081a ak cncl - general - capacity and feasibility modelling - residential capacity results,...

Upload: ben-ross

Post on 13-Apr-2018

215 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    1/53

    ______________________________________________________________________________________

    Residential Capacity (Report of the 013EG, Round 3) i

    Residential Capacity

    Results, Methodology and Assumptions

    Produced by Topic 013 Urban Growth Expert Conferencing Group,

    At the request of the Auckland Unitary Plan Independent Hearings Panel

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    2/53

    ______________________________________________________________________________________

    Residential Capacity (Report of the 013EG, Round 3) ii

    Expert Conferencing Group (Urban Growth 013) Round 3 Participants

    RoleName Provided Evidence on behalf of Topic 013

    Party/Submitter (from Company):

    FacilitatorMr David HillReagan Solomon*

    IHPAuckland Council (RIMU)

    MembersKyle BalderstonDr Doug FairgrayMurray CameronPatrick FonteinAdam ThompsonDavid Gibbs

    Dr Richard BurtonRobert PhilpottPhil OsborneFraser Colegrave /Steve Hoskins*

    Auckland Council (RIMU)Auckland Council (Market Economics)Auckland Council (S&IS)Patrick Fontein (SD4)Property Council (Urbecon)NZIA/Generation Zero (Construkt)

    Auckland 2040Party (McDermott Miller)Housing NZ (Property Economics)Party (Insight Economics)

    ObserversSarah HoldemDavid Hermans

    MfEMBIE

    *AlternatesIndicated with a *, these persons attended in place of the primarymember when absent.

    This is the third round of Expert Conferencing ordered by the IHP involving this broader group.

    Auckland Council staff involved in the process would like to thank all of the participants in the

    group (and their clients) for being so generous with their time and expertise, and Mr David Hill for

    his facilitation.

    Expert conferencings purpose is not to reach consensus on all matters but to narrow the areas of

    disagreement. It is fair to say that there remain areas of disagreement both large and small, that

    are articulated in the body of this report.

    After three intensive rounds of conferencing the group maintains the view that the feasibility model

    is a good decision support tool.

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    3/53

    ______________________________________________________________________________________

    Residential Capacity (Report of the 013EG, Round 3) iii

    Executive Summary

    The Auckland Unitary Plan Independent Hearings Panel (IHP) in its memo of 5 November 2015

    (Enclosed in full at Appendix 1) requested the 013 Expert Group reconvene to:

    .. to develop and agree a revised method to estimate supply that is capable of

    accommodating the changes to density rules as proposed by the Council and its

    forthcoming proposed changes to the spatial application of residential zones.

    The estimates of supply are to be based on the Councils proposed zone rules as

    presented to the Panel in Topic 059/60/62/63 and as modified in its closing comments on

    those Topics, and using the spatial application of those zones that are proposed by Council

    for the 081 Topic.

    The Panel requests that the output from this work include:

    An estimate of demand for residential capacity (expressed as dwelling

    numbers) throughout the Auckland region, for the period to 2026 and to

    2041.

    An estimate of overall supply of residential capacity (in tabular form along

    the lines provided by Dr Fairgray on 28 October) and a breakdown of that

    supply in terms of each type of zone (e.g. Centres, THAB, MHU, Mixed Use,

    etc.).

    An estimate of the years of supply that would be enabled and available ifthe Plan is made operative in 2016, and at 2026, and the trend in that metric

    in the intervening years.

    Presentation of supply in the form of heat maps or similar (e.g. as at 2026

    and 2041) to enable easy visualisation of the spatial spread of supply.

    A supporting document that sets out the method used to estimate demand

    and supply, with any significant diverging views and their implications for the

    estimates recorded in an appendix (along the lines of the approach used in

    the July report)

    The Panel intends to use the outputs from this work to assess the extent to which the

    proposed Plan would be adequate to meet forecast demand for residential capacity for the

    period to 2026, and to 2041.

    The Panel requests the above outputs be provided to the Panel by 26 January 2016 (which

    is when the Council evidence on Topic 081 is due).

    This report summarises the findings of the 013 Expert Group, including divergent views.

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    4/53

    ______________________________________________________________________________________

    Residential Capacity (Report of the 013EG, Round 3) iv

    Table of Contents

    Executive Summary ........................................................................................................................ iii

    Table of Contents........................................................................................................................... iv

    1.0

    Introduction .......................................................................................................................... 1

    Why .................................................................................................................................. 1

    1.

    Scope ............................................................................................................................... 1

    1.2

    Outputs ............................................................................................................................. 2

    1.3

    Project limitations .............................................................................................................. 2

    1.4

    2.0

    Residential Demand ............................................................................................................. 4

    3.0

    Residential Supply .............................................................................................................. 11

    4.0

    How Residential Supply and Demand have been estimated ............................................... 25

    5.0 Areas of agreement and disagreement47

    Panel request for demand and supply estimates

    ................................................. 48

    Appendix A:

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    5/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 1

    1.0 Introduction

    Purpose1.1

    Consideration of the ability of the planning system to provide for Aucklands projected population

    growth needs via enabling sufficient capacity that is both commercially viable is a key concern to

    many. The IHP has directed a group of experts to examine housing supply and demand issues (as

    per their Memo dated November 5th, 2015). This report outlines the findings of this work to date.

    The modelling undertaken by the expert group (E.G. hereafter) is assessing whether the zoning

    being assessed, enables a commercially prudent development market to deliver the quantum and

    type of housing at a price that meets the expected needs of present and future Aucklanders?

    Making these assessments requires assumptions to be made about:

    - How the zoning patterns might translate into allowable development opportunities

    (enabled capacity);- How those enabled development opportunities might translate into actual

    developments (feasible capacity);

    - The quantum of demand expected over time (population growth and make-up)

    - How prices of existing dwellings and the costs to develop new ones vary over time;

    - The ability to pay (incomes) of current and future households (enabling a comparison

    to be made between feasible capacity and anticipated demand) along with some

    limited consideration of assumed household preferences in terms of type and price;

    - The implication of properties being developed to partial capacity in reflection of

    developers preferences abilities.

    Scope1.2

    The instructions of the IHP, indicated that the E.G. report should include supply outputs from the

    Councils final position on the spatial zoning pattern (to be presented in Topic 081) and revised rule

    position (established in closing legal submissions in the relevant Residential, Business and Rural

    Provisions hearings).

    Given Councils decision on out of scope matters, the 081 spatial zoning patterns has not been

    modelled. The results included in this report, including the analysis of supply and demandinteractions by Adam Thompson and Doug Fairgray utilise ACDCv3 run 3.6 which references the

    PAUP spatial zone and the ACAP residential provision. A further run 3.7, which addresses an

    issue with the Apartment typology, has occurred. The results of that modelling is included in the

    evidence of Doug Fairgray and Kyle Balderston however due to time and resourcing constraints it

    is not incorporated these into this report. That said this report covers key methodologies for

    determining supply, demand and how this has been developed by the E.G. It also includes key

    areas of agreement and disagreement amongst E.G. members.

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    6/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 2

    The rules and spatial data patterns modelled are shown in

    Table 1 below.

    Table 1: Modelling Parameters

    INPUT DATA TYPE THIS REPORT

    Spatial Zoning Pattern (incl Precints and Overlays)PAUP (As notified)

    Residential Rules (incl modelled precincts and

    Overlays)

    ACAP (059-063 Closing Legal Submissions Track

    Change Version)

    Business Rules (incl Precincts and Overlays)PAUP (as notified)

    This report provides a description of the processutilised in development of the methodby which

    supply may be determined.

    Outputs1.3

    Outputs of the groups work takes several forms including;

    - The establishment of methods to estimate supply of feasible development from

    enabled development, to estimate dwelling demand from population based forecastsand to compare supply and demand.

    - Tabular and spatial outputs of the feasible development model.

