ISSN (Print) - 0012-9976 | ISSN (Online) - 2349-8846

A+| A| A-

What Does the Recent Indian Consumption Behaviour Tell?

The paper presents alternative estimates of poverty lines and head count ratios for India and its states and union territories, not only with respect to the fixed all-India calorie norm, but also by considering the state-specific benchmarks. These alternative estimates are found to be much higher than the official estimates. Given the extent of divergence of such a large magnitude, it is high time to revisit the methodological issues of estimation of poverty lines and head count ratios.

SPECIAL ARTICLE

What Does the Recent Indian Consumption Behaviour Tell?

G C Manna, S K Samanta, Dipankar Coondoo

The paper presents alternative estimates of poverty lines and head count ratios for India and its states and union territories, not only with respect to the fixed all-India calorie norm, but also by considering the state-specific benchmarks. These alternative estimates are found to be much higher than the official estimates. Given the extent of divergence of such a large magnitude, it is high time to revisit the methodological issues of estimation of poverty lines and head count ratios.

The authors are thankful to the Union Ministry of Statistics and Programme Implementation for making available the household level data on consumption expenditure based on NSS 61st round required for preparing this article. The support provided by Siladitya Chaudhuri in extracting the relevant information from the full data set is acknowledged. Views expressed in this article are of the authors and not of the organisations to which they belong.

G C Manna (gc.manna1@gmail.com) is with the National Sample Survey Organisation; S K Samanta is with the Department of Statistics, Kalyani University, West Bengal and Dipankar Coondoo is with the Indian Statistical Institute, Kolkata.

1 Introduction

I
n Manna (2007), a set of revised estimates of average per capita daily calorie requirements for the Indian population was derived. The suggested calorie norms were rounded off to 2,290 and 2,250 calorie per capita per day for people r esiding in rural and urban areas, respectively. These correspond to calorie norms of 2,400 and 2,100 calories per day per capita for rural and urban underlying the official poverty lines based on which official estimates of incidence of poverty for India in terms of head count ratio (HCR) are made. In that paper, state-specific calorie norms based on state-specific age-sex-occupation distribution of persons were also suggested. These suggested norms were based on (i) a more realistic c lassification of different types of work as argued by the author, (ii) the latest (1993-94 and 1999-2000) age-sex-occupation d istribution of persons, and (iii) the latest (1998) recommendations on calorie norms by the expert group of the Indian C ouncil of Medical Research (ICMR). It may be mentioned here that since a given amount of calories can result from a lternative bundles of food items consumed, corresponding to a given calorie norm there is a conditional distribution of monthly per capita consumption expenditure (MPCE) and a lower prediction limit of MPCE corresponding to that given amount of calories may be defined as the lower limit of the MPCE interval such that a household selected at random from the universe of all households having the given calorie i ntake will have an MPCE in that interval with a given probability. The present paper inter alia focuses on setting state-level lower p rediction limits of household level MPCE and the corresponding lower confidence limits of average MPCE required to achieve the suggested state-specific calorie norms.1 These threshold limits have been worked out by analysing the household level consumption expenditure data collected in the latest quinquennial round of the National Sample Survey (viz, NSS, 61st round, 2004-05).2 The percentages of persons below the said threshold MPCE limits have also been worked out for i ndividual states/UTs and compared with the corresponding o fficial estimates of HCR.

The paper is organised as follows. Section 2 briefly describes the data and the related methodology of estimation of household MPCE and per capita daily calorie intake used in the present analysis. Section 3 discusses the methodology adopted for setting the lower prediction limits of household level MPCE and corresponding lower confidence limits of average MPCE associated with the state-specific calorie norms.

