ENHR Housing Economics Workshop

24
School of the Built Environment Housing Mobility and Tenure Choice with varying constraints and rationing: a model for English regions built from micro household transition data Prof Glen Bramley (Heriot-Watt University, Edinburgh, UK Contact: [email protected] ; +44 (0)131 451 4605) with Prof Michael White (Nottingham Trent University) July 2011 ENHR Housing Economics Workshop

description

ENHR Housing Economics Workshop. - PowerPoint PPT Presentation

Transcript of ENHR Housing Economics Workshop

Page 1: ENHR Housing Economics Workshop

School of the Built Environment

Housing Mobility and Tenure Choice with varying constraints and rationing: a model for

English regions built from micro household transition data Prof Glen Bramley

(Heriot-Watt University, Edinburgh, UKContact: [email protected]; +44 (0)131 451 4605)

with Prof Michael White (Nottingham Trent University)July 2011

ENHR Housing Economics Workshop

Page 2: ENHR Housing Economics Workshop

School of the Built Environment

Overview of Paper

• Paper is about housing tenure choice and outcomes in England

• Reviews literature• Describes general approach and data sources

- micro estimation on BHPS transitions over 7 pairs of waves- macro regional simulation model built on S.E.H., LFS, etc.

• Findings on drivers of mobility, moves to buy and social rent• Simulation model features• Baseline and alternative scenarios• Conclusions

Page 3: ENHR Housing Economics Workshop

School of the Built Environment

Previous Literature

• Fundamentals of ‘choice’ to buy or rent• Embedded within wider housing demand – also hhld

formation, quantity/quality of housing services• Affordability & income; permanent income; relative ‘user

costs’• Transaction costs & mobility (length of stay)• Econometric issues around identification • Credit constraints – savings & wealth/age of purchase/• Demographics – age, family type, ethnicity• Tax & inflation effects• Subsidies e.g. Housing Allowances/Benefits• Location – region, market area, labour market conditions -

migration interactions

Page 4: ENHR Housing Economics Workshop

School of the Built Environment

Our Approach

• While taking much from previous literature, we see some limitations, esp in British context

• Better to focus on flows – active decision making households - path dependence

• Hierarchical sequential approach (Form/move – Buy – Soc Rent)• Mobility models generate flows & also affect ‘expected length of

stay’• Household formation modelled on similar drivers• Social rented lettings clearly rationed• Market rents, prices, unemployment etc linked at subregional

HMA levelplus, in simulation stage,

• Recognition of physical limits to stock-> feedback Vacs->HHFm, PRS

• Recognition of (re-emergence of ) credit rationing post 2007

Page 5: ENHR Housing Economics Workshop

School of the Built Environment

Hierarchical Choice Model

Table 1. Hierarchical/sequential scheme for tenure flows modelling New Household Owner Social Renter Private Renter

Form? Move? Move? Move? Buy? Buy? Buy? Buy?

Social Rent? Social Rent? Social Rent? Social Rent?

Page 6: ENHR Housing Economics Workshop

School of the Built Environment

Mobility Rates

Mobility Rates by Household Age-Type and by Tenure, 2002-06

0.0%

5.0%

10.0%

15.0%

20.0%

25.0%

30.0%

35.0%

40.0%

Age

-Typ

e C

ateg

ory

% move in 1 year

Own

Social

Priv Rent

Page 7: ENHR Housing Economics Workshop

School of the Built Environment

Household Formation

• Reviewed theory and past research, and highlighted trends from data since early 1990s

• Base period ‘profile’ of new households from S.E.H. & BHPS• Modelled propensity to form new household using BHPS

linked to local/subregional market variables• Key demographic drivers include age (younger), migrancy,

marriage, childbirth• Key economic factors include prices (-0.27) incomes (0.31),

unemp (-0.24)• Social lettings supply (0.26)• Smaller effects from prev tenure, ethnicity, qualifs, hsg type• Model is consistent with previous research including Reading

model but adds some extra elements

Page 8: ENHR Housing Economics Workshop

School of the Built Environment

Mobility Models

• Logistic regression to predict 1-year moves by origin tenure• Young are more mobile, esp in rental tenures• Larger households less mobile, tho children & crowding may

trigger moves• Higher (current & perm) income increases mobility in private

market; wealth mixed• Unemployment mixed but mainly positive• Rates in private renting 4-5x other tenures