    - The qualitative analysis contained within this report (about the modelling itself but

    also what the modelling and analysis might mean regarding the balance between

    supply and demand);

    - Assessment of demand for housing

    - Assessment of the likely overall capacity and demand outcomes

    Project assumptions and limitations1.4

    The modelling work is based on a series of key assumptions. Firstly it is assumed that the planning

    systems remains in place, unaltered for the analysis period. This is a necessary assumption to

    make as we cannot predict future planning systems however it is clear that planning is never static

    and changes to Aucklands planning framework will occur in the future.

    All modelling is a necessary simplification of the real world, using reasonable assumptions about

    human behaviour and observed phenomenon. Typically this involves utilising the average

    response or behaviour and accordingly results represent the average outcome this approach

    necessary precludes the wide range of potential outcomes either side of the average.

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    7/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 3

    Both supply and demand are generated exogenously, that is from external sources without

    reference to the other. The approach to supply and demand is to effectively assume that demand

    is fixed, via a high level population projection, and that supply must be generated that matches

    demand, such that the demand fits into the city that the plan facilitates . However, supply and

    demand are interdependent and interrelated in complex ways. This is alluded to in the detailed

    commentary, but has not been modelledas an interactive system over the full period of analysis.

    A number of potentially relevant constraining factors have not been considered. The model is

    largely urban focussed, and tests development feasibility on the basis of average costs and

    prices1, and presumes that sufficient infrastructure capacity exists (or will exist) to service what is

    zone enabled and feasible some members have highlighted that in their experience, this is not

    always the case and infrastructure limitations have precluded otherwise feasible developments.

    1There is disagreement amongst some E.G. members on the sale price information used in the model. The nature of issue is described

    in more detail in section 5.

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    8/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 4

    2.0 Residential Demand

    The Panel requests that the output from this work include:

    An estimate of demand for residential capacity (expressed as dwelling numbers)

    throughout the Auckland region, for the period to 2026 and to 2041.

    Population projections2.1

    The E.G. is in general agreement that Statistics New Zealand (SNZ) population and household

    projections for the Auckland Region are the most appropriate basis for determining long term

    underlying demand for residential capacity (see table 1). The group is of the view that the SNZ

    Medium and High population projections form the lower and upper bounds of the expected future

    housing demand. The E.G. notes that Auckland Plan has been designed around enabling the

    successful accommodation of a high growth which has the benefit of facilitating a high, medium or

    low growth outcomes with minimum downsides2. If a lower than high growth rate eventuates, the

    development market will simply not utilise the capacity enabled (as the demand does not exist),

    and the process of additional supply being enabled will assist in reducing constraint related price

    increases irrespective of growth rates.

    The planning system must as a first principle, demonstrate that the accommodation of the

    exogenous unconstrained projections is reasonably able to be accommodated. The planning

    framework can mould the spatial distributionof projected growth (towards locations most suitable

    and away from locations less suitable) but should not seek to limit it 3. The projections should be

    taken as an indication of the scale and rate of change that might occur without constraint.

    The overall demand from population projections is largely taken as a given. It is the supply side

    (based on the planning system that controls the type and amount of housing) that is adjustable.

    However, supply and demand are dynamic and will adjust or shift in response to various factors.

    There is no real deviation from the E.G. general position on this as articulated by Professor

    Bedford in the first report of this group on population projections, other than to note, that

    projections are not inevitable; if Auckland does not provide all the necessary conditions to

    accommodate this growth (both attractiveness and affordability) then it may not occur in the way

    projected4.

    2The main potential downside is making uncertainty explicit for infrastructure and service providershowever this is largely a when not

    if issue as a fixation on a single figure simply hides the uncertainty already implicit in a single numbera medium projection is just aslower than high growth rate, the high figure (usually) being achieved slightly later.

    3The SNZ projections are themselves based on assumptions about the relative attractiveness of Auckland relative to elsewhere in New

    Zealand and assumptions about its demographic makeup these assumptions are not immutable and the planning system, in the longrun can have an influence on these factors, but not without serious implications. Therefore accommodation of the projected populationas a given but with some flexibility in terms of location, typology and built form (which will influence relative attractiveness, amenity andaffordability as well as relationships with transport and employment) are within the RMAs sustainable management framework are theoptions available to the planning system.

    4Statistics New Zealand have on their website technical material outlining how their population projections for the nation and region are

    produced.

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    9/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 5

    Household projections2.2

    Population growth represents the net sum of future individuals, but people tend to group

    themselves together in their living arrangements, reflecting various family, friendship or shared

    residential arrangements (called households). These can be projected from the projected

    population based on assumptions about demographics, family status, income and ethnicity, (all ofwhich may influence the propensity for individuals to be found in household types, based on past

    observations) where each of those forecast householdsis simply presumed to represent demand

    for a dwellingunder unconstrained ideal conditions5.

    Statistics New Zealand has released its updated household projections from the 2013 base, to

    complement its previously released population projections. These are summarised in Table 2.

    Table 2: Statistics NZ Population and Household Projections 2013-2041.

    The population numbers have previously been reported on6. The table above presents the updated

    household projections, and the projected increase in household numbers to 2026 and 2041. Thelatest SNZ projections use 5-yearly inter-censal periods from 2013 (ie 2013 to 2043). For

    consistency with the previous projections, these have been interpolated to provide projections for

    the periods previously reported (2016, 2021, 2026, 2031 and 2041). Note that the household

    projections released by SNZ extend only to 2038, even though the population projections extend to

    2043. In order to estimate the 2041 household numbers, the projected mean household size has

    been extrapolated from 2038 to 2041, and applied to the 2041 population figures to estimate

    household numbers.

    The household projections show some difference from the previous estimates, as follows:

    a. The base number for 2013 is 498,000 households, some 24,000 (4.6%) fewer than the

    earlier projection base (522,000);

    b. The medium projection shows fewer households by 2026 (639,000 compared with

    669,000 previously), and less growth (141,000 households compared with 147,000). The

    medium projection also shows fewer households by 2041 (778,000 compared with

    5

    Statistics New Zealand provide a detailed overview of their household projection methodology on their website.6Fairgray EIC Topic 059-063, Table 5.1, p38.

    Population (000) 2013 2016 2021 2026 2031 2041 2013-26 2026-41 2013-41

    High 1,597 1,759 1,913 2,069 2,372 420 459 879Medium 1,493 1,583 1,718 1,841 1,962 2,186 348 345 693

    Low 1,569 1,677 1,767 1,853 2,001 274 234 508Source: Statistics NZ 2015

    Households (000)

    High 542 613 680 747 878 182 198 380

    Medium 498 534 589 639 688 778 141 139 280

    Low 525 565 598 629 678 100 80 180Source: Statistics NZ 2015

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    10/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 6

    820,000 previously), and less growth to 2041 (280,000 households compared with

    298,000);

    c. The high projection also shows fewer households by 2026 (680,000 compared with

    704,000 previously), and fewer households by 2041 (878,000 compared with 902,000

    previously). However, the net growth over the period is the same as previously

    estimated; 182,000 from 2013 to 2016, and 380,000 more households to 2041.

    In the context of the medium and long term planning for Auckland, the differences from the

    previous estimates are relatively small. The projected growth by 2026 is only 6,000 households (-

    4%, approximately -500 pa) fewer than previously in the medium projection, and 18,000 fewer (-

    6%, approximately -650pa) by 2041. The projected increase is unchanged for the high projection.

    In addition, the re-basing of the SNZ household projections captures to a degree the changes in

    household formation and structure observed at the time of the 2013 Census. The final estimate of

    498,000 households as at June 2013 compares with the Census night count of 466,800. The

    difference between from Census night and June 30 estimate for 2013 (+6.7%) is broadly in linewith differences at earlier Censuses (1996 + 6.7%; 2001 +9.0%; 2006 +8.4%).

    The final 2013 population figure is below the SNZ Low projection for 2013 (interpolated at

    1,504,000 from the 2006 base projection series). The household total is also below the Low

    household projection. Adjusting for the lower population outcome and the Low projected mean

    household size (2.892), the difference is -18,270. That is, for the population size the expected

    household count would be 516,270 rather than 498,000.

