More People, Less Food

Transcription

1 More People, Less Food The implications of population growth on housing, food production and amenity in the England Diandian Chen Candidate number: Academic supervisor: Jeffery Jones Project supervisor: Roger Martin OR440 Applied Operational Research MSc in Management Science (Operational research) Academic year

2 EXECUTIVE SUMMARY The population of England in 2013 was estimated as 52 million. With a decreasing household size, the household number increased rather faster. However, in the past 20 years, the dwelling completed every year was decreasing. And the housing shortage becomes more and more severe. It is urgent that measures should be taken by authorities to cope with the accelerating population growth This report firstly looks back to 1994 and discusses the historical facts about housing problem in the past 20 years. The population of England in 1994 is estimated as million. Based on the assumption that population was stable since 1994, then: 125,360 households were statutory homeless in 1994 With the house density of 24 dwellings per hectare, 5223 hectares of land were required to allocate all homeless population in The areas of brownfield changed for residential use in 1994 and 1995 can be utilized for 153,900 new dwellings construction. So the house demand from homeless people can be met by hectares of farmland can be saved during 1994 to million pounds agriculture outputs can be produced in these farmlands 3610 hectares of greenbelts can be saved. In the forecasts sections, the land areas change for residential predicted according to the population projection in the next 40 years. The first case is based on the key assumption that dwelling density varies with the population change each year. The second case is based on the key assumption that dwelling density is fixed at a relatively high level of 50 dwellings per hectare. The last case refers to the plans of local authorities that 700,000 houses will be planned in the countryside - including almost 200,000 allocated for the greenbelts. The main findings are listed below: variation of the population projection brownfield area exhausted by the year Total agricultural area change for residential use (hectares) Total agricultural income loss (million) Total greenbelt area change for residential use (hectares) Low Case 1 High principle Low Case 2 High principle Low Case 3 High principle

3 Contents EXECUTIVE SUMMARY... 1 INTRODUCTION Background Purpose Literature review... 3 HISTORICAL FACTS AND ANALYSIS Approaches Data representation ) The overall picture of population growth ) The house building and shortage ) Land type and land usage: ) The household size: ) The dwelling densities: ) Total Income from Farming: Analysis and discussions FORECASTS UP UNTIL Approaches Results and discussions ) Float forecast model ) Fixed forecast model ) 10 years construction on greenfield model LIMITATIONS CONCLUSION AND RECOMMENDATIONS REFLECTIVE CHAPTER REFERENCE Live tables on house building (Online), Office of National Statistics 2014, Available: ABBREVIATIONS APPENDIX Terms of Reference

4 INTRODUCTION 1.1 Background Over the last 20 years, the UK population has grown rapidly. As the major contributory factor for housing expansion, population growth brings pressure for more house-building, much of it on open land. However, the capacity of the previously developed land for house building was decreasing from All vacant and derelict land and building among previously developed land shrunk by 18% from hectares to hectares in Moreover, this trend continues within brownfield as more and more previously developed area change for residential use. Thus, to release the brownfield area from the pressure of growing house demand, the excess demand of new house building has to be allocated on the greenfield land which is land that has not been developed before. Consequently, this will firstly reduce the farmland and affect UK s food production capacity; and secondly reduce greenbelts which have been considered inviolable lungs and recreational areas for the cities. Now many Councils are actively considering building in greenbelt land in the next 20 years. With a stable population, none of this (or very little) would have been necessary. 1.2 Purpose The reason for undertaking this report was to understand better the relationship between UK population growth and housing expansion, and the impact that increased housing has on the community food production and its environment. To quantify the cost of population growth in the last 20 years, in terms of: loss of food production (hectares of farmland); loss of amenity especially the greenbelts land (hectares); and numbers of unhoused people which requires more affordable housing; and to project these costs forward for the next 40 years to 2051 at the top and bottom of the ONS projected population range. 1.3 Literature review In the previous study, many literatures have discussed the population expansion and housing problem. Clearly more household needs more dwellings to live in. The population growth directly leads to demand of new house. However, England is meeting a serious situation that the new house demanded is more than the new house completed. Home Builders Federations (HBF) (2013) published a figure that, at least 220,000 homes need to be completed a year to meet the new household projection, regardless the historical undersupply of houses. In fact, only 107,820 homes have been completed in Obviously, there is a large gap between the house demand and supply. Although there are 700,000 empty homes in England, less than 260,000 among those have been empty for more than 6 months. Therefore, even if utilise the vacant homes only can make up the house shortage no more than three years. 3

