Stephen Byrne (Central Bank), Kieran McQuinn (ESRI), Austin Hughes (KBC), Kevin Timoney (IFAC), Seamus Coffey (UCC), Simon Barry (Ulster Bank).
CSO: Michael Connolly, Chris Sibley, Stephen McDonagh, John Sheridan, Tim Linehan, Ruth O’Shaughnessy.
Loretta O’Sullivan (BOI), Ruth Lennon (D/Fin), Annette Hughes (EY), Ciarán Nugent (NERI)
1. The Quarterly National Accounts for 2020Q2, released on the 7th September, were reviewed in detail, with issues such as the chain-linking of pricing; the composition of Balance of Payment Direct Investment outflows; and the composition of the increase in Gross Value Add in the Industrial sector discussed in some detail.
2. Tim Linehan from the CSO’s Methodology Division presented the new CSO approach to seasonal adjustment. Seasonally adjusted aggregates can be computed either by aggregating the seasonally adjusted components (indirect adjustment) or adjusting the aggregate and the components independently (direct adjustment). In the 2020Q2 Release, seasonal adjustment was conducted using the indirect seasonal adjustment approach. This approach is in line with CSO’s Policy on Seasonal Adjustment and Eurostat’s Recommendations on Seasonal Adjustment.
The indirect approach can give the best results when the component series show very different seasonal patterns, which is a feature of some data series in the Irish National Accounts. Under this indirect approach, individual time series are independently adjusted at the component level. These individual series are then aggregated to compute seasonally adjusted results for GDP, GNP etc. This indirect approach is applied to GDP and GNP and the seasonally adjusted GDP and GNP are derived as the sum of the seasonally adjusted components (such as Personal Consumption, Exports etc.).
Once the statistical discrepancy is taken into account, the sum of the seasonally adjusted components of GDP and GNP add up to the respective seasonally adjusted GDP and GNP series. This preservation of additivity is viewed as an improvement from a user perspective. As part of the seasonal adjustment process using the US Census Bureau’s X-13-ARIMA framework, ARIMA models are identified for each series based on unadjusted data spanning Q1 1995 to Q1 2020. These models are then applied to the entire series (Q1 1995 to Q2 2020). Seasonal factors and the parameters of the ARIMA models are updated each quarter.