august 8, 2009 vol xliv no 32

Whether the given household level data on MPCE follow a log- These various MPCE levels have then been compared with the of
normal distribution – an assumption made in Section 3 – has ficial estimates of poverty lines. Section 6 then presents state
been examined in Section 4. Section 5 presents state-specific es wise alternative estimates of the proportion of persons below dif
timates of threshold limits of household MPCE and average MPCE ferent MPCE levels calibrated in Section 5 and compare their di
as well as estimated average MPCE corresponding to state-specific vergences with the corresponding official estimates of HCRs. The
calorie norms. In addition, Section 5 also gives alternative esti paper is concluded in Section 7.
mates of state-specific poverty lines associated with the calorie
norm underlying the official poverty lines (viz, 2,400 and 2,100 2 Data Set and Its Limitations
calories for the rural and urban populations, respectively) as well As already mentioned, for the present exercise, we have used NSS
as the suggested calorie norm (viz, 2,290 and 2,250 calories for 61st round household level consumption expenditure data. The
the rural and urban populations, respectively) based on the cur survey covered the whole of the Indian Union except a few inacces
rent data set (i e, the NSS 61st round data for the year 2004-05).3 sible areas of the country.4 This data set relates to a representative
Table 1R: Results of Log Normality Test of the Household MPCE Distribution Table 1U: Results of Log Normality Test of the Household MPCE Distribution
for the Rural Areas for the Urban Areas
State/UT PCCR* Number of Sample K Value of χ2 for the State/UT PCCR* Number of Sample K Value of χ2 for the
Households Truncated Data Set Households Truncated Data Set
Full Truncated Observed Critical Value Full Truncated Observed Critical Value
Data Set Data Set Value (α = 0.01) Data Set Data Set Value (α = 0.01)
(1) (2) (3) (4) (5) (6) (7) (1) (2) (3) (4) (5) (6) (7)
Andhra Pradesh 2,425 5,555 142 8 10.637 15.087 Andhra Pradesh 2,256 2,876 76 8 3.101 15.087
Arunachal Pradesh 2,161 1,503 47 5 3.004 9.210 Arunachal Pradesh 2,158 540 61 5 11.709** 9.210
Assam 2,252 3,350 144 6 3.221 11.345 Assam 2,197 900 64 8 8.091 15.087
Bihar 2,238 4,354 184 8 13.606 15.087 Bihar 2,185 1,398 87 8 7.156 15.087
Chhattisgarh 2,312# 1,997 48 5 0.729 9.210 Chhattisgarh 2,233# 799 56 6 5.019 11.345
Goa 2,332 160 29 4 0.477 6.635 Goa 2,295 238 43 5 2.659 9.210
Gujarat 2,335 2,320 56 6 6.083 11.345 Gujarat 2,256 1,955 60 6 0.717 11.345
Haryana 2,207 1,680 53 4 14.225** 6.635 Haryana 2,232 1,040 36 5 5.055 9.210
Himachal Pradesh 2,244 2,143 97 7 9.676 13.277 Himachal Pradesh 2,227 400 32 5 0.852 9.210
Jammu and Kashmir 2,217 1,882 33 5 10.473** 9.210 Jammu and Kashmir 2,163 884 48 6 2.864 11.345
Jharkhand 2,238# 2,379 95 7 4.430 13.277 Jharkhand 2,185# 1,040 30 4 4.515 6.635
Karnataka 2,372 2,880 62 6 25.397** 11.345 Karnataka 2,263 2,227 54 6 6.860 11.345
Kerala 2,318 3,300 64 4 0.579 6.635 Kerala 2,293 1,950 36 5 1.499 9.210
Madhya Pradesh 2,312 3,838 105 9 4.367 16.812 Madhya Pradesh 2,233 2,075 57 7 3.957 13.277
Maharashtra 2,363 5,014 104 7 1.714 13.277 Maharashtra 2,277 4,993 135 9 7.463 16.812
Manipur 2,202 2,177 186 7 11.659 13.277 Manipur 2,193 1,000 72 7 2.850 13.277
Meghalaya 2,218 1,159 44 5 0.895 9.210 Meghalaya 2,167 437 28 4 2.320 6.635
Mizoram 2,187 800 52 4 4.392 6.635 Mizoram 2,201 1,112 37 4 0.314 6.635
Nagaland 2,210 960 59 5 10.566** 9.210 Nagaland 2,150 320 27 4 1.947 6.635
Orissa 2,312 3,836 108 9 9.376 16.812 Orissa 2,257 1,187 45 5 14.530** 9.210
Punjab 2,248 2,433 100 6 2.876 11.345 Punjab 2,266 1,855 55 6 5.134 11.345
Rajasthan 2,194 3,541 132 7 5.332 13.277 Rajasthan 2,209 1,630 63 7 4.013 13.277
Sikkim 2,240 920 48 5 6.555 9.210 Sikkim 2,254 200 41 5 5.976 9.210
Tamil Nadu 2,467 4,159 71 6 5.003 11.345 Tamil Nadu 2,309 4,137 82 8 7.753 15.087
Tripura 2,285 1,760 41 4 5.547 6.635 Tripura 2,195 560 35 4 1.804 6.635
Uttarakhand 2,172# 1,465 58 6 5.802 11.345 Uttarakhand 2,200# 750 28 4 2.018 6.635
Uttar Pradesh 2,172 7,868 126 9 10.828 16.812 Uttar Pradesh 2,200 3,345 103 9 6.135 16.812
West Bengal 2,277 4,988 141 7 12.791 13.277 West Bengal 2,281 2,889 80 8 12.718 15.087
Andaman and Andaman and
Nicobar Islands 2,268 268 35 5 5.461 9.210 Nicobar Islands 2,282 359 52 5 1.295 9.210
Chandigarh 2,475 80 31 4 0.270 6.635 Chandigarh 2,226 300 41 5 4.201 9.210
Dadra and Nagar Haveli 2,281 160 22 4 0.632 6.635 Dadra and Nagar Haveli 2,274 80 27 4 1.491 6.635
Daman and Diu 2,467 80 20 4 4.409 6.635 Daman and Diu 2,270 80 33 4 1.044 6.635
Delhi 2,263 59 23 4 0.203 6.635 Delhi 2,230 1,101 57 5 2.196 9.210
Lakshadweep 2,176 70 21 4 0.642 6.635 Lakshadweep 2,212 129 25 4 0.645 6.635
Pondicherry 2,428 160 26 4 0.618 6.635 Pondicherry 2,313 560 32 4 0.060 6.635
All-India 2,286 79,298 All-India 2,249 45,346
* PCCR: State-specific average daily per capita calorie requirement, Manna (2007). * PCCR: State-specific average daily per capita calorie requirement, Manna (2007).
# PCCR for Chhattisgarh, Jharkhand and Uttaranchal have been respectively taken as the same # PCCR for Chhattisgarh, Jharkhand and Uttaranchal have been respectively taken as the same
for those of earlier undivided Madhya Pradesh, Bihar and Uttar Pradesh. for those of earlier undivided Madhya Pradesh, Bihar and Uttar Pradesh.
k: Number of mutually exclusive classes of MPCE formed to perform chi-square test to the k: Number of mutually exclusive classes of MPCE formed to perform chi-square test to the
truncated data set. truncated data set.
** Hypothesis that MPCE follows lognormal distribution gets rejected. ** Hypothesis that MPCE follows lognormal distribution gets rejected.
Economic & Political Weekly august 8, 2009 vol xliv no 32 39
EPW

sample of 79,298 rural households and 45,346 urban house-(institutional) and durable goods, and “last 30 days” for remainholds, spread over a stratified random sample of 7,999 villages ing items of consumption. Here we have used data for last 30 and 4,602 urban blocks, covered in the central sample.5 For days for all item groups, including the five infrequently purd etails of sample design, etc, see NSS Report Number 508 chased item groups. For each household, MPCE is obtained by (December 2006): Level and Pattern of Consumer Expenditure, dividing the aggregate value of monthly consumption expendi2004-2005. ture of the household by the corresponding household size (i e,

In the survey, expenditure incurred by a household on the number of persons normally living together and taking food domestic consumption during the reference period was con-from a common kitchen). Henceforth, the term MPCE will be sidered as the household’s consumption expenditure. The refer-used to refer to MPCE based on the uniform reference period of ence periods used for collection of consumption data were both last 30 days. “last 30 days” and “last 365 days” for relatively infrequently pur-Apart from the value of total consumption, the total quantity chased items, viz, clothing, footwear, education, medical care of consumption during the last 30 days for each of the detailed

items of food, pan, tobacco and intoxi-

Table 2R: Threshold MPCE Limits and Alternative Poverty Lines for the Rural Areas (2004-05) cants was also collected at the house
State/UT Threshold MPCE Limits (Rs) Associated with Alternative Poverty Line (Rs) Based on Current (NSS: 2004-05) Data Official Poverty Line* (Corresponding to hold level in the survey. Converting
SPCCR# as Per LSM and Different Methods and Calorie Norms Fixed All-India these quantities into calorie levels and
Lower Prediction Lower Confidence As Per LSM and Associated As Per IIM and Associated As Per IIM and Suggested As Per IIM and Official Calorie Norm of aggregating over items, total calorie
Limit of Household MPCE Limit of Average MPCE with SPCCR (PL1) with SPCCR (PL2) Fixed All-India Calorie Norm Fixed All-India Calorie Norm 2,400 Cal) intake during last 30 days for each
(1-α = 0.99) (1-α = 0.99) of 2,290 Cal of 2,400 Cal sample household is obtained. In our
(1) (2) (3) (4) (5) (PL3)(6) (PL4)(7) (8) analysis, we have used these house-
Andhra Pradesh 258.13 656.67 772.18 966.56 784.20 932.78 292.95 hold calorie intake data. The conver-
Arunachal Pradesh 255.07 599.27 742.85 655.24 715.57 767.02 sion factors used for converting quan

tities consumed to calorie intake are

Assam 323.96 609.08 667.79 603.62 621.69 748.00 387.64

Bihar 226.65 441.98 484.22 468.22 492.55 538.81 354.36 reproduced from the NSS report in

Chhattisgarh 190.39 447.60 552.44 567.05 554.73 616.30 322.41 Annexure I. Per capita daily calorie

Goa 492.23 1,155.74 1516.36 ** ** ** 362.25 intake (PCCI) has been calculated by

Gujarat 233.82 643.15 825.16 915.39 886.82 956.66 353.93

Haryana 232.99 675.25 895.11 682.66 728.20 788.56 414.76

Himachal Pradesh 276.57 622.23 722.51 645.95 674.24 741.89 394.28

Jammu and Kashmir 292.47 647.34 810.17 648.21 701.19 781.02 391.26

Jharkhand 236.98 457.48 512.37 530.34 562.99 626.81 366.56

Karnataka 160.33 605.92 862.02 945.58 766.85 1,018.73 324.17

Kerala 426.56 956.16 1,139.31 1,259.72 1,202.42 1,427.51 430.12

Madhya Pradesh 213.84 494.34 575.85 622.67 611.08 729.51 327.78

dividing the household calorie intake by the factor “30 multiplied by household size”. Needless to mention, our results and the conclusion in this paper are subject to the limitations, if any, of this household level data set on MPCE and PCCI.

