Working out the true impact of demonetisation
Indian macroeconomic fundamentals have been given a good scorecard by capital markets, going by the volume of capital flowing into the country. We should probably be planning ahead for policy responses when the flow turns into a flood, similar to what happened in the years preceding the global crisis of 2008.
What is interesting is that this is happening when the most basic metric of macroeconomic performance, the real growth rate, remains a contested figure. The credibility of the current gross domestic product (GDP) series, with base 2011-12, got a new jolt after the revised figure for 2016-17 showed headline GDP growth at 7.1%, the same as the first advance estimate issued in January, based on the first two quarters. Several observers expressed surprise at the seeming implication that demonetization did not have a negative growth impact in the second half of the year.
But the demonetisation impact does show up when the GDP numbers are unpacked. To get an effective handle on growth (in everything that follows, growth is estimated from real inflation-adjusted aggregates), it is best not to go by the reported headline number, which in the new series is based on GDP at (constant) market prices, to give it its full technical term. Instead, it is best to go back to estimating real growth from the gross value added (GVA), as we have traditionally done. GVA happens also to be a GDP aggregate—one that is measured not at market prices (which include indirect taxes like excise and value added tax, net of subsidies), but at valuations excluding those tax impositions.
GVA is more reliable than GDP at tax-included market prices because the net indirect tax add-on poses problems. There are two pieces going into net indirect taxes—the VAT or service tax or excise as the case may be (all to be soon replaced by goods and services tax, or GST), added on at the time the good or service changes hands. The second piece is price subsidies, which function in effect like a negative indirect tax. So price subsidies are subtracted from indirect taxes to obtain net indirect taxes.
In a country like India, where both indirect taxes and subsidies vary from year to year, and are levied by both the Centre and 29 state governments (not to mention local governments in some states), an aggregate including those levies is bound to suffer from measurement problems. Further, to get a growth rate from the real aggregate, you have to deflate the nominal aggregate. Without going into the tiresome details of how that deflation should be configured, let me just say that there are contestable issues with how to handle variations in the rate of levy of both taxes and subsidies. And with the GST looming, it is best to side-step that aggregate altogether.
The switch to GDP at market prices as the generator of headline growth numbers in the new series was driven by persistent pressure from a consortium of international agencies, led by the UN statistical office, to get growth rates of all countries on to a uniform format. Needless to say, this uniform format is based on practices in the developed world, where the pattern of indirect taxes and subsidies is predictably uniform from year to year.
That should not stop us from going back to GVA as our preferred growth indicator for India. I am by no means alone in recommending this. The latest monetary policy statement of 6 April, for example, deals with growth projections for next year in terms of GVA alone. The further advantage of focusing on GVA is that it can be disaggregated into its sectoral components.
GVA growth for 2016-17 was revised downwards from the advance estimate of 7.0% to 6.7%. At the same time, the GVA growth rate for the previous year, 2015-16, was revised sharply upwards from 7.2% to 7.8%. Stopping right there, the figures point to a growth slowdown in 2016-17 of 1.1 percentage points. The GDP at market prices slowed down too, from 7.9% growth in 2015-16 in the revised estimates to 7.1%. So whichever way one views it, growth slowed down in 2016-17 by somewhere of the order of 1 percentage point.
To get a handle on how much of this slowdown might have been a result of demonetisation, we have to disaggregate GVA into those sectors where value addition is driven by factors exogenous to (short-run) domestic policy developments. Here again, others have recommended this, although their preferred lines of partition vary. At least one other commentator, Sajjid Chinoy, excludes the same two sectors as I do—agriculture and public administration—to obtain what might be termed core GVA, analogous to core inflation.
Agriculture is monsoon-driven, and is thus a clear candidate for exclusion. The rationale for exclusion of public administration (lumped together with defence and other services like education and health) is somewhat different. The numbers and salary structures in public employment, although exogenously determined, remain uniform from year to year unlike agriculture, except in years (such as 2016-17) of pay commission salary hikes. These discontinuous jumps raise the sectoral growth rate, when in fact there has just been a (real) salary increase indicative of no underlying productivity growth. So both these sectors are candidates for exclusion from GVA.
When such an exercise is done for 2016-17, we get a (core GVA) growth rate of 6.4%. The core GVA for the previous year, 2015-16, works out to 7.83% (slightly higher than the overall GVA growth rate of 7.8%). Thus there has been a core GVA slowdown of 1.43 percentage point.
How much of this slowdown was due to demonetisation? For an answer to this, or indeed to any question about intra-year growth rates, I always reject quarterly growth rates, and prefer to go with half-yearly figures. Quarterization, once again forced on us by international standards of data dissemination, is not suitable for India, because the boundaries between seasons shift from year to year with respect to quarterly boundaries, and a quarter is too short a period to absorb the impact of such a shift.
The best thing to do is to aggregate the first two and last two quarters into half-yearly estimates for H1 (April-September) and H2 (October-March), respectively. When you do that, you get a core GVA growth rate of 6.9% for H1 2016-17 and 5.9% for H2 2016-17. The average of these two is what gives us that 6.4% growth rate for the whole year in core GVA.
So here is the final summary from the revised estimates as we have them so far. Core GVA, which cleans out exogenously driven sectors like agriculture and public administration, shows a slowdown, from 7.83% (full year 2015-16) to 6.9% (H1 2016-17). There was a further 1 percentage point growth decline to 5.9% growth in H2 of 2016-17. I believe the decline between H1 and H2 is attributable to demonetisation. All these figures could change when, in a little less than two months, the first provisional GDP figures for 2016-17 are issued, with all four quarters fully estimated.
Core GVA as defined here is the aggregate of mining, manufacturing, utilities, construction, domestic trade and transport, and finance, real estate and related services. The decline in core GVA is likely to have been even higher than the 1 percentage point visible so far, because the advance estimates assume informal sector growth to be roughly at par with the formal sector. As is well known, demonetisation hit the informal sector harder than the formal sector, because consumers desperately switched where possible to formal payment channels.
Staying with half-yearly estimates is of course unsatisfactory for those who need more high frequency data. These are the standard domestically generated ones, and the Purchasing Managers’ Index (PMI). The PMI for March 2017 at 52.5 for manufacturing is the highest recorded after October 2016, just before demonetisation. This could reflect a quick spring-back after that disruption, although the March index is generally high because of the purchasing rush at the end of the fiscal year. All of these high-frequency indicators are, however, based on data from the formal sector.
The only hope for a better picture of what happened in the informal sector is in fact the official statistical machinery, which commands the data from which a clearer picture of the informal sector impact could emerge with the final estimates for 2016-17.
A final check on the impact of demonetisation on the informal sector will become possible in June if the half-yearly Financial Stability Report due that month provides a full sector-specific picture of what has happened to defaults on loans to the informal sector (priority sector lending) over the last six months. Of course, these defaults pale in comparison with the mountain of non-performing loans taken by large corporate borrowers. But they did add extra straws on to the back of a severely burdened camel.
All roads on the macroeconomic map of India, no matter where the starting point, lead today to banking sector balance sheets. The problem is finally receiving serious attention. There is in particular an excellent three-path solution offered by Reserve Bank of India (RBI) deputy governor Viral Acharya in a recent speech (reproduced in the March issue of the RBI Bulletin).
Going back to the GDP numbers, no attention should ever be paid to consumption and investment figures reported in advance or first revised estimates. These carry wide error margins even in the final estimates. It is best to stay with the sectoral constituents of value addition at least in the initial estimates, and wait for the final estimates before trying to get a handle on consumption and investment.
Indira Rajaraman is an economist.