Stop the Arc Group – Written evidence (UKH0009)

Analysis of the housing market in England

Summary

i).            This document examines the demand and supply sides of the UK housing market for the period 2002-2018 at the regional level.  It was originally designed to understand the proposal of the National Infrastructure Commission for up to one million new houses for the Oxford-Cambridge Arc region, but eventually involved a nation-wide analysis in order to understand that more local picture.

ii).            Office of National Statistics’ (ONS) data on the growth of populations in the English regions and in Wales showed the highest per capita increases in London, the South-East and East of England.  The lowest rate of increase was experienced in the North-East.

iii).            Rates of increase were always higher than can be accounted for by the natural (birth-death) increase of the local population.  Only immigration into each region, and into the country as a whole, can explain the observed population growth rates.

iv).            Natural birth and death rates vary across the regions, with the highest birth rate (1.54% p.a.) and lowest death rate (0.63% p.a.) in London.  The highest natural net rate of population increase (birth – death rates) was, unsurprisingly, in London (0.91% p.a.), the lowest in Wales (0.039% p.a.) and the South-West (0.044% p.a.).

v).            In all areas except London immigrants contributed more to the recorded increase in local populations than did natural (birth – death) processes of the resident population.  In London (only) the majority (c 70%) of the increase in population size was due to reproduction of the resident population.

vi).            There is a significant inverse correlation between death and birth rates over the period 2002-2018 in all regions except London (Figs 8 & 9 and Table 2).  The reasons for this are not clear.

vii).            Annual per capita population rates of increase are inversely related to the distance of each region from London (Fig. 4).  This relationship also holds when rates of increase are adjusted for the net natural growth rate (birth-death) of the resident population (Fig. 11).  This suggests that new immigrants preferred to settle in the economically active areas in the South and East of England.

viii).            Housing supply over the period 2002-2018 was variable and obviously affected by the 2008 financial crash, but, with the exception of London, the average rate of house-building appears to have kept pace with the average increase in regional populations, with housing provision resulting in average occupancy rates of between 2.70 people per new household (pph) and 2.17 pph, close to the current national average (of 2.4pph).  The shortfall of housing provision in London is estimated to have been between about 16,000 and 22,000 new houses per year.  The latter figure over the 17 year period considered (2002-2018) corresponds to a housing shortfall (>369,000) 27% bigger than the entire housing stock of Oxfordshire (290,000 in 2018).

ix).            House price increases over the period 2002-2018 were highest in London (>£17,000 p.a.) and diminished with distance from the capital city, being lowest in the North-East (<£4,000 p.a.).  Neither a shortage of building land due to high population density nearer to London, nor inadequate provision of new houses outside the capital city seem able to explain this relationship, which appears due to the chronic under-supply of houses in London itself.  How this effect ripples out from the capital city across the nation is not clear.

x).            Unsurprisingly house affordability (the ratio of median house price to median wage locally) is high in London (c. 9.5) and diminishes with distance from the capital, again being lowest in the North-East (c. 5).  The higher salaries of Londoners mean that they can more easily afford houses outside London than can local people.

xi).            The South-West region experienced the greatest per capita increase in population due to immigration than any other region of the country (including London).  For every one locally-born person in the South-West there were more than 15 immigrants each year.  For every one locally-born Londoner there were only 0.44 immigrants (the lowest ratio of any region in England and Wales).  Considering its distance from London, the South-West also had the lowest housing affordability of any region outside the South and East of the country.  This may be explained by the fact that newly arriving immigrants require housing immediately, whereas new-borns will only require houses in two decades’ time, or more.

xii).            The total of the estimated numbers of regional immigrants into England and Wales (c. 253,000 p.a.) is shown to be similar to the Office of National Statistic’s estimate of net inward migration derived from UK Immigration Service data (c. 247,000 p.a.).

xiii).            The information in this document is finally put in the context of both the Redfern Review into the decline of home ownership from 2003 to 2015, which showed no clear link between housing supply and home ownership rates, and the more fundamental issue of the role of land supply in the commercialisation of the property market.

Analysis of the housing market in England

This document presents a preliminary analysis of the housing market in England, and tries to answer questions about what drives housing supply and prices. This has already been the subject of numerous reports and analyses[1], but is here discussed in an attempt to understand the housing market in the Oxford-Cambridge Arc area, and to challenge the assumptions that underlie the reasons given for a housing target of one million new houses across the Arc by 2050.  How does this target fit with Local Plans (most of which end in the mid 2030s), and is there any justification for it?

