原创 2016-01-12 季天鹤央行观察

季天鹤
方正中期研究院研究员、央行观察专栏作家

令人惊心动魄的2016年第1周过去了。尽管去年12月CFETS人民币指数出来的时候，市场就已经预期到央行要有所贬值，但其到来之突然，波动之剧烈，以及市场主体在新一年表现出的购汇冲动，则还是出乎很多投资者意料，特别是股票投资者，一方面是不熟悉的人民币汇率波动，一方面是呱呱坠地的熔断机制，再加上救市限售令到期等因素的刺激，心理防线被冲破，股市两次熔断。

外汇市场和股票市场很不一样。外汇市场在最宏观的层面，乃是一个银行兑付问题。购汇交易中，外币是被兑付的目标，本币是用于兑付的工具，银行的外币资产和外币存款负债之间存在固定兑换率下的准备金率，而银行的外币资产与本币负债之间存在着不固定兑换率下的准备金率。之前的多篇文章都讨论过上述观点。

从具体的数字来看，银行去年11月底的国外资产在4.1万亿人民币。2015年底，银行持有的国外资产还在3.6万亿，2015年8月底则达到历史峰值4.65万亿。从3.6万亿到4.65万亿，再到4.1万亿，这先升后降反映了四个问题。

首先，银行自己早就知道人民币升值前景退却，因而主动增持了外币，而不是像过去一样都卖给人行。比如2012年中期，银行持有的国外资产达到3.2万亿人民币，随后在2013年下降至2.8万亿，但此后在2014年上升很快，年底达到了3.6万亿，这8000亿的增长，都是银行主动持有外汇的结果。

第二，上述银行国外资产迅速下降，说明这些资产的流动性是比较好的，可以应对市场主体对外汇出的需求。第三，在8月汇改之后，人行对银行在外汇方面进行了支持。《8月份的央行数据讲了哪些故事？》一文中仔细讨论了人行当时的行动，即外汇占款的抛售以及外汇储备以外的外汇转移。

第四，银行作为外汇市场的直接参与者，尽管面临市场主体汇出外币以及提取外币现钞的压力，但15年11月底银行的国外资产规模（4.1万亿）依然较2014年底多不少（3.6万亿），这反映出银行自身还有一定的应对外汇流出的能力。再汇出5000亿人民币的外汇，也只不过是把银行打回2014年底的水平。而打回2013年底的2.9万亿，则需要再减少7000亿人民币的外汇。

同时我们还要看到，银行的国外负债在2015年下降非常快，从2014年底的2.5万亿人民币下降至2.15年11月底的1.45万亿，下降1万亿人民币，特别是在2015年8月剧烈下跌近2千亿，此后每月下跌超过1千亿。这反映出，海外人民币加速回流境内，或者偿还境内银行的人民币贷款。按照目前1.45万亿的规模，再下降1万亿基本意味着人民币国际化回到原点，也就是2009年的水平。不过考虑到目前人民币的配置需求比那个时候多了不少，回到09年的规模不太可能，因此银行国外负债下降1万亿已经是非常夸张的情况。

我们再来看央行。央行的情况是国外资产配对准备金存款。2015年11月底，央行国外资产为26万亿元人民币，而准备金存款是20万亿。银行找央行购汇，导致央行国外资产减少，同时银行的准备金存款减少。看起来，目前央行的国外资产比准备金存款要多，但我们知道国外资产是历史上央行结售汇的历史成本，反映了很多历史上美元兑人民币1:7甚至1:8时候的交易。此外，国外资产不都是外汇储备，外汇储备自身的估值也会变化，所以毫不以外，官方储备资产的11月底合计是3.5万亿美元，和26万亿的比值并不是6.6的美元人民币汇率，而是7.4左右，恰在7与8之间。

至于外汇储备2015年底的3.33万亿美元到底有没有注水，我们可以从多个角度观察。首先，央行的外币流动性模板已经说得很清楚了，外汇储备的绝大部分都是证券。其次，从央行十几年来的资产负债表变化来看，关于外汇的大动作都看得很清楚，无论是注资汇金、注资银行、外汇缴纳人民币准备金等，都反映为外汇储备和外汇占款双减，而在其他资产类项目上增记。第三，外汇局近期的多次记者会也指出，外汇储备用作委托贷款等用途后会导致外汇储备规模减少。如果指责外汇储备的会计有“水分”，还需要更严谨的分析和证据，目前我们还没有看到。

