India's foreign exchange reserves touched a record $785.7 billion in the week ended September 4. The main reason is the Reserve Bank's special swap window for foreign-currency deposits, announced on June 8 and closed on August 31. It drew $127.2 billion in FCNR(B) deposits and $136.4 billion in all, against market expectations of $90 billion to $100 billion. The rupee, however, has barely moved. It held at about 95.7 to the dollar from the day the window opened until August 21, and closed at 95.89 on September 18, close to its weakest levels.

The decision to act was justified. With oil prices high, portfolio money leaving and the currency at record lows, inaction carried real costs, and the RBI moved quickly. The question is how the outcome should be assessed. The window has built a large reserve buffer and stopped a slide. It has not restored the currency, and it has left obligations that will run for years. A rescue should be judged on both counts.

Start with the scale. In early August the Governor said that no proposal to close the scheme early was under consideration. Nine days later, the deadline was advanced by a month, and roughly $62 billion came in during the last 10 days of August. The RBI was absorbing the hedging cost that banks would normally bear, which made dollar deposits unusually attractive to depositors and banks alike. For comparison, the two windows of 2013, open from September to November, raised $34.3 billion in all: $26 billion through FCNR(B) deposits and $8.3 billion through bank overseas borrowing. The 2026 windows have raised about four times as much, and the scheme brought in more than it was designed for, which suggests it had no mechanism to keep inflows in proportion to the need.

The domestic effects are already visible. Surplus rupee liquidity in the banking system is reported at ₹10 trillion to ₹11 trillion, a record, and the RBI is absorbing it through sell-buy swaps and reverse repos. One brokerage expects lower wholesale funding costs for banks, though it sees near-term margins staying under pressure. More liquidity than the economy can readily use also lowers lending rates and can push credit growth ahead of underlying demand, at a time when import prices are rising.

The headline reserve figure needs careful reading as well. Market participants say the dollars raised through the window went into reserves and were not released into the spot market, which is one reason the rupee did not respond. Meanwhile, the RBI's net short dollar forward position reached a record $136.8 billion at the end of July, up from $103.3 billion a month earlier, and about $91.5 billion of it runs beyond one year. Netted against these forward commitments, usable reserves are smaller than the gross number suggests. Economists also point out that because the borrowed dollars must be repaid over three to five years, the RBI has to be more cautious in spot intervention.

Then there is the repayment profile. The largest share of the deposits carries five-year maturities, and analysts place repayments between 2027 and 2031. The RBI has taken on the exchange-rate risk on the principal through the swap. Banks carry the interest cost and will need to refinance or buy dollars when the deposits fall due, in conditions nobody can predict today.

None of this means the response failed. It means the outcome is not yet complete.

Concessional swap windows were used in 2013 as well, and the rupee appreciated by about 8.8% by March 2014. Conditions were different then, and the window was one of several measures, sequenced with a credible fiscal signal and an effort to bring the external account under control. Markets could see the direction of adjustment, and the window was sized to the need.

In the present episode, each instrument is defensible on its own, but they do not point in one direction. I have argued elsewhere that policy today is multi-directional rather than cumulative. The window has built reserves without lifting the rupee, and it has left a liquidity surplus and a forward book to be managed, without any stated order of priority among these objectives.

There is a tendency in a crisis to treat a larger inflow as a better result. The more useful test is the net effect of an instrument over its whole life. Measures taken to secure stability should not become a source of instability later.

A set of standing rules for emergency windows would address this, and four elements appear necessary. The first is a cap set according to the assessed external financing gap, with an automatic pause and public explanation when inflows approach it. The second is a published pricing rule stating who bears the cost and for how long; when the central bank absorbs hedging costs, it is committing public resources, and an estimate should be available before the first dollar comes in. The third is a repayment ladder announced at launch, showing maturities by year, bank-level exposures and the intended approach to refinancing, together with the effect on the forward book. The fourth is a fixed closing date, followed by a published review of the window's cost, its effect and its residual obligations.

It may be argued that rules reduce the flexibility that crisis management requires. The rules proposed here would still allow discretion, provided that departures are explained. Their main effect would be on predictability, which markets value, and on speed, since officials would know in advance how to proceed.

The pressures that arrived this year, from oil and from capital moving towards economies at the AI and semiconductor frontier, are likely to return in some form in not too distant  future. When they do, the RBI will be operating with the repayment schedule of this window in the background. Setting the rules now, and publishing the repayment plan, would show that the costs of the rescue have been counted.

Transparency and predictability are, after all, the hallmarks of mature public policy.