2015-11-17 央行观察

央行的数据：外汇与缴税

央行较早时候公布的外汇储备，较9月底增加了113亿美元，而10月的央行外汇占款也增加了533亿人民币，两者基本协调。但我们也关注到，央行10月的国外负债增加了205亿元。由于央行在伦敦发行的50亿元央行票据，已经全部反映在了10月央行“发行债券”的项目里，所以这部分国外负债的情况，应该是央行除伦敦票据之外的一些负债。

央行负债的增加一般有两种情况。一种情况是，在负债增加的同时，增加了某种资产，比如逆回购，在增加央行对银行债权这一资产的同时，也让银行在央行的准备金存款增加。而另一种情况则是，在一种负债增加的同时，减少了另一种负债，比如发行央票，即把银行在央行的存款准备金，挪到央票的项目里。

这里插一句关于货币乘数的评论。由于央行对于基础货币中的银行准备金有很多操作，因此在计算货币乘数的时候，恐怕不能简单考虑M2/基础货币这一比率。因为央行通过央票和回购，已经把很多本来应该属于基础货币的准备金存款，弄到了其他的项目里，导致了基础货币量其实小于无央行操作情况下的规模，比率分母由此人为变小，需要特别注意。

回到央行的外汇操作上来。考虑到央行8月和9月的国外资产减少都在几千亿元规模，因此10月份央行新增的205亿的国外负债，如果是为了膨胀资产图好看，感觉很没有必要。较可能的是商业银行人民币清算的影响，即境外参与人民币业务的银行在央行这里存放备付金，用于新上线的CIPS。

与其说央行国外负债增加是为了做大资产，不如说国外负债增加对应了国内人民币的某种流出。从这个角度看，10月份外汇局将要通报的人民币跨境流出或许会转正。不过，金融机构境外人民币存款10月减少了896亿元，境内外币贷款减少206亿美元，看来外汇去杠杆的进程还在继续。

10月份的外汇占款和外汇储备情况其实在市场预料之中，因为当月在岸-离岸汇率一度出现正价差。离岸市场上，人民币利率也比较高，并且当月似乎还颇发生了几次踩踏行情，有利于人民币汇率稳定，以及央行在低买高卖和套利中赚钱。不过外汇占款从减少几千亿人民币直接翻正，还是有点出乎意料。

除了外汇占款之外，比较大的资产增加项，当属央行对银行的债权增加了754亿，不过这包含了1000亿的逆回购缩量，以及1760亿的中期借贷便利以及抵押补充贷款。此外，尽管有报道称参与股市救灾的国家队已经账面浮盈，但由于央行给其他金融性公司的债权并无变化，看来证金公司还欠着央行这笔钱。

再来看看央行的负债侧。十一长假结束，大家之前取出的纸币回存银行，央行的货币发行也就减少了2383亿。这部分回流到银行的纸币硬币，一部分形成了银行的库存现金，一部分应该回到了央行。如同社会公众到银行存入纸币，会增加在银行的存款一样，银行把纸币还给央行，也有助于银行在央行的存款增加，所以银行在央行的准备金存款，受纸币回存影响，应该有2000亿左右的增加。

此外，考虑到央行还进行了754亿的流动性净注入，以及外汇占款项目的稳定，银行10月份在央行存款应该增加2000余亿。但事实上，银行在央行的存款反而减少了1503亿，央行也在10月降了准。虽然国外负债205亿的增加以及央行债券新增的50亿，会减少银行在央行存款，但和1503亿的规模仍然相距较远。

答案很简单：缴税使财政在央行的存款增加了近5000亿，这在央行的降准公告里也提到了。缴税并不是什么突发事件，每年都是如此。而缴税除了把银行在央行的存款，挪到财政在央行的存款账户之外，也会使得社会公众在银行的存款减少，而这当然会影响到M2，也是解开10月贷款与M2增幅背离之谜的钥匙。