Page 9: ENHR Housing Economics Workshop

School of the Built Environment

Table 3: Effect of different variables on choice to buy by previous tenure (coefficient measuring effect of 1 unit change on log-odds of buying for moving households) Variable description

Short name New to buy Own to buy Social to buy Private rent to buy

Moved area migrant 0.256 -0.402 Migration direction (u-r) migdir2 0.411 Aged <20 oageu20 -1.215 -0.412 -1.396 -0.570 Aged 20-24 oage2024 -0.615 Aged >60 oov60 1.023 No. of children onchild 0.366 0.597 Household size ohhsize -0.145 -0.249 -0.152 -0.953 Single person household osing -0.316 -1.268 Lone parent olpar -0.470 -0.857 Married dmar -1.129 Get married getmar 2.307 0.702 Get un-married getunmar 0.380 Get child getchild 0.440 1.187 Higher education qualifications dhequal 0.211 Working HRP dwork 1.082 0.478 2.452 1.038

Page 10: ENHR Housing Economics Workshop

School of the Built Environment

Student dstud -7.146 Weeks unemployed dnjuwks -0.045 -0.012 -0.108 -0.028 Log Income (household £k) ldinchhrk 0.118 0.717 0.277 Wealth £k dwealthki2 0.075 0.012 0.019 0.020 Low Income (<£10k pa) lowinc -1.857 -0.746 -0.636 Number of cars 0.467 0.551 0.374 Area unemployment % asunem -0.013 House price £k lplqk -0.016 User Cost £k pa -0.263 -0.108 -0.252 Social lets % hhd pslets2 -0.363 -0.117 Private rent/ House price % prent2bm -0.192 0.070 Social Rent/ House price % lhpir 0.309 -1.063 Constant const -2.352 0.232 4.745 -0.152 Nagelkerke R Sq 0.657 0.278 0.456 0.445 % correct predictions 90.4 77.8 89.6 82.9

Note: bold type indicates coefficients significant at 10% level or better.

Page 11: ENHR Housing Economics Workshop

School of the Built Environment

Mobility and Tenure Choice

• Logistic regression fitted to pooled BHPS data on 1-year transitions

• Movers Buying related to mobility (-, via user cost), age (young -), hhd size (-), children (+), working (+), students (-), unemployment (-), income (+), wealth (+), house prices & int rates (-, via user cost), private rents (+ for PR), soc rents (mixed)

• Movers Social renting related to similar factors, but usually with reverse sign (except students; user cost omitted);migrants (-), young (-), sick/disabled (+); low income (+); crowding (+); supply of social lettings signif (+); priv rents (+ for PR); soc rents (?)

Page 12: ENHR Housing Economics Workshop

School of the Built Environment

Forward Forecast

• Model now simulates system forward year-by-year from 2009 to 2021

• Takes inputs from parallel run of Reading Affordability model• Model is recursive – endogenous variables calculated

sequentially with some use of lagged values• Forecasts of household formation and tenure flows and

stocks by household age-type and region• Stock-household reconciliation, affecting vacancies, new

household formation and concealed/sharing households• Model to predict private rents – simple reduced form

regression• Social lettings rationing constraint, now formularised• 9 categories of specific need incl homelessness forecast

each period, based on models for each need

Page 13: ENHR Housing Economics Workshop

School of the Built Environment

Figure 6.1: Schematic Picture of Overall Simulation Model

Page 14: ENHR Housing Economics Workshop

School of the Built Environment

Stock-Household Reconciliation

• Ex poste, system must satisfy identity relationship:Households=Stock-Vacancies-2nd homes+Sharing Households-Shared Dwellings

