In its news release and in the data dissemination tree, Eurostat publishes seasonally adjusted data, corrected for calendar effects, for the euro area and the European Union, with respect to the following "key indicators":
- The gross household saving rate (defined as gross saving divided by gross disposable income adjusted for D8* net);
- The gross investment rate of households (defined as gross fixed capital formation divided by gross disposable income adjusted for D8* net;
- The gross investment rate of non-financial corporations (defined as gross fixed capital formation divided by gross value added);
- The profit share of non-financial corporations (defined as gross operating surplus divided by gross value added).
Non-profit institutions serving households (NPISHs) have been included in the household sector.
In addition, a larger set of time series, including the components of the key indicators, are published in seasonally adjusted terms for the euro area only:
- Final Consumption Expenditure of households (P31) - Compensation of Employees received by households (D1rec) - Gross operating surplus of households (B2B3G) - Property income paid by households (D4pay) - Other current transfers paid by households (D7pay) - Other current transfers received by households (D7rec) - Interests received by households (D41rec) - Property income other than interests received by households (D4Nrec) - Net social contributions paid by households (D61net) - Net social benefits received by households (D62net) - Net adjustment for the change in net equity of households in pension funds reserves (D8net) - Current taxes on income, wealth, etc paid by households (D5pay) - Gross fixed capital formation of households (P51) - Gross value added of non-financial corporations (B1G) - Gross fixed capital formation of non-financial corporations (P51) - Compensation of Employees paid by non-financial corporations (D1pay) - Other taxes on production paid by non-financial corporations (D29pay) - Other subsidies on production received by non-financial corporations (D39rec) - Changes in inventories and acquisitions less disposals of valuables of non-financial corporations (P5N) - Exports of goods for the total economy (P61) - Exports of services for the total economy (P62) - Imports of goods for the total economy (P71) - Imports of services for the total economy (P72) - Final Consumption expenditure of General Government (P3) - Gross Fixed Capital Formation of General Government (P51) - Total Expenses of General Government (TE) - Total Revenues of General Government (TR) - Net lending/net borrowing of General Government (B9)
The corresponding time series in Excel format are available in dissemination tree.
* Adjustment for the change in net equity of households in pension funds reserves
METHODOLOGY
The methodological choices made are explained in the following.
1. Indirect adjustment
Euro area
The method used in the euro area case consists in decomposing key indicators into component series.
These component series are first seasonally adjusted and then used to calculate indirectly intermediate totals and key indicators. This provides an extended and consistent set of seasonally adjusted variables for the euro area.
European Union
In order to provide seasonally adjusted key indicators for the European Union, the component series for non-euro area countries are first aggregated then seasonally adjusted and finally added to the seasonally adjusted data of the euro area to obtain EU times series. These seasonally adjusted data are used to derive the European Union key indicators.
The component series are not published in seasonally adjusted terms, as they may be impacted by exchange rate fluctuations. However, the latter have hardly any impact on ratios such as the key indicators mentioned above as the variations in the numerator and the denominator broadly cancel out.
2. Pre-treatments
The method chosen to pre-treat the series and correct calendar effects is Tramo as implemented in version 2.1 of the Demetra software.
- Choice of an additive or multiplicative model
A multiplicative model has been used by default except for time series which alternate in sign, namely: "changes in inventories and acquisitions less disposals of valuables of non-financial corporations (P5N).
- Correction for outliers
Tramo detects three kinds of outliers: additive outliers (AO), transitory change (TC) and level shift (LS). An additive outlier consists in a single point jump in the data, a temporary change a single point jump followed by a smooth return to the original path, and a level shift a permanent change in the level of the series. Each detected outlier is assessed according to its economic rationale.
- Correction of calendar effects
The presence of an Easter effect has been tested. Each detected Easter effect has been assessed according to its economic rationale. Given the short length of the series only one regressor has been used to test the presence of a working day effect. A working day correction has been implemented in the final model provided (1) it was statistically significant and (2) it improved the quality of the seasonal adjustment model.
3. Seasonal adjustment method and policy
The method chosen to seasonally adjust the series is Seats as implemented in version 2.1 of the Demetra software. This method has been favoured, as tests have shown that it generally leads to smoother and less volatile results than X12-Regarima.
The parameters and the models are revised once a year. ARIMA and regression coefficients are updated each quarterly release. Under special circumstances, such as a sudden economic slowdown, seasonal factors may be frozen until changes can be properly modelized.
4. Parameters and statistical tests
The file below provides the parameters used for the production of 2011Q4 seasonally adjusted time series: Parameters 2011Q4
