RMB update Sep-23-2005: non-USD gap(band) doubled

Consider 10 horses pulling a cart on a straight road, where 9 small horses pull with elastic bands of different length and the 10th horse (which is much bigger and stronger) pulls with a relatively stiff rope. How would this cart move if the 10 horses are moving to the same direction? What if the 10 horses are moving in a slightly different directions with 'random walk' in small step (i.e. statistical random motions along the road). How would the trace of the cart's path look like? Is the path "smoother" if the 10 ropes are (1) all elastic? (2) all stiff? (3) if only the 10th horse with a stiff rope is used and the other 9 released? (4) if the cart is replaced with a horse, which has its own will of movement, and if this cart-horse becomes stronger? (5) what if all the 10 (or 11) horses are tied with ropes and bands of different elasticity?

Now consider making the 9 elastic bands twice as tactile.

The 9 horses are EUR, JPY, KRW, etc. The 10th horse is USD. The cart is RMB (which can also be replaced with an 11th horse). The lengths of the rope and the elasstic bands are the exchange rates. The road is a proxy for the basket peg. Current length of the 10th rope (which is slightly elastic) is 8.09m long. PBoC has just made the elasticity of the other 9 bands/ropes' twice as large.

According to PBOC, more flexibility (see also English media reports) has been introduced into the basket peg regime. As expected the daily trading gap for non-USD was significantly widened, (from previously unannounced 3%) to 6%. It was also said that the management (of USD) will mainly be done through "managing the spread", althoug the gap is (reportedly) kept unchanged. The announcement was made internally on Sep 14, but only posted on the PBC website today

• For USD cash, the buy and sell spread is capped at 1% (defined as from 1% of the central parity), for cash transaction (at tell window) it is capped at 4%
• 外汇指定银行对客户挂牌的美元对人民币现汇卖出价与买入价之差不得超过中国人民银行公布的美元交易中间价(上一日银行间市场美元收盘价，下同)的1％(［现汇卖出价-现汇买入价］/美元交易中间价×100％≤1％)，现钞卖出价与买入价之差不得超过美元交易中间价的4％(［现钞卖出价-现钞买入价］/美元交易中间价×100％≤4％)。在上述规定的价差幅度范围内，外汇指定银行可自行调整当日美元现汇和现钞买卖价。
• PBoC did not talk about the USD gap, presumably it is kept at the old 0.3%. (BBC/FT reported with certainty, but it is not impossible that PBoC might test the widening secretly or announce another widening soon);
• one possible scenario is asymmetric drifting: say, today central parity is 8.10, with buy/sell=7.95/8.25; tomorrow the buy/sell will be 8.10/8.32; that is a change certainly larger than 0.3%! (unless, did I make a mistake?). This is an example for USD cash. For non-cash the effective drift can still be as large as 0.5%.
• The non-USD gap is enlarged from +/-1.5% to +/-3%, effectively doubled from Jul/21 gap (note the +/-1.5% has not been disclosed before, but was observed if one has been monitoring the data)
• 每日银行间即期外汇市场非美元货币对人民币的交易价在中国人民银行公布的该货币当日交易中间价上下3％的幅度内浮动
• For non-USD, there is no cap on the buy-sell spread
• 取消非美元货币对人民币现汇和现钞挂牌买卖价差幅度的限制，外汇指定银行可自行决定对客户挂牌的非美元货币对人民币现汇和现钞买卖价。

I have long suspected the gaps for non-USD have been in generally much wider than that for USD before, i put my estimate at (at least) 0.5%-1%, now it is confirmed to have been 1.5%, and will be doubled. I believe the USD gap would be further expanded soon. The practice has probably been tested all along. If that is the case, part of the puzzle that USD has been largely "pegged" can be explained. There are many parameters in the basket peg. Each currency is one of the parameters (too many parameter is not necessarily a good thing, as it encourages human error, unless there is a strict internal set of rule in PBoC). By effectively 'floating' non-USD currencies, USD/RMB rate can certainly be kept relatively stable. (while the basket weighted average unaffected)

In addition, one of the mechanisms PBoC employed is to increase the transaction cost (typical transaction cost/spread in forex market is 0.2%, for RMB it is higher, close to 1%) making speculation more costly. It will discourage frequent trading and hence reduce short term volatility, the downside is it takes longer to reach the equilibirum point (less efficient market). The high transaction cost would help to reduce the need for cross-rate intervention as suggested by Halpenny in the report by FT, because EUR-USD arbitrage through RMB as a medium can be made prohibitively expensive. This would be a short term fix. For longer term discrepancy in cross-rates, the crawl will slowly narrow the arbitrage opportunity.

