See the latest Gavekal book Avoiding the Punch: Investing in Uncertain Times

Click here Close
Gavekal Research

Four Quadrants: The Growth Question

The chart below shows my time-honoured Four Quadrants matrix. The idea is that there are four basic investment environments depending whether economic activity is expanding or contracting, and whether prices are rising or falling. It is usually obvious which state we have just left, but far harder to judge what state we have entered. Making such a call in real time has been a holy grail of my research and in the first part of this series (see Four Quadrants: A Wicksellian Analysis), I outlined a decision rule to show whether the US economy was in the top or the bottom half of the grid—i.e. to know if prices were structurally rising (top half) or declining (bottom half). The aim of this paper is to show whether the US is on the left side of the matrix (recession or depression), or instead on the right side (actual growth in output).

In my good Wicksellian framework when the spread between the real rate on Baa-rated bonds (computed by deflating the average inflation of the last ten years) and the GDP growth rate is above 250bp, then I consider a recession to be imminent. What this decision rule is telling me is that the “market” rate of interest is significantly above the “natural rate” and so economic activity is sure to contract. Indeed since 1955, the US has experienced nine “official” recessions, and each one was preceded or accompanied by my “Wicksellian spread” going negative (see chart below). The only “false” signal came in 1989-91 when the effect of Communism’s collapse in Eastern Europe was to drive German rates up and leave the US dollar undervalued.

Interactive chart

In economics, as with physics, there are “thresholds” between distinct economic states. At 1° Celsius water is liquid; at 0° Celcius it becomes ice. Similarly, when my Wicksellian spread is 240bp nothing special happens. At 250bp or above, the economy stops. In my opinion these transition phases are an area where far too little research has taken place.

Volatility is the key issue

If inflation picks up when my Wicksellian spread moves below zero and a recession is triggered when it breaches 250bp, then it follows that the “natural rate” is between those two limits. I have previously called this range the “Wicksellian optimum”. Since the 1950s, the only sustained period when the US stayed within that range was 1990-2002, also known as the “great moderation”. Since 2002, while staying broadly in the range, the volatility of the variations has greatly amplified. I conclude that while the level is very important so is the volatility of the results.

The Wicksellian spread for the US economy is now close to the 250bp critical level which could be breached either by the Baa rate rising or the economic growth rate falling, or a mixture of both. So, how to use this information? Remember the first part of this series showed that the US is in the lower deflationary/disinflationary section of the Four Quadrants matrix. The next step is to “cross” my two decision rules (relating to inflation and economic activity) to determine the historical progression through the Four Quadrant matrix, namely;

1) Inflationary growth (shaded orange in the chart below)  

2) Inflationary recession (shaded blue)

3) Deflationary/disinflationary growth (shaded green, this is where the biggest bull markets take place)

4) Deflationary/disinflationary contraction (shaded pink, being the most dangerous for the equities markets)

Interactive chart

Since the 1950s the US saw long episodes of inflationary growth, which lasted until the bust of 1980-82. This period saw a fair few inflationary recessions (shaded blue), which tended to coincide with actual recessions (shaded grey). There were also a good number of disinflationary contractions (shaded pink).

Starting in 1985 (I would contend 1982), there has been a long period of disinflationary growth, and during this time busts have been of the deflationary variety; moreover these periods have tended to be very difficult for equities. Next I apply these insights to portfolio construction using the chart below which shows the relative performance of US equities and bonds.

  • From 1950 to 1982-1985, the core asset class in a portfolio had to be equities. Bonds should have been added only when a recession was looming ( i.e. when the Wicksellian Spread hit 250bp).
  • Since 1985 (and arguably since 1982) the tables have been turned. The US asset of choice has been long-dated bonds; equities only made sense when they were extremely cheap such as in 2012-13.
  • From 1955 to 1982, a 50:50 “balanced portfolio” proved to be a disaster by hugely underperforming the equities market. Since 1982, however, US long-bonds have outperformed equities on a total return basis by 66%, and as a result a balanced portfolio has outperformed both assets with a lower volatility than its components. The explanation is simple: in a disinflationary period, bonds and equities have similar returns but a negative correlation, while in an inflationary period they have a positive correlation, with equities outperforming massively. This outperformance of a balanced portfolio has lasted since 1982 for almost every ten year period.

