Survey Methodology & Estimation

Procedure

2.1 INTRODUCTION

2.1.0 The latest survey on household consumption expenditure (previously known as household
consumer expenditure survey) was conducted during the period August 2022 to July 2023 in
which information was collected from each sampled household in three questionnaires, namely,
Questionnaire: FDQ (Food Items), Questionnaire: CSQ (Consumables & Services) and
Questionnaire: DGQ (Durable Items) in three separate monthly visits in a quarter. Apart from
these, another questionnaire, namely, Questionnaire: HCQ was canvassed to collect information
on household characteristics.

2.2 SAMPLING DESIGN

2.2.0 In HCES: 2022-23, a multi-stage stratified sampling design was used where villages/urban
blocks or sub-units of these were regarded as the First Stage Units (FSU) and the households
were the Ultimate Stage Units (USU). Both the FSUs and USUs were selected with Simple
Random Sampling Without Replacement (SRSWOR). 18 sample households were canvassed
within an FSU.

2.2.1 CONCEPT OF PANEL

2.2.1.0 The survey period of HCES: 2022-23 was divided into 10 panels, each consisting of
three months. The compositions of the panels were as below:

Panel number Time period Panel number Time period

Panel 1 August - October 2022 Panel 6 January - March 2023
Panel 2 September – November 2022 Panel 7 February - April 2023
Panel 3 October - December 2022 Panel 8 March - May 2023
Panel 4 November 2022- January 2023 Panel 9 April - June 2023
Panel 5 December 2022- February 2023 Panel 10 May - July 2023

1
2.2.1.1 In each of these panels, an equal number of FSUs were allotted for survey to ensure
uniform spread of the sample over the entire survey period. Because of arduous field conditions,
panel restriction was not strictly enforced in the Andaman and Nicobar Islands, Lakshadweep,
Ladakh and rural areas of Arunachal Pradesh and Nagaland.

2.2.1.2 In the first month of any panel, Questionnaire: HCQ along with any one of the
questionnaires, i.e., FDQ/CSQ/DGQ were canvassed in the selected households. During the
second month of the panel, any one from the remaining two questionnaires was canvassed and
in the last month, the last questionnaire was canvassed. The sequence of the questionnaires to be
canvassed in each month of a panel for a particular FSU was decided randomly to eliminate bias
that may arise due to the adoption of a particular sequencing for canvassing. Thus, all six possible
sequences, i.e., [(Q1, Q2, Q3), (Q1, Q3, Q2), (Q2, Q1, Q3), (Q2, Q3, Q1), (Q3, Q1, Q2) and
(Q3, Q2, Q1)], where Q1 refers to FDQ, Q2 refers to CSQ and Q3 refers to DGQ, were canvassed
at random in the sample households.

2.2.2 CHOICE OF FIRST STAGE UNIT / SAMPLING FRAME FOR FIRST

STAGE UNIT

2.2.2.0 The sampling frame for urban sector is the list of Urban Frame Survey (UFS) blocks as
per latest Urban Frame Survey and for rural sector, it is the list of villages as per Census 2011
updated by removing those villages which are urbanized and included in latest UFS (till the time
of sample selection). Sometimes, with a view to ensure uniformity in the size of FSUs and
operational convenience, large villages/UFS blocks are notionally divided into smaller units of
more or less equal size, known as sub-units depending on a pre-defined criteria based on
population in the village or number of households in the UFS block. The sector-specific criteria
for sub-unit formation are as below:

Rural Sector
(i) The number of SUs to be formed in the villages (with Census 2011 population of 1000 or
more and except some States/UTs) are decided based on projected present population of the
village. The criteria are given below:

