Precision of Estimates of Gross Value Added Per Worker

Precision of Estimates of Gross Value Added Per Worker

Gross value added per worker (GVAPW) in any sector of the economy is an important parameter used in national accounts statistics since its product with the population of the workforce is used to estimate GDP originating in the sector. The author describes the method used for estimating GVAPW in the services sector in the 57th round of the National Sample Survey, and discusses the sampling error of the estimates as well as the probable reasons for large possible error in some sub-sectors.

G C MANNA

I Introduction

he services sector is a rapidly growing sector in the Indian economy. The percentage share of this sector – comprising industrial categories other than agriculture, manufacturing, mining and quarrying, electricity, gas and water supply, and construction – in GDP at current prices was about 53 per cent during the year 2004-05.1 The sector is a major provider of employment to a large segment of the population, particularly those living in urban India. According to the 60th round (January-June 2004) of the National Sample Survey (NSS), about 14 per cent of the usually working population in rural India and 58 per cent in urban India were engaged in the services sector according to their “usual principal and subsidiary status”. The majority of them were engaged in the unorganised segment. About 66 per cent of the usual status workers in the services sector were found to be working in unincorporated, i e, proprietary and partnership, enterprises2 according to the NSS 55th round (19992000).3

Recognising the unorganised services sector’s importance, the NSS has taken it as one of its regular subjects of enquiry. The enterprises in the sector are covered through what are known as “enterprise surveys” (ESs). The last ES on the services sector, excluding trade and ﬁnance, was conducted during 2001-02 as part of the NSS 57th round. The items of information collected at the enterprise level included number of workers, compensation to workers, operating expenses, receipts, gross value added (GVA), ﬁxed assets, and loans outstanding.

GVA per worker (GVAPW) based on the enterprise surveys is one of the important parameters in use for the estimation of GDP. In fact, the estimates of GVAPW are generally multiplied by number of workers, estimated otherwise, to compute GVA for different sub-sectors in the unorganised segment. Thus, we consider GVAPW based on the NSS 57th round as the variable for analysis. The primary aim of the article is to make an objective assessment of the precision of estimates of GVAPW for different industrial categories covered in the survey and suggest appropriate reﬁnements in the survey methodology for improving the estimates through future surveys.

II About NSS 57th Round Survey

The NSS 57th round (July 2001-June 2002) was devoted to the survey on the unorganised services sector excluding trade and ﬁnance. The term “unorganised sector” was used to represent all enterprises except the following:

(1) units registered under Sections 2m(i) and 2m(ii) of the Factories Act, 1948, and Bidi and Cigar Workers (Condition of Employment) Act, 1966; and

(2) units run by government/public sector enterprises.

In other words, all unincorporated, ie, proprietary and partnership, enterprises, outside the public sector comprised the unorganised sector. For more details on the coverage, sample design and sample size, see Appendix I.

III Some Key Results

Before we turn to the main theme of this article, it may be useful to highlight certain key results based on the NSS 57th round. Considering all types of enterprises and activities under the coverage, during the year 2001-02, an estimated 14.5 million enterprises were found engaged in the unorganised services sector (from now onwards, the term services sector will be used to represent the activities covered in the survey, as described in Section II). About 59 per cent of the enterprises were located in the rural areas.

As many as 84 per cent of the enterprises were own account enterprises (OAEs), that is, they were operating without employing any hired worker on a “fairly regular basis”. Three activity groups, namely, restaurants; transport and related activities; and other community, social and personal services, dominated the total number of enterprises in the country – their shares being 15 per cent, 33 per cent and 21 per cent respectively.

The unorganised services sector employed nearly 26.6 million people. The rural sector had a share of about 53 per cent in the total employment in the unorganised services sector.

Average value of ﬁxed assets per enterprise was estimated at about Rs 1,53,000-Rs 52,000 for OAEs and Rs 6,67,000 for “establishments”, that is, units run by employing at least one hired worker on a fairly regular basis. Average value of ﬁxed

Economic and Political Weekly September 15, 2007

assets was much higher (Rs 2,66,000) for urban enterprises as compared to their rural counterparts (Rs 76,000).

