IN DE­FENCE OF ES­TI­MATES OF JOB LOSSES

Business Standard - - FRONT PAGE - MA­HESH VYAS The au­thor is manag­ing di­rec­tor and CEO, Cen­tre for Mon­i­tor­ing In­dian Econ­omy P Ltd

In July 2017, I wrote in this col­umn that the num­ber of peo­ple em­ployed dur­ing Jan­uary-April 2017 was 1.5 mil­lion less than those em­ployed in Septem­ber-De­cem­ber 2016. The num­ber of em­ployed fell from 406.5 mil­lion to 405 mil­lion. Em­ploy­ment fell fur­ther to 404.6 mil­lion dur­ing May-Au­gust 2017.

I had con­jec­tured that the 1.5 mil­lion fall in em­ploy­ment dur­ing Jan­uary-April 2017 could be at­trib­uted to the Novem­ber 2016 de­mon­eti­sa­tion. I had also added a caveat that the fall could also be be­cause of sea­son­al­ity but since we do not have a long time se­ries it was not pos­si­ble to ad­just the fall for sea­son­al­ity be­fore at­tribut­ing it to de­mon­eti­sa­tion.

This con­jec­ture has been crit­i­cised by Bibek De­broy (Moon­watcher’s Logic,

The In­dian Ex­press, Oc­to­ber 19) as post hoc, ergo propter hoc fal­lacy. Later, it was crit­i­cised by TCA Anant that if I ad­just for sea­son­al­ity then I will not see any fall in em­ploy­ment. Sur­jit Bhalla (“No Proof Re­quired: De­mon­eti­sa­tion and its Con­tents”, The In­dian Ex­press, Novem­ber 8) has com­pared the Jan­uary-April 2017 em­ploy­ment data with the Jan­uary-April 2016 data and shown that em­ploy­ment has gone up and so, em­ploy­ment has gone up post-de­mon­eti­sa­tion. There could be many more crit­i­cisms but I know these three from very em­i­nent economists.

Bibek De­broy’s crit­i­cism that I can­not at­tribute an ob­ser­va­tion to a phe­nom­e­non just be­cause the ob­ser­va­tion came af­ter the phe­nom­e­non is facile. De­mon­eti­sa­tion was a huge shock to the econ­omy. It was ex­pected to have a se­vere short-term im­pact upon the econ­omy. The long-term im­pact is a gam­ble but the short-term im­pact was ex­pected by ev­ery­body. Even the Prime Min­is­ter asked for a lit­tle time to solve a big prob­lem. News­pa­pers nar­rated count­less sto­ries of job losses for weeks af­ter de­mon­eti­sa­tion. The CMIE-BSE real-time mea­sure­ment of un­em­ploy­ment gave us the op­por­tu­nity to ob­serve its im­pact in real time. Even­tu­ally, we will be able to do more rig­or­ous nat­u­ral ex­per­i­ments to un­der­stand this im­pact bet­ter. But, for now, we have a mea­sure of the im­me­di­ate im­pact.

Sev­eral em­i­nent economists have told me that a job loss of 1.5 mil­lion post­de­mon­eti­sa­tion is pos­si­bly an un­der­es­ti­mate. Pos­si­ble. 1.5 mil­lion is a net num­ber. Gross job losses were larger and these were off­set by job gains else­where. This is a reg­u­lar af­fair. What mat­ters is the net in­crease or fall in jobs.

TCA Anant was men­tion­ing what I had al­ready pointed out too, in my piece in Busi­ness Stan­dard on July 11, that the em­ploy­ment num­bers should be sea­son­ally ad­justed. His cri­tique was that I had made only a fleet­ing men­tion and not em­pha­sised this suf­fi­ciently. I may plead guilty to that. He also said that once the num­bers are sea­son­ally ad­justed then I may see no fall in em­ploy­ment. This could be a fair con­jec­ture and I do not dis­agree that im­pact of sea­son­al­ity must be in­ves­ti­gated when we have more data. But, then why doesn’t he, as Chief Statis­ti­cian of In­dia, pro­duce sea­son­ally ad­justed se­ries for the IIP and other fast-fre­quency in­di­ca­tors gen­er­ated by the of­fi­cial ma­chin­ery. Why pick on a very young se­ries gen­er­ated pri­vately?

Sur­jit Bhalla is a statis­ti­cian-wiz­ard, a num­bers ma­gi­cian be­sides be­ing an em­i­nent econ­o­mist. He ad­dresses the prob­lem of sea­son­al­ity by mak­ing yearon-year com­par­isons in­stead of se­quen­tial com­par­isons as I do to say that 1.5 mil­lion jobs were lost af­ter de­mon­eti­sa­tion. He uses the BSE-CMIE es­ti­mates to show that jobs grew by 4.2 mil­lion in Jan­uary-April 2017 com­pared to Jan­uary-April 2016. Since these two are like pe­ri­ods, he seems to sug­gest that there is no sea­sonal vi­ti­a­tion. Fur­ther, since de­mon­eti­sa­tion hap­pened be­tween these two pe­ri­ods, we can say that shock­ingly, jobs grew af­ter de­mon­eti­sa­tion.

Sur­jit is, of course, wrong. He is wrong for two rea­sons. First, he is wrong on an ele­men­tary sta­tis­ti­cal prob­lem. A year-on-year com­par­i­son is not a sea­sonal ad­just­ment or a cor­rec­tion for sea­son­al­ity in any way. A year-on-year com­par­i­son en­sures we are com­par­ing like months, that’s all. This is not a sea­sonal ad­just­ment.

Fur­ther, while it is true that 4.2 mil­lion jobs were added be­tween Jan­uaryApril 2016 and 2017, these were not added post-de­mon­eti­sa­tion. As many as 5.7 mil­lion jobs were added dur­ing MayDe­cem­ber 2016 (a pe­riod that is mostly be­fore de­mon­eti­sa­tion) and 1.5 mil­lion jobs were lost dur­ing Jan­uary-April 2017, that is post-de­mon­eti­sa­tion.

Se­condly, to un­der­stand the ef­fect of an event (such as de­mon­eti­sa­tion) we have to see the out­comes (such as em­ploy­ment) be­fore and af­ter the event. A sea­son­ally ad­justed se­quen­tial com­par­i­son is the right way of do­ing this, not a year-on-year com­par­i­son.

While 4.2 mil­lion jobs were added from Jan­uary-April 2016 to 2017, these were not added post-de­mon­eti­sa­tion

Newspapers in English

Newspapers from India

© PressReader. All rights reserved.