Former NSC chief PC Mohanan warns against using statistics as a political tool, says data is a public good

This debate helped us navigate through the myriad problems in data and data collection methodology.  Another debate is now in full swing in the media on Indian official data, but with uncertain outcomes.

Over the last two months, official data has entered the public consciousness through a series of front-page headlines, in print, visual and online media like never before, most of which are centred around two key data sets -- the GDP data and the NSSO report on employment and unemployment.

The debate is still continuing though it has not bought any clarity on the issues in the absence of the data being denied to the public as is the case with the employment data, or the methodology not fully explained as in the case of the GDP.

It will be useful to look at the key takeaway for the statistical system from the events of the last six weeks starting from the resignation of the last two remaining members of the National Statistical Commission (NSC) and the publication of a part of the NSSO report in a newspaper. While doing so we also examine the relationship between statistical agencies and the government in light of the recent debates.

The aftermath of the resignations of the members of the NSC bought in plenty of commentaries on the integrity and autonomy of Indian official statistics from economists, media and the political establishment, though the members resigned feeling side-lined.

The official reactions came from the vice chairman of NITI Aayog and his aides and also from the finance minister. This was somewhat surprising as the government‘s own gazette notification issued on June 15, 2016, notifying a set of ‘fundamental principles of official statistics’ unambiguously stated that only “the statistical agencies are entitled to comment on erroneous interpretation and misuse of statistics”.

The silence of the statistical agencies, in this case, the Central Statistics Office (CSO) and the NSSO, perhaps points to the subordination of these agencies by others who have a stake in maintaining a particular narrative of the government, no matter the data.

The failure of the official statisticians or the external technical experts associated with CSO and NSSO to respond to the issues has possibly allowed the debate to stray more in to a political zeitgeist rather than a technical one. The initial government reaction towards the leaked NSSO report centred on it being a ‘draft report’ or an ‘unapproved’ one, indirectly admitting the report and the veracity of its contents. Without the official report, the leaked numbers on employment-unemployment had an easy entry into the political arena.        

The argumentative nature of our polity and the wide-spread access to independent media and the failure of the government to address what in effect are technical issues make the current debate a unique one in our country. But at the international level, this is not so.

Examples galore of government dealing with uncomfortable statistics in a wide variety of ways. A recent publication by the Economic Governance Support Unit of European Parliament on ‘Statistical Governance in Greece – recent Developments’, (November 2016) deals with the developments of the Greek statistical system.

It also covers legal proceedings before Greek courts against Andreas Georgiou (head of the Greek statistical office, ELSTAT, between August 2, 2010, and August 2, 2015), among others on the ground of the accusation that he inflated 2009 Greek deficit figures. To quote from the report:

“After almost six years of detailed deliberations, the criminal charge against Mr Georgiou and two senior staff of ELSTAT of falsifying the 2009 deficit (and debt) figures and thereby causing Greece extraordinary damage is still being pursued. This situation pertains even though investigating judges, prosecutors and the Council of the Appeals Court have on six separate occasions recommended, or decided, that the charges be dropped. The most serious reversals have been the proposals by the Chief Prosecutor of the Supreme Court to annul on two separate occasions (September 2015 and July 2017) the acquittal orders issued by the Council of the Appeals Court.  On the 1st of August (2017) the Appeals Court acquitted Mr Georgiou on two of the three misdemeanour charges of violation of duty but convicted him on the third charge of not obtaining prior approval for the release of the 2009 EDP figures from the Board of ELSTAT. The court sentenced him to two years in prison, suspended for three years. On the basis of this verdict and sentence, it would appear that the fact that Mr Georgiou was acting in accordance with the European Statistics Code of Practice was not taken into account by the Court to any significant extent in reaching its decision…” (emphasis added).

As everyone knows one of the major causes of the collapse of the Greek economy has been attributed to the frequent revisions of official statistics. The frequent revisions made them unreliable according to the inspection reports of the Eurostat, the European agency overseeing the official statistics for the European Union.

Prosecuting the chief statistician for following the established procedure of compiling official data on the ground of falsifying numbers as these were not to the taste of the government has no known analogy. But the reaction of the government and to an extent that of judiciary disregarding the European statistical code is difficult to accept. This is an extreme case but very useful to illustrate the centrality of official data.

