2015-05-04 季天鹤央行观察

作者：季天鹤，万门大学CFO，央行观察特约专栏作家

原创不易，转载请在文章开头显著位置注明原作者和来源微信公众号《央行观察》

人民银行调查统计司在《金融工具统计分类框架和设计》一文中专门讨论了非股票证券的回购：现金融入方在未来必定要将该证券购回，因此承担着证券的风险和收益，回购实质是一种抵押贷款。当现金融出方将通过回购协议获得的证券出售给第三方时，回购协议的现金融入方因回购协议约束，其资产方仍然记录着证券资产，而直接购买证券的第三方因买断式购入证券，其资产方也记录着证券的持有，会产生双重记录的问题。为消除这一错误，回购协议现金融出方应在转售证券时记录一笔负证券头寸，既反映了现金融出方“卖空”证券的经济现实，也克服了双重计算的问题。

从资产负债的角度看，现金融出方减少现金并增加债权资产，现金融入方增加现金并增加债务，没有涉及证券，所以融入方资产侧仍保留证券。卖给第三方之后，第三方的资产侧增加了证券，此时证券既体现在了第三方的资产上，也体现在了回购融入方的资产上，出现双重纪录。除此之外，融出方卖掉证券后必然换回另一资产或减少自身负债，但由于融出方没有记录卖掉证券这一资产减少，来中和换回的另一资产，融出方就意外地增加了所有者权益。所以调统司的负证券头寸就可以使得融出方所有者权益不变。

为什么融出现金会在现金融出方的资产侧产生“债权”而在现金融入方的资产侧产生“现金”，而融出证券既不在证券融出方的资产侧产生“证券债权”也不在证券融入方的资产侧产生“证券”呢。如果这样操作，那么回购后，现金融出方减少现金并增加“现金债权”资产，同时作为证券融入方在资产侧增加证券而负债侧增加“证券债务”；现金融入方增加现金并增加“现金债务”，同时作为证券融出方在资产侧减少证券并增加“证券债权”资产。证券卖给第三方后，卖出者也就是回购中的现金融出方/证券融入方资产侧的证券减少，换回另一资产，所有者权益不变，无须特意增加负证券头寸，而“证券”也只存在于第三方的资产侧，回购的现金融入方资产侧只有“证券债权”这一资产，并非证券本身，没有双重计算。

这种等价的记账背后，其实是如何看待例如国债之类的证券的货币性质。对于货币，人们有一套需求供给的分析思路，但往往把货币和商品、货币和其他金融产品区别，所以对于回购中证券融出方融出证券后的状态，人们虽然知道融出方仍然是证券的拥有者，享有证券带来的收益，并且回购到期时证券融入方需要归还证券，但不会想从货币角度把握。国债融入方归还融出方证券的时候，是否要归还当初对方融出的那张国债？在纸币上，这个问题就是储户取钱的时候是否要求银行提供当初存钱时的那张纸币。如果不要求是同一张纸，融出方资产上证券一项也完全没有更改过，那为什么纸会变化呢，总应该有一条会计分录以供审计吧？

所以融入现金和融入国债可以有相同的处理思路，可以尝试采取货币的视角，而且在国债等产品上也较为可行。国债之于国债持有者，正如纸币之于纸币持有者、存款之于储户一样，都是持有者的资产。同时，国债作为被广泛接受的回购抵押品，流动性极高，当然这两者是相辅相成的。用国债作抵押、借来现金做多其他金融产品获利，或者用国债作抵押、借入其他金融产品卖空从下跌中获利，都是完全可行的操作。事实上，上面交易中的“现金”完全可以省去，金融产品和国债的相对价格变化本身就可以看作该产品用国债衡量的价格变化，通过做多做空产品获利的“利”也可以用国债来衡量，国债就成了买卖其他金融产品的“货币”。

因此，抵押品紧缺和流动性紧缺其实是一回事。当回购双方都将融来的金融工具，即国债和现金，都用来购买了其他金融产品，若恐慌出现，则回购双方对国债和现金的需求会同时产生。虽然在目前的市场结构下，证券融入方需要卖出金融产品头寸换成现金，然后再买入国债以归还证券融出方，但像上面提到的，这个现金在理论上可以跳过，直接用国债交易。如果大家都抢着买国债而没有足够的卖方，那么国债的价格当然就会提高，也就出现了金融危机中国债价格上涨的景象。而如果回购中的证券融入方把国债借给第三方而不是售予，那么危机中对国债的需求会因债务关系的增加而增加，因为证券融入方和第三方都想买到国债。

