by Ajay Shah.
In 2013, the Indian government mounted a big defence of the rupee. The authorities appeared to throw everything that they could at the problem: enhanced capital controls on outflows, relaxed capital controls on inflows, exchange rate intervention, restrictive actions which damaged currency market liquidity to make it easier to manipulate the market by trading on the market, and monetary policy tightening. See this article for a narrative from that period, and a set of links to contemporaneous thinking.
It's important to look back at this period and ask: Did the currency defence deliver on its objectives? Can we identify costs and benefits?
As with all episodes of currency defence, we do not observe the counterfactual: What would have happened if the authorities had not mounted this currency defence? In this case, we have some interesting evidence ahead which suggests that the outcome was not influenced by all these actions.
Along the way, we suffered collateral damage of numerous kinds. Many components of a currency defence are effectively an interest rate defence. At a difficult time in the economy, defending the rupee by raising rates hurt the domestic economy further.
In what follows, on each of the graphs, we superpose the dates of various kinds of actions taken by the authorities. This helps us conduct an intuitive causality test: if an action mattered, it should have yielded some impact in its immediate after math. These graphs also help us see through the claims made by authorities. (Event studies have value in this broad area of research, but are problematic here as there is a lot of event clustering).
The graph above shows the USD/INR exchange rate. The first action was undertaken when the USD/INR exchange rate was within range of Rs.60 to the dollar. After that a series of actions were taken, through June, July, August and September. At the bottom, the exchange rate was near Rs.70 to the dollar. After all the actions stopped, the exchange rate got back to the region of Rs.60 to the dollar.
It is useful to see this same graph where the USD/INR exchange rate is indexed to 100 at the left edge. The currency defence began when there was a 5% depreciation and the bottom was a 25% depreciation. After the flow of measures taken by the authorities ended, the exchange rate recovered to a 10% depreciation compared with the starting point.
When we look at the graph above, we could tell two rival stories. One argument is that the actions had no impact. Another argument is that the cumulative impact of all the actions gave a reversal of the exchange rate.
We are able to resolve this debate using international evidence. Let's start at the J P Morgan Emerging Markets Currency Index:
The graph above shows the J P Morgan Emerging Markets Currency Index on the left axis and the USD/INR on the right axis. The two series are remarkably alike!
All the actions taken by the Indian authorities can only have a small impact on the Emerging Markets Currency Index, reflecting the weight of the INR in that index. That small weight simply cannot account for the extent to which the two graphs move together.
The two graphs are so alike! If the actions by the Indian authorities had scored some impact, then the Indian story would have unfolded differently from the overall average of all Emerging Markets. It did not. Therefore all the actions taken by the Indian authorities had no impact on the USD/INR.
In a similar vein, we look at the Citibank index of policy surprise in the US:
The important dates in this series are the dates of policy surprise in the US. These policy surprises -- Bernanke's first speech and then his second speech -- are more likely to have shaped the USD/INR when compared with the actions of the Indian authorities.
The currency defence was an interest rate defence, as all currency defences are. The short rate surged sharply.
RBI has numerous instruments through which monetary policy is implemented. All these reduce to one summary statistic in the form of the 91-day treasury bill rate. The graph above shows a huge increase in the rate, from roughly 7.5% up to 12%: an increase of 440 basis points.
These days, it is fashionable for the authorities to claim that they do not actually have a target exchange rate in mind, but they are only intervening to prevent episodes of high volatility. The graph above shows that squared daily returns on the USD/INR were slumbering when the rupee defence began. Volatility seems to have surged after the government got going. Even after October, volatility had not come back to the levels found in May. (There are two things going on here: the impact of global developments and the impact of the actions taken by the authorities).
Squared daily returns is a poor statistical estimate of volatility. Using IGIDR FRG data, we exploit high frequency data to construct the daily time series of realised volatility of the USD/INR currency futures traded at NSE. This also shows a similar picture: volatility was slumbering when the currency defence began; it's hard to claim that the authorities were focused on `excessive volatility' and not concerned about the level of the rate. Volatility worsened through the period of the currency defence. (There are two things going on here: the impact of global developments and the impact of the actions taken by the authorities).
We use data from IGIDR FRG on USD/INR options trading at NSE to construct the implied volatility every day. This gives a forward looking indicator of future USD/INR volatility as seeen by the market. It shows that volatility was slumbering when the first actions began. Volatility generally worsened after the authorities acted. (There are two things going on here: the impact of global developments and the impact of the actions taken by the authorities).
Many people at the time noticed the impact on stock prices. Nifty dropped by 10% and Nifty Junior dropped by 15% before recovering late in the year. But many other elements of the collateral damage have not been widely noticed. Let's start at stock market liquidity.