William Seltzer in a report prepared for the UN Statistical office (UN, 1994) deals with many possibilities of politics coming in to play in official statistics. The report quotes the Norwegian Central Bureau of Statistics (CBS) as saying ”...CBS tries to avoid provocative comments relating to issues which are hotly debated between our political parties. However, the CBS will (as a policy) not refrain from putting its statistics in a relevant social or economic context, irrespective of how this will be interpreted by various parties…The CBS will show circumspection, while Ministries and Ministers occasionally will have to live with some statements from the CBS that they do not like; extreme caution would lead to a reduction in the social usefulness of the CBS (emphasis added).”

There are many cases of extreme political interferences in controlling of official data which mostly come from countries under dictatorships of an era gone by or in the communist regimes of the past. International agencies, especially the UN statistical division, World Bank, IMF etc have been in the forefront of developing policies and principles that help insulate statistical agencies from as much political pressure as possible and help maintain the integrity and autonomy of the statistical systems in member countries.

Coming to the Indian context, it is instructive to see how data entered the jobs narrative. In a widely quoted interview, Prime Minister Narendra Modi is quoted as saying “…On this issue, more than a lack of jobs, the issue is a lack of data on jobs. Our opponents will naturally exploit this opportunity to paint a picture of their choice and blame us”. He went on to give more details to indicate where and how jobs are being created. "…If we look at numbers for employment, more than 41 lakh formal jobs were created from September 2017 to April 2018 based on EPFO payroll data. According to a study based on EPFO data, more than 70 lakh jobs were created in the formal sector last year.…. In just one year, 48 lakh new enterprises got registered. Will this not result in more formalisation and better jobs? More than 12 crore loans have been given under Mudra (micro loans). Is it unfair to expect that one loan would have created or supported means of livelihood for at least one person? More than one crore houses have been constructed in the last one year; how much employment would this have generated? If road construction has more than doubled per month, if there is tremendous growth in railways, highways, airlines, etc, what does it indicate?"

Quoting pieces of evidence of employment generation with poor statistical legitimacy is certainly due to the absence of independent data from labour force surveys. However, these ‘leading’ and ‘loaded’ questions coming from the Prime Minister if addressed to the national statistical agencies involved in measuring employment through scientific sample surveys would most likely put them under extreme caution.

It is thus easy to understand the reluctance of the agencies to release the NSSO jobs report; if the leaked figures from the report is any indication of the actual findings of the labour force survey.

It is therefore imperative for statistical agencies to keep an ‘arm’s length’ distance from the government. This is not always easy as the statistical system is embedded in the government and no non-governmental agency can replace them as the statistical operations heavily depend on the government administrative machinery.

During the last quarter century, most modern democratic governments have established independent mechanisms to oversee the functioning of the statistical system. In the UK, whose administrative arrangements can find many parallels in our country, these reforms started in the last decade of the last century.

The Labour Party then in opposition promised to give the statistical system autonomy in their election manifesto. The UK has now an independent statistics board to oversee the functioning of the office of national statistics with almost no interference from the government.

Statistical reforms in India actually started a little earlier with the establishment of a National Advisory Board on Statistics (NABS) with the minister for statistics as the chairman and a large number of stake holders as members.

The board was to advise the government on all statistical issues. Rangarajan commission noted that the NABS was intended to be an institutional arrangement for the statistical system as a whole, but it did not succeed because it lacked legal backing. It further recorded in its report that as official statistics play a major role in assessing the performance of government, it is important that such statistics are not only accurate, but are also trusted as such by the layman as well as by its principal users.

The commission recommended the creation of a permanent and statutory apex body – the National Statistical Commission – independent of the government and responsible to the Parliament in respect of policymaking, coordination and certification of quality of core statistics.

Following the recommendations of the Rangarajan commission, a permanent National Statistical Commission came into being in 2006 as an apex and autonomous agency with a very wide mandate that covered almost all the substantial issues relating to quality, periodicity and regulation of all core statistical activities of the central and state agencies.

The NSC was also given oversight powers on all technical aspects of NSSO functioning, hitherto enjoyed by an independent governing council. However, the NSC lacked a legal entity status to ensure statutory validity for its recommendations.