Maharashtra 224.16 650.92 807.83 1,059.18 828.35 1,194.66 362.25

Manipur 300.99 590.67 647.63 ** ** ** –

Meghalaya 431.95 751.47 852.50 1,074.82 1,216.32 1,432.51 –

Mizoram 374.91 820.04 982.70 ** ** 728.57 –

Nagaland 406.95 959.95 1,166.00 1,512.15 1,775.10 2,136.65 –

Orissa 188.32 449.01 526.52 487.96 470.22 539.70 325.79

3 The Methodology of Analysis

For a given population of households, the HCR measures the proportion of persons having MPCE below a given poverty line, where the poverty line is

Punjab 297.17 768.78 924.98 767.54 801.27 890.20 410.38 specified as the estimated average

Rajasthan 285.81 561.50 623.08 558.79 602.33 657.98 374.57 MPCE corresponding to some given per

Sikkim 288.76 740.42 945.26 1232.97 1,348.76 1,603.48 – capita calorie norm. For convenience

Tamil Nadu 267.76 685.47 843.75 1,553.41 956.58 1,290.31 351.86 of discussion, let us treat PCCI and

Tripura 304.09 576.69 682.56 1,019.98 1,029.08 1,229.30 – MPCE as a pair of jointly distributed Uttarakhand 225.02 480.88 568.59 568.19 608.34 725.35 478.02

random variables and denote them

Uttar Pradesh 170.58 433.61 512.81 468.76 522.46 569.24 365.84

West Bengal 238.99 589.21 686.82 617.06 624.48 711.61 382.82

Andaman and Nicobar islands 249.02 852.37 1,264.99 1,366.48 1,426.89 1,728.94 –

Chandigarh 461.07 919.07 1122.35 ** ** ** –

Dadra and Nagar Haveli 358.95 937.55 1317.32 ** ** ** 362.25

Daman and Diu 704.68 1687.12 2355.35 ** ** ** –

Delhi 424.86 1080.81 1505.18 ** ** ** 410.38

as, X and Y respectively. Formally, in the Indian official procedure the poverty line is defined to be a value z of Y such that E(X/Y = z) = x*, where x* is the given per capita calorie norm. Operationally, based on a bivariate data set of observed MPCE class-wise

Lakshadweep 339.22 762.88 1011.90 ** ** ** – means of PCCI and MPCE, the X on Y

Pondicherry 339.70 801.32 1,060.72 1,170.37 1,051.08 1,146.17 – regression line would be estimated and

# SPCCR: State-specific per capita daily calorie requirements as in Manna (2007) and presented in Table 1R . using that z corresponding to x* would

* Source: Press Note dated March 2007 by the Planning Commission. ** Not presented due to doubtful fluctuations in average PCCI over MPCE classes as per Report Number 513, NSS 61st round. be estimated by inverse interpolation. In

august 8, 2009 vol xliv no 32

actual practice, official poverty lines for India for the base year 1973-74 were first calibrated following this procedure using NSS 28th round consumption expenditure data. Poverty lines at current prices for subsequent years are being derived by deflating the base year poverty lines by suitable consumer price index numbers.

Note that, given x*, the conditional distribution of Y has a prediction interval (y*1-α/2, y* α/2) of value of Y such that probability (y* ≤ Y ≤ y* /X = x*) = 1 – α. In other words, a household

1–α/2 α/2

with observed PCCI = x* will have MPCE in the prediction interval with probability (1 – α). Empirically, given a small PCCI interval (x* ± δ), δ > 0 and small, suppose (xi, yi;i = 1,2,...,n) is the set of sample observations. Suppose the conditional distribution of ln(Y) (= Y0) given ln(X) (= X0) in the

less than perfect correlation between PCCI and MPCE, one has the following inequality:

1 (x*0 – x0 )2[^y*0 –tα/2, n – 2 σ^√(1 + n+∑ (x0i – x0 )2)]≤ [^y*0 – tα/2, n – 2 σ^√(n +∑ (x0i – x0 )2)] ≤ z0 1 (x*0 – x0 )2

or equivalently, the lower prediction limit of household MPCE (i e, Y) is less than or equal to the lower confidence limit of the conditional mean E(Y/X = x*), which in turn is less than or equal to the defined poverty line z. Note that if the variance of the conditional distribution of Y0 given X0 = x*0 (^σ2 is a measure of which) is not large, the lower prediction limit of Y, the lower confidence limit of E(Y/X = x*) and z will be close. A more important issue is the following: Suppose, for a later year ^ z is the poverty line at current prices arrived at by deflating the base

corresponding interval is normal with Table 2U: Threshold MPCE Limits and Alternative Poverty Lines for the Urban Areas (2004-05)