The graph above shows the round-trip transactions costs (in basis points) faced when doing basket trades for the full Nifty portfolio on the NSE spot market. These are computed off the full limit order book that's observed at IGIDR FRG. The lowest curve is at a basket size Rs.2.5 million; the next one is at Rs.5 million and the highest one is at Rs.10 million. These `round trip transactions costs' are the sum of impact cost to buy and impact cost to sell.
Stock market liquidity held up reasonably well in the early part of the story. At event 6 (increase in interest rates), stock market liquidity worsened, particularly at larger transaction sizes. As an example, at a transaction size of Rs.10 million, the round-trip impact cost rose from the region of 8 basis points to the region of 12 basis points, a 50 per cent increase. This would, in turn, spill over into an array of downstream consequences such as enlarged no-arbitrage bands, increased costs of dynamic trading strategies, etc.
Market impact cost reflects a combination of implied volatility, funding constraints and asymmetric information. The outcome seen above reflects a deterioration on all three counts.
At first, the currency defence did not seem to make a big difference to the forward looking volatility of Nifty. But from August onwards, the outlook for Nifty became much more risky. (There are two things going on here: the impact of global developments and the impact of the actions taken by the authorities).
Another element of the collateral damage was banks. The increase in the short rates hurt banks who, on average, seem to be carrying an interest rate mismatch. There was a roughly 30% decline in the CMIE banking industry stock price index.
In the 1980s, an extensive literature worked on capital controls and found generally bad outcomes. In recent years, a new literature is utilising the improvements of econometrics of recent years and reopening the same questions. Forbes & Klein 2015 say:
We have a few papers in this rediscovery of the orthodoxy: Patnaik and Shah 2012, Patnaik et. al. 2012, and Pandey et. al. 2015.
A political economy literature in the US suggests that the last one year prior to an election matters disproportionately (link, link, link, link, link, link). The currency defence of 2013 may have been unusually important in securing a clear majority for the BJP in May 2014. But this same currency defence also damaged the economy, and has helped worsen things in the first year of the BJP.
In 2013, the Indian government mounted a big defence of the rupee. The authorities appeared to throw everything that they could at the problem: enhanced capital controls on outflows, relaxed capital controls on inflows, exchange rate intervention, restrictive actions which damaged currency market liquidity to make it easier to manipulate the market by trading on the market, and monetary policy tightening. See this article for a narrative from that period, and a set of links to contemporaneous thinking.
It's important to look back at this period and ask: Did the currency defence deliver on its objectives? Can we identify costs and benefits?
As with all episodes of currency defence, we do not observe the counterfactual: What would have happened if the authorities had not mounted this currency defence? In this case, we have some interesting evidence ahead which suggests that the outcome was not influenced by all these actions.
Along the way, we suffered collateral damage of numerous kinds. Many components of a currency defence are effectively an interest rate defence. At a difficult time in the economy, defending the rupee by raising rates hurt the domestic economy further.
In what follows, on each of the graphs, we superpose the dates of various kinds of actions taken by the authorities. This helps us conduct an intuitive causality test: if an action mattered, it should have yielded some impact in its immediate after math. These graphs also help us see through the claims made by authorities. (Event studies have value in this broad area of research, but are problematic here as there is a lot of event clustering).
Level of the exchange rate
 |
| USD/INR exchange rate |
The graph above shows the USD/INR exchange rate. The first action was undertaken when the USD/INR exchange rate was within range of Rs.60 to the dollar. After that a series of actions were taken, through June, July, August and September. At the bottom, the exchange rate was near Rs.70 to the dollar. After all the actions stopped, the exchange rate got back to the region of Rs.60 to the dollar.
 |
| USD/INR exchange rate, indexed to 100 at start |
It is useful to see this same graph where the USD/INR exchange rate is indexed to 100 at the left edge. The currency defence began when there was a 5% depreciation and the bottom was a 25% depreciation. After the flow of measures taken by the authorities ended, the exchange rate recovered to a 10% depreciation compared with the starting point.
Did the currency defence work?
When we look at the graph above, we could tell two rival stories. One argument is that the actions had no impact. Another argument is that the cumulative impact of all the actions gave a reversal of the exchange rate.
We are able to resolve this debate using international evidence. Let's start at the J P Morgan Emerging Markets Currency Index:
 |
| USD/INR exchange rate superposed with the J P Morgan EM Currency Index |
The graph above shows the J P Morgan Emerging Markets Currency Index on the left axis and the USD/INR on the right axis. The two series are remarkably alike!
All the actions taken by the Indian authorities can only have a small impact on the Emerging Markets Currency Index, reflecting the weight of the INR in that index. That small weight simply cannot account for the extent to which the two graphs move together.
