Central Banking’s Fundamental Misunderstanding of Inflation

Central banks operate under the premise that real growth is exogenous, except for policy mistakes in moderating inflation and the business cycle. The assumption is that they have dominating influence on the impulses to core inflation and the output gap, both controlled through the credit channel.

Many people since 2008 have questioned the elasticity of the output gap to the credit channel. That is not a unique criticism. The typical mechanical explanation for this has been the liquidity trap, which is a recent event, and as of the end of 2012, not relevant to a contemporary description conditions.

Central banks target core inflation, believing the highly volatile food and energy components to be subject to supply shocks, and hence exogenous to their system of control. What they are effectively doing here is taking responsibility for the sticky consumer prices (those that appear inelastic to changes in supply or demand curves), while the flexible prices are assumed beyond their scope of influence. This can be framed by observing that Sticky CPI is virtually the same as Core CPI, and Flexible CPI is, in effect, Headline CPI.

Correlation co-efficient matrix of sticky CPI, flexible CPI, core CPI, headline CPI, and the Fed Funds rate.

Correlation co-efficient matrix of sticky CPI, flexible CPI, core CPI, headline CPI, and the Fed Funds rate.

Flexible CPI is highly sensitive to changes in both supply and demand. Yes, negative supply shocks may be exogenous from Fed control, but demand is very cyclical – which is something that the Fed assumes some control over. This means that the Fed assumes that, through changes in interest rates impacting the credit channel, they can exert influence on the output gap to target sticky CPI.

This is where the argument begins to unravel.

The output gap is virtually uncorrelated with sticky CPI (-.05). It is, however, more correlated (but not particularly well correlated) with flexible CPI (-.18). Thus, it seems that:

  • The Fed can’t really control sticky CPI
  • The Fed can control flexible CPI to some degree via the credit channel on consumption and investment (demand curves), as long as the supply curve remain stable
  • Supply shocks can dominate the variation of flexible CPI

Assuming our intermediate conclusions are true, we have a Fed which ignores flexible prices, instead uses sticky prices as the primary input, but really only is able to materially impact flexible prices, and further only to the degree that there aren’t supply shocks.

We are now compelled to consider core and sticky CPIs are exogenous, making most of CPI exogenous, and limiting central bank control mostly to the spread between sticky and flexible prices.

To test our theory, we can apply impulse response function on the Fed Funds rate to Flexible less Sticky CPI. If our theory is true, it should be negative, while the impulse response of Fed Funds on Sticky CPI should be flat to positive.

Screen Shot 2016-02-28 at 2.18.49 PM Screen Shot 2016-02-28 at 2.23.30 PM

If variance in flexible prices is largely described by supply shocks, then it stands to reason that impulses on core inflation are also supply related, but are more acyclical.

This is all problematic to the assumptions under which modern monetary policy is conducted. If Fed can only effectively target the spread between flexible and sticky prices, it’s never actually addresses the underlying driver of inflation, and only exercises a much broader band of control over inflation via the output gap through rate shocks.

Posted in Uncategorized | Leave a comment

US Recession Callers Are Embarrassing Themselves

Through a combination of quackery, charlatanism, and inadequate utilisation of mathematics, callers for US recession in 2016 are embarrassing themselves. Again.

The most prominent reason for recession calling may well be the Institute of Supply Management’s Manufacturing Purchasing Manager Index. The problem with this recession forecasting methodology is that it doesn’t work.

Figure 1: ISM PMI probit model forecasting recession

Figure 1: ISM PMI probit model forecasting recession

The first problem is statistical. The squared correlation coefficient between recessionary state and ISM PMI is 0.33 (33%) with a standard error of 0.29 (29%). This means that, when you calculate the recession probabilities from ISM PMI, they are only good within +/- 29%. Ouch.

The second problem is that of false positives. We can easily eyeball nine prior examples of the present implied recession probability being matched or eclipsed, but generating no recession. Ergo, using ISM PMI to forecast a recession is a Type I error.

Finally, when the standard error of the model (29%) is most of the useful threshold for determining recessionary state (43%). And, of course, the last reading of the model (26.5%) is less than the standard error.

If anyone tries to talk to you about recession probabilities using ISM PMI as data to back their argument, they’re doing it wrong.

How is any of this even possible, you ask? The answer is that manufacturing is not a material relative to the size of the balance of the US economy. Evidence of this is observed by regressing Real GDP against manufacturing and service ISMs.

