Addendum to Pushing on a String

In discussions with the ever thoughtful HBL (which I now think stands for “♥s Bank Liabilities”), he asked where in H.4 data we would see bank bond liabilities increasing in response to an increase in CB bond acquisitions. And the answer is that you may not see it.

Counting Properly

First, we are concerned with the financial sector as a whole (which I’ve defined as all financial sectors in Flow of Funds data except for retirement funds, insurance, CEFs, Mutual Funds, and ETFs). So “finance” is not just banks but also includes ABS issuers and money market mutual funds. Recall that “bonds” are foreign and domestic corporates, munis, treasuries and agencies.

Second, we are not concerned with just the CB. The issue is bond supply of the non-household sector versus bond demand of this sector. If the CB is buying bonds but the foreign sector is selling, then there is no need for the financial sector to adjust its net bond supply.

Third, what matters is the net bond supply, not the quantity of financial sector bond liabilities. This sector has traditionally held more bonds on the asset side of balance sheet than the number of bonds on the liability side.

What matters is the net position, not the gross position. If a bank sells an agency to a household, this operation is just as effective at reducing household deposit holdings as if the bank issued a new bond to households. It doesn’t matter whose liability the bond is — the financial sector supplies bonds to households by selling bonds to them, regardless of whether it is incurring a new liability or selling a third party liability.

All About Elasticities

Now the larger debate whether the non-financial sector will decrease borrowing as a result of QE. I think borrowing demands will be a function of expectations of return and interest rates. I don’t see a channel from quantity adjustments by the CB to a decrease in borrowing that is not an interest rate channel. And the effects of operations like QE on interest rates are extremely small, and go in the opposite direction of what you would expect if you assume households will borrow less to offset the increase in deposits created by the CB.

Roughly speaking, the Keynesian position is that investment, which is often or primarily funded by borrowing, is going to determine the level of savings.

My claim is that the allocation of household savings among bonds and deposits will be whatever this sector wants, conditional on the overall level of savings and interest rates, and that the other sectors aren’t going to have a lot of say on what that allocation happens to be.

The financial sector will transform the maturity of assets in such a way as to counter any maturity transformations being done by the CB (or other sectors), at the expense of a (small) change in rates.

The change in rates will not substantially alter household portfolio demands.

This is an elasticity argument, based on the assumption that the interest-elasticity of savings demands and the interest-elasticity of investment is substantially less than financial sector interest-elasticity of bond supply.

The reason for this difference in elasticities is that the financial sector operates via leverage, converting small spreads into large profits. In theory, with infinite leverage, the elasticity of supply of bond supply will be infinite. Of course it is not infinite, nevertheless the financial sector is sufficiently more sensitive to rates so that small changes in yield make it profitable for this sector to undo any attempts by the CB to transform the maturity of assets held by households away from what the household sector wants. Households need only offer a small premium to get back to their ideal savings allocation.

Therefore we might as well assume that the household sector is always able to allocate its savings among bonds and deposits however it wants, so that our concerns should be focused on the overall level of savings rather than the allocation of that savings.

Interest rate policy — if it can encourage more investment,  or income policy —  to directly supply savings — are what is important. Portfolio shifts are not important, moreover they do not succeed at changing the portfolio holdings of households.

Update: fixed some typos.

Advertisements
Addendum to Pushing on a String

16 thoughts on “Addendum to Pushing on a String

  1. Thanks for the follow-up post. One minor typo: H.4 should be H.8.

    “If a bank sells an agency to a household, this operation is just as effective at reducing household deposit holdings as if the bank issued a new bond to households.”

    Here you have introduced an additional mechanism by which the private sector’s portfolio adjustment can occur courtesy of the financial sector. (At least, if it was covered in your last post, I didn’t pick up on it). I agree that this could be a contributing factor. I think of it as related to but distinct from the previous mechanism you described. While I was aware of this possibility, I hadn’t checked whether it was evident in the data.

    “What matters is the net position, not the gross position.”

    This makes sense. So if the combined mechanisms you describe explain how the private sector adjusts to QE, we should expect to see a clear downtrend in the black line in your graph as a result of QE (unless other concurrent dynamics in the economy overwhelm that effect). The recent years are a little difficult to discern, but to me the graph evidence appears inconclusive either way.

