Tag Archives: correlation

Stoking the Embers of Inflation

This article was co-authored by J. Brett Freeze of Global Technical Analysis and Michael Lebowitz

Monetary Velocity, an oft-misunderstood metric that quantifies the pace at which money is spent, has recently shown signs of rising after trending lower for the better part of the last decade. Since increasing velocity is frequently associated with inflation, it comes as no surprise the Federal Reserve (Fed) has upped their vigilance towards inflation. While one would think higher interest rates and a reduced balance sheet both currently being employed by the Fed, would hamper inflation, there exists a well-known financial identity that argues otherwise.

In this article, we closely examine the Monetary Exchange Equation with a focus on monetary velocity.  Decomposing this simple formula and extracting the inflation identity shows precisely how the level of economic activity and the Fed’s monetary actions come together to affect price levels. This analysis demonstrates that the broadly held and seemingly logical conclusions are incorrect.

Might it be possible the Fed is stoking the embers of inflation while the world thinks they are being extinguished?

Monetary Exchange Equation

To understand how the Fed’s commitment to continued interest rate hikes and balance sheet reduction (Quantitative Tightening – QT) affect inflation or deflation, the Monetary Exchange Equation should be analyzed closely.  The equation is not a theory, like most economic frameworks based on assumptions and probabilities. The equation is a mathematical identity, meaning the result will always be true no matter the values of its variables. The monetary exchange equation is as follows:


The equation states that the amount of nominal output purchased during any period is equal to the money spent.  Said differently, the price level (P) times real output (Q) is equal to the monetary base (M) times the rate of turnover or velocity of the monetary base (V). The monetary base – currency plus bank reserves, is the only part of the equation that the Federal Reserve can directly control.  Therefore, we believe to form future price expectations, an analysis of the Monetary Exchange Equation using the forecasted monetary base is imperative.

The Inflation Identity

Through simple algebra, we can alter the Monetary Exchange Equation and solve for prices. Once the formula is rearranged, the change in prices (%P) can be solved for, as shown below. In doing so, what is left is called the Inflation Identity.

%P = %M + %V – %Q

Before moving on, we urge you to study the equation above. The logic of this seemingly modest formula is often misunderstood. It is not until one contemplates how M, V, and Q interact with each other to derive price changes that the power of the formula is fully appreciated. 

Per the inflation identity, the rate of inflation or deflation (%P) is equal to the rate of money growth (%M), plus the change in velocity (%V), less the rate of output growth (%Q).  The word “less” is highlighted because in isolation, assuming no changes in the monetary factors (%M and %V), inflation and economic growth should have a near perfect negative relationship.  In other words, stronger economic growth leads to lower prices and vice versa. While that relationship may seem contradictory, consider that more output increases the supply of goods, therefore all other things being equal, prices should decline. Alternatively, less output results in less supply and higher prices.

It is important to note that the inflation identity solves for the GDP deflator, which is one of the price indices on which the Fed relies heavily. While the equation does not solve for the more popular consumer price index (CPI), the deflator is highly correlated with it. The graph below highlights the perfect (correlation = 1.00) relationship between the deflator and the price identity as well as the durable, but not perfect (correlation = 0.93), relationship of CPI to the deflator and price identity.

Data Courtesy: Federal Reserve

Let us now discuss %M, %V and %Q so we can consider how %P may change in the current environment.

%M – As noted earlier, the change in the monetary base is a direct function of the Fed’s monetary policy actions. To increase or decrease the monetary base the Fed buys and sells securities, typically U.S. Treasuries and more recently Mortgage-Backed Securities (MBS). For example, when they want to increase the money supply, they create (print) money and distribute it via the purchase of securities in the financial markets. Conversely, to reduce the monetary base they sell securities, pulling money back out of the system. The Fed does not set the Fed Funds rate by decree. To target a certain interest rate they use open market transactions to increase or decrease money available in the Fed Funds market.

Beginning in 2008 with Fed Funds already lowered to the zero bound, the Fed, aiming to further increase the money supply, resorted to Quantitative Easing (QE). Through QE, the Fed bought large amounts of Treasuries and MBS from primary dealers on Wall Street. Largely through this action, the monetary base increased from $850 billion to $4.13 trillion between 2008 and 2015.

