Distributed Economics

When debugging complex systems, it can be useful to ask, “What’s missing?”. Especially when looking at  the available monitoring information, which tells you not only what the designers of the system thought it was important to know, but also by omission, what wasn’t. Critical realtime systems, we’re talking Nuclear Power plants and their ilk, are monitored fairly closely, but if the right information isn’t provided to the operator, then that won’t necessarily help all that much.

In this sense, it’s quite interesting that  Economics doesn’t seem to have  a definition of what money is, in any satisfactory, scientific sense. Rather it seems to have various descriptions of what money does. The “Money is a matter of functions four, a medium, a measure, a standard, a store.” dogeral that is found in some textbooks and on Wikipedia is the perfect illustration of this. This is not a definition of what money is, it’s simply a list of the things money is used for. This confusion can also be found in the various measures used to quantify the total amount of money in the Economy, which are variously M0, M1, M2, and depending on which country you’re in M3 or M4.

Interestingly, and perhaps damningly in the context of the current credit crisis, no single, monitored central bank measure appears to completely capture the information that is of most interest today – which is what is the total quantity of money that commercial banks and reserve based savings institutions are allowed to create loans against.

M0/Monetary Base

This is the “narrowest” definition of the money supply. The British, who technically invented the entire system, used to use M0, but since 2006 it’s the “Notes and coins and reserve balances” figure (YWMB43D), whilst the Federal Reserve simply refers to this as the Monetary Base, and reports it in the H3 Aggregate Series.  It is defined as the reserves from the Banks held on deposit by the Central Bank, and the total amount of physical currency. This measure probably made more sense under the Gold Standard than it does now, when a significant number of bank deposits don’t in fact carry a reserve requirement, and physical currency is simply printed to meet day to day demand.

Reserves held at the central bank used to be a very fixed, and important part of the system’s regulation, since this directly controlled lending. In that implementation of the system, the Commercial Banks had to deposit the reserve on deposits they were required to hold behind for each loan (i.e. the 10% fractional reserve requirement), at the central bank. For two simple reasons. It kept them honest, and it gave the Central Bank complete control, and indeed visibility, on the monetary expansion that was occurring. Assuming for example, that the Banks were correctly depositing their reserve on deposits at the central bank, then if that amount started growing, the central bank immediately knew that the loan supply was increasing and vice versa.

Only net transaction accounts in the US Banking system carry a reserve requirement currently, so that part no longer works very well. Even so – when the Federal Reserve saw the non-borrowed reserves drop by 25% in December 2007, and indeed go negative the following month, it was a clear warning sign of major systemic issues. The Bear Stearns collapse into bankruptcy would occur 3 months later in March 2008. How significant it is that there is also a fall in the British reserve measure, beginning in July 2006 and reaching its nadir in September 2007 before an abrupt recovery, is hard to say.

M1

M1 is technically physical currency and checkable deposits. Quite what it will be when it is no longer possible, as it will be and to a large extent already is in Europe, to write a cheque on a bank account, is an interesting question. It’s reported by the Federal Reserve as part of the H.6 dataset, along with its cohort M2, and they also helpfully break out the different components of M1 if you scroll down a little.

Broadly then, it’s physical currency and demand, or instant access checkable deposit accounts, which appear to be the Net Transaction Accounts for which there is still a reserve requirement. I have an HSBC internet savings account, it has no cheque book, pays a (very) small interest rate, and which i can transfer to and from using the Internet at will. Is it classified under M1 or M2? Does it have a reserve requirement? Does that even matter?

The last question is perhaps the most interesting. The answer appears to be both yes and no as it happens. Since reserve requirements are no longer used to directly regulate loan expansion – equity capital does that – then it doesn’t strictly matter if an account has a reserve requirement or not. Except of course, that the Banks can at the moment get quite a nice interest rate from the Federal Reserve on their reserves, which is why reserves so dramatically increased a year ago. Now there’s no correspondingly dramatic shift in the components of M1, so it looks like the Banks just dumped all their spare money onto the federal reserve – which lends some credence to the claim that this has inhibited lending. Except of course, that they’re still as busy as ever creating and  selling Asset Backed Securities.

M2

Strictly, M2 is all of M1 plus Bank savings accounts and Retail Money Funds. The Federal Reserve helpfully breaks out the non-M1 components of M2 as part of its H6 dataset, which makes life a little easier.

Now this is where it gets quite interesting. Retail Money Funds, which were a non-existent category at the beginning of 1959, and peaked in October 2008  at slightly over $1 trillion are around 15% of the total quantity of M2. Retail money funds are funds that generally provide an extremely safe return, that is expected to track the safest form of investment (government securities), and never not return their capital investment. They’re also expected to be very liquid, i.e. you can get your money back more or less on demand. This essentially means that they invest in short term debt securities, although only ones of very high quality. Like triple A rated, insured, top of the tranche Mortgage Backed Securities for example?

Leaving aside the interesting co-incidence that the growth of retail money funds begins in the same period as MBS were introduced (late 1973, sales don’t really kick off until the S&L crisis in the early eighties); just the total quantity of retail money funds indicates that some of them must be in MBS securities. There simply aren’t enough other suitably ’safe’ instruments paying a suitable rate of return otherwise. Which means that some proportion of the M2 money supply measure, is actually debt.

This helps to explain the divergence in the growth between M1 and M2, in the USA and other countries, since thanks to the Asset Backed Security loophole, debt is currently growing faster than money within the banking system. It also i think, provides a measure of proof, that the Central Banks really don’t understand what’s going on in the system these days. Because if they did, they would not publish statistics that obviously confuse money with debt, especially in a system where money is regulating the total quantity of debt, and vice versa. Of course, if they did understand that, then Banks wouldn’t be allowed to hold debt instruments in their equity capital holdings either.