2
 Projected Population of the village Number of SUs to be formed
less than 1200 1
1200 to 2399 2
2400 to 3599 3
… …

(ii) For rural areas of Himachal Pradesh, Sikkim, Andaman & Nicobar Islands, Ladakh, Parts of
Uttarakhand (except four districts Dehradun, Nainital, Haridwar and Udham Singh Nagar),
Jammu and Kashmir (seven districts Poonch, Rajouri, Udhampur, Reasi, Doda, Kishtwar,
Ramban) and Idukki district of Kerala; SU is formed in a village if population as per Census
2011 is more than or equals to 500. The criteria for the number of SU to be formed are as
below:
Projected Population of the village Number of SUs to be formed
less than 600 1
600 to 1199 2
1200 to 1799 3
… …

Urban Sector:
(i) SUs are formed in those UFS blocks having more than or equal to 250 households. The
number of SUs to be formed within the UFS blocks is decided by the following criteria:
Number of Households in UFS Block Number of SUs to be formed
less than 250 1
250 to 499 2
500 to 749 3
… …

Thus, the list of Villages / UFS Blocks / Sub-Units (for those villages or UFS blocks where sub-
units are formed within) together formed the sampling frame for First Stage Unit selection.

2. 2.3 STRATIFICATION OF FSUS

2.2.3.0 A State/UT is considered as a basic geographical unit for stratum formation in both the
sectors for HCES:2022-23. The sector-specific stratification criteria are given below:
3
Rural Sector:
(i) A Special Stratum comprising of all the uninhabited villages as per Census 2011 is formed
at All-India level.

(ii) From the remaining villages, two more strata are formed in each State/UT:

(a) Stratum 1: Comprising of the villages within a distance of 5 Kms from the district
headquarter or from a city/town with more than 5 lakh population. This stratum is not
formed if there are less than 50 such villages in the State/UT.

(b) Stratum 2: Rest of the villages.

Urban Sector: UFS blocks within each State/UT are divided into strata based on two criteria
i.e., Population as per Census 2011 and ‘Affluence’ status as indicated in UFS frame. The stratum
compositions are as follows:

Strata Number Strata Composition

Stratum 11: ‘affluent’ UFS blocks of all towns with population less than 50,000

Stratum 12: remaining UFS blocks of all towns with population less than 50,000
‘affluent’ UFS blocks of all towns with population 50,000 or more but
Stratum 21:
less than 3 lakhs
remaining UFS blocks of all towns with population 50,000 or more but
Stratum 22:
less than 3 lakhs
‘affluent’ UFS blocks of all towns with population 3 lakhs or more but
Stratum 31:
less than 10 lakhs
remaining UFS blocks of all towns with population 3 lakhs or more but
Stratum 32:
less than 10 lakhs
Separate stratum for each of the Million plus towns, Odd and even
Stratum 41, 42, 43…
numbers are given for affluent and non-affluent blocks respectively

2.2.4 SUB-STRATIFICATION OF FSUS

2.2.4.0 In the rural sector, three groups of villages are formed within each stratum, except
special rural stratum at all-India level based on the following criteria:

4
 Group Population of the village (as per Census 2011)
1 less than 250
2 250 to 499
3 ≥ 500

2.2.4.1 Further, the sample size for a particular rural stratum is distributed among these 3 groups
in proportion to population and if the group-wise allocation is more than 20, 2 or more sub-strata
are formed in each group. If the number of FSUs in a particular group is very small, no sub-
stratum is formed.

2.2.4.2 In the urban sector, if the sample size allocated for an urban stratum is more than 20, 2
or more sub-strata are formed based on the number of FSUs within that stratum.

2.2.5 SAMPLE SIZE

2.2.5.0 8,723 FSUs in the rural areas and 6,115 in the urban areas have been surveyed in
HCES:2022-23. The total sample size of 15016 FSUs has been allocated to State/UTs in
proportion to Census 2011 population, subject to a minimum allocation of 40 FSUs.

2.2.6 SELECTION OF FSUS WITHIN A STRATUM/SUB-STRATUM

2.2.6.0 The required number of FSUs from each of the sub-strata is selected by SRSWOR
scheme. The selected sample is divided into 10 Panels with 2 dependent sub-samples within each
Panel. Three sequences of questionnaires are canvassed in the FSUs belonging to sub-sample 1
and remaining three sequences are canvassed in FSUs sub-sample 2.