Average annual GVAPW for all types of enterprises taken together was estimated to be Rs 28,160. It was Rs 19,412 for OAEs and Rs 41,086 for establishments. The units/enterprises located in the urban areas had higher GVA per worker (about

1.87 times) as compared to those located in the rural areas.

IV Estimated GVA per Worker

Annual GVAPW by tabulation category/activity, separately for OAEs and establishments and also for unincorporated enterprises vis-à-vis all enterprises, based on the NSS 57th round (2001-02), is presented in Table 1. The following are the main ﬁndings:

(1) Among the unincorporated, i e, proprietary and partnership (PP) enterprises, GVAPW is maximum (Rs 51,917) for hotels and minimum (Rs 13,916) for other services not separately listed.

(2) Considering enterprises of all ownership types (‘all’) including the unincorporated ones, GVAPW is maximum (Rs 86,074) for storage and warehousing and minimum (Rs 16,698) for other services. Hotels comes second in position in descending order of GVAPW, the value being Rs 76,599.

(3) GVAPW for establishments is higher than that for OAEs. This is true for each activity as well as for PP and all enterprises. The ratio between the two values varies between 1.18 (mechanised road transport) and 2.70 (storage and warehousing) in the case of PP enterprises and between 1.26 (mechanised road transport) and 7.52 (storage and warehousing) in case of all enterprises.

V Precision of Estimates of GVA per Worker

As mentioned in Section I, in the national accounting, the estimates of GVAPW are multiplied by the estimated workforce to arrive at GDP at factor cost by industry/activity. Thus, it would be of interest to analyse the precision of the estimated GVAPW presented in Table 1.

For the NSS 57th round (2001-02), like for previous surveys, the government/public sector was outside its coverage. The residual category (termed ‘all’ in Section IV) comprises two broad groups, namely, private organised (PO) and private unorganised. While the former group mainly consists of limited companies and cooperative societies, the latter comprises unincorporated (PP) enterprises. Keeping the use of data in view, it would be desirable to assess the estimates of GVAPW separately for PO and PP enterprises. But since separate estimates for PO have not been published, we will focus on the estimates pertaining to PP and ‘all’ (i e, PO + PP).

Normally, the precision of any estimate can be judged by computing the relative standard error (RSE) of the estimator. The NSS sampling designs normally have inbuilt provisions to compute RSEs of survey results based on two sub-sample estimates. Generally, the primary stage units (sample villages/ urban blocks) in the NSS are selected in the form of two independent sub-samples. This was not the case in the NSS 57th round where two sub-samples were dependent. In such a situation, it is not straightforward to compute RSE based on sub-sample wise estimates. But the value of the indicator D, as deﬁned below, is likely to give some rough idea about the precision (RSE) of the estimated GVAPW based on the whole sample (ie, two sub-samples combined) if the pooled estimate based on the whole sample is taken as the simple average of the two sub-sample estimates of GVAPW:

|y1 – y2| D = ———– × 100 (y1 + y2)

where y1 and y2 denote estimates of GVAPW based on subsamples 1 and 2 respectively.

Table 1: GVAPW by Enterprises in the Unorganised Services Sector, 2001-02

Rural and Urban Combined

Activity Estimated Annual GVAPW (Rs) (Category of NIC 98) Proprietary and All Enterprises Partnership Ents incl Unincorp OAEs Estts Com-OAEs Estts Combined bined