M3

And then there’s M3. Which was somewhat notoriously, and to the great delight of the FRLF, discontinued by the Federal Reserve in 2006. If you want proof that many of the critics of the current banking system also have no idea how it actually works, threads on the Federal Reserves  discontinuance of M3 can be quite enlightening, if for no other reason than their blind assurance that it was a valid measure of the money supply.

But no, if you look at the largest, and towards its end, most rapidly growing components of M3, it’s Institutional Money Funds, followed by something called “RPS”. Institutional money funds are retail money funds writ large, and stuffed with those wonderful top tranche Asset Backed Securities. RPS are,  i believe i’m correct in saying, Repurchase agreements owned by the banks. Repo’s as they’re otherwise called are one of those interesting pieces of financial alchemy, whereby you sell a financial instrument to somebody, for cash, along with a guarantee that you will buy it back at a specified amount sometime in the future. It’s essentially a loan in return for an agreement to buy a loan.

Quite how you can get away with calling that part of any money supply measure, I have no idea.

© cc

cc Credit Crises, Empirical Analysis, Fractional Reserve System

The latest Bank of International Settlement Figures are out for Asset Backed Security Issuance for September 2009. The total amount of outstanding international securitized loans  increased by approximately $1 trillion in the 3 months between June and September 2009. Based on the  USA’s stock market performance this last three months, I wouldn’t expect there to be a drop in issuance this quarter either, especialy since the Federal Reserve Banks are still the buyer of last resort. That puts the world on track for a total increase in issuance this year of around $3 trillion. There are some rather interesting questions hanging around relative valuation of financial instruments given that the unit is american dollars, and the dollar hasn’t being doing well of late, but that doesn’t effect the impact for the USA.

The Federal Reserve  expects to be able to stop buying Mortgage Backed Securities next March. It will certainly be quite entertaining if they try.

The gross figures are quite interesting. They do show that up to this point, securitization is predominantly a problem of developed countries, although issuance is starting to increase in the developing countries. The first credit crisis period also stands out clearly, and stands out as a drop in securitization. So a naive approach to the problem, would be to simply worry about that drop, and find ways to “restart the securitization pipeline”, which is pretty much the policy that has been both advocated and followed over the last year.

Consequently, the Federal Reserve Bank of America, wittingly or unwittingly, has put itself in an interesting situation. They can’t stop buying Mortgage Backed Securities, as soon as they do, the lending pipeline collapses again, and we get Credit Crisis II. Like all sequels, bigger and much worse than the first release. On the other hand, how long can they continue? Especially as the world’s financiers then turn around and just sell more of them. Thereby increasing the total amount outstanding by $1 trillion a quarter.

Any increase in the outstanding amount of Asset Backed Securities, increases the ratio of debt in the system, to money. Money is ultimately what has to be available to pay debt, so as the proportion of debt within the monetary system increases, there is less money to pay more debt. It’s a complex relationship, especially when variable rate interest payments start to get factored in, and it’s fair to say, one that the world’s financial authorities clearly don’t understand very well, if they think increasing the amount of securitized loans is a way out of this.

Outstanding securitization actually peaked in June 2008 at $25,300 trillion, and didn’t return to that peak until June 2009 this year, at $25,900 trillion. We’re now at $26,900 trillion.  However, trying to predict the next crisis point isn’t as simple as looking at the outstanding figures, and going gulp. In the background, the movement of debt instruments into equity capital is also wreaking its own share of havoc.

This is total liabilities and assets for the US Commercial Banks. It is not as simple as liabilities equals deposits, and assets equals loans, since for one thing the equity capital reserve is listed under assets (hence the excess over liabilities.) But it’s close enough for government work. These figures show the other side of the credit crisis, the debt failure induced contraction in the money supply – total deposits, and also the recovery this year, as the securitisation pipeline was restarted.

The contraction in the money supply as debt was removed from the system, either by foreclosure, or by loan repayment – is pure monetary mechanics – it’s a side effect of the other problems in the economy and a consequence of the linkage between money and debt created by the fractional reserve banking process. However, it has its own repercussions. In particular, it worsens the imbalance between money and debt being created by securitization. So in some sense, it’s just as well that didn’t continue too long.

But the price of an at most half trillion dollar increase in the US money supply, appears to have been over $1 trillion of securitized loans. And the money to buy those loans , was provided as a direct claim on the US tax payer. So it would seem, that ignorant of the underlying causes of the problem, the Federal Reserve is failing, to run the red queen’s race, as it tries to stop the housing market in the USA completely collapsing  from a fisher debt deflation.

Quite how long they can continue to sustain the banks with interest payments on their central bank reserves, presumably being funded from the proportion of securitized loans that they are receiving capital and interest repayments on is a very interesting question. What they’re doing to the rest of the economy in the process, even more so.

© cc

cc Credit Crises, Empirical Analysis, Out of Real Time

In the textbook presentation of the Fractional Reserve Banking System, the example of the reserve redeposit process typically shows the expansion from the starting condition of an initial deposit. This is more than slightly misleading. The banking system that we have has been running more or less continuously, depending on which country you’re in, for several hundred years. The expansion from initial conditions can be presumed to be long past.

Theoretically then, we should be in a state where the money and loan supplies are fairly constant, or varying between known limits, and where the total commercial bank debt is 90% of the total money supply, assuming a 10% reserve requirement.

Clearly not in the world we’re living in. Especially the known limit part.

The explanation this blog is exploring, is that the primary reason for this is an interaction between the effects of  allowing loans to be sold in the form of Asset Backed Securities, and using equity capital rather than reserves to control total bank lending.