E(Y0 /X0 = x0 ) = α + βx0 (= ^μ, say) and the estimated linear regression equa State/UT Threshold MPCE Limits (Rs) Associated with SPCCR# as Per LSM Alternative Poverty Line (Rs) Based on Current (NSS: 2004-05) Data and Different Methods and Calorie Norms Official Poverty Line* (Corresponding to Fixed
tion is ^y0 = a + bx0, a and b being the ordinary least squares estimates of α and β. The (1 – α)% interval estimate of Lower Prediction Limit of Household MPCE (1-α - 0.99) Lower Confidence Limit of Average MPCE (1-α=0.99) As Per LSM and Associated with SPCCR (PL1) As Per IIM and Associated with SPCCR (PL2) As Per IIM and Suggested Fixed All-India Calorie Norm of 2,250 Cal As Per IIMand Official Fixed All-India Calorie Norm of 2,100 Cal All-India Calorie Norm of 2,100 Cal)
E(Y0 /X0 = x*0 ), where x*0 = ln(x*), is [^y * 0 ± tα/2,n–2 ^σ 1 (x* 0 – x0)2 + n ∑ (x0i – x0 )2√( )], where ^y0 = a + bx* 0, ^σ is the estimated equa (1) Andhra Pradesh Arunachal Pradesh (2) 273.21 306.92 (3) 822.99 736.71 (4) 1,058.04 897.35 (5) 1,057.78 ** (PL3)(6) 1,042.35 ** (PL4)(7) 856.44 ** (8) 542.89 –
tion standard error and is estimated as Assam 223.47 726.69 974.41 1,056.14 1,104.02 922.32 378.84
^σ = (∑ y0i 2 – a ∑y0i – b ∑x0i y0i )/(n – 2), x0 = 1 n ∑x0i is the sample arithmetic BiharChhattisgarh 182.99 211.66 520.21 664.60 649.55 895.64 ** ** ** ** ** ** 435.00 560.00
mean of x0i ’s, i e, of ln(xi)’s; i = 1,2,...,n; Goa 420.48 1,294.43 1,784.21 ** ** ** 665.90
and tα/2,n – 2 is the α / 2 % critical value of the t-distribution with (n – 2) degrees of freedom. The corresponding (1 –α)% prediction interval of Y0 is given by [^y * 0 ± tα/2,n–2 ^σ 1 (x* 0 – x0 )2(1 + + n ∑ (x0i – x0 )2√ )] (Johnston and Dinardo 1997, Chapter 1, GujaratHaryanaHimachal Pradesh Jammu and Kashmir JharkhandKarnatakaKerala 327.99 331.90 339.38 313.95 193.56 382.86 321.32 1,060.21 872.14 877.14 757.82 492.46 1,045.69 1,021.35 1,428.02 1,155.11 1,181.83 945.47 668.95 1,341.97 1,456.57 1,566.06 1,342.38 932.06 861.88 725.51 1,473.56 1,670.11 1,554.31 1,377.06 987.58 965.89 853.26 1,452.04 1,610.68 1,260.62 1,101.15 ** 722.32 602.19 1,179.53 1,095.12 541.16 504.49 504.49 553.77 451.24 599.66 559.39
Section 1.5.3 for these results). Once Madhya Pradesh 202.97 672.82 920.09 ** ** 869.18 570.15
the lower prediction limit of Y0 /X0 = x * 0 Maharashtra 312.81 1,178.88 1,556.12 2,698.48 2,520.74 1,746.46 665.90
and the lower confidence limit of Manipur 303.40 611.44 701.36 ** ** 636.37
E(Y0 /X0 = x * 0) are obtained, the corre- Meghalaya 442.11 981.68 1,258.38 1,901.24 2,181.82 1,675.17
sponding threshold MPCE limits (i e, Mizoram 400.60 1,031.17 1,350.00 ** ** **
lower prediction limit of household Nagaland 716.60 1,169.25 1,340.54 ** ** **
MPCE (Y) associated with the statespecific calorie norms x * and the lower confidence limit of average MPCE, i e, E(Y/ X = x* ), can be easily derived by OrissaPunjabRajasthanSikkimTamil Nadu 143.50 374.52 243.50 326.08 261.57 513.60 968.35 672.52 912.41 1,023.35 747.26 1,217.92 854.06 1,216.34 1,430.28 810.57 1,247.95 851.42 1,654.84 1,723.18 805.13 1,220.73 1,016.19 1647.81 1587.32 566.23 944.52 761.36 1,334.74 1,178.66 528.49 466.16 559.63 – 547.42
taking antilog. Estimated average MPCE Tripura 261.15 802.63 1,135.60 1,009.67 1,056.45 949.21
associated with the state-specific Uttarakhandcalorie norm x * is obtained as expUttar Pradesh 207.53 195.40 714.59 629.55 1,112.35 808.52 869.72 719.95 942.14 777.11 724.16 ** 637.67 483.26
^ + ^σ2 2[ μ ] . West Bengal 294.31 911.80 1,182.78 1,541.83 1,479.05 1,160.95 449.32
It should be straightforward to Andaman and
Nicobar Islands 562.70 1,559.52 2,021.18 2,249.57 2,117.09 1,271.14

see that if PCCI and MPCE are perfectly correlated, both the intervals defined above collapse to the point

^

y* = a + bx*0 = z0, where z0 = ln(z). In other words, in such a case, the rela-

Chandigarh

Dadra and Nagar Haveli

Daman & Diu

Delhi

Lakshadweep

311.60 1,025.65 1,454.85 1,361.52 1,397.64 1,206.19
534.36 1,355.60 1,827.61 3,400.92 2,638.24 1,398.01 665.90
409.38 1,349.92 2,014.29 2,569.21 2,480.64 1,429.88
373.85 1,117.94 1,472.32 1,406.31 1,445.95 1,159.94 612.91
282.83 734.27 1,010.87 749.83 769.18 710.56

tionship between the calorie norm Pondicherry 253.53 709.18 978.13 1,472.81 1,363.23 1,065.38 –

and the corresponding poverty line # SPCCR: State-specific per capita daily calorie requirements as in Manna (2007) and presented in Table 1U.

* Source: Press Note dated March 2007 by the Planning Commission. is exact. In a more realistic case of ** Not presented due to doubtful fluctuations in average PCCI over MPCE classes as per Report Number 513, NSS 61st round.

Economic & Political Weekly

year poverty line by a consumer price index number. How would

^

z compare with the corresponding z, the lower prediction limit of Y and the lower confidence limit of E(Y /X = x*) for that later year? If the consumer price index number used for deflating the base year poverty line is appropriate, it is easy to see that for later years

^

z would be close to the other three quantities if the variance of the conditional distribution of Y given X = x* does not change much over time. In other words, any observed large discrepancy of these four quantities may be due to use of an inappropriate consumer price index number for deflation or temporally rising variance of the conditional distribution of Y given X = x* or both.

4 Test of Log Normality of the MPCE Distribution

The derivation of the lower prediction limit of household MPCE

(Y) and the lower confidence limit of average MPCE for a given calorie norm or PCCI (X), as discussed in the preceding section, involves the assumption that the conditional distribution of Y/X = x* follows lognormal distribution. In order to test the empirical validity of this assumption, for each state/UT x sector (i e, rural/urban), we consider the household level truncated data set of (X, Y) with X falling within the class interval (x* ± δ), δ > 0 and small, where x* is the state- and sector-specific average daily per capita calorie requirements as suggested in Manna (2007). Application of the Pearsonian Chi-square test of normality to

Table 3R: Proportion of Persons Below the Threshold MPCE Limits and Alternative Poverty Lines for the Rural Areas (2004-05)

State/UT Proportion of Persons (%) Below Head Count Ratios (HCR)/Proportion of Persons (%) Below the Threshold MPCE Limits Associated Alternative Poverty Line

with SPCCR as per LSM Based on Current (NSS: 2004-05) Based on Official Lower Prediction Lower Confidence Data and Alternative Poverty Lines Poverty Line* Limit of Household Limit of Average PL1 PL2 PL3 PL4 (Corresponding to Fixed) MPCE (1 -α = 0.99) MPCE (1 -α = 0.99) All-India Calorie Norm of 2,400 Cal)

(1) (2) (3) (4) (5) (6) (7) (8)

Andhra Pradesh 6.3 73.2 81.8 90.8 82.5 90.0 11.2

Arunachal Pradesh 1.5 42.6 59.6 50.1 57.1 61.9 22.3

the truncated data set of ln (Y)6 reveals that ln (Y) follows normal distribution for most of the states/UTs x sector justifying the assumption of the log normality of the MPCE distribution7 for given value of X = x* . It may be seen from Tables 1R and 1U that in most cases the observed values of χ2 are found to be smaller than the critical values for the chosen level of signifi-

Assam 10.0 69.0 78.4 68.1 71.0 85.7 22.3 cance α = 0.01.