    So I continue to believe that the two dynamics you describe probably work in parallel to the dynamic I describe, and I don’t know the relative magnitude of each in delivering the outcome that we both agree on.

    “Now the larger debate whether the non-financial sector will decrease borrowing as a result of QE.”

    Just in case there’s any confusion, that’s not my debate. I have specifically argued that total borrowing would NOT decrease as a direct result of QE, just that a relative shift would occur between bank lending and non-bank lending.

    “I think borrowing demands will be a function of expectations of return and interest rates. I don’t see a channel from quantity adjustments by the CB to a decrease in borrowing that is not an interest rate channel.”

    I agree with this. It is reasonably widely recognized in the blogosphere that a net private sector debt reduction (assuming bank loans are involved) also reduces the money supply, all other things equal. The concept my post introduced that I had not seen discussed elsewhere is that the private sector does NOT need to reduce its borrowing to make the QE-reactive portfolio adjustment. As best I can tell you have not yet chosen to give any judgment on the validity of that argument (which is fine).

    1. Yes, I didn’t make this point very clear, but even in the original post, I was arguing that it was the net bond position.

      “So if the combined mechanisms you describe explain how the private sector adjusts to QE, we should expect to see a clear downtrend in the black line in your graph as a result of QE (unless other concurrent dynamics in the economy overwhelm that effect). ”

      No, you don’t need that. Wow, now I’m thinking I really failed at communication.

      Look, this is simple:

      Go back to the graph. All you need to know is that the lines above the x-axis — the net bond consumers — are holding steady even as the main suppliers of bonds (government + firms) are swinging about. The sum of all lines = 0. All lines are net bond positions (but households and the CB do not sell bonds).

      Therefore the lines in the middle are adjusting to keep the lines at the top steady.

      That’s all you need to know — the steadiness of the top lines means that households are able to keep their bond holdings at the level they want even in the face of large changes to bond supply coming from the traditional bond suppliers (government + firms).

      Perhaps I should re-write the whole post. In any case, it is not two different mechanisms, it is one mechanism. Finance increases its supply of bonds to the household sector as a result of a small change in rates. At that small change, household asset demands don’t move a lot, and bond supply from the non-leveraged sectors doesn’t move a lot. What moves is finance — the sector most sensitive to changes in rates.

  2. “but even in the original post, I was arguing that it was the net bond position”

    Now I see that you did label your graph ‘net bonds’ in the last post. Since you hadn’t explicitly called out the ability of banks to sell assets to the non-bank private sector (unless I missed it), I didn’t think through the ‘net’ implications at the time.

    “No, you don’t need that [a downtrend in the black line]. Wow, now I’m thinking I really failed at communication.”

    Well the trouble is I think I understand you 🙂 So it could also be that I don’t really, or that I’m failing in my own communication…

    Here’s why I think the black line should be dropping concurrent with QE if your dynamics explain the portfolio adjustments of the private sector:

    1. The black line is (BA-BL)/HD where BA is bond assets of the financial sector, BL is bond liabilities of the financial sector, and HD is household deposits.

    2. You have mentioned two mechanisms by which the financial sector can enable the desired portfolio shift by the non-financial private sector in response to QE

    3. The first mechanism involves BA decreasing — the banking sector sells assets to households and businesses, reducing the money supply

    4. The second mechanism involves BL increasing — the financial sector accepts a larger proportion of its liabilities as bonds rather than money liabilities (deposits & money markets).

    5. No matter what mix of these two effects actually occurs, the spread (BA-BL) should DECREASE as a mathematical result of points (3) and (4) above. As such, the ratio (BA-BL)/HD should also decrease.
    – That is, unless either: (a) another explanation, such as the one I offered, explains the private sector’s adjustment mechanism, or, (b) other dynamics occurring in the economy overwhelm how visible this dynamic reaction to QE is in the data.

    “Go back to the graph. All you need to know is that the lines above the x-axis — the net bond consumers — are holding steady even as the main suppliers of bonds (government + firms) are swinging about.”

    I assume you mean the graph in your last post. I agree that it shows that households and businesses are able to keep their portfolio mix stable, however this result is also entirely consistent with the dynamic I described in my posts. i.e., it is evidence of the conclusion we both reach, without giving evidence of how the adjustment occurs.

    “Perhaps I should re-write the whole post. In any case, it is not two different mechanisms, it is one mechanism.”

    No need to re-write it at least on my account.

