Hair of the Dog

"Too much debt got us into this mess, and now central banks are trying to get us out of the mess by lowering interest rates to get us to borrow and spend more?"

It sounds very much like the cure for hangovers: "take some hair of the dog that bit you". At best, drinking more to cure a hangover merely postpones the problem.

All analogies have problems. No simile runs on four legs. But this dog is legless. It won't hunt.

Assume a closed economy. Canada isn't closed, but the world is, and this financial crisis is a global crisis.

In a closed economy, aggregate net debt is zero. If Peter owes Paul $100, then Peter's liability is Paul's asset; Peter has a debt of plus $100 and Paul has a debt of minus $100, and aggregate net debt is zero. If Peter owns a firm which owes Paul $100, then Peter (or Peter's firm) has a debt of plus $100 and Paul has a debt of minus $100, and again net debt is zero. Government debt is a little trickier: if the government owes bondholders $100, and Paul owns the bond, then Paul has a debt of minus $100, and the taxpayers, who own the government, have a debt of plus $100. But some of those taxpayers are not yet born, and currently living taxpayers might only pay $60 of the taxes needed, in which case aggregate net debt of the currently living is minus $40. So "excessive" government borrowing actually reduces aggregate net debt if you follow that line of argument; but I'm not going to push it.

So in a closed economy, aggregate gross debt can rise and fall, and the distribution of net debt between different people can vary; but aggregate net debt (ignoring unborn taxpayers) is always zero. That's just accounting, but it helps keep our heads clear.

The global financial crisis cannot have been caused by too much global net debt, because global net debt is always zero. If increasing debt played a role in the crisis there are two possibilities: it was an increase in gross debt which caused the crisis; or it was an increase in the variation of net debt across people which caused the crisis; (OK, or maybe both).

Suppose we line up the world's population around the equator, and order everybody to lend $100 to the person to the West of them. Gross debt has increased by $100 per person, but the distribution of net debt stays the same, because everyone owes an extra $100 but is also owed an extra $100. Could this cause problems? Yes. If Peter decides to sell the IOU he just received, spend the proceeds, and leave himself unable to pay his debt to Paul, then Paul may be unable to pay his debt to Mary, and Mary may then be unable to pay her debt to Neil, etc. Each person is certain of the value of his liabilities, but is uncertain of the value of his assets. Each person's perceived value of his net debt may now be negative. If each person values his liability at face value, and assets at market value, then perceived aggregate net debt will now be negative. This reduction in perceived aggregate wealth will cause a reduction in spending, and a global recession. (Similarly, if there were lots of people like Peter, who had previously wanted to borrow and spend but couldn't, the initial increase in aggregate gross debt could have caused a global boom, until the Peters all defaulted).

Now divide the world's population into odds and evens at random, order all the odds to save and pay down $100 in net debt, and order all the evens to spend $100 and increase their net debt. Total gross debt stays the same, but the distribution of net debt now has a wider variation, with odds being net creditors and evens being net debtors. Some of the evens were previously on the brink of insolvency, and will now be pushed over the edge. But some of the odds were also on the brink of insolvency, and have now pulled back from the edge. If it's a small amount, like $100, the two effects will perhaps cancel, and total insolvencies will stay the same. Only if more evens than odds were closer to the brink of insolvency would the change in the distribution of net debt increase total insolvency. But if this were the case, then we get the same result as when gross debt increased: the reduction in perceived aggregate net wealth will cause a reduction in spending and global recession.

So, an increase in total gross debt, or in the variation of net debt, or both, has caused increased insolvencies, a reduction in spending, and a recession. Should monetary policy try to get people to borrow and spend? Isn't that what got us into the mess in the first place?

No. Increased borrowing was not what got us into the mess in the first place. It was increased borrowing and lending. Low interest rates might have made everyone want to borrow an extra $100, but nobody can in fact borrow an extra $100 unless someone is willing to lend an extra $100. Low interest rates encourage borrowing, but they also discourage lending. Low interest rates encourage people to want to increase their net debt, but aggregate net debt is always zero. If we want to blame low interest rates for creating the problem, we must explain how low interest rates caused an increase in gross debt, or increased the variation in net debt. It is not at all obvious that low interest rates would do this.