• ‘Natural’ vacancy rate (3.5%) in private sector; if a region goes below this, adjustments are made to new household formation by (younger) singles in PRS (equiv to half the difference)

• With corresponding increase in multi-adult households, and also in concealed households and sharing (specific needs)

• Argue that concealed effects bigger than sharing under current conditions

Page 15: ENHR Housing Economics Workshop

School of the Built Environment

Private Rents Equation

Table 6.1: Regression Model for Private Rents at Subregional Level 1997-2006 Explanatory variable Coeff B Std Coeff Beta t stat signif

Constant -32.074 -2.562 0.011 Opp cost cap value * int £pw 0.410 0.574 8.913 0.000 Cap value growth 3yr pa £pw -0.013 -0.144 -2.595 0.010 Household income £ pw 0.107 0.210 4.753 0.000 Flow demand % hhd 7.492 0.369 8.395 0.000 Vacancy rate % (priv) 9.085 0.181 4.739 0.000 Model summary

R R Sq Adj R Sq S E Est 0.871 0.758 0.752 22.464

ANOVA Sum of Sq Deg Frdm Mn Sq F Ratio Regression 323047 5 64609 128.029 Residual 102948 204 504.6 Signif Total 425995 209 0.000

In simulation model, vacancy relationship imposed, raising rents where vacancies below natural rate

Page 16: ENHR Housing Economics Workshop

School of the Built Environment

Baseline Scenario

• Modified version of Reading model baseline• Income growth curtailed in this period, leading to incomes

nearly 10% lower than trend by 2014• New private build reduced sharply in 2008-2010, (95,000

comps in 2010) recovering to 150,000 by 2012 & 185,000 by 2015

• New social output static at 17,000 pa. • Altho prices and HPIR fall sharply in 2008-9, we apply a

‘shadow price of credit rationing’ in model to represent ‘effective affordability’ (1.9 in 2009, tapering off to 1.10 by 2015).

• Populations are as per Sept 2009 Reading model, but household growth is endogenous

Page 17: ENHR Housing Economics Workshop

School of the Built Environment

Baseline ForecastTable 5: Simulated Shares of Home Ownership and Private Renting for English Regions 2004-2021

HomeOwnership %

Region 2004 2009 2014 2021 NE 65.88 61.80 63.21 62.99 YH 69.94 68.37 70.37 70.81 NW 71.75 69.28 70.05 70.29 EM 73.90 72.38 75.30 75.32 WM 70.20 69.17 71.42 71.99 SW 73.78 71.85 74.80 74.24 EE 72.83 71.15 74.42 74.15 SE 74.25 71.76 74.42 74.57 GL 56.84 54.45 59.86 62.03 ENG 69.93 67.69 70.46 70.87 Private Renting % Region 2004 2009 2014 2021 NE 8.48 11.88 10.75 11.77 YH 10.07 13.30 11.79 12.28 NW 8.74 12.01 11.73 12.36 EM 9.13 12.57 10.38 11.34 WM 8.56 11.23 9.51 9.72 SW 12.58 14.07 11.56 12.74 EE 10.71 13.05 10.26 10.95 SE 11.75 14.77 12.35 12.35 GL 17.27 21.41 16.25 14.44 ENG 10.81 14.28 11.95 12.18

Page 18: ENHR Housing Economics Workshop

School of the Built Environment

Components of Change in Owner Occupation and User Cost

Components of Change in Owner Occupation, England 2007-2021 (annual flow

numbers)

-200,000

-100,000

0

100,000

200,000

300,000

400,000

2006 2008 2010 2012 2014 2016 2018 2020 2022

Year

Num

ber

New OwnerMove InMove Out

DissolnNet Chg Own

Components of User Cost, England, 2007-2021

-6.00

-4.00

-2.00

0.00

2.00

4.00

6.00

8.00

10.00

12.00

2006 2008 2010 2012 2014 2016 2018 2020 2022

Year

£ 00

0 pa

@ 2

006

pric

es

Price x int ratePrice Growth

Trans cost (PR)User Cost (New)