5 In order to relieve the house shortage, more dwellings have to be built on both greenfield and brownfield. Some studies have indicated several important implications of new dwelling construction. In the first place, according to the policy advice in March 2000, brownfield is preferred than greenfield for the new house building. As the development capacity of brownfield abates, more houses are built in cities with high density. Furthermore, a larger proportion of housing land developed within cities accompany with a larger proportion of new dwellings placed on brownfield area. (Bibby and Brindley, 2006). The overcrowded houses in high density area lead to lower quality of life. Secondly, owing to the limited capacity of developed land, many greenfield areas have been changed to residential use. The area of land in agricultural holdings in the UK fell on average by about 15,400 ha per annum between 1983 and Bibby (2009) argues the residential purpose amount for more than half of this change. As a consequence, some farm lands have been occupied and correspondingly food productions have been reduced. From the historical data, UK self-sufficiency has fallen by 15% since 1995 or 1% p.a. (Department for Environment, Food and Rural Affairs 2010). The other greenfield such as greenbelt are shrinking due to filling new houses. The latest land statistics released shows that (DCLG, 2013), the proportion of new dwellings built on greenbelt has kept at 2% since Reductive greenbelt can contribute to stress and mental health issues (PM, 2011). Because there will be less opportunities to access green area; less opportunities for outdoor sport and recreation; weakened landscapes, visual amenity and biodiversity. The last but not the list, the housing also have implications to climate change. The carbon dioxide equivalents emissions from domestic heating, cooling and lighting lead to residential green house gases. Moreover, green spaces can help reduce excessive urban temperatures building. However housing on green space would increase the temperature of urban area. This report focuses on the forecasts of land change for residential use (up until 2051) and implications of land use change as a result of population growth in England. This is an undeveloped subject in the previous study. The models are completed constructed by the author with the inspiration of the academic supervisor Mr. Jeffery Jones. This quantitative analysis will intuitively produce possible outcome by taking different scenarios into considerations. The comparison between the results generated from different cases may provide some basis for future policy making. In addition, the models are sustainable for future analysis by updating population projection figures and other statistics. 4

6 HISTORICAL FACTS AND ANALYSIS 2.1 Approaches In order to estimate the influence of population growth on housing, it is firstly assumed as a base case, that population was stable since On this assumption, the number of new homes needed in year 1994; the number of houses could have been built on brownfield land available to that date; and how much farmland (if any) would have been needed, can be calculated. Based on the historical facts, the influence of actual population growth on land use and housing in the past two decades is estimated by linear regression via STATA Data representation 1) The overall picture of population growth Since 1971 the population of England has grown by over 7 million people (15.2%). Nearly half of the population growth comes from 2001 onward because of the high levels of net inward migration. Nevertheless births slow down through the 1990s while the total population grows faster year after year since Figure illustrates the components of population change for England in the past 20 years. One of the crucial components is the net migration represented by the blue bar. The other one is the number of natural change calculated as the birth minus the death. During this period, the population change partly leads by the expansion of the European Union in 2004 and In addition, the number of births has also increased during this period, driven by both the immigration of women of childbearing age (15-44) and rising fertility among UK-born women (ONS). The population of England has reached million by the end of 2013 and this figure is continuously increasing Figure2.2.1: Component of population change for England Mid Mid 2012 (thousand) Natural change Net migration & Total change

7 2) The house building and shortage There is no doubt that more population leads to more houses and more land needed for construction. The figure shows the historical data of permanent dwellings completed each year from 1983 to In general the number of dwellings completed follows a downward sloping trend and varies between 100 thousands and 250 thousands dwellings per year. The historical highest level occurred in 1978 when 241 thousand of dwellings have been built. In the past 20 years there is a peak of dwellings completion in 2007 (177 thousands dwellings completed). Then new house completed experience a rapid drop through recession period and reach the bottom of 107 in , , , , ,000 50,000 figure 2.2.2: House building: permanent dwellings completed, England Private Enterprise Local Authorities Housing Associations All Dwellings As Barker (2004) argued the current build rates implies there is a current shortfall of 39,000 homes in England per annum, of which 8,000 are private sector and 31,000 are affordable homes. In addition there is a backlog of around 450,000 households without self contained dwellings. Historically, there was a huge short fall of house building. RPG for England indicates that nearly 155,000 houses per annum should be built, but average annual completions between 1996 and 2001 were just over 140,000, suggesting a shortfall of nearly 15,000 units. (Barker s Review, 2004) 3) Land type and land usage: House construction is mainly considered to be developed on two types of land the greenfield land and the brownfield land. Greenfield land is land that has not been developed before (including agriculture land, amenity area, other greenfield). Brownfield or so called developed area is land previously used for industrial purposes or some commercial uses. The brownfield is always the first choice for constructing new buildings for residential or commercial use. But sometimes suitable brownfield land cannot be fully utilized for some reasons, such as lack of a coherent, long term 6

8 planning strategy; and, in some cases, the costs of site remediation. (CPRE, 2009) Home Builder Federation argues that the countryside is already over developed and house building is concreting over the countryside. Statistically, the developed greenfield residential purpose approximates to 57 % which is more than the industrial and commercial use (20%) or other developed uses (23%), primarily transport. (Bibby, 2009) But from the view of the whole country, only 10% of the total area is developed. Even in the South East (excluding London) only 12% of land is developed. As a matter of fact, only about 2.27% of England is actually built on or concreted over while the rest majority developed area is parks, gardens and natural environments. Between 2000 and 2010, only 0.13% of England s land area is utilized for new dwellings building. (HBF, 2013) The Figure indicates the trends in previously-developed land in England from 2002 to It is noticeable that the vacant or derelict brownfield area decreased from in 2002 to in 2008 and went up again to hectares in It is reported by Home and Communities Agency that according to local authorities assessment 31,160 hectares of brownfield are suitable for housing in (HCA, 2009) Since the figure for the year after 2009 was not provided, from the optimistic aspect the total area suitable for housing is assumed to be hectares in Figure Land/building type (hectares) Change from 2008 to 2009 Vacant derelict land 40,710 39,710 38,170 36,560 34,850 33,600 32,400 33,390 3% Land currently in use 25,400 26,050 25,960 26,930 27,880 28,520 31,350 28,530-9% All land types 66,110 65,760 64,130 63,490 62,730 62,130 63,750 61,920-3% 4) The household size: In general, the household size was getting smaller over the last fifty years; i.e. the number of households grows faster than the number of population. (ONS, 2011) The Figure 2.2.4(LFS, 2012) shows the trend of household size and household number in the UK. Clearly, as population grows bigger, the number of household increases from in 1996 to in 2012, which is a 11% increase during this period. The corresponding household size slightly dropped from 2.43 people per household to 2.33 people per household. Since the household size has been stable at 2.36 for several years, it is reasonable to be expected the same in future. 7