    This difference is one broad indicator of the dwelling shortfall as at 2013, particularly the influence

    of economic conditions on household numbers tight economic conditions and relatively highhousing costs generally likely to have a suppressing effect on household formation. This is picked

    up in Section 4 (below).

    2.2.1 Households by Type and Income

    The SNZ household projections include estimates of each broad type of household, covering

    couples without children, 2-parent and 1-parent families within the family category, together with

    other multi-person households, and single persons. Aucklands household structure for 2013, and

    projected to 2026 and 2041 in the medium and high projections, is shown in Figure 1. The figure is

    most notable for the limited change apparent in household structure out to 2041. This confirms that

    the main feature of household growth is the quantum of that growth, rather than changes in the

    nature of households over the period.

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    11/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 7

    Figure 1: Auckland Households by Type 2013-2041

    Source Statistics NZ 2015

    SNZ does not provide projections of future household income levels. Estimates of household

    numbers in each income band and by type of household have been sourced from Market

    Economics Ltd household model. This model estimates numbers of households of each type and

    in each age band (42 specific combinations), according to the projected demographics of the

    resident population, using the SNZ series. The current (2013) income structure for each household

    type is applied to the projected household numbers, on the basis that population and household

    structures are closely related to household income levels. This provides a basis for estimating

    future household numbers by type and income band.

    The projected size and structure of the future household market are summarised in Tables 2 to 6.

    These show for the medium and high growth futures the projected numbers of households as at

    2026 and 2041, by main household type and income band. The use of household income

    information7 is important as income is a key determinant of that households ability to pay for

    housing income, together with its accumulated wealth, of which historically, a substantial portion is

    its equity in a dwelling8). The ACDC Model provides detail on the expected cost of viable housing

    (as well as its type and size) which is enabled by the Plan. This facilitates (relatively) simple

    matching of demand - reflected by a households ability to pay - and potentially size - based on the

    number and nature of household members.

    7 Note that this approach is not integrated economic modelling the population growth and HH projections are exogenous and

    assumed to be a given, just as the incomes of our projected HH are also taken as a given. We are effectively testing to see if the HHprojected by SNZ can indeed be accommodated by the planning system, rather than forecasting the implications of them not. HHincomes are a function of the broader economic situation including the affordability of housing, and its relative supply in a situationwhere house prices are rising fast this will have a positive effect on some households incomes driving increased consumption and soon, but a negative effect on others. The key point is that these complex interrelationships and feedback loops are not considered in thisreport, but is a potential area for further research.

    8This has largely been the situation in NZ, however as Mr Thompson notes, information from the US suggests this cannot always be

    assumed to exist into the future.

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    12/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 8

    Table 3: Auckland Household Growth to 2026 by Type and Income (Medium)

    Table 3 shows that projected household is somewhat skewed toward the lower income rangeespecially for single persons and couples (including older age groups) and also the higher income

    range especially among couples and households with children. The middle income bands

    account for proportionately lower shares of growth.

    By household type, there is substantial growth in single persons (26.5% of the total), but

    substantial shares of growth are by couples (28.6%) and families with children (36% in

    combination). Nevertheless, there is growth across all household types and income ranges, again

    reflecting the increase in quantum as distinct from changes in the structure of the household

    sector.

    Table 4: Auckland Household Growth to 2026 by Type and Income (High)

    A similar pattern is evident in Table 4, albeit for the larger volume of growth assumed in the high

    future. Figure 2 shows the overall structure of the Auckland household market in 2026 (medium

    and high futures). This total structure is important because it relates directly to the overall structure

    of demand for housing into the medium, including numbers of smaller and larger household types,

    as well as income ranges.

    INCOME BANDSingle

    PersonCouple

    Couple

    with 1-2

    Chn

    Couple

    with 3+ Chn

    1 Parent

    Family

    Multi-

    Family

    Non-

    FamilyTOTAL

    Q1 (

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    13/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 9

    Figure 2: Structure of Auckland Household Market 2026 : Medium and High Growth

    The longer term picture is similar to the medium term, albeit for a larger quantum of household

    growth (Table 5). To 2041, single persons account for a slightly larger share of total growth, as do

    couples, with slightly smaller shares for families with children. Nevertheless, there is limited change

    in the structure of growth in the period beyond 2026.

    Table 5: Auckland Household Growth to 2041 by Type and Income (Medium)

    Table 6 shows a similar pattern for the high growth future out to 2041.

    Table 6: Auckland Household Growth to 2041 by Type and Income (High)

    INCOME BANDSingle

    PersonCouple

    Couple

    with 1-2

    Chn

    Couple

    with 3+ Chn

    1 Parent

    Family

    Multi-

    Family

    Non-

    FamilyTOTAL

    Q1 (

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    14/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 10

    The net increase in households by type provides some guidance as to the nature of demand for

    dwellings. As a generality, single person households and couples require less space than othertypes of households. This often does not translate through to occupancy and/or ownership of

    smaller dwellings, especially as many mature and older couples and single persons opt to remain

    in their family homes. Nevertheless, some 28-30% of the projected net increase in households is

    single persons and couples in the low and medium-low income brackets. By 2026, these segments

    will account for about 20-22% of total demand.

    There is also substantial growth in medium to high income single person and couple households.

    These demand patterns are to a degree captured in the Housing Wed Choose research, which

    indicates preference for detached dwellings (52%), attached dwellings (20%), low-medium rise

    apartments (walk up) 19% and higher apartments (9%).

    2.2.2 Dwelling shortfalls

    The E.G. is of the view that the estimate of total dwellings demanded must include those unbuilt

    dwellings which generate Aucklands housing shortfall. There are many anecdotal examples of the

    negative effects of housing undersupply as well as less extreme effects including delays in

    household formation and family starts. The exact size of the existing shortfall is difficult to

    establish, as there are many methods to do so. However most of them compare the difference

    between an unconstrained forecast, and the current situationthe difference being the shortfall.Mr Thompsonsand Dr Fairgrays estimates of the shortfall are included in Section 4.

    INCOME BANDSingle

    PersonCouple

    Couple

    with 1-2

    Chn

    Couple

    with 3+ Chn

    1 Parent

    Family

    Multi-

    Family

    Non-

    FamilyTOTAL

    Q1 (

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    15/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 11

    3.0 Residential Commercially Feasible Supply

    The Panel requests that the output from this work include:

    An estimate of overall supply of residential capacity (in tabular form along the lines

    provided by Dr Fairgray on 28 October) and a breakdown of that supply in terms of

    each type of zone (e.g. Centres, THAB, MHU, Mixed Use, etc.).

    An estimate of the years of supply that would be enabled and available if the Plan is

    made operative in 2016, and at 2026, and the trend in that metric in the intervening

    years.

    Presentation of supply in the form of heat maps or similar (e.g. as at 2026 and 2041)

    to enable easy visualisation of the spatial spread of supply.

    Notes on the Panels Request:

    - The majority of supply determination (in the urban area) is an output from the ACDC

    Model;

    - The supply information reported here is from ACDCv3.6 and not ACDCv3.7;

    - Supply and demand has not been matched spatially over time, only in aggregate

    (regionally). Regional supply (as at today) can be mapped;

    - years of supply requires two main assumptions that supply is fixed (i.e. the PAUP

    never changes) and demand occurs at the initially projected rate irrespective of supply.

    3.1.1 Feasible Capacity is Unconstrained Supply

    As for the population projections, supply from the ACDC Model is unconstrained by the realities of

    supply and demand interactions. The ACDC model assesses each site individually, without

    knowledge or concern for the results from other assessments, and costs and price assumptions

    (implicitly incorporating currentsupply and demand interactions) calculates feasibility.

    In effect individual site assessment are undertaken for commercial feasibility however when

    considered as a whole there may be implications for costs and prices if all of the supply indicated

    were to be realised. The interactions of unconstrained supply and demand are considered in

    Section 4.

