Reference Period in Household Surveys

A reference period is the time interval before the interview for which a household reports its consumption expenditure. Its length influences what respondents remember and whether occasional purchases fall within the reporting window, thereby affecting estimated monthly per capita consumption expenditure and its distribution.

Why estimates change

The choice of period creates a trade-off between memory accuracy and coverage of irregular expenditure.

• A longer recall period can cause recall lapse, especially for frequent, low-value purchases, leading to under-reporting or inaccurate amounts.
• A very short period reduces memory burden but makes estimates sensitive to the timing of purchases, festivals, illness and other temporary events.
• For infrequently purchased items, a short period produces many zero reports, while a longer period captures more transactions but may increase recall error.
• Incorrect placement of expenditure inside or outside the stated interval, called telescoping, can also bias reported consumption.

Alternative recall designs

Household surveys may use a single period for all items or match different periods to different consumption patterns.

• Under the Uniform Reference Period, consumption of all items is collected using a 30-day recall period.
• The Mixed Reference Period uses a 365-day recall for selected infrequently purchased items and a 30-day recall for the remaining items.
• The Modified Mixed Reference Period adds a 7-day recall for selected frequently consumed food items, while retaining 30-day and 365-day periods for other groups.
• Mixed designs seek to reduce recall error for frequent purchases while adequately covering durable and other occasional expenditure.

Implications for welfare measurement

Reported expenditure for different reference periods is converted to a common monthly basis, but this arithmetic does not remove the underlying reporting effects.

• Changing the reference-period design can alter estimated consumption levels, expenditure shares and the measured rural-urban or inter-group differences.
• It can consequently affect poverty and inequality estimates derived from the expenditure distribution, even when actual living conditions have not changed.
• Valid comparisons across surveys require attention to reference-period comparability, questionnaire design and valuation procedures.