Figure 1. Map of the English regions

Level of statistical analysis

The present analysis uses data for the English regions and for Wales.  The attraction of using large areal units is that variable changes at a local level tend to average out over larger areas, so that the regional patterns can be taken as the ‘average’ of many local changes within each region.

  The English regions are shown in the map in Fig.1.  No Scottish data are used in the present analysis and data for Wales are combined into a single, country-wide figure.

The red dots in the map of England (blue for Wales) identify the centroids (geographical centres) of the regions.  The distance of each of these centroids from London is shown in Table 1. 

Table1.   Distances of regions of England and Wales from London

London itself is given a distance of 5km from the centroid of London because some analyses use logarithmic fits to the data (such fits cannot handle zeroes).  This figure of 5kms seems a reasonable one to allow for the urban sprawl of the capital city.  It doesn’t make a great deal of difference to the overall conclusions.

The single figure for Wales is the distance from London to the centroid of the Mid-Wales region shown on the map.

The mean distance from London to the South-East region (which includes Oxfordshire) is relatively small (38kms) because of the shape of the region, which curls around London.  The region’s centroid is therefore quite close to the boundary of the London region.

The housing market is clearly an example of supply and demand economics.  Prices go up as demand outstrips supply.  This document explores the regional changes in both demand and supply.

Housing demand

There are several ingredients in demand; natural birth and death rates of the resident or ‘indigenous’ population; in-migration and out-migration into and out of regions within the country (called ‘internal migration’); and finally immigration and emigration into and out of each region, and the country as a whole, from countries overseas.  These collectively contribute to the overall rate of increase in the total population in each area.

Changes in regional populations over time

Population figures for each region are produced by the Office of National Statistics[2] (ONS) and the present analysis uses the data from 2002 to 2018.  These are shown in Fig. 2.  Trends in regional populations over time are more or less linear, and the equations for each are shown to the right of the graphs.  The slope of each regression is therefore the increase in each region’s population per annum.

Figure 2.  Total regional populations of England and Wales from 2002 to 2018.  Equations on the right are for the linear regressions of total population against year.  The regression slopes are therefore the average net annual increases in population per region.

This increase must be expressed relative to the total population size in order to get a per capita rate of change over time.

Per capita population increase per annum (expressed as a percent) is plotted against the average population in each region for the period 2002 to 2018 in Figure 3.  Since adjustment has already been made for total population, the expectation is that this graph should be horizontal – i.e. that the per capita population rate of change per annum should be independent of total population size.  Figure 3 also includes as a horizontal red line the average (for England and Wales) natural net rate of increase for the same years, 2002-2018.  This net rate (of 0.03%) is defined as the difference between the Crude Birth Rate (CBR, the recorded births divided by the total population at the time of those births) and the Crude Death Rate (CDR, the recorded deaths divided by the same population total).  This horizontal line thus represents the average contribution to the rates of increase shown in the rest of the figure of the natural demographic processes (birth – death) of the current resident population in each region (regional variation in these natural demographic processes is examined later).

Two features stand out in Figure 3:

Figure 3.  Relationship between regional population growth rates (per capita, per annum) and total population size. The red line indicates the natural (birth-death) rate.

1) In all regions except the North-East, the net population change is always above the average natural (birth – death) rate (0.03% p.a., the horizontal red line in Fig. 3).  Thus immigration from outside the country must be contributing to population change in all but one region.

2) Net population change is highest in the largest populations in the London and the South-East regions.  Thus, if net natural rates of increase are constant across the regions, immigration into London and the South-East must be greater than elsewhere.

The London effect

The same per capita increase data are plotted against distance from London in Figure 4.

Figure 4.  Relationship between regional population growth rates (per capita, per annum) and total population size. The red line indicates the average natural (birth-death) rate.

Figures 5 & 6.  Crude Birth Rate (CBR, Fig. 5, left) and Crude Death Rate (CDR, Fig. 6, right) by region of England and Wales, 2002 – 2018.

It appears that the capital city is having a major effect on rates of population change across the nation.  The farther a region is from London, the lower its population’s rate of change per annum.  This effect peters out only in the North-East, more than 400km away from London.  Only here could the recorded rate of change be accounted for entirely by the natural birth and death rates of the resident population. 

The effect of geographically varying demographic rates

Figures 3 and 4 show a single average value for the net natural rate of population increase across England and Wales (the red lines), but this natural rate varies in both time and space.

Figure 5 shows variations of the Crude Birth Rate (CBR)[3],[4] from 2002 to 2018 by region, and Figure 6 shows the Crude Death Rate (CDR)[5],[6], both expressed as per 1,000 of the resident population. 

