Validity, Clarity and Reliability[1]– These were the three ‘ity’s’ of good quality data that Akanksha’s (our research coordinator) presentation to the 2012 batch of CCS’ summer interns focused on. This got me thinking about the public finance data I work with everyday. To give you a little background, I’ve spent the last 10 months working on a project that requires me to explore, understand and analyze the Detailed Demand for Grants -a document that compiles all requests for funding by a government department in a given fiscal year; to specifically calculate how much the government spends on education. Sounds like a simple enough task right?

However, my problems started early on.  Since education is a concurrent subject- not only does the MHRD sponsor some central schemes, but each state government also sponsors their own set of local schemes. Moreover, the Department of Education is not the only department that funds education. As it turns out between 14 to 30 other Ministries and Departments (depending on the state) sponsor various Educational Schemes both nationally and at the state-level.  Some of the larger spenders include the department of Social Welfare and Justice, the Department of Planning and Public Works, the Department of Tribal Welfare, the Ministry of social justice and empowerment, the Ministry of labour and employment, the Ministry of Rural Development, and the Ministry of Youth Affairs and Sports.   Thus, to get a complete picture of the expenditure on education, one needs to collect the Detailed Demand for Grants for all these 30 Ministries at both the Centre and State level.

My first challenge was finding these documents –most of these documents weren’t available online. If they were, more often than not the links were broken. If I got lucky and did find them online and the links worked, it would only be the most recent document that was available and not the past years’ Demands.  My next stop was the NIPFP library where I spent several afternoons manually going through volumes of budgets to find what I was looking for.

The next step was analyzing these demands and actually understanding the government’s accounting system. This brought with it a new set of challenges:

I assumed that since I had managed to get my hands on the most recent Demands for Grants publically available I would have the latest data on government spending. Turns out that there is a minimum of a 2 year time lag between when these amounts are spent and when these documents are published. Thus, the most recent actual expenditures I had were for the fiscal year 2009-10.

My next challenge was the lack of uniformity in the accounting systems between the Centre and states. The Detailed Demand for Grants uses the standard coding structure as prescribed by the Comptroller and Auditor General (CAG) of India.  However, there is little or no uniformity beyond minor head level of classification as some States continue to follow the old series of coding structure (as was in use prior to 1st April, 1987). This leaves a lot open to the interpretation of the user and reduces the overall clarity of the document. Additionally, even the unit of reporting i.e rupees thousands, rupees lakhs and rupees crores differs across states which makes standardizing and comparing this data twice as hard.

The Detailed Demand for Grants is a head-wise classification of public expenditure- the Major heads are intended to represent the major functions of the government, the sub-major heads are intended to represent the sub-functions of the government and so on. However, it also includes Major Head 2552: Lump sum provision for North Eastern Areas and Sikkim which in my mind is a geographical classification as opposed to being a functional classification. So how well does the Demand for Grants really measure what it was really intended to measure i.e what is the validity of the data?

Despite all these set-backs, I was excited about finally getting to the bottom of these documents when this made news- the CAG got its crores and lakhs mixed up ! Agreed, the mix up was not in the Demand for Grants. But, this forced me to think- what was the reliability of the numbers in the Demand for Grants? Moreover on a recent visit to the MHRD, I was informed that these Demands used to be typed on MS Word until about 2 years back (which was when they moved to MS Excel) and all calculations were done manually on a calculator and then entered into the Word document! – Which greatly increases the scope for human error.

While tools like the RTI have helped make public finance data more accessible, the concern that it meets Akanksha’s three ‘ity’s’ of good data remains.  Hopefully the new account structure that the Government intends to implement over the next couple of years will be a step in the right direction to good quality data….one ity at a time.

 


[1] Validity: The degree to which a measure reflects the concept it is intended to measure

  Clarity: Is the data sufficiently well defined that all users will interpret it similarly

  Reliability : The degree to which measurement method would collect the same data each time in repeated observations

Post Disclaimer

The opinions expressed in this essay are those of the authors. They do not purport to reflect the opinions or views of CCS.