The two graphs are so alike! If the actions by the Indian authorities had scored some impact, then the Indian story would have unfolded differently from the overall average of all Emerging Markets. It did not. Therefore all the actions taken by the Indian authorities had no impact on the USD/INR.
In a similar vein, we look at the Citibank index of policy surprise in the US:
 |
| USD/INR versus the Citi US surprise index |
The important dates in this series are the dates of policy surprise in the US. These policy surprises -- Bernanke's first speech and then his second speech -- are more likely to have shaped the USD/INR when compared with the actions of the Indian authorities.
All currency policy is monetary policy
 |
| The 91-day treasury bill rate |
The currency defence was an interest rate defence, as all currency defences are. The short rate surged sharply.
RBI has numerous instruments through which monetary policy is implemented. All these reduce to one summary statistic in the form of the 91-day treasury bill rate. The graph above shows a huge increase in the rate, from roughly 7.5% up to 12%: an increase of 440 basis points.
Volatility
 |
| Squared returns on USD/INR exchange rate |
These days, it is fashionable for the authorities to claim that they do not actually have a target exchange rate in mind, but they are only intervening to prevent episodes of high volatility. The graph above shows that squared daily returns on the USD/INR were slumbering when the rupee defence began. Volatility seems to have surged after the government got going. Even after October, volatility had not come back to the levels found in May. (There are two things going on here: the impact of global developments and the impact of the actions taken by the authorities).
 |
| Realised volatility of USD/INR futures |
Squared daily returns is a poor statistical estimate of volatility. Using IGIDR FRG data, we exploit high frequency data to construct the daily time series of realised volatility of the USD/INR currency futures traded at NSE. This also shows a similar picture: volatility was slumbering when the currency defence began; it's hard to claim that the authorities were focused on `excessive volatility' and not concerned about the level of the rate. Volatility worsened through the period of the currency defence. (There are two things going on here: the impact of global developments and the impact of the actions taken by the authorities).
 |
| Implied volatility of USD/INR |
We use data from IGIDR FRG on USD/INR options trading at NSE to construct the implied volatility every day. This gives a forward looking indicator of future USD/INR volatility as seeen by the market. It shows that volatility was slumbering when the first actions began. Volatility generally worsened after the authorities acted. (There are two things going on here: the impact of global developments and the impact of the actions taken by the authorities).
Collateral damage
Many people at the time noticed the impact on stock prices. Nifty dropped by 10% and Nifty Junior dropped by 15% before recovering late in the year. But many other elements of the collateral damage have not been widely noticed. Let's start at stock market liquidity.
 |
| Round-trip transactions costs for Nifty basket trades on the NSE spot market |
The graph above shows the round-trip transactions costs (in basis points) faced when doing basket trades for the full Nifty portfolio on the NSE spot market. These are computed off the full limit order book that's observed at IGIDR FRG. The lowest curve is at a basket size Rs.2.5 million; the next one is at Rs.5 million and the highest one is at Rs.10 million. These `round trip transactions costs' are the sum of impact cost to buy and impact cost to sell.
Stock market liquidity held up reasonably well in the early part of the story. At event 6 (increase in interest rates), stock market liquidity worsened, particularly at larger transaction sizes. As an example, at a transaction size of Rs.10 million, the round-trip impact cost rose from the region of 8 basis points to the region of 12 basis points, a 50 per cent increase. This would, in turn, spill over into an array of downstream consequences such as enlarged no-arbitrage bands, increased costs of dynamic trading strategies, etc.
Market impact cost reflects a combination of implied volatility, funding constraints and asymmetric information. The outcome seen above reflects a deterioration on all three counts.
 |
| Implied volatility of Nifty |
At first, the currency defence did not seem to make a big difference to the forward looking volatility of Nifty. But from August onwards, the outlook for Nifty became much more risky. (There are two things going on here: the impact of global developments and the impact of the actions taken by the authorities).
 |
| CMIE Bank Index |
Another element of the collateral damage was banks. The increase in the short rates hurt banks who, on average, seem to be carrying an interest rate mismatch. There was a roughly 30% decline in the CMIE banking industry stock price index.
What can we learn from this episode?
- USD/INR seems to have responded to global events and all the actions of the authorities seem to have had little impact.
- The entire focus of economic policy in that period was on fighting the INR depreciation. Every day, new tools were being bandied about and implemented. There was a reverential approach to the power of central banks -- currency intervention, capital controls, choking off financial markets, arranging lines of credit, etc. But when we look back, we see that the central bank was ineffectual in delivering on the goal.
- The cost of all the actions taken by the government was large, and the payoff obtained from the actions was elusive. Ex post, we see that the cost of mounting the defence was much larger than the costs envisaged at the outset by proponents of the currency defence.