2.2.7 FORMATION OF SUB-DIVISION

2.2.7.0 It has been experienced that in some of the selected FSUs, the actual present population
is significantly higher than the projected population/Census population that causes operational
inconvenience for listing of all the households. In such a situation, the selected FSU is notionally
sub-divided into several smaller units, called Sub-division. The criteria for determining the

5
number of Sub-divisions to be formed in the selected rural (except areas mentioned in Point ii
Para 1.2.2) /urban FSU is given below.
Approx. present population of the selected SU Number of Sub-divisions to be formed
less than 1500 1
1500 to 2399 2
2400 to 3599 3
3600 to 4799 4
… …

2.2.7.1 The criteria for determining the number of Sub-divisions in rural areas mentioned in Point
(ii) of Para 1.2.2 is as below:
Approx. present population of the selected SU Number of Sub-divisions to be formed
less than 750 1
750 to 1199 2
1200 to 1799 3
1800 to 2399 4
… …

Only one Sub-division is selected randomly after forming the required number of Sub-divisions.
Further, listing and selection of households are done in the selected Sub-division unit only.

2.2.8 FORMATION OF SECOND STAGE STRATA (SSS)

2.2.8.0 In order to ensure proper representation of households of different economic categories

in the sample, all the households of a selected FSU are classified into three second stage stratum
(SSS) with the following criteria:
SSS Rural Urban
Households with land possessed > X, Households owned one or more
where X is such that 5% of the households four-wheeler cars for non-
SSS 1 had land possessed more that X as per NSS commercial use as on date of listing
77th round SAS survey. This was whose combined purchase price was
determined at State/UT level more than Rupees 10 lakhs

6
 SSS Rural Urban
Households with land possessed > Y and
Households owned one or more
less than or equal to X, where Y is such
four-wheeler cars for non-
that 20% of the households had land
SSS 2 commercial use as on date of listing
possessed more that Y as per NSS 77th
whose combined purchase price was
round SAS survey. This was determined at
less than or equal to Rupees 10 lakhs
State/UT level
SSS 3 Remaining households Remaining households

A total of 18 households are surveyed from each of the selected FSUs. Allocation of the sample
households among the three SSS is done in proportion to the number of households listed in the
respective SSS.

2.2.9 SELECTION OF HOUSEHOLDS

2.2.9.0 The sample households from each SSS are selected by the SRSWOR scheme.

7
3. ESTIMATION PROCEDURE

3.1 NOTATIONS:
s Subscript for sth stratum
t Subscript for tth sub-stratum
r Panel number [ r= 1(1)10]
m Subscript for sub sample
i Subscript for ith FSU [SU/ village /panchayat ward/ block]
j Subscript for jth second stage stratum in an FSU
Subscript for kth sample household under a particular second stage strata within
k
an FSU/sub-division
Total number of sub-divisions formed in the sample FSU. D1 = 1, if no Sub-
𝐷1
division is formed in the SU
N Total number of sub-unit (FSU) in any rural/urban sub-stratum
Number of sample FSUs surveyed including ‘uninhabited’ and ‘zero cases’ but
n
excluding casualty for a particular sub-stratum
H Total number of households listed in a second-stage stratum of an FSU
h Number of households surveyed in a second-stage stratum of an FSU
x, y Observed value of characteristics x, y under estimation
X̂ , Yˆ Estimate of population total X, Y for the characteristics x, y

Let, ystijkrm be the observed value of the characteristic y for the kth household of the jth second
stage stratum of the ith FSU of rth panel and mth sub-sample for the tth sub-stratum of sth stratum.
For ease of understanding, a few symbols have been suppressed in following paragraphs where
they are obvious.

