Testing seasonal adjustment with Demetra (for example, changes in inventories at current prices)

1
Testing seasonal adjustment with Demetra (for
example, changes in inventories at current
prices)

• Yankov V.M.
• Rosstat, Russian Federation

2

• Rosstat currently (Q3. 2011) conducts seasonal
  adjustment of time series of key macroeconomic
  indicators using the software Demetra 2.2
• In the process of data analysis it was found that
  updating of the seasonal model every time when
  new observations are added to the time series
  leads to significant changes in the values ??of
  the seasonally adjusted series
• In this regard, a decision that seasonal
  adjustment is carried out using a fixed model,
  which will be updated once per year, was made.
  This eliminates the significant revisions in the
  seasonally adjusted series.

3
Change in inventories / quarterly data from Q1
2003 to Q3 2011
4
A graph of the changes in inventories
The presence of seasonality is clearly visible as
some instability of changes in inventories and
the declining trend during the crisis period of
2009-2010
5
Check for a sliding seasonal factor
6
Main quality diagnostics
7

• Check for residual seasonality after adjustment

8
Model stability
9
Residual analysis
10
Assessing the possibility of publishing the
results

• The users of statistics may have trouble
  interpreting the results of seasonal adjustment
  due to changes in the results when adding new
  points of time series
• Changes in inventories is the most unreliable
  component and is subject to adjustment in
  compilation of GDP, calculated by different
  methods.