BSPS Seminar: Household Formation Economic and housing market influences on household formation: a review. Prof Glen Bramley (Heriot-Watt University, Edinburgh, UK Contact: [email protected]; +44 (0)131
Download ReportTranscript BSPS Seminar: Household Formation Economic and housing market influences on household formation: a review. Prof Glen Bramley (Heriot-Watt University, Edinburgh, UK Contact: [email protected]; +44 (0)131
BSPS Seminar: Household Formation Economic and housing market influences on household formation: a review. Prof Glen Bramley (Heriot-Watt University, Edinburgh, UK Contact: [email protected]; +44 (0)131 451 4605) 16 December 2013 School of the Built Environment Background • Migration & household formation are central to sub-national demographic forecasts and important for physical & service planning and especially for housing • Traditional approach reliant on extrapolative projections remains popular • There has been an economic critique of this, arguing that labour and housing markets influence these trends • Speculate about reasons for reluctance to incorporate these in projections – unfamiliarity with econometrics – predicting the predictors – ‘need’ vs demand – taint of uncertainty • Problems which can result – ‘circularity’ and underprovision - out of phase with cycles - persistent discrepancies households vs dwellings - lack of realism about adjustment mechanisms in market – inappropriate planning between related geographical areas School of the Built Environment Objectives • Review literature on economic influences on household formation • Highlight particular findings from this literature • Consider economic-based forecasts for (domestic) migration and household formation at sub-regional scale • Introduce one such modelling framework • Demonstrate application of this model in England with particular reference to relationships between supply and household growth • Comment on recent household numbers and projections in the light of this • Suggest some ways forward School of the Built Environment Earlier Review • Bramley, Munro & Lancaster (1997) reviewed economic influences on household formation for DOE • Drew on range of earlier studies, esp US work • Confirmed importance of demographic fundamentals (age, sex, mar/ptnr status) & demographic events • Main arena for econ infl is younger non-family adults, altho’ marriage/partnership & fertility may be affected also • Income elasticities ranged 0.05-0.40, but much higher for young nonfamily (0.3-1.8) • Relatively inelastic with housing costs (-0.01 to -0.28); income & price offsetting • Some evidence that social housing (rationed) has direct supply effect • Higher educn; skills; ethnicity & culture; benefits? School of the Built Environment DAE/DTLR Model & other work c.2000 • Peterson et al (DETR, 1999) used aggregate GHS time series data to model household formation as part of wider economic ‘need’ model, updated in DTLR (2002) • Found income effect of 0.33 (higher due to use of consumption), also –ve influence of unemployment • Sensitivities quoted in 1999 DETR Household Projections • Other studies involving micro-modelling of household transitions included Ermisch, J. (1999) Prices, Parents and Young People's Household Formation. Journal of Urban Economics 45, 47-71 Clark, W.A.V. and Mulder, C.H. (2000) Leaving Home and Entering the Housing Market. Environment and Planning A 32, 1657-1671. School of the Built Environment DCLG ‘Affordability’ Model • This model drew on work by Andrew & Meen (2003), and included household formation function in ‘Affordability Model’ (ODPM 2005, Meen et al 2007, Meen 2011) • Micro-simulation based on probit model fitted to BHPS data (part of wider tenure choice model) • Also found demographic variables most important • Incomes, unemployment & housing cost played modest role - however, housing cost only tested at regional level - model only applied to under-35s • Similar approach subsequently adopted in Leishman et al (2008) Scottish Affordability Model • Meen & Nygaard (2008) & Nygaard 2011 looked at effects of different international migrant flows School of the Built Environment Further studies and EHN • Bramley, Champion & Fisher (2006a) explore household transitions and relationship of migration and mobility with household formation, finding effects (often indirect, via mobility) of range of economic variables • Bramley et al (2006b) modelled LA level aggregate headship x age in Scotland, finding effects of rental tenure, income, class, house prices • Bramley et al (2010) Estimating Housing Need study for DCLG modelled new household transitions (for under/over 40s) using logit in BHPS micro data with housing & labour market variables attached at SAR district level • Found effects from recent migrancy, tenure, qualifications, working, area unemployment, house price, income, & social lettings • Incorporated in regional simulations of housing need outcomes (with linked inputs from CLG-Reading ‘Affordability’ model) - although additional direct feedback from vacancy rates was needed School of the Built Environment EHN Supply Scenarios Table 7.1: Impacts of four supply scenarios relative to baseline, 2009-2021 Impact Summary Extra Extra Social Net Addn Extra Private Net Addn Household Growth New Household Formation stock-hhd reconcil adj Change Own Occ Hhlds Change Soc Rent Hhlds Change Priv Rent Hhlds New Social Lettings Hhlds 'Rationed Out' of SR Total Need backlog Private Vacancies Social Vacancies Hi Social 268,845 -3,005 235,285 208,963 93,188 130,292 246,491 -146,829 282,479 -25,138 -167,902 8,770 13,258 Hi Private Med Both Hi Both 9,021 200,003 267,201 435,243 292,695 430,697 347,247 406,299 567,487 19,175 162,753 231,404 309,844 270,525 364,352 113,216 168,231 232,168 13,734 187,448 251,721 201,766 44,215 78,227 17,358 216,960 287,731 2,699 -10,827 -19,684 -90,787 -184,554 -252,150 46,701 36,015 48,693 -5,026 5,689 6,628 School of the Built Environment Other Recent Literature • Several studies claiming clear evidence of cyclical recession effects (from incomes and labour market) on household formation (Lee & Painter 2013, Dyrda et al (2012), Paciorek (2013) • Some of these also point to effect of housing costs (Paciorek 2013) or sub-prime crisis • Studies focused on longer term decline of owner occupation, suggesting real situation compounded by declining young headship (Rosenbaum 2013) • Studies comparing ownership rates x ethnic group misleading for same reason (Yu & Haan 2011, Nygaard 2011, Yu & Myers 2010) School of the Built Environment ‘Gloucestershire’ Model • Model arose out of feasibility study into sub-regional housing market models undertaken for former NHPAU 2009-10 • Operationalised in study for Gloucestershire County & Districts in 2011, used to inform SHMA • A medium-term model geared for policy simulations with particular focus on new build, household growth, affordability, housing needs • Geographical framework of 102 Housing Market Areas (HMAs) based on LA Districts developed in parallel NHPAU research (Jones, Coombes et al) • Econometric functions for key variables based mainly on aggregate panel data, but some based on micro-models • Other exogenous or intervening variables projected in simpler mechanistic fashion • Simulation model implemented in Excel workbook • Similar model subsequently developed for New Zealand School of the Built Environment Main Behavioural Components of Model • • • • • • • • Real house prices (mix-adj) New private build completions (mix-adj) Migration gross flows x 4 age groups Household formation (headship) x 3 age groups [micro – BHPS] Household income (proxy-based prediction) & low income Social housing lettings Private rents Housing needs incidence [micro – EHS/S E H] School of the Built Environment School of the Built Environment School of the Built Environment Examples of Migration Equations Variable Description Varname Adjacent out-migration Lagged in-migration International in-migration Population 000s Relative price index (1.0) Adjacent relative price Mortgage interest rate Household income £k pa