季天鹤看懂经济 2016-07-07 08:04

在我们分析中国金融体系在宏观层面的诸多操作时，贯穿始终的分析视角是流动性的层级结构。但是，随着我们把视野拓宽到更广阔的金融工具、更全球的金融体系、更多样的金融交易之后，我们发现层级结构的流动性乃是一个更宏大叙事的微观部分。本文将在此前的基础之上，讨论层级结构背后更宏大的叙事。

简要回顾流动性的层级结构

在中国，流动性的层级结构由高到低，包括央行资产侧的流动性、央行负债侧的流动性、境内银行负债侧的流动性、离岸银行负债侧流动性等层。例如，人行外汇干预离岸人民币市场时，人行资产侧外汇减少、负债侧人民币准备金存款减少、境内银行负债侧人民币存款减少、离岸银行负债侧人民币存款减少，涉及4个流动性层面。不能认识到上述层面，就无法理解央行的离岸存款准备金操作。

离岸银行负债侧流动性和余额宝权益侧流动性在同一层次。6月末余额宝规模已经超过8000亿元，而上述权益均可用于购买商品和服务。如果央行突发奇想，购买余额宝，那么央行负债侧流动性、银行负债侧流动性和余额宝权益侧流动性会同时扩张。这种扩张方式，和央行为证金提供贷款的机制是一样的，只不过证金的负债不是流动性，因此央行贷款给证金只涉及两个层面的流动性。

认清流动性的分层结构在流动性分析中非常重要。例如法定准备金制度事实上是作用于央行负债侧流动性，和银行负债侧的流动性无关。又例如国家发行债券给银行是作用于央行负债侧流动性，和银行负债侧流动性也无关，不会引发M2的减少。但是国家向个人和企业征税则是同时影响央行负债侧和银行负债侧流动性，这时会影响到M2。

M2增速放缓和银行间流动性紧张也没有直接关系，因为两者是在不同层面的流动性，而同理M2迅速扩张也并不意味着银行间就能宽松。13年钱荒当中人们提出的“货币超发为什么还会钱荒”的问题，其实是由于没有理解这两者完全不在一个流动性层面上，“货币超发为什么还会钱荒”这个问题本身就是一个伪问题。

流动性的网格结构

随着我们把视野拓宽到更广阔的金融工具、更全球的金融体系、更多样的金融交易之后，我们发现层级结构的流动性乃是一个更宏大叙事的微观部分。例如，人行在中行有国家外汇存款，中行在人行有外汇存款准备金存款。在这一对关系中，显然有央行负债侧外币流动性和银行负债侧外币流动性，但哪个层次的流动性更高呢？

又例如，在欧央行和美联储的货币互换中，欧央行在美联储有美元存款，美联储在欧央行有欧元存款，这两者的负债侧谁高谁低呢？如果外国央行开始在中国的银行开设存款账户，那么我们会发现一个更大的环，即人行-中国银行-外国银行-外国央行-中国银行-人行。在人民币存款转账时，人行负债侧是可能的最高顶点，但在外币存款转账的时候，外国央行负债侧是可能的最高顶点。

一家银行内部的总行和分行，也可能是环上的不同节点，我们有可能看到人行总行在香港汇丰有人民币存款，香港汇丰在中银香港（香港人民币清算行）有人民币存款，中银香港在人行深圳中支有人民币存款，深圳中支在人行总行有人民币存款。另一个环是人行在新加坡星展银行有人民币存款，星展银行在工行新加坡分行（新加坡人民币清算行）有人民币存款，工行新加坡分行在工行总行有人民币存款，工行总行在人行总行有人民币存款。

一旦产生了环，由高到低的层级结构就受到了冲击。在层级结构里，人们不考虑具体的交易，也可以指出哪个流动性层面高，哪个流动性层面低。但是，在环当中，如果不清楚具体的交易为何，那么我们就无法确认任何的流动性层次。这其实涉及到了货币理论最深层面的内容，即什么是流动性。

