The method used in the calibration of the SCR property shock is the same as the method used in Solvency II (the so-called ‘level II advice’ method). The approach followed in the level II advice calibrates the property shock by deriving the lower percentiles of the unadjusted index data (using non-parametric methods), without fitting a specific distribution.

South African data was used in the calculation to compare to the result obtained from UK data. The FTSE/JSE SAPY Index, consisting of the top 20 market capitalisation weighted property funds, from 2005 onwards was used (this gave 1693 daily values). From this JSE property data, a 23.24% property shock was derived (the 99.5thValue at Risk over a 1 year time horizon was 23.24%). The data had, however, a couple of limitations that are discussed in the section below.

Given the limitations in the South African data, it was decided to retain the Solvency II property shock of 25%. As noted by the SAM steering committee task group, it is in general difficult to calibrate and compare property shocks across countries because of the illiquid nature of property markets and the part of the economic cycle currently experienced in each country.

The data used in the calibration of the property shock is of key importance – if the data has shortcomings, then the shock derived from the data will have shortcomings.

As noted above, various limitations with using the FTSE/JSE SAPY data to calibrate a reliable property shock were identified.

In short, the main limitations were:

Differentiating by property class – The JSE data did not provide a breakdown for all property classes. UK data did not indicate significant differences between property classes (office, retail, industrial, city, etc.) and therefore the Solvency II property shock was not calibrated for individual property classes. However, given the data limitations it was not possible to determine if the same holds true for South Africa, i.e. whether applying the same shock to different property classes is reasonable or not.

Country specific factors – Country specific impacts on property, such as immigration and crime rates, are not allowed for in the property risk calibration. Even within the same country, the property risk can vary substantially between regions, due to these types of factors.

Number of years covered – The FTSE/JSE SAPY index only covered 6 years, compared to UK data that span across approximately 22 years. We know that economies are cyclical, and 6 years may not be enough to adequately account for this cyclicality.

Kurtosis & variance – Compared to the UK data set used in Solvency II, the South African data was more leptokurtic (fatter tails and more peaked) and had a much higher variance. This may be due to the SA data covering too short a time period in comparison to the UK data.

In addition to the above, it should be noted that the volatility of property prices is not explicitly allowed for in the property risk calculation. This may cause the property risk to be understated during periods where property prices are more volatile than usual.

An area of uncertainty worth pointing out in a discussion around Property risk, is where shocks on leases should be performed in the SCR calculation (to the extent that the right of use asset is approved by the Prudential Authority). If the value of the lease asset is a function of property prices, it may make sense to shock it under Property risk – this might be the intention of the Financial Soundness for Insurers (FSI 4.1) paragraph 7.2b that requires ‘immovable property rights’ to be included in the property risk capital requirement. The interpretation and appropriate approach remain uncertain.