%V – Velocity is calculated as nominal GDP divided by the monetary base (Q/M). Velocity measures people’s willingness to hold cash or how often cash turns over. Lower velocity means that people are hoarding cash, which usually happens during periods of economic weakness, credit stress, and fear for the going-concern of banking institutions. In contrast, higher velocity tends to result in people avoiding holding cash.  This typically happens during times of economic growth, lack of credit stress, and rising interest rates.  During such periods, the opportunity cost of physically holding cash increases, as cash holders are incentivized by rising interest rates on deposits and/or productive returns on money in other investments.

Unlike the monetary base, velocity is influenced by the Fed through interest rates but not directly controlled by the Fed.  The graph below shows the relationship between the Fed Funds effective rate and velocity.

Data Courtesy: Federal Reserve

The deflation of the Great Depression occurred as credit stress, weak economic growth, and bank failures created an acute demand by the public to hold money. It was deemed safer to stuff your mattress with cash than to trust a bank to hold it for you. The effect was a sharp decline in velocity (%V). When coupled with inadequate growth in the monetary base (%M), the combination overwhelmed the inflationary impact of lower output growth (%Q).  The result of this effect was deflation (%P).

Similar to the Great Depression, velocity dropped precipitously during and coming out of the Great Financial Crisis of 2008.  Determined not to make the same perceived mistake, the Fed under Ben Bernanke increased the monetary base substantially. After cutting Fed Funds to zero and executing three rounds of QE, its balance sheet increased from $800 billion to over $4 trillion as shown below. Partially as a result of these actions, the GDP Deflator never registered a negative year-over-year reading but, and this point is critical, neither did it spike higher as was forecast by critics of QE.

Data Courtesy: Federal Reserve

%Q – Like velocity, the level of economic output (Q) is not directly set by the Fed, but it is influenced by their actions.  The supply and demand for credit, and therefore related economic activity, ebbs and flows in part based on the Fed’s interest rate policy. Historically the Fed will raise rates when economic growth “runs hot,” and inflationary pressures are on the rise. Alternatively, the Fed lowers rates to spur economic activity, incentivize borrowing, and boost inflation (or avoid deflation) when economic conditions are weak or recessionary.

The Fed’s influence on output (Q) varies over time as it is heavily dependent on the composition of economic growth. Currently, with demographics and productivity providing little support for economic growth, debt (be it government, corporate or household) has been a predominant driver of economic activity. In such an environment, one can presume that the level of interest rates plays a bigger role in determining output (%Q) than one in which debt is not the primary driver of growth. This factor helps explain why double-digit interest rates in the 1970’s, although painful, did not crush economic growth yet investors today are fretting over a 3.0% yield on Ten-Year Treasury Notes.

Current Monetary Dynamics

Since 2015 the Fed has increased the Fed Funds rate six times, bringing it from the range of 0.00-0.25% to its current range of 1.50-1.75%. In October of 2017, it began reducing the size of its balance sheet (QT). Thus far, they have allowed their balance sheet to shrink by $128 billion. To be very clear, it is this dynamic of the balance sheet reduction that alters the implications of Fed actions on the expected change in prices. This is a policy action with which our system is entirely unfamiliar.

Given that the Fed appears firmly in favor of continuing to tighten policy for the remainder of 2018 and throughout 2019, we assess how changes in the monetary base (%M) might affect inflation.

As a reminder: %P = %M + %V – %QTherefore, if we can accurately model the change in the monetary base (%M), velocity (%V), and output (%Q), we should be able to form expectations for the rate of price change (%P).

As stated earlier, recent economic growth has largely been driven by debt, both for current economic activity and the debt remaining from prior consumption. Given that higher interest rates disincentivize new borrowing and make servicing existing debt more expensive, the Fed’s actions should limit GDP growth. However, we must recognize that the surge in fiscal stimulus and recent tax reform should provide economic benefits to offset the Fed’s actions. Whether the Fed fully offsets or partially offsets fiscal policy is an important consideration.

This leaves us with velocity (%V). As mentioned, velocity tends to increase as rates increase and the money supply declines. After a long decline in monetary velocity, we are witnessing a change, albeit subtle thus far, alongside tightening monetary policy.

