Dormant for a while, the debate over India's demonetization program of last fall has been revived by new evidence. The new evidence on note returns and GDP vindicates the critics, and has the defenders in strategic retreat.
To recap: On November 8, 2016, India's Prime Minister Narendra Modi shocked the nation by announcing the immediate "demonetization" of the two largest rupee currency notes (Rs 500, worth about $7.50, and Rs 1000, worth about $15). Noteholders would have only 50 days to turn them in for new Rs 500 and Rs 2000 notes. The move, Modi promised, would sharply penalize holders of unaccounted "black money," namely tax evaders, bribe-takers, professional criminals, and terrorists. Their currency hoards would become worthless — a welcome one-time wealth loss — or they would expose themselves to detection by trying to swap or deposit large batches. Anyone depositing a large sum in old notes would face scrutiny by tax authorities.
In order to keep the move a surprise (the better to catch the black money holders), new notes to replace all the discontinued notes had not been printed in advance. The cancelled notes represented 86% of the currency in circulation, and more than half of M1 (currency plus checking deposits), India having a highly cash-intensive economy with half the population unbanked. As criminals were far from the main users of currency, the impact was unavoidably felt well beyond the black-money set. A serious currency shortage immediately arose, with predictable consequences. Honest wage laborers in the huge cash economy went unpaid, honest construction projects came to a standstill, honest shopkeepers saw sales dry up, and honest businesses failed. Honest people wasted billions of hours waiting in queues to exchange old notes for the trickle of new notes.
As Shruti Rajagopalan and I noted in November last year, there was also a fiscal angle: for every billion of old rupee notes not turned in (for fear of being scrutinized), the government could issue a replacement billion and pocket it as one-time seigniorage revenue. For example:
if 20% of the old notes are never turned in, the government's revenue windfall is up to Rs 2.9 trillion ($42.5 billion).
The destruction of the private wealth of non-redeeming old-note holders, combined with the revenue windfall to the government, makes the currency policy effectively a large capital levy, a massive one-shot transfer of wealth from the private to the public sector.
We speculated: "The wealth transfer to government may help to explain Prime Minister Modi's enthusiasm for the currency cancellation and re-issue, despite its likely ineffectuality against black money."
Economically literate defenders of demonetization have been fewer than critics. The most prominent defenders have been the well-known trade economist Jagdish Bhagwati of Columbia University together with his former students Vivek Dehejia and Pravin Krishna, and with his Columbia colleague Suresh Sundaresan.
In a December 2016 piece in the prominent Times of India, Bhagwati, Krishna, and Sundaresan (hereafter BKS) praised the demonetization program as "a courageous and substantive economic reform that, despite the significant transition costs, has the potential to generate large future benefits." BKS recognized that "the process is inconvenient, and subjects many households to hardships," but thought it worthwhile for "potentially increasing transparency and expanding the tax base and revenues to the government from taxes and surcharges." The fiscal angle was foremost: since "unaccounted deposits will be taxed, this will yield a windfall for the government permitting large increases in social expenditures." In addition, it would promote a "switch into digital transactions" and "put a major dent in counterfeiting."
In an Op-Ed published on December 27, Bhagwati, Dehejia, and Krishna (hereafter BDK) defended the demonetization program entirely on the grounds that it would impose an effective capital levy. It was, they wrote, "a policy designed, in effect, as a one-time tax on black money." They noted that the government's revenue gain would not come just from replacing unreturned notes. Under "voluntary disclosure" rules promulgated after the initial announcement, depositors of old notes who acknowledged their holdings as illegitimate would also pay: "deposits of unaccounted money will be taxed at 50% — with a further 25% taken by the government … as an interest-free loan for a period of four years." Thus there would be a one-time fiscal gain to the government not only from notes never returned, but also from notes returned under such terms.
BDK proposed the size of the revenue gain as a sufficient criterion for judging the success of the program: "at least from the perspective of its effectiveness in dealing with the black money issue, success has to be measured by the sum of tax revenue generated [from the 50% tax on acknowledged black deposits] and black money destroyed [i.e. revenue from replacing unreturned notes]." For the sake of illustration, they supposed (calling it an "estimate") "that one-third [Rs5 trillion] of the approximately Rs15 trillion in demonetised notes is black money." Then if by the end of the turn-in period "Rs1 trillion is unreturned, as is believed, and we further assume that only half of the remaining Rs4 trillion of black money that is returned falls within the tax net, the net gain works out to Rs1 trillion of black money destroyed and 50% times 2 trillion = Rs1 trillion in tax revenue." With such a total fiscal gain of Rs2 trillion, "the government could reasonably claim this as a successful outcome."
In a commentary on the BDK piece, Rajagopalan and I pointed out that, from the economist's point of view, the costs of any measure must be taken into account before judging it worthwhile or efficient. What matters is effectiveness per unit cost. Unlike the earlier BKS piece, BDK had simply neglected the costs incurred in collecting revenue or suppressing black money through demonetization. We noted a think tank's estimate of Rs. 1.28 trillion in losses during the transitional period from expenses of printing new notes, lost income of those waiting in queues, additional costs to banks tied up with exchanging currency, and (the largest item) lost business sales due to the currency shortage. It was then too early to replace the estimate of lost business with measured effects on GDP, but we noted that one percentage point of lost annual growth equals Rs1.45 trillion. These costs need to be set against the revenue. Even if the revenue were as high as Rs2 trillion, collecting it at a deadweight cost of 64% or more would be a very bad bargain.