The more likely result is, businesses will be encouraged to convert RMB directly into non-USD currencies, and more contracts will be denominated in non-USD currencies, which are what PBoC wanted.

There are merits of this regime, as it becomes extremely hard for speculators to second guess the basket and trends, while maintaining the desirable degree of overall flexibility, even on USD. It also complicates the jobs of speculators, as it will be a multi-currency attack rather than a simple work of choosing one currency (e.g. USD/RMB).

The problem of this, however, is that the inertia to shift the USD/RMB rate is so big that it is not too different from the old USD-peg regime (high volatility, i.e. daily change, to non-USD and relatively unchanged USD rate). In fact, the widening of the non-USD gap with a narrow USD gap means a comparatively more rigid peg to the USD. As it is the overall basket that is managed, relative rigidity means real rigidity. This is not a healthy solution for China in the long term. It could even be argued as a reverse step from July 21. Unless, my speculative interpretation of "managing through spread" above is correct.

Hopefully, some measures would be introduced to mediate the USD and no-USD volatitily in near future. i.e. widening the USD gap. But if my gut feeling is right, USD rate change would be less than 2.5% p.a.

update Sep 24: Japanese Minister of Finance Watanabe commented on the 10 times difference in USD and non-USD gap, saying it does not make sense as a long term solution (in fact, the difference is 20 times, not 10 times. 0.3% vs 6%). he also mentioned that he discussed with PBoC and they are aware of the problem. His reasoning is based on the cross rate implication as discussed above. However, I would tend to think it is worth experimenting to see how the market behaves under such constraint (it is academically interesting). There are a few reasons that it might not be as straightforward as one has thought, because

1. If RMB transaction becomes large enough in the international forex market, then one needs to consider the interaction of direct eur-usd and the indirect (arbitrage) eur-RMB-usd rates. But RMB volume will be tiny for at least another decade.
2. The time lag effect of non-USD leading USD movement (hence the stabilizing effect) for an "pseudo"-USD peg (see horse-cart analogy above)

update Sep 25: PBC published a Q&A elaboration to explain the differential gap and flexibility in transaction charge (buy/sell spread). It explained why USD/RMB rate has been more or less pinned to 8.09-8.11. It also implied that with the enlargement in spread, the daily crawl in USD would be larger than the past 2 months. It also confirmed my conjecture above of "assymetric buy/sell" rate setting by the commercial banks

