Devil in the detail: Data collection a weak link in new GDP calculation

Improvement in methodology of GDP calculation or for that matter the Base Year is usually a tame routine affair. However, the launch of new GDP Series (Base 2011-12) in India has been anything but tepid, what with everyone from RBI Governor to Chief Economic Adviser publicly expressing their struggle with understanding the new growth numbers. As such GDP cannot be measured in an ‘empirical ‘sense. One may count with certainty the number of cows in a dairy farm or weight of food grains in a warehouse. On the contrary, GDP can only be estimated and not measured with absolute certainty. The output of this estimation is expected to be reflective of the aggregate economic activity of the nation. [caption id=“attachment_2278838” align=“alignleft” width=“380”] Reuters[/caption] To the extent GDP is defined as market value of final goods and services produced in an economy, the Income Based Approach that CSO has recently adopted is more aligned to the classical concept of GDP. The newly adopted framework is an international best practise. Thus per se the intent should not be doubted. However, certain doubts remain as to the weather data sources being used for this advanced approach are fully supportive of this advancements or not. What exactly is GVA? Under the recently adopted income-based approach, GDP is expressed as aggregate gross value added (GVA) plus product taxes(such as excise, sales, service tax) minus product subsidies(such as petroleum, fertiliser subsidies). The value added in an economic activity is the output less the value of inputs. The value created is distributed within various entities involved in the process of value creation. The value added on economic activity at a country level is termed GVA. Thus Gross Value Added (GVA) is given as the sum of a) compensation of employees; b) consumption of fixed capital which is often captured from ‘depreciation’ in corporate profit and loss (P&L) account; c) operating surplus which may be approximated by Profit After Tax (PAT) adjusted for property income and current transfers to trade channels and d) Production Taxes less Subsidies. Production taxes/subsidies needs to be differentiated from product taxes/subsidies since the former is related to the overall production process itself and includes items such as Land revenues, stamp/registration fees while the latter is specific to actual units produced. Well over 90% of the GDP is driven by GVA. If one precisely knew the P&L of all business entities (registered as well as unregistered businesses from the largest listed corporate to the smallest self-employed person) in India, the GVA could have been calculated ‘almost perfectly’ with the assumption that businesses are correctly reporting their profitability. However, this is a near impossible task for even more developed economies. The residual portion of the GDP consisting of taxes net of subsidies may be more accurately measured. For estimating operating surplus and consumption of fixed capital on an economy wide basis understandably the key data source is company financial statements. Of course, various nation-wide surveys are used not only for estimating employee compensations but also for supplementing the information available from company financials. Let’s look at each of them in some detail. Corporate information used in GVA In the GDP calculated as per 2004-05 series, the key source of information about private corporate sector has been RBI Study on Company Finances, which considered financial results of around 2500 companies (around 1500 non-financial corporates). The new series justifiably attempts to increase the coverage of the corporate sector and has used the MCA21 database maintained by the ministry of corporate affairs. Approximately 14 lakh companies are registered with MCA, of which 9.8 lakh companies are active. Post filtering for data availability, 5 lakh companies have been analysed and used for GDP estimation for 2011-12 and 2012-13. While coverage of the private sector has improved, it has some downsides as well. Not the least of the shortcomings is the instability in the data, since as of 2013-14 the number of common companies was reduced to just 3 lakhs. This is an outcome of companies not reporting possibly because they are closing down their operations. Thus, if out of 5 lakh companies 2 lakh have not reported, it should normally set alarm bells ringing about the economy. How the current methodology addresses this ‘survivor bias’ in the data is not clear. However, the much bigger problem is the quality of the financial statement of a lot, but of courses not all, of these companies. As such both under-reporting and over-reporting corporate profits will now distort GDP calculations since GVA is heavily dependent on such variables. At least three entities have published research on earnings management and quality of financial reporting of Indian listed entities. The first titled ‘Study on the state of Corporate Governance in India’, done under the auspices of Indian Institute of Corporate Affairs, analysed 2,315 corporates over 2006-11. The second study, conducted by SEBI-DRG titled ‘Earnings Management in India’, studied 2,229 firms over 2008-11. Both these studies focussing on non-banking/non-financial services corporates had comparable findings. While both studies acknowledged the prevalence of ‘earnings management’ issue among some Indian corporates, they specifically highlighted that the issues of doubtful financial reporting increases as the size of the firm decreases. The third study conducted by India Ratings & Research (Ind-ra), which focussed exclusively on the BSE 500 corporates, agrees with the findings of the two above reports. In addition, the Ind-Ra study found that possible discrepancy in reporting financial number increases at the time of economic slowdown. During stressful years, questions about reliability may be raised about variables such as PAT and EBITDA of even BSE 500 corporates. So it is possible that companies over-report profit in bad times to secure financing and under-report profit in good times to pay lower taxes. More research on corporate reporting behaviour and its motivations are required to draw stronger conclusion. Still as of today some doubts may be raised of the reliability of such corporate numbers and by extension one may conjecture whether GVA is getting over-estimated to some extent. Frequency and timeliness of surveys The usage of surveys to supplement information is unique neither to the new methodology adopted by CSO nor to India. To substantiate and supplement the data from other sources, under the new approach certain survey results and the time by which those results are available become critical. It is the author’s conjecture that the sensitivity of final published GDP figure to such survey inputs may be higher as per the new approach than the erstwhile approach. Annual Survey of Industries (ASI) is critical for GDP estimation. When the new methodology was launched in 2014, the ASI data of 2011-12 was available. Ironically, results of this annual survey are typically available with a two-year lag. As of March 2015 only Volume I of ASI 2012-13 is made available. It is not clear whether the latest GDP figures have fully incorporated the ASI 2012-13 to the extent it is available. Another survey critical to the new approach is Employment and Unemployment Surveys of National Sample Survey Organisation (NSSO). This survey is conducted once in five years with the last one being conducted at 2011-12. These surveys reflect structural changes in the economy. As such, if the economy moves at an even pace of growth (or for that matter decay) statistical outputs based on such surveys reflect reasonably the ground level realities. However, if the economy undergoing the structural change or the economic growth is volatile or is at inflection point then questions may be raised about using inputs from some dated surveys. The years 2010-11 and 2011-12 were of robust economic activity with GDP growth rate of 8% plus (as per the 2004-05 series). In 2012-13 the economy nose-dived. Clearly data wise this was a period of economic volatility. Estimates used in calculating GDP based on survey conducted prior to 2012 may create an upward bias in the estimates. To the extent, global approach has been adopted the focus should be two fold. In the short term, which is next two-three years CSO may consider focussing on clearly articulating the various specific assumption and critical inputs specific to the period for which GDP is estimated. In the long term CSO should consider deploying more people and technology resources to improve the frequency of various surveys as well as reduce very substantially the time lag between the survey and the publication of the reports. And last but not the least, stronger measures in line with the ones proposed in the original draft version of Corporate Act 2013 may be implemented to ensure higher quality of financial reporting but all business entities in India. The author is senior director - corporate director, India Ratings. Views are personal