Wednesday, 9 October 2013

Do Banks Have to Compete for Funds with Non-Banks?

In a recent blog discussion, I made the comment that banks have to compete for funds with non-bank financial institutions (NBFIs).  Somebody questioned this by pointing to the fact that when investors transfer funds to NBFIs, this doesn't actually remove funding from banks, it merely transfers bank deposits around.  I think the person who made the comment understands the distinction but it does highlight a common confusion, so I thought I'd say something on it.

First of all, we can note that in a very simple model, banks create their own funding and do not need to worry about having to compete with anyone else.  In this model, banks have only one liability - bank deposits.  There is also no cash, so funds cannot be withdrawn.  The only way that bank deposits can change is therefore if their loans or investments change.  Banks compete with other banks of course, but in the simple model we can ignore this.  In this world therefore, banks do not need to worry about the liability side of their balance sheet - it looks after itself.  Once deposits are created by lending, they are always there.

If households wanted to switch out of bank deposits into NBFI investments, they would transfer funds to the NBFIs, but this wouldn't reduce the overall amount of bank deposits.  It simply transfers them from one person to another.

In the real world, banks have complex balance sheets.  In particular, their assets are funded by a range of liabilities.  These included transactions accounts, term deposits, wholesale deposits and repo, and capital market instruments.  They also have equity.  The balance between these different types of funding is important.  Whether for regulatory or business reasons, banks need to manage the mix.

Unlike in the simple mode, this means that banks cannot just forget about the liability side of their balance sheet.

Again, I want to ignore the fact that banks compete with each other and look at the impact of NBFIs.  Let's imagine, for some reason, that NBFIs start to offer higher returns on certain types of savings bonds (perhaps they have found a way to get round regulatory restrictions on holding certain types of higher yield assets).  Investors now might decide to run down their term deposits with banks and place more money into NBFIs.

This does not of course reduce the overall amount of bank funding.  However, it does involve a change in the composition of bank balance sheets away from longer term funding and into demand deposits.  This matters to banks and if the switch is material and persistent, banks will be forced into raising the rates on their term deposits, in order to attract in more term money.

Therefore, even though the totality of bank liabilities is determined primarily by what happens on the asset side of their balance sheet, banks do not avoid having to compete for funds with NBFIs and other borrowers.

For more on this, see my earlier post: Banks, Non-Banks and the Interest Rate Effect


  1. Nick,

    "Somebody questioned this by pointing to the fact that when investors transfer funds to NBFIs, this doesn't actually remove funding from banks, it merely transfers bank deposits around."

    As I showed in my blog some posts back, the amount of deposits can also get reduced if NBFIs become more competitive and in the process induce borrowers to borrow from them rather than the bank. Borrowers would reflux deposits when loans are due and banks' balance sheet reduces and that of non-banks' increases. So inducing borrowers to bank with NBFIs is also important. It is necessary for banks as a whole to keep their business with them. So without going into term deposits and other types of funding one can show deposits are reduced.

    1. Yes. I'd agree that would happen. There are all sorts of things that happen in a more realistic picture. Although the simple model is useful in illustrating an important point, it can be quite misleading. I know you and JKH have stressed this point before and, of course, it's key to what Tobin is saying.

    2. Hi Nick, I was also curious about your thoughts on reflux and NBFI lending. One of your comments on JKH's Tobin thread was you didn't think reflux was significant. Yesterday, I was looking at the large numbers from Barclays via FTA ( on the NBFI lending ($5 to $12T depending on definition of "moneyness" or funding duration for dollar claims). If we assume that propensity to borrow is fixed, then at least this amount has refluxed back to banks. Is this the right way to think about it or would a more detailed accounting show this to be too simplistic?

    3. Hi,

      I'm not sure what I said on reflux in the comment you're referring to. This sometimes depends on context. I see reflux as the process by which the non-bank sector gets rid of an excess supply of money by paying down loans. This is the process which I believe Kaldor emphasised as a reason for seeing money as demand determined. My view though is that this process plays a relatively small part in matching demand and supply of money. This is based on the observation that usually the people with money are not the people with debts. I tend to think that more of the adjustment happens through changes in the type of assets people hold - switching money for non-monetary claims on banks - and I think the actual demand for money adjusts as well.

