Lies, damned lies and statistics

The Collection of Statistics Bill doesn't address the root of the problem which is how we can use data on actual production that is available instead of just estimates and surveys.

National income accounting is inherently beset with complexities and conceptual problems. On a number of points it involves an element of judgement regarding inclusion, exclusion or valuation, which often tends to be somewhat arbitrary, giving rise to divergent views. Therefore, despite the rigours of methodology, procedures and statistical tools
used, NIA is finally an estimate of quantification of economic activity.

In this context, the prime goal of NIA should be to increase the element of computation of actuals, as against the use of estimates and surveys to fill up data gaps. The aforesaid can be achieved by harnessing the power of IT.

The present method, purportedly adapted by the Central Statistical Organisation in proposing a new Act (the Collection of Statistics Bill), is a compartmentalised approach to the problem. It will only add to the already long list of statutes enacted by Parliament. On the other hand, an integrated and holistic approach can help achieve the desired
outcome.

Under the provisions of the Companies Act, 1956, all companies are mandated to make detailed filing to the Registrar of Companies at specified periods/events. Under the e-governance project of the ministry of company affairs -- namely MCA-21 which is being implemented in phases -- all these filings are required to be made electronically.

Along with the aforesaid filings, companies can be statutorily required to file data on production and sales, as required by the CSO, by making appropriate changes in the draft Companies Bill (and its Rules). Once this data is available in electronic form, the CSO can directly extract it from the MCA portal. Non-compliance by companies should carry heavy penalties. The CSO and MCA will have to ensure confidentiality of this data.

Coming to the informal and service sectors, the data on value addition in these sectors are believed to be of dubious quality. In these sectors also, NIA can move from 'guesstimation' to computation of actuals by following the Saudi model: every entity consuming electricity is required to furnish production data to the electricity supplier, which
is used for statistical compilation. No data, no electricity  -- simple!

Another area in NIA that can move from estimation to computation of actuals, is the measurement of household savings in physical assets.

Given

the vast improvement in data availability of housing loans disbursed to individuals by banks and housing finance companies, the savings of the households in physical assets can be directly measured from the balance sheets of these institutions instead of using the residual method.

These measures, if implemented, can dramatically improve the quality of NIA and also reduce the time lag.

Let us now look into some issues pertaining to the statistics on household financial savings:

Currency

The currency held by households is estimated on the basis of past trends of the proportion of currency held by households to total currency in circulation. The proportion of currency held by households is estimated  to be 0.93 by RBI, and has been in use since 1985-86 (CSO, 1989).

This may not be appropriate in the present scenario for several reasons. For one, after the mid '80s, the financial sector has undergone innumerable and structural changes, viz, in products, services and also in payment and settlement systems. These changes impact the currency holding behaviour of all economic agents, including households. The ratio of currency in circulation to broad money (M3) can be used as a proxy to ascertain the relative levels of currency usage over time.

Cu/M3 has shown a secular declining trend from 22.3 per cent in 1985-86 to 16.4 per cent in 2004-05.

Speaking of households, the total value of transactions through cards for the year 2004-05 was Rs 77,120 crore (Rs 771.2 billion). It was Rs 35,870 crore (Rs 358.7 billion) in 2003-04) and the number of ATMs installed as of March 2005 was 17,642 versus nil (or negligible) in the year 1985-86.

The rapid expansion in the use of credit, debit and smart cards and ATMs, though predominantly in urban areas, lowers the tendency of households to hold currency. Therefore, the method used to estimate the financial assets of households in currency, needs a re-look.

Shares, debentures and mutual fund units

The 'estimate' of financial savings of households in shares, debentures and mutual funds units for the year 2003-04 (P) is Rs 492 crore (Rs 4.92 billion). As a percentage of GDP it is negligible. How credible is this data?

Recognising the emergence of new institutions like depositories, regulators and so on, and extracting data directly from them should be an ongoing and automatic process and should not wait for committee recommendations.

The views are the author's and do not reflect those of Sebi, where he works