Unemployment age 25-44 Younger adults/25-55 Social rented tenure Private build output rate Adjacent private build rate High occupational class Single adult non-eld hhd White British persons Net residential density pph Sparsity ha/persons Students IMD low income score Adjacent IMD low income Distance city centre km Greenspace /land area Air quality index Climate index (warm/dry/etc) Scenic areas access Cars per m of road Constant pgomal_s pgin_1 pintmin npopk prrlprc3 rlpric_s mint hhinck asunem pyngla psrla prppcmp3 ppcmp_s hiclas hh1 pwhiteb netdens2 spars01 pstud01 imdlwinc imd_s dist150k pgreenh air climate scenic carspm _cons Children Young Adults In Out In Out pgin014 pgot014 pgin1524 pgot1524 0.621 0.326 0.360 0.728 -0.107 -0.183 0.226 0.133 0.000 0.000 -0.929 -0.365 0.724 -1.961 -0.086 -0.036 -0.166 -0.110 -0.049 -0.063 0.058 0.088 0.049 0.157 0.063 -0.262 -0.062 -0.052 0.332 0.369 -0.569 0.240 -0.861 0.314 -0.039 0.674 -0.126 -0.298 0.008 -0.058 -0.077 0.630 0.015 10.418 5.568 0.358 0.213 0.053 1.159 1.450 -0.176 -0.862 0.471 0.972 -0.116 -0.477 0.564 -0.037 -0.293 -0.327 -13.057 -0.446 7.193 School of the Built Environment Influences on Migration • Structural effects – in-migration -> out-migration; size of area/popn; adjacent out-migrn -> in-migrn • Geographical effects - sparsity & counter-urbanisation • Demographic effects – singles vs couples; younger (like attracts like); ethnic effects • Socio-economic effects - employment -> mobility and moving towards opportunity by younger groups; students • Income –ve? but poverty more -ve • Tenure - social renting -> less in-migrn • Housing market – relative house price -> -ve for in-migrn • Housing supply – strong +ve effects on in- & net migrn; but –ve diversion effect of adjacent supply • Environmental effects, esp climate +ve School of the Built Environment Household Formation (HRRs) - Elasticities Variable Description Male gender Get un-married High occupational class Sick or disabled Student Previously private renting Previously in lone parent hhd Previously in couple family Previously in multi-adult hhd Married Previously social renting Ethnic minority Social Lettings Aged 25-29/aged 25-55 Aged 30-34/aged 25-55 Migrant (between localities) Has own children Acquired child Unemployed (indiv) Lwr quartile house price (£k) Indiv Income £k Varname omale getunmar hiseg dsickdis dstud prevpr olpar ocfam omult omar prevsoc oethnic pslets oage2529 oage3034 migrant onchild getchild dunem lplqk dincindrk HRR age 15-24 HRR age 25-59 HRR age 60+ hr1524 hr2559 hr60ov 0.220 0.512 0.354 0.020 0.017 0.006 0.003 0.271 0.169 0.019 0.016 -0.151 0.059 -0.019 -0.433 -0.165 -0.008 -0.356 -0.105 -0.019 -0.015 -0.120 -0.789 0.065 0.026 0.048 -0.014 0.000 0.144 0.044 -0.055 -0.046 -0.030 0.101 0.134 0.085 0.009 0.014 -0.005 -0.002 -0.174 -0.046 0.239 0.221 0.170 School of the Built Environment Influences on headship • Range of expected age & household type background effects; also migrant (+0.10), student (+0.27), ethnic (-0.014) for younger group • Income elasticities 0.24 / 0.22 /0.17; also high SEG. • House price -0.174 / / -0.046 [in retrospect, should have also modelled age 25-34 separately] • Unemployment marginal -ve • Tenure (previous): priv rent +0.17 / 0.02 / 0.02 : soc rent +0.07 / 0.03 /0.05 • Social lettings supply +0.14 /+0.04 / -0.06 • Vacancy rates – no consistent/significant effects (but necessary to impose some feedback in simulation) School of the Built Environment Household Growth Rates England 19912001-11 2001 PrevProj Growth% 7.1% 9.9% Number pa 135,688 202,329 2011-21 PrevProj Growth% 10.3% Number pa 232,960 2001-11 NewProj 7.7% 157,965 2011-21 NewProj 10.0% 220,528 2001-11 GAM 7.3% 149,232 2011-21 GAM 7.9% 173,230 Previous 2008-based projections envisaged higher growth, due to higher int migrn; ‘Reality’ of shortage of supply has led to much lower growth up to 2011. Gloucs Model tracked actuality reasonably. Looking forward, new interim projections envisage resumption of similar growth, but GAM predicts a lower likely outturn, due to