More or less consistently across the regions, birth rates rose and death rates fell for the first half of the period shown.  Then birth rates fell and death rates rose in the second half of the period.  Changes in birth rates were more pronounced than changes in death rates and their combination produces a peak in the net rate (birth – death) at around the mid-years of the sequence, 2010-2012, as shown in Figure 7, where the net rate is expressed as a percentage (i.e. per 100 population), thus in the same units as the growth rates in Figures 3 and 4. 

The most striking regional difference in Figs 5 and 6 is between London (the red curves in each figure), and the regions.  The capital city has both a very much higher birth rate and a very much lower death rate than elsewhere.  The resulting net rate in London is between two and three times greater than in any of the other regions (Fig. 7, which shows the average for all regions except London as a thick black line with red dots).

Figure 7. Net rate of natural population change (CBR-CDR) expressed as a %

The reasons for these variations in birth and death rates are not at all obvious.  The period concerned involved a global financial crash (2008), a period of very high net immigration into the country, especially from the EU8, the countries of Eastern Europe that entered the EU in 2004, and a period during which the UK Government attempted severely to restrict net immigration. 

Although the rates shown in Figs 5 to 7 are for the populations currently resident in each region it is likely that immigrants in one year become residents the next year, and so eventually contribute to the denominators of the per capita rates shown. 

Figures 8 & 9.  Relationship between the Crude Death Rate and the Crude Birth Rate/1000 population for the East of England (Fig. 8, left) and London (Fig. 9, right).

Another curious feature of these demographic data is that over the period shown there is an inverse correlation between crude death and crude birth rates for all regions except London.  One example of the former is shown in Fig. 8, with that of London in Fig. 9.  The correlations for all the regions are shown in Table 2.

Apart from that for London, all correlations in Table 2 are significant at the 5% level or better.

Table 2.  Values of the Pearson correlation coefficients (r) for the regional relationships between CDR and CBR (2 examples in Figs 8&9), levels of significance, p (one-tailed test) and mean regional net rates of increase 2002-2018.

Adjusting for geographically varying demographic rates

The regional data shown in Figs 5 to 9 and Table 2 can be used to correct the population growth rates shown in Figs 3 and 4 to examine the relationships between growth rates adjusted for net natural population change (birth-death) regionally and population size and distance from London, as shown in Figs 10 and 11.  Thus Figures 10 and 11 show the same relationships as in Figures 3 and 4, adjusted for regional variation in net natural population growth rates.  The relationship between overall growth rate and population size (Fig. 3) disappears once allowance is made for this natural net rate of change (Fig. 10, r = 0.208, n.s.), but that between the same overall growth rate and distance from London remains when the same allowance is made.  This relationship is significant when all data points are included (r = 0.599, p<0.05), and becomes even more so if the point for London is excluded (r=0.77, p<0.01), as shown in Fig. 11.

Figure 10.  Relationship between regional population growth rates (per capita, per annum) adjusted for net natural (birth-death) rates per region and total population size.

As before, the fact that all growth rates, now adjusted for natural population increase, remain positive can be explained only by net immigration into the country over the period 2002 to 2018.  Net immigration appears to be greatest for those regions closest to London (Fig. 11).  London, even with its high net natural growth rate (several times that of any other region; Table 2), still shows a positive adjusted growth rate that again can only be explained by net immigration into London from outside England and Wales.

Figure 11.  Relationship between regional population growth rates (per capita, per annum) adjusted for net natural (birth-death) rates and distance from London.

The South-West, with the second lowest net natural rate of growth (Table 2), shows the highest adjusted growth rates (Figs 10 and 11), indicating a high level of immigration into this region.  The population size of the South-West is the sixth largest of the ten regions, so the fact that the adjusted rate of increase is so high cannot be explained by the influx of people into a particularly small resident population.  It seems the South-West is favoured as a place for immigrants to settle for other reasons (the South-West’s population is more than double that of the North-East where there is no sign of a similar disproportionate influx).

The conclusion from this section is that a great deal of the demand side of the housing market is driven by net immigration into the country as a whole, concentrated on regions near to London, but having an effect on regions of England and Wales far away from the capital city.

Housing supply

The Office of National Statistics also produces data on the supply of houses in England.  Here we use the data for net additions (i.e. new builds plus other houses coming into occupancy minus demolished houses) for the English regions[7].  Wales presents its data differently and only, it seems, in the form of new builds which are here treated in the same way as the data for England[8]; the figures refer to the whole of Wales and not to each Welsh region separately.

Figure 12 shows the time series of additions to each region’s housing stock between 2002 and 2018.

Figure 12.  Net additions (England) or new additions (Wales) to the total housing stock, 2002 to 2018.