M2和贷款：没有背离之谜，只有口径差异

M2和存贷款数据，本身来源是不同的统计体系，这一点是分析该问题时首先需要注意的。M2是央行资产负债表-存款性公司资产负债表-其他存款性公司资产负债表的体系，而存贷款则是人民币信贷收支表的统计口径。其次，M2和存贷款数据包含的项目本身也不一样。

M2包含了M0（流通中现金），M1-M0（企业存款+机关团体部队存款+农村存款+信用卡类存款），M2-M1（城乡居民储蓄存款+企业存款中具有定期性质的存款+信托类存款+外币存款+非存款类金融机构在存款类金融机构的存款+住房公积金存款），其中对于外币存款其实有一些争议，因为有些资料也提到外币存款不算M2，还需要继续核实。

存贷款口径下的存款，包括了两个M2-M0中不包括的项目，一个是境外人民币存款，而另一个是财政性存款。而M2当中，则包括了“流通中现金”这个不包含在存贷款中的存款的项目，所以，在考虑存贷款口径下的存款与M2的时候，要考虑到上述三个项目带来的扰动。

此外，除了项目口径之外，M2和存贷款还有机构口径的差异。M0作用于“存款性公司”，而存款性公司就是银行+信用社+财务公司，不包括信托和租赁等。而存贷款数据则作用于“金融机构”这一口径，比“存款性公司”要大，因此需要使用“存款类金融机构人民币信贷收支表”来考察，减少口径不合。

存贷款体系下，存款和贷款的关系是平衡的，因为信贷收支表的来源和运用是相等的。这和M2体系下“国外净资产+国内信贷=M2+非M2存款+债券+实收资本+其他”一样。因此，如果想了解每个月贷款和M2之间的关系，只需要考虑贷款和存款如何在信贷收支体系下受到其他项目影响，以及存款如何和M2之间如何弥合差异，就可以了。

如果考虑同比增长率的话，还要看基期的情况，总体上比较复杂，特别是中国的统计口径每年有修正，而且2015年初的修订规模还不小，此外对于冲存款等操作的限制、取消存贷比等，都会对同比数据造成较大扰动，因此不如直接分析同一个口径下的月度变化来得简单直接。

首先我们从贷款入手。贷款在10月份以金融机构的口径，增加了5136亿，而以存款类金融机构的口径，则增加了4602亿。此外，存款类金融机构还有债券投资增加5152亿，股权和其他投资增加1412亿等，最终使资金运用增加11354亿，而这当然也等于资金的来源。而资金来源中，“其他”项增加了5597亿，剩下的基本就是存款和流通中货币的变动了

因此，我们看到，10月份的存款类金融机构，在各项存款上增加6674亿，其中住户存款减少5921亿，非金融企业存款增加275亿，非存款金融机构存款增加5737亿，机关团体存款增加2368亿，此外还要考虑到流通中货币减少1122亿。上述的四个存款，相当于从“各项存款”中去掉财政性存款以及境外人民币存款。

所以，考虑前述几种存款的增加和流通中货币的减少，由信贷收支表推断的类似M2口径，也就是考虑四个存款和一个货币，相当于-5921+275+5737+2368-1122=1137亿，和M2的增加1200亿就比较接近了。从2015年以来的历次月度变化来看，两者误差不大。有兴趣的读者可以自己做一下比较。

上述讨论和前文关于货币乘数的讨论一样，想说明在讨论数据的时候，需要注意各方面操作的影响。货币的创造和货币的流动都很重要。贷款创造的货币被用来缴税了，看起来和贷款没有创造货币一样，但事实上只是账目的调整。12月财政大支出的时候，必然又会出现贷款和货币量的背离。

四大行：存款搬家和贷款缩表

前面提到，10月金融机构人民币存款增加5136亿，但我们发现，四大行（工农中建）境内存款减少4038亿，而中小行则增加6313亿。在细项上，四大行个人存款当月减少5406亿，单位存款当月减少1435亿，而反观中小银行，个人存款增加261亿，单位存款更是增加4335亿。