9 Figure Number of household and household size in UK ,000 26,000 25,000 24,000 23,000 22,000 Average household size (number of people) Numbers of households (in thousands) 5) The dwelling densities: From 1989, the building density was increasing from 25 dwellings per hectare to 43 dwellings per hectare in (Figure 2.2.5a) Especially for the brownfield area, new dwellings were built at an average density of 53 dwellings per hectare in 2011 compared to an average density of 33 dwellings per hectare for other previously undeveloped land. ( 2013) Figure 2.2.5a: Density of new dwellings built, England, 1989 to 2011 (dwellings per hectare) 0 On all land On Non Previously-Developed Land On Previously-Developed Land According to the proposal of policy in March 2000, the brownfield is preferred than greenfield for new house building. As the result of an increase in the density of new residential development, there is a higher proportion of new dwellings built within cities. In turns, higher propotion of new dwellings constructed within cities leads to more new houses built on brownfield cites. (Bibby and Brindley, 2006). It can be seen from the following graph, a higher dwelling density accompany with a larger proportion of brownfield (53%) occupying all land change for residential use in On the contrary, in the year (1994) with lower dwelling density, the corresponding 8

10 proportion of brownfield is 46%. Figure 2.2.5b land use change for residential use in 2001 land use chang for resedetial use in 1994 Brownfi eld 53% Agricult ure 31% Other Greenfi eld 16% Brownf ield 46% Agricul ture 37% Other Greenfi eld 17% 6) Total Income from Farming: Population growth not only directly raises the demand for food but also indirectly lead to decrease in food production through agriculture land use change. Therefore, the housing problem puts pressures on food security in some extent as well. In order to measure the costs of agriculture production reduction, the total income from farming is introduced, which represents the income that is left to farmers when all costs are deducted. It deducts interest, rent and labour costs. National statistics (2014) delivers the historical figures of total Income from farming which is estimated to have risen between 2012 and 2013 by 13% ( 630 million) in real terms, to 5.5 billion. With the trend of agricultural land area in the past decades (See Figure 2.2.6), the corresponding of income from farming per hectare can be calculated (Total income from farming divides by agricultural land area). The agriculture incomes are measured in real terms of 2013 level. Graphically, it greatly varied during 1990s and kept an upward trend after There is seems no regular pattern for the agriculture income per hectare. Therefore, the historical average income of 250 per hectare can be calculated to measure the level of agriculture production in money term. 9

11 Figure :Total income from farming in real terms (pounds per hectare) income per hectare (pound) Utilised agricultural area (thousand hectare) Analysis and discussions To start with, this report firstly looks back to 1994 and discusses the historical facts about housing problem in the past 20 years. The population of England in 1994 is estimated as million, 126,500 more than the population in Based on the assumption that population was stable since 1994, the new houses demand afterward is owing to the outstanding homeless population in The office of deputy prime minister defines the statutory homeless as those households which meet specific criteria of priority need set out in legislation, and to whom a homelessness duty has been accepted by a local authority. Rather than sleeping rough, statutory homeless people are more likely to be threatened with the loss of their current accommodation. These people become homeless mainly due to 'Parents or friends no longer able or willing to accommodate'. Some other reasons may be the end of an assured tenancy or experiencing acceptances. It is reported that the number of households accepted by the local authorities as statutory homeless in 1994 was 125,360 in England. Therefore, the number of dwellings demanded in 1994 can be approximated to the number of homeless household. As previously illustrated, the housing density in 1994 is 24 dwellings per hectare. Thus the land area needed for allocating these new houses can be calculated as the number of homeless household (125,360) divide by the dwelling density (24). i.e 5223 hectares of land are required to allocate all homeless population in In fact, there were 122,710 (see Figure 2.3.1) dwellings completed during 1994, which nearly meet the number of homeless households. Only a small gap of 2650 dwellings needs to be filled in Therefore, without any population growth all the house demand of homeless people can be satisfied by Afterwards, no more land is required for new house construction. 10