Clearly the financial crash of 2008 had a dramatic effect on housing starts, an effect that continued for several years.

Because there is no direct relationship between housing starts and any recent increase in local population (at best there would be a lag effect, as Local Plans respond to population changes), the most useful statistic to extract from these time series is the average number of new additions for the period concerned.  These are shown in Table 3.

Table3.  Average annual house additions by region, 2002 – 2018.

The relationship between these figures and the mean annual changes in population over the same period of time is shown in Fig. 13.

Figure 13. Relationship between mean net house additions and mean annual population increase for 2002-2018.

The regression line in Fig. 13 is calculated excluding the data-point for London, which is clearly below the relationship for the rest of the country.  For all regions excluding London the average rate of housing growth apparently kept pace with the average increases in population size.  The slope of the regression line indicates a net increase of 0.37 houses for every additional person, and therefore implies an occupancy rate of 2.70 (= 1/0.37), people per household, pph, not far from the national average of 2.4 pph at present[9].

The graph indicates with a double-headed blue arrow the housing shortfall for London – i.e. the under-supply of the London housing market.  This is 15,990 houses per annum.  That means that to house all additional Londoners at the same pph density as the rest of England, London would have had to provide almost 16,000 extra houses per year over and above what it built on average between 2002 and 2018 (i.e. >27,000 houses p.a., Table 2).

The regression in Fig. 13 does not go through the origin.  In the long-term it probably should do so, because if there is no net addition to the population there would be no need for net addition to the housing stock.

Figure 14. As in Fig. 13, but with the regression line forced through the origin

Fig. 14 provides an alternative equation for the same data, with the regression line now forced through the origin.  The slope of the new regression (0.46) indicates an occupancy rate of 2.17pph, and the London shortfall now rises to 21,761 houses per year, making a total of 369,937 houses ‘un-built’ in London over the 17 years considered.  This is equivalent to 1.3 times the current housing stock of the entire county of Oxfordshire (of 290,000 houses in 2018).  With chronic under-supply of houses in London, it is hardly surprising there is a housing crisis, and the next question arises as to what effect does the London housing crisis have on the rest of the country?

The effect of housing demand and supply on house prices

The ONS also produces data on mean house prices by region and on the ratio of house prices to earnings locally.  Fig. 15 shows trends in house prices over the same period, 2002-2018.

Figure 15.  Changes in median house prices, England and Wales, 2002 - 2018.

As in the case of population changes over time (Fig. 2), the slopes of the regressions in Fig. 15 represent the average increases in house prices per annum by region.  There was a slowing down of the increase in house prices after the 2008 financial crash, but acceleration again several years later to above the average long-term trend. 

The average annual increases (i.e. the slopes of the regressions in Fig. 15) are plotted against average increases in local populations in Fig. 16.  House prices increase at a faster rate, the higher the annual additions to the local population.

Figure 16.  Average annual increase in median house prices vs average annual increase in regional populations, 2002 – 2018

These same rates of house price increase are plotted against mean distance from London in Fig. 17.

Figure 17.  Average annual increase in median house prices vs distance from London.

One possible reason for the relationships shown in Figs 16 and 17 is that large population densities in London and the surrounding areas force up land prices and therefore house prices.  Is competition for scarce building land the underlying cause of these relationships? 

Figs. 18 and 19 show the relationship between population density (people per km2) and distance from London.

Fig. 18 shows that London has a far higher population density than any of the regions, but the relationship, if any, outside London is unclear. 

Fig. 18. Population density (people per km2) and distance from London.

Fig. 19 shows the data for all regions except London. There is no obvious trend in population density with distance from London.  The North-West has the highest density outside London and is almost the farthest away from it.  Thus the relationship between house price increases and distance from London (Fig. 17) does not seem to be due to any greater shortage of land for house building near to London than at any distance from it.

Fig. 19.  As for Fig. 18 but with the point for London omitted. 

If neither land shortage (Figs 18 and 19) nor an inadequate supply of regional houses for regional population increases for all regions except London (Figs 13 and 14) can explain the increase in house prices near to London (Figs 16 and 17), the only obvious conclusion is that the chronic under-supply of houses in London itself (Figs 13 and 14) is the underlying reason.  Sheer numbers (demand) are driving up house prices in and around the capital city.

House affordability

A key indicator in the housing debate is the ratio of mean house prices to mean gross earnings per household, often referred to as ‘affordability’.  The best ratio to use is that of the median value of each term.  Again these data are reported by the ONS and two graphs of the results are shown in Figs. 20 and 21; the ratio against the population rate of increase in Fig. 20 and the distance from London in Fig. 21.

Figure 20.  Ratio of median house price to median gross household income vs population rate of increase.

