Understanding the Yield Curve
One of the puzzles of recent financial history has been the repeated failure of a flattening or inversion of the yield curve reliably to signal economic doom. Repeatedly since 1990, economists and others have greeted this phenomenon as a harbinger of forthcoming economic trouble, and just as repeatedly it hasn’t turned out that way.
For laymen, the yield curve is a graph that shows the relationship between yields and maturity dates for bonds, certificates of deposit or other investments to mature within a given amount of time. The longer it takes for the instrument to mature, the greater the yield. When economic forces cause a shorter maturity to produce a greater yield than a longer one, the yield curve is said to be inverted, according to bankrate.com, and economists start looking for the culprit.
In fact, as economists are far too fond of repeating, an inverted yield curve has predicted 11 out of the last six recessions since the early 1960s. Five times since the last recession, it has missed entirely.
Prior to the mid-1980s, this signal was almost always right, and for good reason. An inversion of the yield curve ought to be an extremely powerful brake on the credit cycle, which in turn ought to be a powerful brake on the underlying economy.
Let’s get the history right first. The National Bureau of Economic Research has recognized seven recessions since 1960, the latest of which was March to November 2001. (Contrary to popular belief, the NBER’s definition of a recession isn’t simply two successive quarters of negative gross domestic product growth, but rather “a significant decline in economic activity spread across the economy, lasting more than a few months, normally visible in real GDP, real income, employment, industrial production and wholesale-retail sales”).
The NBER’s Declared Recessions
Trough April 1960 February 1961 December 1969 November 1970 November 1973 March 1975 January 1980 July 1980 July 1981 November 1982 July 1990 March 1991 March 2001 Nov 2001
There are many ways the yield curve can flatten or invert, which will provide greater or lesser degrees of volatility. I have chosen to look at the history of the five-year minus 12 month curve, mainly because it is only moderately volatile.
The table below shows the 11 instances of sustained flattening or inversion since 1960
Flat/Negative Yield Curve
Start Finish Did the Signal Work? September 1965 February 1967 Not successful signal April 1968 January 1970 Successful signal March 1973 December 1974 Successful signal August 1978 May 1980 Successful signal October 1980 September 1981 Successful signal February 1982 April 1982 Successful signal February 1989 December 1989 Not successful signal October 1995 January 1996 Not successful signal August 1998 January 1999 Not successful signal Aug 2000 January 2001 Successful signal December 2005 August 2007 Not successful signal
Source: CEIC Data, ColdWater Economics
Six successes out of 11 tries is no better than a coin-toss. But as the table shows, what’s happened is that a signal which was successful five times out of six during 1960 to 1982 subsequently failed four times out of five between 1982 and the present day. In other words, this is a signal which truly used to work, and then truly didn’t.
An inverted yield curve really ought to work as a very good signal for economic trouble, because traditionally it tends to check growth in banks’ assets, and banks’ assets tend to be the mirror image of the liabilities on corporate and household balance sheets. Slow the growth of one, and you slow the growth of the other, and thus slow the economy. The hinge of this, of course, is banks’ usual practice of borrowing short and lending long. When the price of short-term liquidity is greater than can be recouped from lending long, the lending tends to stop. More generally, an inverted yield curve promotes holding cash/near cash over buying/creating financial assets. Does it work? Since 1975, there have been five major credit slowdowns, measured on a six month average month-on-month momentum basis: they are listed below:
Major Credit Slowdowns: Month-on-Month momentum 6 month average
Start of Slowdown Trough Recovery Starts 4Q74 2Q75 4Q76 1Q80 3Q80 1Q82 1Q90 3Q91 2Q93 2Q01 1Q02 3Q02
The short answer is that since the 1980s there is no statistically significant correlation between the shape of the yield curve and changes in lending behaviour. The graph below illustrates that breakdown
I suspect there are two reasons for this divorce between the signals from the financial markets and what happens in lending markets and the real economy:
Bank credit is no longer principally financing the ‘real assets’, including both real estate and the plant and equipment upon which immediate and future economic activity depend. Once again, the easiest way to appreciate this is to chart changes in bank lending against changes in private sector investment spending: between 1960 and 1980 there appears to be a reasonable correlation (0.46 over 80 observations), but since 1980, that correlation dwindles markedly (0.20 over 110 observations). In other words, the linkage between what the banks are doing and what the real economy is doing is loosening. There’s a reason for this.