流动性不是定义的结果，不是设计的结果，实践是检验流动性的唯一标准。贝壳布匹、金银铜铁，只要人们使用起来，就是流动性。余额宝权益是否是广义流动性或狭义流动性，不取决于其是否在M2里，而取决于人们是否能用它方便购物。同样，流动性的层次也无法事先确定，不同的交易里面，有不同的流动性层次，流动性层级结构是特定流动性操作的展现。

一个个环叠加在一起，就是一个网格结构。央行、银行总行、银行分行、个人、企业、基金等等，全都是网格的不同节点。连接节点的“线”是债权债务关系或者股权权益关系。两个节点之间的线并不只有1条，而是可以有多条。如果我们给整个流动性结构照一张静态照片，我们会看到很多节点，以及节点之间许多条不同颜色的线。

上面提到的线，显然不是银行负债侧和储户资产侧之间的线。事实上银行负债侧存款和储户资产侧存款指的乃是一个东西。上面所说的线，乃是指类似银行资产侧和银行负债侧存在的线。储户跨行转账或者取现，会同时影响到银行的特定资产和特定负债。而我们知道，银行的特定资产是另一个机构的负债（比如央行），那么上面所说的线，也可以看作是不同机构负债之间的关系，即银行负债和央行负债被线拴在了一起。

同样颜色的线，意味着其穿过的所有节点（也就是机构）的特定负债有同时变动的能力，例如离岸银行负债侧的流动性（离岸人民币）和人行负债侧有同时变动的能力，他们之间就被同样颜色的线串在一起。变动的能力不等于变动本身，比如一家离岸银行的本行存款转账，就不会带动人行负债侧的变化。

变动的能力是怎样决定的？答案是清算结构。离岸人民币存款之所以和人行负债侧联系，因为跨行转账这个操作可能会涉及到人行负债侧的清算。离岸银行A和离岸银行B可能分别在境内银行甲和境内银行乙开户，也可能共同在境内银行亥开户。如果是前面一种情况，那么A和B之间的转账就要涉及到甲乙间的转账，而甲乙间的转账会涉及到央行负债侧。而如果是后一种情况，那么A和B之间的情况就到亥为止，前面提到的“线”就触及不到央行负债侧。清算结构的等价表述，当然就是准备金结构。

清算结构不是无限延伸的，也不是必然成环的。人行负债侧和人行资产侧完全可以是隔绝的关系。人行如果愿意用外汇干预人民币外汇市场，市场参与者如果愿意获得央行的外汇，那么人行负债侧人民币存款就和人行资产侧美元存款（这当然是境外某个机构的负债）相联系，而如果人行不想干预，那么人行负债侧人民币存款就和人行资产侧美元存款隔断，也就和人行美元存款所在的那个机构的负债侧没有关系。在这个情况下，人行用自己的T账户中的竖杠，把清算结构导致的线隔断了。

但人行负债侧的美元存款呢？美元存款和人民币存款不在同一种颜色的线上，美元存款对应着美元的清算结构，当然也就和人民币存款清算结构不是同一根线。人行可以通过资本项目管制来切断美元这条线，使人行在境外的美元存款不会受到境内银行在央行的美元存款的拉动，但资本项目管制和T账户的竖杠不是一个机制，毕竟，银行在央行的人民币存款无论如何转账，也转不出央行在境外的美元存款。这里我们看到了结售汇这个操作的威力，这相当于让操作者切换了清算结构，从人民币的条线变成了美元的条线。

总之，流动性的网格结构，包括节点和线两部分。静态来看，节点是一张张资产负债表，线是一个个清算结构。动态来看，一张张资产负债表的规模和机构不断沿着各式各样的清算结构扩张和收缩。在特定的交易里，我们会发现特定的层级结构，但一旦脱离交易，我们看到的只有网而没有层级结构。这不禁让我想起《孙子兵法》中的一句话:“水因地而制流,兵因敌而制胜。兵无常势，水无常形”。