12 Figure 2.3.1: Homelessness figure and dwellings completed, England , , , ,000 50, homeless household dwellings completed household number change As a matter of fact, the number of household growth continuously increased after 1994 and breaks through 150 thousands households a year in This is due to both increasing fertility among UK-born women the number of immigrant women of childbearing age (15-44). The number of household growth reaches the peak of nearly 200 thousands in 2011, a huge 40% increase than the level of household growth in This figure drops back to 163,771 in On the contrary, the house market was hit by the financial crisis in 2008 and started to shrink afterwards. Therefore, large gaps between the number of household growth and dwellings completed give rise to severe housing shortage in the past few years. This circumstance will become worse in the future if no coping methods being taken. Compared with the situation under assumption of no population growth after 1994, the increasing population brings about the land change (both brownfield and greenfield) for residential use in the large extent. And in turn, result in the agriculture output reduction. Figure demonstrates the hectares of land use change for different types of land. The blue area shows a decreasing trend of the agriculture land area change for residential use over the past 20 years. In addition, the brown area and green area represents the corresponding figures for brownfield and other greenfield which indicate the similar trend. However, without the population change, it is not necessary to occupy greenfield for new house building. Statistically, the areas of brownfield changed for residential use in 1994 and 1995 are 2880 and 2820 hectares which can be utilized for 153,900 new dwellings construction with a density of 27 dwellings per hectare. This construction capacity is more than enough to meet the demand of 125,360 homeless households in In this way, hectares of farmland and hectares of other greenfield can be saved during 1994 to Compared to the level of 17 million hectares in 1994, total utilized agriculture area decreased by 0.146%. As the 11

13 farmland area shrinks year by year, the agriculture outputs foregone each year accumulate to 63 million pounds in total during this period. The total area of other greenfield land lost due to change for residential use is hectares. Figure 2.3.2: land area change for residential use (hectare), England, Agriculture Other Greenfield Brownfield 12

14 population (thousand) FORECASTS UP UNTIL 2051 In this section, the land areas (both greenfield and brownfield) change for residential use each year is predicted according to the population projection in the next 40 years. Based on the forecasts of land use change, the hectares of farmland area and other greenfield foregone and the number of agriculture income reduction due to new house building are calculated. 3.1 Approaches With the population projection from ONS and the estimated effects of population growth on land use and housing, the forecast model is constructed (up until 2051) by EXCEL. The relationships between the movements of variables are estimated by the regressions generated in STATA 11. The forecasts are carried out by considering three different cases. In each case, three scenarios are analysed, depending on the variant population projections the high variant population, the low variant population and the medium variant population which is also called principle population projections. Figure 3.1.1: compares the trend of each type of population projection for the next 40 years from 2012 to 2051 (the latest figure released by Office for National Statistics). The estimated population of England is 53.5 million in The Office for National Statistics (ONS) has accounted for several different variables in their projections. Under the highest projection, assuming high fertility, high life expectancy and high migration, England's population will reach 74m in 2051 while under the lowest projection, with all those factors at lower than expected levels, the population will only grow to 58m. 75,000 Figure 3.1.1: England population projection ,000 65,000 60,000 55,000 low variant population high variant projection principle projection 50,000 To start with the first case called the float forecast, the main assumptions and calculations are listed as followings: 1. Regarding the NPPF Brownfield first policy for building land, new houses are firstly considered to be built on brownfield and then greenfield. 13

15 2. The dwelling density is linearly related to the population change each year. (Appendix I) It is reasonable to consider that as the population growth accelerates the dwelling density will increase to meet the high demand. The regression result implies that a thousand increase in annual population change lead to an increase of in dwelling density. 3. The area of brownfield change for new house building varies with the dwelling density each year. The relationship between the brownfield change and the dwelling density is estimated by the linear regression results from the historical time series. (Appendix II) 4. The household size was decreasing from 2.42 people per household in 1996 to the recent 2.36 people per household. (ONS LFS, 2012) Roughly speaking, household size scales down by 0.06 during the past 16 years. Thus the number of people per household is assumed to decrease by 0.04 for every decade in the next 40 years. 5. One household needs one dwelling to live in. So the number of household change equals the number of dwellings demand. 6. If plan to build new dwellings on greenfield, 71.5% of greenfield for new housing are agricultural lands. 7. It is assumed that no more than 30,000 hectares of brownfield is available in the next 40 years. Once the brownfield suitable for housing is used up, the excess demand for housing will be filled in Greenfield area. 8. The agriculture income is assumed to be 228 pounds per hectare in the real term of 2013 level for the next 40 years. this number is the average output level in the past 30 years. 9. After the ten year plan, 2 per cent of dwellings are assumed to build within the Green Belt. Because this percent has been hold since 2004 (LUCS, 2013) The second case is based on the key assumption that dwelling density is fixed at a relatively high level of 50 dwellings per hectare. So forecast in this case is called fixed forecast in short later. The purpose of this model is to investigate how the high density construction can save the greenfield in different scenario. Other 14