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

Hotels (H1) 25945 54556 51917 24900 80224 76599 Restaurants (H2) 16961 30493 22434 16961 31274 22809 Hotels and Restts

combined 17023 32702 23704 17019 37697 26140

Storage and

warehousing (I1) 19879 53591 41692 16688 125523 86074 Mech road transp(I2) 37240 43957 40218 37245 46991 41615 Other transp and

related (I3) 18534 39831 20224 18545 58222 22132 Communication (I4) 17913 22625 19704 17889 28732 22150 Transp, storage and

commn (I) 24668 39293 28784 24663 45418 30686 Real estt, rent and

business (K) 27288 53765 39441 27231 61187 43577 Education (M) 14671 20387 18515 14616 27728 24744 Health and social work (N) 26996 49011 35513 26826 60617 41959 Other services (O) 11681 24672 13916 11529 40294 16698

All combined 19501 34613 24922 19412 41086 28160 Estimated no of workers

(million) 15.745 8.809 24.554 15.837 10.719 26.556

Source: NSS (2002a).

Table 2: Precision of Estimated Annual GVAPW, 2001-02

Rural and Urban Combined

Activity Precision (D) of Estimated Annual GVAPW (Per Cent) (Category of NIC 98) Proprietary and All Enterprises Partnership Ents incl Unincorp OAEs Estts Com-OAEs Estts Combined bined

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

Hotels (H1) 17.7 6.5 10.5 14.9 13.4 9.8 Restaurants (H2) 0.7 5.7 4.3 0.7 4.7 3.7 Hotels and Restts

combined 0.6 4.8 3.8 0.6 3.4 1.4

Storage and

warehousing (I1) 34.5 76.2 44.1 40.8 74.7 94.9 Mech Road Transp (I2) 0.9 2.1 1.9 0.8 13.6 8.0 Other transp and

related (I3) 0.1 7.7 0.2 0 28.5 5.0 Communication (I4) 16.4 8.3 5.2 16.7 9.6 13.1 Transp, storage and

commn (I) 3.4 1.9 1.0 3.5 2.5 0.5

Real estt, rent and

business (K) 3.4 46.6 23.5 3.3 52.6 30.6 Education (M) 3.6 2.9 0.6 3.3 10.4 12.5 Health and social work (N) 16.5 6.4 15.3 16.0 35.3 35.4 Other services (O) 4.3 7.0 5.4 5.2 63.7 24.4

All combined 4.8 6.5 0.0 5.1 10.3 1.4

Economic and Political Weekly September 15, 2007

Table 2 presents the values of D for the estimates presented in Table 1. If values of D (which is a percentage) are less than 10 to 15, the estimated GVAPW is likely to be fairly reliable. We ﬁnd that estimates for OAEs and establishments combined are fairly reliable for all activities except storage and warehousing; real estate, rent and business activities; health and social work; and other services. Two other important facts emerge. One, estimates for establishments are relatively inferior to those for OAEs in most cases. Two, estimates for all enterprises are generally inferior to those for PP enterprises.

Among the activities for which the value of D is relatively high, sample size, that is, number of enterprises surveyed, is small (Table 3) in the case of storage and warehousing (sample size: 729) when compared with the overall sample size of 2,44,376 enterprises surveyed during the round. But in the case of other activities, the sample size is fairly large, particularly for OAEs. For establishments for which values of D are larger, sample size varies between 7,411 (other services) and 10,297 (real estate, rent and business services), which is not that small. We perhaps need to investigate whether it is the sampling error or non-sampling error that has really contributed to the higher value of D in these cases.

VI Estimated Aggregate GVA, Enterprise Survey vis-à-vis NAS

It may be interesting to see to what extent the alternative estimates of GVA based on ESs and national accounts statistics (NAS) differ at the activity level. Although the methods for compilation of such GVA estimates are different according to the two methods, it is desirable that the alternative estimates do not differ by more than a certain percentage. Due to the limitation of data availability, we could consider one activity, hotels and restaurants, for this analysis. We note from Table 2 that the estimated GVAPW for this activity according to the NSS 57th round is quite reliable.