面对国债紧缺，国债的发行方未必能像央行一样灵活地供给自己的负债，所以国债市场没有货币市场一样有“最终贷款人”平息危机。当然持有大量国债的某些机构，如养老基金、保险公司、商业银行甚至央行，可以借机卖出所持有的债券，但这些机构的很多头寸都是用于持有到期的，未必可以轻松改变用途，因此能用于交易获利的国债数量有限。而这种持有到期的需求，如同窖藏黄金一样，也可以被看作是“货币贮藏”的需要，九十年代末美国财政盈余时期对国债减少的忧虑也来于此。在国债总量不变的情况下，国债贮藏越多，可用于“流通”的国债越少，而在金融危机中，避险需求导致国债贮藏增加，而同时可供大家抢夺的“流通”国债又在变少，国债价格自然暴涨。

不同期限国债的货币意义也不同。货币视角下，永续国债就如同持有目前央行发行的纸币无法从央行兑换任何东西一样，成为一种不可兑换的“纸币”，而金本位时代可以兑换黄金的纸币则相当于一种随时可赎回的国债（不知是否问世过），而一般有期限的国债则居于两者之间。此外，我们也可以理解为什么日本政府可以维持令人乍舌的负债率。传统意义认为这是由于对日本政府的信心、货币政策的配合、以及配置资产的需要等，但考虑到央行纸币和准备金存款这些央行永远不用偿还的负债的存在，以及中国6000亿银行库存现金对应了110多万亿活期存款和可以提前支取的定期存款的现实，或许仅仅流通需求本身就已经足以支持一个非常高的国债存量，国债持有者不会想去挤提财政部要求其还本付息，而是无压力接受续发，财政部的税收能力和现金余额看上去完全不重要。

如果把国债看作纸币，那么国债能否实现货币扩张？事实上如果回顾货币扩张的原理，我们会发现，首先被定义的是货币本身，而货币包括了作为银行资产的纸币与作为银行负债的存款，银行无论通过直接借出纸币的方式还是直接扩张存款的方式来扩张资产负债表，其负债侧扩张的结果都是存款的增加而非纸币的增加。如果只有纸币是货币，那么就不再有货币扩张原理，而只有印钞了。在国债的情形下，除非国债持有者的债务能够如存款一般流通，从而增加了“国债货币量”，否则所谓回购扩大货币量、增发影子货币的说法，似乎更像是指的国债成为货币本身。不过或许财政部门大概会希望市场永远不要进化，国债持有者的债务永远不能像国债一样流通，国债也就不会被替代，可以维持较高的余额了。

在过去，连银行存款的流动性都很差，工资都要取出纸币来发放，这也为什么在中国，储蓄帐户居然连M1都算不上，因为这是1994年制定货币量统计时候的现实。但现在，这种发放只在春节的时候体现了，而银行存款真正流动起来成为了货币的主体，甚至还有余额宝这样以银行存款作为准备金的货币创造机制。对国债而言，凭证式国债和储蓄国债只能提前赎回或到期还本付息，没有流动性也没有成为货币的可能，只是现在流动性好了，能做的事情多了，才有了货币化的前景，或许有一天，火车票都能用国债来买了。

The People's Bank of China's (PBOC) Statistics and Research Department specifically discusses repurchase transactions of non-equity securities in the article "Statistical Classification Framework and Design of Financial Instruments." The cash borrower in a repurchase transaction is required to buy back the security in the future, thus bearing the risk and return of the security. Repurchase transactions are essentially collateralized loans. When the cash lender sells the security acquired through the repurchase agreement to a third party, the cash borrower, due to the constraints of the repurchase agreement, continues to record the security asset on its asset side. Similarly, the third party that directly purchases the security records the ownership of the security, leading to a dual recording issue. To eliminate this error, the cash lender in the repurchase agreement should record a negative security position upon reselling the security. This reflects the economic reality of the cash lender "shorting" the security and overcomes the issue of dual calculation.

From the perspective of assets and liabilities, the cash lender decreases cash and increases debt assets, while the cash borrower increases cash and increases debt liabilities. As securities are not involved, the asset side of the cash borrower retains the security. After selling to a third party, the third party's asset side increases with the security, while the asset side of the cash borrower in the repurchase agreement still reflects the security asset due to the initial constraints of the repurchase agreement, resulting in dual recording. In addition, when the cash lender sells the security and subsequently acquires another asset or decreases its own liabilities, the cash lender inadvertently increases owner's equity because the asset decrease due to the sale of the security isn't recorded, offsetting the increase from acquiring the other asset. Therefore, the negative security position managed by the Statistics and Research Department can keep the cash lender's owner's equity unchanged.