Interactive chart

The problem with this methodology is that the lags inherent to the US statistical system are long and the final version of the GDP often differs sharply from the first estimate. Hence I am proposing a more timely proxy for the Wicksellian spread in the shape of my US recession indicator (see Gauging The Chances Of A US Recession). The chart below shows the relationship between the two “decision tools”.

Interactive chart

Since my US recession indicator has historically led the Wicksellian spread by about 100 days and with a very high correlation, the result should be a tool that is less theoretically pure, but hopefully more timely. This point can be illustrated by considering the US government bond market in the chart below.

Interactive chart

Interestingly, when I substitute my US recession indicator for the Wicksellian Spread, the tool posits that since 1982 the US experienced a number of inflationary busts, while none were registered using my Wicksellian spread. Moreover, the tool shows all of these recessions ending with a disinflationary bust. Since 2002, the growth periods using  the revised tool are mostly recorded as inflationary, while they are largely  disinflationary using the Wicksellian spread (see the second chart above). So what gives between the two signals? I tend to believe that the US recession indicator offers a better facsimile of the underlying realities, yet from an asset allocation perspective the timing of the shift toward the left side of the matrix does not change much between the two approaches.

The red line in the immediate chart above represents the normalized deviation from the actual bond yield and its model, centred at zero (see What To Do With US Bonds). It can also be seen that each time a disinflationary bust unfolded in the US, there was an accompanying huge decline in long bond yields. Should the US be looking at a re-run with a new deflationary bust looming then US long bond yields are likely to fall from about  3% at the start of this year to below 2.20%, for a 40% or so capital gain. While the experience in Germany and Japan has been that long bonds lose their diversifying power when yields reach 1.3% or below, I would contend that long-dated treasuries yielding 3% still offer diversifying power.

The next step is to look at the performance of the S&P 500 versus its normalized model. Once again, I will use my usual valuation model for the S&P 500 which is shown in the left hand chart below (see Philosophical Dominance, Profits And Stocks).

Interactive chart

The pink periods are dangerous for the stock markets, while the blue ones (inflationary recessions) are a little less dramatic. At the start of most pink periods, the stock market has tended to be overvalued and the bond market undervalued. This may explain the negative correlation which has been visible between the two assets in the last two financial crises.

Interactive chart

The great bull markets have occurred during disinflationary booms (green periods), while inflationary booms have only offered decent equity returns when the market started from an undervalued level. Also, since inflationary booms tend to end with a deflationary bust, it takes courage to stay invested during such a period.

Lastly, consider the relative performance of US equities compared to long duration treasuries. What is clear is that during both blue and pink periods (inflationary  and disinflationary busts) long-bonds have hugely outperformed equities. It should also be noted that between 1981 and the present, US shares have underperformed long-dated treasuries by a robust 75%, while their relative performance has been about the same since 2003, even though this phase contained two lengthy growth periods. Also, the recent inflationary periods spurred unsustainable bubbles in housing followed by an ongoing bubble in credit and financial assets. It is my contention that the misallocation of capital in this cycle took place less in the equity markets, than in the credit markets. The most recent boom, in particular, was the direct outgrowth of ZIRP and resulting scramble for yield. Hence, it is in the credit markets where I would expect the most damage once the tide eventually fully recedes.

However, just because the core of the problems lies with credit, this does not mean that equities are immune to trouble. Since we have entered a “pink” disinflationary bust period, it is likely that both shares and low quality bonds will massively underperform. Equities have already started to underperform, being down already 12% versus their relative peak, just before we entered the pink phase last year. Moreover the high yield bond benchmark has been making new lows for the cycle in recent days.


After many attempts, this is the first time that I have found a credible method to “map” the four quadrants such that a real time positioning is possible. During the course of this exercise I have checked many other financial relationships not detailed in this article. However all of them point to the same conclusion that the US is entering a dis-inflationary bust period.

As such, few assets will do well other than US government bonds and, during the crash period at least, gold. Also, during the crash-period the trade-weighted dollar can be expected to rise, potentially substantially. I would strongly advise investors not to hold an equity portfolio that is not hedged with US treasuries and if this is not possible they should substantially raise cash, which should be kept in US dollars.

Interactive chart