从历史上看，10月份的中小银行存款一般都是负增长或者小额净增长，原因在于9月底银行为了业绩而扩张资产负债，导致10月份一般有一个收缩的过程。但今年10月中小银行存款增长6000亿，则是历史首见。这方面的可能的原因，可能是在利率进一步市场化之后，市场主体把在四大行的低利率存款，搬到利率较高的中小行去，也就是存款搬家。

但另一个原因，或需要从贷款侧观察。十月份的四大行各项贷款居然减少656亿，这是从2010年以来第一次出现负值。其中短期贷款更是减少2027亿。而中小银行境内贷款则增加4772亿，超过了9月份，其中中长期贷款增加2330亿。因此，有理由相信，四大行在10月遭遇的，除了存款的转移之外，还有存款还贷去杠杆。

去杠杆的原因，有可能是之前发行置换债券累积创造的存款，在10月被用来集中还贷，使四大行的贷款余额减少，而还贷也使存款余额减少。不过，由于中小行的贷款和存款都增加较多，似乎置换对于中小行的影响和大行有天壤之别，这一点看上去有一些奇怪。但由于缺乏更多的细节信息，因此还不能确定原因。

而如果我们相信四大行确实受到了置换的影响，那么四大行和中小行的贷款存量差不多，但前者10月份贷款余额减少，后者增加较为明显，不太可能是大行的贷款客户突然都跑到了小银行，而更应该解读为贷款总体的需求还可以，中小银行的贷款增量比较自然，只不过四大行由于置换或者某些原因显得很差，拉低总量使其不及预期。

此外还需要关注的是，四大行和中小行在央行的人民币存款总共减少了近4300亿，而存款性公司整体在央行的存款应该只减少了1500亿。其间的差额，很大程度上反映为开行、交行和邮储在央行存款的增加。这很可能意味着目前这几家银行今后准备进行一些大的动作，所以先把备付金收进来，未来在贷款当中慢慢任其流走。

总体来看，10月份的数据还有一些亮点，包括非银行金融机构在银行存款增加，意味着市场风险偏好回升；中小银行贷款数据还不错；财政在央行的存款虽然当月增加，但同比下降了18%（8000亿），支出力度还是相当明显的；外汇占款和外汇储备都不温不火，意味着国内主体心态趋于稳定，只剩下离岸主体去和央行作交易对手了。

之后两个月的最大看点，恐怕是财政存款到底能支出多大的一波。现在存贷比也没了，在财政存款回流的支持下，再搞出个银行间流动性紧张恐怕说不过去。此外，10月底以来联储加息态度对于人民币走势的利空，和人民币加入SDR对人民币走势的利好，孰强孰弱也需要观察，从目前走势来看外汇占款可能还会重新减少，但幅度应该不会回到8、9月份的水平。

Foreign exchange and tax payment.

The People's Bank of China (PBOC) earlier released data on foreign exchange reserves, which increased by $11.3 billion compared to the end of September. Additionally, the PBOC's foreign exchange holdings increased by ¥53.3 billion in October, with both figures showing close coordination. However, it's worth noting that the PBOC's foreign liabilities increased by ¥20.5 billion in October. This increase in foreign liabilities could be due to factors beyond the issuance of ¥5 billion central bank bills in London, which is already accounted for in the October PBOC's "bond issuance" category. This suggests that there are some other foreign liabilities for the PBOC apart from London bills.

There are generally two situations in which the PBOC's liabilities increase. One situation is when, while liabilities increase, certain assets are also increased, such as reverse repurchase agreements. By increasing the PBOC's assets in the form of bank claims, the reserves held by banks at the PBOC also increase. Another situation involves increasing one type of liability while decreasing another, such as issuing central bank bills. This involves shifting bank reserve deposits at the PBOC to the central bank bills category.

