In the sections describing the factor model we see clearly that a large part of the relative performance of areas can be explained by systematic risk factors, and that these are a means to model the well-known ‘ripple effect’ in a rigorous quantitative framework. This is interesting, but does not directly answer the homebuyer’s question ‘where should I buy to get outperformance’? The reason is that the homebuyer typically is constrained - they already know the area where they have work, friends, and family and only want to know where to go within those constraints. As an extreme example, a young family of 5 can fit into a 2-bed apartment in prime London - I have had neighbours like this - but they are looking to time a move to somewhere better suited, and want to know the numbers. The question for them is: ‘if we delay a year can we afford a better house in the suburbs, or should we do it now?’. There is a temptation to look at recent history and mentally project it into the future, but what’s really needed is a forward-looking model.
London makes for an interesting example. There is a wide range of options from prime central London to the edge of the commuter belt, which maybe for practical purposes is about 1 hour out. This approximately extends to Brighton, Oxford, and Cambridge, and within our cases Colchester is pretty much at this limit. See for example https://www.totallymoney.com/commuter-hotspots/tool/ for more detail. So near-enough to London we have five of the entire spectrum of nine (WC, NW, SE, CR, CO) actually feasible to live in and commute from. If perhaps we have a £1m budget and intend to invest to up this limit there is a very broad range of choices. Even within the London areas (E,SE,SW,W,NW,N,WC,EC) there are districts spanning approximately £4000/m2 to £28000/m2, a factor of 7, nearly 3 octaves in music. To pose this as a quantitative problem: we specify our investment horizon i.e. how long we intend to hold it for, add knowledge of the current phase (or equivalently the ‘hot’ price point in £/m2) and where that will have rotated to at the price horizon. This gives a price band of interest, and we should probably focus on the middle of it if possible to capture maximum cyclical performance from now to horizon.
Consider a fairly typical five-year horizon. Five years back in mid-2014 the phase angle was approximately 166 degrees, corresponding to a price of £9400/m2 - this subsequently rotated fairly steadily down to around 30 degrees now in 2019, which corresponds to a price of £1,400/m2 i.e. a very cheap level not found anywhere near London. Note that if the rotation were steady and the period is about 16 years, we would have foreseen a rotation of 23 degrees per annum and so a total change of 5 x 23 or 115 degrees, whereas we have seen 136 degrees - slightly more rapid than expected based on a simple constant rotation rate. Not having perfect foresight, we would have considered the sweet spot for a 5-year horizon to be around 166-2.5 x 23 = 109 degrees, which corresponds to £4200/m2. This is about the cheapest price in London, to be found in Thamesmead SE28/SE2, a new estate which because of its remoteness – and perhaps also its bleak ambience – has not proved a popular place in which to live. This has led to a spiral of decline and part of the district is now perceived as a ‘sink estate’. (https://hidden-london.com/gazetteer/thamesmead/)
So had we thought cyclically, this is where we would have invested - or we could have gone further afield and found more remote, more superficially pleasant places perhaps, but lets just follow the logic through with this example. First consider the indices for the top quintile by price in W (West London), and compare it with the bottom quintile in SE (Southeast London). W quintile 5 is down about -0.04 in log terms, while SE quintile 1 is up about +0.28, so a relative performance of .32, which in normal percent is 38%.
Given that this is quite a dramatic result, it’s a good idea to do a ‘sanity check’ or data audit on a few datapoints. Taking repeat sales in the postcodes EC,WC,SW3,SW7 (central London, South Ken, Chelsea) and selecting ownerships starting 2014, ending 2019, we get 19 records. Ranking on return and looking at the central 3 to robustly reject the outliers, we see the median ranked 9/19 returned of -4.85%. this pretty much tallies with our index.
Turning to Thamesmead we take repeat sales in SE2 and SE28, and again select holding periods 2014-2019, giving 10 repeat sales. Ranking on return and looking at the central 3, the median return is 46.6% or a log return of 0.38. Again this pretty closely tallies with the index, bearing in mind the small sample in this fairly crude ‘sanity check’.
What should we conclude?
And finally: the phase described here has come to an end, and London’s primest sectors should henceforth outperform the rest for some time in the early 2020s unless there is a major tectonic shift as a result of either brexit - or perhaps a new socialist era.