How long before we are building enough houses?

Storys about house-building have been all over the media, actually the DCLG live tables show for the last quarters data year on year there has been a fall in housebuilding.  However undoubtedly over the summer housebuilding has picked up, the NHBC data shows a 30% increase between the last three months data and the previous three months.  Much of this I think is because of the unusually bad winter and I expect some reversion to the mean.

The NHBC data however is raw, not seasonably adjusted, and only covers about 80% of completions.  So i’ve imputed the rates of increase seen in their data to the DCLG data to estimate that over 2013 there will be something like 123,000 houses completed, a modest 7% increase.

Lets assume that the rate of increase seen in the last few months continues into 2014, and lets make the heroic assumption it keep increasing without a housing crash, will it be enough?

To this end ill present a model where the whole of England is treated as one local plan area applying the usual local plan residual methods to it, including the ‘Sedgefield method’ for calculating backlog.  Ill assume a base year for housing need for 2010, and a plan year zero of 2013 (so backlog of need is calculated from 2010-2013, the plan year one (date of adoption) would be 2015 so the end period of the 15 year plan would be 2028.

I assume that annual hh formation was 230,000 a year 2010-2012 and 245,000 from theron based on the latest TCPA analysis of the census and hh formation, I add 7% technical adjustment to account for frictional  vacancies, demolitions and second homes.  I stress this is back of the thumb,

Clearly the rate of increase over the last 6 months will revert to the mean.  If housebuilding increased by 20% YoY to 2028 we would be in the millions of units per year by then,  The rate of increase in housebuilding is governed by the rate of profit of house-builders, if this is high they can retain profits, raise equity or borrow.  I generously assume a rate of increase of 15% YoY for the next two years and 10% per annum thereafter.  In the spreadsheet I expose these as variables so you can change them.

Even with this increase you will be short for many years so I apply the Sedgefield method to calculate how the annual backlog is added to the net need over the next 5 years.

By this calculation the need+ backlog dramatically rises from 286,000 in 2014 when we would be building around 241,000 homes to a maximum of 582,000 by 2020 by which time we would be building around 259,000 homes, thereafter the net need would fall because of the relentless increase in housebuilding so that by 2029 we would be building around 423,000 homes per annum finally meeting need so that the next year the number of new homes needed would fall dramatically to 411,000. By 2024 we would be buiding enough homes per annum to meet net annual new HH formation but it takes another 5 years to fully clear the backlog.

If you reduced the growth assumptions to less optimistic 10% and 8% per  annum it takes till 2030 to break even. If they are reduced to 5%/annum then you never get there as by 2030 the sedgefield adjusted need rises to 1 million per annum and you would only be building 280,000.

What this demonstrates is that the Sedgefield method is too much of a fast moving goalposts and even with up to date plan coverage of 100% they would all soon be rendered out of date and para 14 of the NPPF applies – no longer a plan led system.  The analysis shows that phasing the backlog over 10 years would be more realistic.  Clearly there was no regulatory impact assessment on the Sedgefield method.

It also shows that even with fairly heroic assumptions about recovery by housebuilders it will not be enough, a housebuilding programme, including public housebuilding, on a post war scale is needed.


Was Wynne Godley ‘Old Fashioned’? – Krugman’s Straw Man Excuse for Not Reading Him

Krugman has written a piece on the NYT article on the increasing influence of Godley’s work.  Perhaps he felt under pressure to do so to avoid looking old fashioned and in a classic example of transference accuses Godley of being ‘old fashioned’.  Clearly a blogosphere troll to which Ramanan has already done two spot on response (here and here) and we can expect many more.  I hope this leads to as good an examination of Krugman’s assumnptions here as that of assumptions on loanable funds, but in that case he could wrongly state it doesn’t really matter, my neo-classical models would be the same.  Here it really does matter as the assumptions of ‘maximising’ which he relies go to the heart of the neoclassical project. Krugman accuses Godley of reviving ‘Hydralic Keynsianism’ (ironic because the IS-LM models Krugman uses was one of the earliest results from this school) and states that there were

 some notable predictive failures of hydraulic macro, failures that it seemed could have been avoided by thinking more in maximizing terms.

The aim of the hydralic models of Phillips was to use system dynamics to develop a stock-flow consistent model of aggregate demand and the whole economy. But this early work did not include a mathematical framework for ensuring stock-flow consistency between different sectors of the economy.  Godleys great advance was to develop an accounting framework to ensure this consistency. Krugman gives as examples Freidman’s permanent income consumption function and NARIU.  Again ironic because there were few greater challengers to Freidman’s ideas than Wyne Godley.  The assumption here I think is that Godley’s ideas were not sufficiently ‘microfounded’ – wheras Freidman’s ideas, then embodied in the Real Business Cycle models – could predict 70s stagflation, and this of course evolved into the ‘New Keynsian’ macro models which Krugman champions. Lets take these claims in turn.  Firstly that Godley had a ‘naive consumption function’ that doesn’t take into account booms or wealth effects.  Lets look at one of Godley models from 1996

We assume, fairly conventionally, that real consumption is some proportion (less than one) of expected real income plus another proportion of the opening real wealth stock …or, solving out lagged wealth recursively, consumption can be written as a function of current and lagged income with the coefficients constrained in a particular way, to sum to unity

This clearly is someone who had read the literature on consumption functions and shows fairly clearly that Krugman has not read any Godley.   Indeed the whole ‘portfolio balance’ approach to consumption which Tobin pioneered was considerably advanced by Godley. As Randall Wray points put In Tobin in and out flows of funds were considered exogenous leaving agents solely to consider portfolio balancing, in Godley’s models however all flows are considered endogenous. On the NARIU issue, Godley never agreed with it and offered alternative explanations of 70s stagflation.  In Godley, W. and Lavoie, M. 2007. Monetary Economics: An Integrated Approach to Credit, Money, Income, Production, and Wealth, they deny the accelerationist approach and develop in chapter 10 an alternative stock-flow consistent approach to modelling stagflation.

So on both counts Krugman is building a straw man.  New Keynsian macro relies on a rational expectations model whereby agents maximise and this achieves gneral equilibrium.  Godley on the other hand models on the basis of

procedural rationality, with agents reacting to past disequilibria relative to norms (Monetary Economics P. 493)

Stock flow consistent modelling is able to fully explain price changes outside of equilibrium without heroic assumptions of rational agents being able to understand the models assumptions. This makes them useful for simulation in a way that New Keynsian models are not.    You can of course assume that individual pursue behavior in a broad optimizing manner, indeed Fishers PHD thesis was to construct a ‘hydralic’ model that simulated a walras type general equilibrium, but SFC models in addition can explain violent disequilibrium processes including economic crises.   ( I will be exploring n more detail the SFC approach to price determination in a future post)