Bihar 6.0 66.2 74.7 71.7 76.2 83.0 42.1

Chhattisgarh 11.0 73.8 84.2 85.4 84.4 88.3 40.8 5 Threshold Limits of MPCE and

Goa 18.1 74.4 77.1 ** ** ** 5.4 Alternative Poverty Lines

Gujarat 2.2 68.7 82.8 87.6 86.7 88.8 19.1

Haryana 0.6 51.2 71.8 52.3 56.8 62.3 13.6 average MPCE, given a calorie norm, are
Himachal Pradesh 1.1 47.2 60.3 50.7 55.0 62.1 10.7 random variables by themselves having a
Jammu and Kashmir 0.9 45.8 65.0 50.0 54.8 62.3 4.6
Jharkhand 7.9 68.8 78.1 80.1 83.8 88.6 46.3 probability distribution, it would be of in-
Karnataka 1.8 80.8 93.2 94.8 90.0 95.6 20.8 terest to know the corresponding threshold
Kerala 12.9 65.9 75.4 76.2 76.2 76.3 13.2 MPCE limits (i e, lower prediction limit of
Madhya Pradesh 10.5 72.7 82.2 85.3 84.6 90.5 36.9 household MPCE and lower confidence limit
Maharashtra 4.7 75.1 84.6 92.2 85.6 94.3 29.6 of average MPCE associated with the given
Manipur 0.1 55.9 67.9 ** ** ** 22.3 per capita calorie norm). These threshold
Meghalaya 7.4 74.9 85.3 95.0 97.5 97.8 22.3 limits should indicate the minimum values
Mizoram 2.7 65.8 81.1 ** ** 52.9 22.3

Since both household MPCE and estimated

Nagaland 0 59.9 75.6 77.7 79.3 81.5 22.3 calorie norm. We have derived these thresh-
Orissa 16.6 73.3 81.9 78.1 76.1 82.9 46.8 old MPCE limits associated with the state-
Punjab 1.2 56.7 70.2 56.6 59.8 68.2 9.1 specific per capita daily calorie require-
Rajasthan 5.2 60.1 69.7 59.5 67.0 74.1 18.7 ments (SPCCR) for each state/UT x sector for
Sikkim 1.6 69.2 81.7 89.8 90.0 90.7 22.3
Tamil Nadu 5.9 78.7 86.9 90.6 94.8 94.6 22.8 the year 2004-05 as per the methodology
Tripura 14.2 78.3 87.9 96.6 96.7 98.0 22.3 discussed in Section 3. These estimates
Uttarakhand 0.4 41.4 56.5 56.1 62.2 75.3 40.8 corresponding to the confidence coefficient
Uttar Pradesh 2.9 49.0 64.9 56.5 66.2 72.4 33.4 1-α = 0.99 are presented in Tables 2R/U
West Bengal 2.7 69.5 81.2 72.8 73.7 82.7 28.6 (see columns 2 and 3) (pp 40, 41). The lower
Andaman and prediction limits of household MPCE are
Nicobar islands 0 51.4 73.4 73.7 73.9 74.9 22.9

of MPCE required for achieving the given

Chandigarh 9.1 62.7 74.7 ** ** ** 7.1

Dadra and Nagar Haveli 38.6 85.7 91.0 ** ** ** 39.8

Daman and Diu 12.9 75.1 80.1 ** ** ** 5.4

Delhi 8.0 71.3 79.7 ** ** ** 6.9

found to be smaller than the official poverty lines (PL) for most of the states/UTs for both rural and urban areas. But, disturbingly enough, state-specific lower confi-

Lakshadweep 0.1 27.8 43.5 ** ** ** 13.3 dence limits of average MPCE are found to be

Pondicherry 23.8 68.8 81.6 85.2 81.2 85.0 22.9 much higher than the official poverty lines

All-India@ 5.6 65.4 76.9 --82.7 28.3 violating the inequality mentioned in Sec

* Source: Press Note dated March 2007 by the Planning Commission.

tion 3. One reason for this could be the use

** Not presented due to doubtful fluctuations in average PCCI over MPCE classes as per Report Number 513, NSS 61st round. @ All-India estimates in columns 2-4 and 7 are derived as the weighted average of state-wise estimates for which estimates are of different state-specific calorie norms (viz, presented, with weights as the proportionate shares of the individual states/UTs in estimated total projected population for January 2005. SPCCR) in the derivation of these threshold

august 8, 2009 vol xliv no 32

MPCE limits instead of using the fixed all-India official calorie norm (2,400, per capita per day for rural and 2,100 for urban) as is the case for estimating the official PL. But this argument may hold well, at least partially, with regard to only those states/UTs for which SPCCR is higher than the official calorie norm and not certainly for the remaining states/UTs.

We have also estimated the alternative PLs at current prices using the latest (NSS 61st round: 2004-05) data that take into account the current consumption pattern of the people. We derive four alternative PLs by jointly considering the methodo logy of

estimation and calorie norms. As regards the former, we consider two methods, viz, (1) the Least Square Method (LSM) as discussed in Section 3 which has been applied to the truncated data set, and (2) the Inverse Interpolation Method (IIM) used for the estimation of official PL of the base year 1973-74 applied to the full data set. Finally, four alternative PLs are derived as follows: PL1 based on LSM and calorie norm SPCCR; PL2 based on IIM and calorie norm SPCCR; PL3 based on IIM and fixed all-India calorie norm (viz, 2,290 calorie for rural and 2,250 calorie for urban) proposed in Manna (2007); and finally, PL4 based on IIM and fixed all-India

calorie norm of 2,400 calorie for rural and

Table 3U: Proportion of Persons Below the Threshold MPCE Limits and Alternative Poverty Lines 2,100 calorie for urban underlying official for the Urban Areas (2004-05)

poverty lines.

State/UT Proportion of Persons (%) Below Head Count Ratios (HCR)/Proportion of Persons (%) Below the Threshold MPCE Limits Associated Alternative Poverty Line Note that PL1 and PL2 are important since