16 assumptions made in this case are the same as those in the first case except the calculation of brownfield area change for residential use. With assumptions of fixed high dwelling density in the future, the brownfield change is assumed to be consistent at 1878 hectares per year after The last case refers to the plans of local authorities currently considering building in countryside and greenbelts. The main steps of establishing this model are the following: 1. One household needs one dwelling to live in. So the number of household change equals the number of dwellings demand. 2. Regarding the NPPF Brownfield first policy for building land, new houses are firstly considered to be built on brownfield and then greenfield. 3. CPRE (June 2014) revealed a significant figure that 700,000 houses will be planned in the countryside - including almost 200,000 allocated for the Green Belt. It is assumed that the 700,000 houses in countryside will be completed in a decade from 2014 to i.e 70,000 houses built in countryside and 20,000 in greenbelt per year. Therefore, during this period 70,000 of new dwellings will be allocated in the countryside every year. Then the other new houses will be constructed on the brownfield. 4. From 2014 on, the dwelling densities on greenfield and brownfield are assumed to be 53 dwellings per hectare and 33 dwellings per hectare respectively in 10 years time. 5. After the 10 years plan, the dwelling density continues to be linearly related to the population change each year and the area of brownfield change for new house building each year varies with the dwelling density. The relationship between the brownfield change and the dwelling density is again estimated by the linear regression results from the historical time series. 6. The household size is 2.36 people per household at the moment and will kept the same in the next ten years. This number is assumed to decrease by 0.04 for every decade thereafter up until If plan to build new dwellings on greenfield, 71.5% of greenfield for new housing are agricultural lands. 15

17 8. It is assumed that no more than 30,000 hectares of brownfield is available in the next 40 years. Once the brownfield suitable for housing is used up, the excess demand for housing will be filled in greenfield area. 9. The agriculture income is assumed to be 228 pounds per hectare in the real term of 2013 level for the next 40 years. The expected number of new house will be built in greenfield each year and the excess demand of new house will be met by new housing on brownfield area. 10. After the ten year plan, 2 per cent of dwellings are assumed to build within the Green Belt. 3.2 Results and discussions The models explained previously generate large time series figures, which are presented by graph in this section. Each graph forecasts how the lands change for residential use along with the population change in various scenarios. The brown line indicates how many hectares of brownfield change for residential use every year. The green area represents the numbers of greenfield changed for residential use under which the blue shaded area estimates the proportion of greenfield change that is farmland. 1) Float forecast model As the term suggests, float forecast depends on the float dwelling density, which is affected by the population growth every year. In the first case, (Figure 3.2.1) under the circumstance of low variant population the brownfield area will be used up by 2025 from when the new houses demand have to be filled in Greenfield land. Fortunately, as the population growth slows down, the number of Greenfield change for residential used reach the historical height of 3531 hectares in 2026, then gradually goes down to zero after There seems to be not much concern about agriculture land loss due to population change before However the greenfield loss, especially the agricultural area foregone, cannot be neglected afterwards. Up until 2051, hectares of agriculture land will be changed for residential use which amount to 103 million pounds of agricultural output loss in total. In addition, 1426 hectares of greenbelts will be lost in total. 16

18 Figure 3.2.1:land change for residential use with low variant population (hectare) 4,000 3,000 2,000 1,000 0 Greenfield change Agriculture change Brownfield change The circumstances under high variant population (Figure 3.2.2) And principle population projection (Figure 3.2.3) seem to be more serious. In high variant population scenario, there is a higher need for greenfield to make up the unsatisfied housing demand from brownfield after Before the brownfield used up in 2028, there is as much of the brownfield change as the farmland area turns into the residential areas. Afterwards, more than 6500 hectares of greenfield will be transferred to residential area each year. The total farmland loss due to new house building is expected to be hectares until 2050 which account for 0.53% of the total utilized farmland area in i.e. Food production capacity drop by 0.53% at the end of This land use change will lead to 230 million agriculture income loss which is more than double the amount in low variant population scenario. Moreover, 2598 hectares of greenbelts will be lost in the next 40 years in total, which is a 0.16% drop comparing to the total area of greenbelts in ,000 7,000 6,000 5,000 4,000 3,000 2,000 1,000 0 Figure 3.2.2: land change for residential use with high variant population (hectare) Greenfield change Agriculture change Brownfield change In principle population projection scenario, the area of brownfield that change for residential use will be exhausted by As the capacity of brownfield to build new 17

19 house take up more than half of the housing demand, the land change of Greenfield keeps lower than 2000 a year before Similarly, the land reduction of agriculture area continues around just over 1000 hectares a year. This figure will reach a peak of 4806 hectares of land change for residential use in 2026 and will gradually decrease year by year. Consequently, 186 million of agriculture income will be lost in total. 6,000 5,000 4,000 3,000 2,000 1,000 0 Figure 3.2.3: land change for residential use with principle population (hectare) Greenfield change Agriculture change Brownfield change 2) Fixed forecast model In the second case, land use forecast mainly based on the assumption that the average dwelling density in the following 50 years will keep at a high level of 50 dwellings per hectare. The new dwellings demanded are firstly built on the brownfield and the access housing demanded will be satisfied by building on the farmland and other greenfield. For the scenario of low variant projection, the high dwelling density enables the available brownfield to satisfy the majority of new house demand before 2021 shown in Figure After that, the new house demand will lead to no pressure on greenfield area during 2022 to2028 when the brownfield area needed for new houses decrease to zero Comparing with the same scenario in the first case, increasing the dwelling density can put off the date that brownfield is used up by 4 years. Afterward, the dwelling constructions in greenfield peak at 1246 hectares per year (among which 892 hectares land is farmland). Only 30 million pounds of agriculture production will be lost due to the housing problem. The lost of greenbelts (819 hectares) is minimized in this scenario as well 18