recession and very low new build output in early years. School of the Built Environment Regional Household Growth Household Growth Comparisons between Models by Region 2011-21 16.0% 14.0% 12.0% Percent 10.0% PrevProj NewProj 8.0% GAM 6.0% 4.0% 2.0% 0.0% NE YH NW EM WM SW EE SE GL Region New projection and GAM show lower growth for Y&H, SW and EE than 2008 projn New projection, but not GAM, show somewhat lower growth for NE, NW, EM New projection shows similar for SE (& WM) but GAM shows signif lower New projection shows much higher growth for London, but GAM shows much lower! London figures strain credibility School of the Built Environment Young Adult Headship Headship Rates for 20-29 Year Olds, Selected English Regions 1992-2012 .4000 .3500 Headship .3000 NE2029 .2500 EM2029 .2000 SE2029 .1500 GL2029 .1000 .0500 .0000 1990 1995 2000 2005 2010 2015 Year Source: Fitzpatrick et al (2013) Homelessness Monitor 2013 CRISIS, based on Labour Force Survey Comment: London & SE rates have fallen significantly since early 1990s; EM & NE fell a bit later; all regions blipped up in 2010 but dropped back in 2012 School of the Built Environment Tenure and New Household Flows Estimated number of new households forming, by tenure of first destination 2002-2010 (000s) Tenure 2007 2008 2009 2010 2011 131 76 183 72 44 229 40 48 208 55 71 268 75 48 259 Total 400 390 345 296 394 Source: Survey of English Housing and English Housing Survey Reports. Note: years refer to financial years 2007/08 etc. 381 Own Social rent Private rent 2002-06 avg 118 92 190 Overall new household formation slumped in 2008-09, recovered in 2010 Biggest drop in owner occupation – only partially recovered by 2011 Generally lower level of access to social renting as well Situation ‘saved’ by rise of private rented lettings (BTL) School of the Built Environment Identity Relationship • There is an identity relationship between households and dwellings (sometimes called the ‘Holmans identity’) • In change form, this states that ΔHH ≡ ΔDWG - ΔVAC - ΔSEC + ΔXSHR [the change in households is identically equal to the change in dwellings (‘net additions’) minus the change in vacancies minus the change in second homes plus the change in ‘excess sharing households’] • This helps to explain recent events in household numbers game • If the supply of dwellings is dramatically reduced, and vacancies cannot go much lower, and second homes don’t change very much, and sharing is pretty rare, then… household growth will inevitably fall, mainly through mechanism of new household formation, mainly affecting younger adults (age related dissolutions unaffected) • This shows that household growth will be strongly influenced by dwelling supply, particularly in a ‘tight’ situation - in a looser market you may see more change in vacancies and demolitions School of the Built Environment Model Simulations of Response Household Growth Response to New Build Impact On Gloucs WoE SEGAS SEGAS England Scenario Gloucs only low Glouc, WoE & Adj Low G, WoE, Adj & SEGAS low All South Low All England Low 2016 0.21 0.29 0.27 0.26 0.15 2021 0.97 0.96 0.91 0.87 0.78 2031 0.77 0.87 0.80 0.81 0.87 Gloucs WoE SEGAS SEGAS England Gloucs only High Gl, WoE, Adj High Gl, WoE, Adj & SEGAS High All South High All England High 0.21 0.29 0.28 0.26 0.14 1.08 1.00 0.95 0.90 0.83 0.70 0.87 0.80 0.82 0.85 London London V High 0.38 0.95 1.00 England More SR (AH) 0.77 0.69 0.77 School of the Built Environment Implications of Simulation • As expected (on basis of past research, theory, and ‘identity’), household formation responds to new build supply • Response is lagged, takes time to build up; still quite low at yr 5 (20-30%) • After 8-10 years, response level is high, 80-100% • After 18-20 yrs, response level fades somewhat (70-85%) • Local variation in response rates, also depending on contextual/adjacent supply changes – London particularly high • Local responses strongly affected by migration • Building more social/affordable housing would have earlier positive impact on household growth, but