Introducing MOAN -Method for Objective Assessment of Need 2/2

In this second part ill introduce the component of the MOAN Spreadsheet Model.   First Part here.

Having taken the 2014 based Household Projetions (with a 10year baseline) it then applies a number of correction factors.

The figures are split into 2016-2031 (15 years from 2018) as teh current plan period, and then the figure from 2032-3052 (2 years) totalling 25 years.

The concept is that current plans adopted in the next 2 years are based on current methods – to avoid disrupting current plan formation, but that after that plans plan for the long range, and include a shortfall from 2016-2032 based on any gap between the current plan and the new method, as well as strategic allocations for large sites, such as large urban extension townships and Garden towns and Cities, that will complete beyond 2033. Plans need not allocate every site post 2033, they can quite legitimately leave a proportion of total allocations (around 20% from experience of previous plans) in a rolling manner to be brought forward in local plans and neighbourhood plans every 5 years.  The key though being that we marks out the broad locations for strategic scale development for the remaining 80% to ensure we can have joined up long range infrastructure and transport planning across whole regions.

Looking at 2016-2031 here is an example of the spreadsheet based modifications for Warwickshire,  The figures are in thousands with one decimal place (100s) which the correct degree of precision for large scale planning.

The first elements is to convert from households to homes.  This includes a factor for frictional valencies and second homes.  A fairly standard 3% in total.  As with any element an argument could be made that there should be regional variations.  for example in the south West of England second homes could be much higher. However in the first draft it adopts a KISS (keep it simple stupid) approach.  The second adjustment factor is for household formation suppression during the great recession. Empirically comparing similar periods before and after the Great recession this is around 9.4%.


These tw0 factors converts from a baseline OAN to a Homes requirement.

Now we need to convert from homes to an OAN target.

We need a figure for lag/non-completions.  this always used to be 10% but in recent years it seems to be a lot higher.  Again ideally LPA by LPA adjustments but as no national dataset we use the DCLG assumption of 10-20% and apply a mid range 15%.

The next element is to make an assumption that in the long run to 205s as a nation we will not only meet the needs of newly forming households but also start biting into the backlog of housing shortfalls created over half a century.  This, in the long run, would be the same things as a market signals and affordable housing ‘Uplift’

We have already adjust for concealed households before the Great Recession, however even before that there was still a big shortfall of people who could not afford to form new households.  The simplest thing here is to re-use a modelling assumption derived from the departments affordable housing economic model by the former NHPAU.  That an uplift of 12.5% would restore affordability to 1987 levels within 20 years.  There may be more recent runs, we don’t know, but as a starting point it applies this 12.5% uplift.

The final element involves applying a second order derived migration effect of 10.5%. uplift.  This accounts for where additional housebuilding creates a multiplier effect from supply chain effects,  expenditure by builders and suppliers etc. on services etc.  for the Oxford Cambridge Corridor study Cambridge Econometrics has estimated the effect at around 10.5%.

In growth areas this uplift will produce a total figure. In other areas however there will be net out migration from land constrained and low employment areas to growth areas, which will have enhanced housing targets and enhanced employment growth figures.  Therefore there will have to be a final round of ‘reconciliation’ to sure that all in and migration flow nationally sum to unity.  This is not included in the current first draft MOAN model as  this can only be done after the determination of where enhanced employment and housing growth areas will be – which can only be done with larger than local planning.

If an area which is land constrained still has an employment surplus it may still suck in migrants (many from abroad) who will enforced share to replace out migrants going to new homes, and in many cases new jobs, in growth areas. – such a the Oxford-MK- Camridge Growth Corridor.  If the donating area  does not have an employment curplus its population will decline or not grow as quickly as it might.  Much of this migration effect is accounted for nationally in the ONS projections, however it does not account for the enhanced in migration is newly designated growth areas and pro-rata internal in migration from non growth areas.

A column needs to be added to the spreadsheet for this reconciliation – and is not yet done.  It will basically take the lowest two quartiles of an index of an growth increases by LPA and weight the 10.4% national uplift according to this index (a ranked index how land unconstrained multiplied by the  growth (homes) for each LPAeach LPA is).    This will require some further GIS based modelling.  currently the model over estimates National Modelled OAn need by 10.4% as a result.