The significance of the relationship in Fig. 21 is increased if the single point for the South-West is omitted from the regression calculation, as shown in Fig. 22 (correlation coefficients of r = 0.90 and r = 0.98 respectively).

Figure 21.  Ratio of median house price to median gross household income vs distance from London (all points).

Given the distance from London, South-West house affordability is exceptionally low (i.e. a high ratio of house prices to (local) median earnings).  This may be the result of the disproportionately large amount of net immigration into the South-West over and above the natural increase of the local population (Fig. 11 and Table 4).

A Londoner effect?

When well-paid Londoners buy homes outside the capital, they can afford houses that local people cannot.  This effect is illustrated in Fig. 23 which shows both the local affordability of Fig. 22 and the affordability to Londoners (obtained by dividing median local house prices by median Londoner gross salaries).

As expected, Londoners have lower ratios than local people and this difference tends to increase the farther away from London.  Houses in the South-West are still more expensive (even for Londoners) than expected on the basis of its distance from London, but the ratio here for Londoners falls on the line of the ratio for local people.  Londoners in the South-West have a significant advantage over local people in the South-West for whom house affordability is exceptionally high in relation to their distance from the capital (previous section).

Figure 23.  Ratio of median house price to median gross household income vs distance from London (omitting South-West) for both local people (blue points) and Londoners (red points).  Unsurprisingly, the higher gross salaries of Londoners maks houses more affordable to them at all distances. Beyond about 150kms Londoners have price:income ratios lower than any other part of the country, i.e. they can afford houses local people cannot.

Figure 23 probably under-states the difference between the affordability of local houses by local people and by Londoners because the very much higher population of the capital mean there are very many more high net worth individuals (i.e. those above the median gross salary) in London than there are locally.  If it is this sub-section that seeks houses outside the capital the ‘Londoner graph’ in Fig. 23 falls even farther below the local graph.

The importance of migration in the UK housing market

Figure 11 shows that population growth rates (per capita, per annum) adjusted for net natural (birth-death) rates per region are all positive, indicating a significant rôle of net immigration into the UK contributing to increases in regional and national population totals.

This section examines whether official figures for net immigration into the UK can be reconciled with these regional net immigration estimates.  In theory the two sets of numbers are from independent sources.  Net immigration figures arise mostly from the Office of National Statistics’ International Passenger Surveys (IPS) ‘based on respondents’ stated intention to come to or leave the UK for at least one year’ – in other words, from data from Immigration Controls at ports of entry into the country.  Population figures used to produce Fig. 11 arise from periodic census and other data, presumably interpolated between official census dates to produce annual figures; birth and death rates come from Registrars in the various regions.

Table 4 shows the average net rates of increase from Fig. 11 (Table 4, column 2) and the average population sizes in the various regions (Column 3).  From these are produced (by multiplication) the estimates of the average annual immigration into each region (Column 4).  The total across all regions is 253,148 immigrants per annum into England & Wales for the period 2002-2018.   For comparison, the Table also shows the net natural rate of increase (Column 5) and the net natural increase in population arising from this net rate (Column 6), totalling an annual average increase of 168,679 due to the natural demographic (birth-death) processes of the resident populations.  In all areas except London, immigrants each year exceed the natural increase of the local population (the ratio is shown in Table 4, last column).  The South-East, South-West, and East regions all have more immigrants per year than does London, but London has by far and away the greatest increase in total population size due to the high net natural (birth-death) rate of increase of its resident population.

The Office of National Statistics’ (ONS) data on net immigration into the UK are from a document produced by the Migration Observatory[10].  Data are broken down into Non-EU, EU and UK components and are shown in Table 5 and Fig. 24.

The average annual net migration figure for 2002-2018 is 247,294 (Table 5, last column), essentially the same as the total of the calculated regional immigration figures in Table 4 (i.e. 253,148).

Figure 24.  Total net migration into the UK (red line) and its three components (data from Table 5)

With the proviso that the ONS net migration figures are for the UK rather than just England & Wales it is thus possible to reconcile the increases in regional populations estimated to be due to immigration with the UK total net immigration figures.

One Million New Houses and the Oxford-Cambridge Expressway

The National Infrastructure Commission’s (NIC) Partnering for Prosperity Report (2017) put forward the twin ideas of one million new houses and an expressway between Oxford and Cambridge[11] (an expressway is an almost-motorway standard road, carrying traffic at mile-a-minute speeds but with less traffic and more junctions than a motorway).  In February 2018 the NIC released a final version[12] of a planning options Report on the Ox-Cam Arc which included a map of how the one million houses might be allocated across the Local Authorities concerned[13].  This map, based on a previous housing analysis by Savills[14], is reproduced here with modifications as Figure 25. 