A brief note on the money multiplier: Due to the PBOC's various operations involving bank reserves in the base money, the calculation of the money multiplier might not be as simple as M2/base money ratio. This is because the PBOC has, through central bank bills and repurchase agreements, moved many reserve deposits that should belong to base money into other categories. This makes the actual size of the base money smaller than it would be under a scenario without PBOC operations, artificially reducing the denominator of the ratio, which needs special attention.

Turning back to the PBOC's foreign exchange operations, considering that the foreign assets of the central bank decreased in August and September, the increase of ¥20.5 billion in foreign liabilities in October might not be aimed at inflating assets for cosmetic purposes. A more plausible explanation could be related to the impact of RMB clearing by commercial banks overseas. These banks hold reserve funds at the PBOC for the newly launched Cross-Border Interbank Payment System (CIPS).

Rather than interpreting the increase in foreign liabilities by the PBOC as an attempt to expand assets, it might indicate a certain outflow of RMB within China. From this perspective, the People's Bank of China might report a positive trend in RMB cross-border outflows in October. However, it's noteworthy that overseas RMB deposits of financial institutions decreased by ¥89.6 billion, and foreign and domestic currency loans decreased by $20.6 billion in October. This suggests that the process of reducing foreign exchange leverage is ongoing.

The foreign exchange holdings and foreign exchange reserves in October were within market expectations. The onshore-offshore exchange rate had a positive spread at one point in the month. In the offshore market, RMB rates were relatively high, and there were several instances of apparent speculation, favoring RMB exchange rate stability. The central bank might have earned profits from buying low and selling high, but the direct shift from a decrease in foreign exchange holdings to a positive increase was somewhat unexpected.

Apart from foreign exchange holdings, the significant increase in assets could be attributed to the PBOC's increase in bank claims by ¥75.4 billion. However, this includes a ¥100 billion reduction in reverse repurchases, as well as a ¥176 billion increase in medium-term lending facilities and collateralized supplementary loans. Additionally, despite reports suggesting that the national team participating in stock market rescue operations has realized paper profits, the PBOC's claims on other financial institutions remained unchanged, implying that these institutions still owe money to the central bank.

Looking at the liability side of the central bank, after the end of the "Golden Week" holiday, people returned the banknotes they withdrew earlier to banks, resulting in a decrease of ¥238.3 billion in the PBOC's money issuance. A portion of these banknotes and coins returning to banks would have become banks' cash reserves, while another part would return to the central bank. Just as when the public deposits banknotes in banks, increasing bank deposits, returning banknotes to the central bank would also contribute to increasing banks' reserves at the central bank. Therefore, the impact of banknote return might lead to an increase of around ¥200 billion in banks' reserve deposits at the PBOC.

Furthermore, considering that the PBOC injected a net liquidity of ¥75.4 billion and that the foreign exchange holdings were stable, bank deposits at the PBOC should have increased by over ¥200 billion in October. However, in reality, bank deposits at the PBOC decreased by ¥150.3 billion, and the PBOC also lowered the reserve requirement ratio in October. Although the increase of ¥20.5 billion in foreign liabilities and the ¥5 billion addition from central bank bond issuance might reduce banks' deposits at the PBOC, the gap between these factors and the decrease of ¥150.3 billion is still considerable.

The answer is quite simple: tax payments increased the deposits of the treasury at the central bank by nearly ¥500 billion. This was also mentioned in the PBOC's reserve requirement ratio reduction announcement. Tax payments are not unexpected events and occur every year. Besides transferring deposits from banks to the treasury at the central bank, tax payments also lead to a reduction in deposits held by the public in banks. This, of course, affects M2, providing the key to understanding the divergence between loan growth and M2 increase in October.

M2 and Loans: No Mystery of Divergence, Only Methodological Differences

The sources of M2 and loan data are from different statistical systems, which is the first thing to note when analyzing this issue. M2 is calculated using the system of the PBOC's balance sheet - the balance sheets of deposit-taking companies - other balance sheets of non-deposit-taking companies. On the other hand, deposit and loan data is based on the statistical methodology of the RMB credit and debit statement. Additionally, M2 and deposit/loan data include different items.