20 Figure 3.2.4: land change for residential use with low variant population (hectare) Greenfield change Agriculture change Brownfield change 2,000 1,500 1, The circumstance of high variant population projection in this case (Figure 3.2.5) is quite similar with that in the first case. Building new houses at the density of 50 dwellings per hectare not seems to be enough to satisfy the housing demand in the future. There will be even more area of greenfield than brownfield changing for residential use each year. And the pressure of building on countryside will become bigger if the dwelling density cannot be improved to the higher level. In 2050, approximately 3609 hectares of greenfield (2% of which are greenbelts land) are occupied a year for new house constructions and hectares of farmland will be lost due to the housing problem in the next 50 years. This number amounts to 187 million pounds of agriculture income loss in total. This may imply that, if the population growth follows the high variant projection, the new dwellings can be built at a higher density on the brownfield until 2028 in order to save more greenfield areas. Figure 3.2.5: land change for residential use with high variant population (hectare) Greenfield change Agriculture change Brownfield change 6,000 5,000 4,000 3,000 2,000 1,000 0 The housing issue mitigates if the population growth follows the principle projection. Less than 1500 hectares of greenfield is needed for housing per year before the brownfield exhausted in The number of greenfield area changing for residential use rocket to 4461 hectares a year in Then it will gradually go down 19

21 to 2281 a year by The accumulated farmland lost is hectares which amount to107 million agriculture incomes loss. Moreover, the aggregate greenbelts land change is 1630 hectares Figure land change for residential use with principle population projection (hectare) 3,500 3,000 2,500 2,000 1,500 1, Greenfield change Agriculture change Brownfield change 3) 10 years construction on greenfield model Moving on to the third case, the number of dwellings built on greenfield is set to a specified level for the next decades. As CPRE argued, the government plans to build 700,000 new houses in countryside. But CPRE has not clarified by which year this target will be met. Therefore, it is assumed that in the following ten years, new house demand will be satisfied by 70,000 new houses building in greenfield area each year and then the access demand will be satisfied by building on brownfield. After the 10 years period when this target has been hit the new house demand will be satisfied by the developed land area first and then the greenfield area. As before, once the hectares of brownfield is exhausted, new house constructions can only rely on developing greenfield. In the low variant population projection scenario (see Figure 3.2.7), 700,000 new dwellings in countryside can solve the problem of high house demand in the next ten years. After 2023, once the 700,000 target is completed, new house demand can mainly rely on the vacant brownfield site. However, the brownfield is expected to be used up by the year Thereafter, the only way to meet the house demand of growing population is to largely develop greenfield again. By the year 2051, there will be hectares of agriculture area lost due to new housing construction, which is amount to 100 million pounds agriculture output loss in total. And 7202 hectares of greenbelts (0.44% drop than the level in 2013) will be lost. 20

22 Total Figure land change for residential use with low variant population (hectare) 3,000 2,500 2,000 1,500 1, Greenfield change Algriculture change Brownfield change In the scenario of high variant population projection (see Figure 3.2.8), there is an astonishing result. Even with 70,000 new dwellings constructed in the countryside each year, there is still a high excess demand for new dwellings. Start from 2016, even more than 2500 hectares of brownfield are needed each year to satisfy the total house demand. After 2024 when all available brownfield sites are exhausted, there will be continuous demand of more than 4500 hectares greenfield every year. In this case, hectares of agriculture area are expected to be developed for residential use, a 0.46% agriculture land decrease than the total utilized agriculture area in Consequently, 197 million pounds of agriculture income and 5842 hectares of greenbelts will be foregone. 6,000 5,000 4,000 3,000 2,000 1,000 0 Figure land change for residential use with high variant population (hectare) Greenfield change Algriculture change Brownfield change At last, in the principle population projection scenario (Figure 3.2.9), the 700,000 21

23 house on greenfield still cannot solve the problem in the long run. Before the brownfield is exhausted by 2029, nearly the same amount of greenfield and brownfield are required to meet the growing population. The demand for greenfield peaks at 3917 hectares in Then as the population growth decelerates, the demand for developing Greenfield gradually fall to 3612 hectares in 2050 among which 2583 hectaress are expected to be farmland. On the whole, approximate hectaress of agriculture area will be lost. Consequently, 156 million pounds of agriculture income and 6444 hectares of greenbelts are expected to be lost in the next 40 years. 4,500 4,000 3,500 3,000 2,500 2,000 1,500 1, Figure land change for residential use with principle population projection (hectare) Greenfield change Algriculture change Brownfield change 22