Figure 2: Using ISM surveys to forecast/now-cast RGDP

Figure 2: Using ISM surveys to forecast/now-cast RGDP

The coefficient for service ISM is over 36 times that of the manufacturing ISM. In fact, if you want to forecast recessions using ISM data, non-manufacturing ISM gets you much closer.

Figure 3: Probit model for recessions using non-manufacturing ISM

Figure 3: Probit model for recessions using non-manufacturing ISM

The squared correlation coefficient is 0.48 (48%) vs 0.33 for manufacturing, and the standard error is 0.23 vs 0.29 for manufacturing. The useful threshold for determining recessionary state (54%) is 2.3x the standard error, instead of just 1.48x for manufacturing ISM. Unfortunately, the data only starts in the late 1990s, so we are unable to make a meaningful comparison of false positives.

Oh, and the last reading of recession probability from non-manufacturing ISM is 5.4%.

If we’re going to use just a single variable for recession forecasting, then we should use a high signal, leading variable. The Leading Index for the United States, which is an amalgamation of state leading indexes, which themselves are a compilation of individual state-level economic indicators, is a good bet. Of course, it’s probably cheating a little methodologically, because it already is composed of many variables, but it makes for easy econometric work. Let’s try it.

With a squared correlation coefficient of 0.53, it’s the best fit of them all. The standard error, 0.21, is very small relative to the recession threshold of 59%. The last reading is a 6% probability.

If we’re serious, however, we won’t just use one variable. Univariate analysis makes for good charts and slides, but it’s a poor analytical method. It is also far more useful to look at recessionary probabilities simultaneously at different forecasting intervals. Here is a probability estimate using a multivariate model of my favourite leading indicators across 5 time-frames.

Figure: Recession probabilities

Figure 5: Recession probabilities

Most 2016 recession callers already exist on a diet of crow from recession calls in prior years. I’ve been debunking their calls for years. This is just one more episode.

Posted in Uncategorized | 11 Comments

The Time For a Hike is September

Back in December of last year, in the midst of the downdraft of oil, I wrote that the deflation in headline prices we were about to see was symmetrically opposite to what we saw in the energy-driven headline inflation during 2011:

In April of 2011, Ben Bernanke was universally lambasted and lampooned for claiming that inflation, which was accelerating and running above 3%, was “transitory”. He used this view to justify loosening monetary policy. The next few months of CPI were not favourable to the Fed chairman’s views: it peaked at 3.8% (nearly double the implicit target at that point) in September of 2011, sparking a feverishly pitched cacophony of criticism that the Fed chair was out of touch, and tone-deaf in his theoretical ivory tower to the practical realities on the ground.

This, however, proved to be Bernanke’s finest hour. Yes, even more so than the extraordinary measures taken during the height of the credit crisis. His detractors then, of which there were still many, included people and institutions on the brink that needed the Fed to extend them a hand. In September of 2011, the chairman stood very much alone in his call for moderated inflation now that the acute disaster removed influential institutions and people from needing the Fed to act in order to survive.

Oil has fallen farther and faster than most anyone, including I, anticipated. Headline CPI has consequently been nudged farther into negative territory than I’d expected. But what hasn’t happened: the deflation in oil has not bled into everything else, causing a broad and general price decline. Wages have accelerated. Median and trimmed mean measurements of CPI have already recovered — and exceeded — the 2% inflation target. The largest component of spending and price indexes, shelter, continues to accelerate, and is now at the fastest inflation rate since 2008.

Figure 1: Various measurements of inflation

Figure 1: Various measurements of inflation

Perhaps most convincingly, even headline CPI m/m annualized has normalised, despite the Sturm und Drang of the energy component. But, as I’ll argue further, investors and the Federal Reserve are far better off served consuming their energy from the Sturm und Drang period of Haydn.

The Taylor Rule is the most eloquent expression of monetary policy in the United States. It explains that interest rates should be optimized on two variables: inflation, and how far national income is from potential output. There is a third term, of course: the natural real rate. It is often supposed to be 2% in the long-run, and perhaps this is a reasonable assumption.