At the macro-economic level Asset Backed Securities have effectively increased the total amount of loans outstanding in the economy, and have done that within the part of the loan supply that comes from fractional reserve operations, rather than say somebody going out and buying a Corporate bond or government security. The end effect of this is that the ratio of commercial bank originated loans to deposits is no longer a fraction of total deposits, but rather a multiple. For example, in addition to the approximately $10 trillion commercial bank assets (loans to the rest of us)  held by the United State’s commercial banks, must be added the outstanding amount of Asset Backed Securities that have been sold on by the commercial banks. The net outstanding total for Asset Backed Securities for the USA as of June 2009, was $6,5 trillion. Back of the envelope, and assuming that most ABS is held as part of equity capital, then probably at least $5 trillion is being held outside of the commercial banking system, by pension funds, insurance companies, and other entities that need a store of money.

The immediate question this raises then, is how many of these things are there out there, and which currencies or countries are worst effected?

I’m personally convinced you can get a rough idea just by counting building cranes and uncompleted luxury condo blocks in any major city at the moment. Tracking down actual figures is a little tricky. Most Central Banks don’t seem to publish separate statistics on ABS sales, or if they do they’re not readily identifiable. However, the Bank of International Settlements does put out some high level aggregated data. Quite what their source for it is, I have no idea, but it seems to be the best there is at the moment.

The chart above shows the quarterly totals of newly issued debt securities from table 12a, which includes Asset Backed Securities and Pfandbriefe (covered bonds). The data is quite interesting. For one thing, it indicates that the 2008 credit crunch was primarily a failure in issuance of Asset Backed Securities. The drop, early in the year, suggests that at $1 trillion in capacity was removed from the system, which is a lot of loans. Even though the visible symptoms of the credit crunch were in areas like letters of credit, and revolving credit, it was very probably a knock on effect from the removal of Asset Backed security credit. The other takeaway from the chart is the subsequent, central bank and government financed recovery to 2006 levels. Somewhat akin to treating gangrene with a band aid, really.

Working out which are the worst afflicted countries is problematic. Whilst $6.5 trillion is a lot of debt, the USA is a lot of country, and a very powerful economy. Ireland on the other hand, is a rather soggy island on the left hand side of Europe, and they’re currently sitting on $546 billion of the things. With a population of 5 million, that’s just over $100,000 each.

An alternative to normalizing by population is to normalize by GDP. The problem there is that loan financed activity, building luxury condominiums for example, is counted as part of GDP. It’s the Exon Valdez disaster problem again – measured purely by its contribution to GDP, cleaning up an ecologically catastrophic oil spill was economically beneficial because of the extra work and spending it caused. Normalised by population Ireland comes out 4 times worse than the USA, normalised by GDP the USA is twice as bad as Ireland. On either measure, Greece, Iceland, Ireland, the Netherlands, Spain and the United Kingdom, and of course the USA, all stand out. No surprises there.

So where does this all end up? Asset backed securities have proliferated throughout the western developed economies, so almost all countries are involved to a greater or lesser degree. Norway for example, which is sitting on the proceeds from their share of North Sea Oil, has still managed to accumulate an outstanding, per head of population amount of around $35,000. The closest the banking system has ever come to generating this kind of situation before, was probably in 19th century America, a period known for rampant bank failure. There were assorted reasons for this, but a fair amount of it was caused by deliberate fraud, in the form of banks making loans that exceeded their deposits, typically to colluding cronies, sorry, ‘business associates’ of the men controlling the banks. It seems possible that the total loan supply might at some points have exceeded the total money supply in the US during this period, even with the gold standard, but who knows. The American financial system at that period was regarded as a thieves charter by British economists, and even by some Americans.

In our modern age, though it is strictly illegal for individual Banks to lend more than their deposits; Asset Backed Securities have unfortunately managed to allow the entire banking system to do it.

No collusion required.

© cc

cc Empirical Analysis, Lending

Here is a thought experiment. Assume that the money supply is completely constant, there are no loan supply induced fluctuations, the government is behaving itself, and there is simply a fixed quantity of money in the banking system. In a period of steadily rising production, due to technological and social change, what happens to prices?

They go down.

Deflation, to use the technical term, occurs because as more and more things, assets, food, granite kitchen tops, hit the market, there is less and less money available per thing, and the price drops. Or to put it another way, in a world with a constant money supply, if you inflate production, you deflate costs. It doesn’t really have anything to do with the social organisation of production, and it’s  highly debatable if it’s a good or a bad thing in terms of the effect on human behaviour. It’s just  math.

Over the last 20 years, there has been a lot more stuff available, at all levels of society. This is the age of the nail gun. The fully automated factory, and the completely unautomated Indian call centre.  There are more people, producing more things than there have ever been on the face of this planet, so the real question of the age should be, why hasn’t there been massive deflation -  rather than mild inflation, and raging asset inflation.

The simple answer, and the one born out by the central bank statistics is the money supply has been increasing. But that just raises the issue of why? And if we accept the idea that it’s not the central banks, and that is something that is also born out by the statistics, then the question is also who?

Equity capital is at the moment, the only real control on the part of the money supply within the commercial banking sector. To clarify, the chart below shows the difference between the ‘reserve’ of deposits that banks may be required to keep against deposits, and the equity capital that acts as a reserve against loan loss.

Under the Basel treaties, individual banks are required to keep equity capital reserves that are at least 10% of their loans.

Reading the Basel treaties, it’s apparent that their main concern is risk. Basel is all about reducing the risk of any individual bank defaulting, and to achieve that, it mandates equity capital reserve requirements, since it’s the equity capital that gets used when a bank suffers losses to ensure that depositors don’t lose their money. So if the only concern is what happens at individual banks, then it makes sense to  regulate that they have adequate equity capital reserves, and that those reserves are held in safe, secure financial instruments, that also provide the bank with some kind of return to compensate it for the ‘cost’ of maintaining a stack of otherwise useless money. (From the individual bank’s selfish point of view.)

Like mortgage backed securities for example.