It is not obvious, but it is nevertheless conceivable, that low interest rates could have done this. Suppose low interest rates had a bigger effect on debtors than on creditors. When interest rates fall, debtors want to borrow more, creditors want to lend less, and everyone wants to spend more. In a Keynesian world (at least in the short run) that increased desire to spend will increase output and income and so actual spending will increase. Conceivably, the increased income could have caused an increase in gross lending, if creditors had a lower marginal propensity to spend than debtors, so creditors save and lend more and debtors spend and borrow more.

Let's try to get an order of magnitude estimate of the size of this effect. If there were a difference in marginal propensity to spend of 10 percentage points, and if income increased by 10%, then gross debt would increase annually by 1% of GDP. That's tiny, especially given the massive size of the boom that caused it, and the massive size of the difference in propensities to spend. Financial innovation looks far more plausible as a cause of the increased gross debt. Low interest rates were not the cause of the problem in the first place.

The same argument applies to low interest rates as a cure. If excessive gross debt is the underlying cause of the problem, there is no reason to think that low interest rates, by curing one symptom of the problem (insufficient spending), will make the underlying cause worse. Low interest rates encourage borrowers to borrow more; they also encourage lenders to lend less. It is not at all obvious that low interest rates will cause gross debt to increase further.

Again, there may be indirect effects on gross debt, via differences in marginal propensities to spend, or reducing the rate of default. But these side-effects on gross debt are just that: side effects of the cure, and not part of the way the cure works. Low interest rates cure a recession both by increasing the demand for loans and by reducing the supply of loans; they encourage an increase in spending through both channels, and can be effective even if gross debt stays exactly the same.

Low interest rates are Aspirin, not hair of the dog.

Comments

[JKH]

Nick,

Very interesting again - an important perspective on the interpretation of the debt problem – both sides considered, just like double entry book keeping, and assets and liabilities in banking.

But let’s qualify how low rates might cause less lending. Banking is a spread game. And so long as the spread isn’t constrained by the 0 lower bound, on it goes.

Now the question is how much of a constraint is the 0 bound on commercial bank lending? I haven’t looked at the spreads closely enough to know.

The constraint now is more capital uncertainty than low rates I think.

But I agree that low rates are aspirin, just as low rates, quantitative easing and qualitative easing are all deflation fighters more than inflation catalysts.

[Nick Rowe]

Thanks JKH! Yes, it's double-entry bookkeeping.

In my (over)simplified mental model, there are no financial intermediaries: Paul lends $100 to Peter. Suppose we add a bank, so Paul lends $100 to the bank (so the bank has a debt of $100 to Paul) and the bank lends $100 to Peter. Gross debt is now $200, rather than $100 with no bank. And a 10% reduction in the perceived value of assets (due to insolvency risk) would now lower aggregate net wealth by $20 instead of $10. So I think this magnifies the effect of insolvency risk. (Though of course, insolvency risk may be lower with banks' risk-pooling etc.)

Now cut interest rates. Let's hold the bank's spread constant for the minute. The bank is no more nor less willing to intermediate than before, but Peter now wants to borrow more from the bank, and Paul wants to lend less to the bank. Both Peter and Paul want to spend more, so that part of the story is unchanged. But there is pressure on the spread to widen if the bank's borrowing rate is pushed down to zero? I am going to have to think about that one a bit more.

Yes, risk is now a bigger deal than yield, and that's why qualitative easing might make a bigger difference. Or maybe liquidity is a bigger deal still (I did some earlier posts on this). So by buying risky/illiquid assets and selling safe/liquid assets the central bank could twist the yield curve along the risk/liquidity dimension.

By the way, is liquidity very important? Assuming it is, why is liquidity so very important? Why is it more important in a crisis? I've had these questions running through my mind ever since I did those posts on liquidity.

[JKH]

Nick,

Here is a little (actually big) problem with fair value accounting (which I mentioned in a previous comment) that relates tangentially to your story here.

Suppose I have a financial asset that is actually a debt security issued by a bank. Suppose the bank then gets into some trouble related to credit issues (I know – hard to imagine, but humour me).

The market value of my bank bond goes down – say from 100 to 80.

