资源描述
14thAnnualDemographia International Housing Affordability Survey: 2018Rating Middle-Income Housing AffordabilityAustralia Canada China (Hong Kong) IrelandJapan New Zealand SingaporeUnited Kingdom United StatesIntroduction byFelipe Carozzi, Paul Cheshire and Christian HilberLondon School of EconomicsData for 3rdQuarter 2017 14th Annual Demographia International Housing Affordability Survey (2017: 3rd Quarter) A 14th Annual Demographia International Housing Affordability Survey INTRODUCTION Measuring Affordability: Alternative Perspectives Felipe Carozzi, Paul Cheshire & Christian Hilber London School of Economics Felipe Carozzi Paul Cheshire Christian Hilber Britains Office of National Statistics reported that houses were the most valuable asset in the UK at 5.5 trillion, accounting for 62% of the UKs total net worth at the end of 2015: up from 48.7% 20 years previously. This compared to net asset values of Equity and Investment funds of a mere 115billion.According to the Halifax the value of the UK housing stock had risen another 500billion to 6tn by November 2017, up from just over 4tn in 2007. No wonder the British a country largely of homeowners are obsessed with the asset value of housing while at the same time complaining about the real crisis of housing affordability. This is of course the first paradox of housing affordability: housing is both an asset and a good providing a flow of housing services a place to live. The interests of house owners do not align with those of would be house owners. Rising house prices relative to incomes pit the old against the young and the rich against the poor. Before we can have useful debates or even give a balanced assessment of the issues we need good measures. Here Demographia has done wonders over the past decade to focus public debate on the inequity of rising house prices relative to incomes. As Oliver Hartwich in his Introduction to the 13th edition last year said “Demographias median multiple approachfirmly established a benchmark for housing affordability by linking median house prices to median household incomes. It is not a perfect measure because it does not account for house sizes or build quality. But it is the only index that allows a quick comparison of different housing markets, and it is the best approximation of housing affordability measures we have to date.” 14th Annual Demographia International Housing Affordability Survey (2017: 3rd Quarter) B We agree. Apart from the median multiple being simple and useful, it is also the only measure out there for purposes of international comparison. The point of this introduction is to explore how it fares when we pull it about: using more precise definitions of what spatial housing markets might really be, accounting for differences in housing unit size and looking at the impact of housing affordability for households at different places in the overall distribution of incomes. We do this using data for Britain: the cradle of housing unaffordability and the originator of the ideas and mechanisms of planning which have contributed so much to the problem: Green Belts and planning by unpredictable political processes! What we do and what we find i) Replication Demographia did not originate the median multiple (MM) as a measure of housing affordability but they have done great work popularising it. As academics, however, we believe in replicability. So, our first task was just to see if, using no more than the information about sources and methods in last years issue, we could replicate the 3rd Quarter 2016 MMs for the UK housing markets shown in the 13th Survey1. Our exact sample of transacted houses was not identical and our estimates of median household incomes are a bit different but the broad conclusion is that the replication was successful. The correlation between our estimated MMs and those reported by Demographia is 0.92. Both sets of MMs are shown in columns 3 and 4 of Table 1 and illustrated in the following figure. ii) Housing markets The next issue is what is a housing market. Elementary urban economics tells us that housing markets are spatially bounded and their extent is determined by the need to travel to jobs.Almost all house purchases are paid forout of incomes and to earn those incomes people must travel to work.So, housing markets are essentially coincident with broadly defined urban areas Functional Urban Regions. Until very recently, although there were official definitions and data for such areas in the US, they were not defined in many other countries. Now since 2014 the OECD has defined such regions on a comparable international basis but provides only a limited set of data for them not including house