Doesn't it matter that the transfer in this case is coming from bad actors whose welfare one may disregard? No. In the above reckoning, as in standard tax analysis, the pure wealth-transfer losses of taxpayers don't figure in the deadweight loss calculation, which only counts the costs associated with extracting the transfer.
BDK had noted in passing the argument "that the short- to medium-run economic impact post 8 November will be contractionary" due to a "temporary liquidity shortage induced by an insufficiently fast replacement of old notes with new notes." But they dismissed on theoretical grounds that "this is not necessarily the only outcome possible." Government could avoid a currency shortage by promptly providing new notes, they reckoned. This was a very odd line to take seven weeks into demonetization, given that the government was not in fact providing new notes sufficiently fast, and when the evidence of currency shortage was plain to see. Alternatively, they proposed, hoarded currency could come out from under mattresses, be deposited in banks, and actually expand M1 "via the classical money multiplier." This was an odd line to take given that expansion of deposits (even should it happen) would not remedy the currency shortage being suffered by the unbanked half of India's population.
In March 2017, Bhagwati was quoted by the Indian newspaper Firstpost making the surprising claim in an email interview that demonetization had actually promoted economic growth: "On the effects of demonetisation on growth, I should say that I was the one economist who had argued (with my co-authors), from first principles, that demonetisation would increase, not diminish, growth. And that is exactly what appears to have happened." The factual basis for saying that it appeared to have happened was not clear.
In a March 30 piece, BDK cited a new 2016Q4 GDP report as showing that GDP had suffered "only a modest dip … of roughly half of a percentage point" below pre-demonetization projections. This was not an increase in growth. But they counted it a victory compared to "the economic disaster that the critics had imagined."
The debate over demonetization was revived this month (September 2017) after the Reserve Bank of India finally announced the count of returned currency. It announced that 99 percent of the discontinued notes, Rs 15.28 trillion out of Rs 15.44 trillion, had been returned. As Vivek Kaul has noted, "The conventional explanation for this is that most people who had black money found other people, who did not have black money, to deposit their savings into the banking system for them."
The trivial size of unreturned currency of course obliterates BDK's projection of a government seigniorage windfall.
What about BDK's other projected source of revenue, the 50% tax on acknowledged black deposits? Whereas in BDK's scenario, black currency holders would make Rs2 trillion in voluntary-disclosure deposits, which would yield Rs 1 trillion in revenue, the actual collections under the scheme were reported in April at Rs 23 billion, or 2.3% of the BDK-imagined sum. Such paltry revenues mean that demonetization, from the fiscal perspective, was all pain and no gain.
The accumulating evidence on economic growth, meanwhile, has become damning. Between July and September 2016, India's GDP grew 7.53 percent. Between January and March 2017 it grew 5.72 percent. Former head of the Reserve Bank of India Raghuram Rajan, now returned to the University of Chicago, links the drop to demonetization: "Let us not mince words about it — GDP has suffered. The estimates I have seen range from 1 to 2 percentage points, and that's a lot of money — over Rs2 lakh crore [i.e. trillion] and maybe approaching Rs2.5 lakh crore." Kaul adds that GDP does not well capture the size of the informal cash sector, where the losses from demonetization were greatest.
In response to the RBI report and GDP data, and to their credit, BDK have substantially retreated from claims of success to what can be regarded as the claim that there is still a chance to break even. They have recently written:
First off, it must be conceded that if demonetisation is to be judged narrowly on the basis of the triple rationale originally advanced … , it would at best be unclear if it could be accounted a success. For, little black money was literally "destroyed" and there is scant evidence that the policy had much if any impact on counterfeiting or terror finance.
Although they acknowledge that they "overestimated the quantum of black money that would ultimately be unreturned" and thus overestimated the seigniorage gain, they still contend that the "money deposited into bank accounts can also generate fiscal gain, as these will invite the scrutiny of tax officials." For about two-thirds of the deposits of old currency by value, even though no admission was made and thus no 50% tax was paid, the sums deposited were large and "are mostly open to scrutiny by tax officials." Thus it is conceivable that tax investigators may eventually squeeze taxes and fines out of them, and it is premature to rule this out.
Conceivable, but unlikely. The investigatory capacity of the tax authority is finite and it already has its hands full, as Vivek Kaul spells out in detail.
BDK concede: "Should, however, the government fail in identifying and taxing black money deposits in any significant quantity, we can all conclude that demonetisation will have failed in achieving its primary goal." Welcome as this reasonable concession is, the converse does not follow: whether even significant eventual revenue counts as success depends on how it compares to the sizable costs of demonetization. A deadweight burden of less than 100% seems highly unlikely.
BDK add an odd coda. They acknowledge transition costs in the program actually followed, but suggest that it could have been otherwise:
in principle, had demonetisation occurred without transition costs — for instance, if old notes could have been seamlessly converted or deposited within a few days after 8 November, or if demonetisation had been pre-announced to occur with a lag, allowing time for an orderly remonetisation — there could only have been largely upside gain without any downside cost.
It is hard to square this with BDK's earlier statements that demonetization without secrecy would have been pointless because it would not have caught out the black money holders. Are BDK saying that if the aim had been merely to introduce new notes with better anti-counterfeiting features, the Modi government's demonetization program was an unnecessarily costly way of doing it? Well yes, the critics have said that all along. High transition costs were a feature and not a bug of the dramatic scheme to penalize black money by surprise.