The implication of full fair value accounting for the bank is that it will mark down the value of that debt security on its books to a level that reflects “fair value”, which in this case is fairly easily observable as the market price of the security.

The implication of marking down the value of that liability for the bank is that, other things equal, earnings and the book value of the bank’s capital get marked up by 20.

So the effect of the bank’s credit risk deterioration is that its capital position improves (again, other things equal)!

This pathological result is one of the many objections that were raised “back in the day” when fair value accounting was first introduced. It’s one of many problems with the idea.

In your story, suppose Peter as borrower is the bad credit risk that is intermediated by the bank to Paul’s exposure as original lender. Assume further that unlike a normal bank, this intermediation is segregated and not commingled with other credits risks, so that Paul’s exposure as borrower is still totally to Peter via the bank, which is really only performing a broker type intermediation function.

Suppose each of Peter, Paul, and the bank subject themselves to fair value accounting. Lender Paul will lose 10, borrower Peter will gain 10 (due to a decline in the value of his liability), and the bank will be flat. This assumes credit risk flows through in this contrived case on a segregated basis from borrower Peter to the bank and then to Paul.

And the system will be flat.

I should get back to you on liquidity at some point. It’s such a huge and important topic that it makes my head explode. But what really causes this is the disconnect I’ve seen been the nature of blogosphere discussions on the subject (which are quite interesting) and what actually goes on inside big banks (which is also interesting).

[Nick Rowe]

JKH:

Wow! I hadn't realised that implication of fair value accounting. But it makes sense.

Let's ditch the intermediary for a minute, get back to just Peter and Paul, and suppose Peter has a 10% chance of defaulting on his loan to Paul. I said in my post that perceived net worth has declined by $10. That was really a digression from my main point in the post, and perhaps I should not have said it, but it's interesting anyway, so let's explore it now.

The important question is this: will Peter and Paul behave differently if Peter owes Paul $100 with a 10% default risk, than if Peter owed Paul $90 with no default risk? It's behaviour, especially their total demand for goods, which matters in this case. We choose the accounting system which best proxies for their behaviour.

If Peter still spends as if he owed $100, and Paul spends as if he were owed $90, then it is AS IF aggregate net worth were lower by $10. On the other hand, if they both spend as if Peter owed $90 to Paul, then fair value accounting, applied to both assets and liabilities, gives us a good proxy for their behaviour. And I am going to have to think which way they would behave. I need to find the right thought-experiment so I can apply standard theory of behaviour under uncertainty. I must think more about that one.

I would really like to hear your thoughts on that disconnect on liquidity discussions. Until a few weeks ago, I had what were probably the standard macroeconomists views on liquidity. They went roughly like this:

There are two (maybe three) types of assets: money, and all the other stuff (or maybe distinguish bonds from real capital). Money is very liquid; all the other stuff is less liquid. The fact that money is more liquid than all the other stuff explains why money exists (why we use monetary exchange and not barter); and it explains why people still hold zero-interest money while all the other stuff pays positive rates of interest. But it is not important for macroeconomists to worry about liquidity differences between the various subsets of "all the other stuff". Finance theorists may want to think about this stuff, from the microeconomic perspective, but the macroeconomic implications are as little relevant as the difference between cars and bicycles.

Then the financial crisis came along, and T-bills suddenly switched from the "other stuff" camp into the "money" camp. So the standard open market operation of monetary policy suddenly became much less meaningful (or meant something very different). And the liquidity-slope of the yield curve suddenly steepened massively. And some markets dried up completely, which made me realise the parctical importance of market volume for liquidity: it couldn't just be modelled as a fixed broker's commission -- it was endogenous. And maybe there were self-fulfilling multiple equilibria of trading volumes. And maybe changes in liquidity, and/or changes in the value people placed on liquidity, was the main thing that was causing asset prices to fall or rise, not just changes in risk, or risk-aversion.

So yes please, tell me more about this disconnect between blogosphere and real world discussions of liquidity. Is it like the disconnect between me a month ago and me now?

