Abstract
For about 125 items of food, the Consumer Expenditure Survey (CES) schedule of the Indian National Sample Survey asks the interviewer to obtain both quantity and value of household consumption during the reference period from the respondent. This would appear to put a great burden on the respondent and call for reduction in the number of items in the schedule. But it is likely that interviewers actually proceed by asking the respondent to recall quantity and price usually paid (instead of quantity and value) and multiplying the two to get value, item-wise, as price usually paid might be easy to recall, and survey protocol does not disallow it. Whether this is done and, if so, how frequently, is of obvious importance to the planners of this important living standards survey, as efforts to reduce the number of items to lighten the respondent burden continue. In this study, a method, using unit records for vegetable items in the NSS’s 2011–2012 CES, is devised to estimate the proportion of interviews in which values of vegetable consumption were in fact determined by the multiplication method. The findings suggest that this method was much more prevalent than “independent” recall of values. This paper concludes that the survey would benefit if the price-and-quantity method were explicitly laid down as the method to be followed to obtain value, as it releases valuable interview time for more important items. Research using unit values from NSS CES data as a proxy for prices would also interpret and use such unit values better, were the practice to be followed uniformly.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
The NSS CES serves mainly as a living standards measurement survey. The estimated distribution of consumption expenditure over households is used in the study of poverty and inequality in the country, while the item-wise expenditure estimates are used in the weighting scheme for construction of consumer price index numbers.
“…Several studies on end-digit preferences … suggest that a strong preference exists for 0 as the last terminal digit in a number set” [7]. See also Preese [10]. Digit preference in medical data is discussed in Indrayan and Holt [5], who point out that “the net effect of the digit preference is that the values ending with 0 or 5 have higher frequency (spikes) than what are actually present. This can cause bias in the result”. Digit preference in age reporting has attracted the attention of demographers for a long time (see [6, 11]). “Digit preference has been quantified in demography using statistics that are based on comparing the size of the enumerated population to the population one would expect to see in the absence of digit preference” [1]. Rounding of income data and its impact on data quality have been studied in depth in Hanisch [4].
There is plenty of evidence in the data of values being reported after multiplication and rounding of the product to the nearest rupee. For instance, among 2857 “potato” records with q = 2500, there are 46 records with v = 22 and 51 records with v = 23. Such unusual v figures (when people usually report figures like Rs. 20 or Rs. 25) are difficult to explain until one sees that if 2500 g is multiplied by Rs. 9/kg, one gets Rs. 22.50, which may be rounded to 22 or 23 according to personal preference. (Unit value is exactly Rs. 9/kg in another 978 “potato” records, showing that Rs. 9/kg is a common price for potatoes.)
Whether an interview is an M1 or an M2 interview might be expected to depend mainly on the interviewer but may occasionally depend on the household too (or on both). Given the population H of households under survey coverage, and the set of interviewers engaged for data collection, this gives rise to a probability of an interview being a M2 interview which we call p. (In a very simple situation where the interviewers determine the interview type and the household does not play a role, p will be the proportion of what may be called the M2 interviewers in the set.)
In all there were 50,657 households reporting at least one vegetable item with q a multiple of 3. This breaks up into A1, A2, A3, etc. The An’s declined rapidly with increase in n. Thus, A6, A7, …, A10 were, respectively, 143, 33, 8, 2 and 1. We have excluded the households with n ≥ 8 from our analysis because of the low frequencies A8, A9 and A10.
Such unit values have been used, for instance, by the Tendulkar Committee [2] for the purpose of setting state and national poverty lines for India, to generate state-to-all-India and rural-to-urban price indices.
References
Alho J, Spencer B (2005) Statistical demography and forecasting. Springer series in statistics. Springer, Berlin
Commission Planning (2009) Report of the expert group to review the methodology for estimation of poverty. Government of India, Planning Commission, Delhi, p 2009
Deaton A (1998) Getting prices right: What should be done? J Econ Perspect 12(1):37–46
Hanisch JU (2007) Rounding of income data: an empirical analysis of the quality of income data with respect to rounded values and income brackets with data from the European community household panel. Peter Lang Pub Incorporated, New York
Indrayan A, Holt MP (2016) Concise medical encyclopaedia of biostatistics for medical professionals. CRC Press, Boca Raton
Institute for Resource Development (1990) An assessment of DHS-I data quality. DHS Methodological Reports No. 1, Columbia
McNabb DE (2014) Non-sampling error in social surveys. Sage Publications Inc., Thousand Oaks
NSSO (2014) Level and pattern of consumer expenditure 2011–2012, NSS 68th round (July 2011–June 2012). Ministry of Statistics and Programme Implementation, Government of India, Delhi, p 2014
Perali F (2003) The behavioral and welfare analysis of consumption. Kluwer Academic Publishers, Amsterdam
Preese DA (1981) Distributions of final digits in data. Statistician 30(1):31–60
Shryock HS, Larmon EA, Siegel JS (1973) The methods and materials of demography, vol I. United States Bureau of the Census, Washington
Slesnick D (1998) Empirical approaches to the measurement of welfare. Journal of Economic Literature 36(4):2108–2165
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Chaudhury, P., Mukherjee, D. Are Values Reported Using Quantities and Prices in Consumption Expenditure Data?. J Stat Theory Pract 13, 40 (2019). https://doi.org/10.1007/s42519-019-0040-0
Published:
DOI: https://doi.org/10.1007/s42519-019-0040-0