    3.1.2 Further Development of the Feasibility Model(ACDC Version 3)

    Previous work by the 013 E.G worked on methods to determine supply, essentially filtering the

    enabled capacity from the Capacity for Growth Study Model (CfGS) via a further Auckland

    Development Capacity Model (ACDC) that has attempted to replicate commercial development

    decisions about how much the enabled development would cost to build, and how much it might

    sell for. If the difference between costs and price is sufficiently positive, the enabled capacity is

    considered feasible. Each round of conferencing has further iterated and refined this process

    making direct comparisons between the outputs of different model versions difficult. Each versionof the model is however fully functional and different outputs from those model versions are

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    16/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 12

    comparable between themselves (e.g. Version 2 outputs are directly comparable with Version 2

    but not version 1 or 3)

    Residential Supply is largely calculated in a three step process.

    1. Firstly, the proposed zoning, precincts and overlay rules, spatial locations, and cadastral

    pattern is fed into the Capacity for Growth Study (CfGS) model which (based on manualinterpretation of how all this fits together) calculates the plan enabled capacity and

    provides much of the necessary base data for the next stage.

    2. Secondly, the results of the CfGS are fed into the Auckland Council Development Capacity

    (ACDC) model which calculates if these enabled opportunities are commercially feasible.

    3. Lastly, as the CfGS does not calculate capacity on all sites in the region9, and the ACDC

    model either excludes some sites10 or assesses them as normal when they are not11,

    some manual accounting must be made to provide a picture of the overall supply ofresidential capacity from publically available sources or estimates.

    The E.G. has spent considerable time refining the ACDC model. The model is now at Version 3

    which determines commercially feasible supply. The next section outlines the main changes made

    to the ACDCv3 Model from ACDCv2.

    3.1.2.1 Description of major changes made to model

    The key changes made to Version 3 from Version 2 are summarised below:

    A range of development typologies is tested on every site these are small,medium and large sized, houses, terraces and apartments, (9 in total) tested as

    both infill and redevelopment (as appropriate, resulting in up to 18 developments

    per site)this compares with both previous model versions (1 and 2) where a only

    single optimised development was tested as infill or redevelopment;

    The scale of the typologies tested are controlled by the most binding of the sites

    zone controls and practicality i.e. 6 storey apartments are not tested in the SHZ,

    rather 2 storey big house developments with a dwelling up and downstairs (as the

    zone controls are binding). Conversely, in the THAB zone and above, House

    typologies are limited to 3 storeys and need at least 200m2 of land, even though theheight limits and density controls are more enabling (in this case practicability is

    binding, going taller or smaller would push the effective development typology

    9 Rural capacity, expected supply from the Future Urban Zone, and various un-modelled precincts and special areas in particular.

    10Housing New Zealand, and sites zoned for residential use but currently used for activities such as churches, schools and cemeteries

    including undesignated uses.

    11

    SHAs are in the ACDC model but are developed and tested as if the SHA does not exist resulting in results considerably lower thanwhat the SHAs are either delivering or expected to deliver based on developers reports. The difference in result is largely due to scaletreating the SHA as a single development site (as is the practice in the SHAs) or as individual sites (as per the modelling)

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    17/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 13

    towards a terrace or apartment with different cost, price and yield characteristics

    which are already tested in those typologies).

    Input dwelling floor areas have been calibrated to the ceiling prices to ensure the

    sale price of the dwellings are at or below the ceiling prices (i.e. all tested

    developments are sellable);

    Ceiling prices and sizes, and the sales location category have been reviewedagainst a range of data sources and have been confirmed as reasonable;

    From these development options that are feasible (if any), a single development is

    chosen to report a range of scenarios are possible depending on the choosing

    method applied (e.g. a focus on profitability gives a different mix to a focus on

    affordability) providing a feasible supply range, better enabling consideration of

    demand and demonstrating the flexibility within the enabled supply.

    The amendments made to the model have generally improved the reliability and utility of its

    outputs. However it has also introduced significant added complexity and detail to the input

    assumptions calculations and output data.

    3.1.2.2 Why test multiple developments per site?

    A number of members questioned if testing per site single development option was appropriate,

    when, under the PAUP, a range of development opportunities were enabled (i.e. on sites enabling

    4 storey apartments it is still possible to build single storey houses, or two storey terraces, which

    may be more profitable than the maximum outcome). E.G. members also queried whether the

    single optimised development tested was truly optimised, and the testing of a range of sizes

    would also be an improvement enabling the model to find the sweet spot rather than relying onexperience and professional judgement. E.G. members also advocated that a very wide range of

    development be tested in very fine size/price increments to enable the production of a dataset that

    represented the very lowest cost feasible dwelling possible per site12to better appreciate the Plans

    ability to deliver affordable dwellings.

    These concerns have been taken into account and balanced against practical concerns of coming

    up with the multiple assumptions required for multiple developments and dealing with the resulting

    data outputs has resulted in the development of the approach developed.

    The agreed approach is to now test a range of dwelling typologies (House, Terrace and

    Apartments) at a range of different sizes (small, medium and large) resulting in a minimum of 9

    different developments in total, all within (at least as far as possible) the twin limits set by the

    relevant zoning parameters and practicality. This may result in none or many of those

    development options being feasible. It also provides for a wider range of potential outcomes across

    sites, neighbourhoods and the region as a whole.

    12See Adam Thompson and Patrick FonteinRe:ACDC15 Model Commissioning and Application, 12.10.2015 memo to IHP.

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    18/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 14

    3.1.2.2.1 Zone Controls vs Practicality

    Once the determination was made to apply a multiple development approach, it was realised that

    establishing the limits on development under each different development scenario would be

    required. This is because apartments can be established at far higher densities than houses, and

    small houses can be established at greater densities than large ones, but various rules would

    preclude high densities from being achieved particularly in lower density zones.

    This was resolved by establishing practical limits for the intensity for each typology in each zone,

    alongside the various zoning limits, and the model develops only up to the most restrictive of the

    twin limitations. The effect of this can be illustrated from the (pre-feasibility testing) modelling of the

    modelling of medium houses, terraces and apartments density (average site area per dwelling by

    zone) as shown in 3 below:

    Figure 3: Average Land Area per Dwelling from unfiltered Capacity recalculations in ACDCv3

    As can be seen, in the zones where density is binding, all typologies have the same density (zone

    rules are binding, especially on the more intensive typologies, terrace and apartments). In the more

    enabling zones, practicality binds on the lower density typologies (houses and terraces) as they

    reach their practical limits. The effect of dwelling floor area on the average amount of land

    required to accommodate them is also discernible particularly for houses in the MHS and

    Apartments in THAB for quite different reasons.

    In MHS, the combination of building coverage and storeys (along with outdoor space and parking

    expectations) serve to require more land to fit in more floorspace. In THAB, the building envelope

    is also the controlling factor, but the amount of apartments that can fit in the envelope is more afunction of their size (while still being limited by design criteria but to a lesser degree).

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    19/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 15

    3.1.2.2.2 Typology Definitions

    Some debate was had regarding the testing of apartments and terraces in zones such as Single

    House, Large Lot and Rural and Coastal, where the rules are considered to be designed to

    preclude such developments. While it may certainly be the intention of the zoning to preclude the

    intensive or high rise form that these typologies are usually associated with, the definition of the

    typologies (being varying forms of physical (de)attachment) could potentially be realised within the

    low density constraints of the zoning under certain circumstances, and provided the site has

    sufficient room for establishment of more than a single house, an attachedarrangement may be

    possible, at least in theory.

    However, in most instances in these lower density zones, the differentiation is relatively nominal as

    capacity values are low (1 maybe two dwellings per site)

    capacity values are common to all typologies (as density controls are binding in these

    zones you can only have 1 dwelling per

    irrespective of built form)

    the build costs are greater than for an equivalent house and sale prices are lower than an

    equivalent house

    result in only rare instances where the typology is feasible at all, and is never more

    profitable than a house.

    Table 7 below outlines the possible combinations and outlines the definitions further:

    Table 7: Typology Definition Matrix (Text)

    Typology Definition Attached

    May

    Share

    Walls

    MayShare

    Floor/

    Ceiling

    House

    DetachedNot attached to any other dwelling,

    direct access to separate ground level outdoor

    space.