小结

一旦意识到流动性的网状结构之后，我们就有了分析流动性的通解。首先，在理论上，所有的机构间都可以建立债权和股权关系，同时还可以持有一些不是任何机构债权股权的实物（典型的就是黄金）。于是我们先有了下面的一张理论关系表。其中，每一个项目都是右面机构在下面机构拥有的资产，比如第二行最后一格就表示“Kyle Bass在人行有人民币存款”。

在这个非常理论的结构中，我们完全看不到任何层级结构，可以在上面随便画圈或者画层级，比如我在工行北分有人民币存款，工行北分在联储有人民币存款，联储在我这里有人民币存款。这个表格还有一个绝妙之处，就是对角线居然为黄金白银等不是任何机构负债股权的“纯资产”也留下了位置，对应了金属本位时代的情况。复本位制或者平行本位制自然反映为多种“纯资产”以及针对上述“纯资产”的各类存款。在未来，比特币何时取代黄金的位置，或者在黄金旁边开辟自己的位置，是十分需要人们关注的点。

上面的联储在我这里有人民币存款当然是不实际的。所以，我们可以考虑现实当中的情况。在现实当中，很多项目肯定是不存在的，我们可以得出一个更现实的关系图。我们看到，虽然很多项目都在现实中不存在，但这种不存在不是绝对的静止的，而是随着时间推移可能动态变化的，比如余额宝横空出世、比如人民币国际化等等，都对下面的表产生了修改。

在上面的现实的表的基础上，我们就可以考察具体的交易了，相当于是在上面画各种各样的线，也就能够找到各种各样的层级结构。图中画了几个我们熟悉的层级结构，一个是离岸人民币相关的条线，一个是余额宝相关的条线，一个是美元清算的条线。箭头意味着层级结构里面由低到高的层次。现实当中的情况怎样，取决于现实当中的清算安排，这也是为什么我们看到王永利老师总是提出主张，让境外银行的人民币清算账户开在境内商业银行负债侧而非人行负债侧，因为确实可以做出上述两种安排出来。

未来除了“存款”这种流动性之外，还会有更多形式的流动性。若没有余额宝份额横空出世，下一种流动性本来应该是国债等债券。债券或者说更广义的证券已经有了很成熟的托管体系和借贷抵押体系。托管体系也分为一级二级甚至更多级的托管，托管资产不入托管机构资产负债表，托管资产和客户的资产提取权一一对应，类似于银行体系下100%准备金制度。借贷抵押体系会导致部分准备金的制度，即客户的证券提取权大于借入证券机构持有的债券规模，这种情况下整个借入的证券都会进入借入证券机构的资产负债表，尽管债券公允价值的损益变动以及利息收益可能会传递给证券出借人。当然，目前证券还没有支付方面的功能，应用场景还较为有限，距离“流动性”的境界还有距离。

总之，流动性在现在的情况已经超过了绝大多数人的日常认知。即使对于很多专业金融研究人士来说，如果不专攻这一领域，也很难理解流动性到底是怎样来去生灭的。但事实上，流动性从来没有脱离过人们实实在在的生活，其所基于的会计和清算原理以及其执行的功能在昨天、今天和明天也都一直适用。只不过随着科技和人们的变化，流动性的具体运行和表现会不断演化和变化，它也就成了利率汇率研究领域时时刻刻都需要直面和展望的永恒主题。期待未来更丰富的流动性形式和流动性运行方式出现。

When analyzing the various operations of the Chinese financial system at the macro level, a continuous analytical perspective that runs through it all is the hierarchical structure of liquidity. However, as we broaden our perspective to encompass a wider range of financial instruments, a more global financial system, and a more diverse set of financial transactions, we realize that the hierarchical liquidity structure is but a micro part of a grander narrative. Building upon the foundation of the preceding discussions, this article will delve into the larger narrative behind the hierarchical structure.

