integration between onshore and offshore markets they show that the basis trade within the no-arbitrage band only 56 percent of the time(Table 1). This fact, together with the large, absolute positive and negative values for the basis(peaking at 1, 795 and-1, 235 pips espectively), suggest users of offshore renminbi face relatively high basis risk, which serves to discourage them from using the CNh as a substitute for CNY. This awareness of basis risk may have increased with the sharp rise in the volatility in the basis in the second half of 2011, which may help explain the temporary stalling of CNH deposit growth as users became more cautious in holding CNH in place of CNY(Figure 4) C. Arbitrage between Onshore and offshore markets and the role of capital controls The estimation results confirm that arbitrage works to steadily narrow the basis when shocks air processes followed by the basis when outside the band. The coefficient on move it outside the no-arbitrage band. This is reflected in the highly significant coefficient estimates are fairly close to one, indicating that th e convergence process is relatively slow and may reflect limits on the scale of arbitraging capital flows owing to capital controls The estimation results reveal an asymmetry in the speed with which capital inflow and outflow work to narrow divergences in the offshore and onshore exchange rates. Arbitrage is much slower when the CNh is stronger than the Cny than when it is weaker, specifically When CNH trades at a premium to CNY, arbitrage takes an average of 25 days to close half the gap back to the band (the"half life")(Table 1). Capital outflows from the mainland are needed for this arbitrage, and work to increases the supply of offshore renminbi liquidity. This was the case in the November 2010-May 2011 episode When CNh trades at a discount to CNY, arbitrage takes an average of 6 days to close half the gap back to the band. This involves capital inflows to the mainland, reducing the supply of offshore renminbi liquidity. This was the case in September 201 1-October 2012 The faster rate of convergence in the latter case-when CNh trades at a discount to CNY- implies that capital controls are less restrictive with respect to arbitraging capital inflows to the mainland than outflows from the mainland. This difference may reflect the fact that recent liberalization measures have focused more on easing constraints on inflows than outflows(see section Iv below), such as the opening of channels for renminbi denominated FDI and QFll that can be used to bring offshore renminbi funds onshore D. Sources of divergence of onshore and offshore renminbi exchange rates The movement in the basis outside the no-arbitrage band will be driven by a number of factors that, along with capital controls, contribute to limited integration of onshore and offshore markets. The literature points to a range of potential factors, including market expectations for the renminbi, market risk sentiment, capital controls and offshore renminbi liquidity. For example, arbitraging capital flows would lead to changes in offshore renminbi liquidity that would be negatively correlated with the basis. This appears to be the case, with7 integration between onshore and offshore markets. They show that the basis trade within the no-arbitrage band only 56 percent of the time (Table 1). This fact, together with the large, absolute positive and negative values for the basis (peaking at 1,795 and -1,235 pips, respectively), suggest users of offshore renminbi face relatively high basis risk, which serves to discourage them from using the CNH as a substitute for CNY. This awareness of basis risk may have increased with the sharp rise in the volatility in the basis in the second half of 2011, which may help explain the temporary stalling of CNH deposit growth as users became more cautious in holding CNH in place of CNY (Figure 4). C. Arbitrage between Onshore and Offshore Markets and the Role of Capital Controls The estimation results confirm that arbitrage works to steadily narrow the basis when shocks move it outside the no-arbitrage band. This is reflected in the highly significant coefficient on the autoregressive processes followed by the basis when outside the band. The coefficient estimates are fairly close to one, indicating that the convergence process is relatively slow and may reflect limits on the scale of arbitraging capital flows owing to capital controls. The estimation results reveal an asymmetry in the speed with which capital inflow and outflow work to narrow divergences in the offshore and onshore exchange rates. Arbitrage is much slower when the CNH is stronger than the CNY than when it is weaker, specifically:  When CNH trades at a premium to CNY, arbitrage takes an average of 25 days to close half the gap back to the band (the “half life”) (Table 1). Capital outflows from the mainland are needed for this arbitrage, and work to increases the supply of offshore renminbi liquidity. This was the case in the November 2010-May 2011 episode.  When CNH trades at a discount to CNY, arbitrage takes an average of 6 days to close half the gap back to the band. This involves capital inflows to the mainland, reducing the supply of offshore renminbi liquidity. This was the case in September 2011-October 2012 period. The faster rate of convergence in the latter case—when CNH trades at a discount to CNY— implies that capital controls are less restrictive with respect to arbitraging capital inflows to the mainland than outflows from the mainland. This difference may reflect the fact that recent liberalization measures have focused more on easing constraints on inflows than outflows (see section IV below), such as the opening of channels for renminbi denominated FDI and QFII that can be used to bring offshore renminbi funds onshore. D. Sources of Divergence of Onshore and Offshore Renminbi Exchange Rates The movement in the basis outside the no-arbitrage band will be driven by a number of factors that, along with capital controls, contribute to limited integration of onshore and offshore markets. The literature points to a range of potential factors, including market expectations for the renminbi, market risk sentiment, capital controls and offshore renminbi liquidity. For example, arbitraging capital flows would lead to changes in offshore renminbi liquidity that would be negatively correlated with the basis. This appears to be the case, with