The recent low in velocity was achieved in the third quarter of 2014 at a level of 4.35.  The monetary base peaked in the same quarter at a level of $4.049 trillion. From that point of inflection, the Fed waited five quarters before beginning to raise rates in a slow, incremental fashion.  As expected, once the Fed began raising rates and subsequently reducing its balance sheet, velocity gradually increased further as shown below.

Data Courtesy: Federal Reserve

The scatter plot below highlights the near-perfect correlation (r-squared = 0.9414) between %M and %V.

Data Courtesy: Federal Reserve

While %M and %V are highly correlated, it is important to grasp that the directional changes of %M and %V are not one for one. Note the formula on the graph that solves for %V given a level of %M is as follows:  %V = (-1.0983 * %M) + 6.9543

For instance, a 5% increase in %M would result in a change in velocity of 1.4628 [(-1.0983 * 5 ) + 6.9543 = 1.4628]. Without regard for output growth, the monetary components of the identity would produce a 6.4628% ( 5 + 1.4628 ) increase in prices.  If we assume a 3% output-growth-rate, %P will equal 3.4628%.

The relationship between positive monetary growth and velocity is well known, as it has been the status-quo of inflation-forecasting for the better part of the last 60 years. Interestingly, however, the response of velocity (%V) is vastly different for a declining, as opposed to increasing, monetary base. Using quarterly observations beginning in 1960, the annual change in the monetary base (%M) has been negative in only 21 of 233 quarters (9%).  Given the infrequency of money supply declines, do policymakers, economists, and market participants fully understand the ramifications of a sustained decrease in the monetary base?  

The table below showing how velocity (%V) reacts to changes in %M, assuming constant output (%Q), helps us better understand the relationship between %M and %V.

First, as demonstrated above, a reduction in the monetary base has a much larger magnitude-of-impact on velocity than an increase in the monetary base. Second, there exists what we call a positive “convexity” effect.  At larger increments of percentage changes in the monetary base, the differential between the effects on %P widens. As shown, a 10% increase in %M, assuming a constant %Q, results in %P of 3.5%. However, a 10% decline in M results in a %P of 5.4%, almost 2% more despite an equal change in M. As shown, the “convexity” gap widens further when the money supply changes at greater rates.

Using the Federal Reserve’s guidance on the pace of QT, and assuming a constant %Q, we modeled the change in the monetary base (%M), the change in velocity (%V), and the resulting change in inflation (%P).

The forecast above is somewhat conservative as it is solely based on QT and doesn’t incorporate any further open market operations in the Fed’s quest to increase the Fed Funds rate.

This is where forming inflation expectations gets a little more complicated. If we assume the Fed follows through on their proposed actions, how much can economic growth offset increases in %P? When considering that important question, our primary concern, is that if economic growth weakens as a result of higher interest rates or other factors, the outlook is for higher inflation. In the example above, consider that by May of 2021 prices are expected to rise by 6.7% annually. Now recalculate that number for one percent (%Q) economic growth and %P increases further to 7.9%. On the other hand, assuming 4% economic growth would leave %P in May of 2021 around current levels of 2.7%.

Further Considerations

  • Economic Growth (%Q) – Weaker growth is inflationary, while stronger growth reduces inflation. Will economic growth stumble with higher interest rates? Will tax cuts and fiscal stimulus keep growth humming along despite higher rates?
  • Monetary Base (%M) – Close attention should be paid to the Fed’s pace of QT. Further, we must also gauge the Fed’s intent to continue raising interest rates as this action also reduces %M.
  • Velocity (%V) – Given the liquidity tightening actions the Fed is taking, and will likely continue to take, we should expect that the increase in %V has the potential to shock the markets, first bonds with equities close behind.
  • Federal Reserve – How does their tightening posture change based on the factors above? Will they realize the pitfalls embodied in their policy and change course? Will they make a grave mistake by ramping up their inflation vigilance, not understanding that they are the ones stoking inflation’s embers? Alternatively, might it be possible the Fed is trying to thread the proverbial needle by carefully balancing economic growth with the monetary supply?