We see that as compared to the estimated aggregate GVA based on the ES, the NAS estimate for the activity of hotels and restaurants during 2001-02 is 65 per cent higher for private unorganised enterprises and about 110 per cent higher for all private enterprises. The difference is alarmingly large. It would be useful to identify the reasons for such variation with a view to taking remedial measures. Further, an extension of this study by covering other activities may also be useful.

VII Concluding Observations

We have seen that the precision of estimated GVAPW for the activities of storage and warehousing; real estate, rent and business activities; health and social work; and other services (urban only) is relatively low. This is true even for PP, i e, private unincorporated enterprises, if we leave aside “other services”. Thus, the precision of the estimated GVAPW at the corresponding sub-activity/compilation category level used in national accounting is likely to be affected further, at least for some compilation categories. We need to reorient our survey designs suitably to improve this situation in future.

It has been observed that the estimated GVAPW for all enterprises is generally inferior to that for proprietary and partnership enterprises. This is rather expected since the ‘all’ category includes private organised enterprises, consisting of limited companies and cooperative societies, which are relatively few in number but are likely to be quite heterogeneous in terms of GVAPW. The use of standard NSS sampling designs based on an area frame approach may not be appropriate to capture them in adequate number. In our opinion, it would be desirable to delink the surveys of such private organised enterprises based on a single stage design by considering an appropriate list frame on the lines of the Annual Survey of Industries (ASI). The National Statistical Commission has recommended the launching of a separate survey of such big units. Instead, it may be cost effective to continue to work with the ASI by enlarging its coverage to include private organised enterprises and other big units in the services sector.4

We have also found wide divergence between the ES based estimate of aggregate GVA and that (GDP at factor cost at current price) based on the NAS for the activity of hotels and restaurants considered for the analysis. It seems useful to extend this study to other activities and ﬁnd out the reasons for such divergence so as to incorporate the necessary corrective measures.

Finally, after having analysed the precision of estimates of GVAPW for different industrial categories based on the NSS 57th round, we strongly feel that integration of so many service sector activities in one round of the NSS needs a relook. The case becomes stronger if we have to think of generating reliable estimates of GVAPW by activity at the state level for the preparation of estimates of state domestic product by state governments.

Table 3: Values of Precision (D) of Estimated GVAPW vs Sample Size (n)

Rural and Urban Combined

Activity	Values of	All Enterprises Incl
		Unincorporated
		OAEs	Estts	Combined
(1)	(2)	(3)	(4)	(5)
Storage and warehousing	D	40.8	74.7	94.9
	n	417	312	729
Real estate, rent and	D	3.3	52.6	30.6
business service	n	20625	10297	30922
Health and social work	D	16.0	35.3	35.4
	n	15020	7662	22682
Other services	D	5.2	63.7	24.4
	n	32891	7411	40302

(n = number of enterprises surveyed)

Table 4: Aggregate GVA according to Alternative Sources, 2001-02

Activity: Hotels and Restaurants

Category of Enterprise (1)	Aggregate GVA (Rs Crore) ES NAS* (2) (3)	Per Cent Difference** (4)
Prop and part enterprises All (other than govt/public sector)	11,745 13,303	19,395 27,934	65 110

Notes: * Courtesy: NAD, CSO; value ﬁgures under NAS include ﬁnancial intermediary services indirectly measured (FISIM) amounting to Rs 913 crore, which ideally should be deducted. However, adjustment has not been done since its apportionment by category of enterprises is not feasible. ** Col 4 = 100 x (Col 3-Col 2)/Col 2; actual difference would be a little less (see the previous footnote).