41. Loans fully secured by mortgage on occupied residential property have a very low record of loss in most countries. The framework will recognise this by assigning a 50% weight to loans fully secured by mortgage on residential property which is rented or is (or is intended to be) occupied by the borrower.

(Clause 41 of the Basel Capital Accord)

Those were the days. Bet they rewrite that for Basel 3.

In fact, there’s a serious problem with allowing any kind of debt based financial instrument into equity capital, however tempting the associated income stream may be, and that is that as soon as you do, it allows the commercial banks to do the one thing that they’re not supposed to be able to do without central bank involvement – create money.

Consider 2 banks:

Bank A creates a $900 Mortgage Backed Security, and sells $800 of it to the depositor at Bank B for $1000. How much would it really be  worth? I’ve yet to be able to track down how much they sold these things for, but it’s a guaranteed (or at least it was before the loan insurers blew up), financial instrument paying anywhere between 5% and 15% a year for 5-25 years. So $1000 seems like a bit of a bargain. It doesn’t actually matter as long as its more than the underlying loans were made for.

Bank A takes the $100 tranche of the MBS it retained, and uses it to increase its equity capital. It counts it as 50% so it only increases its reserves to $150. It takes the $100 and pays that out as bonuses, which get deposited in its employee’s bank accounts, at its branches. One of the perks of being a bank employee is free banking from your employer, and usually reduced rate loans too.

Assuming the strictest case, i.e. that the deposits at Bank A are being held in net transaction accounts, then Bank A can now lend 90% of its deposits ($990), provided that this doesn’t exceed 10 times its equity capital. The new loan then gets deposited at Bank B.

The end result of all this sleight of loan, is that the total money supply has been increased by $90, and the  total loan supply has been increased by $900 MBS and $90 additional loan capacity in the banking system. This makes for a fairly slow leak as these things go, since it can take months or even years to issue the underlying loans. It’s worth noting that the central bank, who supposedly control the money supply, played  no part at all in any of this. It’s the commercial banking system that is out of control here.

So the money supply is increasing, but this isn’t effecting inflation nearly as much as it should. In fact its masking what would have been some fairly dramatic deflation over the last 20 years due to productivity improvements. Unfortunately however, the total loan supply is increasing much faster.

Which is where the real problem lies.

© cc

cc Asset Inflation, Credit Crises, Prologue

There is a lot of discussion on the net, good and very accurate discussion, about the reserve requirements, and the current almost absence of same. The Eric deCarbonnel post (hat tip Andre. G. and Mickanomics) has a nice summary of the situation in the US. There is one unfortunate problem though with these discussions -  fractional reserve requirements no longer have anything to do with controlling how much money a bank can lend.

What has replaced it, is the regulatory capital requirement. Regulatory or equity capital is a completely separate pile of money that the owners of the bank are required to provide when a bank is started, and maintain at a minimum ratio to their loan book, as the bank is run. The FDIC requires that this be at least $2 million when a bank is founded, plus the anticipated startup costs.  The Basel treaties then specify what percentage or leverage ratio the Bank is allowed to maintain between its equity capital and its loan book, and also what kind of financial instruments the Bank can hold the equity capital in.

There are two things then that control the immediate size of the loan book, of any bank in a Basel treaty conforming country. The amount of money it has on deposit (including savings accounts, and term deposits), and the size of its equity or regulatory capital. The fractional reserve requirement, i.e. the difference between the total amount of money it has on deposit, and the total money it has lent out, is relevant only insomuch as it is not allowed to lend out more money than it has on deposit. Just to ensure that complete confusion reigns, the recommended maximum total capital ratio, ie. the leverage ratio of equity capital to amount being loaned, is 10%. The same as the textbook fractional reserve requirement.

This is the 21st century. Weren’t we supposed to have moon bases by now?

It makes some sense, if you only think about the situation at the individual banks. If a bank makes a loan that doesn’t get paid back, and the loan loss can’t be covered out of profits, then equity capital is there, to in theory, prevent the depositors losing the money in their accounts. This is what is being referred to when there is talk about ‘recapitalizing’ the banks. Obviously too, given that this is all that stands between the depositors, national deposit insurance funds, and significant losses in the case of loan defaults, it’s critical that the equity capital be held in extremely safe forms of financial instruments.  This is why so much of the discussion and rules around equity capital are framed by the risk of default represented by the underlying securities, and the desirability of keeping that as low as possible. Solutions to the current crisis that involve increasing the equity capital are also being floated.

You can see the mental process path that’s being followed here. If banks take a hit to their equity capital, it reduces their loan capacity. If they take a hit that’s bigger than 10%, they have no equity capital left, and then their depositor’s money is endangered. Do everything possible to prevent that happening, the depositor’s money is now safe, and banking can go back to being boring again.

Unfortunately, system design doesn’t work well, when only the individual entities are considered, and not the interactions between them. Concentrating on the risk of individual bank losses, ignores the larger issues being created by a system whose core set of rules, embodies a relationship and an interdependence between loans and the money supply. Especially, if owing to “financial innovation”, some loan instruments are being allowed to masquerade as money.

The simple answer then, to the question  – what is the limit on commercial bank lending? -  is that for any given Bank, it is the  lesser of the total of their deposits (minus a very small fractional reserve requirement that applies only to Net Transaction accounts), or ten times their equity capital. It follows then, that the instantaneous limit for all the Banks, is simply the sum of their deposits, or equity capital allowance.

But that’s merely the instantaneous limit. It tells us  nothing about the way that the system will evolve over time, as loans are made, money is created, loans are repaid and money is destroyed, within the commercial banking sector. Especially within a larger system, that allows debt to be turned into financial instruments, and those instruments to be used as part of the bank’s equity capital.