This is where we get disagreement amongst practitioners. If we were to take the common 2% natural real rate assumption as static fact, we have actually long been ready to hike:

Figure 2: Taylor Rule with 0%, 1% and 2% real natural rate assumptions

Figure 2: Taylor Rule with 0%, 1% and 2% real natural rate assumptions

Indeed, a small fraction of poor interpreters of the Taylor Rule have called for hiking as early as 2011. We know with reasonable certainty that the Federal Reserve made a spectacular call — and one that the European Central Bank simultaneously blew — to hold rates on account of the transitory energy inflation. The natural real rate was most certainly not 2%. Most importantly, the nature of the transitory energy-driven inflation is that it is actually detrimental to the natural real rate.

If energy-driven headline inflation is detrimental to natural rate, then an energy-driven headline deflation may in fact be constructive to the natural real rate. The most concrete reasoning we can estimate that the natural real rate is greater than 0%, however, is in the loan growth data.

Figure 3: Real liability growth. Red: non-financial corporate sector; Blue: household sector.

Figure 3: Real liability growth. Red: non-financial corporate sector; Blue: household sector.

It’s tricky to say exactly how high it would be. Ask yourself this: how elastic would those series be to changes to the Fed Funds rate?

The answer is: not very. Why? Because these liabilities are not priced at the cash rate. Corporate bonds, auto loans, mortgages, and the vast majority of economic sensitive lending are benchmarked off 5s, 10s and 30s. The Federal Funds Rate for next quarter barely matters, unless it is really wrong.

5s, 10s and 30s lifted off from ZIRP two years ago:

Figure 4: 5s, 10s and 30s

Figure 4: 5s, 10s and 30s

They initially lifted off somewhat excessively (as I pointed out in my August 21, 2013 piece). Short rates are going up already. Belly rates will be the sum of expectations on the short rates. Long rates will be more about reinvestment policy for the foreseeable future.

Economically sensitive borrowing has long forgotten about ZIRP — even with this downdraft in inflation expectations. That’s why it’s time to start assuming the natural real rate is no longer 0% — and if that’s no longer 0%, then by most accepted functions, such as the Taylor Rule, ZIRP is no longer appropriate.

The only excuse not to hike we are left with is the lack of apparent urgency to do it, as my friend Igor suggests:

I have each of a theoretical and empirical argument. We actually have no idea if ZIRP is hurting us – and we won’t until the after the fact. The Fed is not a purely reactionary institution, and it shouldn’t be. Rather, it has historically taken upon itself the obligation to smooth out the cycles, exchanging amplitude for magnitude, preferring longer, shallower cycles over shorter, deeper cycles. The idea is to hike earlier into strength so you don’t have to hike later into weakness.

But my best argument is an empirical one. I estimate how loose monetary and fiscal policy are relative to the variables used as inputs into their policy: unemployment, the output gap and inflation:

Figure 5: Measurements of Monetary and fiscal policy accommodation

Figure 5: Measurements of Monetary and fiscal policy accommodation

For the macro variables as they currently stand, both monetary and fiscal policy are looser than their historical analogues. Monetary policy is particularly loose: around 1.8% looser than similar conditions historically.

The first few hikes are not a binary switch that flip us to tight policy. The tightening cycle already began during the Taper Tantrum. A 25bps hike in the Fed Funds rate is but a small adjustment to an analogue dial which has been already long expected by longer interest rates. Monetary policy will still be phenomenally — historically — loose.

I don’t know if the Fed will hike in September. Before very recently, I’d have guessed that it will. But they should. And, after the brief Sturm und Drang, the last vestige of the financial crisis will sneak out the back door and go home to his place in the economic history books. Hopefully it will not be viewed as too late, but it will join its long-time friend, fiscal policy, who left the party too soon.

Posted in Uncategorized | 3 Comments

Breaking the real rho even

The only rational way to value capital is by estimating your required risk premium against a safer benchmark expected return. This is why the correlation between investment grade bonds, Treasury yields and the S&P 500 earnings yield follow such similar shapes.

Screen Shot 2015-06-06 at 7.48.31 AM

There are many forces pulling these yields together. Investors choose between bonds and stocks, along with safety and return seeking. The corporate sector chooses to capitalize itself by the cheapest option – issuing bonds or equity.

When the risk-free yield goes down, if all other things are equal (principally, the equity risk premium), the yield on equities will follow. If the risk-free yield declines more than the equity earnings yield, the equity risk premium is rising.

Despite a tight long-term correlation, the shorter-run change in these same yields are negatively correlated. From 2004 (when our TIPS data begins), the correlation between nominal 10y yields and the equity earnings yield is -0.4.