The last thing the Fed wants to do is generate higher inflation with reduced economic growth, otherwise known as stagflation. As such, we must pay careful attention to Fed speeches and FOMC meeting minutes to glean a better understanding of how their policy expectations might change.

Stagflation, while historically rare, has proven to be an unfriendly investment climate for stocks and bonds. We think it is critical for investors to take their cues not only from Fed actions and talk, but also from economic growth outlooks and, maybe most important, changes in %V.


Assuming that further reductions in the money supply, higher velocity and weaker output ensues, we can confidently declare that inflationary pressures will increase. For investors, it is extremely important to be cognizant that such a conclusion runs counter to the popular narrative that slower growth and higher interest rates are deflationary.  We do not want to take too much for granted in assuming the economists at the Fed are aware of these dynamics, but the potential for the Fed and investors to be caught by surprise while inflationary pressures rise is palpable. If the Fed were to be stoking inflation under a belief they are taming it, such an event would be a central banking error of historic proportions.

As emphasized above, it is essential to note that the size of their balance sheet is significantly larger than it ever has been. Therefore, sustained reductions in the balance sheet stand to be more significant than anything witnessed in the past. Given the Fed’s tone, alongside the identity and the factors discussed in this article, the potential for sharply increasing velocity is a distinct possibility. We venture that the equity and fixed income markets will not look favorably upon such an event, nor the increasing potential for stagflation.

Perhaps it is time to reconsider the Fed’s actions in this new light.

Viewing Employment Without Rose-Colored Glasses

Fed Officials Worry Economy Is Too Good. Workers Still Feel Left Behind” – New York Times 4/27/2018

This coming Friday the Bureau of Labor Statistics (BLS) will release the monthly employment report. Consensus expectations from economists are for an unemployment rate (U3) of 4.1% which is nearly unprecedented in the last fifty years.

On April 26, 2018, the Department of Labor reported that a mere 209,000 people filed for initial jobless claims. This weekly amount was the lowest since 1969. The data point is the lowest in almost 50 years and remarkable when normalized for the number of people considered to be of working age (ages 15-64).

Low initial jobless claims coupled with the historically low unemployment rate are leading many economists to warn of tight labor markets and impending wage inflation. If there is no one to hire, employees have more negotiating leverage according to prevalent theory. While this seems reasonable on its face, further analysis into the employment data suggests these conclusions are not so straightforward. This was recently raised by the New York Times as highlighted in the lead quote above.

Strong Labor Statistics

The following chart highlights initial jobless claims adjusted for the working age population (ages 15-64).

Data Courtesy: St. Louis Federal Reserve

As shown above, there is only one person filing an initial jobless claim for every thousand people in the workforce. This is less than half the average (dotted line) of the last 50 years. Further, when one considers seasonal workers that will always be filing claims, regardless of the health of the economy, this number may be reaching the lowest point conceivable.  Regardless, the current low rate of jobless claims is unprecedented.

The U-3 unemployment rate, as calculated by the BLS, is also at a level that implies an incredibly strong labor market. Except for the year 2000 when it dipped to a low of 3.8%, the most recent reading of 4.1% is the lowest since 1969.

Adjusting Labor Statistics for Reality

The data mentioned above suggests that the job market is on fire. While we would like nothing more than to agree, there is other employment data that contradicts that premise.

If there are very few workers in need of a job, then current workers should have pricing leverage over their employers.  This does not seem to be the case as shown in the graph of personal income below.

Data Courtesy: St. Louis Federal Reserve

In addition to the weak wage growth, we are also troubled by another labor statistic, the participation rate. This indicator measures employed people and those “looking for work” as a percentage of those aged 16 and older. During economic recessions, the ratio tends to decline as unemployed workers get discouraged and stop looking for work. Conversely, it tends to increase when the labor market is healthy.

The participation rate graphed below shows that, despite nine years of economic recovery since the 2008 financial crisis, the participation rate has trended lower and clearly broken the trend from the prior 20 years.

Data Courtesy: St. Louis Federal Reserve

Closer inspection of the BLS data reveals that, since 2008, 16 million people were reclassified as “leaving the workforce”. To put those 16 million people into context, from 1985 to 2008, a period almost three times longer than the post-crisis recovery, a similar number of people left the work force.