目前的外汇市场是怎样的状况？尽管动机不同，但在具体操作层面上，银行的客户在做几件事情：向国外汇出外汇，在国内用人民币存款购汇形成外币存款，用人民币存款兑换外币现钞，此外还可能有衍生产品的交易。其中，衍生产品交易本身不涉及当期外币和人民币数量的变化，但会导致一些交易从而对人民币汇率产生影响。

购汇形成外币存款虽然也会导致一些银行间外汇交易，但购汇而不汇出使银行的外汇存款准备金率降低，而不会导致银行资产侧外汇离开银行。与之类似的还有一些“不可交割”的购汇产品，不占用额度，人民币汇率涨跌表现为人民币投资本金的增减，这样的产品虽然让银行客户可以从人民币汇率中获得损益，银行也会有一些对冲需要，但对银行来说恐怕构不成外汇流出的压力，甚至对外汇存款准备金的压力都不存在。

比较关键的就是汇出外汇和提取现钞。后者目前在新闻中被频繁报道，但这个交易速度实在太慢，银行无法提供现钞也不会被认为是很可怕无法等的事情，客户也会被劝说购买现汇形成外汇存款。如果劝说成功，银行面临的，不是提取外币现钞导致的外汇准备金率分子变小，而是外汇准备金率分母的变大。如果银行的外汇存款准备金率在5%，那么分子减小1%会使准备金率降至4%，而分母增加1%则会使准备金率降至4.95%，影响可谓天壤之别。

而汇出外汇则必定是人行打击的重点。这里又分为两种情况。基于实体需要的会受制于额度或者真实情况，但套利则不然。以目前离岸在岸的价差来看，只要套利机会存在，美元就会不断跑到境外。根本的解决方法，还是把两地的价差收敛到一定程度。阻碍套利资金流动虽能一定程度上缓解问题，毕竟不能根治。

如果人们把2万亿美元汇到境外，则人行外汇储备下降2万亿美元。如果人民币维持在7的水平，那么将有14万亿元人民币消失。这对于股票投资者而言或许很难理解，毕竟在股票市场里，股票买家把钱给股票卖家，钱只是流动而不会消失。但在外汇市场，银行售汇导致的乃是人民币存款的消失，而对外付款则是银行外汇资产的和外币存款负债的同步消失。

14万亿人民币已经是10%的货币总量，同期住户的活期存款是19万亿，非金融企业的活期存款是16万亿。如果市场疯狂到用14万亿人民币换美元，意味着每个人和企业的活期存款将近减半，或者市场主体损失定期存款的利息来换取基本没有利息的美元，或者把钱从股市中各种撤出，把股票卖给国家队或者各种抄底外资。而离岸人民币市场恐怕已经回到了NDF时代。

而对银行来说，14万亿的人民币存款消失，变成外汇汇出，意味着银行在央行的准备金也减少了14万亿人民币。这意味着在银行的资产负债两侧，同时少了负债和资产。考虑到准备金利率仅1.62%，还不如余额宝回报率，因此银行应该很愿意甩掉低回报的准备金资产，并且还甩掉了高成本的人民币存款资产，利润率说不定还能有些提高。

如果人民币在合理均衡水平上维持基本稳定，会导致一系列结果。央行外汇储备减少，银行国外资产增加，但银行国外资产的增加会被外汇汇出所部分抵消。银行负债侧有很多人购汇并且汇出，但大多数人只是把人民币换成美元后放在银行不动，银行承担外汇存款准备金的变化。央行的支持使外汇存款准备金率分子变大，市场主体的购汇使其分母变大，但由于准备金率小于1，因此在10%的准备金率上，央行对银行提供1千亿美元支持，能够应对不汇出的情况下10000亿美元的购汇，并维持准备金率不变。

