Balancing act

The level of output which an economy can produce at a constant level of inflation is its potential GDP. The potential GDP of an economy is dependent on monetary factors and its productivity. An economy’s demand for products can differ from the its ability to produce those products for various reasons. When it does, either an inflationary or recessionary gap develops.

For example, when consumers or governments spend at high levels and businesses invest, the demand for and production of, goods can temporarily exceed an economy’s potential to produce them. The economy will ultimately produce the goods, however at increased prices. In this situation actual GDP temporarily exceeds potential GDP causing an inflationary gap. The inflationary forces and authorities reactions to them, ultimately cause the gap to close. When consumers or governments do not spend and businesses invest at low levels the process is reversed, actual GDP is below potential GDP and a recessionary gap develops. The deflationary forces and authorities reactions to them ultimately close this gap as well.

Despite what most economic models assume, economies rarely operate in equilibrium where actual and potential GDP are equal. Instead they are in a constant state of flux: flipping between inflationary and deflationary gaps. Monetary and fiscal authorities seek to balance the gaps and reduce extremes through the use of fiscal and monetary policies. 

Since the 2007 financial crisis the US economy has been in a recessionary gap (chart 1), much of the world has as well. The US government and the federal reserve system has been attempting to spur spending and investment through low interest rates, asset purchases, and fiscal stimulus. While many uncertainties remain, more observers are beginning to see green shoots for US and global economies. Since an inflationary gap may soon develop, now is a good time to take a quantitative look at inflationary gaps to determine if they hold any predictive ability for equity markets. In this edition of Quantessential views we take a statistical approach to inflationary gaps and develop trading strategies around them. Specifically, we attempt to determine whether investors can accurately predict a market top during an inflationary gap period.

Chart 1: Inflationary and recessionary gaps

Source: FRED, Yahoo, Vital Data Science Inc.

The quantitative approach to trading inflation gaps

Since 1945 there have been 9 periods of inflation gaps and most have ended in a recession. The median duration of an inflationary gap has been 11 quarters, the mean duration is 10 quarters (indicating a probability distribution with a slight negative skew), and the standard deviation of inflationary gap durations is almost 7 quarters.

The reason why inflation gaps generally end in recession, and hence usually contain a market top, is a result of strong economic conditions which create inflation. Central banks respond to the inflation by raising interest rates. At some point the combination of increased debt costs and price escalation cools the economy off, decreases demand, and ultimately causes a recession. Before the recession, equity markets usually start to decline. During inflationary gaps, interest rates generally begin by trending up, then trend down as the economy begins to cool.

Vital Data Science analyzed inflation gaps, interest rates, and SP500 prices using a factorized bayesian model in a attempt to determine advantaged periods for long/short equity positions. The results are summarized in chart 2.

Chart 2: Probability distribution of inflation gaps bayesian model

Source: Vital Data Science Inc.

With data going back to 1945, a 3 node bayesian network model was developed using two parent nodes, Fed fund rate and actual vs potential GDP, and one child node, SPY. Fed funds rate was factored into two levels: up trend and down trend. Actual vs potential GDP data was factored into three levels: inflationary gap - early stage ( <= 6 quarters), inflationary gap - late stage (>6 quarters), and deflationary gap. SP500 data was factored into three levels: up trend, top, and down trend. Factorization was done using a proprietary algorithm, which characterized trends by looking at local bottoms and tops.

The goal was to determine if any combination of levels in the parent nodes would yield higher probabilities of success in either long/short SPY positions. As shown in chart 2, it is that it is difficult to predict market tops accurately based on interest rate action and inflationary/deflationary gaps alone. Further, regardless of inflationary gap, deflationary gap, and interest rate action, equities tend to trend upwards. However, there are times when a market top is more likely to occur. If an inflationary gap has gone on for 6 or more quarters and interest rates are decreasing, then based on historical market behaviour there is a 30% of hitting a market top and 70% chance the market is already in a downtrend - not a good time to be long equities. Bear markets followed by recessionscan last longer than a year making this information is useful and investable.

Our analysisalso showed that market bottoms occur most frequently during recessionary gaps and when interest rates are decreasing, proving Warren Buffet’s quote, “be greedy when others are fearful, be fearful when others are greedy” statistically accurate. Early stages of inflationary gaps (<6 quarters) are generally bullish equities regardless of the interest rate environment. Timing inflection points within inflationary gaps or interest rates (i.e. when an uptrend changes to a down trend) had no predictive abilities according to our analysis.

Application in today’s market

Our simplistic inflationary/deflationary gap and fed funds rate model predicts likelihood of a continued uptrend in equity markets, inline with calls form many analysts. We, however, remain a little cautious. There are many variables which this particular analysis did not consider: high equity valuations, high debt levels (globally), low global GDP growth, and high geopolitical tension and political gridlock in the US. When we layer these variables into our analysis, the result is less certain.

We have positioned ourselves long US equities (using a value approach to stock selection). We are hedged with a position in gold bullion, and as we have closed out our short CDS swap position recently we are holding a larger than normal cash position as we search for opportunities - the topic for the next post.


Any material provided in this note is for general information use only. You should not act based solely upon the materials provided herein. Vital Data Science Inc. advises you to obtain professional advise before making investment decisions. Your use of these materials is entirely at your own risk. In no event shall Vital Data Science Inc, its officers, directors or employees be liable for any loss, costs or damages whatsoever.