3.2 FORMULAE FOR ESTIMATION OF AGGREGATES FOR A

STRATUM  SUB-STRATUM:

3.2.1 Questionnaire LHQ (Rural/Urban):

3.2.1.1 For estimating the number of households in a stratum  sub-stratum possessing a
characteristic:

8
 𝑁
𝑌̂=∑10 2 𝑛
𝑟=1 ∑𝑚=1[ 𝑛 × ∑𝑖=1 𝐷1 × 𝑦𝑖𝑟𝑚 ]

where yirm is the total number of households possessing the characteristic y in ith FSU of rth panel
and mth sub-sample respectively.

3.2.2 Formulae for Estimation of Aggregates for a stratum * sub-stratum

Questionnaire HCQ, FDQ, CSQ and DGQ:

3.2.2.1 For jth second-stage stratum of a stratum  sub-stratum:

N nj  
hij
H ij
Yˆj  r 1 m 1[  y
10 2
 D1 * ijkrm ]
nj i 1 
 hij k 1 

where nj is the number of sample FSUs with non-void jth second-stage stratum.

3.2.2.2 Aggregate Yˆ is obtained combining all the second-stage strata:

3
Yˆ   Yˆj
j 1

Note: Values of j for all the questionnaire is 3.

3.3 OVERALL ESTIMATE FOR AGGREGATES FOR A STRATUM× SUB-

STRATUM:

3.3.1 Overall estimate for a stratum × sub-stratum ( Ŷst ) will be obtained as

10 2 𝑠𝑡𝑗 𝑛 𝑁𝑠𝑡 𝐻𝑠𝑡𝑖𝑗𝑟𝑚 ℎ
𝑠𝑡𝑖𝑗𝑟𝑚
𝑠𝑡 = ∑𝑟=1 ∑𝑚=1[∑𝑖=1
𝑦̂ × 𝐷1 × ∑3𝑗=1 × ∑𝑘=1 𝑦𝑠𝑡𝑖𝑗𝑘𝑟𝑚 ]
𝑛𝑠𝑡𝑗 ℎ𝑠𝑡𝑖𝑗𝑟𝑚

3.3.2 Overall Estimate for Aggregates for a stratum:

𝑦̂𝑠 =∑𝑡 𝑦̂
𝑠𝑡

3.4 OVERALL ESTIMATE OF AGGREGATES AT STATE/UT/ALL-

INDIA LEVEL:
The overall estimate Ŷ at the State/ UT/ all-India level is obtained by summing the stratum

estimates Yˆs over all strata belonging to the State/ UT/ all-India.

𝑌̂ = ∑ 𝑦̂𝑠

9
3.5 ESTIMATION OF MONTHLY PER CAPITA CONSUMER
EXPENDITURE (MPCE) FROM THREE QUESTIONNAIRES:

3.5.0 The monthly per capita consumption expenditure (MPCE) of a household, defined as “the
total value of monthly consumption of goods & services of the household divided by the number
of members of the household” is based on the data collected from those households who provided
information for all the three visits, i.e., the households from whom data pertaining to all the three
questionnaires, namely, FDQ, CSQ and DGQ are available. The multiplier computed from the
data of third visit is used for generation of the estimates. Household size recorded at the time of
canvassing of food questionnaire (FDQ) is considered for calculating MPCE.

3.5.1 Let E1, E2 & E3 be the total value of expenditure on food, consumable & services and
durables items respectively for a particular sampled household, compiled from the three
questionnaires, namely FDQ, CSQ & DGQ. Further, assume that P1, P2 & P3 be the number of
household members (i.e., size of the household) recorded during the time of canvassing FDQ,
CSQ and DGQ questionnaire, respectively.

3.5.2 Then the total monthly expenditure of the household, say, TE is computed as:
TE = E1 + (E2/P2) ×P1 + (E3/P3) ×P1

3.5.3 Consequently, the MPCE of the household is computed as:

MPCE = TE/P1.

3.6 ESTIMATES OF RATIOS:

3.6.0 Let Ŷ and X̂ be the overall estimates of the aggregates Y and X for two characteristics y

and x respectively at the District/State/ UT/ all-India level.