      I do think that reflux occurs however. The process I described in my August post on Repo Volumes and Asset Values (see penultimate paragraph) can be seen as debt repayment to get rid of surplus cash. And I'm certainly not suggesting that loans do not get repaid, nor that overall leverage does not get reduced - rather that I don't necessarily see that happening due to an excess supply of money.

      It does talk in the FTA article about the fact that the large exposures recorded within the shadow banking sector reflect the lengthy collateral chains. Shortening of the chains due to dis-intermediation would reduce recorded loan volume with little noticeable effect on end borrowers. This process wouldn't normally be thought of as reflux any more than a reduction in inter-bank indebtedness would.

      As the article notes, it is difficult to tell how much end borrower borrowing is involved. However, even if some end debt is being repaid, it shouldn't necessarily be interpreted as being driven by an excess supply of money.

      Thanks for the reference to the article though - I'd missed that and there's a few interesting bits in it. Might almost be worth another post, but I'd probably need to look into the figures a bit more.

    4. I actually think Ramanan's accounting example may be a bit more on point, but I think it could use a bit more explanation. As he said, NBFI's steal business from banks by inducing people to borrow with them instead of banks. However, just focusing on that statement alone, one could think (as I initially did) that what's relevant here is solely the interest rate NBFI's offer on loans versus bank loans - the deposit rate has nothing to do with it. This is true in a sense: the decision to borrow from a bank vs. NBFI will mostly just involve a comparison of the interest rates on the two loan alternatives - the deposit rate doesn't come into play much at all. However, the deposit rate is relevant in an indirect but critical way: the only way NBFIs can offer loans is to first attract deposits, and to do this, they need to offer deposit interest rates that are competitive with banks'. (NBFIs could also borrow deposits from banks, but it's cheaper to attract the deposits from investors.)

      Nick, the implication of your example isn't clear to me. It seems like you're still talking at the aggregate bank system level, and comparing that system to the aggregate NBFI system (since you say to ignore competition among banks). As such, if we continue to assume that deposits don't leave the aggregate banking system level, then term vs. demand vs. other maturity funding types doesn't seem relevant. In other words, the banking system offers higher interest rates to retain funding. But at the system-level in your example, there is no concern of retaining funding. Bank liabilities simply shift between different entities. If I am not misunderstanding, then maybe your example is more relevant at the individual bank level?

      The real problem the banking system as a whole faces seems to be highlighted in Ramanan's example. There are 3 key ingredients:

      1) NBFIs acquire bank deposits by offering the private sector a better asset to hold
      2) The destruction of bank deposits via repayment of loans
      3) The replacement of those bank liabilities with the liabilities of competing financial institutions, accomplished by, for example, NBFIs offering better interest rates on loans than banks

    5. I would agree that the things you've highlighted are relevant to the overall picture of how banks and NBFIs interact. I didn't talk about these aspects because I wanted to focus on one specific issue. This is to do with the idea that total bank liabilities are determined solely by the asset side of bank balance sheets (loans and other things) and therefore that banks (as a whole) do not need to compete with NBFIs on the liability side. To look at this I needed to ignore the impact of bank loan repayment and the impact of competition between banks. That is not to say thought that I think those aspects do not matter to the wider picture. And maybe it's a bit artificial to consider the one aspect in isolation from the others.

      The particular point I was interested in looking at was that even if the totality of bank liabilities is fixed by the asset side, the mix of liabilities is not. Therefore, even if banking was a monopoly and even if it did not suffer from loss of lending business, it could still not prevent its liability holders from shortening the tenor of their holdings without being forced to raise savings rates.

      The issue of NBFIs taking loan business from banks is an interesting one. If NBFIs expand their lending, two things can happen. First, borrowers might switch from banks to NBFIs. Normal demand and supply analysis would tell us that we would expect to see an increase in the overall number of loans and a decrease in lending rates, with the mix depending on the relevant elasticities. However, in practice what we also see with expansionary lending is a relaxing of credit standards. This leads to an increase in loan volume and an increase in (average) lending rates.

      I think this second effect is quite an important one. It seems to me that periods of significant expansion in lending by either banks or NBFIs is often accompanied, or even driven by, a relaxation of lending standards. That is why I tend to think that whilst growth in NBFI lending might lead to some reduction in bank lending, the effect in general is limited.