The 10y Treasury yield serves as a discount rate or benchmark for equities. We can axiomatically specify the difference between the 10y and equity earnings yield as the sum of:

  • Expected differences in future cashflows (earnings growth in this case)
  • The market’s required risk premium for assuming equity risk

In the short-run, we expect new information about the economy to favour or disfavour these terms, driving the competition between these two yields.

Correlation matrix between 1y change in yields

Correlation matrix between 1y change in yields from 2004 to present

The expected differences in future cashflows between fixed income and equity instruments can largely be explained by expected inflation. When we isolate the breakeven inflation from 10y nominal yields, the negative correlation against the equity earnings yield widens to -0.74. Change in nominal yields, however, are more largely explained by change in real yields: the correlation there is 0.68.

Screen Shot 2015-06-06 at 7.54.47 AM

Where this leads us: nominal 10s are the defacto discount rate (rho) to all other assets. When the common discount rate goes up, all other things being equal, we can expect all yields to follow. The exception is in the equity risk premium simultaneously contracting. When reals adjust upward, rho adjusts upward for everyone. When breakevens adjust upward, inflation sensitive assets experience a positive counteracting force.

When rates rise, whether it is manifested out of breakevens or reals is the fulcrum upon which way the earnings yield sways.

Posted in Uncategorized | Leave a comment

Say ‘Good-bye!’ to the “New Normal”

“Secular Stagnation.” “The New Normal.”. “The New Neutral.” “Lower for Longer.”

These are the phrases which have been used to characterize the U.S. economic situation, and by proxy, interest rates.

The thinking is both reactionary and rational: the after-math of the largest and deepest financial crisis the world has experienced since the Great Depression is a long and deep hangover. The direct explanations for cause get a little murky here. Most agree it is some combination of lowered inflation expectations, an increase of risk-averse behavior by corporations and households, the aging of the dominant baby-boomer generation, and the slow repair of damaged household income- & balance-sheets.

Instead of having coming up with a list of why the United States will remain mired in this situation, purveyors of the new normal should instead have formulated a list of criteria required to get out. If they had, they wouldn’t be missing the economic ascent to escape velocity.

“Escape velocity!?”, you say? Yes. The proof is in the jobs pudding.

Nominal GDP — national income and output — is easily approximated by multiplying Average Weekly Hours x Average Hourly Earnings x Payrolls:

Figure 1: Nominal GDP vs Average Weekly Hours x Average Hourly Earnings x Payrolls

Figure 1: Nominal GDP vs Average Weekly Hours x Average Hourly Earnings x Payrolls

Nominal GDP, as estimated by January’s employment report, is now running at the fastest rate (5.18% y/y) since 2006. For January, that number is 8.35% annualized. The deflationary scare is just that — a scare. The growth looks even more impressive when deflated by the Core PCE Price Index (Core Personal Consumption Price Index):

Figure 2: Real GDP vs (Payrolls x Average Weekly Hours x Average Hourly Earnings) / Core PCE Price Index

Figure 2: Real GDP vs (Payrolls x Average Weekly Hours x Average Hourly Earnings) / Core PCE Price Index

At the time of writing, we lack the Core PCE data for January, but the liftoff is already evident as of December’s data: it was running at the fastest clip since the 2nd quarter of 2004.

For buyers of secular stagnation, this isn’t supposed to happen. Let us recast our earlier explanations of the “New Normal” as the criteria required to escape:

Reversing Lowered Expectation on Future Inflation

It’s certainly true that short-term inflation expectations have been quickly fallen by way of collapsing oil prices. Long-term inflation expectations have actually rebounded strongly since 2012, however, and have remained stable in the face of the great noise in oil:

Figure 3: Inflation expectations (5y & 30y)

Figure 3: Inflation expectations (5y & 30y)

Long-term inflation expectations remain firmly anchored just above the 2% target set by the Federal Reserve. The criterion that inflation expectations need to recover enough that investors and consumers aren’t rather incentivized to save for fear they could invest and consume more in real terms later seems satisfied.

Reversing Risk-averse Behaviour by Corporations and Households

We can easily observe the risk-averse behaviour of corporates by looking at the investment level. As confidence rises, the corporate sector deploys more capital into investment, above the value merely needed to sustain their existing business. Similarly, we can observe the personal savings rate of households as a proxy for their risk-aversion: the more they fear for their future earnings, the more they save.