with SPCCR as per LSM Based on Current (NSS: 2004-05) Based on Official
Lower Prediction Lower Confidence Data and Alternative Poverty Lines Poverty Line*
Limit of Household Limit of Average MPCE (1 -α = 0.99) MPCE (1 -α = 0.99) PL1 PL2 PL3 PL4 (Corres ponding to Fixed) All-India Calorie
Norm of 2,100 Cal)
(1) (2) (3) (4) (5) (6) (7) (8)
Andhra Pradesh 2.4 56.2 71.1 71.1 70.3 58.8 28.0
Arunachal Pradesh 1.0 44.4 61.0 ** ** ** 3.3 PL4 are also useful for a meaningful com-
Assam 0.9 35.8 57.3 65.3 70.0 52.4 3.3 parison of the average MPCE levels across
Bihar 7.9 47.1 62.4 ** ** ** 34.6 states because these poverty lines corre-
Chhattisgarh 7.2 48.0 64.7 ** ** ** 41.2 spond to a common all-India basket of con-
Goa 4.4 67.3 83.1 ** ** ** 21.3 sumption for each state/UT.
Gujarat 1.1 60.6 78.1 81.7 81.4 71.3 13.0 Estimates of these alternative PLs are
Haryana 3.3 49.3 67.8 75.9 77.4 65.4 15.1 presented in columns 4 to 7 of Table 2 R/U
Himachal Pradesh 1.4 32.9 51.7 37.0 40.1 ** 3.4 while the official PL is shown in column 8 of
Jammu and Kashmir 0.2 28.7 49.9 40.5 51.7 25.0 7.9 the same table. We find that each of the four
JharkhandKarnatakaKeralaMadhya Pradesh Maharashtra 4.3 10.7 2.8 6.1 4.1 24.5 66.3 56.7 53.3 68.2 40.4 77.5 75.7 69.8 80.6 44.5 80.8 79.9 ** 94.1 53.3 80.3 78.4 ** 94.0 34.7 71.9 60.5 66.6 84.7 20.2 32.6 20.2 42.1 32.2 alternative PLs is much higher than the official PL for each state/UT x sector. It is also of interest to note that for rural areas, even PL3 that corresponds to a lower (2,290) cal-
Manipur MeghalayaMizoram 0.1 0.9 0.3 37.0 45.6 48.0 54.6 62.2 70.4 ** 90.9 ** ** 93.7 ** 42.2 81.3 ** 3.3 3.3 3.3 orie norm is much higher than the official PL based on a higher (2,400) calorie norm. As regards PL4 derived afresh based on the
Nagaland 4.7 40.1 50.8 ** ** ** 3.3 latest (2004-05) NSS data following exactly
Orissa 6.6 42.7 62.7 67.5 67.2 48.5 44.3 the official procedure, ideally, one would
Punjab 1.0 52.4 65.8 67.1 65.9 50.9 7.1 expect it to be closer to the official PL. But
Rajasthan 2.7 45.7 62.9 62.7 72.7 55.5 32.9 the findings are otherwise. In fact, for some
Sikkim 0.4 42.9 68.7 85.9 85.8 76.6 3.3 of the states/UTs, PL4 even exceeds twice the
Tamil Nadu 2.3 62.7 79.1 85.6 82.6 69.9 22.2 official estimates. As discussed earlier in
Tripura 1.6 53.3 72.4 66.6 68.9 63.6 3.3 Section 3, the reasons for the above differ

they take into account the SPCCR which were derived by considering the latest agesex-occupation distribution of the people in the respective states/UTs and their calorie requirements. On the other hand, PL3 and

Uttarakhand 1.4 44.0 73.5 55.8 61.0 44.7 36.5 ences could be many, viz, use of inappro-
Uttar Pradesh 5.3 48.8 64.4 56.9 62.0 ** 30.6 priate consumer price index numbers for
West Bengal 3.5 54.2 68.7 79.9 78.5 67.9 14.8 deflation of base year PL or change in the
Andaman and
Nicobar Islands 1.1 62.7 75.5 79.7 77.3 49.1 22.2 consumption behaviour of the people,
Chandigarh 0.9 36.0 49.9 46.8 48.0 42.0 7.1 change in the variance of the conditional
Dadra and Nagar Haveli 15.8 58.0 76.8 95.5 91.5 60.4 19.1 distribution of MPCE, given the calorie
Daman and Diu 0.1 83.1 93.1 96.8 96.4 85.4 21.2 norm or some combination of these. It is
Delhi 1.2 55.2 71.2 69.3 70.4 57.3 15.2 high time that these methodological issues
Lakshadweep 3.6 24.3 45.2 25.5 26.9 22.6 20.2 are addressed.
Pondicherry 0.4 42.5 60.9 82.0 79.5 65.4 22.2
All-India@ 4.0 55.4 70.9 - - 66.2 25.7 6 Head Count Ratios

* Source: Press Note dated March 2007 by the Planning Commission.

HCR, i e, proportion of persons below a

** Not presented due to doubtful fluctuations in average PCCI over MPCE classes as per Report Number 513, NSS 61st round. @ All-India estimates in columns 2-4 and 7 are derived as the weighted average of state-wise estimates for which estimates are

given PL, is an important indicator of inci

presented, with weights as the proportionate shares of the individual states/UTs in estimated total projected population for January 2005. dence of poverty to the government for

EPW

EPWERF

august 8, 2009 vol xliv no 32

launching welfare programmes for the poor. Needless to per capita daily calorie requirements. The paper also presents emphasise, HCR should reflect the true poverty situation pre-state-specific alternative estimates of poverty lines and correvailing at a parti cular point of time for meaningful policy sponding HCRs not only with respect to the fixed all-India formulations. In Table 3 R/U (see columns 4 to 8), we present calorie norm but also by taking into account the state-specific the HCRs corresponding to the alternative PLs as well as the of-per capita daily calorie requirements. The analysis is based ficial PL for the year 2004-05. Columns 2 and 3 of the table give on the latest available large sample household consumption the proportion of persons below the threshold MPCE levels as-expenditure data collected through the National Sample Sursociated with SPCCR. vey in 61st round (2004-05). A major finding is that the alter-

We observe that during 2004-05 nearly 6% of all-India rural native poverty lines and head count ratios derived afresh based population and 4% of all-India urban population had MPCE on the latest data are much higher than the official estimates. smaller than the lower prediction limit of household MPCE associ-The observed difference between the two sets of estimates ated with the SPCCR. As regards the HCR, the alternative HCRs in-is very large and calls for re-visiting the methodological issues cluding the proportion of persons below the lower confidence of estimation of poverty line and head count ratio. During limit of average MPCE associated with the SPCCR are found to be the early 1970s, the share of food items in total household much higher than the official HCR. For the year 2004-05, the consumption expenditure was about 73% for rural India official estimate of HCR is placed at 28.3% for rural India and at and 65% for urban India. By 2004-05 this percentage share de25.7% for urban India. As against this, the alternative estimate of clined to only about 55% for rural and 43% for urban, clearly HCR based on PL4 that adopts the same official calorie norm is indicating a drastic change in consumer preferences that has found to be as high as 82.7% for rural India and 66.2% for urban taken place over time. Given this, it is perhaps necessary to India.8 The divergences are alarming. switch over to a more recent base year and redefine and recali

brate the base line poverty lines at the earliest. Equally impor

7 Concluding Observations

tant is the issue whether the calorie norm should continue to be One important issue of discussion of this paper is the setting up taken as the sole criterion for defining and determining the of a lower prediction limit of household MPCE and lower confi-poverty line ignoring altogether other dimensions of living and dence limit of average MPCE associated with the state-specific human development.

Notes

1 The lower confidence limit of average MPCE for a given calorie norm is the lower limit of the confidence interval of average MPCE corresponding to that calorie norm within which the estimated average MPCE will lie with a given probability.

2 NSS 61st round also collected data on employment-unemployment.

3 These poverty lines have been derived using the inverse interpolation method as adopted for working out the official poverty lines for the base year 1973-74. This official procedure is also described briefly in Section 3.

4 Areas excluded in the survey were Leh (Ladakh) and Kargil districts of Jammu and Kashmir; interior villages of Nagaland situated beyond five kilometres of a bus route; and villages in Andaman and Nicobar Islands which remain inaccessible throughout the year.

5 Fieldwork for the central sample was carried out by the regular field investigators of national sample survey organisation. There was a matching state sample for which the responsibility of data collection vested with the respective state

Annexure 1: Procedure for Estimating Calorie Equivalence with Regard to Items of Consumption*

In the present exercise nutrient contents of different items are taken largely from the book entitled Nutritive Values of Indian Foods by C Gopalan, B V Ramasastry and S C Balasubramanian, subsequently revised and updated by B S Narasinga, Y G Deosthale and K C Pant in 1991.