在汇出流比较稳定或者影响较小的情况下，银行资产侧的外币少于负债侧的外汇存款，在人民币贬值时，这一资产负债缺口会导致银行本币计价下的汇兑损失。当然，这一汇兑损失会因外币存款利率很低而本币资产收益较高而受到一定程度弥补。如果在岸银行在央行中间价的调控下继续有序交易而没有恐慌，而央行的中间价小幅对美元贬值，并且能有效缓解跨境套利的压力，那么上述情况下，银行估值损失有限，外汇准备金率分子减小也会转变为分母扩大，央行在应对外汇汇出当中，可以争取比较长的时间。

维持人民币在合理均衡水平上基本稳定是很有希望的，但价格方面的稳定会以数量方面的变化作为代价。相对于人民币升值期间人行的无上限扩表，贬值期间的缩表会有下限，市场也会密切关注。不过3万亿美元的规模非常大，巨额的购汇会对金融体系会造成基础性变化。人民币货币量的减小，一定会体现在利率和实体经济上，促使人们思考要不要为了汇率上的一点收益放弃本币债权股权乃至实体投资的收益。

这个时候，一个意外的问题会出现，也就是在汇率波动使人民币货币量趋于下降的背景下，人行要不要努力把货币量增长目标继续维持在13%？要不要通过银行信贷创造更多的人民币出来让大家抛？让我们翘首以盼，期待两个月后人民银行给出的答案。

The heart-pounding first week of 2016 has passed. Despite market expectations for a devaluation when CFETS RMB Index was introduced in December of the previous year, the sudden arrival, intense volatility, and the buying impulse demonstrated by market participants in the new year have surprised many investors. Especially for stock investors, unfamiliarity with RMB exchange rate fluctuations and the abrupt circuit breaker mechanism, combined with the stimulus of measures like lifting stock sale restrictions, breached psychological defenses, leading to two stock market circuit breaks.

The foreign exchange market and the stock market are quite distinct. At the macro level, the foreign exchange market revolves around the issue of bank solvency. In currency exchange transactions, foreign currency is the target to be exchanged, while the domestic currency is the tool for exchange. The interbank foreign currency assets and foreign currency deposit liabilities have a fixed reserve ratio under fixed exchange rates, whereas the reserve ratio varies for interbank foreign currency assets and domestic currency liabilities under fluctuating exchange rates. Numerous articles have previously discussed the above viewpoint.

Examining specific numbers, as of the end of November, the foreign assets of banks stood at 4.1 trillion RMB. At the end of 2015, banks held foreign assets of 3.6 trillion, which peaked at 4.65 trillion by the end of August 2015. The shift from 3.6 trillion to 4.65 trillion, and then to 4.1 trillion reflects four issues.

Firstly, banks were aware in advance that the prospect of RMB appreciation had receded, prompting them to actively increase foreign currency holdings rather than selling to the central bank as they did in the past. For instance, in mid-2012, banks' foreign assets reached 3.2 trillion RMB, and then decreased to 2.8 trillion in late 2013, only to rise rapidly in 2014, reaching 3.6 trillion by the year's end. This 800 billion increase was a result of banks' active foreign exchange holdings.

Secondly, the rapid decrease in the aforementioned foreign assets indicates good liquidity for these assets, capable of meeting market demand for foreign exchange. Thirdly, after the August exchange rate reform, the central bank supported banks in terms of foreign exchange. An article titled "What Stories Do the Central Bank's Data in August Tell?" delves into the actions of the central bank at the time, which involved selling foreign exchange settlement, as well as transferring foreign exchange beyond foreign exchange reserves.

Lastly, as direct participants in the foreign exchange market, despite facing pressures from market participants' outward remittances and foreign currency cash withdrawals, the scale of foreign assets held by banks at the end of November 2015 (4.1 trillion) was still significantly larger than at the end of 2014 (3.6 trillion), reflecting banks' capacity to handle outflows. Even remitting 500 billion RMB in foreign exchange would only bring the banks back to the level at the end of 2014. To return to the 2.9 trillion at the end of 2013, a reduction of 700 billion RMB in foreign exchange would be needed.