This in due course diluted the importance of the commission as an apex body with many of its recommendations yet to be implemented by the official agencies and gradually treated as simply advisory in nature. The need for the commission to depend on the ministry for administrative support in the absence of its own administrative set-up led the ministry and its officials to treat the commission as an inconvenient appendage.

The gradually diminishing importance of the commission with deliberate intent was cited as the key reason for the resignation of its members. As noted in their letters, the ministry of statistics in its eagerness to promote new initiatives announced a national economic census of all establishments and several other surveys besides the introduction of a national policy on official statistics without any reference to the commission. Some of these initiatives like the economic census which are very resource intensive are being repeated without critically examining the results of the previous exercises and learning lessons from them.

Commentators have also critically viewed the involvement of NITI Aayog in the release of the back series of GDP and its effort to defend the non-publication of the NSSO report as clear examples of government getting involved in official statistics disregarding the commission specifically set up for this purpose.

This is a most unfortunate development when the need is for the statistical agencies to become more professional and absorb the new horizons opening in the digital world.

Unlike in the past debates on NSSO survey data, where the issues remained mostly confined to the technicalities of the data generating processes and the interpretations, the current debate has less to do with them.  Attempts to bring in technicalities were lost in the steadfast refusal to release the survey report for research scrutiny. The entire efforts were to discredit the numbers purported to be from the report and contradict them with alternate narratives not always using numbers.

The other numbers dragged into the debate are the GDP numbers, the holy cow of the official statistical system. The GDP is a highly summarised number that describes how an economy functions and the entire set of related numbers form what is collectively called the national accounts statistics.

The GDP computation procedure follows a very rigorous methodology detailed in the system of national accounts, a product of the joint efforts of experts from international agencies under the guidance of the UN Statistical Division published in 2008 after several years of consultations. India adopted this latest version substantially in its new series.

The  GDP has a starting year called the base year and the yearly GDP and quarterly GDP allows the computation of growth rates used as an indicator of the performance of the economy during the reference period. The base year gets revised at periodic intervals to reflect the changing structure of the economy more realistically and avail of better data sources.

The GDP of India was revised in 2015 to a new base year corresponding to 2011-12. This revision for the first time not only used the concepts given in SNA 2008 but also made use of the corporate sector data from statutory annual company filings besides a host of other improvements in data and methodology.

The CSO came out with a detailed methodology paper when the new series was developed. Though there were some questions on the numbers from economists, the CSO did explain the methodology that produced these numbers.

The GDP numbers entered the debate when the CSO extended the new series backward to 2004-05. Without going into the merits of the methodology used to construct the back series for which the company filings data were not available, the timing and manner of the release of the new back series that covered the period when the present opposition was in power raised questions.

For one, the new series was released under the guidance of the NITI Aayog which was never in the past involved in the release of government statistics. Second, the new series provided a picture of economic growth during the previous regime quite opposite of what came out in another exercise in GDP back-casting attempted by a few experts in a report commissioned by the NSC.

The contesting pictures produced by the two sets of figures have been the basis of the debate on GDP and the GDP numbers entered the political arena as soon as they were released. In the normal course, the NSC or the official statisticians from CSO could have explained the situation and ended the controversy or rescued it from political hands.

An unfortunate fallout of the debate on GDP has been the questioning of the credibility of the entire gamut of official data. Economists have pointed out certain anomalies in the GDP numbers that appear to be contradictory to events in the economy like the demonetisation or the differing sectoral structure in the old and new series and thus failing the ‘smell test’.

However, the system that produces these numbers are in the hands of extremely competent professional statisticians. Possibly better communication between the CSO and the experts could have helped clear these questions just as the release of the NSSO report and the accompanying micro-data would have allowed the economists and statisticians to provide answers on the employment question far more convincingly than politicians or chartered accountants.

Clearly, the issues raised by economists on the GDP data require careful analysis in a transparent manner as was the case with the NSS data reported in the ‘Great Indian Poverty Debate’. However, the political nature of the discourse with ‘contrarian’ views strictly not in fashion appear to minimise the chances of this happening.

The absence of a functioning statistical commission has left the field open for people with little interest in technicalities to take the centre stage. In such a situation the current debate is unlikely to benefit the statistical system. The only course open for the official statistical system is to show professionalism and openness and start a dialogue with the stakeholders. National statistics is a ‘public good’ and allowing it to be a tool of politics even for a short time will damage it and repair works could take a long time.