Yˆ
Then the combined ratio estimates (Rˆ ) of the ratio ( R  ) is obtained as Rˆ 
Y
.
X Xˆ

10
3.7 ESTIMATION OF ERRORS:

3.7.1 Formula for estimated variance (for Rural/Urban)

3.7.1.1 FSUs and USU (households) are selected by SRSWOR method. If ith FSU has been
selected, hi unit is selected from this particular FSU x SSS by SRSWOR method. For simplicity
of formula some obvious suffixes (rth panel, mth sub-sample) have been suppressed in following
formulas for RSE (%) calculation.

(a) Formula for aggregate Ŷ (for Rural/Urban)

ℎ
∑ 𝑖𝑗 𝑦
̂
𝑌𝑖𝑗 = 𝐻𝑖𝑗 𝑦𝑖𝑗 ∗ 𝐷1𝑠𝑖 and 𝑦𝑖𝑗 = 1 𝑖𝑗𝑘⁄ℎ
∗ ̅̅̅
𝑖𝑗

Vaˆr (Yˆ )  Vaˆr (Yˆs)   Vaˆr (Yˆstj)

s s t j

2 1 1 n 1 n
̂ (Ŷ
var stj ) = Nst ( n – Nst
)(1/(nstj - 1)) ∑1stj (Hstij *D1sti * ̅̅̅̅̅
ystij – nstj
∑1 stj Hstij *D1sti ̅̅̅̅̅
ystij )2+
stj

Nst nstj
Hstij 2 *D1sti 2 ( 1 1
nstj
∑1
hstij
- Hstij *D1si
) s2 wij
1 𝑠𝑡𝑖𝑗 ℎ
where 𝑠 2 𝑤𝑖𝑗 = ∑𝑘=1 (𝑦𝑠𝑡𝑖𝑗𝑘 – ̅̅̅̅̅̅)2
𝑦𝑠𝑡𝑖𝑗
(ℎ𝑠𝑡𝑖𝑗 −1)

(b) Formula for ratio R̂ (for Rural/Urban)

̂ is unbiasedly estimated by V(𝑌̂ − R𝑋̂)
It may be noted that X2 MSE(𝑅)
V(𝑌̂ − R𝑋̂) = v(𝑢̂) where uijk = (yijk – R xijk ) , U𝑖 = (Y𝑖 – R X𝑖 ) and U = (Y − RX) = 0 at
domain level (State).
̂2 𝑀𝑆𝐸
𝑋 ̂ (𝑅̂) = 𝑉̂ (𝑈
̂ ) at R = 𝑅̂
1 1
𝑌̂
𝑠𝑡𝑖𝑗 =
̂
∗ ∑𝑘 𝑦𝑠𝑡𝑖𝑗𝑘 ∗ 𝑛𝑠𝑡𝑗 ∗ multiplier 𝑎𝑛𝑑 𝑋𝑠𝑡𝑖𝑗 = ∗ ∑𝑘 𝑥𝑠𝑡𝑖𝑗𝑘 ∗ 𝑛𝑠𝑡𝑗 ∗ multiplier
𝑁𝑠𝑡 𝑁𝑠𝑡

1
MSˆE ( Rˆ )  2  MSˆEst ( Rˆ )
Xˆ s t
𝑛 𝑛
̂𝑠𝑡 (𝑅̂) =∑𝑗 𝑁𝑠𝑡 2 ( 1 −
𝑀𝑆𝐸
1
) (𝑛
1
∑ 𝑠𝑡𝑗 (𝐻𝑖𝑗 𝐷1𝑠𝑖 ̅̅̅̅
𝑢𝑖𝑗 - 𝑛
1
∑1 𝑠𝑡𝑗 𝐻𝑖𝑗 𝐷1𝑠𝑖 ̅̅̅̅
𝑢𝑖𝑗 ) 2
𝑛 𝑠𝑡𝑗 𝑁𝑠𝑡 𝑠𝑡𝑗 −1) 1 𝑠𝑡𝑗