Why does cash borrowing generate "debt assets" on the asset side of the cash lender while generating "cash" on the asset side of the cash borrower, whereas borrowing securities doesn't create "security debt" on the asset side of the securities lender or "securities" on the asset side of the securities borrower? If operated in this manner, after the repurchase, the cash lender, which borrowed cash, decreases cash and increases "cash debt" assets. Simultaneously, the asset side of the cash borrower, which is also the securities lender, increases with the security while its liability side increases with "security debt." The cash borrower increases cash and increases "cash debt," while the asset side of the securities borrower, which is also the cash lender, decreases with the security and increases with "security debt." After selling securities to a third party, the asset side of the cash lender/ securities borrower in the repurchase agreement decreases with the security. This leads to acquiring another asset and maintaining unchanged owner's equity. There's no need to intentionally increase the negative security position, and "securities" only exist on the asset side of the third party, while the asset side of the cash borrower in the repurchase agreement only holds the asset "security debt," not the security itself. Thus, dual recording is avoided.

Behind this equivalent accounting is how securities like government bonds are viewed from a monetary perspective. When it comes to money, people have an analytical approach to demand and supply. However, they often distinguish between money and goods, money and other financial products. Therefore, when considering the status of the securities borrower in repurchase transactions, although people understand that the borrower still owns the securities, benefiting from their returns and needing to return the securities upon maturity, they don't tend to approach it from a monetary angle. When the securities borrower returns the securities to the lender, should they return the original government bonds that were lent out initially? In terms of paper currency, this would be equivalent to whether a depositor withdrawing money should receive the exact same paper currency they deposited. If the same paper isn't required, the asset side of the securities lender doesn't change in terms of the securities item, so why does the paper change? There should be an accounting entry for audit purposes.

So, why does borrowing cash create a "debt claim" on the asset side of the cash lender and "cash" on the asset side of the cash borrower, while borrowing securities doesn't create a "security debt claim" on the asset side of the securities lender or "securities" on the asset side of the securities borrower? If we consider securities as paper currency, government bonds and cash can follow the same treatment. Attempting to view it from a monetary perspective could be more feasible, especially for products like government bonds. Government bonds, for the bondholder, are like holding paper currency issued by the central bank, with an inability to exchange it for anything, just as paper currency is unredeemable at the central bank. Government bonds that could be redeemed for gold during the gold standard era could be seen as a form of government debt that could be redeemed at any time (if they ever existed). Regular government bonds with set terms fall between these two cases. Furthermore, we can understand why the Japanese government can maintain an astounding level of debt. Traditional views attribute this to confidence in the Japanese government, coordination with monetary policy, and asset allocation needs. However, considering the existence of central bank currency and reserve deposits as liabilities that the central bank never needs to repay, and the reality that 600 billion yuan in bank cash corresponds to over 110 trillion yuan in demand deposits and withdrawable time deposits, the idea that these deposits might account for part of the debt base doesn't seem crucial.

If we consider government bonds as paper currency, can government bonds lead to monetary expansion? In fact, when reviewing the principles of monetary expansion, we find that it's defined based on money itself, which includes bank assets like paper currency and bank liabilities like deposits. Whether a bank expands its balance sheet through direct lending of paper currency or by expanding deposits, the result on the liability side is an increase in deposits, not an increase in paper currency. If only paper currency is money, then the principles of monetary expansion cease to exist, and it's only about printing money. In the case of government bonds, unless the debt of government bond holders can circulate like deposits, thus increasing the "government bond money supply," the idea that repurchase can expand the money supply or lead to the issuance of shadow money seems to imply that government bonds become money themselves. Perhaps the fiscal department might hope that the market never evolves, and government bond holders' debt can never circulate like government bonds, allowing government bonds to remain high.

In the past, even bank deposits had poor liquidity, and wages were paid out in paper currency. This is why in China, savings accounts couldn't even be classified as M1, as was the reality when money supply statistics were formulated in 1994. But now, such payouts only occur during the Spring Festival, and bank deposits have truly become the mainstay of money, even leading to mechanisms like Yu'e Bao, where bank deposits are used as reserves for money creation. For government bonds, both coupon-bearing government bonds and savings bonds can only be redeemed early or upon maturity. They lack liquidity and the potential to become money. It's only now that liquidity has improved and more possibilities have arisen, leading to the prospect of monetization. Perhaps one day, even train tickets could be purchased using government bonds.