Economic and Political Weekly September 15, 2007

Appendix:Brief Background of NSS 57th Round Survey Notes

The ﬁeld work of the NSS 57th round was carried out during July 2001-June 2002. The survey covered enterprises other than those in the government/public sector and aimed at providing estimates of the total number of enterprises, employment, ﬁxed assets, operating expenses, receipts, value added, loans, GVAPW, etc. The 57th round covered broadly all the activities of hotels and restaurants (tabulation category H of NIC 98); transport, storage and communication (I); real estate, renting and business activities (K); education (M); health and social work (N); and other community, social and personal service activities (O). The survey did not cover the service sector enterprises pursuing the activities of wholesale and retail trade, repair of motor vehicles, motorcycles and personal and household goods (G); ﬁnancial intermediation (J); public administration and defence (L); private households with employed persons (P), and extra-territorial organisations and bodies (Q). A stratiﬁed multi-stage sampling design was adopted for selection of sampling units.5 The ﬁrst stage units were villages (panchayat wards in Kerala) in rural areas and urban frame survey (UFS) blocks in urban areas. The ultimate stage units were enterprises. Within each selected ﬁrst stage unit, all big non-agricultural enterprises having 200 or more workers were separately listed and the schedule was canvassed in all such enterprises that were eligible for survey. Relatively larger ﬁrst stage units were subdivided into smaller segments, a minimum of four in number. Three segments, one with the maximum number of enterprises, and two more at random, were selected for the survey. The ultimate stage units were selected independently from each of these segments after stratifying the listed enterprises by broad activity group6 x enterprise type.7 In all, a maximum of 36 enterprises were selected from each ﬁrst stage unit. At all-India level, 6,513 villages and 9,356 UFS blocks were surveyed. A sample of 92,528 enterprises from rural India and 1,51,828 enterprises from urban India was netted for collecting detailed information.

Email: mannagcm@yahoo.com

[This article expresses the author’s personal views and does not represent the opinions of the organisation with which he is associated.]

1 See press release dated January 31, 2006 by the Central Statistical Organisation.

2 In the NSS, traditionally, an enterprise is deﬁned as an undertaking engaged in production and/or distribution of goods or services mainly for the purpose of sale, whether fully or partly. Two units/branches at two different sites of the same concern are considered as two separate “enterprises”.

3 See Table 9.1, NSS Report No 460

4 See Manna and Bhattacharjee (2004).

5 See sample design in NSS 2002a, 2002b for details.

6 Nine broad activity groups (BAGs) were formed: storage and warehousing (BAG 1); hotels (BAG 2); post and telecommunications (BAG 3); mechanised road transport (BAG 4); real estate, renting and business activities (BAG 5); health and social work (BAG 6); restaurants (BAG 7); education (BAG 8); and other community, social and personal services (BAG 9).

7 Enterprise type: own account enterprises (i e, units without any hired worker on a fairly regular basis) – 1; establishments (i e, units/enterprises with at least one hired worker) – 2.

References

NSS (2002a): NSS Report No 482: Unorganised Service Sector in India, Salient Features, National Sample Survey Organisation.

– (2002b): NSS Report No 483: Unorganised Service Sector in India, Characteristics of Enterprises, National Sample Survey Organisation.

Manna, G C and J P Bhattacharjee: ‘Enlarging the Scope of Annual Survey of Industries’, Economic and Political Weekly, Vol XXXIX; No 48.

SAMEEKSHA TRUST BOOKS

Inclusive Growth K N Raj on Economic Development

Essays from The Economic Weekly and Economic and Political Weekly

Edited by ASHOKA MODY

The essays in the book reﬂect Professor K N Raj’s abiding interest in economic growth as a fundamental mechanism for lifting the poor and disadvantaged out of poverty. He has also been concerned that the political bargaining process may end up undermining growth and not provide support to those who were excluded from access to economic opportunities. These essays, many of them classics and all published in Economic Weekly and Economic and Political Weekly, are drawn together in this volume both for their commentary on the last half century of economic development and for their contemporary relevance for understanding the political economy of development in India and elsewhere.

Pp viii + 338 ISBN 81-250-3045-X 2006 Rs 350

Available from

Orient Longman Ltd

Mumbai Chennai New Delhi Kolkata Bangalore Bhubaneshwar Ernakulam Guwahati Jaipur Lucknow Patna Chandigarh Hyderabad

Contact: info@orientlongman.com

Economic and Political Weekly September 15, 2007