Reporting and derivation of data on financial transactions related to banks' securities holdings

This paper contributes to the ongoing efforts by the European authorities to reduce the reporting burden for banks by assessing the statistical methods currently used to compile data pertaining to financial transactions on securities holdings. Based on statistical information collected by the Bank of Italy, we compare data on purchases of securities (net of sales and redemptions) reported by banks with transaction estimates based on indirect (i.e. balance sheet) methods envisaged within the methodological framework of datasets compiled by the European System of Central Banks (ESCB). Although the direct method of collecting data on transactions is more costly for reporting agents, it produces results which are fully aligned with current statistical methodological standards (European System of Accounts 2010, ESA 2010). By contrast, the indirect method is a simplified and less costly approach. The recent development of high-quality data sources such as the ESCB integrated system for the market prices of securities – the Centralised Securities Database – has boosted the attractiveness of indirect methods, since they have the potential to deliver accurate and reliable estimates. The significance of the differences between direct collection and indirect compilation of these data is analysed in detail for listed ISIN securities that are actively traded on exchanges, by also considering the impact of price volatility and trading activity. From an aggregated perspective, all indirect methods produce results which are comparable and consistent with the ESA 2010 methodology for all instrument types. There are some minor differences for equity instruments, due to the higher price volatility and trading activity associated with these instruments, but the overall aggregated dynamics are also well captured by indirect methods in these cases. The results thus support implementing simplified reporting solutions that would reduce the burden of statistical data collection without jeopardising statistical quality. It should also be noted that the differences can be expected to be even smaller if the methods are applied at a monthly frequency (as may be the case in future in the context of the ESCB Integrated Reporting Framework, for instance) instead of at a quarterly frequency, as in our exercise.