    Whether it’s one or two mechanisms is really semantics, unless I’m completely missing something. So instead, I’ll call it one mechanism with two sub-mechanisms, one related to a shift in financial sector asset holdings, the other to a shift in financial sector liabilities. Both shifts are part of the financial sector accommodating the non-financial private sector’s portfolio preferences.

  3. No matter what mix of these two effects actually occurs, the spread (BA-BL) should DECREASE as a mathematical result of points (3) and (4) above. As such, the ratio (BA-BL)/HD should also decrease.

    Take a look at the second graph. As the CB was buying bonds (decreasing the bonds available to households), the government sector + rest of the world was selling, so the sum of these three factors was actually an increase in the number of bonds available for households to purchase. They kept their holdings constant but the insurance/investment sectors added to their bond holdings.

    The only issue I have with your dynamic is that I don’t see why, say, if the foreign sector is accumulating bonds, household borrowing should decrease. You need some casual link. On the other hand, if households are willing to pay a premium for purchasing bonds, then I do see a mechanism by which the financial sector will sell them bonds. The securitization craze is a good example. IMO, as the foreign sector was accumulating treasuries there was a huge demand for AAA debt, and the financial sector supplied this debt with MBS. Clearly there is a “channel” through which asset demands of households will be met. But I don’t see a channel by which CB bond purchases lead to a reduction in household borrowing.

    If there was such a channel, then why would households not decrease their borrowing when China was buying bonds, but they would decrease their borrowing if the CB is buying bonds? Who cares who buys the bonds? All that matters, in the end, is the allocation of savings by households among bonds and deposits. If bond holdings are unchanged, then deposit holdings will be unchanged.

  4. “Take a look at the second graph. As the CB was buying bonds (decreasing the bonds available to households), the government sector + rest of the world was selling, so the sum of these three factors was actually an increase in the number of bonds available for households to purchase.”

    I consider this to be a potential example of what I labeled “other dynamics occurring in the economy overwhelm how visible this dynamic reaction to QE is in the data.” Maybe the increased government debt net of QE is making it harder to see the results of the dynamics that you describe in the data. For now, I still consider your graphs useful but inconclusive (with respect to evidence that the dynamics you describe are responsible for the portfolio adjustments).

    “The only issue I have with your dynamic is that I don’t see why, say, if the foreign sector is accumulating bonds, household borrowing should decrease…. But I don’t see a channel by which CB bond purchases lead to a reduction in household borrowing.”

    That is not what I am arguing at all. I thought I had explicitly responded to this multiple times, including in my first comment above on this post.

    If you want the full details of my argument, please read both these posts:

    http://www.thoughtofferings.com/2010/10/how-loanbond-choice-helps-private.html

    http://www.thoughtofferings.com/2011/04/further-evidence-that-private-sector.html

    But, I will try to put up a post that is a bit more concise and concrete soon, as I’m sure my explanations have lots of room for improvement.

    1. HBL,

      I apologize for misrepresenting your argument. I read the loan-bond choice blog. Actually, I think we are talking about the same effect. Note that the loan-bond choice is actually within the “finance” sector as I have defined it. A household does not have the option of going to the bond markets. In the U.S. bank loans and bonds are different markets. Bank loans are for small businesses, student loans, and securitized loans against real estate. Large firms can access the bond markets and obtain funding that is not collateralized. This is because banks, when they lend, are levered and so must lend against good collateral. However, this is country specific. In other nations, bank lending can compete directly with bond lending. I’m going to focus on the U.S. at the moment.

      However, a bank does have the option of securitizing the loan and creating a bond. ABS issuers that sell bonds backed by credit card loans, mortgage loans, or auto loans — those are all financial sector entities in my model. When I consolidate the financial sector, the “loan-bond” choice shows up primarily as the financial sector shifting its liabilities.

      But I think in the end we are describing the same effect.

      We would be describing a different effect if firms had the option of borrowing from banks or borrowing from the bond market. In practice they do not have this option, but in other countries or with other institutional arrangements there is this option. In that case, what you would see (in my charts), is an increase in firms bond issuance and a decrease in firm bank borrowing.

