Thursday 7 February 2013

Comparing prices and rents in Central London with new data

Posted by Dr Philippe Bracke, LSE and SERC 

As homeownership rates fall across the country, ‘Generation Rent’ seems here to stay. But beyond the media coverage, we know surprisingly little about the private rented sector – even down to the most basic issue: price dynamics.

The National Statistician's Review of Official Housing Market Statistics, issued in September 2012, has identified the lack of an official private rental index as one of the three major issues currently affecting housing market statistics. That ‘data gap’ makes it harder for policymakers to make effective decisions.

When official sources are inadequate, private datasets can be very useful. In recent research for SERC I analyse rental data from John D Wood & Co., a real estate agency. My main focus is on how prices and rents relate. The price-rent ratio – or its inverse, the rental yield – is an important determinant of people’s tenure decisions. And for investors, yields are fundamental to assess the viability of real estate investments.

Having access to individual rental contracts, as well as individual sales, I can compare prices and rents for properties with similar characteristics. I start by computing average prices and rents within “cells” of properties with homogeneous characteristics (eg, 2-bedroom flats in SW7). Using data for the period from 2004 to 2011, I identify two patterns:
  1. Bigger properties display lower yields (higher price-rent ratios): this applies to houses vs. flats, but also to big flats vs. small flats. 
  2. Properties located in more expensive neighbourhoods deliver lower yields.
These patterns are well-known among market practitioners. Moreover, they are not limited to these years or to Central London specifically: in the nineties, the Joseph Rowntree Foundation produced a report on rents and yields that highlighted the same stylised facts. However, these patterns have not been the subject of academic studies, so far.

Standard methodologies cannot take into account unobservable “quality” differences between properties. Results will be biased if, for instance, within a same cell properties for sale have higher quality than properties for rent. To overcome this issue, I run a more sophisticated analysis on a subset of the data. For each rental property in the dataset, I search the Land Registry (which contains all sales in England) and check when the property was last sold. If the property was sold less than six months before the beginning of the rental contract, I match its price and rent and compute its rental yield directly. Reassuringly, the results that I obtain with this subset of the data confirm the two patterns described above.

What explains these persistent differences in yields between properties of the same city? One reason could be unobserved costs. If transaction costs rise less than proportionally with prices, for instance, trading smaller properties becomes relatively more expensive. These extra costs would offset the higher gross yield for smaller properties so that ex-post returns are equalised. Or maybe ex-post returns are truly different: more expensive properties and exclusive areas display lower rental yields because they are perceived as safer, or, alternatively, investors expect to gain from pure price appreciation, not from rental income.

Our understanding of the private rental market is still in its infancy, at least for academics. But with an open-minded approach to data, we are starting to learn more.