Packaged Retail Investment and Insurance-based Investments Products (PRIIP) - Scenari di performance con traiettorie alla Cornish-Fisher

The European Regulation n° 1286 of November 26, 2014 (PRIIPs regulation) will be effective from January 1st 2018 and will oblige products manufactures to produce a three-page document (the Key Investment Document or KID) for several products categories offered to retail investors. In 2020 the KID will also replace the KIID currently compulsory for UCITS funds. One of the key points intro-duced in the new regulatory document is the presence of ex-ante performance scenarios while in the KIID only past performances are represented. In this article, the methodology proposed in the level II of the RTS is analysed and backtested. In par-ticular, the aspect of the European legislator's willingness to quantify the expected returns of invest-ment in four market conditions - stress, unfavourable, moderate and favourable - and two variants of the model proposed by the European Commission were tested and compared with a model based on the simple normality of returns by adopting a backtesting procedure in a time window from 1980 to 2017 on a sample of over 9,000 assets. The work proves that the model proposed by the European Commission for the calculation of perfor-mance scenarios for category 2 PRIIPs (linear products) may not to be considered applicable for those products with a maturity less than one year (e.g. Forward/Future contracts with an underlying ex-change rate and expiration up to six months) when the calculated parameters are such that they do not allow the compliance of the bijectivity of the quantitles distribution function. Furthermore, the results of the backtesting show how the use of Cornish-Fisher's expansion to forecast future performance is compliant with the hypothesis of normality of returns and that between the two proposed variants. We have verified that the variant in which the underlying factors have a zero drift (described in the RTS modification proposals published on November 14th , 2016) is more reliable than the one that foresees a non-zero drift (described by the RTS of June 30th , 2016 and March 8th, 2017).