Question
The "observed" attribute on each time series denotes how the measurement should be interpreted and controls how Data Buffet performs high-to-low frequency conversion.
Answer
Where do I see this?
- In Mnemonic 411, as "tab:Overview » row:Observed Attribute."
- In the basket editor, activate the optional header field with "Options » Fields » General » Conversion Method."
- In the basket output, as the "Observed:" field.
Values of the "observed" attribute
LAST series measure a quantity at the end of a period.
- Annual population is usually measured as of December 31 (but check Mnemonic 411 for the particulars of a specific series).
- A monthly employment series measures the employment at the end of the month.
- "Balance" and "stock" series are usually LAST.
- Financial share price indexes are usually LAST, because users generally want to know the closing value of the day, week, or month.
- Year to date (YTD) series, such as XHPRNR.US, are always set to LAST.
BEGINNING series are the reverse of LAST, series, with measurement as of the start of the period. Such series are uncommon.
SUMMED series measure quantity over the extent of an interval.
- Trade data, such as exports value or volume, is the sum of export activity over a period. For example, TRLEXEXMUM.INZL.
- "Transaction" and "flow" series are usually SUMMED.
AVERAGED series measure the average activity over an interval.
- Price indexes (consumer price, house price, etc.) are usually, but not always, AVERAGED. For example, XHOFHOPIQ.US.
Advisory: Some series, by the nature of their measurement, cannot be sensibly converted (for example, MBA quarterly foreclosure starts and RealtyTrac monthly foreclosure inventory). Data Buffet does not have an observed attribute to express this, and cannot stop you from performing a conversion. Please examine the written documentation in Mnemonic 411.
Implications for high-to-low frequency conversion
LAST series (monthly to quarterly): pull the last observation that occurs during the target interval.
SUMMED series: sum the observations over the target interval.
AVERAGED series: take the arithmetic average of observations.
- Q3 = (Jul + Aug + Sep) / 3
References