[JKH]

Nick,

I can give you a high level summary right now. The two major disconnects I’ve found are:

1. Bank liquidity management tends to differentiate between two types of liquidity: funding liquidity and market liquidity. Funding liquidity is the general area of how a commercial bank is positioned to fund its balance sheet – what it’s liability maturity structure looks like, it’s asset maturity structure, it’s liquid asset profile, it’s capital profile (capital provides long term liquidity), it’s contingent liquidity profile (e.g. contingent credit commitments), etc. etc. Market liquidity is the general area of how particular instruments trade in the market and how this affects their pricing and availability in different environments. Market liquidity is more pertinent to trading and investment banking operations. I find that blogosphere discussions tend to focus on market liquidity and often completely ignore funding liquidity issues. Funding liquidity might be viewed as the institutional portfolio profile of liquidity whereas market liquidity is more of an instrument profile. Funding liquidity and market liquidity converge as major issues in a universal bank that has both commercial and investment banking operations.

2. The second disconnect I’ve found is far more conceptual and theoretical. There have been a series of blogosphere discussions on what is known as “maturity transformation”. Maturity transformation is viewed by some as the root of all evil in terms of combining credit banking with the risks of short funding through maturity mismatches. The discussions get quite intricate in terms of combining related issues such as fractional reserve banking and fiat money theory and Austrian economics. What I find difficult about these discussions is that they have a heavy ideological bias as a debate about liquidity. There is a connection with real world banking in the sense that all commercial bank risk systems employ transfer pricing systems that attempt to identify maturity transformation on both an interest rate risk and liquidity risk basis (they are not always the same). So the phrase does mean something to me. But the discussions from there tend to move to a much higher plane of banking ideology debate rather than an assessment about how banks actually attempt to manage the issue in the context of the system as it exists.

[Patrick]

"explain how low interest rates caused an increase in gross debt, or increased the variation in net debt"

The theory I've heard is this: when Alan Greenspan lowered interest rates and promised to keep them low, the global savings glut went looking for safe investments somewhere other than US treasuries. They found it in MBS. After all US housing prices never go down, and the ratings agencies said it was all good as money. So huge sums of money pile into the now toxic securities we all know an love. Greed takes hold, the agency problem rears its ugly head, and they start handing out mortgages to anyone with a pulse because the investment banks want to pump out CDO to meet the demand.

At least, that is the story told by this NPR piece:

http://www.thisamericanlife.org/Radio_Episode.aspx?episode=355

[Nick Rowe]

JKH: Thanks. You have taught be quite a bit there

What you call "maturity mismatch" is what I've been calling "duration mismatch" (borrow short, lend long)?

Yes, I'm partly aware of the debate over maturity mismatch, and its relation to fractional reserve banking. I associate the arguments against it partly with Austrians, partly with free banking/private money proponents, but I think its roots go back to Henry Simons (??) in the 1930s? Anyway, my view is that maturity mismatch is one of the main things banks do, and ought to do. It's (mostly) why banks exist. People (non-bank people) want to borrow long and lend short (for various good reasons that I won't go into), and the job of banks is to make it possible for them to do it. But it's a very dangerous job; there are always two equilibria, one with a bank-run.

So "Funding liquidity" is the name bankers give to their attempts to manage the risks from maturity mismatch? By "risk" I mean both the "bank run" risk, and the interest rate risk (which are not unrelated, because if you can sell your long-maturity assets at a low interest rate/high price you can handle a bank run).

The liquidity I have been thinking about is more what you call "market liquidity".

And what I now realise is that funding liquidity is related to market liquidity. That is new (to me). If a bank knows it can quickly sell its long assets in a deep liquid market, in the event of a bank run, its funding liquidity concerns aren't a big problem. Right?

So the people who deal with a bank's funding liquidity care very much about the market liquidity of the bank's assets, right?

If I am right on all the above, my head is now a lot clearer on these issues.

My main remaining question is this: why are assets traded so much?? Why don't we (mostly) just buy assets and hold them till we retire? Is it just unproductive noise, with speculators trading back and forth between themselves? Or are we seeing the results of banks' (in the widest sense of banks') dynamically managing their funding liquidity?

Patrick: I'm going to think about your comment before replying. Brain can't multi-task.

[JKH]

Nick,

The key distinction for risk in banking is between commercial and investment banking. Their respective risk profiles are sometimes called “structural risk” and “trading risk”. Regulators also often refer to the “banking book” and the “trading book”. Universal banks combined commercial and investment banks. The major Canadian banks all have the universal bank model.

