The main estimates of this exercise – the decomposition of ICIO* GFCF by shares of using industry – can present inaccuracies due to data limitations. For instance, to transform asset-industry investment data into a GFCF matrix, assumptions have to be made to convert assets to industries. While the solution taken in this estimate is to broadly match assets to industries, i.e. construction assets correspond to the construction sector, a more thorough approach to capture such correspondence would be to calculate through asset to product information and thus (capital) use tables. Because converting assets to products can be significantly country specific and data for doing so is sparse, this approach was not taken for this cross-country dataset.
[...] There is an inherent time inconsistency between CFC and GFCF. While GFCF is a flow of investment in a given year, CFC is a constructed flow of usage of the stock of capital, which in this exercise is proxied as the sum of the last 5 years of GFCF.