M2 includes M0 (currency in circulation), M1-M0 (enterprise deposits + government and institutional deposits + rural deposits + credit card-type deposits), and M2-M1 (urban and rural resident savings deposits + enterprise deposits with a fixed-term nature + trust deposits + foreign currency deposits + non-deposit financial institution deposits in deposit-taking financial institutions + housing provident fund deposits). It's important to note that there is some controversy around foreign currency deposits in M2, as some sources claim they are not included and further verification is required.

In the deposit and loan system, deposits include two items not covered in M2-M0: overseas RMB deposits and treasury deposits. Meanwhile, M2 includes the item "currency in circulation," which is not included in deposit and loan data. Therefore, when considering deposits from the deposit and loan perspective in relation to M2, one must take into account the disturbance caused by these three additional items.

Additionally, apart from item differences, M2 and deposit/loan data also differ in terms of institutional perspective. M0 operates on "deposit-taking companies," which comprise banks, credit cooperatives, and finance companies, excluding trusts, leases, etc. Meanwhile, deposit and loan data operates on the "financial institutions" perspective, which includes more entities than "deposit-taking companies," making it necessary to use the "RMB Credit and Debit Statement for Deposit-Taking Financial Institutions" to ensure alignment.

Under the deposit and loan system, deposits and loans are balanced, as indicated by the source and use of funds in the RMB credit and debit statement. Similarly, under the M2 system, "net foreign assets + domestic credit = M2 + non-M2 deposits + bonds + paid-in capital + other," reflecting the balance between sources and uses of funds. Therefore, to understand the relationship between loans and M2 each month, it's necessary to consider how loans and deposits are affected by other items in the credit and debit statement and how deposits are reconciled with M2.

If considering year-on-year growth rates, it's also important to consider the base period's conditions, which can be quite complex. China's statistical methodology is subject to annual revisions, with a significant revision in early 2015. Additionally, factors like restrictions on deposit expansion, removal of loan-to-deposit ratios, and so on, can lead to significant disturbances in year-on-year data. Therefore, analyzing monthly changes within the same methodology might be simpler and more straightforward.

Starting from loans, in October, loans increased by ¥513.6 billion from the perspective of financial institutions and ¥460.2 billion from the perspective of deposit-taking financial institutions. Furthermore, deposit-taking financial institutions had an increase of ¥5,152 billion in bond investments, ¥1,412 billion in equity and other investments, ultimately resulting in a net increase of ¥1,134 billion in fund use, which is also equal to the source of funds. Among the sources of funds, the "other" category increased by ¥559.7 billion, and the rest mainly pertained to changes in deposits and currency in circulation.

In this light, we see that in October, deposit-taking financial institutions increased their deposits by ¥667.4 billion, with household deposits decreasing by ¥592.1 billion, non-financial corporate deposits increasing by ¥27.5 billion, deposits from non-deposit financial institutions increasing by ¥573.7 billion, and deposits from government and institutional entities increasing by ¥236.8 billion. In addition, currency in circulation decreased by ¥112.2 billion. These four types of deposits essentially exclude treasury deposits and overseas RMB deposits from the "various deposits" category.

Therefore, considering the increases in these several types of deposits and the decrease in currency in circulation, the similar M2 perspective derived from the credit and debit statement, which considers four types of deposits and currency, would be approximately -¥5,921 + ¥275 + ¥5,737 + ¥2,368 - ¥1,122 = ¥1,137 billion, which is quite close to the ¥1,200 billion increase in M2. Looking at the historical monthly changes since 2015, the discrepancies between the two are not significant. Interested readers could perform a direct comparison.

This discussion, much like the earlier discussion on the money multiplier, aims to emphasize the need to consider the impact of various operations when discussing data. Both the creation and circulation of money are important. Although it might seem that money created through loans has been used to pay taxes, rendering it similar to money not created through loans, it's just a matter of adjusting accounts. When significant fiscal spending occurs in December, a divergence between loans and money supply is likely to emerge once again.

