UK retail is seen as polarised between prime and secondary. But Béatrice Guedj and Tony Christie show how a multivariate analysis can provide a more nuanced picture

Polarisation has long been a buzzword in the UK retail market. Since the crisis, investors have become even more aware that core retail locations have performed better and expect that they will continue to do so. At the other end of the spectrum, investors have thrown out any forecasts of a recovery in secondary or tertiary locations. In fact, secondary assets seem to be slipping into a downward spiral as no real recovery is in sight. Many investors are realising they should not have been drawn in by the siren song of ‘secondary catch-up' in the debt-driven boom period before the crisis.

In fact, the word polarisation is never defined clearly. One view is that it means big centres, often out of town, with a wide range of modern stores. On the other hand, it is used in reference to the high streets of small and medium-size towns that have vacant shops and are in decline. It also conveys the image of large centres reinforcing their dominance over time, attracting both consumers and retailers, while other locations lose their momentum and in many cases become deserted high streets.

Such a core periphery phenomenon has re-emerged in the European real estate landscape. In the UK, as well as in parts of continental Europe, investment markets have been characterised by increasing competition for prime locations, compressing yields for core assets, while capital values in secondary locations have continued to slide.

The simplistic view is that by picking core retail assets, one can easily reap attractive returns over time. This would mean securing assets with (i) easy access to a large and wealthy catchment area, (ii) modern units with a wide range of prime retailers and (iii) restricted supply. Of course, attractive returns also depend on a specific risk stance. However, Grosvenor believes it is possibile to be both too bullish on prospects for growth in successful locations and not optimistic enough about the prospects and opportunities in what are perceived as weaker locations. In an uncertain environment where any recovery is likely to remain subdued, the key is to have the right underwriting capability to pay the ‘fair price' and to have the correct management strategy in the long term.

The UK retail property market is the largest in Europe in terms of stock and investment volumes. As such, it offers one of the largest range of assets but also locations. Additionally, it is highly liquid, transparent and arguably the more mature. In terms of assets, there are standard shops, department stores, supermarkets, shopping centres, retail parks, in-town versus out-of-town; retail is a broad and well established asset class in the UK compared with other countries in Europe.

From an occupier standpoint, this market is also very attractive. CBRE's annual survey of 326 major international retailers confirms that the UK is the most desired location for expansion.

Over 25 years, total returns in the retail sector were 8.1% per annum, according to IPD. From a risk-return perspective (measured by the average returns on the standard deviation), retail provides 0.9 compared with 0.7 for all property in the UK. This high value of the risk/ return ratio, as well as a similarly high Sharpe ratio for the retail sector, explains why the retail sector is in demand from investors.

Another reason is that it has always provided a sustainable income return well suited to matching long-term liabilities. Given that pension funds favour sustainable income yields combined with the defensive side of income returns of the retail property sector, the strong competition observed over the recent past is not likely to fade, especially for ‘safe assets' in the UK.

Within the UK retail sector, performance differences are mainly driven by capital value movements during the downturn rather than by differences in income returns . The effect is most striking in the shopping centre sub-market which offers the largest and most stable cash flow. It has more volatile capital values than its peers. In fact, the average picture behind the capital value fall over the past cycle hides huge dispersion between assets and idiosyncratic features.

The picture can be very different when considering specific criteria such as (i) the location (London versus other regions), in-town versus out-of-town, (ii) size of the scheme, or (iii) age of the scheme. Implicitly, these differences between criteria reflect wealth effects based on the catchment area, competition effects, the quality of the supply as well as the retail - in other words, the fundamentals of retail property.

The economics of polarisation
This background explains our motivation to characterise the UK retail hierarchy and where possible allocate a score to retail locations - to appraise the magnitude of the retail polarisation but also to find out its defining factors. We do this in the knowledge that retail property is always somewhat correlated with macro-economic fundamentals.
Firstly, if polarisation exists it has been generated by structural economic processes over a long time. The emergence of ‘winning' versus ‘losing' locations has been driven by the increasing diversion in productivity across regions over the past 20 years. An analysis of the ‘beta convergence' (a macroeconomic statistical approach) demonstrates that the UK, compared with other European countries, has not been characterised by a regional convergence. This has been generated by a huge concentration of highly skilled activities in London since the early 1990s. The dispersion of productivity per head has increased over time. Looking at the economic geography of the UK, the dispersion of productivity per head can be clearly seen.

The dispersion in income per capita - which is a good proxy for retail sales - has been somewhat less marked across the regions over the same period due to transfer payments. This explains why the dispersion in rent has also been minor compared with what might have been expected. However, cheap credit and the ability of households to borrow have triggered a virtual catch-up effect in some locations and explain rental growth increases. Since the financial crisis, the deleveraging effect has unveiled the hidden reality of differences in productivity and sources of households' purchasing power.