The major risks across all banking are credit risk, liquidity risk, interest rate risk, foreign exchange risk, and operational risk.

The maturity transformation discussions I’ve seen generally fall into the structural liquidity risk category. The other liquidity discussions seem to focus on liquidity risk as it pertains to trading books.

What I’ve referred to as funding liquidity might also be called structural or banking book liquidity. “Funding” is not meant to imply a focus that is restricted to the liability side. It is simply that liability management is the core liquidity management activity for the structural book. Policies typically require that a stock of liquid assets be maintained as precautionary liquidity. It is fundamental to acknowledge that this full asset-liability infrastructure is far more important that the so-called cash reserve with the central bank in terms of the ongoing substance of bank liquidity management.

Banks slice and dice all of these risks across both books. Notwithstanding the miserable job the system has done in dealing with a 100 year risk flood, bank risk data collection and analysis tends to be fairly sophisticated. But if anything, the infrastructures have allowed banks to fall into the trap of excess precision instead of sound judgement.

The duration mismatch to which you refer is really a structural interest rate risk perspective rather than liquidity per se. E.g. the duration of a 10 bond that resets its rate according to 3 month LIBOR is very short whereas the duration of a 10 year 0 coupon bond is 10 years. Yet both have significant liquidity protection in terms of their 10 year maturity – the 3 month floater certainly has much greater liquidity value than a 3 month LIBOR deposit with essentially the same interest rate duration.

Structural interest rate risk management is known in big banks as “gap management”. It’s a big deal along with structural liquidity management. An interesting aspect of gap management is that the zero bound for interest rates has been a critical issue for banks for the past 15 years. This is because historic spreads between the prime rate and rates on some of the more passive core deposit and transaction accounts were quite wide. Spread compression began long before the fed funds rate or the Canadian bank rate got down to current levels. The zero bound isn’t a problem just for central banks.

[Nick Rowe]

Patrick: Let's address the question of how a global savings glut, and central banks' response to it, would affect gross debt.

First, assume everybody in the world wakes up one morning and decides to increase savings by 10% of income. This lowers the "natural" (equilibrium) rate of interest. Assume central banks lower the rate of interest accordingly, which discourages savings and encourages investment. Assume all the increased investment takes the form of investment in housing, and that this is spread proportionately (to income) across all households. The result is that each household spends 10% more of their income on buying houses, and 10% less of their income on buying current consumption. No change in gross debt. It's all self-financed.

In order to make the story work (that an increased desire to save causes an increase in gross debt), you need to assume that the increased desire to save doesn't affect everybody; it only affects the people who are currently creditors, not those who are currently debtors. In this case interest rates fall, and creditors save more and debtors borrow more, and so gross debt increases.

But if it is the current debtors who have the increased desire to save, then gross debt will decline.

So, a savings glut will not (in general) cause an increase in gross debt. Gross debt is caused by differences between people: some want to borrow and others want to lend. Anything which increases the differences between people (with respect to their desires to spend or lend) will increase gross debt.

And if central banks did not respond to a savings glut by decreasing interest rates, the fall in spending below output would cause a recession, deflation, and the sort of problems we are seeing now.

[Nick Rowe]

JKH:
Thanks. OK, I have been using the words "duration mismatch" when I should have been talking about "maturity mismatch". Duration mismatch creates the risk that short interest rates will rise; maturity mismatch creates the risk that you can't roll over the liabilities (bank runs). I don't think they are quite distinct however; a run on the bank would be no problem if the bank could sell its long assets at a "good" price (i.e. low interest rate).

I'm interested in what you say about low interest rates over the last 15 years compressing the spreads. If I understand you correctly, you are saying that the zero lower bound/liquidity trap problem, which economists see as a problem beginning this year, bankers see as a problem beginning 15 years ago? Let me try to re-state this in economists' language.

There are "bonds", which pay nominal interest rate i%, and money (currency) which pays 0%. The opportunity cost of holding money is i%, and this represents a tax on holding money. Friedman's optimum quantity of money argued for pushing that tax down to zero, by pushing inflation negative. (It has been known for some time that this would create a danger of a liquidity trap). As i gets lower, the tax on currency falls, and currency is a competitor to what banks do, and this pushes down banks' spreads and profitability.