    6. Hi Nick, thanks for your reply. Also to your comment @14 October 2013 02:20. Agree. Over the last few years, growth and contraction of bank lending and shadow bank lending have tended to move together; often characterized as the securitizaion effect. Also, this thought experiment: what would happen if a NBFI's ABCP became untradable and thus unredeemable (it happened in Canada and the US) ... well obviously deposits didn't increase again in the banking sector. To what extent promoters/investors know that they are really contingent equity rather than depositors (replacing another set of depositors)?; let's call this the equity effect. So reflux ("competitive lending") may be a much smaller effect than the "securitization" and "equity" effect ("co-opetitive lending").

    7. Just one point on ABCP. A seizing up in the market for ABCP shouldn't in itself prevent outstanding CP being redeemed, as these facilities should all have backstop liquidity facilities which would fund redemption provided the underlying assets were not defaulted. I think actually in many cases banks bought the underlying assets back onto their own balance sheets in a process which would have technically increased deposits.

      However, I agree on the general point that there is more of a tendency for bank lending and shadow bank lending to move together than in opposite directions.

    8. Yes, that's right for the US, but of course the Fed helped enormously by providing general backstops. Canadian ABCP had very weak liquidity provisions and regulatory regimes:;
      Similarly, for auction rate securities in the US:;

    9. Hi Nick, I don't disagree that the composition of banks' liabilities is irrelevant. But I guess I need help to see why your example illustrates this. You wrote:

      "even if it did not suffer from loss of lending business, it could still not prevent its liability holders from shortening the tenor of their holdings without being forced to raise savings rates."

      I guess what I am saying is that if the banking system has no threat of losing their original "funding" (which is what we're calling the creation of deposits in the first instance), then the tenor of their liabilities seems somewhat irrelevant or artificial. Banks would have no need to offer shorter or longer term liabilities in the first place, since they know they would not be at risk of losing the deposits, as they'll will end up right back with them. Any liabilities they create are in a sense long-term funding, at least for the life of any corresponding loans. Do you see what I'm saying, or do I have this incorrect?

      Assuming the former, then I think you need to introduce a threat to banks' loan business. Just introducing the mere presence of a so-called NBFI isn't enough, since again they'll end up holding the deposits and the banks maintain 'long-term' funding. However, if you also stipulate that the NBFI has the potential to steal banks' loan business, then the bank realizes 'hey, people are going to move their deposits to NBFIs, which is going to enable the NBFI to take away some of my business. to prevent this, i need to lock in my deposits, by offering a higher interest rate, maybe at a longer tenor.'

    10. Sorry, meant 'relevant' in the first sentence.

    11. ATR,

      I think I see what you're saying, which is that if the totality of liabilities is fixed anyway, it doesn't matter if some of those are notionally demand deposits as opposed to time deposits, because at the end of the day they are all perpetual in effect.

      That's an interesting way of looking at it which I hadn't thought of. I don't think it invalidates my general point, but it does perhaps call into question how useful it is to imagine a monopoly bank. Perhaps a monopoly bank could take the view that its demand deposits are really undated debt.

      There are perhaps a couple of reasons why we still might want to make the distinction though. First, aside from any commercial considerations, there are regulatory requirements on banks regarding their liability mix. Secondly, there is always the option for non-banks to demand cash from the banks. Obviously the more liquid its liabilities, the greater the risk of this.

      But maybe again this just suggests that the assumption of a monopoly bank is an unhelpful one.

    12. Right - you put it more succinctly than I did, and I'm with you on the rest of your reply. If anything, your point about cash demand is also unambiguously correct. Demand for cash is an ever-present threat to the banking system as a whole, which requires banks to find alternate sources of funding.

      On that point, I think we could take a step back and frame it as - what are the various threats that could force the banking systems' composition of liabilities to change? One is very clearly cash demand. It seems another is loan competition from NBFIs. Another could be regulation, as you say.

      There's an interesting distinction between cash demand and loan competition, however. With a withdrawal of cash, aggregate bank liabilities stays the same in quantity, but the mix is different (x% deposits and y% borrowings from some source). In contrast, if NBFIs steal loan business, as in Ramanan's example, then aggregate bank liabilities change in quantity, but the mix stays the same (still all deposits). I'm not sure if this distinction should mean something...