more moderate later peak • (Note that examples are mainly pressured South; England level responses slower initially) School of the Built Environment Household Formation vs Migration Share of Household Growth Attrib to Household Formation Scenario Gloucs only low Glouc, WoE & Adj Low G, WoE, Adj & SEGAS low All South Low All England Low 2016 0.42 1.24 0.70 0.91 0.98 2021 0.73 0.86 0.70 0.74 0.99 2031 0.18 0.19 0.05 0.17 0.98 Gloucs only High Gl, WoE, Adj High Gl, WoE, Adj & SEGAS High All South High All England High 0.44 1.27 0.73 0.95 0.98 0.75 0.84 0.70 0.73 0.99 0.09 0.24 0.04 0.16 0.98 London V High -0.19 0.39 -0.10 More SR (AH) 0.99 0.98 0.99 School of the Built Environment Comments on Household Formation Share • At national level, virtually all of difference between scenarios in household growth is attributable to net household formation, and none to migration (given fixed international migration & balanced internal migrn) • At local level, this share is quite variable, and it also varies over time e.g. Gloucs relatively low, West of England rel high • In long run, local household growth responses mainly dominated by migration • London responses strongly dominated by migration School of the Built Environment Migration & Growth Scenarios • Higher international immigration would...raise household growth, esp in London, worsen affordability (& need), but reduce household formation - hhd increase % only 0.26-0.34 of popn increase % (vs 0.45 in New Zealand) (ie. elasticity of hhd wrt popn) • Higher economic growth (+0.5% pa) would raise household growth by 7-13,000 pa (2-6%), ignoring any induced extra international migration - in 2031 household numbers would be 0.6% higher (vs GVA 10.5% higher) - note offsetting effects of higher prices (similar to NZ model) • Combination of these would raise hhd growth by 14-26,000 (8-11%) - 2031 hhd numbers 1.5% higher (vs. 2.2% more popn) - affordability would be 1.4% worse in 2016 but +0.3% by 2031 - household formation would be suppressed by 15-25,000 pa School of the Built Environment Other Scenarios • Ending credit rationing completely could greatly increase new build (45-65%), affordability (20-25%) and household growth (3%-46%) [but treatment of this factor in model is crude, with lack of pre-2007 experience to calibrate it] • More Buy to Let activity would worsen affordability to buy (-1 to -7%) but net effect on household growth/formation slight • Relaxing planning controls over size mix would lead to a moderate increase in housebuilding numbers (1-9,000 pa), associated with a general reduction in dwelling size, and a modest increase in household growth (3-4,000 pa), with stronger effect in London • Very high London supply (doubling plan numbers) would raise output a lot (10-20,000 pa) and would increase household growth (4-20,000 pa, 26-72%), almost entirely thru’ migration (i.e. little extra net household formation); affordability would be a bit better (1.3-2.2%) - but even this would not match the 2013 Household Projections!! School of the Built Environment So What (is to be done)? • • • • Demographers, planners & economists need to talk We need to talk to Boris about his figures Traditional household projections are necessary but not sufficient Recent turbulence has exposed weaknesses in process, exacerbated by austerity cuts in analytical capacity in government • Planning policy guidance (2013) rightly emphasizes a range of measures of (in) adequacy of housing numbers to be presented through SHMA, including household projections, affordability, price trends, housing needs (incl concealed hhds) and employment growth • ‘Planning’ in full sense requires longer forward look and comparisons of options with outcome performance measures • Such a forward look will be more meaningful if it is based on models which take account of economic feedback effects School of the Built Environment References Andrew, M. and Meen, G. (2003). “Housing Transactions and the Changing Decisions of Young Households in Britain: The Microeconomic Evidence”, Real Estate