prices or even household incomes. As a result, while Demographia is able to use core-based urban regions for the US and a few other countries, it does not in Europe. Here they have had to use data for Eurostats official administrative regions. In Britain these are a mixture of Counties, Unitary Authorities and even Government Regions, such as Greater London. These, in economic terms, are a disparate group, seldom corresponding to actual housing markets, so our next step was to estimate MMs for areas more closely matching housing market areas. In Britain, there is a set of widely used Travel to Work Areas 1Sources: In addition to the data referenced in Demographias 13th Edition we used: Annual Survey of Hours and Earnings (ASHE) and the Effects of Taxes and Benefits on Household Income dataset from Office for National Statistics to calculate median household income and income distributions.; National Statistics Postcode Lookup Centroids from Office for National Statistics to identify TTWAs; Domestic Energy Performance Certificate Register published by Department for Communities and Local Government to calculate house size and price per m2 in England and Wales; Policy paper tax and tax credit rates and thresholds for 2016-17 from HM Treasury to calculate after tax and national insurance household income. 14th Annual Demographia International Housing Affordability Survey (2017: 3rd Quarter) C TTWA) that we take as functional urban areas here.2 We focus only on TTWAs that were located within the administrative regions used by Demographia and that had a significant town in them. We can see from Table 1 that while some Demographia Housing Markets such as Aberdeen, Cardiff or Liverpool & Merseyside corresponded to just one TTWA, others such as Birmingham & West Midlands, covered several TTWAs and at the extreme, in London Ex-urbs, there are 12 TTWAs with MMs in our replications varying from 4.6 (Peterborough) to 7.8 (Brighton). These compare to our replication MMs for the corresponding Demographia market of 6.0. One should remember, however, that our TTWAs in the London Ex-urbs do not cover its whole area. Note: The figure illustrates our replication of the Demographia median multiples. The vertical axis represents the Demographia multiple as included in the 2016 report. The horizontal axis represents ourreplication multiple. The line corresponds to fitted values of a linear regression. The correlation between variables is 0.92. 2These are not strictly urban regions since the criterion used to define their boundaries is self-containment: 75%, of the employed population resident in them also work in them. Unlike urban regions therefore their extent necessarily covers the whole country and some may be quite rural: the Orkney Islands, for example, constitute a TTWA. 14th Annual Demographia International Housing Affordability Survey (2017: 3rd Quarter) D TABLE 1: ALL HOUSING MARKETS IN THE UK BY GEOGRAPHY Multiple (Median House Price/Median Household Income): 2016 3rd Quarter House Market Multiple 1 2 3 4 5 6 7 Demographia TTWA Demographia Replication TTWA Replication2 TTWA2 Aberdeen Aberdeen 4.551 4.318 4.744 Birmingham & West Midlands Birmingham 4.753 4.323 4.667 4.526 4.844 Wolverhampton and Walsall 4.275 4.420 Dudley 4.233 4.415 Coventry 4.354 4.403 Blackpool & Lancashire Blackpool 4.084 4.357 4.332 4.129 4.054 Preston 4.304 4.110 Lancaster and Morecambe 4.443 4.329 Burnley 2.750 2.676 Blackburn 3.633 3.371 Bournemouth & Dorset Dorchester and Weymouth 8.863 6.938 6.453 6.764 5.893 Poole 7.430 7.360 Bournemouth 7.136 7.240 Bristol-Bath Bristol 6.189 6.413 6.175 6.661 6.368 Bath 7.719 7.129 Cardiff Cardiff 5.017 4.531 4.037 4.680 4.094 Derby & Derbyshire Derby 4.132 3.868 3.968 3.963 4.035 Chesterfield 4.092 4.213 Dundee Dundee 4.076 3.666 3.744 Edinburgh Edinburgh 4.370 4.671 4.750 Falkirk Falkirk and Stirling 3.576 3.133 3.248 Glasgow Glasgow 4.043 3.845 3.502 Hull & Humber Hull 4.277 4.328 4.086 4.103 4.010 Grimsby 4.057 3.908 Leeds & West Yorkshire Wakefield and Castleford 3.826 3.837 3.934 4.005 3.912 Leeds 4.295 4.574 Bradford 3.720 3.860 Leicester & Leicestershire Leicester 4.972 5.004 4.884 5.131 5.056 Liverpool & Merseyside Liverpool 5.055 3.677 3.763 3.638 3.655 London (Greater London Authority) London 8.494 8.328 8.163 9.185 8.773 14th Annual Demographia International Housing Affordability Survey (2017: 3rd Quarter) E London Exurbs (E & SE England) Peterborough 7.054 6.014 4.624 6.084 4.386 Luton 6.560 7.259 Southend 5.969 6.108 Medway 5.111 5.284 Milton Keynes 5.365 5.398 Brighton 7.788 8.621 Portsmouth 4.971 4.978 Southampton 5.929 5.972 Isle of Wight 5.085 5.232 Bedford 5.462 5.346 Oxford 6.365 6.269 Cambridge 6.759 6.423 Manchester & Greater Manchester Manchester 4.462 4.132 4.306 4.357 4.560 Middlesbrough & Durham Durham and Bishop Auckland 4.130 3.599 3.410 3.244 3.026 Middlesbrough and Stockton 4.260 4.020 Newcastle & Tyneside Newcastle 4.332 4.112 4.150 4.180 4.165 Newport Newport 4.620 4.346 4.213 3.846 3.865 Northampton & Northamptonshire Northampton 5.080 5.036 5.171 4.831 5.004 Kettering and Wellingborough 4.559 4.410 Corby 4.785 4.636 Nottingham & Nottinghamshire Nottingham 4.346 4.336 4.114 4.294 4.159 Mansfield 3.998 3.829 Perth Perth 4.452 4.232 4.408 Plymouth & Devon Plymouth 7.072 6.207 5.233 5.867 5.087 Exeter 6.336 6.020 Sheffield & South Yorkshire Sheffield 4.267 3.819 4.080 4.016 4.387 Doncaster 3.675 3.567 Barnsley 3.408 3.417 Stoke on Trent & Staffordshire Stoke on Trent 4.843 4.431 3.728 4.454 3.877 Stafford 4.659 4.350 Swansea Swansea 4.903 3.854 3.482 3.678 3.254 Swindon & Wiltshire Swindon 6.928 5.787 5.405 5.556 5.258 Telford & Shropshire Telford 5.810 5.636 4.879 5.176 4.412 Warrington & Cheshire Warrington and Wigan 5.110 5.072 3.718 4.866 3.721 Chester 4.755 4.609 Warwickshire Leamington Spa 5.551 5.586 6.901 5.287 6.301 Median Market 4.686 4.341 4.426 4.490 4.413 Note: Table reporting our replication of the Demographia MM (column 4), the calculation of MM for TTWAs (column 5), the MM after size adjustments as described in the text (column 6) and a column combining both TTWA and unit size adjustments (column 7). 14th Annual Demographia International Housing Affordability Survey (2017: 3rd Quarter) F The basic conclusion of this exercise is that, as expected, administratively defined regions are very varying in their relationship to real housing market areas and so can conceal a big range of affordability within them: about one quarter of the variance in TTWA MMs is not explained by regional level measures. iii) Does adjusting for the size of houses make a difference? Our initial thoughts were that once one controlled for their size, houses in the more expensive markets would be considerably less affordable than they appeared to be on the simple MM measure. The reasoning was that because they were more expensive relative to incomes, they would also be smaller, and simply measuring the median house price would not reflect that. Further thinking suggested, however, a countervailing force: while space might be more expensive, incomes are also generally higher in more expensive, larger cities, and research shows that people spend more on it as they get richer: they want bigger bedrooms for example and perhaps a spare one, possibly an additional bathroom.This might tend to make houses bigger where people are richer even though the unit cost of space may be higher. In fact, research estimating income elasticities of demand for housing space suggests that peoples spending on space in houses rises faster than incomes if income increases by 10% spending on housing space increases by about 20%. Indeed, as one of us has frequently argued, this is one of the main drivers of increasing real house prices over time as incomes rise in the face of constraints on the space for houses imposed by restrictions on urban growth. These two forces might work against each other, therefore, meaning that adjusting for differences in house sizes might make only a small difference to affordability. The data on the price of space in houses is only available for England and Wales3 so in the columns of Table 1 showing the results, 6 and 7, we have had to exclude the Scottish markets. What we find is that whether we compare the Demographia Markets or the TTWAs we prefer, controlling for size makes not a lot of difference to measured affordability using the simple MM. The median house in Britain is very small 83.9 m2: new houses are even smaller at 76m2. This compares to 137m2 in Denmark or 214m2 in the US, according to RIBA,so clearly you get a lot less house for your money in Britain than in Denmark or the US, but the difference in this across British markets is not so big. We would argue the most revealing comparison is between our replication MM for the TTWA with the size adjusted TTWA MM. To estimate this we assume that all markets have the same median house size of 83.9m2. On this measure in the least affordable market the London TTWA affordability deteriorates from an MM of 8.2 to 8.8. In a low income and more affordable market such as Hull, TTWA affordability is almost the same on both measures while in one of the most affordable TTWAs in all of Britain Burnley in Lancashire affordability actually improves once the size of houses is taken into account. 3 Data on the area of houses is collected for purposes of estimating energy efficiency: for this
展开阅读全文