    13. Given the particular point I wanted to look at here, I'd perhaps prefer to restate it as: can actions by non-banks create funding pressures for banks? I think the answer to this is yes, but it manifests itself as pressure on the amount of term debt, not overall debt. This still allows for the possibility that there are other things going on as well, including competition between banks, which is obviously an important factor as well.

      I'm not sure I understand the point in your last paragraph. A withdrawal of cash does result in a reduction in liabilities - deposits go down by the amount of cash withdrawn.

    14. Agree with restatement.

      To illustrate what I was saying, suppose the banking system has 100 loans and 100 deposits. Suppose the 100 deposits are withdrawn as cash. To meet this obligation, the bank has to borrow 100 cash from the Fed, so their new balance sheet will look like 100 loans and 100 Fed borrowings. Liabilities stay the same. Or alternatively, if they borrow in the interbank market, they'll have 100 fed funds borrowing as their liability.

    15. OK - I see what you're saying. I think the assumption usually goes that cash withdrawal is covered out of liquid assets - that's the basis for the new Liquidity Coverage Ratio, for example - but of course banks will go for whatever mix of borrowing or asset reduction is considered best. Also, obviously, interbank borrowing won't cover a net withdrawal of cash from the sector - ultimately someone has to cover that from running down assets or borrowing from outside the sector.

  2. An even simpler way of looking at it is to argue that a money market fund (for example) is just a bank that isn't regulated as such. When "deposits" migrate to money market funds, they remain in the extended banking system. Instead of using the interbank market for funding transfers, the banks need to use other instruments (repo, emission of commercial paper) to maintain the circular flows you see in a pure banking financial system.

    If you are modelling the real economy, the distinction between the real banking system and shadow banking is somewhat artificial.

    1. I totally agree with that Brian.

      I didn't specify money market funds as my NBFIs, even though they would have been an obvious choice, because I wanted to avoid the issue of whose liabilities count as money. This varies between jurisdictions with MMFs treated as broad money creators in the US and Euro zone, but not in the UK. But the whole question of whether they should be inside or outside the money boundary only reinforces the point that the distinction is artificial.

  3. In my thinking, we can distinguish between banks and NBFIs with two characteristics:

    1. Does the NBFI have a deposit account at a bank for customer funds? If so, then it is just another individual or agent-acting-for-an-individual. The NBFI would never increase or decrease bank deposits by any action it took.

    2. Does the NBFI offer unlimited withdrawals of deposits upon customer demand? If the NBFI offers this privilege, then it is offering the same service as a bank, allowing both customer and loan borrower instant access to the identical backing funds. The NBFI making this offer is really a "bank".

    Are there additional distinguishing characteristics separating banks and NBFIs?

    Maybe these two characteristics are not the same distinguishing characteristics that you are thinking of.

    1. Roger,

      The key distinction between banks and NBFIs that I am looking at here is that I'm taking banks to mean those institutions that provide transaction accounts. That means that non-banks can make payments between themselves by transferring transaction account balances. My NBFIs are those that do not provide such accounts. This means that if a non-FI wants to reduce its investment in a NBFI, this must involve withdrawing some other asset from the NBFI; it cannot be done by transferring the claim on the NBFI to someone else.

      This is not necessarily the same as the normal definition of a bank, but it's needed to illustrate the point I wanted to make.

  4. Nick,

    Thanks for the reply.

    Ramanan, in his comment, makes the point that lending by NBFIs can reduce bank deposits. I think that is correct in the following way:

    Assume that banks became permanently non-competitive vs NBFI lending (perhaps due to government regulation). The banks would no longer make any loans.

    Next, assume that all previous bank loans are repaid. Repayment of bank loans is a draw-down of the money supply to the extent that deposits have been created by loans. Bank deposits will therefore be reduced by loan repayments.

    Over time, as all bank loans are repaid, the entire money supply attributable to lending is reduced to zero. The only money supply left would be the money supply provided by government or based on other mechanisms.

    Banks would be reduced to deposit transferring facilities without a loan department capability. Money supply expansion due to bank loans would be an historical curiosity.

    Surprisingly, this may be happening in the United States due to the new bank rules. I don't yet see loan draw-down in the Fed series TOTLL but a definite leveling is in progress.