Economics, 31(1): 117-138 Andrew, M., Bramley, G., Leishman, C., Watkins, D. & White, M. (2010) NHPAU Sub-Regional Market Modelling Feasibility: Main Report on Model Testing and Feasibility. NHPAU/DCLG. Bramley, G. (2013) ‘Housing market models and planning’, Town Planning Review, 84:1. Bramley, G. & Watkins, C. (1995) Circular Projections: Household Growth, Housing Need and the Household Projections. London: Council for the Protection of Rural England. Bramley, G. & Watkins, C. (1996) Steering the Housing Market: new building and the changing planning system. Bristol: Policy Press Bramley, G. (2012) 'Housebuilding, demographic change and affordability as outcomes of local planning decisions; exploring interactions using a sub-regional model of housing markets in England', paper presented at European Network for Housing Research Conference, Lillehammer, Norway, June 2012. Bramley, G., Champion, T. & Fisher, T.(2006) ‘Exploring the household impacts of migration in Britain using panel survey data’, Regional Studies 40:8, 907-926. Bramley, G., Karley, N. K., & Watkins, D. (2006b) Local Housing Need and Affordability Model for Scotland – Update (2005-base). Report 72. Edinburgh: Communities Scotland. Bramley, G., Munro, M. & Lancaster, S. (1997) The Economic Determinants of Household Formation: A Literature Review. DETR, London. School of the Built Environment Dyrda, S., Kaplan, G., & Rios-Rull, J.-V. (2012) Business Cycles and Household Formation: the Micro vs the Macro Labour Elasticity. NBER Working Paper No. 17880. Ermisch, J. (1999) Prices, Parents and Young People's Household Formation. Journal of Urban Economics 45, 47-71. Glaeser, E., Gyourko, J. and Saks, R.E. (2006) Urban growth and housing supply, Journal of Economic Geography 6, 71-89. Glaser, K. and Grundy, E. (1998) Migration and Household Change in the population Aged 65 and Over, 1971-1991. International Journal of Population Geography 4, 323-339. Holmans, A. (2009) ‘Flows and Households Formed 27.1’ Technical paper. Cambridge Centre for Housing and Planning Research.; Jones, C., Coombes, M., & Wong, C. (2010) Geography of Housing Market Areas: Final Report. Research Report to DCLG. London: DCLG http://www.communities.gov.uk/publications/housing/geographyhousingmarket Lee, I.O., Painter, G. (2013) ‘What happens to household formation in a recession?’, Journal of Urban Economics, 76, 93-109. Leishman, C., Gibb, K., Meen, G., O’Sullivan, T., Young, G., Chen, Y., Orr, A. and Wright, R. (2008) Scottish model of housing supply and affordability: final report, Edinburgh: Scottish Government School of the Built Environment Meen, G. & Andrew, M. (2008) ‘Planning for housing in the post-Barker era: affordability, household formation and tenure choice’, Oxford Review of Economic Policy, 24:1, 79-98. Meen, G. & Nygaard, A. (2008) International Migration and the Demand for Housing. International Centre for Housing and Urban Economics, University of Reading. Meen, G. (2011) ‘A long run model of housing affordability’, Housing Studies, 26:7-8, 1081-1103. Nygaard, C. (2011) ‘International migration, housing demand and access to homeownership in the UK’, Urban Studies, 48:11, 2211-2229. Paciorek, A. D. (2013) The Long and the Short of Household Formation. Finance and Economics Discussion Series. Divisions of Research & Statistics and Monetary Affairs. Federal Reserve Board, Washington DC. Rogers, W.H., & Winkler, A.E. (2013) The Relationship between the Housing and Labour Market Crises and Doubling-Up: an MSA-level analysis 2005-2010. Discussion Paper 7263, Forschunginstitut zur Zukunft der Arbeit/Leibnitz Information Centre for Economics. Rosenbaum, E. (2013) ‘Cohort Trends in Housing and Household Formation since 1990’ in J. R. Logan (ed) The Lost Decade: social change in the US since 2000. Russell Sage Foundation. Yu, Z., & Haan, M. (2012) ‘Cohort progress toward household formation and homeownership; young immigrant cohorts in Los Angeles and Toronto compared’, Ethnic and Racial Studies, 35:7, 1311-1337. School of the Built Environment