Simultaneously, we must also consider the situation of the central bank. The central bank's foreign assets are matched against reserve deposits. By the end of November 2015, the central bank's foreign assets amounted to 26 trillion RMB, while reserve deposits stood at 20 trillion. Bank purchases of foreign exchange lead to a reduction in the central bank's foreign assets, while banks' reserve deposits decrease. Although the current foreign assets of the central bank seem greater than its reserve deposits, we should note that foreign assets represent the historical cost of the central bank's history of buying and selling foreign exchange, reflecting many past transactions when the USD to RMB exchange rate was around 1:7 or even 1:8. Furthermore, foreign assets don't solely constitute foreign exchange reserves; the valuation of foreign exchange reserves itself changes. Therefore, the total official reserve assets at the end of November were around 3.5 trillion USD, resulting in a ratio not of 6.6 USD to RMB as the 26 trillion suggests, but rather around 7.4, falling between 7 and 8.

Regarding whether the foreign exchange reserves of 3.33 trillion USD at the end of 2015 were artificially inflated, we can examine this from various angles. Firstly, the central bank's foreign exchange liquidity template is quite clear; the majority of foreign exchange reserves consist of securities. Secondly, looking at the central bank's balance sheet changes over the past decade, it's evident that significant movements related to foreign exchange, such as capital injections into China Investment Corporation and banks, as well as foreign exchange used for depositing renminbi reserve requirements, all lead to the simultaneous reduction of foreign exchange reserves and foreign exchange settlement. Thirdly, recent press conferences by the State Administration of Foreign Exchange have pointed out that using foreign exchange reserves for entrusted loans and other purposes would lead to a decrease in foreign exchange reserves. If there is any suspicion of "padding" in the accounting of foreign exchange reserves, more rigorous analysis and evidence would be needed; currently, we have not seen such evidence.

What is the current state of the foreign exchange market? Although motivations may differ, on the operational level, bank clients are doing several things: remitting foreign exchange overseas, using renminbi deposits to purchase foreign exchange to create foreign currency deposits, exchanging renminbi deposits for foreign currency cash, and potentially engaging in derivative product transactions. Among these, derivative product transactions themselves do not involve changes in the quantity of current foreign currency and renminbi, but they can influence the renminbi exchange rate through certain transactions.

Although buying foreign exchange to create foreign currency deposits could lead to some interbank foreign exchange transactions, doing so without remitting the foreign exchange decreases the foreign exchange reserve requirement ratio for these deposits, without causing the foreign currency assets to leave the bank. Similar to this are some "non-deliverable" foreign exchange purchase products, which do not occupy quotas and changes in the renminbi exchange rate are reflected as increases or decreases in the principal of the renminbi investment. Although such products allow bank clients to profit from renminbi exchange rates and banks require some hedging, they might not exert pressure on banks for foreign exchange outflows, and may not even exert pressure on foreign exchange reserve requirement ratios for foreign currency deposits.

The more critical aspect is remitting foreign exchange and withdrawing cash. The latter has been frequently reported in the news recently, but the pace of these transactions is quite slow. Banks' inability to provide cash also wouldn't be considered a frightening or urgent situation. Clients could be persuaded to purchase foreign exchange to create foreign currency deposits. If successful, banks would not face reduced foreign exchange reserve requirement ratios due to a decrease in the numerator resulting from providing less cash, but rather the denominator of the reserve requirement ratio would increase. If the foreign exchange reserve requirement ratio for bank deposits is 5%, then reducing the numerator by 1% would lower the ratio to 4%, while increasing the denominator by 1% would reduce the ratio to 4.95%, demonstrating a significant difference.

Remitting foreign exchange, on the other hand, would certainly be the central bank's focus. This can be divided into two scenarios. Transactions based on genuine needs could be limited by quotas or actual situations, but arbitrage is different. Looking at the spread between offshore and onshore rates at present, as long as arbitrage opportunities exist, USD will continuously flow overseas. The fundamental solution is to converge the spread between the two markets to a certain extent. While inhibiting the flow of arbitrage funds could alleviate the problem to some extent, it cannot entirely solve it.