Or to put it another way, what controls the total amount of equity capital in the system? The old reserve requirement, and its associated commitment to deposit a part of that reserve with the Central Bank, provided the central bank with direct control over the money and loan supplies.  What is the equivalent control over the total quantity of equity capital in the system, since it now effectively regulates the money supply? The quick answer, which will be explored in considerably more detail in the next post, is that there isn’t one.

The regulators were concentrating on local risk at the banks. They appear to have overlooked the systemic risks associated with removing control over the money supply.

The hardest problems to correctly diagnose in distributed systems almost invariably involve two or more concurrent bugs interfering with each other. In the banking system at the moment, we have the Asset Backed Security bug, allowing banks to create an arbitrary quantities of loans; and the Equity Capital snafu, which has essentially introduced a slow leak into the money supply. Next post, i’ll run through a detailed example of how they interwork with each other.

© cc

cc Credit Crises, Fractional Reserve System, Limits

Viewed from a computer scientist’s perhaps somewhat jaundiced perspective, the history of Fractional Reserve Banking and the many changes that have been made to it over the centuries is one of neither careful design, nor careless conspiracy  - careful conspirators typically don’t publish their meeting’s minutes online. Rather it is a history of kludges.

Every time the banking system has blown up, either in the form of some kind of speculative credit bubble, or through a collapse in the value of money through general inflation, the proximate cause has been identified and “fixed”.  Almost inexorably, or so it seems, thereby setting the larger economic system on track for the next problem.

One of the classic problems acknowledged in the textbook implementation of the Fractional Reserve Banking System is the inherent problem in relying on statistical multiplexing of a small reserve to satisfy deposit demands. What happens if a Bank hits a statistical outlier, and more than 10% of the bank’s depositors turn up on the same day asking for funds?  The other 90% of the bank’s deposits is represented by loans, and not immediately available. The probability of this happening depends on a number of factors, but two obvious, and historically relevant ones,  are the size of the bank, and its geographic coverage. The smaller the bank in terms of the number of its depositors and loan recipients, and the more locally based its depositor base, the more vulnerable it is to external events. For example, a bank that takes all its depositors money and makes five large  loans with it, is much more exposed if one of them defaults, than one that makes one hundred.

As a side note, this also means that the stability of a fractional reserve based banking system itself, depends in part on the distribution of monetary tokens within the community. A society where all the monetary tokens are controlled by a small number of people, is potentially far more vulnerable to banking crises than one where ownership of monetary tokens is more evenly distributed.

The pattern of a large number of small, locally based banks, was especially the case in the 18th and 19th centuries, when there a much higher incidence of bank runs than even today. Customer demand for money that exceeded the available reserves could just happen  randomly, because at the end of the quarter money had to be transferred from depositors to lenders at different banks. Ludwig von Mises also mentions this problem occurring in the Austrian Banking system in The Theory of Money and Credit.  It could happen because some event acted to co-ordinate borrower’s demand for funds – say a local crop failure requiring food to be purchased with savings, which might also put pressure on the bank from both directions, if the crop failures also caused local  farmers to default on their bank loans. It could even happen because the owners of the town’s other bank started a rumour, in order to get customers to switch their deposits.

From the perspective of the people running the banks though, this is regarded more as an annoyance than a real issue with the system itself. After all the depositor’s money is out there somewhere  in the form of a loan which will eventually be paid back. If the depositors would only wait, then they could receive their funds. Silly people for not understanding modern banking concepts and panicking. This is what comes of letting the common people have bank accounts. Etc.

The solution that was found in Great Britain in the 19th century was to introduce the concept of a central bank, in their case the Bank of England, as the lender of last resort. If a bank didn’t have enough funds to satisfy instantaneous demand, the Bank of England would provide funds to do so, and the Bank could then repay the Bank of England when its loans came due. Problem solved.

From 1866 onwards there were no financial crises in Britain at least for the rest of the 19th century. The banking system in the United States of America on the other hand, which at that time consisted of a large numbers  of very small banks,  saw repeated local bank panics, and occasionally some non-local ones with their associated major financial crises. It wasn’t until the early 20th century, that the United States copied the British system and setup the Federal Reserve Banks to act in a similar role as lenders of last resort in 1913. Following the wide spread bank failures during the Great Depression there,  a further support in the form of deposit insurance was introduced to the American system (and subsequently copied in Europe), and since then, there have been remarkably few bank failures, at least until the Savings and Loan crisis in the 1980’s.

From a distributed systems perspective, these are both good and valid solutions to the problem created by excessive demand for instantaneous funds.  Essentially what both these changes did was to distribute the problem represented by instantaneous demand for funds that exceeded the available local reserves of a small bank across the entire banking system.  While the available funds/loan balance of  single  bank could be badly de-stabilised by a single customer with a large deposit moving their account to another bank say, within the context of the entire banking system, this is a non-event.

Unfortunately, either lulled into a sense of complacency by the absence of bank panics, or perhaps because it was something that was never explicitly acknowledged,  the Central Banks appear to have forgotten the other purpose the individual bank’s reserve requirement played in the monetary system.

Providing a limit on the total quantity of loans made available by all the commercial banks.

© cc

cc Credit Crises, Fractional Reserve System, Lending

What is the limit on the amount the Commercial Banks are allowed to lend?

The story of credit crises is usually told from the demand side. The irrational exuberance of speculators down the ages drives inflated prices for over priced assets.  Inevitably the bubble bursts. Inevitably everybody pays the price. But the supply side of this formula always seems to get overlooked.  Why are the commercial banks able to supply a seemingly  limitless quantity of  loans to fuel the credit bubble? Why does a speculative bubble in Beanie Baby’s merely cause general embarrassment, while one in real estate cause massive economic dislocation?

This is a deceptively simple question, in the context of the larger fractional reserve banking system, since it is not just the quantity of the  loans that is potentially problematic. Through the fractional reserve process, commercial bank lending also has potential money supply implications. If there is no limit on the total quantity of commercial bank loans, then there is effectively no limit on the commercial banks’ ability to create money through the fractional reserve re-deposit process.