Figure 4: Investment level (blue, LHS), Personal savings rate (red, RHS)

Figure 4: Investment level (blue, LHS), Personal savings rate (red, RHS)

The savings rate is at the lowest point since the end of the recession. The investment level is at the highest point since the end of the recession, and is even approaching the series average.

The criterion that the risk-aversion in the corporate and household sectors be reversed seems satisfied.

Reversing the Demographic Decline

The growth in peak-aged population occurred in 1986 — over 2.5 million per year were added to the 25-54 cohort. From 2008 through 2013, this cohort actually declined – the losses on a y/y basis were at their highest, 720k, in December 2011.

Figure 5: Growth in 25-54 population

Figure 5: Growth in 25-54 population

The decline is over. The Millennial generation is now growing into this peak-aged cohort faster than Baby Boomers are aging out. This trend will continue. This criterion is satisfied.

Repairing the Household Balance Sheets

The crest of the household debt bubble saw Household Debt Servicing as % of Disposable Income at over 13%. It has declined all the way down to less than 10% — the lowest recorded level.

Figure 6:  Household Debt Service Payments as a Percent of Disposable Personal Income

Figure 6: Household Debt Service Payments as a Percent of Disposable Personal Income

Disposable Personal Income is up over 20% since the peak in debt servicing. Meanwhile, household debt declined from nearly $14 trillion to just over $13 trillion. It has since started growing again — strong evidence that balance sheets have repaired to (or below) their equilibrium debt servicing levels.

Figure 7: Household debt

Figure 7: Household debt

The criterion of repaired household balance sheets is satisfied.


The inversion of the reasons which formed the conditions for the “New Normal” are the predicates for achieving escape velocity out of it. These predicates have become satisfied just as the growth trajectory has busted higher. This points to saying “Goodbye!” to the “New Normal”, and welcoming in a new regime.


Posted in Uncategorized | 7 Comments

Only Bernanke knew Oil is not inflation

In April of 2011, Ben Bernanke was universally lambasted and lampooned for claiming that inflation, which was accelerating and running above 3%, was “transitory”. He used this view to justify loosening monetary policy. The next few months of CPI were not favourable to the Fed chairman’s views: it peaked at 3.8% (nearly double the implicit target at that point) in September of 2011, sparking a feverishly pitched cacophony of criticism that the Fed chair was out of touch, and tone-deaf in his theoretical ivory tower to the practical realities on the ground.

This, however, proved to be Bernanke’s finest hour. Yes, even more so than the extraordinary measures taken during the height of the credit crisis. His detractors then, of which there were still many, included people and institutions on the brink that needed the Fed to extend them a hand. In September of 2011, the chairman stood very much alone in his call for moderated inflation now that the acute disaster removed influential institutions and people from needing the Fed to act in order to survive.

This moment in time represents a difficult choice by the Federal Reserve. It presented Bernanke the opportunity to make a big mistake by raising rates — one that ECB President Trichet took, sending the countries under his monetary care into a deflationary credit spiral that is still viewed as an intractable problem three years later.

The basis for the moderating inflation view was that the price increases were driven by food and energy prices, which make up 23% of consumer expenditures. Obviously put another way, 77% of consumer expenditures are not food and energy. When inflation peaked in September 2011 at 3.8%, wages were growing at just 2%. This brings us to a critical juncture on energy’s impact on inflation: inflation is not just a “price spiral”, but a price-wage spiral.

Food and energy prices are short-run inelastic — mainly because they are not easily substitutable. In English, this means that people need to purchase food and energy regardless of the price. When these prices rise faster than their wages, they are forced to reduce unrelated spending. This depresses other prices, and ultimately forces a moderation of inflation. Economic activity also moderates, and the new equilibrium is set with higher energy expenditures as a fraction of income. When confronted with a price increase of non-substitutable consumer staples, this acts like a new consumption tax. No-one will argue that increasing consumption taxes will increase consumption.

Bernanke was right. Holding the line with Zero Interest Rate Policy in the face of 3.8% CPI is now indefeasible. His decision spared the United States from a new recession, and the unnecessary excess economic suffering of millions that the Eurozone has experienced. While the Eurozone’s unemployment rate has increased nearly 3% since, it has fallen 3.2% in the United States, for a relative spread of more than 6%. As if a switch had been flipped, inflation in the United States sank like a stone to below 1% by October of 2013. This is what makes Ben Bernanke perhaps the best central banker in history. Any reasonable candidate for the job of Fed Chairman would have done what Bernanke did during the crisis (+/- a few weeks). The difference is manifested three years later, when he understood and correctly guided policy around what his most prominent colleague at the ECB did not.