      So a theory that would include _all_ of these effects is basically the model of total bond supply and total bond demand. I was focusing on the financial sector being the “swing” bond supplier. But really all bond suppliers can adjust to CB bond purchases. Firms can sell more bonds, the foreign sector can sell more bonds, etc. As long as other sectors are selling more non-deposit liabilities to the household sector, and as the household sector’s total savings is unchanged as a result of QE (except for small effects to due yield changes), then household deposits will also remain unchanged.

      In that case, your loan-bond choice model is more general than my model, since it points out that anyone can increase bond supply to offset CB bond purchases.

      Hope that makes sense.

      1. Yes, I think that mostly makes sense. As I’ve said all along I think the two perspectives are compatible with each other 🙂

        In my first post I definitely questioned how interchangeable the two markets (loans and bonds) were (and admitted to limited industry knowledge), and I understand that some borrowers may have only one kind of option. However, I wonder if the degree of crossover even in the US could potentially surprise.

        For example, in my first post I mentioned peer to peer consumer lending whether direct or pooled (clearly small in scale now, but still an example of “bond-like” consumer borrowing), and securitization of bank loans (which you also mention here). In my second post I quoted an article that mentioned a flexible company borrowing via bonds, “trying to lock in low interest rates while getting the flexibility to repay debt any time, as it would with a loan.”

        But practically speaking, there might be some limitations to this adjustment process, both with respect to time lags and how interchangeable the types of borrowing are.

  5. Oliver says:

    Your description seems somewhat analogous to the MMT description of ‘forcing’ private (household) sector indebtedness during ‘the great moderation’ except that in your story it works via banks’ spreads instead of a through the fiscal stance. Both ‘push’ the creation of ‘money things’ (the endogenous argument) further down the liquidity hierarchy, first from gvt. (via fiscal) to banks (controlled via monetary) and, in the above case, further to the shadow banks or into corporate papers / risk. I find your chart indicates this effect quite clearly.

    Slightly tangential, but Randy Wray describes the hierarchy story quite nicely in this paper http://www.levyinstitute.org/pubs/wp_656.pdf (just in case you or other readers were not familiar). It helps me picture the paths that ‘money’ creation can take when influenced by policy even if that is in itself no proof for the endogeneity of its creation.

    Does that make sense?

  6. Oliver says:

    whoops, i actually intented to post my comment on HBL’s site. I was reading both and slipped onto your’s. Sorry!

  7. Anders says:

    rsj, these two posts are hugely interesting and seem very important, but they are very compressed and challenging to follow, with various steps and assumptions unstated. If you ever do a third post (I would be very glad if you would) which sums things up and reflects some of your dialogue in the comments, I for one would appreciate it if you could try to anchor it in slightly more concrete terms from time to time, and also cover the following points (apologies if these seem obvious; it may be that your intended audience is purely maths/economics grads).

    It seems you are saying, at a really high level, that demand / supply functions for bonds is more or less exogenous for all sectors except for the financial sector, which is entirely endogenous and serves as a residuum / balancing item, to bring net bonds issued/owned to zero, by adjusting its stance automatically if any other sector changes its desired holdings of bonds.

    1. Why is it that the financial sector always fulfil this role automatically? Is it something to do with incentivisation, or should one view it as a passive ‘mirror’ of positions taken against it, like an accounting identity? I am struck by the contrast with macroeconomic sector surplus / deficit positions where different sectors can serve as balancing items under different conditions: for example sometimes one talks of a government ‘accommodating’ a private sector desire to save more (ie government is the residuum), and at others one talks of a government surplus ‘forcing’ the private sector into more borrowing (private sector is the residuum). A crude analogy would suggest that no single sector always fulfils the residuum role when it comes to bond holdings.

    2. What is so special about bonds? A private sector agent’s surplus at the end of a period (thinking Lavoie-style circuitism) can become an increase in deposits or other financial assets – should one regard equities as included with bonds in your discussion? (I do appreciate that equities aren’t covered under QE, which was where your posts started.)

    3. I struggle to grasp your point about elasticities. I am used to thinking about elasticity of demand or elasticity of supply, but under your thesis the financial sector can be either a net supplier or a net holder of bonds, and needs to be able to move smoothly from one position to the other. So your concept of elasticity for the financial sector isn’t simply EoD or EoS, but spans both.

    4. When you talk of “supplying” a bond, I assume this should cover both issuing a new bond (primary market) as well as selling an existing bond (secondary market)?

    5. Philosophical point: why are you expressing this principle in terms of bonds and not deposits? I appreciate bonds and deposits are logical complements of each other, but why is it that there is a principle to be expressed in bond terms?