I'm not sure if the above paragraph correctly captures what you might be saying: the zero lower bound means that low interest rates compress banks' spreads and force(?) banks into riskier enterprises. That's an idea worth exploring.

[Patrick]

Nick,

I'm not an economist, but it looks like the narrative you present is not addressing what really happened. To be fair, I have no idea if worldwide gross debt is or is not a problem, but it certainly seems like net debt in the US and probably Canada is a big problem. I'm not married to the explanation put forward in the NPR piece, but it does make some sense to me and if it's wrong I'd like to understand why.

Working backwards; in the US there was an increase in net debt. I don't know if it represents a change in worldwide gross debt, but maybe that doesn't matter. In particular US mortgage debt increased significantly. That debt had to be issued by a creditor somewhere (e.g. China). Since they did lend the money to the US the question I'm trying to understand is what would motivate them to lend the money to the people they did for the purpose of buying houses?

At the time interest rates on safe investments like treasuries were barely keeping up with inflation. It seems a cogent argument to say that since safe gov't debt wasn't very attractive, then the 'savings glut' went looking for a new home.

If it found a new home in AAA rated MBS that people believed to be as safe as US treasuries, and yielded a much higher rate of return, wouldn't this motivate the people with money to lend even more money? After all, they believed there was essentially no risk and the returns were great.

Supposing the above is true, it probably still doesn't explain the bubble. After all, by definition bubbles can't be explained by fundamentals. But it might explain the initial conditions that then allowed the irrational actors and market failures to step in and make a huge mess.

Which comes around to the 'hair of dog' issue; if it was an increase in net debt within the US that created the conditions for such a disastrous bubble, then might not more net debt create more problems down the line? And what about paying off all the debt? Isn't that a problem too?

My layman's take (which may very well be wrong) is that it's a very different kind of debt. In the housing bubble case, the people with money where lending to creditors who bought unproductive assets (residential houses in the exurbs), who then, in many cases, leveraged the asset to buy unproductive consumer goods and gas guzzling SUVs. In the case of deficit spending by a gov't, presumably the money is spent on public works that enhance productivity in some way (one hopes).

[Declan]

First of all Nick, thanks for making explicit the link between inequality and gross debt levels - it's obvious when you explain it in terms of gross vs. bet debt, but it helped clarify my thinking on the link between debt levels and inequality.

To be honest, I'm not convinced that low interest rates don't lead to higher gross debt levels. You seem to be saying that a reduced willingness of creditors to lend will offset the increased willingness of debtors to borrow, but you seem to be overly discounting the increased capacity to borrow that low interest rates provide (and the concomitant increased willingness of creditors to lend at a given rate). Low interest rates will also drive up house (asset) prices (housing bubble, anyone) again creating an increased willingness to lend (against the higher collateral values) for a given interest rate.

Additionally, you seem to be thinking solely of lending as being done people with money and not considering bank lending which simply creates the money out of thin air (subject to capital requirements, which themselves are likely to be lower in a low interest rate environment). In fact, with interest rates low, you might find banks aggressively trying to expand their lending volume to compensate for reduced spreads along the lines of what JKH (who really knows his banking!) was suggesting.

I guess what I'm saying is that lower interest rates increase peoples willingness to borrow, but they also increase peoples capacity to borrow which in turn increases creditors willingness to lend and if individual creditors aren't willing to step in by lending more of their money directly (choosing to spend it instead of earning some measly interest from the bank), the money will be lent anyway as banks increase their leverage. The result is more debt.

I could see that you could argue that the root cause of the increase in gross debt is really changes in regulations and financially industry structure and inequality and whatever else, not lower interest rates directly, but in reality, I'm not sure you can separate out lower interest rates from all these indirect effects so easily as I suspect they are all interdependent in complex ways, but I'd need to give that some more thought. Finally, if government decides to intervene to lower interest rates by doing lending (either directly, or by guaranteeing loans) that creditors otherwise aren't going to do, isn't this undermining the very mechanism you expected to prevent lower interest rates from leading to an increase in gross debt?