It is also notable that discussion of the problem of credit supply last year, concentrated on the supposed fear that the banks have of the risk in lending to each other in the inter-bank markets, rather than the possibility that they might simply not have had any money to lend.

Hypothetically then, the question can be framed in at least three ways. Either there is a limit on bank lending, in which case it might be reasonable to suppose that it is that limit that is preventing inter-bank lending, and not a sudden outbreak of pathological loan aversion in bankers; or there isn’t, in which case the money supply is being controlled by the commercial bank’s individual lending policies  rather than the central banks, which would be contrary to every foundational Economics textbook.  The third possibility  is that there is supposed to be a limit on lending, but some elements of the financial industry have found a way around it. However, given the entanglement that fractional reserve lending creates between the money and loan supplies, how exactly could this even be detected?

Going back to the previous post, a side effect of the fractional reserve banking system is that it creates two different kinds of debt within the economy. One is straightforward transfers in exchange for IOU’s of various kinds, either between individuals, businesses primarily in the form of bonds, and the government (treasuries). The other are the loans from the commercial banks, which represent a deposit somewhere in the banking system. For want of any better ideas, I’m going to channel physic’s particle/anti-particle structure and call them anti-deposits.  Commercial bank originated loans for long. Debt that results from a direct transfer of monetary tokens will be referred to as transfer debt.

If there is something going wrong uniquely within the fractional reserve banking system, then one expected outcome might be a change in the proportion of anti-deposits  to transfer loans within the general economy. If the amount of outstanding commercial bank loans is increasing faster than the quantity of transfer loans for example. Conversely, if the actual cause of the systemic problems that we’re currently living through is in fact government transfer debt, or increases in “fiat money” as I was confidently told by an Austrian economist back a few months, then we would also see a change, but it would be an increase in the amount of government originated transfer debt relative to commercial bank originated anti-deposits.

The quantity of outstanding debt within the United States economy is reported on a sector basis in the Z1 Flow of Funds report, Table D.3 The chart below shows the per sector information on Debt Outstanding.

From the explanatory notes in the current report. Foreign debt represents amounts borrowed by foreign financial and non-financial entitities in U.S markets only. Domestic financial sectors consist of government-sponsored enterprises, agency and GSE backed mortgage pools and private financial institutions.

The sectors aren’t totally clean with respect to being able to divide into transfer versus anti-deposit (bank originated debt). The domestic financial sector debt is for the most part being derived from bank loans in the forms of mortgages, since it includes Fannie Mae and Freddie Mac, who buy mortgages from U.S. banks and repackage them as bonds. Presumably also student loans, under the guise of Ginnie Mae.

Even so, it’s noticeable  from the chart above that bank originated debt is growing faster than other sources. Comparing proportionately the amounts owed by each sector in 1975 versus 2008, shows this quite clearly:

Proportionally, the last 30 years have seen a shift away from government debt towards commercial bank debt, primarily in the form of mortgages, since that is the main component of the Domestic Financial Sectors sector. They have also seen an extraordinary rise in the total amount of debt, as was shown in the original flow of funds chart, and it also appears that this new debt has predominantly originated from the commercial banks, chiefly in the form of mortgages.

In theory, the fractional reserve mechanisms don’t allow banks to issue arbitrary amounts of debt.  Originally they were  limited to a fraction of their deposits, now they are limited to a multiple of their equity capital. So what exactly has gone wrong that is allowing the commercial banks both to create so many loans, and to increase the money supply?

© cc

cc Credit Crises, Fractional Reserve System, Limits, Prologue

Central banks  are charged with regulating the money supply. Regulating not in the sense of rules, and laws, although they do form part of the framework it operates in. Regulating in the sense of controlling the quantity of. This is from the leaflet,  Modern Money Mechanics, published by the Federal Reserve Bank of Chicago in 1968, and last revised in June 1992:

From the standpoint of money creation, however, the essential point is that the reserves of banks are, for the most part, liabilities of the Federal Reserve Banks, and net changes in them are largely determined by actions of the Federal Reserve System. Thus, the Federal Reserve, through its ability to vary both the total volume of reserves and the required ratio of reserves to deposit liabilities, influences banks’ decisions with respect to their assets and deposits.

Bank’s assets are for the most part their loans.

The  commercial banking fractional reserve system introduces a rather strange money particle/loan anti-particle recursive relationship between some, but not all, of the loans in the monetary system.  This isn’t true for loans made to companies that borrow money through the bond market for example. They borrow money by selling an IOU, as does the Government when it borrows money using treasury certificates.  It’s only within the commercial banking system that money takes on this strange duality, and can expand and contract with the extension and default of loans.

What am i trying to say there. If a loan goes bad in the corporate bond market, then the company goes bankrupt, and my ownership of that bond gets me into the queue of creditors for a claim of assets and not much else. There are no money supply implications. The money i gave the company in return for a bond, is out there somewhere with whoever the company gave it to in return for goods, services, or a passport for tax free exile.  If a bank loan based on my bank deposit defaults however, I still have my deposit. But there is a deposit out there somewhere that should really disappear. That the current implementation doesn’t allow this to occur is one of the stresses on the system.

The bigger problem though, is that since there isn’t a fixed quantity of money within the commercial banking system, it depends after all on how much lending they’re allowed to do, it’s critically important to scrutinise what’s happening within the system quite carefully.  For example, if an expansion of credit is occurring, is it because the system is expanding within it’s allowed limits, or has it found a a way to circumvent them?

The Federal Reserve seems to believe that by controlling the money supply, it also controls the loan or credit supply.  It controls the money supply, by requiring the banks to hold a small portion of their total deposits either in physical bank notes, or on deposit with the federal reserve bank. Let’s look at money supply regulation first.