Brief Overview of the Hierarchical Liquidity Structure

In China, the hierarchical liquidity structure spans from high to low, encompassing liquidity on the central bank's asset side, liquidity on the central bank's liability side, liquidity on domestic bank liability sides, and liquidity on offshore bank liability sides. For instance, when the People's Bank of China (PBOC) intervenes in the offshore RMB market, it involves reducing foreign exchange assets on the PBOC's asset side, decreasing RMB reserve deposits on the PBOC's liability side, reducing RMB deposits on domestic banks' liability side, and decreasing RMB deposits on offshore banks' liability side—engaging four levels of liquidity. Without recognizing these levels, understanding the PBOC's offshore reserve requirement operations becomes difficult.

The liquidity on the liability side of offshore banks and the equity side liquidity of investment funds like Yu'E Bao (a money market fund) exist on the same level. By the end of June, Yu'E Bao's scale had exceeded 800 billion RMB, and the aforementioned equity could be used to purchase goods and services. If the PBOC were to suddenly decide to purchase Yu'E Bao, liquidity on the PBOC's liability side, liquidity on banks' liability side, and equity side liquidity of Yu'E Bao would all expand simultaneously. This expansion mechanism is similar to the one where the PBOC lends to China Investment Corporation (CIC); the only difference is that CIC's liabilities aren't liquidity-based. Therefore, when the PBOC lends to CIC, it only involves liquidity at two levels.

Recognizing the layered structure of liquidity is crucial in liquidity analysis. For example, the required reserve ratio system actually affects liquidity on the central bank's liability side and is unrelated to liquidity on banks' liability side. Similarly, when the government issues bonds to banks, it affects liquidity on the central bank's liability side and is unrelated to liquidity on banks' liability side—it doesn't lead to a decrease in M2 money supply. However, when the government imposes taxes on individuals and businesses, it affects liquidity on both the central bank's liability side and banks' liability side, potentially impacting M2.

The slowdown in M2 growth and interbank liquidity tightness are not directly related since they are liquidity aspects on different levels. Likewise, rapid expansion of M2 doesn't necessarily imply an easy interbank environment. The question raised during the money shortage in 2013, "Why would there be a money shortage when there's excessive currency issuance?" is actually a pseudo question due to the lack of understanding that these two aspects exist in entirely different liquidity levels. The question itself is flawed.

The Grid Structure of Liquidity

As we broaden our perspective to encompass a wider range of financial instruments, a more global financial system, and a more diverse set of financial transactions, we realize that the hierarchical liquidity structure is but a micro part of a grander narrative. For example, within the relationship between the PBOC and the Bank of China, the PBOC holds foreign exchange deposits with the Bank of China, while the Bank of China holds RMB reserve deposits with the PBOC. In this relationship, there clearly exists liquidity on the central bank's liability side related to foreign currency and liquidity on banks' liability side related to foreign currency. But which level of liquidity is higher?

Similarly, in currency swaps between the European Central Bank (ECB) and the Federal Reserve (Fed), the ECB holds deposits in dollars with the Fed, and the Fed holds deposits in euros with the ECB. Which side's liquidity is higher in this case? If foreign central banks were to open deposit accounts with banks in China, a larger loop would emerge: PBOC - Bank of China - Foreign Bank - Foreign Central Bank - Bank of China - PBOC. During RMB deposit transfers, the PBOC's liability side may be the highest point, but during foreign currency deposit transfers, the foreign central bank's liability side may be the highest point.

A bank's head office and its branches can also be nodes on this loop. For instance, we might find that the PBOC's head office holds RMB deposits with HSBC in Hong Kong, HSBC holds RMB deposits with Bank of China (Hong Kong), Bank of China (Hong Kong) holds RMB deposits with the PBOC's Shenzhen branch, and the PBOC's Shenzhen branch holds RMB deposits with the PBOC's head office. Another loop could involve the PBOC holding RMB deposits with DBS Bank in Singapore, DBS Bank holding RMB deposits with ICBC Singapore Branch (RMB Clearing Bank in Singapore), ICBC Singapore Branch holding RMB deposits with ICBC head office, and ICBC head office holding RMB deposits with the PBOC head office.

When loops emerge, the hierarchical structure's integrity is challenged. In the hierarchical structure, by not considering specific transactions, we can still pinpoint which liquidity level is higher or lower. However, within loops, if we're unclear about specific transactions, we can't confirm any liquidity hierarchy. This touches upon the most fundamental aspects of monetary theory—what is liquidity?