This is why Ben Bernanke’s 2011 triumph is relevant today. The same framework for understanding inflation through commodity prices and wages that successfully predicted the deceleration of inflation against the tidal wave of popular belief now finds itself in the inverse position: the expectations of inflation are very low, and despite low commodity prices, it expects inflation will accelerate. Much like the consumption tax analogy when prices are rising, it is credible to argue that reducing consumption tax will be a net positive impulse on consumer spending. No-one will argue that decreasing consumption taxes will decrease consumption. In fact, axiomatically, the savings rate necessarily must rise in proportion to the savings on food and energy if that were so. And, that, my friends, is simply un-American.

While real wage growth was -1.89% in September of 2011, it is +0.57% as of October 2014. Statistically, a 1% change in the price of oil has been historically consistent with 0.043% change in the CPI. To contrast, a 1% change in rent is historically consistent with 0.8% of headline inflation. Rent has last grown at 3.34% y/y, which exerts an impulse of 2.6% on headline inflation. This is a much more important, sticky and longer term driver of inflation than oil.

Wages and rent growth both exceed inflation. Last Friday’s Employment Situation report showed that November payrolls expanded by 321k (far exceeding the 2010-present average of 175k), and wages jumped at a compounded annualized pace of 4.9%. If Ben Bernanke were still Fed chair, and he was asked to describe the low inflation environment, he’d probably call it transitory. And he’d probably be right.

Posted in Uncategorized | 3 Comments

Why do the Japanese want growth?

It seems to be a universal truth that growth is good. There almost can be no bad growth. But what exactly is the purpose of growth, and why do we want it?

Instead of viewing economic growth as the objective, we should rather view growth as progress towards our real objectives.

As you probably already know, Japan has begun “QE-zilla”, where the central bank will go on a buying binge, including private sector assets like equities that have been considered taboo until now by other major central banks. Why are they doing this? To answer this, let’s discuss the nature of the success “Quantitative Easing” has had in the United States.

For growth, you need fixed investment. This means building out real things like roads or houses, and increasing capacity to make the things we consume. The problem the United States encountered coming out of the Great Recession was that there wasn’t enough of fixed investment because the prospective yield on it was lower than buying financial assets. If you have the opportunity between a prospective yield of 4% which requires you to build a new factory, and one of 10% which requires you simply purchase it online in your brokerage account, ceteris paribus, you will always select the latter investment. Quantitative Easing works by compressing the prospective yield on financial assets through raising their prices so that it makes new, fixed investment more attractive.

To sum, for growth, you need fixed investment. For fixed investment, you need equities’ prospective yield lower than that of fixed investment. For equities, you need fixed income prospective yield lower than that of equities. This is the chain of risk premiums that real growth is predicated on compressing.

What does Japan need growth for? Perhaps they instead should be glad that they do not want for baser needs that less developed countries have. The emerging markets require high growth to bring their majorities out of poverty. The United States requires moderate growth to sustain the living standard of its growing population. Perhaps Europe requires some growth to sustain the living standard of its aging population.

We can formulate a hierarchy of needs, like Maslow’s, except for economies. Developing economies are at the bottom, where they require growth just to reliably feed their people. Above them are the emerging markets, which require growth to push their majorities out of poverty. In the middle, you have the United States, which requires it to sustain its middle-class living standard through population growth. At the top, you have Japan. With outstanding living standards, virtually no inflation and high income, have they not, in Maslow’s terms, satisfied the predicates for self-actualization? Have their baser needs not been met, and allowed them the freedom to explore and accomplish the whims of the soul, heart and mind? Also like Maslow’s hierarchy, it is improbable — and perhaps impossible — for a country to progress up the hierarchy without solving for the more basic needs below it.

Perhaps the Japanese don’t need economic growth, but rather, they need to self-actualize to fulfill the maximum extent of their destiny as humans and a culture, and not in an international pissing contest about numbers on a spreadsheet — which they’ve arguably already won.

Economic growth is the progress towards goal of cultural self-actualization. When it is confused for the goal, no real progress can be made.

Posted in Uncategorized | 4 Comments

Why do stock markets crash?

