The last few years have seen huge growth in cross-border property investment in Europe and 2006 is likely to set a new record. According to CBRE, cross-border investors accounted for almost half of the €89.7bn invested in European real estate in the first half of 2006.

While part of this capital transfer has been for tactical reasons, as investors in weak markets such as Germany have gone in search of higher returns, the main aim for most investors has been the strategic goal of diversifying the risks on their domestic property portfolio. In theory, if foreign markets are only weakly correlated, or synchronised with the domestic market, then cross-border investment should smooth out the returns on a property portfolio and enable investors to achieve superior risk-adjusted returns.

While immature markets in central and eastern Europe have recently grabbed much of the headlines, western Europe has inevitably been the main target for cross-border investors, because it has a much greater stock of investment property. Schroders estimates that the total value of investment property in the ‘old EU15' countries is roughly ten times that of investment property in the ‘new EU10' countries, Bulgaria, Romania, Russia, Turkey and Ukraine.

The dominance of western Europe raises the issue, however, of whether the diversification which investors are seeking is really achievable. Doesn't the creation of the euro and a single monetary policy and globalisation in the financial service, IT and telecoms sectors mean that Europe's office markets all rise and fall together? Isn't there a risk that, in common with foreign equities, the diversification benefits of foreign property prove to be illusory?

In order to address this issue we decided to look at the average correlation across eleven major office markets using real estate service provider CB Richard Ellis (CBRE) data on prime rents from 1985-2005. We focused on offices because they are the largest sector in most European countries. The 11 cities are Amsterdam, Brussels, Copenhagen, Helsinki, Lisbon, London City, Madrid, Milan, Munich, Paris, and Stockholm. (The choice of cities was constrained by our decision to look at trends over 20 years). Figure 1 shows the average correlation in rents across the eleven cities measured over rolling six year periods. For example, the latest observation covers the years 2000-2005 inclusive.

The analysis leads to two main conclusions. First, the average correlation coefficient in prime office rents is low at around 0.5, so that most investors ought to achieve significant diversification by buying assets in another west European city. Looking in more detail at individual cities, it appears that Belgian and French investors probably have most to gain from cross-border investment, because Brussels and Paris have tended to be quite idiosyncratic office markets and have low correlations with office rents in other cities. By the same token, Brussels and Paris should be attractive targets for most foreign investors. At the other end of the spectrum, Spanish investors may have the least to gain because Madrid has the highest average correlation coefficient (0.7) with office rents in other cities, although even in this case cross-border investment would still reduce volatility.

The second conclusion is that the evidence of convergence in Europe's office markets is ambiguous. Although the average correlation coefficient across the eleven office markets has risen since 2000, it is only slightly higher now than at its previous peak in 1994. Arguably, the data suggests that there is no long-term convergence in office rental cycles across western Europe and that changes in the level of synchronisation are cyclical.


n the mid-1990s the average correlation declined sharply because some office markets recovered quite rapidly, while others suffered a protracted downturn. More recently, the average correlation coefficient has increased because all cities suffered a fall in prime office rents at some point in the last five years. Looking forward, our forecast suggests that the average correlation coefficient will again decline as prime rents in London, Madrid and Stockholm bounce back quickly, while office rents in Amsterdam, Lisbon and Milan remain stuck in the doldrums.

At first sight the lack of convergence in office rental cycles looks odd given fairly strong evidence that European business cycles at the national level have become more synchronised, particularly among the major Euro zone economies. The European Commission regularly monitors the degree of synchronisation by measuring real output gaps - the difference between actual and potential GDP of the member states (the green line in figure 1).

However, if the same analysis is then repeated using the appropriate regional economic data (eg, Upper Bavaria for Munich, Ile de France for Paris - the yellow line in fig 1) then the behaviour of office rents in the 11 cities becomes more understandable. The regional economic data highlights the fact that there is still a wide range of regional economic performance in western Europe, even if the national business cycles of France, Germany and Italy appear to have become more synchronised.

If we then turn our attention to the retail sector, then the potential for diversification in ‘old' Europe appears to be even greater (figure 2). The average correlation cefficient in prime shop rents across nine cities in western Europe was 0.25 between 2000-2005 and over the long-term, the rolling correlation has averaged just 0.15. (The retail data are from Cushman and Wakefield and covers the same set of cities as the office analysis, except Helsinki and Lisbon).

The even lower level of synchronisation in retail rents compared with office rents, reflects a number of influences. First, although consumer spending accounts for the majority of GDP, the two are not identical and trends in retail sales reflect not only changes in income, but also house prices and fears over future unemployment. Rising house prices, for example, partly explain low savings rates and strong consumer spending in Spain and France.

Second, there is greater variation across Europe in national and regional planning controls on retail development than on office development. While it is currently very difficult to get permission for new out-of-town retail development in Germany and the UK, the planning regime in Sweden is generally quite liberal and controls on retail development in Belgium, the Netherlands and Spain have recently been loosened.

The third reason why the correlation in retail rents is so low is that the pace of structural change in retailing still varies significantly in Europe. In Italy and Spain the stock of shopping centre and retail park space is relatively small and multiple retailers are keen to take space in new developments, in order to win market share from small independent retailers. By contrast, in the Netherlands, Nordic region and the UK, multiple retailers are already dominant and many of the most interesting investment opportunities involve the extension, or re-development of old shopping centres.

In conclusion, we believe that there is little evidence of a common European real estate cycle. Regional economies are not highly correlated and market rents are subject to local supply constraints and shocks. Admittedly, we haven't yet extended the analysis to cover trends in property yields, capital values and total returns.

In addition, we probably need to take account of movements in exchange rates, given that most foreign property exposure is either unhedged, or only partially hedged. Those are issues for future research. Nevertheless, we are confident that our research into rental cycles is robust and that there are major diversification opportunities for cross-border property investors in old Europe.

Mark Callender is head of European research and Alex Krystalogianni is head of European forecasting at Schroders PIM in London