    No No No

    Terrace

    Horizontally attached to at least one other

    dwelling, direct access to separate ground level

    outdoor space.

    Yes Yes No

    Apartment

    Vertically attached to at least one other

    dwelling, may not have direct access to

    ground level or Table 2: Typology Definition

    Matrix (Text)

    outdoor space

    Yes Yes Yes

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    20/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 16

    Figure 4: Development Typology Matrix (Visual)

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    21/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 17

    3.1.2.2.3 Second Dwelling Conversions

    The modelling of dwellings in the CfGS and ACDC Models does not consider the potential for

    additional dwellings to be created via second dwelling conversions which is explicitly enabled in all

    residentially zoned sites with an existing dwelling excepting THAB. By our definition these

    converted dwellings would become Terrace or Apartment typologies (as the second dwelling must

    adjoin the primary dwelling).

    A significant amount of further work would be required to establish the costs of conversion (which

    would depend on many factors including the age and form of the existing structure and how the

    conversion might be undertaken) and as the dwellings are not able to be legally separated, cannot

    be sold in the same way as is presently considered by the ACDC model, requiring a consideration

    of the difference in sale price and/or rental yield from a single dwelling, and the rent (or increased

    sale price, based on retail yield) from a converted dwelling.

    Due to the wide range of potential solutions to a conversion problem (only a minimum size for

    each dwelling is specified, and may involve the addition of completely new floorspace, notnecessarily subdivision of existing floorspace) in any given specific existing dwelling situation, the

    wide variety of existing dwelling forms and structural details, this results in a very wide variability of

    potential costs of conversion and variation in potential returns. This practical issue compounded by

    the highly personalised decision making process for owner occupiers to add (usually) rental

    accommodation to their dwelling would make any modelling relatively indicative.

    This is an area for further work, but the potential for dwelling conversions to act in contradictory

    ways to both potentially preclude comprehensive (re)development by way of increased

    improvement value and/or return to existing owners form the existing improvements, and

    conversely enable a significant source of additional affordable dwelling supply, should be noted.

    3.1.2.3 Calibration of Price Ceilings and Sales Locations

    One of the key issues raised early on was the issue of whether the initial assumptions, largely

    developed for ACDCv2 were appropriate. In ACDCv2 Sales locations (10 categories of location)

    were established on the basis on sales information from Auckland Councils Sales Record Audit

    File, of recent standalone dwellings average price within a 2013 CAU geography. A range of other

    factorings, and size data was also established on the basis of the sale location classifications.

    In setting up ACDCv3, data was purchased from CoreLogic to review and update these Sales

    Location Classifications and the further refinements for each dwelling typology as initially proposed

    by Mr Fontein.

    In summary the initial assumptions proposed by Mr Fontein were found to be reasonable with the

    exception of the suggested floor area of some developments, which if sold for the $/m2 rate (which

    was confirmed as generally consistent with the data) would exceed the (confirmed) retail price for

    that typology in that location. The Sales Location categories remain as per ACDCv1 and V2.

    Using the set ceiling prices multiplied by a factor (e.g. if the standard dwelling is $800k, and our

    typology is a large terrace that sells for 70% of that (or $560,000), the largest possible dwelling of

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    22/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 18

    the typology was able to be calculated. Only those that were above the ceiling price were adjusted,

    the other initial size settings provided by Mr Fontein were maintained.

    The calculation is shown below, using an example of a terrace assuming a Sale Location Ceiling

    Price of a standard standalone dwelling of $800,000, a typology sale price factor of 0.7, and at a

    sale price per m2 rate of $3000m, then the maximum floor area the terrace can be to come in at or

    below our factored ceiling is ~186m as shown below:

    Maximum Saleable Floor area = ROUNDDOWN((800000*0.7)/3000=186.66),0)

    Some pertinent examples of the calibration process are shown below, where the LUT input

    parameters were being compared to data from long run building consent information averaged by

    inferred building type and sales location.

    Figure x illustrates the relationship between observed attached (equivalent of Terrace and

    Apartments) development size and the averaged size range in the LUTs. Size increases slowly as

    the location value increases (with some fluctuations in observed values due to low numbers, but

    LUT values are well aligned with the general trend).

    Figure 5: Attached (Terrace and Apartment) input sizes vs Building Consents

    .Figure illustrates the relationship between observed Detached (aligns to House) development size

    and the average size range in the LUTs. The relationship between dwelling size and location value

    is much stronger in the observed data than for attached, perhaps reflecting the volume of data

    points but also the much stronger relationship in the detached house typology between land and

    floorspace value due to the limited amount of floorspace able to be generated per unit of land are

    under the House form (c.f. attached developments that are able to use less land per unit allowing

    more smaller dwellings to be feasibly sold to recover the initial land cost). Interestingly, the

    observed data suggests the detached dwelling size to be much larger than the LUT inputs were

    able to be set to, due to the influence of the Ceiling Factors. However the sales data and the $m

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    23/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 19

    information from that confirmed that houses of the average consented size could not be sold below

    the price ceilings we set, and so the input dwelling size was reduced accordingly.

    Figure 6: Detached (House) input sizes vs Building Consents

    3.1.2.4 Choosing between multiple viable developments.

    As section 3.1.2.2 states ACDCv3 tests a range of development opportunities on the modelled

    sites. The model now outputs numbers reflecting that each site now has at least nine slightly

    different chances to pass the feasibility test, all of which are under the price ceiling, instead of just

    a single optimised one which may be excluded post-run due to being priced above the ceiling.

    As the decision to develop can be taken only once (once developed the land is unavailable for

    redevelopment until the improvements depreciate to replacement value again, which would

    generally be beyond our forecast horizon), a single development must be chosen from those sites

    where more than a single option is viable.

    There are a number of potential approaches that could be taken to this issue:

    Single criteria focus (Kyle Balderston has taken this approach, focusing on information inthe supply data to determine a scenario, using multiple single criteria scenarios to

    determine a potential range)

    Multiple weighted criteria (Doug Fairgray has developed this approach, effectively

    advancing on the single criteria approach using data in the to produce a single parcel

    output taking account of more than one criteria different weightings would produce

    different scenarios, and therefore a potential range)

    Multiple weighted criteria including demand (Adam Thompson has advocated for this

    approach, effectively applying a multiple weighted criteria approach also accounting for

    information not in the supply data, i.e. demand).

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    24/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 20

    3.1.2.5 How the Model Works (description of the FME Workbenches)

    The ACDCv3 Model is made up of 3 main workbenches as well as a number of preparatory

    components.

    Stage 0: CfGS and ACDC Input Converter

    Stage 1: Feasibility Calculator

    Stage 2: Filtering and Exclusions

    Stage 3: Chooser.

    3.1.2.5.1 CfGS and ACDC Input Converter

    This is a prerequisite to running the ACDC Model. The CfGS provides the model with the base

    information on capacity and is also combined with various other input data sources in the ACDC

    Inputs Converter to provide the detailed base parcel data needed.

    3.1.2.5.2 Stage 1: Feasibility Calculator

    This component is essentially the evolved portion of the model, and is very similar to ACDCv2. The

    key difference is that the model now clones each parcel, with each close being tested for a

    different development typology. A considerable amount of further work has been undertaken to

    ensure that the LUTs and calculation steps account for the variation in developments for each

    clone/typology. This component takes approximately 10 hours of run time.

    3.1.2.5.3 Stage 2: Filtering

    This component is almost unchanged from ACDCv2 and effectively removes valid parcels that aredeemed to be unreportable post feasibility calculations for a variety of reasons including:

    - Site is used for a use deemed immutable and capacity even if feasible is

    considered unreportable (e.g. infrastructure, schools, churches, etc.)

    - Site is within Housing New Zealand ownership and targeted for redevelopment

    (capacity on these sites is accounted for in total supply by addition of HNZ

    redevelopment targets)

    - Designations deemed to preclude development13 (e.g. includes Ministry of

    Education, etc., but excludes airport noise designations and road widening).