Source: CEIC Data, ColdWater Economics
The Rise of Derivatives and the Hegemony of Finance
I think the divorce between the real economy and the financial economy is captured by and in the derivatives market. It seems most likely to me that the collapse since the 1980s of the yield curve as a useful predictive tool and the collapse of the linkage between changes in banks’ balance sheets and trends in corporate and household balance sheets both share the same parentage in the rise of derivatives products. Partly this is because derivatives products allow banks (and others) to hedge against the liquidity crunch suggested by an inverted yield curve, provided one can find a counterparty.
The first thing to take on board is how utterly central to bank balance sheets, and even more, to bank profits, derivatives have become. The International Swap Dealers’ Association’s attempts to track the size of the market started back in 1987, and by the end of that year the notional value of swaps and interest rates options outstanding totalled a quaintly restrained US$866 billion. To cut a long story short, by end-June 2007, the notional amount outstanding had risen 400 times to US$347.1 trillion.
Source: International Swap Dealers’ Assn
There are two contexts in which that extraordinary number begins to make sense. It is 39 times the total outstanding issue of US Federal Government securities, and it is just under 20x the total outstanding bank credit of US, the Eurozone and Japan combined.
Source: International Swap Dealers’ Assn, CEIC Data, ColdWater Economics
Important milestones for this market are:
June 1990: swaps & options notional outstanding equal to outstanding US Federal govt securities.
June 1999: swaps and options notional outstanding equal to 10 times US Federal government securities outstanding
June 2003: swaps and options notional outstanding equal to 20 times US Federal govt securities outstanding, and 10 times total bank credit in the US, Japan and the Eurozone combined.
June 2005: credit default swaps alone have greater notional value than entire bank credit in US, Japan and the Eurozone combined.
June 2006: credit default swaps alone have 2 times notional value of the entire bank credit in US, Japan, Eurozone
There has long been a strain of more or less swivel-eyed commentary on the rise of derivatives that has stressed the enormous off-balance sheet liabilities thus carried by the banking system, and the danger of this increasing systemic financial risk. I do not intend to join with that commentary, or add to it. Rather, my observations are these:
When the notional amount of swaps outstanding so dwarfs the underlying market for securities on which they are presumed to be priced (ie US treasuries), no one should be surprised if those securities (ie US treasuries) become either unreliable indicators (which is explicitly what happened to the 30-year treasuries), or alternatively lose the power to explain what is happening in financial markets.
When the notional amount of these derivatives dwarfs the amount of total bank credit outstanding, it is clear that we should no longer expect much in the way of linkage between the financial economy and the real economy. The principal activity and preoccupation of banks becomes playing financial markets, so the idea that banks’ balance sheets are simply or even mainly reflections of non-financial corporate and household balance sheets lapses. Rather, banks’ balance sheets become mainly reflections of other banks balance sheets.
By extension, when the amount of credit default swaps outstanding becomes greater than the actual amount of bank credit out there (which has been the case since the second half of 2005), it should not be greatly surprising that out there in the real world, bad credit allocation decisions get made. Those credit decisions have, after all, only a rather minor part to play in the overall financial game of credit default swaps.
My belief is that the great derivatives bonanza has been underwritten by a 25 year bull market in bonds which is now over. Counterparties willing to trade the interest payments from a 10-year bond are always going to be easier to find if there is an assumption that the capital value of the bond is likely to rise anyway as yields fall.
And when that is no longer a sure thing? Well, it is at least interesting to notice that the emergence of credit-default swaps occurred precisely during the period since 2003 when the decades-long assumption of falling bond yields was becoming questionable. At the time it was commonly described as ‘the search for yield’. One might alternatively call it the ‘search for a yield curve’.