Some economists may be tempted to push back on this analysis by claiming the drop in the participation rate is attributable to a large number of baby boomers retiring. While it is true 10,000 boomers will reach age 65 daily from the year 2011 to 2030, we must also consider that 11,500 children a day will turn 16 during that same period. Not all 16-year-olds will seek work or be gainfully employed, but we must also consider that many baby boomers will stay in the workforce.  According to recent Pew Research surveys, boomers do not believe “old age” begins until the age of 72. Taken together, this suggests the demographic explanation may not explain the inconsistencies found in the “full-employment” assumption.

Why are so many people struggling to find a job and terminating their search if, as we are repeatedly told, the labor market is so healthy? To explain the juxtaposition of the low jobless claims number and unemployment rate with the low participation rate and weak wage growth, a calculation of the participation rate adjusted unemployment rate is revealing.

When people stop looking for a job, they are still unemployed, but they are not included in the U-3 unemployment calculation. If we include those who quit looking for work in the data, the employment situation is quite different. The graph below compares the U-3 unemployment rate to one that assumes a constant participation rate from 2008 to today. Contrary to the U-3 unemployment rate of 4.1%, this metric implies an adjusted unemployment rate of 9.1%.

Data Courtesy: St. Louis Federal Reserve and Real Investment Advise

Enter the Phillips Curve

The Phillips curve, named after William Phillips, is a simple measure describing the inverse relationship between the unemployment rate and wage inflation. The logical premise behind the Phillips Curve is that, as unemployment drops and workers become harder to find, workers can demand higher wages. Conversely, when unemployment rises, the supply of workers is greater, and therefore wages fall. The Phillips curve follows the basic tenets of the supply and demand curves for most goods and services.

Many economists and media pundits have pronounced the Phillips curve relationship dead as it relates to employment. They deem it an economic relic that has ceased to provide reliable results. Has a basic, time-tested law of supply and demand ceased to work in the labor markets, or are economists measuring the inputs incorrectly? 

There are a large number of social and economic factors that affect wages and the supply of workers. We do not ignore those factors, but it is a good exercise to observe the Phillips curve relationship if one uses the more “realistic” unemployment rate (9.1%) shown above. Further, we substitute wage growth one-year forward for the traditional method of using current wage growth. The logic here is that it takes time for employees to apply the leverage they gain over employers to boost their income.

The first graph below shows the traditional Phillips curve as typically displayed (U-3 and recent three-month wage growth). The second is a modified Phillips curve which uses the adjusted U-3 from above and one-year forward wage growth.

Both graphs contain their respective R-squared (R²), which shows the statistical relationship between the two factors. The traditionally calculated Phillips Curve (first graph) demonstrates that only 28.84% of the change in wages was due to the change in the unemployment rate. Visual inspection also tells us the relationship between wages and unemployment is weak. It is this graph that has many economists declaring the Phillips curve to be irrelevant. The second graph, with our adjustments, is statistically significant as 70.47% of the changes in wages were due to the change in the unemployment rate. This graph visibly confirms that the Phillips curve relationship for employment continues to hold when more representative data is used.

Recently, Federal Reserve Bank of Chicago President Charles Evans stated, in relation to the Phillips curve, “We don’t have a great understanding of why it’s gotten to be so flat.” Mr. Evans, perhaps employment is not as strong as you and your Fed colleagues think it is.

If one believes that the laws of supply and demand continue to hold true, then the revised Phillips curve graph above argues that the unemployment rate is in reality much closer to 9% than 4.1%. To believe that the Phillips curve is useless, one must be willing to ignore a more rigorous assessment of labor market and wage data. The only reason economists and Fed officials voluntarily ignore this data is that it belies the prettier picture of the economy they wish to paint.


One of the main factors driving the Federal Reserve to raise interest rates and reduce its balance sheet is the perceived low level of unemployment. Simultaneously, multiple comments from Fed officials suggest they are justifiably confused by some of the signals emanating from the jobs data. As we have argued in the past, the current monetary policy experiment has short-circuited the economy’s traditional traffic signals. None of these signals is more important than employment. Logic and evidence argue that, despite the self-congratulations of central bankers, good wage-paying jobs are not as plentiful as advertised and the embedded risks in the economy are higher. We must consider the effects that these sequences of policy error might have on the economy – one where growth remains anemic and jobs deceptively elusive.