Leaving aside the banking system’s day to day liquidity issues in terms of satisfying customer’s demands for access to the funds in their deposits, it is arguably the case that it doesn’t matter how much the banks keep in reserve, or how much of that is on deposit at the central bank, as long as it is a fixed portion of the local bank’s reserves.

As long as some known fraction of the money represented in deposits is required to be held at the Federal Reserve (or in bank notes and coins), then the Federal Reserve can calculate what the total money supply is simply by multiplying. It can set what level it thinks is appropriate for that total, and adjust the reserve amount appropriately, or it can inject money to increase it, again by the known percentage that is anchored by the reserve. The only difference from that simplistic perspective, between a 1% and 10% requirement is how much you have to multiply by to get the total money supply. This is the textbook presentation of the reserve based system.

The statistics on commercial bank reserves, and the Monetary Base for the United States of America, are reported in Report H3 by the Federal Reserve. The Monetary Base is the simplest statistic on the quantity of money in the system, and is defined as the total  of commercial bank reserves held at the Federal Reserve, plus the total of printed/coined currency. To wit:

As you can see, things start to get a little funky on the right hand side there when the Federal Reserve starts trying to resolve the 2008 credit crisis. Let’s look at things before that happens though. This graph is showing the totals for all physical currency and the reserves. It is the reserves that are supposed to be regulating the money supply, “through its ability to vary both the total volume of reserves and the required ratio of reserves to deposit liabilities”. So all other quantities of money, within the fractional reserve system, should be derivable as some multiple of the reserve. (This incidentally, is i suspect where the Rothbard fallacy may originate. Individual banks lend a fraction of their deposits, however the federal reserve regulates their deposits, as a multiple of the reserve.)

Which unfortunately is not what is being shown in the chart above. Then again, what relationship should the total quantity of physical currency have, to the bank reserves? Physical currency isn’t controlled any more, in terms of a fixed relationship to reserves, rather it’s printed as needed. Even so, it’s reasonable to expect that that need would in some sense reflect a somewhat consistent percentage of the total quantity of money – which is essentially physical currency plus the sum total of all bank accounts. So the chart seems to show that the federal reserve is successfully keeping bank’s reserves constant, but either there has been a consistently growing demand for physical notes over bank accounts in the last four decades, or something else is going on.

The consistently growing demand for physical banknotes theory isn’t entirely discountable as it happens. The US dollar is accepted world wide, and a significant amount of physical cash is held outside the USA. That can be deduced simply from the large amounts that were found during and after the invasion of Iraq.

Digging a little deeper. M1 is the base measure of the actual money supply, and includes physical currency, demand deposits and other checkable deposits. Demand deposits are all bank deposits which allow immediate access, rather than savings accounts and similar, which require deposits to remain for a specified period of time. Looking at the historical data on the components of M1, from the Federal Reserve’s H6. Table 4, and please note the y-axis on this chart is in $ billions, whereas for the other two it is $ millions:

Interestingly, demand and checkable deposits do look like they are holding approximately stable in the last couple of decades at least. In fact they’re holding stable at roughly 10 times the reserves held by the central bank. So, the reserve requirement does appear to be successfully regulating the quantity of money held in demand and chequeing accounts. These in other words appear to be the Net Transaction Accounts, for which the banks are required to keep 10% either on deposit with the federal reserve, or present as physical cash in the bank vaults. So why then, is physical cash going up so much? After all, if the Federal Reserve is actually printing physical cash to increase the Money Supply, then the deposits wouldn’t be staying relatively constant, but would in fact be increasing to match.

Or in other words. Is physical cash being printed in order to increase the money supply, or is physical cash being printed because the money supply is increasing?  One last chart.

This chart shows the Total Reserve, the Monetary Base, M1, and the Total Liabilities of the US Commercial Banks from the H8 data series, table b1152a. Total Liabilities, is the sum of all the money deposited at the Banks by their customers, plus loans from other Banks. That table starts in 1973, and yes those are the right units.

What is the money supply? Is it just the checking accounts most people use for day to day cash, the Net Transaction accounts? Does it include savings as well? Physical cash in Iraq and elsewhere? There are economic arguments for and against counting each of those in fact, that’s why there are the different measures, Monetary base, M1, M2 and before it was discontinued M3.

If the money supply is just the amount held in Net Transaction Accounts, then the Federal Reserve is regulating it successfully. But that’s the only thing that is being regulated. Physical cash is clearly expanding, but that’s nothing besides the total amount of cash represented by deposits with commercial banks. The electronic money supply, the money that is represented as electronic 0’s and 1’s in a computer database somewhere is clearly growing exponentially. And has been for some time.

One definition of the money supply that isn’t explitly reported, but would seem to be fairly important in a fractional reserve system, would be the amount of money that the banks can lend a fraction of. As you might guess from the above chart, it certainly isn’t the amount held in Net Transaction Accounts.

So just as this post attempted to dissect the deposit side of the fractional reserve money/loan dichotomy in the context of the larger system, next time, we’ll look at the loan supply.

©cc

cc Empirical Analysis, Prologue

The Reserve Requirement and the Equity Capital requirements are two very different things, which often, and not at all surprisingly, get confused.    The best explanation I’ve found so far for the reserve requirement for the US banks is provided by Federal Reserve Board’s Reserve requirements page. It is fairly clear from that description there that the reserve requirement is the difference between the amount that the bank has on deposit and the amount it can lend, and is required so that the Banks can handle day to day demands for cash and transfers.

The reserve requirement today is 10% for Net Transaction Accounts above $44 million in deposits, 3% below that. A net transaction account is broadly a current account. There is no longer a reserve requirement for time deposits, which are presumably savings accounts with access restrictions.This amount of money either has to be present in the Bank’s vaults, as physical cash, or on deposit at the Federal Reserve, as a reserve.