Big Four Banks: Deposit Relocation and Loan Contraction

As mentioned earlier, in October, financial institutions' RMB deposits increased by ¥513.6 billion. However, we observe that the four major banks (Industrial and Commercial Bank of China, Agricultural Bank of China, Bank of China, and China Construction Bank) saw their domestic deposits decrease by ¥403.8 billion, while smaller and medium-sized banks increased by ¥6,313 billion. In terms of specific categories, household deposits at the four major banks decreased by ¥540.6 billion, while corporate deposits decreased by ¥143.5 billion. In contrast, smaller and medium-sized banks saw household deposits increase by ¥261 billion and corporate deposits increase by ¥4,335 billion.

Historically, smaller and medium-sized bank deposits in October tend to either decrease or show small net increases due to a contraction process following the expansion of balance sheets by financial institutions at the end of September for performance purposes. However, this year's ¥6,000 billion increase in smaller and medium-sized bank deposits marks a historic first. This could be attributed to market participants, after further interest rate liberalization, transferring their low-interest-rate deposits from the major banks to smaller banks offering higher rates, essentially engaging in "deposit relocation."

However, another reason could be related to loan demand. The four major banks experienced a rare decrease of ¥65.6 billion in their loan balances in October, the first negative value since 2010. Short-term loans, in particular, decreased by ¥202.7 billion. Meanwhile, smaller and medium-sized banks' domestic loans increased by ¥477.2 billion, surpassing September's numbers, with medium- to long-term loans accounting for an increase of ¥233.0 billion. As such, there's reason to believe that, in addition to deposit relocation, the four major banks have undertaken deleveraging measures involving deposit repayment through loan contraction.

The reason behind this deleveraging could be related to the accumulation of deposits resulting from the issuance of replacement bonds, which were then used to repay loans, leading to a decrease in both loan and deposit balances for the major banks. However, given that smaller and medium-sized banks experienced significant loan and deposit growth, the impact of replacement bonds on them seems starkly different from their effect on major banks. This raises questions, but without further details, the exact reasons remain uncertain.

If we believe that the major banks are indeed affected by the replacement bonds, their loan and deposit balances would be roughly equivalent to those of smaller and medium-sized banks. However, major banks' loan balances decreased in October, whereas smaller and medium-sized banks saw notable increases, particularly given that the former's total loan balance is near that of the latter's, minus the effect of replacement bonds.

Furthermore, the decline in total RMB deposits by the four major banks and smaller and medium-sized banks at the PBOC amounted to nearly ¥430 billion, whereas total deposits by deposit-taking companies likely decreased by only ¥150 billion. This difference likely reflects increased deposits by Bank of China, China CITIC Bank, and Postal Savings Bank of China at the PBOC. This might indicate that these banks are preparing for significant actions in the future, initially bringing in reserve funds and gradually allowing them to flow away during loan operations.

In conclusion, the data from October reveals some positive aspects, including the increase of non-bank financial institutions' deposits at banks, suggesting a rise in market risk appetite. The loan data for smaller and medium-sized banks is still favorable, while the increase in treasury deposits at the central bank in the same month remains a considerable 18% decrease (¥800 billion) in comparison to the same period last year. Foreign exchange holdings and reserves remain stable, indicating stabilized sentiment among domestic entities, while offshore entities engage with the PBOC.

The main focus in the next two months is likely the extent of government deposit spending. With the removal of the loan-to-deposit ratio and support from the return of treasury deposits, it's unlikely that an interbank liquidity shortage similar to previous years will occur. Additionally, the impact of the Fed's rate hike attitude on the RMB exchange rate, coupled with the positive influence of the RMB joining the SDR, warrants observation. From the current trend, foreign exchange holdings will likely decrease but not return to the levels seen in August and September.