The IMF provides an online tool to help countries convert data into the SDMX format required for their NSDP.

#1 Dimension, Attributes and Time Periods appear as header columns

#2 If any Dimensions or Attributes have fixed values for the whole dataset, these may be placed in a Header section.

Shown below are fixed values for DATA_DOMAIN, REF_AREA and COUNTERPART_AREA

 

#3 Multiple Frequencies are supported in the same worksheet (Annual and Monthly for example).  The frequency of the reported values is derived from the date formats. 

Note: When using SDMX Central to derive the Frequency from the Time Period column, use the highest frequency reported in the FREQ Dimension.

When the data is read, the value for the dimension FREQ will be derived from the corresponding date format. The following rules are used:

 

#4 It is permissible to have multiple worksheets with data. This mechanism can be used to report different frequencies of data per worksheet. 

 The Header section of each worksheet must be consistent in terms of layout, as shown in the image below.

 

#5 Observation Attributes may be reported in the header section, which applies this as a default value for all observations. 

#6 The header section should be separated from the data section by a blank row

 

#7 Reported values appear in the data section

#8 Dimension values are mandatory.  If a value is not reported, this will result in an error. 

#9 Extra rows and in the spreadsheet may be added to improve the readability for the user. Blank rows are permitted but with certain restrictions.  Blank rows may appear before the header section and between the header section and data section.  However, a blank row may not exist within the header section.  If a blank row is encountered in the header section, then this is assumed to indicate the end of the header section and this may cause your spreadsheet to be read incorrectly.

#10 Blank columns indicate that no further information should be read from that row. IMF SDMX Central will read from the first column of information in a row until it reaches a blank cell (unless it’s a data row). The image below shows a spreadsheet where column H is blank and row 9 is blank. This would mean that in the data section only data for 2001 and 2002 are read (columns F and G). Columns I and J will not be read. However, all the 3 series (rows 7,8 and 10) will be read.

#11 It is permissible to have columns in the data section that are not dimensions, attributes or data but may contain additional information for the reader of the spreadsheet.  The image below shows a spreadsheet where column F is for additional notes for each data row. The presence of this column will not prevent the data (in columns G to J) from being read even though the data rows themselves do not have a value for the row.

 

#12 It is permissible to have entire rows in the data section that are there to indicate what the data represents. This has the restriction that the text in these rows must not be in a column that indicates a Dimension, Attribute or Value. The image below shows row 7 being used to explain what rows 8 and 9 represent. The text for row 7 is in column A, which is now used as additional information.

 

#13 It is permissible to have rows that do not report any observations. In the image above, row 10 reports no values (cells H10, I10, J10 and K10 are all blank).