If people send 2 trillion USD abroad, the central bank's foreign exchange reserves would decrease by 2 trillion USD. If the renminbi remains at a level of 7, approximately 14 trillion RMB would disappear. This might be difficult for stock investors to understand since in the stock market, buyers pay sellers, and money simply changes hands without disappearing. However, in the foreign exchange market, bank sales of foreign exchange lead to the disappearance of renminbi deposits, and foreign payments result in the synchronous disappearance of foreign exchange assets and foreign currency deposit liabilities.

14 trillion RMB amounts to 10% of the money supply, with household demand deposits at 19 trillion RMB and non-financial corporate demand deposits at 16 trillion RMB during the same period. If the market becomes frenzied and 14 trillion RMB is exchanged for USD, it means nearly halving the savings of individuals and corporations in demand deposits. People might be persuaded to purchase foreign exchange to create foreign currency deposits, which wouldn't be deemed as a very terrifying or urgent situation. If persuasion succeeds and individuals buy foreign exchange to create foreign currency deposits, the challenges for banks would be more about an increased denominator of foreign exchange reserve requirement ratios rather than a decreased numerator due to cash withdrawals. If the foreign exchange reserve requirement ratio for banks is 5%, then a decrease of 1% in the numerator would lower the ratio to 4%, while an increase of 1% in the denominator would still maintain a ratio of 4.95%, showing a significant impact.

Regarding banks, the disappearance of 14 trillion RMB in renminbi deposits and its conversion into foreign exchange remittance would mean a decrease of 14 trillion RMB in banks' reserve deposits with the central bank. This translates to simultaneous decreases in both liabilities and assets on banks' balance sheets. Considering the meager reserve rate of 1.62%, which is lower than even the return on Yu'EBao, banks would likely be eager to shed low-yield reserve assets and high-cost renminbi deposit assets, potentially boosting their profit margins.

If the renminbi is maintained at a reasonably balanced level and is essentially stable, a series of consequences would follow. The central bank's foreign exchange reserves would decrease, while bank foreign assets would increase, albeit offset to some extent by the foreign exchange remittance. Many bank clients would purchase and remit foreign exchange, but most individuals would simply exchange renminbi for USD and keep the funds in the bank. The shift in foreign exchange reserve requirement ratios, driven by the central bank's support, would be countered by market participants' foreign exchange purchases, with the reserve ratio being smaller than 1. Nevertheless, assuming a 10% reserve requirement ratio, the central bank's support of 100 billion USD could handle 1 trillion USD in purchases without remittance, maintaining the reserve requirement ratio unchanged.

In situations where foreign exchange outflows are relatively stable or have limited impact, the foreign currency assets on the asset side of banks would be lower than the foreign exchange deposits on the liability side. During renminbi depreciation, this asset-liability gap could result in banks incurring foreign exchange translation losses in terms of their base currency. Of course, such losses could be somewhat offset by low foreign currency deposit rates and higher returns on domestic assets. If onshore banks continue orderly trading under the guidance of the central bank's mid-point rate and avoid panic, and if the central bank slightly devalues the mid-point rate against the USD and effectively alleviates cross-border arbitrage pressures, then under the aforementioned circumstances, the valuation loss for banks would be limited, the reduction in the numerator of the foreign exchange reserve requirement ratio would transform into an increase in the denominator, and the central bank could buy more time in managing foreign exchange remittances.

It is quite hopeful to maintain the renminbi at a reasonably balanced level and essentially stable. However, price stability might come at the cost of quantity changes. In comparison to the boundless balance sheet expansion during the period of renminbi appreciation, the contraction during the devaluation period has its lower limit, and the market will pay close attention to this. Nonetheless, a scale of 3 trillion USD is incredibly substantial, and massive foreign exchange purchases could bring foundational changes to the financial system. The decrease in the renminbi money supply will be reflected in interest rates and the real economy, prompting people to consider whether to sacrifice returns on renminbi-denominated debt, equity, and even physical investments for a slight yield in exchange rates.

At this juncture, an unexpected question arises: Should the central bank continue to maintain the target of 13% money supply growth in the context of declining renminbi money supply due to exchange rate fluctuations? Should the central bank create more renminbi through bank credit to encourage people to sell? Let's eagerly anticipate the answer from the People's Bank of China two months from now.