Given that wages translate directly to personal consumption, a reliable interpretation of employment data has never been more important. Oddly enough, it appears as though that interpretation has never been more misleading. If we are correct that employment is weak, then future rate hikes and the planned reduction in the Fed’s balance sheet will begin to reveal this weakness soon.

As an aside, it is worth noting that in November of 1969 jobless claims stood at 211,000, having risen slightly from the lows recorded earlier that year. Despite the low number of claims, a recession started a month later, and jobless claims would nearly double within six months. This episode serves as a reminder that every recession followed interim lows in jobless claims and the unemployment rate. We are confident that the dynamics leading into the next recession will not be any different.

Nowhere to Hide

The following article was originally a PowerPoint presentation that highlights several aspects of recent price movements across assets classes and within equity industry sectors. Many investors are unfamiliar with these relationships and their importance. While the current correction may prove only to be a speed bump on the way to higher prices, close inspection of asset class and security interactions often hold important clues about the future. The information contained in these pages argues for caution.

Click on any table below for a full size image making them easier to read.

Nowhere to Hide

The messages from shifting cross asset and S&P 500 sector correlations

Why Correlations Matter

Correlation is a statistical measure that quantifies the relationship between two financial assets or securities. A correlation of +1.0 is perfect, meaning the two securities or assets move one-for-one with each other. A correlation of -1.0 means they move exactly opposite of each other. As correlations move away from +/- 1.0 the relationship weakens. A correlation of zero quantitatively implies no relationship in the movement between the two instruments.

Correlations between and within asset classes plays an integral role in portfolio management. From a big picture perspective, changing correlations can be a signal that broader market trends are changing. Investors may reduce risk during such periods. Also of importance, changing correlations may increase or decrease the value of “hedges” within a portfolio. For instance, investors tend to assume they are taking a more defensive posture moving technology to utility stocks, or from stocks to bonds when they sense a downturn coming. While such trades have been effective in the past, correlations allow us to observe changes and develop opinions about the future.

The following charts and notes provide recent and historical context on how correlations have changed since the equity market turned lower in late January. Whether these changes turn out to be a dependable warning of trend change, or a multi-month anomaly, is unknown. What is known is that the market is not behaving as it has for the last few years and investors should pay close attention to correlations for more market insight.

Under Appreciated Price Action

This graph, courtesy of Goldman Sachs, shows how correlations between S&P 500 stocks have increased at a rate greater than anytime in the last 40 years except 1987.

Cross Asset Correlations

S&P 500/UST Correlation

Given the popularity of formal and informal risk parity strategies, this graph showing the well below average correlation of the S&P 500 versus 10 year UST yields should be of vital concern if this equity sell-off continues and the correlation remains low.

S&P 500 Sector Correlations

To further highlight the uniqueness of current sector correlations versus the S&P 500, this graph compares the current period (green dots) versus the prior year (orange squares) and the prior 15 years (gray triangles).

Something is different this time

This graph serves as a reminder that passive investing has grown significantly over the past 10 years. In our opinion this popularity will play a role in making it more likely that correlations between asset classes and sectors will behave differently in the next downturn than they have in the past. As such alternative hedging strategies should be considered now.


  • Passive funds and strategies have increased the likelihood that future correlations between asset classes and the S&P 500 and its constituents are higher
  • Equity and fixed income correlations have increased recently, rendering fixed income hedges for equities not as dependable
  • Gold and commodities as measured by the CRB index have also not been as good an equity hedge as in the past
  • Long equity volatility (VIX) has thus far proven a good ballast for stock and fixed income hedging
  • Traditional safety, low beta, equity sectors have been well correlated to other sectors and the equity market as a whole
  • Higher beta equity indexes (Russel 2k and the NASDAQ) have moved nearly perfectly in line with the S&P 500
  • It is too early to tell if the market is topping or just taking a breather, but the signals discussed in this article and others we did not highlight, should be taken seriously