𝑁 𝑛 1 1
+∑𝑗 𝑛 𝑠𝑡 ∑1 𝑠𝑡𝑗 𝐻2 𝑖𝑗 ∗ 𝐷1𝑠𝑖 2 ( ℎ − ) 𝑠𝑢𝑖𝑗 2
𝑠𝑡𝑗 𝑖𝑗 𝐻𝑖𝑗 ∗𝐷1𝑠𝑖

1 ℎ𝑖𝑗
where 𝑠𝑢𝑖𝑗 2 = (ℎ ∑𝑘=1 (𝑢𝑖𝑗𝑘 − ̅̅̅̅̅̅
𝑢𝑖𝑗 ) 2
; ̅̅̅̅ 𝑦𝑖𝑗 - 𝑅̂ ̅̅̅̅
𝑢𝑖𝑗 = ̅̅̅ 𝑥𝑖𝑗
𝑖𝑗 −1)

11
3.7.2 Estimates of Relative Standard Error (RSE):

 
Vaˆr Yˆ  MSˆE Rˆ  
 
RSˆE Yˆ 
Yˆ
 100 RSˆE Rˆ 
Rˆ
100

4. MULTIPLIERS:

4.0 The formulae for multipliers at stratum/sub-stratum/second-stage stratum level for a

Questionnaire type are given below.

Questionnaire Sector Formula for multipliers

N
st
LHQ Rural/urban
n st

N H
st stij
Rural/urban * D1 *
HCQ, FDQ, CSQ n stj h
stij
and DGQ
j = 1, 2,3

Note: Multipliers are computed on the basis of information available in the listing Schedule
LHQ, irrespective of any misclassification observed between the listing Schedule and
detailed enquiry Schedule.

12
Table A1: Number of FSUs and number of households surveyed
No. of fsu’s
No. of sample households
State/UT (villages/blocks)
Rural Urban Rural Urban Rural + Urban
Andhra Pradesh 357 235 6,245 4,025 10,270
Arunachal Pradesh 158 80 2,581 1,440 4,021
Assam 338 140 6,045 2,517 8,562
Bihar 759 199 13,602 3,564 17,166
Chhattisgarh 160 120 2,867 2,141 5,008
Delhi 20 200 305 2,931 3,236
Goa 20 18 360 323 683
Gujarat 320 317 5,726 5,560 11,286
Haryana 156 139 2,796 2,472 5,268
Himachal Pradesh 80 60 1,406 1,036 2,442
Jharkhand 219 140 3,927 2,458 6,385
Karnataka 378 339 6,688 5,701 12,389
Kerala 217 199 3,870 3,507 7,377
Madhya Pradesh 479 320 8,551 5,644 14,195
Maharashtra 655 653 11,596 11,163 22,759
Manipur 143 126 2,572 2,261 4,833
Meghalaya 120 60 2,132 1,079 3,211
Mizoram 80 120 1,439 2,157 3,596
Nagaland 111 60 1,996 1,079 3,075
Odisha 379 140 6,732 2,453 9,185
Punjab 177 160 3,076 2,754 5,830
Rajasthan 497 260 8,724 4,438 13,162
Sikkim 80 40 1,411 720 2,131
Tamil Nadu 416 393 7,447 6,917 14,364
Telangana 198 180 3,553 3,233 6,786
Tripura 179 100 3,222 1,800 5,022
Uttarakhand 98 60 1,700 1,073 2,773
Uttar Pradesh 1096 598 19,611 10,627 30,238
West Bengal 599 419 10,715 7,421 18,136
Andaman & Nicobar Islands 37 20 644 356 1,000
Chandigarh 20 20 360 360 720
Dadra & Nagar Haveli and Daman & Diu 20 20 350 324 674
Jammu & Kashmir 99 100 1,761 1,772 3,533
Ladakh 20 20 359 360 719
Lakshadweep 14 20 252 355 607
Puducherry 20 40 359 711 1,070
All - India 8,723 6,115 155,014 106,732 261,746
Note: Special rural stratum allocation at all-India level is included in rural allocation.

13