24 LIMITATIONS It is important to recognise the limitations, as well as the strengths, of the analysis methods and measurements. Due to the constraints of time, there was no feedback from CPRE on the time assumption of 700,000 dwelling construction plan. However, assuming 10 years construction period with 70,000 dwellings completed every year seems to be reasonable. This is amount to utilize 2121 hectares of greenfield for new house building each year which is quite close to the historical average level of 2270 hectares of greenfield change for residential use per year. Owing to the constraints of resource, there is no historical time series data indicating the amount of green belt area changed for residential use. Only the figures of other greenfield land changed for residential use is available. In addition, the statutory homeless household figure is considered as the number of unhoused population. Because the number of people sleeping rough is difficult to work out, there are no historical estimates for this group of unhoused people. Homeless Link reports that approximately 2414 people are counted to be sleeping rough in Considering the number of people sleeping rough is only a very small portion of homeless people, they are not taken into account when calculate house demand. In the historical facts analysis part, the relationship between variables such as population growth and dwelling density are estimated by the linear regressions. Although the population growth may be the most significant factor account for the dwelling density change, there should be some other influential factors which have been omitted. For instance, the type of the accommodations, the type of the land s previous use, the technologies of construction and the location of the constructions may also affect the dwelling density. Some factors cannot even be modeled by quantitative approaches such as the relative legislations on dwelling constructions. Similarly, the movement of the brownfield change for residential use is assumed to be influenced only by the dwelling density. In fact, it should also involve in other accounts, such as land use efficiency, re-use of existing buildings and mixed use development. Therefore, simplifying the regression model by omitting these factors may lead to the bias on the estimates of brownfield change. On the whole, the predicted movement of brownfield is desirable. In the forecast analysis section, refer to the greenfield land changed for residential use, the proportion of greenfield occupied by farmland is assumed to be fixed at 71%. This fixed proportion may lead to some bias because the utilization of farmland maybe changed with the movement of agricultural market. For instance, during recession period, the area of farmland changed for residential use drop by 45% from 1190 hectares in 2007 to 650 hectares in Therefore, unexpected economic situation cannot be modeled in the forecasts. However, the forecast analysis in this report can reproduce outcome in the future by 23

25 updating the population projection figures or by substituting various dwelling densities. 24

26 CONCLUSION AND RECOMMENDATIONS The population growth brings about big pressure on house shortage in England. Compare to twenty years ago, the population has grown by 11% from thousand in 1994 to thousand in mid With the decreasing household size, the number of households in England grows even faster. However, as the population growth accelerates, the number of dwellings completed goes down. In 2012 the number of dwellings competed is 88,000 which only can meet half of the household growth (164,000). House shortage becomes more and more severe in England. However, if population can be stabilised from 1994, there will be a different story. Suppose no more population growth after 1994, the total house demand roots in 125,360 homeless household. The areas of brownfield changed for residential use in 1994 and 1995 can be utilized for 153,900 new dwellings construction with a density of 27 dwellings per hectare. Therefore, the house demand of all homeless households in 1994 would be met by the end of In this way, hectares of farmland and hectares of other greenfield can be saved during 1994 to Compared to the level of 17 million hectares in 1994, total utilized agriculture area decreased by 0.146%. The aggregate agriculture output could have been generated from these lost farmland accumulate to 63 million pounds in total during this period. From the aspect of amenity, The greenbelts can be accessed was reduced by 3610 hectares in total from 1994 to 2001 The following table compares the main findings from the forecasts model (See details in attached Excel) Table 5.1 variation of the population projection brownfield area exhausted by the year Total agricultural area change for residential use (hectares) Total agricultural income loss (million) Total greenbelt area change for residential use (hectare) Low Case 1 High principle Low Case 2 High principle Low Case 3 High principle Firstly, compare the date by which the brownfield will be exhausted, the float density forecast seems to be most worrying. Especially under the scenario of low variant population projection, because the population growth decelerates, there will be less pressure on the construction density. Thus, both brownfield and agriculture 25

27 areas may be utilized in relative low efficiency. The overall circumstances in fixed density forecast look better. The brownfield exhausted dates in 3 projection models are all extended later (or at least the same) than the dates in the first case. From the aspect of sustainable development, to utilize the brownfield more efficiently is vital significant because the area of brownfield suitable for housing is not unlimited. The plan of building 700,000 new houses in the countryside can extend the brownfield exhausted dates in the scenario of low variant and principle projection. However, if the population follows the high variant projection, the intensive development in countryside cannot solve the problem of exhausting brownfield. Even worse, with the high variant population projection, 4082 more hectares of agricultural land will be lost in third case than in the fixed high density case (68036 hectares). From the aspect of total agricultural area change for residential use, high fixed density model is superior under all population projections. Especially in low variant scenario the agricultural income loss can be reduced to just over 9 thousand hectares, which amount to 30 million pounds output loss in total from 1994 to If the population growth follows the principle projection in the next 40 years, high density construction can reduce the agriculture area change by 45% compared with the variant dwelling density case (70334) Considering the total agriculture output income, the high dwelling density leads to the least income loss in all three scenarios. This is because the high density constructions on brownfield protect the greenfield. Because once the farmland is developed for residential use, the food production could be generated from that land will be lost forever. The later the greenfield is utilized, the more food production can be saved. The last but not the least, the greenbelts land change for residential use rockets in the third case. This is mainly due to the 20,000 dwellings of annul construction on greenbelts during 10-year-plan. Under principle population projections, 6444 hectares of greenbelts will be converted into housing construction, which is nearly triple the number in the first case and fourfold the number in the second case. The figure suggests that in order to protect the greenbelts in England, All in all, to control the population growth is crucial for the future sustainable development. As our land resource is limited, the land for residential use, commercial use and agriculture use should be balanced to meet the first need of the population. A big population needs both more food production to feed and more land to live. So the best way to relieve the food shortage and house shortage is to slow down the growth of the population and to solve the problem from the roots. If the population cannot be effectively controlled, the second best way to tackle the problem is to use the land more efficiently. The Royal Commission on Environmental Pollution (2011) demonstrates that rather than anxious about the total size of the UK population, It should raise more concern about how and where people choose to live which presents the main environmental challenge from demographic change. 26