Figure 25.  Projected new houses across the Oxford-Cambridge Arc by 2050.  The pie charts show houses in current Local Plans (mostly to the 2030s) in pink, houses to accommodate ‘land-constrained markets’ (i.e. for London commuters) in dark blue, and houses ‘unlocked’ by the Arc development plans, including the Ox-Cam expressway, in dark orange.  Yellow splashes show percentage increase in housing stock from the 2017 levels (map modified from12). N.B. apart from Oxfordshire, the boundaries shown are not ceremonial counties.

The total number of houses in all the pie charts in Fig. 25 is 1,021,000 (230,000 in Local Plans, 238,000 for London commuters and 553,000 ‘transformational scenario’ or ‘expressway-unlocked’ houses), close to the aspirational one million figure.  The Local Plan estimate in this map (230,000) is exceeded by the Local Plans already in place across the Arc (estimated by Stop the Arc from numerous Local Authority documents to be about 316,000) but it is not clear how many of the projected 238,000 London commuter houses are incorporated into existing Local Plans.  At present, therefore, less than one third of the one million total are in Local Plans, leaving more than two thirds (1,021,000 – 316,000 = 705,000) to be built between the end of current Local Plans and 2050.  This would require a significant increase in the currently high rate of house delivery across the Arc. 

It is clear from later articles and publications that the houses will be distributed across the entire region shown in Fig. 25 and will not be restricted to the region along any expressway.

Figure 25 also shows the percentage increase (over the 2017 figures) in the housing stock brought about by the proposed increases shown.  The smallest increase of 66% will be in the currently relatively high density area of Aylesbury Vale, Central Bedfordshire and the Unitary Authorities of Milton Keynes, Bedford and Luton.  The largest increase (105%) will be in Oxfordshire (the City and all District Councils).  These figures should be compared with the average ONS-projected increase of 16% in the UK population by 2050.  Thus the Arc housing stock will increase by between four and more than six times the projected national average increase by 2050.  These extremely high rates of growth, unprecedented over such a large area, are justified in light of the aim to make the Arc region a high-growth, high-tech area, an ambition eagerly promoted by estate agents and developers at numerous conferences on Ox-Cam Arc developments.  Relaxation of certain planning controls (for example over green belt development), the introduction of new powers (of compulsory purchase, or permitted development) and new off-site construction methods are all ingredients in the Radical Regeneration Manifesto put out by a consortium of developers in 2019[15], and many of these were echoed in the Planning for the Future proposals from the Ministry of Housing, Communities and Local Government following the 2020 budget[16] and have subsequently found their way into the new Planning White Paper[17].

Fig. 25 assumes an average of about 1.9 people per house (1,915,000 people in total), well below the current UK average household size of 2.4 in 2017[18], but in line with the expectation of a fall in people per household in the future[19] (a fall likely to be slowed down in and near London by high house prices).

The Oxford-Cambridge Expressway was officially cancelled in March 2021 but the job and economic ambitions remain the same[20] even though, in July 2021, the Minister of State for Housing denied the one million houses figure for the Arc remains a Government target[21] - a figure that the NIC’s own ‘careful research’ had shown was essential to deliver the job and economic ambitions that the Government, and Minister, still claim. 

Importantly here the Expressway had been the essential ingredient in the highest, transformational growth scenario of the NIC’s Partnering for Prosperity Report.  With the expressway now removed, the transformational scenario should no longer be considered as an option, but there are no signs that this has yet happened.

Discussion and Conclusions

The following section assumes that the relative contributions to population growth of new immigrants into England and Wales and of children born to current residents (some of whom may have been immigrants in previous years) are as suggested in this document, and draws conclusions about future population changes that may occur.[22]

The average regional population weighted growth rate of England and Wales for the period 2002 to 2018 was 0.76% p.a.  Over the same period, the year-on-year increase in UK GDP averaged 1.70% (range +3.3% to -4.2%)[23].  Thus UK GDP was growing faster than the population of England & Wales, something that presumably attracted more immigrants each year.  The forecast for UK growth in 2020 was only 1.1% before the outbreak of coronavirus[24] and is now likely to be considerably less (a recession is anticipated).  This will make the UK a less attractive destination for immigrants in future.

