原创 2015-12-16 季天鹤央行观察央行观察

季天鹤方正中期研究院研究员，央行观察专栏作家

信贷市场正在重新扩张，财政支出预计将持续发力。人民币的稳定滞后于经济的企稳，央行在汇率方面会有更多后招...

央行重新陷入汇市纷扰
11月份的央行又陷入了人民币贬值和外汇流出的局面当中，行情似乎一下又回到了9月。首先我们看到央行11月外汇储备减少了872亿美元（约5600亿元人民币），央行外汇占款也减少3158亿元。但两者相比较，就会发现央行的外汇占款减少，比外汇储备下降的人民币水平要少很多，因此这一部分大概反映了估值的变化。

不过耐人寻味的是，外汇局的发布会提到，外汇储备参与了一些“走出去”项目的融资。这些融资按照国际货币基金组织关于外汇储备的定义，也会从外汇储备规模内调整至规模外。这一操作，在11月的表上似乎也有征兆，即央行其它国外资产增加549亿，这部分增加的含义，应该是央行代持但自己不能任意使用的某种国外资产。

这一块央行自己不能动用的资产背后有三种可能，第一种可能是，某家银行本来在央行有人民币存款，在11月中的某一天突然变成了外币存款。如果这部分存款被汇到国外，那么央行就会发现自己的外汇占款减少，同时银行在央行的存款减少。但如果这部分存款以外币的形态依然呆在人民银行的负债侧，人民银行成了一个代持者，那么人民银行资产侧会有外汇占款减少而其他国外资产增加，对银行而言则是在央行的人民币存款减少而外汇存款增加。

银行想要间接持有外币资产，除了找央行代持之外，应该还可以找中行来代持，即银行把外币资产存入中行，同时成为了中行的外币债权人，而中行则直接在海外持有这部分外币资产。关于这一机制更细致的讨论，可以参考《如何辨析“藏汇于民”与“资本外逃”？》以及《汇金与中投：精彩的注资大戏》等文章。

第二种可能是，某家同时可以在银行和央行开户的非银行金融机构，忽然把自己在银行的人民币存款变为了外币存款放在了央行。这一举动会使银行在央行的存款减少，非银金融机构在央行的存款增加，同时还改了央行资产侧国外资产的分类。这一可能性的表现，就是央行负债侧“不计入储备货币的金融性公司存款”增加了533亿。

第三种可能性就不涉及负债侧了，而是央行对某个境外实体进行了注资或贷款等活动，但这一境外实体可能是央行自己建立的，也可能是其他中资机构建立的，因而不能满足国际上对于外汇储备的定义，于是央行就把这部分国外资产从储备里拿出来，单独看作是“其他国外资产”，而和央行负债侧无关。当然，数据背后的可能性很多，以上三种可能性，都是在央行没有主动扩表的前提下进行的推测。

我们再来看看央行的其他资产项目。央行的对银行借款方面，中期借贷便利新增1003亿，抵押补充贷款517亿，常备借贷便利新增0亿，逆回购100亿，总计1620亿，接近央行资产负债表的1530亿。对其他金融性公司的债权没有变化，看来央行对证金的贷款还没有收回。而银行在央行的存款减少5045亿，则主要受到货币发行增加864亿，非银金融机构存款增加533亿，政府存款增加2958亿，以及国外负债减少249亿的影响。
其他存款性公司10月的债权之谜在《10月份的央行数据讲了哪些故事？》中，我们提到了M2和信贷收支表贷款之间的关系，指出M2和信贷收支表的存款之间存在口径的差异，而如果想从信贷收支表的贷款来推出M2的变化，可以分成两步，第一步是看看在信贷收支表的体系里贷款和存款增量的关系，第二步是把信贷收支表里的存款往M2的方向调整。由于信贷收支表里面贷款和存款的关系非常清楚，因此我们没有特别考察银行的贷款项目乃至资产侧。

但是现在，央行现在已经公布了10月的存款性公司概览以及其他存款性公司资产负债表，我们惊奇地发现，10月份银行对非金融机构的新增债权居然高达21828亿，而对其他居民部门债权仅增加1616亿。后者和存款性金融机构人民币信贷收支表中对住户的贷款的增幅一致，但前者和人民币信贷收支表中对非金融企业及机关团体贷款的总增幅则相差甚远。