Introducing MOAN -Method for Objective Assessment of Need 1/2

With this the last day of consultation on the Government’s proposed national method for calculating Objectively Assessed  Need  this post finished by series critiquing the method and suggesting a more soundly based and simpler ‘fudge free’ alternative.


The starting point for this model is the spreadsheet proposed by the ONS – the 2014 based household projections.

Tis gives projections for every local planning authority in England and then conveniently groups them by county and region to 2039

This spreadhsheet is then extended first to include household to homes conversions factor then a homes to OAN conversion factor.  The spreadhsheet is then extended in time to cover 35 years (to 2052).  Why 35 years?  Because experience of post war planning shows that 35 years is the typical build out time of a New Town.  So if Garden Cities/New Towns are part of the solution then we have to plan over this period.

The result is a model where the resulting outturns for each planning authority and teh compoents of each element of the OAN fare fully transparent.  In other words it is not a ‘black box’ model/  A charge that could be laid at the DCLG OAN Method.

Transparency could be further raised if the OAN model  itself modelled the components of household formation change in a single model.  All element are published, and it is possible to put them together in a single model or series of models (using R or such like) as the GLA has done.

Two notes on the baseline.  It uses a 10 year baseline for protecting households, as the DLG rightly recommends.  The GLA have modelled 5 and 15 years years baselines as well as a ‘central’ 10 year baseline.

A second qualifier.  The ONS, in a source of eternal frustration, releases sub national population projections every 2 years and sunn national household projection every 4 years.  Quite why it does this is unclear however I think it may have been to avoid disrupting old structure plan timetables by releasing data on a predictable medium term cycle linked to plan revision. It is possible to reconstrict teh baseline using 2016 based sun-national projections, as the GLA and many private consultancies have done, however you then can lose the elegance of the ONS tables with its summaries for counties metropolitan areas and regions.  For simplicity I have stuck with the 2014 based data, and consider that changing national OAN numbers every two years is too short a cycle that would disrupt plan making.

The starting point of the model is household growth using the ONS method.This is of course just a projection, not a forecast, least of all a target, as the ONS always stress.  This can  therefore produce distortions of you then go on to use it in a ‘predict and provide’  manner simply basing the targets n past trends.  for example areas with little housebuilding will see a low increase in household formation.  Areas where people are forced to ‘bunk up’ and share will see low increases in headship rates.

The best way to deal with these issues is to make corrections to thehousehold formation projections rather than to abandon them.  The demographic method using household formation is the only demographically based game in town.  Imagine the scenario where a country did build enough housing, and that all overcrowded and concealed households could afford a home.  What would the household formation rate be in these circumstances?

This is the underlying philosophy of the MOAN method.  England fully and completely meeting its housing needs in the period of 35 years

This is a different philosophy than the government/s atomistic approach of calculating each LPA’s need and adding them up as its baseline, with an ‘uplift’ based on affordability.  Rather MOAN calculates OAN as it would be in a supply=demand England.  Then once this provisional baseline is calculated a calculation i made of the deliverability of housing in land constrained areas, and the deliverability of compensatory housing in employment growth areas.  The assumption being that need will migrate from land constrained to employment growth areas.  The best way of doing this is through a strategic plan informed by modelling.  Though in the absence of strategic plans a model can provide a baseline.

Finally an adjustment is made for the second order effects of housing growth.  Areas with additional housebuilding will experience additional growth from a multiplier effect, from construction, household goods, supply chain effects and impact on services.  This will suck in additional in migration.

To ensure there is no double counting a ‘reconciliation’ is made within the model  to ensure all in and out migration figures add to unity nationally.  This is done by increasing migration from other ‘areas.

As you can see this treats household formation, employment growth and migration as part of a singe integrated and consistent model rather than separately atomistic and likely mutually inconsistent approaches/  Ideally it should be done in an iterative way where the impact of different growth scenarios in different areas can be assessed, the essence of planning., rather than attempting to use a spreadsheet to solve all national houing problems without human input.

The advantages of a single stock-flow consistent model though are legion.  It can be for example integrated with other economic, transport and combined land use and transport models without fear of internal consistency, because MOAN uses the same kind of integrated modelling assumptions used in these kinds of models.

The second part of teh post to follow this afternoon.