Liquidity is not a defined result or a designed outcome; practice is the sole standard by which to assess liquidity. Cowrie shells, cloth, gold, silver, copper, iron—anything people use becomes liquidity. Whether Yu'E Bao's equity is broad liquidity or narrow liquidity depends not on whether it's in M2 but on whether people can conveniently use it for purchases. Similarly, liquidity levels cannot be predetermined; different transactions involve different liquidity levels. The hierarchical liquidity structure is an exhibition of specific liquidity operations.

When each loop is stacked upon another, it forms a grid structure. Central banks, head offices, branch offices, individuals, corporations, investment funds, and more—all represent various nodes on the grid. The "lines" connecting nodes are relationships of debt or equity. There isn't just one line connecting two nodes; there can be multiple. If we were to capture the entire liquidity structure in a static image, we'd see many nodes and numerous lines of different colors connecting them.

The aforementioned lines clearly don't represent the connection between banks' liability side and depositors' asset side. In fact, bank liability side deposits and depositor asset side deposits are referring to the same thing. The lines mentioned earlier refer to relationships similar to the bank's asset side and bank's liability side. Depositor interbank transfers or withdrawals simultaneously affect a bank's specific assets and specific liabilities. And as we know, a bank's specific assets are another institution's liabilities (e.g., the central bank), thus, the lines mentioned above can be seen as relations between different institutions' liabilities—binding bank liabilities and central bank liabilities together.

Lines of the same color indicate that all nodes (or institutions) these lines pass through have the ability to simultaneously change specific liabilities. For instance, offshore bank liability side liquidity (offshore RMB) and central bank liability side liquidity have the ability to change simultaneously. They are interconnected by lines of the same color. However, the ability to change doesn't equate to actual change. For instance, a local bank's interbank deposit transfer won't trigger changes on the central bank's liability side.

How is the ability to change determined? The answer lies in the clearing structure. Offshore RMB deposits are connected to the People's Bank of China's (PBOC) liability side because cross-bank transfers might involve the clearing of the PBOC's liability side. Offshore Bank A and Offshore Bank B might each have accounts with Domestic Bank A and Domestic Bank B respectively, or they could both have accounts with Domestic Bank C. In the former case, transfers between A and B would involve transfers between A and B and would consequently involve the PBOC's liability side clearing. However, in the latter case, the situation between A and B stops at C, and the aforementioned "lines" wouldn't extend to the PBOC's liability side. An equivalent representation of the clearing structure, of course, is the reserve structure.

The clearing structure isn't infinitely extendable nor is it inevitably circular. The PBOC's liability side and the PBOC's asset side could be completely isolated. If the PBOC wishes to intervene in the foreign exchange market, and market participants desire foreign exchange from the central bank, then the RMB deposits on the PBOC's liability side would be linked to the PBOC's asset side deposits in USD (which are liabilities of some overseas institution). However, if the PBOC chooses not to intervene, the RMB deposits on the PBOC's liability side would be disconnected from the USD deposits on the PBOC's asset side, as well as from the institution's liabilities holding USD deposits. In this scenario, the PBOC uses the vertical bar in its T-account to sever the lines created by the clearing structure.

But what about the USD deposits on the PBOC's liability side? USD deposits and RMB deposits don't lie on the same line; USD deposits correspond to the USD clearing structure, which isn't the same line as the RMB clearing structure. The PBOC can sever the USD line through capital account controls, preventing the movement of USD deposits held by overseas banks from influencing USD deposits held by domestic banks at the PBOC. However, capital account controls and the vertical bar in the T-account aren't the same mechanism. After all, no matter how RMB deposits are transferred between banks, they won't alter the PBOC's USD deposits held abroad. Here, we can see the power of foreign exchange operations, which allows operators to switch the clearing structure from RMB to USD.

