Consequences of poor GDP measurement


Business Standard, 28 May 2018


Has Indian GDP been overestimated from 2012 onwards? This is ordinarily an obscure debate of interest to a tiny set of economists. If GDP were mismeasured, this may change the thinking of global portfolio managers, but apart from that, nothing much seems to be at stake. But GDP measurement drives macro policy, and numerical values of GDP directly go into fiscal planning. When GDP is overestimated, it puts pressure on tax collection and in the government borrowing programme.

In the tiny community of economists, there is an even smaller group of people who have an interest in economic measurement. That community has been absorbed in a debate about the difficulties of GDP measurement in India. When we compare with the firm data, GDP is overestimated. When we compare against a large number of independent measures, GDP is overestimated. If there is over-reporting of GDP growth in each year from 2012 onwards, this cumulatively adds up to a large overstatement of the level of GDP in 2018.

Many economists have retreated from taking the Indian GDP data seriously until the discrepancy is resolved. Alternative output proxies are utilised, through which business cycle conditions can be roughly measured, without requiring the use of official statistics. But as a card-carrying economist, it gives me immense sadness to think that I cannot trust official statistics. GDP is a core concept of macroeconomics. A macroeconomist who is deprived of GDP data is like a doctor who is deprived of a thermometer.

This is edge-of-the-seat excitement for macroeconomists. But for most people, this is a storm in a teacup. Whether GDP is mismeasured or not has little consequence in any practical sense. It may matter for the bragging rights of the administration, and it may shape the allocation of capital to India by global portfolio managers, but for the rest, nothing seems to turn on the numerical values of GDP. No manager in India waits for a GDP release, or uses this as a critical input into decisions about production, inventory or investment.

However, macro policy is flying blind if GDP is not properly measured. When a business cycle downturn is coming, we want the fiscal deficit to enlarge and vice versa, but this is infeasible in India as we have shabby measurement. For the monetary policy committee (MPC) to think about monetary policy, they need to know the state of the business cycle. For the Finance Commission to do its work, it requires SDP estimates. Weakness in measurement is hobbling our policymakers.

There is one place in the economy where numerical values of GDP play a mission critical role: The Budget process. In February 2018, Budget makers take a stand on their projected value for nominal GDP in 2018-19. That numerical value feeds into the entire Budget process. The first draft of the Budget for 2018-19 can be made by taking all ratios to GDP for last year and applying them to the new GDP value. As an example, if a certain tax yielded 3 per cent of GDP last year, and if next year's GDP estimate is Rs 150 trillion, we immediately get a first cut of the Budget estimate for the coming year for that tax: 3 per cent of Rs 150 trillion, or Rs 4.5 trillion.

The Budget process is shaped through and through by ratios to GDP. Whether we think about taxation, expenditure, deficits, borrowing or debt, in all these areas, the Budget process uses the projected value for nominal GDP and multiplies this with a sound value for the ratio to GDP. As an example, the discussion on the fiscal deficit is only conducted as a percentage of GDP. Once the Budget makers agree that they want a fiscal deficit of 2 per cent of GDP, this is multiplied by the GDP forecast to get the Budget estimate for the fiscal deficit in the coming year, in rupees.

What would happen in the Budget process if GDP were overestimated? Tax targets would then be set too high. There would be gnawing problems in achieving those tax targets. This would result in many small decisions to increase tax rates in order to get back to normative estimates for tax collections.

India has the remarkably bad concept of tax collection targets assigned to senior managers of the tax administration. When GDP is overestimated, the targets sent to the CBDT hierarchy are too high. Perhaps this gives a greater propensity to use harsh tactics in collecting taxes.

Similar problems show up with deficits and the borrowing programme. When GDP is overestimated, a borrowing plan that appears reasonable in terms of the borrowing/GDP ratio is one that involves asking for too much debt from the economy. This results in stress where the market is not able to absorb the borrowing.

The scenario of overestimating the level of GDP thus results in three predictions: Pressure to raise tax rates in tax policy, pressure to use harsh tactics in tax administration, and difficulties in executing the borrowing programme. These three problems are indeed present in varying degrees in the observed reality around us.

This is a reminder of the importance of sound economic measurement. GDP measurement was once the prestige project of India's best economists. Many of the top economists of India of the previous two generations devoted enormous effort to building the statistical system, and they were proud of the edifice they had constructed.

Unfortunately, the economists of my generation have lost interest in problems of measurement. We are now just users of data; we don't take an interest in how it is made. The editors and referees of foreign journals think that the data emanating from the government is always kosher, and academic economists are content with producing wrong research that is career-maximising. GDP data is, however, a mission-critical input into the fiscal process. We must diagnose what is wrong with GDP measurement, and build a capable statistical system.


Back up to Ajay Shah's 2018 media page
Back up to Ajay Shah's home page