资产的增加无非对应另一种资产的减少，或者一种负债的增加。10月份M2中存款部分的增量仅2319亿，而银行“对非金融机构及住户负债”中的其他负债项目增加17065亿，远高于往常的千亿水平，那么有可能上述的2万亿债权中，只有一部分是符合信贷收支标准的贷款或者债券，并且影响了存款和货币量。而剩余的扩表，则是和非金融机构达成的一种特殊安排，资产侧不算贷款，负债侧不算存款。这种安排给银行能带来什么好处呢？资产侧高于负债侧的利差？还是扩大资产本身能带来一些好处？这或许需要对银行和相关统计口径的进一步了解了。
信贷和货币创造依然向好11月金融机构表现着实不俗。除了各项贷款增加7089亿元之外，更关键的在于债券投资增加10720亿，股权和其他投资也增加了4947亿。正因为这样，虽然外汇买卖减少了2213亿元，但境内存款仍然增加1.5万亿元。不过境外存款连续第5个月减少，当月减少规模较10月增加至1027亿，超过了9月的1000亿元，显示离岸人民币的流动性还在继续收缩。同样连续下降的还有外币的存款和贷款，在11月分别下降了203亿美元和262亿美元。

离岸人民币存款的收缩也在“其他存款性公司资产负债表”上有反映，即“国外负债”一项从7月到10月每个月超过千亿元人民币的下降。除此之外，我们看到银行在9月和10月都在用自己的国外资产在应对汇出，外汇信贷收支中的外汇买卖也均为负值。但在11月，我们看到外汇买卖为增加101亿美元，似乎意味着银行从央行处购买了外汇，银行11月的国外资产降幅可能会缩窄。在贷款方面，七家大行的境内人民币贷款可能继续受到置换的影响，11月仅增加1241亿元，其中短期贷款减少1261亿元，而小银行的境内人民币贷款增加6030亿，其中短期贷款增加2680亿。这再一次印证了我们在《10月份的央行数据讲了哪些故事？》中提到的观点，即小银行贷款的不俗表现，应该是更接近于不考虑置换的情况，因此目前的信贷需求应该是比数据要好的。

而《10月份的央行数据讲了哪些故事？》中提到的开行、交行和邮储在央行存款增加可能会在未来慢慢流走，在11月也成为了现实。工农中建四大行在央行的存款减少664亿，小银行减少890亿，而开交邮三行则减少了2916亿元。考虑到11月银行在央行的存款总体上减少约5000亿，因此可以看出，开交邮三家银行确实通过了各种操作，让10月份进来的流动性，又重新流向其他银行。
展望下个月从目前看来，新的情况在于人民币重新开始贬值，而央行似乎没有明显干预的迹象，美元兑人民币汇率已经突破了811汇改后的高点。由此看来，12月的数据里，外汇占款还是会减少，但减少的规模应该和11月差不多，甚至可能会因为美联储加息落定，市场情绪重新稳定而有所缩减，虽然回到10月的情况可能性不大。

同时，财政存款的支出看上去并没有之前预计的迅速，可能是由于财政一边支出，地方政府又在一边发债，导致财政存款的减少看上去不够大。例如，财政支出1万亿，使银行在央行存款增加1万亿，银行负债侧增加1万亿存款。但是，财政又向银行发债，使银行在央行存款减少1万亿，债券资产增加1万亿，而财政在央行存款增加1万亿，负债增加1万亿债券，财政在央行的存款规模回到了财政支出以前。

如果财政不及时进行置换操作，即以偿还银行贷款的方式，把财政在央行存款重新变为银行在央行存款，那么这1万亿就不能及时流回银行间，看上去就是财政没支出。不过好在上述的置换与不置换，都不会影响银行的负债侧也就是广义货币，因此实体经济还是能够享受财政1万亿支出带来的新增货币量。所以银行买国债+财政支出的组合，其实能实现广义货币量的扩张。这大概也是为什么在存款性公司概览里，对政府债权和财政存款合并成一个“对政府债权（净）”，得以和“对非金融部门债权”等项目并列的原因。

The credit market is undergoing a renewed expansion, and fiscal expenditures are expected to continue their efforts. The stability of the Renminbi (RMB) lags behind the stabilization of the economy, and the central bank is likely to employ more strategies concerning the exchange rate...

In November, the central bank once again found itself entangled in the fluctuations of the foreign exchange market. The situation seemed to revert back to that of September. First of all, we observed that the central bank's foreign exchange reserves decreased by $87.2 billion (approximately ¥560 billion RMB) in November, and its foreign exchange position also decreased by ¥315.8 billion. However, upon comparing these two figures, we can notice that the reduction in the central bank's foreign exchange position is much smaller than the decrease in foreign exchange reserves. This likely reflects changes in valuation.

Interestingly, the State Administration of Foreign Exchange mentioned during a press conference that foreign exchange reserves were involved in financing certain "going abroad" projects. According to the definition of foreign exchange reserves by the International Monetary Fund (IMF), these financings could lead to adjustments from within the scope of foreign exchange reserves to outside the scope. This operation seemed to have some signs in November's data. The central bank's holdings of other foreign assets increased by ¥54.9 billion, which should represent a kind of foreign asset that the central bank holds on behalf of others but cannot use freely.