In summary, the grid structure of liquidity consists of nodes and lines. Statically, the nodes represent balance sheets, while the lines are clearing structures. Dynamically, the scales of the balance sheets and institutions continuously expand and contract along various clearing structures. In specific transactions, we can identify specific hierarchical structures. However, once we move away from transactions, we see the grid without hierarchical structure. This reminds me of a quote from "The Art of War" by Sun Tzu: "Water shapes its course according to the nature of the ground over which it flows; the soldier works out his victory in relation to the foe whom he is facing. Therefore, just as water retains no constant shape, so in warfare there are no constant conditions."

Conclusion

Once the network structure of liquidity is realized, we have a universal solution for analyzing liquidity. First, theoretically, all institutions can establish debt and equity relationships, and they can also hold some physical assets that aren't the liabilities or equities of any institution (typified by gold). This leads to a theoretical relationship chart. In this chart, each entry represents assets held by the institution on the right side under the institution on the bottom; for example, the last entry in the second row indicates "Kyle Bass holds RMB deposits at the PBOC."

In this highly theoretical structure, we can't see any hierarchical structure at all. We can freely draw circles or hierarchies on it; for example, I have RMB deposits at Bank North Branch, Bank North Branch has RMB deposits at the Central Bank, and the Central Bank has RMB deposits at my account. This chart has a remarkable feature—it leaves space even for "pure assets" like gold and silver that aren't the liabilities or equities of any institution, reflecting the situation during the gold standard era. The gold standard or parallel gold standard naturally translates into various "pure assets" and various types of deposits corresponding to these "pure assets." In the future, when Bitcoin replaces gold or establishes its own position alongside gold, it's a point that requires attention.

The aforementioned scenario where the Federal Reserve holds RMB deposits at my account is unrealistic. Therefore, let's consider the real-world scenario. In reality, many entries certainly won't exist, allowing us to derive a more realistic relationship graph. We see that although many entries don't exist in reality, this non-existence isn't absolutely static but could change dynamically with time. For instance, the emergence of Yu'E Bao and the internationalization of the RMB have all caused modifications to the table.

Based on the above realistic table, we can examine specific transactions, essentially drawing various lines on it and identifying various hierarchical structures. The diagram illustrates a few familiar hierarchical structures—one related to offshore RMB, one related to Yu'E Bao, and one related to USD clearing. The arrows indicate hierarchical levels from low to high. The real-world situation depends on the actual clearing arrangement; this is why we often hear Wang Yongli advocating for overseas banks' RMB clearing accounts to be held by domestic commercial banks' liability side rather than the PBOC's liability side, as both arrangements are indeed feasible.

In the future, beyond forms of liquidity like "deposits," there will be more forms of liquidity. If not for the emergence of Yu'E Bao, the next form of liquidity would likely be bonds like government bonds. Bonds or the broader category of securities already have mature custody systems and lending collateral systems. Custody systems come in first-tier, second-tier, and even higher tiers, and custody assets aren't included in the balance sheets of custodian institutions. Custody assets correspond one-to-one with customers' asset extraction rights, akin to the 100% reserve requirement system in the banking system. The lending collateral system can lead to partial reserve systems, where customers' rights to extract securities exceed the scale of bonds held by the borrowing securities institution. In this scenario, the entire borrowed securities would enter the asset side of the borrowing securities institution's balance sheet, although changes in the fair value of bonds and interest income might be passed on to the lenders of the securities. Of course, currently, securities don't serve payment functions, and their application scenarios are quite limited, still a considerable way from the ultimate realm of "liquidity."

In conclusion, liquidity in the current situation has exceeded the understanding of most individuals. Even for many professional financial researchers, unless they specialize in this field, comprehending how liquidity ebbs and flows remains challenging. However, liquidity has never been detached from tangible reality; the accounting and clearing principles upon which it is based and its functional execution have been applicable yesterday, today, and will continue to be applicable tomorrow. With technological advancements and societal changes, the specific operation and manifestation of liquidity will evolve and change, making it an eternal theme that the field of interest rate and exchange rate studies must constantly face and anticipate. We look forward to the emergence of even more diversified forms of liquidity and ways of operation in the future.