Generally, intake of calorie from an item is derived from the quantities reported as consumed

Economic & Political Weekly

EPW
august 8, 2009

governments and union territory administrations. The analysis in this paper is based on the data of central sample.

6 See Annexure 2 for the explanatory note on computation of χ2 from the observed household level data sets.

7 Finally, the test for log normality of the MPCE distribution failed – sometimes only marginally – in case of the rural areas of Haryana, Jammu and Kashmir, Karnataka and Nagaland and the urban areas of Arunachal Pradesh and Orissa. Thus, one has to be careful while using the results pertaining to these states x sectors.

8 Obtained as the weighted average of state-wise estimates of HCR with weights as the proportionate shares of the individual states/UTs in the total estimated projected population for January 2005. Projection figures are based on the document “Population Projections for India and States, 2001-2006,” Report of the Technical Group on Population Projection constituted by Natio nal Commission on Population (May 2006) and published by the Office of the Registrar General of India. However, since the figures for January

by the sample households. The calorie contents of each item per unit of quantity are shown in the following table.

For some items having variable food content, average calorie contents per rupee are given in the table instead of per unit of quantity. For such items the unit has been shown as “Re”. These figures multiplied by the value of consumption give the total quantities of nutrients derived from that item.

The chart of calorie contents remains practically the same as that for the 55th round of NSS

vol xliv no 32

2005 are not available in the document, average of projected figures for October 2004 and March 2005 has been taken.

References

Government of India (1979): “Report of the Task Force on Projections of Minimum Needs and Effective Consumption Demand”, Perspective Planning Division, Planning Commission, January.

  • (1993): “Report of the Expert Group on Estimation of Proportion and Number of Poor”, Perspective Planning Division, Planning Commission, July.
  • (2006): “NSS Report No 508”, Level and Pattern of Consumer Expenditure, 2004-05.
  • (2007): “NSS Report No 513”, Nutritional Intake in India, 2004-05.
  • Johnston, J and J Dinardo (1997): Econometric Methods, Fourth Edition, McGraw-Hill Companies, Inc.

    Manna, G C (2007): “On Calibrating the Poverty Line for Poverty Estimation in India”, Economic & Political Weekly, Vol XLII, No 30.

    (1999-2000). The only differences are those due to change in unit of collection of data, e g, “litre” in place of “glass”. For items for which the unit (as explained above) is “Re”, the fall in value of the rupee since the 55th round (1999-2000) has been adjusted by deflating the nutrient contents by the appropriate consumer price index (viz, CPI for Agricultural Labourers for rural and CPI for Urban Non-Manual Employees for urban areas). Thus for these items, there are two sets of nutrient content coefficients – one for rural and the other for urban households.

    Code Item Unit Calories per Code Item Unit Calories per Code Item Unit Calories per
    Unit (kcal) Unit (kcal) Unit (kcal)
    (1) (2) (3) (4) (1) (2) (3) (4) (1) (2) (3) (4)
    101 Rice – PDS kg 3460 184 Pork kg 1140 247 Other fresh fruits (urban) Re 8.2
    102 Rice – other sources kg 3460 185 Chicken kg 1090 250 Coconut, copra Kg 6620
    103 Chira kg 3460 186 Others: birds, crab, oyster, 251 Groundnut Kg 5670
    104 Khoi, lawa kg 3250 tortoise, etc kg 900 252 Dates Kg 3170
    105 Muri kg 3250 190 Potato kg 970 253 Cashewnut Kg 5690
    106 Other rice products kg 3460 191 Onion kg 550 254 Walnut Kg 6870
    107 Wheat/atta – PDS kg 3410 192 Radish kg 170 255 Other nuts Kg 4100
    108 Wheat/atta – other sources kg 3410 193 Carrot kg 480 256 Raisin, kishmish,
    110 Maida kg 3480 194 Turnip kg 290 monacca, etc Kg 3080
    111 Suji, rawa kg 3480 195 Beet kg 430 257 Other dry fruits Kg 3060
    112 Sewai, noodles kg 3520 196 Sweet potato kg 1200 260 Sugar – PDS Kg 3980
    113 Bread (bakery) kg 2450 197 Arum kg 970 261 Sugar – other sources Kg 3980
    114 Other wheat products kg 3460 198 Pumpkin kg 250 262 Gur Kg 3830
    115 Jowar and products kg 3490 200 Gourd kg 120 263 Candy, misri Kg 3980
    116 Bajra and products kg 3032 201 Bitter gourd kg 400 264 Honey Kg 3190
    117 Maize and products kg 3420 202 Cucumber kg 130 280 Turmeric Gm 3.49
    118 Barley and products kg 3360 203 Parwal, patal kg 200 281 Black pepper Gm 3.04
    120 Small millets and products kg 2615 204 Jhinga, torai kg 170 282 Dry chillies Gm 2.46
    121 Ragi and products kg 3280 205 Snake gourd kg 180 283 Tamarind Gm 2.83
    122 Other cereals kg 2615 206 Papaya: green kg 270 284 Curry powder Gm 0.8
    139 Cereal substitutes 207 Cauliflower kg 300 285 Oilseeds Gm 4.5
    (tapioca, jackfruit seed, etc) kg 2090 208 Cabbage kg 270 286 Other spices Gm 3.6
    140 Arhar, tur kg 3350 210 Brinjal kg 240 290 Tea : cups no 27
    141 Gram (split) kg 3720 211 Lady’s finger kg 350 291 Tea : leaf Gm 0
    142 Gram (whole) kg 3720 212 Palak/other leafy vegetables kg 260 292 Coffee : cups no 40
    143 Moong kg 3480 213 French beans, barbati kg 244 293 Coffee: powder Gm 0
    144 Masur kg 3430 214 Tomato kg 200 295 Cold beverages:
    145 Urd kg 3470 215 Peas kg 930 bottled/canned litre 320
    146 Peas kg 3150 216 Chillis: green kg 290 296 Fruit juice and shake litre 250
    147 Soyabean kg 4320 217 Capsicum kg 240 297 Coconut: green no 60
    148 Khesari 150 Other pulses 151 Gram products kg kg kg 3450 3400 3600 218220221 Plantain: green Jackfruit: green Lemon kg kg no 640 510 10 298 Other beverages: cocoa, etc, (rural) 298 Other beverages: cocoa, etc (urban) Re Re 19 17.1
    152 Besan kg 3400 222 Garlic gm 1.45 300 Biscuits (rural) Re 53.4
    153 Other pulse products kg 3400 223 Ginger gm 0.67 300 Biscuits (urban) Re 48.1
    160 Milk: liquid litre 1000 224 Other vegetables (rural) Re 70.6 301 Salted refreshments (rural) Re 46.1
    161 Baby food kg 3570 224 Other vegetables (urban) Re 63.6 301 Salted refreshments (urban) Re 41.6
    162 Milk : condensed/powder kg 4960 230 Banana no 116 302 Prepared sweets (rural) Re 35.3
    163 Curd kg 600 231 Jackfruit kg 880 302 Prepared sweets (urban) Re 31.8
    164 Ghee kg 9000 232 Watermelon kg 160 303 Cooked meals no 1200
    165 Butter kg 7290 233 Pineapple no 460 304 Cake, pastry Kg 5000
    166 Ice-cream (rural) Re 10.9 234 Coconut no 660 305 Pickles Gm 4
    166 Ice-cream (urban) Re 9.8 235 Guava kg 510 306 Sauce Gm 0.6
    167 Other milk products (rural) Re 55.2 236 Singara kg 1150 307 Jam, jelly Gm 2.5
    167 Other milk products (urban) Re 49.7 237 Orange, mausambi no 50 308 Other processed food (rural) Re 46.1
    170 Vanaspati, margarine kg 9000 238 Papaya kg 320 308 Other processed food (urban) Re 41.6
    171 Mustard oil kg 9000 240 Mango kg 740 310 Pan : leaf no 2.2
    172 Groundnut oil kg 9000 241 Kharbooza kg 170 311 Pan : finished no 3.7
    173 Coconut oil kg 9000 242 Pears, naspati kg 520 312 Supari Gm 6.55
    174 Edible oil: others kg 9000 243 Berries kg 530 331 Toddy litre 380
    180 Eggs no 100 244 Leechi Kg 610 332 Country liquor litre 380
    181 Fish, prawn kg 1050 245 Apple Kg 590 333 Beer litre 380
    182 Goat meat/mutton kg 1180 246 Grapes Kg 710 334 Foreign liquor or
    183 Beef/buffalo meat kg 1140 247 Other fresh fruits (rural) Re 9 refined liquor litre 380
    *Source: NSS Report No 513: Nutritional Intake in India: 2004-05.
    46 august 8, 2009 vol xliv no 32 Economic & Political Weekly
    EPW