28 Polices should be made to encourage the effective use of land by re-using land that has been previously developed. For instance, reclaim the empty houses left by the people have gone. Moreover, in the urban area with high density of population, increase the number of mixed-use building can create more space for residential and commercial use. Aiming at the use of brownfield land, authorities may make policy or plan about setting constraints on the number of bungalows or lower storey buildings in urban area. 27

29 REFLECTIVE CHAPTER I was major in economics during my undergrad studies. To explore the essence of the problem and to discover the relationship between objects is a kind of art work to me. When I first time saw this topic, the implication of population and housing problem and the potential use for policy making of this project excited me very much. Population issue is a global problem which builds up pressures on various areas such as food shortage, sustainable development, biodiversity global climate change and so on. Although it is impossible to investigate all these areas in a short period studying only one aspect of the population matters in the last two months brought me a harvest. With the help of my supervisors who are kind and rich experienced,the project started smoothly. My project supervisor Mr. Roger Martin has provided a very clear outline of work which helps me to specify the aims and complete the research step by step. In addition he and my academic supervisor Mr. Jeffery Jones also suggested some useful resources to me which contributed to my later analysis. In the data collection period, I spend a lot of time on searching for the time series data of population growth, land use and unhoused population. When I searched these key words on the internet, less relative figures can be found. I solve this problem by reading the similar subject literature and tracing the original data resources from the previous work. But the data from different resources have quite different format. It is essential to recognize the high quality figures and edit them in the way that can convey my idea. For example, adding up the monthly data to get yearly data. To build up basic knowledge about housing I consulted my best friend Xin who is major in real estate. She taught me some backgrounds of land classification and recommended one of the key references to me Barker s review. In my opinion, sometime friends can be more helpful than Google, because they can provide clear and definite answer according to your need. In addition, the communication is a procedure to open up my mind throng exchange ideas from others. The biggest challenge of this report is to clarify the logic of forecast model since no previous literatures can be referenced. At first, I didn't have many ideas about how to construct a forecast model. From discussing with my supervisor and exchanging our thoughts every week, my train of thought became more and more clear. It is Mr. Jones who gave me lots of help and inspirations. He firstly suggested me to consider different cases according to various dwelling densities. In addition, he pointed out the limitation of brownfield area suitable for housing and the priority of utilizing different type of land. Then I tried to realize those ideas by constructing models in Excel. But this world is infinitely completed that we cannot capture all the changes in trends or conditions and relationships between objectives in a simple science model. Therefore, the different case and scenarios are helpful to predict the possible outcomes under various conditions. As the forecast is only an estimated outcome in 28

30 the future, the result cannot assure that specific circumstances will happen. Nevertheless, the forecast outcomes can provide the basis of corresponding coping measurement. Through this project, I learnt how to apply the academic thinking in solving practical problems. What I had learnt from Msc Management science is not only the useful technical skills to generate the result of model but also the way to analyse the issue and to think critically. As I said at the start of this section, this project can be treated as a piece of art work for a scientist. My supervisor Mr. Jones taught me not only the academic analysis skills as well as the attitudes towards science. I used to think the art work such as painting or dancing is creative and makes life more colourful. But He told me the scientist can find the essence of things by in-depth analysis and investigation which is a creative process. Just like what I have done in this project. I constructed the model through innovative thinking and scientific exploration. Learning science is just exploring a way to make life better. Therefore, the rational thinking can also bring us a colourful life. 29

31 REFERENCE Agriculture in the United Kingdom 2013, national statistics, 29 May 2014 CPRE Annual Review 2013/14, Campaign to protect rural area, June 2014 Labour Force Survey (LFS), Office for National Statistics, November 2012 Population and Household Estimates for the United Kingdom, Office for National Statistics March March 2013 Land Use Change Statistics in England: 2011, Department for Communities and Local Government, Crown copyright, 2013 Housing crisis statistics, Home Builder Federation, October 2013 R. Swann, et.al, Population growth and housing expansion in the UK: some preliminary considerations, Population Matters, January, 2013 Kate Barker, Review of Housing Supply: delivering stability: Securing our future housing needs, HM Treasury, Rough sleepers count, Homeless Link, 2014 Space and amenities, Population Matters, S.dpuf Demographic Change and the Environment, Royal Commission on Environmental Pollution, Feb 2011 Population Projections (Online), Office of National Statistics 2014, Available: Live Table on Homelessness (Online), Office of National Statistics 2014, Available: Live tables on house building (Online), Office of National Statistics 2014, Available: 30

32 Live Table on Land Use Change Statistics (Online), Office of National Statistics 2014, Available: -statistics 31

33 ABBREVIATIONS LFS ONS HBF NPPF CPRE DCLG LUCS Labour force survey Office of National Statistics Home Builder Federation National Planning Policy Framework Campaign to Protect Rural England Department for Communities and Local Government Land Use Change Statistics 32

34 APPENDIX Appendix I Regression result of average dwelling density on population growth. Appendix II Regression result of brownfield area change for residential use on dwelling density. Appendix III Details about Excel Model, please see in the attached CD 33