  5. ATR,

    Got held up with a few things so couldn't reply earlier.

    Yeah your 13 October 2013 15:30 summary of me looks fine.

    I'll probably add a bit more.

    Even a bank is constrained like an NBFI in the sense that it has to attract deposits. So imagine a bank which can find itself with plenty of lending opportunities but is restricted by the amount of deposits it has raise. It can try other methods of financing as well but these things cannot rise relative to other things without putting the bank under risk. The fact that loans create deposits is although a bit useful, a bank cannot assume that the deposits remain with it. If someone keeps at at the bank it already means the bank has kept the person or the institution attracted enough to bank with it.

    However at a macro level, this is not an issue really - in the case we assume no NBFI and only banks. Even here individual banks have no Widow's Cruse.

    If we include banks are in competition with NBFIs, the interest on deposits both matter and not matter to put it in an oversimplified way. I think the best way is to write a model for competition.

    I guess since people wish to hold money more than NBFI liabilities gives banks some edge.

    I think the best way to say this is to write a model of competition - both price competitiveness and nonprice competitiveness but I guess it becomes too messy.

    1. So it's not like NBFIs have to pay more interest. Imagine a situation in which banks are napping and not being competitive. NBFIs can then take advantage of this and offer more services and become popular. People may not getting less interest simply because NBFIs offer better services.

      So banks have to become more competitive and in a non-hypothetical situation unlike the above - competing both in price and services.

    2. “…but is restricted by the amount of deposits it has raise. It can try other methods of financing as well but these things cannot rise relative to other things without putting the bank under risk.”

      I believe I agree with this, but let me try to restate it and see if you agree. I interpret you as saying that a bank will decide to take on lending opportunities based largely upon risk-return profile. Part of that calculation is determining the funding cost of the loan. It can compare the cost of funding with deposits versus funding through other means, such as the interbank market or the central bank’s lending facility. This is the relevance of deposits, and the only relevance. It’s one form of funding among many. If the risk-return profile of the loans are only satisfactory if deposits are used but not with the other sources, then the bank is surely restricted in taking on these lending opportunities if it cannot attract and retain enough deposits.

      Continuing to assume away NBFI competition, what if the loans are still profitable even if they’re funded in the interbank market? In that case, could we say deposits aren’t as much of an issue, since banks can fund themselves in the interbank market instead? I’d argue against myself and say no, because other banks could offer higher deposits rates, but still lower than the interbank rate, and thereby offer lower interest rates on the loan, stealing loan business away from the other bank… (Example: assume the interbank rate is 1%, and a 1% spread above this is enough to make a given loan attractive for Bank A. So Bank A is willing to make the loan at 2% assuming it funds in the interbank market. Bank A also offers .2% on deposits. However Bank B offers .5% on deposits, and therefore everyone takes their deposits to Bank B. This means Bank B can fund itself at .5%. Assuming a 1% spread above funding costs is enough for Bank B, Bank B can therefore offer 1.5% on the loan. As such, Bank B steals the loan business from Bank A)

      “If we include banks are in competition with NBFIs, the interest on deposits both matter and not matter to put it in an oversimplified way. I think the best way is to write a model for competition…I think the best way to say this is to write a model of competition - both price competitiveness and nonprice competitiveness but I guess it becomes too messy.”

      I agree services should be included, but if we take that out for simplicity, I think my summary of your post captures the reason why pretty well. Basically NBFIs can attract deposits away from banks, if banks don’t offer attractive enough returns on deposits, and then use those to lend at better rates than banks offer.

    3. ATR,

      A few points worth making here.

      It is useful to analyse the lending and borrowing activities separately using an internal reference cost of funds. The interbank rate is a good starting point. This would mean, using your numbers, that taking deposits at 0.2% generates a profit of 0.8% and lending at 2.0% generates a profit of 1.0%.

      If Bank B decides it is worthwhile to take deposits at 0.5% (perhaps it has a developed a neat way to cut its admin costs on retail deposit taking), it won't necessarily want to cut its lending rates, as it might make more sense to get rid of the excess funds on the interbank market.

      If the interbank market was seen as completely risk free for bank lenders and a bottomless source of funds for bank borrowers, then internal reference funding costs would always be at the interbank rate. In that case, any developments in the deposit-taking business should have no impact on rate at which a bank is prepared to lend nor the amount it wishes to lend.