    Annexure 2: Explanatory Note on Computation χ2 of From the Observed Household Level Data Sets

    To start with, household level full data set for each state/UT x sector is truncated by considering the allowable variations (δ) in PCCI as 25 calories (per capita per day), plus or minus the state-specific daily per capita calorie requirements (x*) as in Manna (2007). Suppose the population of households in the truncated data set, with PCCI falling within the class interval (x*±δ), is grouped into k mutually exclusive classes according to their value of MPCE, the proportion of households falling in the i-th class being Pi; i = 1, 2, …., k. For the sake of simplicity and uniformity, to start with, value of k is taken as 12 for each state/UT x sector and the various class-limits of y are taken as same as the corresponding values of MPCE as per the following MPCE classes (in rupees) as used in NSS Report Number 508, NSS 61st round – viz, for rural sector the classes are 0-235, 235-270, 270-320, 320-365, 365-410, 410-455, 455-510, 510-580, 580-690, 690-890, 890-1,155, 1,155 and more and for the urban sector the classes are 0-335, 335-395, 395-485, 485-580, 580-675, 675-790, 790-930, 930-1,100, 1,100-1,380, 1,380-1,880, 1,880-2,540, 2,540 and more. Let Pi denote the estimated population proportion under the null hypothesis that ln (Y) follows normal distribution. In fact, Pi is estimated as:

    ^^

    log yiU – μ log yiL – μ

    Pi = Φ – Φ

    ( ) ()

    σ^σ^

    where Φ (z)= Probability (Z ≤ z) being the cumulative probability of the normal deviate Z, yiU = upper limit of household MPCE for i-th class,

    = lower limit of household MPCE for i-th

    yiL

    class,

    ^

    μ = sample mean of ln (MPCE)’s for the truncated data set of sample households, and

    ^

    σ = standard deviation of ln (MPCE)’s for the truncated data set of sample households. Let n be the total number of sample observations in the truncated data set for a given state/ UT x sector. Then, under the null hypothesis,

    k

    ∑ (fi -npi )2/npi ~

    i=1χ2 distribution with (k-3) degrees of freedom; where n denotes the total number of sample observations (i e, no of sample households) in the truncated data set and fi denotes the number of sample observations for the i-th class.

    It may be mentioned here that the truncation of the data set as per the method discussed above sometimes – particularly for smaller states and UTs – resulted in only a fewer number of sample households which is not found adequate to perform the chi-square test. In such cases, the allowable variation in PCCI was increased such that the required minimum number of sample observations for any state/ UT x sector are obtained. Finally, allowable calorie variations other than 25 calories used were as follows: (a) for rural areas, 10 calories for Jammu and Kashmir and Uttar Pradesh; 50 calories for Mizoram and Sikkim; 100 calories for Andaman and Nicobar Islands; 200 calories for Pondicherry; 300 calories for Chandigarh, Dadra and Nagar Haveli, Delhi and Lakshadweep; and 500 calories for Daman and Diu; and

    (b) for urban areas, 50 calories for Assam, Bihar, Chhattisgarh, Himachal Pradesh, Meghalaya, Nagaland and Tripura; 100 calories for Arunachal Pradesh, Andaman and Nicobar Islands, Chandigarh and Pondicherry; 200 calories for Goa, Sikkim and Lakshadweep; 300 calories for Dadra and Nagar Haveli; and 500 calories for Daman and Diu.

    Further, while performing the chi-square test on the truncated data file, if the expected frequency npi for any class was found small (less than 5 or so), the adjacent classes were coalesced (which resulted in the reduction of value of k) to ensure that each of the expected frequencies was greater than 5. Further, in a few cases the adoption of specified class-limits of MPCE resulted in smaller values of expected frequencies (less than 5 or so) in one or two classes with higher values in the other classes. In such situations, the specified MPCE classlimits were modified to obviate this difficulty. The states/UTs where such modifications were made included Delhi, Goa, Jammu and Kashmir, Meghalaya, Nagaland, Andaman and Nicobar Islands, Chandigarh, Dadra and Nagar Haveli, Daman and Diu and Pondicherry in case of rural areas and Meghalaya, Mizoram, Nagaland, Tripura, Dadra and Nagar Haveli, Lakshadweep and Pondicherry in case of urban areas.

    SAMEEKSHA TRUST BOOKS

    Global Economic & Financial Crisis Essays from Economic and Political Weekly

    In this volume economists and policymakers from across the world address a number of aspects of the global economic crisis. One set of articles discusses the structural causes of the financial crisis. A second focuses on banking and offers solutions for the future. A third examines the role of the US dollar in the unfolding of the crisis. A fourth area of study is the impact on global income distribution. A fifth set of essays takes a long-term view of policy choices confronting the governments of the world.

    A separate section assesses the downturn in India, the state of the domestic financial sector, the impact on the informal economy and the reforms necessary to prevent another crisis.

    This is a collection of essays on a number of aspects of the global economic and financial crisis that were first published in the Economic & Political Weekly in early 2009.

    Pp viii + 368 2009 Rs 350

    Available from

    Orient Blackswan Pvt Ltd

    Mumbai Chennai New Delhi Kolkata Bangalore Bhubaneshwar Ernakulam Guwahati Jaipur Lucknow Patna Chandigarh Hyderabad Contact: info@orientblackswan.com

    Dear Reader,

    To continue reading, become a subscriber.

    Explore our attractive subscription offers.

    Click here

    Back to Top