• The spread had been restricted at 0.2% after Jul/21, which is smaller the the gap of 0.3%. There is a possibility that the bank would lose on a tranaction, if it square the contract the following day. As a result, the commercial banks tend to maximize the spread by setting the buy and sell quotes to the extremity of the permitted values, and effectively there is only one price everyday (no crawling)
• 7月21日后，银行对客户挂牌的美元对人民币现汇买卖价不得超过央行公布的美元交易中间价上下0.2%，而银行间外汇市场美元对人民币的交易价在央行公布的美元交易中间价上下0.3%的幅度内浮动,前者幅度小于后者幅度，银行存在经营亏损的可能。假定某天银行对客户挂牌价格，人民币对美元升值了0.2%，从前一天的8.0890升值到8.0728，从客户手中收购1万美元付出80728元人民币，到银行间市场平盘时，发现平盘价格却升值了0.3%，达到8.0648，1万美元只能卖得80648元人民币，该笔交易直接亏损80元人民币。此外，7月21日后银行对客户美元对人民币现汇价格仍以央行公布的上一日收盘价为买卖中间价，加之最大买卖点差（即浮动幅度）只有上下0.2%，银行为规避经营风险，不得不每天一开市，就把价格挂到规定浮动幅度的两极，客观上造成银行在美元挂牌价格上只能一日一价，无法灵活调整。
• "Management by spread has 3 merits: 1) ensure that the central parity is enforced; 2) encourage price movement within the day (for the commercial banks) 3) allows assymetric buy-sell quote setting (effectively changing the central parity?)
• 价差幅度管理的优点在于：一是坚持了美元挂牌价格必须以央行公布的交易中间价为指导价；二是银行增加了自主性和灵活性，只要在足够大的价差范围内，银行牌价就可实行一日多价；三是只规定价差，银行可在价差范围内制定不以交易中间价为中心的非对称牌价，有利于银行根据当天市场供求情况和自身头寸灵活应对。
• [after Jul/21]...when a commercial bank is determining the EUR rate, it will need two parameters, the USD rate (from previsous day) and the USD/EUR rate from international market (real time), multiply them to obtain the real time rate for EUR/RMB... from Jan/1 to Jul/31 of 2005, the maximum daily 2.2%, therefore the 1.5% restriction we have set before will lead to loss of the bankss. That is why we enlarge the gap to become +/- 3%
• 7月21日配合人民币汇率机制改革，中国人民银行发布了调整汇价管理办法的通知，其中，银行间即期外汇市场上欧元、日元和港币等非美元货币对人民币交易价在中国人民银行公布的该货币交易中间价（即上一日央行公布的收盘价）上下1.5%的幅度内浮动。该文同时规定，银行对客户的非美元对人民币买卖中，买卖中间价由银行根据国际外汇市场行情自行套算，这相当于与国际外汇市场完全接轨。例如，银行在制定欧元对人民币汇率时要找到两个参数，一是国际外汇市场欧元对美元的实时汇率，二是央行公布的上一日美元对人民币的收盘价，前一个参数每时都在变化，一天波幅往往很大，后一参数一天之内固定不变，因此，两个参数相乘得出的欧元对人民币的汇率波幅就完全取决于国际市场欧元对美元的波幅。据我们统计，2004年1月1日至2005年7月31日期间，国际外汇市场欧元对美元汇率日波动最高幅度达2.2%。假定某天欧元对美元涨幅超过2%，银行对客户欧元对人民币汇率的涨幅相应也超过2%，而银行间外汇市场欧元对人民币一天最多只能涨1.5%，理论上银行可能面临0.5%的亏损率。为此，《通知》中将银行间外汇市场欧元、日元、港币等非美元货币对人民币交易价浮动幅度从上下1.5%扩大至上下3%。
• Before Jul/21 the gap for EUR was 10%, JPY 1%, USD 0.3% (I thought it was smaller?), now non-USD gaps are unified to 3%......the USD gap is significantly lower, but it is reasonable because USD is more liquid, and the currency which is more liquid tend to be less volatile, so a smaller gap is justifiable...
• 为什么银行间即期外汇市场美元浮动幅度不作调整，只调整非美元浮动幅度？是否有进一步调整美元浮动幅度的打算？ 答:首先，非美元浮动幅度高于美元幅度是我国外汇市场长期管理实践的延续。汇率机制改革前，我国银行间外汇市场非美元的浮动幅度就比美元大，如欧元为上下10%、日元为上下1%，而美元只有上下0.3%。汇率机制改革后，根据外汇市场多年运行实践，从增加汇率灵活性的需要出发，将非美元浮动幅度统一到1.5%。其次，美元浮动幅度较小符合外汇市场美元流动性高的客观国情。目前，我国外经贸活动80%左右以美元计价和结算，反映到国内银行间外汇市场的即期交易中，美元交易量占绝对优势，美元的流动性远远高于欧元、日元等非美元币种。国际外汇市场的经验表明，流动性越高的货币，市场越容易出清，价格的波动性就越小。国内外汇市场美元流动性较高客观上要求价格波幅较低，相反欧元和日元流动性不足，客观上只能用更大的波幅给交易者提供风险补偿。第三，汇率机制改革以来的市场运行情况表明，美元浮动幅度是适当的，它不但满足了市场参与者的交易需求，也保证了人民币汇率水平的基本稳定。