Page 19: ENHR Housing Economics Workshop

School of the Built Environment

Comments on Baseline

• 2004-09 unusual for seeing decline in OO and large rise in PRS

• This will reverse, particularly up to 2016, but later growth of OO slows again as user cost rises

• Net growth quite sensitive to moderate changes in gross flows

• Some regional differences e.g. more persistent shift from OO to PR in poorer regions

Page 20: ENHR Housing Economics Workshop

School of the Built Environment

Differing Supply Policy ScenariosHome-Ownership Rate Under Different Supply Policy Scenarios

67.0

69.0

71.0

73.0

2000 2005 2010 2015 2020 2025

Year

Perc

ent o

f Hou

seho

lds

Baseline

High Supply (both)

More LCHO

More SR+LCHO

V Low Supply

Private Renting Rate Under Different Supply Policy Scenarios

9.0

11.0

13.0

15.0

2000 2005 2010 2015 2020 2025

Year

Perc

ent o

f Hou

seho

lds

Baseline

High Supply (both)

More LCHO

More SR+LCHO

V Low Supply

Page 21: ENHR Housing Economics Workshop

School of the Built Environment

Differing Financial Contextual Scenarios

Home-Ownership Rate Under Different Contextual Scenarios

66.0

68.0

70.0

72.0

74.0

2000 2005 2010 2015 2020 2025

Year

Perc

ent o

f Hou

seho

lds

Baseline

Less Credit Rat

Persistent Credit Rat

Higher Social Rents

More Inequality

V Low Supply & CR

Private Renting Rate Under Different Contextual Scenarios

7.0

9.0

11.0

13.0

15.0

2000 2005 2010 2015 2020 2025

Year

Perc

ent o

f Hou

seho

lds Baseline

Less Credit Rat

Persistent Credit Rat

Higher Social Rents

More Inequality

V Low Supply & CR

Page 22: ENHR Housing Economics Workshop

School of the Built Environment

Backlog NeedUnmet Need Rate Under Different Supply Policy Scenarios

4.00%

6.00%

8.00%

10.00%

2000 2005 2010 2015 2020 2025

Year

Perc

ent o

f Hou

seho

lds

Baseline

High Supply (both)

More LCHO

More SR+LCHO

V Low Supply

Unmet Need Rate Under Different Supply Policy Scenarios

4.00%

6.00%

8.00%

10.00%

2000 2005 2010 2015 2020 2025

Year

Perc

ent o

f Hou

seho

lds

Baseline

Less Credit Rat

Persistent Credit Rat

Higher Social Rents

More Inequality

V Low Supply & CR

Page 23: ENHR Housing Economics Workshop

School of the Built Environment

Scenario Impacts

• Overall new build supply makes relatively little difference to tenure outcomes; somewhat counter-intuitive

• Priv renting mainly mirror image• Main reason is effective quantity rationing in private sector,

suppressing marginal household growth at expense of private rental sector

• Providing more subsidised housing does impact on tenure, e.g. more LCHO->more OO, but quite small impact even in medium term; more social renting would displace private renting.

• Contextual financial conditions, esp credit rationing, would have bigger effect – persistent CR would see OO stalled or falling

• Inequality – not big effect at level modelled• Rents – effects shown may be too strong (?)

Page 24: ENHR Housing Economics Workshop

School of the Built Environment

Concluding comments

• Findings of both stages of modelling support need take account of institutional & quantity rationing effects

• Nature and extent of recent changes go beyond conditions observed in base period (2003-07); hence estimated functions not sufficient to deal realistically with some of strains on system; need for additional feedback mechanisms in simulation

• General need to measure & model credit rationing & operation of private rented sector more effectively

• General supply policies have less impact on tenure than expected, although they do help to reduce need (gradually)

• Credit rationing has bigger effects on both tenure and need• Maybe governments should be more concerned with need

outcomes than with tenure per se