Therefore, falls in turnover and in rents have been larger in locations where: the purchasing power was low; the past catch-up effect was not driven by pure gains in productivity; or where levels of supply were not aligned with levels of wealth (retail productivity per capita). Polarisation is clearly a reality but the myth was, in fact, the earlier ‘catch-up effect'. So the original factors or retail polarisation are to be found in increasing dispersion in regional productivity. Therefore, polarisation and the gap between ‘winning' and ‘losing' locations is set to increase as long as recovery lacks momentum. Fiscal austerity will also act to widen regional rental differences.

We think that appraising the magnitude of the polarisation is implicitly about a UK retail hierarchy, and can use this to pick the best assets but also identify the best strategy ex ante rather than ex post through the property cycle in the long term. More specifically, an asset has a life cycle and can be purchased as core, core-plus or value-add if one thinks about a repositioning of a retail scheme.

The UK retail hierarchy
In order to fully understand UK retail hierarchy, we have used a multivariate analysis called principal components analysis (PCA). Each UK retail location can actually be described by hundreds of separate variables. Indeed, the retail hierarchy is sometimes thought of in terms of catchment size or floor space, or type of retailers. PCA allows us to analyse all of the potential factors driving the UK retail hierarchy and to identify the most critical ones. It also allows us to measure the specific impact of the key variables. In terms of investment strategy, once we understand the key factors determining the UK retail hierarchy, we can accurately model rental values and identify over and under-rented locations. There is one other advantage of using PCA: it groups retail locations within the UK retail hierarchy. The acquisition and management strategies are quite different for each of these groups.

As is intuitive, the PCA puts into perspective two set of main variables. The first set comprises property variables while the other set (second axis) is made up of socio-demographic variables. The PCA gives the elasticity of each variable, positive or negative. Interestingly, the property variables are more important than socio-demographic variables in explaining the retail hierarchy. This suggests that assets can always be repositioned with appropriate capital expenditure. As shown by the chart, all property variables reinforce themselves (by being oriented on the same direction of the axis) while the socio-demographic variables oppose themselves (social grades AB have the opposite statistical effect of social grades DE). Finally, we have accurately constructed the retail hierarchy by using the key locational factors we have identified. Such a retail hierarchy mirrors three possible strategies for an investor.

Tier 1 is made up of retail locations where the vector of attributes mainly comprises a high share of social grade AB, a larger share of young people. These retail locations would be perfect for a defensive strategy. Investing in these retail locations would mean purchasing large stable income streams over time but at a high price. These retail locations comprise a high share of core products, defensive in the downward phase of the cycle, and providing some upside in the recovery phase of the property cycle.

Tier 2 is made up of retail locations where the property characteristics are the key drivers. The vector of attributes is made up of the overall floorspace provision, the retail mix, the fashion supply. These retail locations comprise core as well as value-add products. As shown in the figure, there is a high dispersion of retail locations among this subset. This high dispersion, driven by a higher standard deviation of some specific coefficients, would mean there is a possibility to reposition a scheme by reconfiguring the property, improving the retail mix or specific supply, such as leisure.

Tier 3 is made up of retail locations where the vector of attributes mainly comprises a high share of social DE, large families with young children, a high share of older people. These retail locations usually comprise small catchment areas with low purchasing power or big catchment areas with dominant retail locations with large schemes performing well. The strategy for these locations includes ensuring a high proportion of discount retailers and exploiting the value uplift that usually takes place towards the end of the cycle. In the latter case, one has to be careful to exit at the right time.

Tier 1 and tier 2 usually comprise the winning locations, while tier 3 would comprise a non-negligible share of what are considered losing locations. Identifying structural patterns, such as tier 1, tier 2 and tier 3, and quantifying the drivers by tier products means optimising the stockpicking exercise as well as the underwriting process. The key to successful underwriting is being able to model the sustainable rental value for each location. Our analysis has enabled us to identify the key structural variables that determine sustainable rental values with a high degree of accuracy. This ensures that we always buy assets with some potential for rental uplift, irrespective of macro-economic conditions.

So, talking about polarisation is, in effect, a shortcut, as it does not provide the hierarchy of the retail landscape. Using multivariate analysis provides a much more accurate method of ranking retail locations and determining their rental value. Moreover, it provides an innovative way in which we can combine top-down and bottom-up approaches for a sound retail strategy.

Béatrice Guedj (left) is head of research and strategy, and Tony Christie is retail fund manager, UK, at Grosvenor Fund Management Europe