Immigrants contributed 60% to the overall increase in the population of England and Wales over the period 2002 to 2018, the balance being the natural increase of the resident population (some of whom may have been immigrants in previous years).  The key difference between these two sources of population increase is that most immigrants arrive already educated, are mainly at an age ready to contribute to national GDP, and need housing immediately (although a higher proportion of recent immigrants, compared with residents, rely on rented accommodation rather than owner occupation[25]).  Newborns need to be educated, will require another 18 or more years before they can contribute to the economy, and only then may enter the market for housing.  Thus the immediate demand for local houses will be affected by the relative mixes of immigrants and newborns as regional populations increase.

Predicting future levels of immigration is extremely difficult but such predictions, usually made by the Office of National Statistics, contribute to estimates of population growth and future housing needs in the form of Strategic Housing Market Assessments (SHMA) or Objective Assessments of Housing Need (OAHN).  The more rapidly any region has grown in the recent past, the greater the future housing needs assessments (which are partly based on extrapolating recent population trends).  This is a recipe to increase not decrease the current inequalities across the nation, because such inequalities increase as more slowly growing regions are ‘condemned’ to more slow growth in future, and faster growing regions are expected to continue their high rates of growth.  If the majority of population change is due to the very variable rates of immigration (which is a function of the global employment market), extrapolating past trends into the future is doomed to result in inaccurate predictions of future populations.  What is needed is almost the reverse of what we have at present.  Currently faster-growing regions should grow more slowly in future, and vice versa.  This would involve a policy decision rather than a planning one.  The argument that areas ‘need’ the level of housing predicted by SHMA to accommodate the offspring of residents is mostly fallacious.  The ONS-predicted average increase in the UK population to 2050, of about 16%, is split more or less equally between the offspring of current residents, and new immigrants. Thus any predicted growth in local housing above 8% is for immigrants into the region.  New immigrants do not have any obvious sense of place (although they do tend to settle where previous immigrants from the same origin countries are already established) and so they could settle anywhere in the UK, given employment opportunities there.  Creating new employment opportunities outside the South-East of the country is again a political decision rather than a planning one.

Despite these objections to current methods of assessing future housing needs it seems that house building in the UK kept pace with the (SHMA-determined) housing requirements of locally increasing populations for the period considered, 2002 – 2018, in all regions except London.  Outside London, enough new houses were built for occupation rates of between about 2.2 and 2.7 people per household (pph), reasonably close to the current national average of 2.4 pph. 

There was, however, a chronic shortfall of new houses in London, leaving the capital short of between about 16,000 and 21,000 new houses per year over the same period of time (figures in line with the emerging London Plan that identifies a 20,000 houses per year shortfall[26]).  What happened to the Londoners needing those houses is unclear although, unusually across the nation, 69% of the net increase in London’s population was provided by new-borns to current residents.  As pointed out above, children only need houses after a lag of two decades or more and, when their time comes to enter the housing market, they may well be forced to leave the capital city to find somewhere to live.

Given the apparently sufficient supply of houses outside London, the question arises as to why house prices rose more rapidly in regions closest to London than in those farther away from the capital.  There are several possible explanations.  The first and most obvious is that previous generations of Londoners were forced out of London for the reasons just mentioned – an inadequate supply of houses in the capital itself. 14.6% of all the workers in the East and 12.6% of all workers in the South-East regions commute to London to work (2011 census data[27]), the two regions outside London showing the most rapid increase in house prices after London itself (East Midlands’ workers are the next most likely to commute to London, but only 1.05% of them do so).  A second reason is that various changes in legislation resulted in the expansion of the ‘buy-to-let’ market, such that 20% of all mortgages in Q2 2016 were of this type[28].  ‘Buy-to-let’ is seen as an investment in property with the expectation of a rise in value of that property in the medium to long term, and is likely to be more common in regions with many people able to raise the necessary funds (e.g. by equity release on their existing properties) and with many other people (including recent immigrants) desperate for accommodation.    Finally, there was considerable inflow of foreign capital into the UK housing market, seen as a good investment by overseas buyers27.  London and the regions around it would be the obvious targets for such investments. 

The ten years from 1996, when house building kept up with housing demand but house prices rose by approximately 150%, show that house prices do not obey the strict rules of simple supply and demand.  The short-term dynamics of house prices in the UK were discussed in the Redfern Review1, but much of the increase in the cost of housing is due ultimately to the fact that houses are built on that one commodity of which there is a fixed supply – land[29].  The quantity of land cannot respond to the demand for it, however high that demand might be.  Smart investors realise this and invest in property solely for an anticipated increase in the value of the land on which the property sits.  Indeed, properties bought by speculators may be left unoccupied and un-maintained, in the expectation of an increase in the value of the land alone27, Chapter 5.  