There are three possibilities behind these assets that the central bank cannot utilize directly. The first possibility is that a bank initially held RMB deposits with the central bank, but on a certain day in November, those deposits were suddenly transformed into foreign currency deposits. If these deposits were sent abroad, the central bank would find its foreign exchange position reduced, and the bank's deposits with the central bank would decrease. However, if these deposits remained as foreign currency on the liability side of the People's Bank of China, turning it into a holder, then the central bank's foreign exchange position would decrease while other foreign assets on the asset side would increase, resulting in the reduction of the central bank's RMB deposits and the increase of foreign currency deposits for the bank.

Banks, apart from having the central bank hold such foreign currency assets, could also approach Bank of China (BOC) to hold these assets on their behalf. In this scenario, the bank places foreign currency assets with BOC, becoming a foreign currency creditor of BOC, which directly holds these foreign currency assets overseas. For a more detailed discussion of this mechanism, refer to articles like "How to Distinguish 'Hiding Foreign Exchange Reserves Among the People' from 'Capital Flight?'" and "CIC and CIC International: A Wonderful Drama of Capital Injection."

The second possibility involves non-bank financial institutions that can hold accounts both with commercial banks and the central bank. Suddenly, they might transform their RMB deposits with commercial banks into foreign currency deposits held with the central bank. This action would cause the bank's deposits with the central bank to decrease while non-bank financial institutions' deposits would increase. Simultaneously, the classification of foreign assets on the asset side of the central bank would change. This possibility is reflected in the increase of ¥53.3 billion in the category "Financial Corporations' Deposits Not Included in Monetary Reserves" on the liability side.

The third possibility doesn't involve the liability side; rather, the central bank might have injected capital or provided loans to a foreign entity. However, this foreign entity might have been established by the central bank itself or other Chinese institutions, which doesn't meet the international definition of foreign exchange reserves. Therefore, the central bank isolates this portion of foreign assets from its reserves and treats it as "Other Foreign Assets," unrelated to the liability side of the central bank. Of course, there are numerous possibilities behind these data, and the three possibilities mentioned above are speculative scenarios assuming the central bank didn't actively expand its balance sheet.

Now, let's explore other asset items of the central bank. In terms of borrowing from banks, the Medium-term Lending Facility increased by ¥100.3 billion, Collateralized Supplementary Lending increased by ¥51.7 billion, the Standing Lending Facility remained unchanged, and reverse repurchase agreements increased by ¥10 billion, totaling ¥162 billion, which is close to the central bank's balance sheet of ¥153 billion. The claims on other financial corporations remained unchanged. However, banks' deposits with the central bank decreased by ¥504.5 billion, primarily influenced by an increase in currency issuance of ¥86.4 billion, an increase in non-bank financial institutions' deposits of ¥53.3 billion, an increase in government deposits of ¥295.8 billion, and a decrease in foreign liabilities of ¥24.9 billion. The mystery of October's claims of other financial corporations In the article "What Stories Did the Central Bank Data for October Tell?" we mentioned the relationship between M2 and the credit allocation table loans, noting that there are differences in definitions between M2 and the credit allocation table's deposits. If one wants to infer changes in M2 from the credit allocation table's loans, it can be done in two steps: first, understanding the relationship between loan and deposit increments in the credit allocation table system, and second, adjusting the deposits from the credit allocation table towards M2. Since the relationship between loans and deposits in the credit allocation table is very clear, we didn't particularly examine banks' loan projects and even the asset side.

However, now the central bank has published an overview of deposits by financial institutions and the asset and liability tables of other deposit-taking financial institutions for October. Surprisingly, we found that in October, banks' new claims on non-financial institutions reached an astonishing ¥2.1828 trillion, whereas their claims on other resident sectors increased by just ¥161.6 billion. The latter matches the increase in loans to households in the credit allocation table of deposit-taking financial institutions in RMB, but the former differs significantly from the total increase in loans to non-financial enterprises and institutions in the credit allocation table, indicating a significant gap.

