Liquidity on the currency futures vs. the Nifty futures

India is the second country of the world, after Brazil, where the currency futures are more liquid than the currency forwards. Today when I glanced at the order book of the near month rupee-dollar futures, I was struck by the big numbers that are visible:

The tick size of this market is 0.25 paisa, so the top five prices cover 1.25 paisa on each side. So there's nothing interesting about the prices: I focus on the quantities. I'm used to generally seeing quantities at each prices running all the way to 1000 to 2000 contracts which is $1m to $2m. Today I was surprised to see two quantities with much bigger values: $6m and $11.3m. This is huge. [NSE currency futures page] I was also impressed at the fact that $32.6m and $42.4m of orders are sitting on this order book. These are big numbers.



For a comparison, I popped over to look at the near month Nifty futures, which is the biggest financial product in India. This order book shows:

In your mind you have to divide the quantities by 10 to make them comparable. Hence, the spreads and the quantities in the top five are not as impressive as those of the currency. But the total orders present on the book here are staggeringly large when compared with the currency.



Theory tells us that liquidity should vary with asymmetric information and volatility. Both Nifty and the currency are macroeconomic underlyings with relatively little asymmetric information. But Nifty is more volatile. So if both markets worked well, we would expect Nifty to be less liquid. Some key differences are obviously visible: foreigners trade on the Nifty futures but are banned from the currency futures, and Nifty options are available while currency options are not yet available (though without foreigners). When these differences go away, the picture will change further.



A paper idea: When a central bank shifts to a more transparent framework on currency trading, or when a central bank steps away from currency trading altogether, asymmetric information on the currency market goes down so currency impact cost should go down. I wonder if one can find some natural experiment of this fashion. Matters are complicated because the date on which a float commences if often not the date on which the float is announced. This can be dealt with by identifying the date on which the exchange rate regime actually changed, as opposed to what is claimed by the central bank.