With those other than genuine house seekers in the property market it should be no surprise that Government attempts to bring down house prices in areas considered at greatest need are failing.  The housing needs formula contains a factor to boost house numbers in places where the ratio of median house prices to median incomes (the ‘affordability ratio’) is high, in the expectation that this might drive house prices down.  Figure 26, with data for all English Local Authorities for the period 2009-2019, shows the relationship between the percentage change in the affordability ratio (on the y-axis) and the percentage change in housing stock over that period (one point for each Local Authority).  The local boost of housing stock (determined by the formula) should have reduced the affordability ratio over time; the bigger the increase in stock, the larger should have been the fall in the affordability ratio: the expectation is that the line in Fig. 26 should therefore have a negative slope.  It doesn’t; in fact the line has a significant and positive slope (although the scatter of points renders the graph useless for making predictions).  It seems that no amount of house-building can overcome the distortions of the housing market brought about by the large community that sees houses as speculative investments rather than as essential roofs over people’s heads.

Fig. 26.  Relationship between the percentage change in housing affordability (the ratio between median house price and media wage) and the percentage change in housing stock for all English Local Authorities, 2009-2019 (based on ONS data).

Regional growth trends are dependent upon the success of local businesses which are competing with each other in both national and international markets.  Whilst from the public’s point of view it might seem better to spread businesses and investments across the country evenly, businesses often prefer to grow locally (the ‘agglomeration effect’).  They compete with others on cost, quality, customer service and timing, things that might be deliverable independent of the companies’ geographical locations, but their success also depends upon factors over and above these that may well be geographically constrained.  These additional factors contribute to a ‘differentiated value proposition’, essentially factors that make businesses relatively more attractive to customers.  Thus, for example, Astra Zeneca chose to locate its new research facility on the Cambridge Biomedical Campus just South of the City.  One of the main reasons for doing so was the proximity of the site to the expertise within the Campus and University nearby.  Reasons such as these are used to justify selecting the Oxford-Cambridge Arc region for further development, with its two internationally highly rated Universities and the technological expertise of Cranfield.  But 46% of the money from the UK’s seven Research Councils goes to Oxford, Cambridge and London, and just 20% to the North[30].  A fairer distribution of research funding could be one step towards rebalancing the North-South divide (unfortunately, a committee set up in 2017 to redress this imbalance was “overwhelmingly dominated by representatives based in the South-East”).   Other sites for Arc type developments are available in England, and a number have additional advantages (building land, cheaper housing and good public transport) over the Oxford-Cambridge Arc[31].  A later study identified a potential growth hub in each of the regions outside the South-East, including in Scotland and Wales[32]

To reduce inequality, both businesses and houses should be located in the currently deprived areas of the country, so that they too can make a net contribution to national GVA.  Investing away from the over-heated South-East has been shown independently to produce a ‘win-win’ situation for both the North and South of the country, rather than a predicted ‘lose-lose’ if South-East investment continues as at present[33].  History shows that a variety of interventions are required in order to avoid free market forces distorting national economies.   Housing provision of the last few decades has shifted from mostly supply-side interventions that dominated post second World War thinking, with Local Authorities building almost half of all new homes each year (in 1969), to demand-side interventions that have predominated since the 1980s (e.g. Right-to-Buy and Help-to-Buy schemes), including those that increase the rental sector (e.g. the originally favourable tax advantages of Buy-to-Rent schemes).  The astronomic increases in house prices since the 1990s are due either to a failure of some interventions (for example, houses lost from the social sector under right-to-buy schemes were not replaced on a one-for-one basis) or to the apparent ‘success’ of others (for example, it is generally accepted that help-to-buy schemes increase house prices). 

In the 1930s, only 2% of the cost of a house was for the land on which the house was built.  Today that figure is near 70%[34].  Much of the increase in house prices is due to the increase in the cost of building land (the actual costs of construction have risen more or less in line with inflation).  As John Stuart Mill first pointed out, the value of land for building is a benefit created by society (through the building of cities) that flows to only one person, the land-owner, who may have done little or nothing to increase the land’s commercial worth[35].  And it is land and its ownership that has distorted the economics of house-building in recent years.   There is a seemingly obvious solution, beyond the introduction of a land value tax suggested in the final Chapter of 27.  Much land is still publicly owned and vast amounts, approximately 10% of the entire UK land area, have been sold to the private sector since the 1980s30.  Is it time for Government intervention once more on the supply side of the housing market?  The Government still has supplies of public land and it recently relaxed the rules on Local Authority borrowing for house-building.  Resurrecting the house-building capabilities of Local Authorities will take time, but a steady supply of public sector, social housing is the only sure way to provide the right sorts of houses in the right places for poorly paid key workers and others for whom secure renting, or owning their own home, is presently an impossible dream.

August 2021

28