    This component also tags excluded parcels with a primary reason and consumes and filters

    parcels excluded from the Stage 1 component (reasons include a lack of capacity to test (i.e. CfGS

    identified site as having potential of 0), a lack of necessary data required to make calculations (e.g.

    missing valuation information) or a variety of other issues.

    13Refer CfGS Methodology and Assumptions report

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    25/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 21

    3.1.2.5.4 Stage 3: Chooser

    This step is optional and runs a single criteria focused selection from all feasible development

    available per parcel. This provides a single development option per parcel depending on the

    choosing scenario, and converts the closed outputs into a format consistent with previous model

    versions.

    3.1.3 Feasible Dwelling Supply (ACDC version 3 Results)

    For the information contained in this section, the 5 different individual criteria developed by Kyle

    Balderston which in combination produce a range have been utilised.

    The selection is only from those developments that are viable (and below the ceiling), these

    developments are ranked according to the criteria and the top (or bottom) ranked development

    chosen for that site on sites with zero or one viable developments that is the top ranked

    development for all scenarios. The criteria are designed to answer the question Of the Feasible

    Development options on the site (within the range tested) which is the scenario that delivers the

    :

    - Maximum % Return: This is the output which is most consistent with the actor

    based approach to the modelling to date. While this is arguably the most likely

    (or developers first choice) development other factors may need to be

    considered14including demand, and the nature of the developer;

    - Lowest Project Cost: This represents the lowest capital outlay for the

    developer which may be an important consideration especially for small firms or

    individuals which dominate the construction industry;

    - Largest Dwellings: Size of dwelling affects possible dwelling yield (by

    decreasing it, also making this scenario the least number of dwellings scenario)

    and also impacts on costs and sale price. It is often anecdotally suggested, that

    big houses are the most profitable (as price increases faster than build costs),

    and the outputs of this scenario are closely aligned with the maximum return

    scenario correlating with that view;

    - Maximum number of Dwellings: This scenario focusses on supply of dwelling

    units from each development, focussing on maximum supply of dwelling units;

    - Cheapest Dwellings: This scenario focusses on identifying the development

    that produces the lowest priced dwelling units, focussed on the affordability of

    dwellings to the end purchaser Mr Thompson identified this as a key

    consideration in his earlier evidence.

    14Such as the required capital outlay (as per the Lowest Project Costs scenario), or an alternative approach based on gross or net

    dollar return, which may be an applicable approach for some self-financed or small scale developers.

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    26/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 22

    - Version 2 (High Price Ceiling): This is included as a cross check and to

    illustrate the variability in model outputs due to variations in input assumptions.

    3.1.3.1 High Level Results (TABLES)

    3.1.3.1.1 High level Results from ACDCv3.6 using PAUP zoning and ACAP Rules

    The outputs using just single criteria indicate a range of currently viable capacity estimates, from

    197,706 if the criterion is to select the largest dwelling, through to 255,881 dwellings if the sole

    criterion is to maximise the number of dwellings (Table 4)15.

    Table 8: High Level Results

    3.1.3.1.2 Detailed Results from ACDCv3.6 using PAUP zoning and ACAP Rules

    These outputs can be disaggregated to examine different typologies by different criteria. The

    possible mix of dwellings includes ranges from very few apartments, a small share of terraces and

    a dominance of houses (highest return and largest dwellings criteria), to a somewhat more even

    mix across the typologies. The lowest priced dwellings criterion has a 7% apartments, 11%

    terraces and 83% houses mix (Table 5), while the maximum number of dwellings criterion shows

    8% apartments, 18% terraces and 74% houses.

    15As noted earlier in this report ACDCv3.7 has been run which address how the Apartment typology. This is described in Kyle

    Balderstons evidence along with the modelling results.

    Choosing Scenario

    (ACDCv3.6)

    Total Feasible

    Capacity (n)

    Average Sale

    Price ($)

    Average

    Floorspace (m)

    Cheapest Dwellings 248,836 814,054$ 128.2

    Largest Dwellings 197,706 985,356$ 176.5

    Lowest Project Cost 222,212 861,307$ 136.7Maximum percentage return 209,931 974,559$ 174.6

    Maximum number of dwellings 255,881 825,429$ 132.3

    ACDC Version 2 (HPC) 144,165 850,475$ 119.7

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    27/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 23

    Table 9: Dwellings by Typology

    A wide range of outcomes is also possible in terms of dwelling prices. Table 6 shows the indicated

    dwellings in each price band. Clear contrasts are evident. If lowest price was the sole criterion fordevelopment, then around 15.4% of new dwellings would be in the under $500,000 band, and

    29.5% in the under $600,000 band. Similarly, the lowest project cost criterion and maximum

    number of dwellings would see 24-26% of new dwellings in the under $600,000 band.

    In contrast, the highest return criterion and the largest dwelling criterion would both see very few

    dwellings under $600,000, with development focussed in the $700-$1,100,000 price bands.

    Table 10: Dwellings by Price Band

    The results are useful for showing what is possible under the provisions of the PAUP. They depict

    the bounds under different assumptions. For instance if development occurred solely based on the

    % return to a developer, then the scenario implies a mix of predominantly large houses, and largeterraces, with a few medium houses, terraces and apartments. However, if development were

    Lowest Price

    DwellingsShare %

    Lowest

    Project CostShare %

    Maximum

    Number of

    Dwellings

    Share %Highest %

    ReturnShare %

    Largest

    DwellingsShare %

    Large Apartments 3,105 1.2% 1,855 0.8% 5,850 2.3% 1,835 0.9% 772 0.4%

    Medium Apartments 6,978 2.8% 2,219 1.0% 7,011 2.7% 18 0.0% - 0.0%

    Small Apartments 6,593 2.6% 36 0.0% 6,499 2.5% - 0.0% - 0.0%Total Apartments 16,676 6.7% 4,110 1.8% 19,360 7.6% 1,853 0.9% 772 0.4%

    Large Terraces 14,973 6.0% 10,217 4.6% 24,567 9.6% 19,472 9.3% 12,654 6.4%

    Medium Terraces 6,718 2.7% 4,939 2.2% 16,486 6.4% 32 0.0% 15 0.0%

    Small Terraces 4,883 2.0% 485 0.2% 4,679 1.8% - 0.0% - 0.0%

    Total Terraces 26,574 10.7% 15,641 7.0% 45,732 17.9% 19,504 9.3% 12,669 6.4%

    Large Houses 69,269 27.8% 80,952 36.4% 79,025 30.9% 185,139 88.2% 183,946 93.0%

    Medium Houses 90,589 36.4% 78,040 35.1% 89,606 35.0% 3,435 1.6% 319 0.2%

    Small Houses 45,728 18.4% 43,469 19.6% 22,158 8.7% - 0.0% - 0.0%

    Total Houses 205,586 82.6% 202,461 91.1% 190,789 74.6% 188,574 89.8% 184,265 93.2%

    TOTAL AL L DWEL LIN G 2 48 ,8 36 100.0% 222,212 100.0% 255,881 1 00.0% 209,931 1 00.0% 197,706 100.0%

    Source: ACDC15 Model 22-1-16

    Dwelling Typology

    SELECTION

    LowestPrice

    Dwellings

    LowestPrice

    Dwellings %

    Lowest

    Project Cost

    LowestProject Cost

    %

    MaximumNumber of

    Dwellings

    MaximumNumber of

    Dwellings %

    Highest %

    Return

    Highest %

    Return %

    Largest

    Dwellings

    LargestDwellings

    %

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    28/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 24

    driven more by the number of dwellings, then a more diverse mix would ensue, with considerably

    more medium and small houses, apartments and terrace houses.

    Note that each set of figures is based as if a single criterion were the predominant guide on all

    development which is viable. Development decisions typically take into account a variety of factors,

    which include but are not limited to the return on investment, perceptions of the size of the market

    for each typology, the location of the development opportunity, the circumstances of the developer,

    and so on.

