements for adults (but may be other ways of achieving them…) And out-movers replaced by in-movers into poorest neighbourhoods: so net effect?
Moving the other way? into poor neighbourhoods Weinhardt 2009 – ingenious, opportunist method Given difficulty of getting into social housing – move is exogenous in timing & non-self selecting re area Identify most deprived neighbourhoods as those with 80% or more in social housing: highly correlated with deprivation English kids tested at 10/11 & 13/14 – KS2 & KS3 Compare school performance at KS3 of kids moving in between 10 and 14 with those moving in after K3 As usual ‘apparent’ neighbourhood effects – kids moving to ‘worst’ neighbourhoods do worse at KS3 But - control for KS2 & range of other factors – All sign of neighbourhood effects disappears academic performance does not suffer from moving into deprived area
Cohort studies Oreopoulos (2003) Canada, 30-year tracking – origin in range of social housing neighbourhoods Neighbourhood of origin had NO significant effect on labour market success or earnings Bolster et al (2007) Britain, 10-year tracking Neighbourhood of origin had NO significant effect on labour market success or earnings (perverse sign) van Ham & Manley (2009) 10 year tracking & labour market outcomes – test for tenure mix effects/social housing: for social housing concentrations – NO effects: weak effects for owner occupiers – but self-selection e.g. house prices? Evidence is neighbourhood effects are at most very weak + not straightforward + positive as well as negative
Benefits of ‘Specialised’ Neighbourhoods? People do choose the neighbourhoods in which they live Observe persistence of sorting: bigger the city, more specialised its neighbourhoods: since ancient Rome Early ‘designed’ mixed communities quickly re-sorted 1)Direct welfare: consumption benefits e.g. ethnic neighbourhoods, liberal professionals; young singles; young child raisers. Mutual support + demand generates appropriate local goods & services 2)Productivity gains: access to labour market Informal information & job matching: Most effective search method – esp. for non-local language groups More important the larger the city (Ioannides and Loury, evidence of positive agglomeration economies & neighbourhood sorting) Bayer et al (2009): good evidence neighbours play important role in job matching - more important for less skilled 3)And evidence (Luttmer, 2005) income relative to neighbours source of welfare so more homogeneous neighbourhoods?
Dynamics of Neighbourhood Segregation Neighbourhoods selected subject to constraints – income But talk of ‘local community’ misleads: not a stable set of families. More like occupants of a bus… Segregation by incomes, demographics, politics, ethnicity, religion etc – But significant local income mixing As houses become vacant, new occupants Cities – subject to shocks (growth, change in income distribution…) constant change
Effects of Neighbourhood Improvement Harlesden City Challenge – W. London 10, 000 population: 5 year programme: £37.5m Training (useful); crime reduction; job creation; physical upgrading…. At end of period unemployment in ‘local community’ risen & higher relative to other comparable areas Sample survey of ‘Stayers’, ‘Inmovers’ & ‘Outmovers’ Participated in training? Stayers 13%; Inmovers 6%; Outmovers 37% Outmovers had more skilled, enjoyable and better paid jobs…& 6 times more likely to have full time job Get on get out! And don’t judge a regeneration programme on basis current residents
Conclusions Residential segregation very persistent over time: & self-selection given income – people choose & evidence suggests - carefully Causation from income to neighbourhood sure – nice neighbourhoods cost more & the nicest cost a lot more Plausible that increased income inequality more intensive residential segregation via house price dispersion ‘Mixed communities’ policy treats a symptom and not a cause: treating fever with leeches instead of looking for causes of poverty What are the benefits? Evidence so far shows ‘communities’ have little/no impact on Production/life chance outcomes but may help job matching Benefits may come through redistribution of Consumption/Quality of life opportunities (if policy goal?), but inadequate evidence & generally likely inefficient mechanism compared to resource transfers No obvious evidence that poor derive positive externalities from living together with nice, educated and affluent: if not - just well-meant paternalism?
Conclusions What are the costs of mixed community policies ? Direct welfare costs through de-specialisation for both consumption & production Higher housing supply costs in high land price communities: Fewer units/lower quality greater long run inequality Diversion of resources away from direct interventions (e.g. schools, education, training, families, policing) or redistribution Evidence costs of policy substantially outweigh any benefits And a displacement activity But concentrated poverty ugly? Mixing makes well-intentioned affluent feel better: + costs disguised But poverty & social immobility real - need to act on causes: Including having area based policies e.g. schools or crime