- The Ministry of Finance debated the views of many advisors, and chose to go with the folks who advocated a big muscular currency defence.
- I feel that the academic literature on capital controls and currency policy does not adequately come to grips with the mess that we get in the real world when we try to do these things. It's easy to bandy about capital controls or currency trading. Such `heterodox thinking' has become fashionable. Capital controls are not irrelevant; they can deliver pricing distortions and reduce market efficiency. What's important to ask is: Do capital controls deliver on the goals of macroeconomic policy? The answer in India seems to be: No.
- Some advisors at the time said that it was possible to `squeeze the shorts' and `hit the speculators' without contaminating monetary policy. But all currency policy is monetary policy, and all real monetary economists know this. Every currency defence becames an interest rate defence. The policy rate went up by 440 basis points, dealing a body blow to a weak economy. By August 2013, I felt the UPA was going to lose the next election.
- When the authorities defend the rupee, this protects foreign currency borrowers. In addition, raising rates hurts local currency borrowers. This is different from the conventional moral hazard argument, that currency policy creates moral hazard in favour of more unhedged foreign currency borrowing. The new thing I understood in 2013 was that a currency defence is not only good for unhedged foreign currency borrowers; it hurts local currency borrowers. These two factors come together to shape the incentives of firms on choosing between borrowing in local currency vs. borrowing abroad. If you borrow in USD, you will get support from the government in extreme events; if you borrow in INR, you will get hit by the government in extreme events.
- Episodes like this interact with the bureaucratic politics of Indian finance. As an example, currency futures trading started in 2008, against the protests of RBI which feared the loss of turf (the role of SEBI on exchanges). RBI was also concerned about the possibility that exchanges might achieve liquidity and market efficiency, and thus undo decades of hard work in preventing a genuine currency market. The currency defence of 2013 was used as an opportunity to hamper the working of currency derivatives trading on exchanges. Even today -- June 2015 -- some of those restrictions imposed in the summer of 2013 have not been reversed.
- Market liquidity got worse once the currency defence got going. This may have implications for the decision to stay exposed through a currency defence.
- It takes a long time for the damage caused by a currency defence to heal. Some economists think you can switch off and switch on finance at a whim. But the working of the financial system is ultimately about the organisational capital in financial firms. When private firms want to build a business, organisational capital can only be built slowly, and after such episodes, private firms may not feel like committing resources to build a business. Private firms lose respect and trust in the authorities when they see such episodes, and hold back from investing in building the business.
In the 1980s, an extensive literature worked on capital controls and found generally bad outcomes. In recent years, a new literature is utilising the improvements of econometrics of recent years and reopening the same questions. Forbes & Klein 2015 say:
large interest rate increases, major reserve sales, large currency depreciations, and new controls on capital outflows are ineffective at improving these three outcome variables and often imply substantial tradeoffs. More specifically, sharp increases in interest rates and new controls on capital outflows appear to significantly reduce GDP growth, have no consistent effect on unemployment, and are correlated with higher inflation. Large depreciations also appear to initially reduce GDP growth and be correlated with higher inflation, but there is some evidence that they can yield a lagged improvement in growth and reduce unemployment, especially during the 2000s.
We have a few papers in this rediscovery of the orthodoxy: Patnaik and Shah 2012, Patnaik et. al. 2012, and Pandey et. al. 2015.
A political economy literature in the US suggests that the last one year prior to an election matters disproportionately (link, link, link, link, link, link). The currency defence of 2013 may have been unusually important in securing a clear majority for the BJP in May 2014. But this same currency defence also damaged the economy, and has helped worsen things in the first year of the BJP.
ReplyDeleteThis is a very tough Paper from the pen of a very erudite Scholar so much so 'lesser mortals' like me would take several weeks to comprehend the substance of the variety of arguments let alone appreciate the applicability or the limitations of the same.
As such, I wish to make only a general observation:
I was lucky to be a M.B.A. student in the Katholic University of Leuven in 1979 and was in Europe from 1975 to 1985 so could observe the ramifications of the Floating Exchange Rate Regime ushered in from around 1973. The “Snake in the Tunnel” was adopted by Six European Countries to arrest wild gyrations of their Currencies (in my opinion with reasonable success) However, there was a run some times on the British Pound as also the US Dollar (both measured versus the then Deutsch Mark DM). I well remember the Bank of England making efforts to defend the Pound also aided by The Bundes Bank and the Federal Reserve. The well known magazine “ECONOMIST” was a staunch critic of such multilateral defence measures. However, it is fair to say the arguments of “ECONOMIST” were qualitative unlike the Quantitative Research results of the learned Author.
My dumb enquiry is: How can we say how well some measures addressed the various parameters unless we compare “measures scenario” with “absolutely no measures scenario” (do nothing strategy!)