An increase in assets corresponds to a decrease in another asset or an increase in a liability. In October, the increment in deposits in the M2 component was only ¥231.9 billion, while the increase in other liabilities of "Liabilities to Non-Financial Institutions and Households" of banks reached ¥1.7065 trillion, far exceeding the usual hundred-billion-level increments. It's possible that among the ¥2 trillion claims mentioned earlier, only a part adheres to the credit allocation standards of loans or bonds and affects deposits and the money supply. The remaining expansion might represent a special arrangement with non-financial institutions where assets are not counted as loans on the asset side and liabilities are not considered as deposits on the liability side. What benefits could this arrangement bring to banks? A higher interest rate spread on the asset side compared to the liability side? Or perhaps the expansion of assets itself brings certain advantages? Further understanding of banks and relevant statistical methods might be necessary to explore this. Credit and monetary creation remain positive In November, financial institutions exhibited commendable performance. Apart from an increase of ¥708.9 billion in various loans, the more critical aspect was the increase of ¥1.072 trillion in bond investments, as well as an increase of ¥494.7 billion in equity and other investments. Due to these factors, although foreign exchange transactions decreased by ¥221.3 billion, domestic deposits increased by ¥1.5 trillion. However, overseas deposits continued their fifth consecutive month of decrease, with a decrease of ¥102.7 billion in the month, surpassing October's increase of ¥100 billion, indicating a continued reduction in the liquidity of offshore RMB. Similarly, foreign currency deposits and loans also continued to decrease, with decreases of $20.3 billion and $26.2 billion, respectively, in November.

The contraction of offshore RMB deposits is also reflected in the "Asset and Liability Tables of Other Deposit-Taking Financial Institutions," specifically in the item "Foreign Liabilities," which has been decreasing every month from July to October by over ¥100 billion. In addition, we see that banks were using their foreign assets to cope with outward remittances in September and October; both the foreign exchange transactions in the credit allocation table and the foreign exchange transactions in banks' foreign assets were negative. However, in November, we observed an increase of $10.1 billion in foreign exchange transactions, suggesting that banks purchased foreign exchange from the central bank. The decline in banks' foreign assets in November might narrow. Regarding loans, the domestic RMB loans of the seven largest banks might still be affected by loan replacement, with an increase of only ¥124.1 billion in November, of which short-term loans decreased by ¥126.1 billion. Meanwhile, domestic RMB loans of small banks increased by ¥603 billion, including an increase of ¥268 billion in short-term loans. This once again confirms the viewpoint we mentioned in "What Stories Did the Central Bank Data for October Tell?"—the performance of loans in small banks is likely closer to the scenario without loan replacement, indicating that current credit demand might be better than the data suggests.

In the article "What Stories Did the Central Bank Data for October Tell?" we mentioned that the Bank of Communications, the Agricultural Bank of China, the China Construction Bank, and the Industrial and Commercial Bank of China might gradually divert deposits from the central bank in the future, which also became reality in November. The deposits of the four major state-owned banks decreased by ¥66.4 billion, and those of small banks decreased by ¥89 billion, while the Bank of China, China Construction Bank, and Postal Savings Bank of China decreased by ¥291.6 billion. Considering that banks' deposits with the central bank decreased by around ¥500 billion in November, we can see that the three banks indeed directed the liquidity that entered in October back to other banks through various operations. Looking forward to the next month, the new development is that the Renminbi has begun to depreciate again, and there seem to be no clear signs of intervention by the central bank. The USD-RMB exchange rate has already surpassed the high point after the exchange rate reform, reaching above 8.11. Consequently, the data for December will likely show a further reduction in foreign exchange position, but the decrease might be similar to November's or even smaller, possibly due to the stabilization of market sentiment following the Federal Reserve's interest rate hike decision. However, it's unlikely that we will return to the situation in October.

At the same time, it appears that the rapid depletion of fiscal deposits hasn't materialized as previously expected. This could be due to the fiscal expenditures being countered by local governments issuing bonds, making the reduction in fiscal deposits appear less significant. For example, if fiscal expenditures amount to ¥1 trillion, it increases banks' deposits with the central bank by ¥1 trillion. However, the fiscal authorities then issue bonds to banks, which decreases banks' deposits with the central bank by ¥1 trillion and increases bond assets by ¥1 trillion. The fiscal authorities' deposits with the central bank increase by ¥1 trillion, resulting in an increase of ¥1 trillion in their liabilities. This brings the scale of fiscal deposits with the central bank back to where it was before the fiscal expenditures.

If the fiscal authorities don't conduct timely replacement operations, i.e., converting fiscal deposits with the central bank back into bank deposits with the central bank by repaying bank loans, then the ¥1 trillion won't flow back to the interbank market in time, making it appear as though the fiscal expenditures haven't occurred. Fortunately, whether the above replacement occurs or not, it won't impact banks' liabilities, i.e., broad money. Therefore, the real economy can still enjoy the new money supply brought about by the ¥1 trillion fiscal expenditures. Thus, the combination of banks purchasing government bonds and fiscal expenditures can indeed lead to an expansion of broad money. This is probably why in the overview of deposit-taking financial institutions, claims on the government and fiscal deposits are merged into a single item "Claims on the Government (Net)," placed on par with "Claims on Non-Financial Sectors" and other items.