    This suggests that the eventual outcome, if all of the capacity identified here as being currently

    viable is developed, is likely to be an amalgam of the mixes shown. Certainly the results are not

    near final, but they indicate the scale of development indicated as currently viable, and also reflect

    the range of possible outcomes.

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    29/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 25

    4.0 How Residential Supply and Demand have been estimated

    The Panel requests that the output from this work include:

    A supporting document that sets out the method used to estimate demand and supply, with

    any significant diverging views and their implications for the estimates recorded in an

    appendix (along the lines of the approach used in the July report)

    Comparing Supply with Demand4.1

    Demand has been estimated using the approach outlined in Section 2. Supply has been estimated

    using the ACDCv3 model as outlined in Section 3. This section outlines how the two may be

    compared, contrasted, and potential influenced by the other.

    As noted in Section 2 (Demand) and 3 (Supply) consideration of supply and demand in isolation

    can be problematic the trade-offs between the range of options open to both suppliers andconsumers of housing are functions of the interactions between them.

    While there has been insufficient time to develop a fully integrated model that would consider all of

    the matters necessary to describe an outcome resulting from these trade-offs at a high level supply

    of dwellings by price and type as estimated by the ACDCv3 under a range of scenarios can be

    compared with demand from households for dwellings based on their incomes and needs

    (converted to dwelling price and type) also based on a range of demand scenarios (largely

    affecting quantum of houses rather than strong variations in incomes or household nature).

    Household prices or types where demand from households for dwellings appears to be unsatisfied

    by feasible supply may help to indicate potential regulatory (i.e. planning adjustments) or non-

    regulatory interventions (e.g. additional social housing requirements).

    4.1.1 Methodology 1 (Adam Thompson)

    Economics of Housing Demand

    A population or household forecast does not in itself enable an estimate of demand for

    dwellings. This is because the composition or size of households is influenced by the housing

    market, in the sense that as the price of housing increases, household size also tends to increase.

    The increase in dwelling size is evident in Auckland. By implication, the quantity of housing

    demanded will be lower when dwellings prices are high (as fewer people will able to afford to buy

    or rent) and higher when dwelling prices are low. This appears to be the case for Auckland. Table

    below shows the change in household size and composition for Auckland.

    The purpose of the table is to show the difference between the forecast household size and count,

    and the actual household size and count. Statistics NZ forecast (2006 base) household size to

    steadily decrease over the past decade, from 2.90 persons per dwelling in 2006 to 2.82 persons

    per dwellings in 2015. This was due to a benign economic environment and changing

    demographics, most notably the baby boomers entering retirement and changing from largerfamilies to couples and single households, and later family starts driving observed HH size

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    30/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 26

    2006 2013 2014 2015

    Forecast Population (2006 Base) 1,371,000 1,534,400 1,557,600 1,580,800

    Forecast Household Size (2006 Base) 2.90 2.86 2.84 2.82

    Forecast Households (2006 Base) 472,760 529,100 544,620 556,620

    Population Estimates 1,373,000 1,493,200 1,526,900 1,569,900

    Actual Household Size 2.90 3.00 3.03 3.07

    Actual Households 473,450 497,730 504,310 511,940

    Difference Between Forecast and Actual

    Households (Dwelling Deficit)690 -31,370 -40,310 -44,680

    Source: Statistics NZ, Urban Economics

    reductions in the period prior to the forecast, conditions that were (and still are) expected to

    continue.

    Under this forecast household size projection, given the actual population growth (from population

    estimates), expected households should have increased from 473,450 in 2006 to 556,700 in 2015,

    an increase of 83,250 households.

    What actually occurred however was the opposite, household size increased steadily, from 2.90 in

    2006 to 3.07 in 2015.

    Under this forecast, given the actual population growth, Statistics NZ expected households to

    increase from 473,450 in 2006 to 511,940 in 2015, an increase of 38,490 households. The final

    row shows the difference between the expected household formation rate and the actual

    household formation rate, which was -44.760 over this period.

    Given that the external factor of high dwelling prices was not accounted for by Statistics NZ, this

    suggests that if dwelling prices had not increased so rapidly, there would have been an additional

    44,760 households formed, and therefore there would have been (ex ante) demand for an

    additional 44,760 dwellings (or in other words, 44,760 households would have chosen to reside in

    their own dwelling, rather than form larger households (e.g. complex households) if prices had

    not increased so rapidly.

    In terms of the economics of housing demand, it can be concluded that the demand for housing is

    presently in dis-equilibrium and there is a deficit of approximately 45,000 dwellings (as every

    household requires a dwelling). A report16 prepared by Statistics NZ estimates that 8% of

    dwellings in Auckland are crowded. This largely confirms the findings outlined.

    The PAUP should therefore be amended to provide for an additional 45,000 lower price dwellingsin order to meet this deficit. In my opinion meeting this deficit is required under the social and

    economic welfare provisions of the RMA.

    Table 11: Household Size and Formation Rates

    16Goodyear, R & Fabian, A (2014). Housing in Auckland: Trends in housing from the Census of Population and Dwellings 1991 to

    2013.

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    31/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 27

    In terms of the economics of housing demand, it is generally accepted in the economic literature

    that the price elasticity of demand for housing is in the order of -0.7%17This means that a 1%

    increase in the average price of dwellings results in a -0.7% demand for dwellings, as fewerhouseholds can afford to own or rent a dwelling within the limits of their income, and are forced to

    choose between moving to another City or to share a dwelling.

    For Auckland this means that a 10% increase in the average price of dwellings results in a

    decrease in annual demand (assuming 10,000 to 13,000 dwellings required) of approximately 700

    1,000 dwellings. The foregoing is relevant to understanding the future demand for housing

    because demand is significantly influenced by price, in the sense that, whilst population may be

    increasing rapidly, if housing is very expensive relative to incomes, then fewer people will be able

    to afford to own or rent their own dwellings, and will therefore either establish larger households, or

    move to another city. This is clearly evident for Auckland in the foregoing table and indicates asignificant market failure (or more accurately government failure in regard to supply constraints).

    When evaluating the demand for housing there is a need to decide between an ex ante and ex

    post demand evaluation methodology. These are defined as follows:

    Ex ante demand is the aggregate demand before consumers have interacted with the

    marketplace.

    Ex post is the aggregate demand afterconsumers have interacted with the marketplace.

    The practical implication of the difference between and ex ante and ex post housing demand

    evaluation methodology, is if ex post is chosen, then demand is assumed to be what is evident in

    the current market, and therefore households that have significant equity, as a result of unusually

    large price rises, will demand more expensive dwellings.

    Conversely, if ex ante is chosen, then demand is assumed to be what households can afford to

    spend on a dwelling based on their income and ability to service a mortgage.

    Using an ex ante methodology maximises economic welfare because it does not benefit

    households that have already interacted with the market (i.e. purchased a dwelling before prices

    have significantly increased) and create a cost for those that have not already purchased and

    would struggle to purchase (or rent) a property based on the current high prices.

    An ex ante demand methodology would also reduce the risk of some households having significant

    negative equity if prices drop. At present, 23% of households in the US18have negative equity.

    17For example, theprice elasticity of the demand for housing services in North America is estimated as negative 0.7 by Polinsky and

    Ellwood (1979), and as negative 0.9 by Maisel, Burnham, and Austin (1971).18

    , George R. Carter III, U.S. Census Bureau, Housing Units with Negative Equity, 1997 to 2009

    https://en.wikipedia.org/wiki/Price_elasticity_of_demandhttps://en.wikipedia.org/wiki/Price_elasticity_of_demand
  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    32/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 28

    Price Bracket ($,000's)

    Current

    M edian Price

    Scenario

    Percentage

    CurrentM edian Price

    Scenario

    M edian Price

    "Five by 2030"

    Scenario

    Percentage

    M edian Price"Five by 2030"

    Scenario

    $0-$99 150 0% 390 0%

    $100-$199 370 0% 19,850 6%

    $200-$299 3,170 1% 80,940 24%

    $300-$399 20,120 6% 77,060 23%

    $400-$499 38,320 12% 59,660 18%

    Sub-Total less than $500 62,130 19% 237,900 72%

    $500-$599 44,270 13% 36,020 11%

    $600-$699 40,030 12% 18,320 6%$700-$799 39,470 12% 12,100 4%

    $800-$899 29,820 9% 7,850 2%

    $900-$999 27,460 8% 4,860 1%

    $1,000+ 89,460 27% 15,600 5%

    Total 332,640 100% 332,640 100%

    Source: Auckland Council, Urban Economics

    The five by 2030 objective that is to be included in the Auckland Plan indicates a median price of

    $382,500. This is consistent with an ex ante demand evaluation methodology because it reflects

    the ability to pay based on income rather than with large capital gains.

