Business Today - - CONTENTS - By Pronab Sen

Cur­rent sur­veys pro­vide a broad em­ploy­ment sce­nario but miss out on in­sights re­quired for in­clu­sive pol­i­cy­mak­ing.

For the long­est time, we sim­ply have not had a method for fre­quent col­lec­tion of jobs data. The main source of data is the National Sam­ple Sur­vey (NSS), which is col­lected ev­ery five years. The NSS used to have an an­nual round, which rep­re­sented the coun­try as a whole but could not break the data down to states as the sam­ple size was small (40,000). Re­gard­ing jobs, there is more in­ter­est in state-wise data than the national statis­tics. That is why the data was not used very much, and my mid2000 an­nual sur­vey was stopped. Since then, we only have five-year jobs data.

Along with that, the Labour Bureau used to col­lect data essen­tially on wages. But after the global fi­nan­cial cri­sis in 2009, the bureau was told to start a sur­vey on job losses. So, it se­lected nine in­dus­tries (across labour­in­ten­sive sec­tors) to mea­sure job losses. As it started col­lect­ing jobs data, the Labour Bureau stopped col­lat­ing wage data in 2009.

Then the bureau started an an­nual sur­vey like the NSS sur­vey some­times in 2011 to mea­sure job cre­ation. The first cou­ple of an­nual sur­veys were not good but then it was re­vised, and now the sur­vey is good enough.

So, once the pe­ri­odic labour force sur­vey by the National Sam­ple Sur­vey Of­fice (NSSO), which gives an­nual statis­tics of em­ploy­ment state-wise (ur­ban and ru­ral) and quar­terly es­ti­mates for ur­ban ar­eas, sta­bilises, the Labour Bureau would stop its an­nual labour sur­vey and go back to wage sur­vey. Wage data is im­por­tant be­cause it re­flects the bar­gain­ing power of the labour force, and in a way, in­di­cates the em­ploy­ment sit­u­a­tion.

Be­sides gov­ern­ment data, the Cen­tre for Mon­i­tor­ing In­dian Econ­omy (CMIE) is the only pri­vate agency that comes out with labour data. The CMIE sur­vey is not de­signed for mea­sur­ing em­ploy­ment but it is a house­hold sur­vey, and its main fo­cus is on con­sump­tion. Be­cause it is do­ing a house­hold sur­vey, it is get­ting em­ploy­ment data as well.

The prob­lem with the CMIE data set is that it is sur­vey­ing the same house­holds and that kind of sam­ple be­comes non-rep­re­sen­ta­tive after a cou­ple of years. CMIE has un­der­taken the sur­vey for four to five years. So, by now, the sam­ple has be­come non-rep­re­sen­ta­tive.

Vari­a­tions and Short­com­ings There would be vari­a­tions in the re­sults of all the sur­veys men­tioned ear­lier. There would be vari­a­tions re­gard­ing sam­ples, but most im­por­tantly, there would be vari­a­tions re­gard­ing con­cepts.

The Labour Bureau has now started fol­low­ing the same con­cept as the NSS so that the re­sults are com­pa­ra­ble. CMIE, how­ever, is some­what dif­fer­ent.

You can mea­sure em­ploy­ment in dif­fer­ent ways. The sim­plest way is to ask if one has been em­ployed most of the time dur­ing the year. That gives you an idea of the larger em­ploy­ment pic­ture, but it does not tell you if peo­ple are work­ing full time.

There are more sen­si­tive mea­sures, which are ac­tu­ally bet­ter. One mea­sure is that you use weekly data. You take pre­vi­ous week’s data and break it into 14 half­days and ask the per­son if he/she was em­ployed dur­ing those pe­ri­ods.

There are many short­com­ings of a sur­vey data. The first, of course, is that the sam­ple has to be rep­re­sen­ta­tive. For a sam­ple to be rep­re­sen­ta­tive, you must have some way of know­ing the full pop­u­la­tion. The only place where we get that kind of data is the cen­sus. In fact, al­most all th­ese sur­veys are based on the last cen­sus data. Now, as we start mov­ing away from the cen­sus, the qual­ity of sam­pling drops. But there is noth­ing you can do about it; you can­not have a cen­sus ev­ery year.

Th­ese sur­veys are based on re­sponses from in­di­vid­u­als, and when you ask them ques­tions, there could be what you call re­call er­ror. When you ask any­one if he/she was em­ployed for 183 days, the per­son may not re­mem-

ber the ex­act num­ber of days. So, the shorter the pe­riod, the more ac­cu­rate the re­sponses could be. But the prob­lem with shorter time span is that you do not cap­ture the sea­son­al­ity.

The third prob­lem is how peo­ple in­ter­pret ques­tions. It is most com­mon among the women sur­veys. Women, who con­sider them­selves as homemak­ers, could be car­ry­ing out a lot of eco­nomic ac­tiv­i­ties such as be­ing in­volved in an­i­mal hus­bandry or do­ing em­broi­dery. But when you ask them if they have been work­ing, they would say no be­cause they think th­ese ac­tiv­i­ties are part of their house­hold work.

Cur­rent Sta­tus We do not have the data to de­ter­mine what is the cor­rect em­ploy­ment sce­nario to­day. The only data we have is the Labour Bureau and the CMIE data. The two do not match, but both are show­ing neg­a­tive growth. So, to that ex­tent, at

Al­most all sur­veys are based on the last cen­sus data. As we start mov­ing away from the cen­sus, the qual­ity of sam­pling drops. But you can­not have a cen­sus ev­ery year

least the di­rec­tion seems to be con­sis­tent.

An­other prob­lem with the jobs data is that only know­ing the over­all pic­ture does not serve the en­tire pur­pose. The over­all pic­ture hides the dif­fer­ences, which can be sec­toral and ge­o­graph­i­cal.

I can eas­ily have a sit­u­a­tion where there is a mas­sive loss of jobs in one state and a gain in an­other state. If I have an over­ar­ch­ing national level pol­icy, it may not ad­dress the prob­lem be­cause the prob­lem can be a lo­calised one. So, you do need a cer­tain de­gree of gran­u­lar­ity in data to be able to frame poli­cies cor­rectly.

I think the pe­ri­odic labour force sur­vey that the NSSO is do­ing should be able to solve the prob­lem to a sub­stan­tial ex­tent be­cause it is an all-In­dia ex­er­cise, and gives you state-level and oc­cu­pa­tion-wise es­ti­mates.

Newspapers in English

Newspapers from India

© PressReader. All rights reserved.