Requiring the Banks to have some amount of physical cash, and/or a fixed percentage deposit of their total deposits at the Federal Reserve, is a way of synchronising the entire system. From a distributed systems perspective, this can be very important, if you want agreement between independent agents on say, a fixed ceiling on the amount of money or credit that the entire system can supply. There’s a nice distributed systems proof called the Fisher consensus problem, which shows that it is impossible to guarantee agreement without some element of synchronisation (which generally means a central point of some kind) between the distributed members of a system. However, it’s not entirely clear that the Federal Reserve understands it this way.

There is a very interesting paper from the Federal Reserve by Joshua N. Feinman, called Reserve Requirements: History, Current Practice, and Potential Reform. The paper reviews current and past reserve requirements and the reasons for them, at least according to the Federal Reserve.

Perhaps the most interesting thing about the paper is what it doesn’t say. At no point does Feinman discuss reserves as an integral part of systemic regulation of the loan and money supplies. In the main, the paper is about avoiding undue quantity of reserves – because of the cost to the banks of maintaining them (cost of deposits, with no loan income to match), and providing the necessary liquidity to meet day to day requirements. Which it seems, the Federal Reserve believes it has solved by being the lender of last resort, and providing shortfall funds to the banks whenever they need them.

A Bank’s equity capital is something quite different. This is the capital that was used to found the bank – it is a completely separate pile of funds to the deposits that are held by the bank. Nominally, this is the money the founders had to provide in order to be allowed to setup a bank in the first place. It is also referred to as Tier 1, 2 or 3 capital following the Basel accords. Equity capital is also generally what is meant when the problem of recapitalizing the banks is referred to.

When Banks take a loss on their loans, they first cover the loss from profits. In the absence of profits, they use their equity capital. That in and of itself can create problems, given that money then has to be brought in to replace the equity capital, which nominally at least reduces the amount on deposit, and hence the amount that can be lent.

Which returns to the problem of controlling the loan or credit supply for the whole banking system – i.e. how much in total are the Banks allowed to lend – and what in fact controls it? The reserve requirement – as implied by Feinman, or the amount of equity capital held by the banks, as implied by the Tier 1 and 2 capital ratio’s reported in every bank’s call report?

Who really controls the loan supply? The federal reserve? Or the commercial banks?

©cc

cc Fractional Reserve System, Prologue

All Banks licensed to operate within the United States of America have to file detailed quarterly reports on their lending and deposit status with the Federal Financial Institutions Examinations Council (FFIEC). These are known as banking Call Reports, and since 2001, are available for download, and public examination.

If there’s any kind of guide to reading and understanding them, I’ve yet to find it. Still, it’s amazing what can be done with python scripts, and a little patience.

There are two ways to get call reports – either as a consolidated set of files for all the banks per quarter, or individually per bank. It’s easier to understand call reports by looking at the individual consolidated ones, but for systemic analysis, the consolidated files can’t be beaten. It’s also useful to know that in the consolidated files, the Bulk POR report has the names, addresses and FDIC/IDRSSD numbers for all the Banks, which can otherwise be quite hard to find.

So, taking the Citibank FDIC #27606 call report for December 2001, we can  find the ratio of deposits, reserve and loans – at least, we can with a little decoding.

Deposits are known as liabilities in the alternative universe of Bank accounting. Total liabilities, are RCON2948; Total Equity Capital is RCON3210, and Total loans and leases are RCON2170. The forms aren’t completely consistent. For Wachovia, the RCON3210 field is missing, but RIAD3210 is listed as Total Equity Capital, Total Liabilities come under RCFD3300, Total loans are still under RCON2170 though. So in tabular form:

Then there is the question, which Citibank? For the big banks, there often seem to be several separate listings. There is also Citibank Delaware, Citibank New York State, Citibank Nevada, Citibank (South Dakota), and a couple of others.27606 is Citibank USA, using the form for a Bank with domestic offices only.

Another interesting question, in which field are the reserves? Equity capital is normally used to refer to the capital invested in a firm by its owners. So this would presumably be the capital reserve, which is the money used to found the Bank. It also matches the Tier 1 Capital entry, RCON8274.  Presumably then, the reserve of deposits referred to in textbooks, is not listed separately, but is simply the excess of deposits over loans. For which, for Citibank at least in December 2001, would appear to be  -$109,040,000. Interesting.

Actually, it gets worse further down the form. In RC-R the Regulatory Capital section, where Total Risk Weighted Assets are $2,470,549,000 (RCONA223).

Wachovia’s numbers also seems a little strange, since and at least in 2001, they were somewhat over capitalised. However, they seem to be reporting some kind of change to Equity Capital of $6,819,394,000 and it looks like they spent 2001 merging with First Union, which might explain it.Their Total Assets (loans) really are exactly equal to their Total Liabilities(deposits). I’m afraid most engineers, almost innately, tend to find exact matches like that suspicious as hell.

So from the empirical evidence, it seems that Banks can lend out as much as they have on deposit, and the Equity Capital reserve is in fact, the shareholders capital investment into the Bank, and is completely separate to the money deposited by the customers. Banks are required by the Basel treaties to maintain a minimum leverage ratio between theri Equity Capital and their Loans, which Citibank is within the limits of. (RCON7204-6). Presumably this explains why Citibank can get away with lending more money than it has on deposit, it’s still within its ratio with respect to its equity capital.

All the same, it seems a little strange. So, pausing only to drop this lot into a nice little sqlite database, next time I think we’ll look at the figures for some of the banks that have failed this year, versus some hopefully healthy ones, and also look at the situation across the eight years of data available.

One thing can be said though. The Murray Rothbard claim, that Banks can lend ten times their deposits, is shall we say, not supported by the available data.

©cc

cc Empirical Analysis, Fractional Reserve System, Prologue