      In practice, of course, it is more complicated and in particular no bank would want to rely on the inter-bank market for core funding. Rates paid on other funding sources do therefore matter.

    4. This comment has been removed by the author.

    5. ATR,

      I'd argue that non-price factors are far more important that price considerations. (Price as in interest rate here).

      That of course doesn't mean price is not important but just that the picture is far more complicated.

      In practice price seems more important because it is under the assumption that financial institutions are highly competitive with each other. Nonetheless that means they have to be competitive and hence don't possess a Widow's Cruse.

      Tobin's analysis errs because he uses some neoclassical analysis of what determines lending at the macro level. However in reality, between banks and NBFIs there is competition but for them together the volume of lending is determined typically by creditworthy demand (except in credit crunch scenarios).

      Nonetheless, the underlying message of Tobin is that by hook or crook, banks should keep depositors deposit with them and keep them happy is correct. This is true for a single bank or banks collectively. Tobin's base case is some sort of perfect competition so he emphasizes price but I'd use imperfect competition but also emphasize the role of price.

    6. Nick,

      Fair points, but I’m not sure I agree with this: “In that case, any developments in the deposit-taking business should have no impact on rate at which a bank is prepared to lend nor the amount it wishes to lend.”

      Let me try to restate what I was trying to say. In general, banks will look for the cheapest form of liabilities to grow their assets. Deposits should be cheaper than money market borrowing. So whether we view deposits as sources of funds for retail lending or bank asset base growth more generally, they’ll enable the bank to be more profitable as a whole (vs. more expensive money market borrowing). As bank executives decide on the bank’s lending strategy, they’ll take into consideration a holistic view of the bank’s cost of funding. If banks typically acquire liabilities via deposits as opposed to interbank borrowing, then surely that will impact the interest rate at which a bank is able to make a loan. The price of the loan will depend on the bank’s funding costs, and so if it can fund itself cheaper, then the price of the loan can be lower.

      Discarding my example, I would still say that the relevance of deposits for banks to be competitive in the loan business has to do with cost of funding relative to other forms of financing.

      The outcome of my “scenario 1” seemed pretty uncontroversial: If loans are only profitable by financing mostly with deposits at some price, then a bank that cannot acquire sufficient deposits is obviously doomed to fail.

      My “scenario 2” seemed a bit more complex: If loans can be profitable even if the bank finances mostly in the interbank market (which we can assume is a bottomless source of funds as you say), then to what extent are deposits relevant? I’d say they’re still relevant in the sense that if they are cheaper than interbank borrowing, then financing with deposits will make banks more profitable as a whole. If they can be more profitable, then to some extent, they should be able to outcompete less profitable banks (who may be more reliant on interbank borrowing) in the loan business. As such, the latter type of bank will need to improve their ability to retain and attract deposits, which would include examining the interest rate they’re paying on deposits (or savings accounts, time deposits, etc).

    7. I can try to restate it with your internal reference cost of funds - let me know if this makes sense to you. Let's say the bank thinks going forward, they can grow liabilities at a rate of 80% deposits and 20% interbank borrowing. The cost of deposits is .5% and the cost of interbank borrowing is 1%. So the average cost of funds for the bank = .8*.5 + .2*1 = .6%. This bank can be more competitive than a bank funding with a higher percentage of interbank borrowing.

    8. My comment which you quoted was specifically in the context of the assumption of a deep, risk-free interbank market. With that assumption, the opportunity cost of funds for a lending bank is the inter-bank rate. Taking the funds raised through deposits, it can either place them in the inter-bank market at 1% or lend them out. So if it lends them out at 1.5%, the incremental income from lending is 0.5% - it doesn't matter what rate is paid on deposits.

      Taking an extreme example, if the bank found it was able to raise deposits at 0% (perhaps the public suddenly decided all the other banks were very risky), it wouldn't make sense to make loans at 1%, because it could place the funds in the interbank market at that rate with no risk (by assumption).

      But the point is that this is based on the assumption that banks will happily deposit or borrow unlimited amounts in the interbank. Once we drop that assumption then we would expect to see the internal reference cost of funds also reflect the bank's ability to fund from our sources. Banks with good access to deposits might therefore be expected to have lower internal benchmark funding rates.