    1. Thanks. I should make another post about this, but here are some responses:

      “Why is it that the financial sector always fulfil this role automatically?”

      It does not need to only be the financial sector, but it will primarily be the financial sector, because this is the sector that employes leverage, which means that a small spread in instruments corresponds to a large profit. Non-financial firms do not borrow in order to lend, they borrow in order to invest and produce. Financial firms borrow in order to lend, that is they transform financial assets from having the characteristics that non-financial borrowers want to hold to having the characteristics that non-financial savers want. As soon as a small spread is provided, this transformation will occur.

      You still have some arbitrage in the other sectors — for example, firms can borrow in order to buy back shares, when equity becomes more expensive than debt, but the financial sector is the one that can do this quickly and on a massive scale in response to small changes in spreads.

      “What is so special about bonds? ”

      I looked at the bond/deposit trade-off because the CB intervention is to increase the quantity deposits by decreasing the quantity of bonds (e.g. buying bonds with newly created deposits).

      “I am used to thinking about elasticity of demand or elasticity of supply, but under your thesis the financial sector can be either a net supplier or a net holder of bonds, and needs to be able to move smoothly from one position to the other. So your concept of elasticity for the financial sector isn’t simply EoD or EoS, but spans both.”

      Yes! When you are talking about long lived goods, I think this is the appropriate concept to use. Elasticity of net supply. If a good is uniquely produced by a small group and cannot be re-sold, then you can conceptualize a population of suppliers interacting with a population of demanders (everyone else).

      But with long lived bonds, that is not the case. Everyone who has a bond or may be willing to issue a new bond is a supplier. Therefore everyone is a supplier.

      Everyone who may want to buy a bond (everyone) is a demander. Everyone is also a demander.

      Therefore rather think that each person has some indifference price at which they will be neither a (net) supplier or a (net) demander, but will be satisfied with their position. Then imagine the ratio of suppliers to demanders, which increases with the yield. As the yield rises, some suppliers switch to being demanders. The equilibrium price is that when the ratio of suppliers to demanders is 1.

      So, in this case, we can look at the change of suppliers/demanders in response to a percent change in the yield. Then, we can disaggregate and discuss population subgroups. For example, among households, what will a 1% increase in yield do? Then the elasticity argument is that among the financial sector, a change in yield will cause a larger swing in the ratio of suppliers to demanders than among the non-financial sector — because this sector is leveraged and therefore more sensitive to the yield. Therefore it will primarily be quantity movements in the financial sector that causes the adjustment to equilibrium, rather than quantity movements in the non-financial sector. And therefore the quantity of deposits held by the non-financial sector isn’t going to change a whole lot, either.

      “When you talk of “supplying” a bond, I assume this should cover both issuing a new bond (primary market) as well as selling an existing bond (secondary market)?” — Yes.

      “Why are you expressing this principle in terms of bonds and not deposits? I appreciate bonds and deposits are logical complements of each other, but why is it that there is a principle to be expressed in bond terms?”

      In the post, I graphed the relative holdings — e.g. bonds/deposits of each subgroup. But conceptually, it’s easier to apply supply and demand considerations in the bond market than to look at a “deposit market” — at least for me.

  8. Anders says:

    rsj, thanks for the reply. I’m still grappling with the sensitivity point, as a purely amateur economist. I have tried to build a very simple XL model to illustrate your point, but it doesn’t seem to give the result I was expecting; do you have an email address (or the inclination) for me to send it over?

    I am yet another recentish MMT convert; the inflation concern (significant underfunding of deficits will alway leads to inflation) appears to be the #1 challenge for MMT – so your argument about private non-bank preferences for bonds vs deposits being fairly robust in the face of underfunded deficits, seems critically important.

    Cheers

  9. Anders says:

    Thanks – I have sent a spreadsheet which tries to show, non-algebraicly (in order to preserve one’s intuitions), demand / supply of bonds by sector as a function of market yield – which I think is a way of showing the market clearing level of yield in order to balance the net demand/supply at zero.

    When I try to show elasticity of net supply, I get a radical / asymptotic discontinuity in the banking sector, which ties into my unease with the “elasticity of net supply” concept.

    Should such a discontinuity not be seen as problematic? Is there a different, better way of conceiving of elasticity here?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s