New GDP series faces fresh questions after NSSO discovers holes
New Delhi: A key database introduced in India’s new gross domestic product (GDP) series has now been found to be full of holes, raising fresh questions over the controversial and contested GDP numbers in Asia’s third-largest economy.
A study conducted by the National Sample Survey Office (NSSO) in the 12 months ended June 2017 and released last week has found that as much as 36% of companies that are part of MCA-21 database of companies and are used in India’s GDP calculations could not be traced or were wrongly classified.
The results were so disappointing that two detailed reports based on the survey had to be junked. It is worth noting that these companies were deemed as “active companies” by the ministry of corporate affairs (MCA), which includes any company that has filed returns at least once in the past three years on its list of active firms.
Statisticians say the use of the untested database in India’s national accounts also raises troubling questions about the decline of the Central Statistics Office (CSO), which was once a globally renowned institution, and the reliability of India’s official statistics.
“This is a devastating blow for CSO,” said R. Nagaraj, a professor at the Indira Gandhi Institute of Development Research in Mumbai. “Some of us had repeatedly asked CSO officials to verify the MCA-21 numbers before using them in national accounts, but they finalized the new series without adequate scrutiny and debate.”
The key change in the new GDP series launched in 2015 was the use of MCA-21, which CSO sourced from MCA. Even at the time it was being introduced in the national accounts calculations, several economists had raised questions on this issue (see “The truth behind India’s new GDP numbers”, Mint, 2 April 2015). Nagaraj was among the first to raise red flags on this.
Critics argued that the database includes many fictitious or shell firms that exist only on paper. They also said the methodology used to plug in the MCA-21 numbers in the national accounts tends to lend an overestimation bias in the GDP numbers. They demanded the MCA-21 data be released to researchers and the public so that the unit-level data could be examined. Even those who thought the new GDP series represented a great methodological leap by CSO made the same demand.
So far, India’s national accounts statisticians at CSO have defended the use of the new database although they stopped short of making it public. But now, their own colleagues from NSSO have warned about the presence of a large number of ghost firms in the database.
NSSO got into the act while carrying out a survey on the service sector (74th round), supposed to be a first-of-its-kind survey on the service sector. The MCA-21 database was used as part of the sampling frame for the survey as it had addresses and other details of firms. Business registers in states that had such registers and data from the last economic census were the other parts of the sampling frame.
This survey strategy was approved by the National Statistical Commission (NSC) two years ago, and later even a tabulation plan for the two reports that were to be generated on the basis of this survey was approved by it.
The disappointing results from the field led NSC to suggest a short technical report instead of the two reports it had sanctioned earlier.
That NSSO report highlights the gravity of the problem. Fifteen percent of the firms listed in the MCA-21 frame either could not be traced or were found to be closed. Another 21% were found “out of coverage”, suggesting they were no longer operating as service sector firms though they had registered—and were being captured in national accounts—as such. Still others, which were correctly identified, did not respond to survey questions, or didn’t maintain proper accounts on the basis of which enumerators could capture data on those firms.
“The problem of non-response was severe in case of units chosen from MCA frame,” the report says. “About 45% of MCA units were found to be out-of-survey/casualty while EC/BR (economic census/business register) frame had about 18% of such cases.” The report shows that many of them are fake or shell firms, said Nagaraj. “They remain legally registered but without producing goods and services.”
P.C. Mohanan, former NSC member and former NSSO chief, said the MCA-21 database did not receive the scrutiny it should have. Mohanan, who resigned from NSC in December over the suppression of an NSSO jobs report, was at NSC when the 74th round results first came in.
“The CSO should have done some kind of critical scrutiny and validation before using the MCA-21 database in the new GDP series, either through quick surveys or by comparing with other databases, or consultations with accountants familiar with company filings,” said Mohanan. “That kind of critical examination was always done by CSO whenever they introduced new databases earlier, but I am not sure if MCA-21 data was examined in the same manner. This was all the more important because unlike other databases such as ASI or NSSO surveys used by CSO, this database is not publicly available.”
Nagaraj said the database should be made public, and that a statistical audit by independent experts was sorely needed.
The then chief statistician, T.C.A. Anant, under whose watch the new GDP series was prepared, declined to comment, saying he was a member of the Union Public Service Commission now, and would not like to speak to reporters. A spokesperson for the ministry of statistics and programme implementation did not comment on the implications for the GDP series.