Source: CEIC Data
The enforced unwinding of these credit swaps is, I suspect, only the first stage of a very long and probably very slow unwinding which will creep across the broader derivatives market as the assumption that “warehousing” bonds is ultimately a profitable business gets proved wrong. Already, one can see how the unwinding of those derivatives is gradually dragging up rates throughout the credit spectrum.
I detail the widening of credit spreads in each Heads Up (see below, in the Capital Market Spreads graphs), but I fear that those charts presuppose a return to ‘normality’ which misunderstands the broader dynamic at work. Looking at the range of credit yields taken together, it is striking how the widening of yields was first confined largely to the worst credits, but is gradually working its way up the credit spectrum to the better credits.
I now suspect that the credit spreads are more likely to narrow by US treasury yields rising than by the yields on lesser-credit bonds falling
The removal of the assumption of a bond bull market will certainly be painful and transformative for the financial services industry, but since, as we’ve discovered, the balance sheet of commercial banks no longer reflects the balance sheet of household and corporate sectors, one need not rush to pessimism about the impact of this transformation on the non-financial economy.
Why will US bond yields rise? Because the Chinese will have to let the yuan rise to choke off inflation, which in turn will curb the private sector savings surplus it recycles into US bonds. Because demand from China, India and (probably) Indonesia will broadly maintain demand pressure on a broad range of commodities. Because ‘Asian deflation’ in manufactured goods is broadly over (reflecting the slower pace of capacity build-up). Because Asia, particularly China, India and Indonesia, have only just embarked on the period of intense infrastructural build-up which invariably accompanies the arrival of a car-driving economy.
And because . . . . . because there are no ‘bond hawks’ out there, even now
The News Flow:
Japan: Sumitomo Mitsui FG reports that sub-prime related losses mounted to ¥99b in the 9 months to December, vs the ¥32bn it had reported in the 6m to Sept.
Japan: Production of vehicles rose 1.5 percent year-on-year in December to 969,457 vehicles, while domestic demand fell 9.8 percent year-on-year to 367,786 vehicles
Japan: Mizuho Securities is now likely to book as much as ¥250b in sub-prime mortgage-related losses in the current fiscal year, double the amount projected in November.
Japan: Industrial production rose a seasonally adjusted 1.4 percent month-on-month in December, the first gain in two months. Shipments rose 1.6 percent month-on-month and inventories fell 0.5 percent month-on-month. However, manufacturers expect output in both Jan & Feb will drop, and METI says output trend is “inclined to be flat.”
Japan: Elpida Memory lost ¥8.9b in 4Q, with sales down 34 percent year-on-year to ¥94b. In response, company plans to slow production of DRAM chips this year, before accelerating again in 2009 based on 65 nanocircuits. No chipmaker can afford to stay in business with the current prices, which are below production cost.
Korea: Kookmin Bank says it is negotiating with mid-tier Kazakh bank JSC Bank CenterCredit, hoping to buy a 30 percent stake this year for about W600bn, with a further 20.1 percent to be bought in 2011.
Korea KFI's business confidence index fell 8.2 pts month-on-month in December to 94.8, the first time in seven months that it has been below the boom/bust line of 100.
Taiwan Bilateral trade with mainland rose 15.3 percent year-on-year in Jan-Nov to US$92.68b, on which Taiwan registered a surplus of US$41.89b, up 19.4 percent year-on-year.
Taiwan MasterCard's biannual survey of consumer confidence gives Taiwan a score of 29.7, the lowest since SARS in 2003. Poor score reflects 'pessimism especially in terms of the prospects for employment, quality of living and the economy in 1H08.'
India:Following the Reserve Bank of India’s noting in its quarterly monetary report that banks' net interest margins are extremely high by international standards, commercial banks begin to mull bringing down loan and deposit rates even within structure of existing monetary policy
India: Net profit growth is slowing at the four big software companies in 4Q07: Tata Consultancy was up 18.9 percent year-on-year; Infosys was up 25.2 percent; Wipro was up only 11.6 percent; and Satyam was up 28.6 percent. These growth rates compare with the 40-50 percent rates seen during the same period last year.
Michael Taylor is an independent economist at Coldwater Economics