    In order to estimate the overall demand for housing, in terms of price, I therefore consider that

    income (ability to pay) of existing residents is the preferred approach.

    The following Figure and Table 3 show the distribution of all existing dwellings by price in

    Auckland (blue). It also estimates a possible overall price distribution if the five by 2030 objective

    was achieved (yellow).

    Figure 7: House Price DistributionCurrent vs Five by 2030

    Table 12 House Price DistributionCurrent vs Five by 2030

    Additional low cost housing needed

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    33/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 29

    Scenario $0-$499$500-

    $599

    $600-

    $699

    $700-

    $799

    $800-

    $899

    $900-

    $999

    $1,000-

    $1,499

    $1,500-

    $1,900$2,000+ Total

    Apartment Large 1 1,291 543 1,835

    Apartment Medium 18 18

    House Large 386 18,450 35,814 8,465 47,358 68,774 5,336 556 185,139

    House Medium 1,203 1,280 461 304 184 3 3,435Terrace Large 1,515 1,370 10,056 6,435 96 19,472

    Terrace Medium 32 32

    Total 0 386 19,653 37,095 10,459 49,032 80,337 12,317 652 209,931

    Percentage of Total 0% 0% 9% 18% 5% 23% 38% 6% 0% 100%Source: Auckland Council ACDC 2015

    The main implication of this is that under the current price scenario only 19% of dwellings are

    priced under $500,000, whereas under the five by 2030 scenario requires 72% of dwellings to be

    priced under $500,000. The ACDC15 maximum percentage return results show that only 1% of

    net additional dwellings can be built for less than $600,000. This indicates that the urban portion of

    the city will, under this scenario assessing the PAUP, will not enable enough low cost dwellings to

    achieve the regional five by 2030 objective or equilibrium demand, which are broadly equivalent.

    Matching Demand with Supply

    The Maximum Profit Scenario outputs from the ACDC15 model show that only a small number of

    dwellings are likely to be built for less than $600,000 (only 386 dwellings or less than 1%). The

    most useful outputs results are the maximum percentage returnbecause this is the option that the

    developer would in most cases choose. This is subject to there being a market demand in that

    price point; sometimes a developer will choose a lower intensity development, such as a two lot

    stand alone subdivision, rather than a more profitable and more intense development, such as a 6-unit terrace development, because of either expertise or access to capital/finance. These model

    outputs are summarised below.

    Table 13: Maximum percentage profit by sale price ($,000) rangefeasible dwellings under the

    PAUP

    Dr Fairgray estimates that approximately 22% dwellings sold are for less than $500,000 based on

    historic sales data. 14% of properties listed for sale on Trademe (at XXXX) are for less than

    $500,000. This suggests that around 15-20% of dwellings in Auckland are available for less than

    $500,000.

    To achieve the five by 2030 objective or an efficient housing market requires a much higher

    number of dwellings to be enabled for less than $500,000. The five by 2030 objective requires

    around 72% to be in this lower price range. The estimated deficit suggests that there is an

    immediate need for 45,000 dwellings for less than $500,000.

    This does not appear to be achievable based on the ACDC15 model results.

    Table 4 Maximum Percentage Return Scenario: Net Additional Feasible dwellings under PAUP

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    34/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 30

    I would suggest that the PAUP should enable a minimum of 120,000 dwellings for less than

    $500,000, to meet both the historic deficit and future demand growth over the next decade. This is

    based on:

    100% of the existing deficit of 45,000 dwellings.

    50% of future (10 year) growth, including a five-year supply buffer.

    Tenure

    As a general comment, I do not believe that tenure is relevant to the question of supply and

    demand, as some countries function well with a high rate of rental tenure. The objective should not

    be to increase ownership per se, rather to provide a market that enables inexpensive housing.

    4.1.2 Methodology 2 (Douglas Fairgray)

    To examine the demand side of the housing equation, I have focussed on both the volume of

    demand for housing in terms of dwelling numbers, and the structure of that demand in terms of

    dwelling pricethat is, distribution of that demand across price points.

    For this, I have drawn on the projected increase in household numbers over the period 2013-26

    and 2013-41 for both medium and high population projections. It also takes account of the

    estimated shortfall in dwellings as at Census 2013, which is relevant in terms of making provision

    for all of Aucklands housing needs. This is on the basis that demand for housing is manifest as the

    ownership and the rental of permanent private dwellings, with the PAUP approach to meeting the

    requirements of the Auckland population based on one dwelling per household.

    The purpose of this current analysis is to develop estimates of housing demand in Auckland to

    2026 and 2041, in terms of total dwellings, and the price structure. This is to provide a suitable

    basis for comparison with the dwelling supply side data.

    4.1.3 Key Issues

    Assessment of the price structure of housing demand is not a simple exercise, and a number of

    factors need to be taken into account. There are several dynamics in the market, there are many

    segments within the market to which specific price structures apply, and circumstances change

    over time. Particular issues include:

    a. The interface between the rental and owning components of the market, where dwelling

    prices (especially) directly influence decisions to purchase or be tenants, and decisions on

    household formation. Households circumstances and ability to be renters or owners vary

    over time, especially as they progress through key life stages. This means that current or

    historic levels of ownership (vs rental) do not necessarily reflect the future levels, or

    represent aspirational levels. Over time, the general trend is for dwelling ownership rates to

    be higher for older households which generally (but not solely) reflects a move to become

    owners rather than renters.

  • 7/26/2019 081a Ak Cncl - General - Capacity and Feasibility Modelling - Residential Capacity Results, Methodology and Assu

    35/53

    ______________________________________________________________________________________

    Feasible Residential Capacity 31

    b. The current price structures in the market do not necessarily fully represent the demand

    from each segment. This is especially the case for first home buyers / (generally) younger

    households who are not active in the market as buyers because they cannot afford to be. It

    is reflected in the lower levels of dwelling ownership identified by the 2013 Census,

    compared with previous Censuses, for these segments especially. In combination, these

    two factors suggest that the current price structure (2015) as reflected by the numbers of

    sales in each price/value band is truncated and understates demand in the lower value

    ranges.

    c. A third factor is on-going change by existing households, both owners and renters, as they

    change dwellings over time. Commonly referred to as the churn in the market, this

    process includes (but is not limited to) owners of dwellings purchasing higher value

    dwellings as they move through the life cycle especially in the 40-65 age groups. It also

    includes households in the more mature age groups who purchase smaller and/or less

    valuable dwellings later in life, including those who move into retirement villages19and thelike. This process means that there is regular availability of existing dwellings to be

    purchased, for owner occupiers or for rental purposes.

    d. Part of this process arises from a fourth factor, that over time some existing households will

    pass on20.

    e. Demand for housing for each segment in the market may be met from both new dwellings

    and the existing housing stock. Simply, an increase in demand for (more) dwellings arising

    from an increase in household numbers does not mean that new dwellings would be to

    meet the needs of new households. Rather, the pattern is for new dwellings to be taken upby existing households and new households, and the movement of existing households

    through the housing stock as buyers, sellers and renters is an important catalyst for

    existing dwellings to become available to new households. This is especially because new

    households tend to be younger and more price sensitive than the average, while new

    dwellings are more expensive than the average which means a common process is for

    existing households to move into new dwellings, and thereby make available existing

    dwellings to new households. This is particularly important when considering the needs of

    new households it is not as simple as estimating the price structure relevant to new

    households, and assuming that such a price structure should be a