[JKH]

Nick,

Assume W is a benchmark short term wholesale rate.

Then banks charge a rate on prime rate loans of W + x.

They pay a rate on liabilities such as savings accounts of W – y.

X and y are the spreads above and below W for pricing these assets and liabilities.

Interest margin compression threatens liability spreads when W declines to the level y. Any further drop must compress the spread y due to the 0 bound for (W – y).

As I said previously, because W was high in the earlier era of higher rates, the spread y was allowed to be quite wide.

I randomly Googled the following web page, which graphs the Canadian bank rate over the past 70 years:

http://www.bcrealtor.com/d_bkcan.htm

(I’d be interested to know where you go to get long term histories of Canadian interest rate information; the Bank of Canada site seems to be a bit of a joke in this regard compared to the Fed’s data base).

I’ll use the bank rate as the indicator for the benchmark short term wholesale rate W.

(Canadian chartered banks tend to use the 30 day bankers’ acceptance rate, but the bank rate is close enough for discussion purposes.)

Policy interest rates declined steadily on a trend basis since “Volcker’s Peak” in 1981. Through various interest rate cycles since then, there have been lower highs and lower lows. The key peaks since 1981 have been, roughly, 1990, 1994, 2001, and 2007/8.

What you notice is that the bank rate first came down as low as 5 per cent about 15 years ago. It had been higher than 5 per cent for the previous 20 years. So 5 per cent bank rate is an interesting reference point. It doesn’t tell the full story as a number, but it’s indicative of an important inflection point in Canadian banking.

The bank rate fell to around 5 per cent at that time and then moved up for a brief period again. This mirrored cyclical Fed tightening starting from a 3 per cent funds rate in early 93 and ending at 6 per cent in 94. In Canada, the 95 Quebec referendum was the final interest rate spike point, after which rates fell. But the underlying trend was still to disinflation and lower rates in the long run.

It was around this time also that there was a sea change in the Canadian mortgage market. The variable rate mortgage became very popular. I don’t recall the precise timing in relation to the 93-95 up cycle in rates, but this shift in mortgage preference caused a significant change in bank interest rate risk – because, other things equal, it exposed bank interest margins to declining asset rates. It sort of made sense – the consumer had caught on to the game of disinflation and lower rates as well, and chose to short fund (like banks!). The maturity structure shift in bank mortgage portfolios from fixed to variable rate was absolutely massive.

So two things were happening to bank balance sheets strategically – they were becoming more exposed to declining rates on the asset side due to these mortgages, and they were becoming more exposed to declining rates on the liability side, because (W – y) was approaching the zero bound.

This put bank interest margins under threat of compression in a low rate environment – unless steps were taken. Some banks started hedging some of this interest rate exposure with interest rate swaps (on a sort of delta hedging probability basis). Interest rate swaps became exceedingly useful risk management tools around this time.

(Ironically credit default swaps rather than interest rate swaps became the ultimate derivative weapon of mass destruction. Interest rate swap systemic risk has remained reasonably benign in recent years because the credit risk equivalent on interest rate swaps arises from relatively small percentages of nominal amounts (percentage differentials between fixed and floating rates) – differentials that have been contained in a secular environment of generally low interest rates – unlike the potentially massive percentage losses on nominal amounts now being generated in the case of credit derivatives.)

I wouldn’t connect what I’ve described too closely with the more recent debates about “global savings gluts” and low rates and risk seeking as a result of low rates. It’s just an observation about zero bound risk in commercial banking spreads in comparison to central banks dealing with the same issue, and the way in which banks manage their structural interest rate risk partly in response to this.

By the way, zero bound spread compression has something to do with the recent reluctance of Canadian banks to match bank rate decreases fully with prime rate decreases.

Finally, I do think duration and maturity mismatches (or interest rate risk and liquidity risk) are quite distinct, although they are dynamically interactive. Every financial instrument embeds risk from these and other sources. As you say, an interaction exists naturally that is critical for risk management. My problem again with some of the blogosphere discussion in this area is the failure to differentiate the two risks before reintegrating them in an attempt to understand where the risk is really coming from. This important distinction shouldn’t be conflated in the simple banking notion of “borrowing short and lending long” or “maturity transformation”, in my opinion.

[Nick Rowe]

Declan: Thanks.

Let me try to summarise. The naive argument is from the perspective of an individual borrower: "if interest rates go down, of course I borrow more". What economists can (and should) point out is that this perspective ignores the general equilibrium constraints. From the individual lender's perspective: "if interest rates go down, I lend less". Since for every dollar borrowed there's a dollar lent, these two perspectives (individual borrower's and individual lender's) contradict one another.

From the market or general equilibrium perspective, if interest rates go down, that must represent either a fall in the natural rate of interest, where aggregate demand for consumption plus investment go down, or a decision by the central bank(s) to cut the market rate of interest relative to the natural rate.

From the market or general equilibrium perspective, the effect of lower interest rates on gross debt is ambiguous. There are lots of possible indirect effects of lower interest rates on gross debt. Differential impacts on borrowers and lenders would be one type of indirect effect. I think the indirect effect via asset prices (that you mention above) is interesting, since increased asset prices do increase collateral. An increase in financial intermediation (with no change in ultimate lending and borrowing) would increase gross debt. And if we think of governments as one of the players in the game, sort of like a financial intermediary that both borrows and lends, then the recent decisions by governments to extend credit, financed by borrowing, would increase gross debt, as you point out.

Yes, figuring out all these indirect effects is getting too complex for me too. But the main point of my post was just to demolish the naive view, which assumes that the market experiment is just an individual borrower experiment times the number of borrowers.

JKH: This is interesting and serious stuff. I think I am going to try a post on this topic. You obviously know waaay more than me about the banking and finance sector (and waaay more than any economist as well, judging by the amount they are learning from you on other blogs). I am going to try to add my comparative advantage as an economist, and try to place your insights into some sort of macroeconomic/monetary policy context. This will massively over-simplify things in some ways, but I hope it will also add something.

[JKH]

Look forward to such a post, Nick. Simplicity is great if it can help the subject of banking risk find more traction within a macroeconomic framework.

Hoping also that bank CEOs don’t descend on your university with pitchforks in an attempt to stop the unveiling of the vault (like Geraldo and Al Capone).

:)

[Adam Smith]

A Specific Application of Employment, Interest and Money

Plea for an Adventure in a New World Economic Order

Adam Smith, Karl Marx, John Maynard Keynes and Alan Greenspan: a Unified Perspective

Abstract:

This tract makes a critical analysis of credit based, free market economy, Capitalism, and proves that its dysfunctions are the result of the existence of credit.

It shows that income / wealth disparity, cause and consequence of credit, is the first order hidden variable, possibly the only one, of economic development.

It solves most of the puzzles of macro economy: among which Business Cycles, Stagflation, Greenspan Conundrum and Keynes' Liquidity Trap...

It shows that Adam Smith, John Maynard Keynes, Karl Marx and Alan Greenspan don't contradict each other but that they each bring a meaningful contribution to a same framework for understanding macro economy.

It proposes a credit free, free market economy as a solution that would correct all of those dysfunctions.

In This Age of Turbulence People Want an Exit Strategy out of Credit, an Adventure in a New World Economic Order.

Read It.

[JKH]

Nick,

Here is something on the fair value issue that's recently come out and could affect the financial rescue package, because of Volcker's influence on Obama.

http://www.cfo.com/article.cfm/12960181/c_12959021?f=home_todayinfinance

Note in the article the specific reference to the same liability fair value problem I pointed out above:

"The report's authors cite, for instance, a provision in 157 under which a company can boost its reported earnings by becoming less creditworthy. Thus, in paragraph 15, FASB states that the fair value of a company’s liability must reflect the risk that the company won’t pay it back. Thus, as the risk that companies won’t pay back their debts rises, their reported liabilities actually decrease—and may even provide an earnings boost."

Also note that this sort of thinking may not be consistent with the idea of gutting commercial banks of their bad assets at so-called market or fair value. It's probably more consistent with the original TARP idea in terms of pricing bad assets.

[Nick Rowe]

JKH: that's interesting. You can see the logic, but it's a crazy logic! The main point of a balance sheet is to see if the company can pay its debts, so valuing the debts at 50% on the $1 because others think there's a 50% probability the company will default, defeats the whole purpose.