With the S&P 500 off 5% from all-time highs, and the 10-day volatility in the 87th percentile, some observers are anticipating a 1987-style market correction. In this piece, we analyze the conditions that presaged prior violent corrections.

Starting with what the most recent level of volatility provides us, we split the sample of 10-day historical annualized volatility readings into 2% buckets, and calculate the median, minimum and maximum 10-day annualized volatility which immediately follows:

fv vs rv

We estimated best fit lines using a robust linear fit to exclude outliers. The R code to generate this plot is here.

Although much of the narrative seems to indicate that an elevated level of volatility is a harbinger of higher levels of future volatility, the median case actually shows that future volatility is a fraction (2/3rds in this case) of what was recently experienced, plus a base level (of 3.32). There are outliers, of course, but they are exogenous to the relationship between realized and future volatility. To gain insight into the volatility spikes, we must look elsewhere.

annualized annual vol

The average annualized volatility out of recession is 13.88%, while it spikes to 17.3% in recession. Volatility has only eclipsed 20% outside of recession (or impending recession) twice — the crash of 1987, and the Eurozone meltdown in 2011. Because recessions are the most common cause of volatility, we measure the probability of that kind of event with our proprietary macro model, and include University of Oregon’s Professor Jeremy Piger’s Recession Probability model for additional signal.


Absent a recession, we turn to theory. David Merkel proposes using a discounted cash flow model, whereby the stock market’s Net Present Value is determined by the expected returns in stocks (book value growth and cumulative earnings) discounted by a risk premium for uncertainty to a marginally safer asset. A materialization of this view is spreading the earnings yield and investment grade bond — a variable sometimes referred to as LERP, or the Levered Equity Risk Premium. LERP is the difference in expected returns between stocks and bonds, plus the risk premium required by investors to assume the uncertainty of owning the equity component.

If this is true, we should be able to explain future equity outperformance of bonds by using the starting LERP value and the subsequent period earnings growth. Combined with realized inflation, we can:

expected return parity

More difficult is knowing exactly where the expectations differential ends and the risk premium component begins within LERP. Using the long-run historical realized equity risk premium in the above equation (3%), plugging in market inflation expectations (1.85%), and the yield of VFSTX as its expected return, we could solve for the market’s expected EPS growth, which is 1.49%:

Where x1 = 1, x2 = 1.85, y = 3….
y = -8.26 + 3.06(x1) + 3.45(x2) + 1.22(x3)
3 = -8.26 + 3.06(1) + 3.45(1.85) + 1.22(x3)
3 =  1.1825 + 1.22(x3)
3 – 1.1825 = 1.22(x3)
(3-1.1825)/1.22 = x3
x3 = 1.49

Returning to the 1987 analogue, we can plug in our variables, and contrast them with now. LERP had reached the most extreme reading ever: -6.23%. 5-year inflation expectations, as measured by the Cleveland Fed series, was 4.3%. Using the same technique as above, it implied a 5-year compounded annual EPS growth of 12.3%. Spoiling the party, however, was the Fed, who jacked up rates from 5.95% to 7.62%. This had two effects:

  1. It increased the minimum required return by investors to over 7.62%, which they could get merely by holding cash.
  2. Decreased inflation expectations.

At the same time, actual realized EPS growth up until that point had been very muted in 1987, starting the year negative y/y, and only maxing out at 6.5% towards the end – well below the estimates built into the market of 12.3%. Further, with the spectre of fighting inflation very fresh, future monetary policy expectations became even firmer, as evidenced by 3s-1s curve briefly reaching a whopping 1.41%.

Using these methodologies to analyze prices, it is a very simple numeric case to make against owning equities near the top of the market in 1987.

In contrast to now, the Treasury curve is already steep (policy normalization is already priced in), inflation expectations are at generational lows and have little room to decline, LERP is +1%, and the EPS growth required to beat a 2% short-term corporate bond yield could be estimated using the above methodology at somewhere around 1.5% per year.

If we get a correction anywhere near the magnitude of 1987, or even 2011, the environment and context will bear little resemblance to the mid-cycle corrections of the past.

Posted in Uncategorized | 13 Comments

Dear Ms Yellen: Raise the inflation target

Dear Ms Yellen:

Congratulations on both your appointment as well as becoming the first woman to head the Federal Reserve. This will potentially make you, at times, the most powerful and important person in the world.

We are more alike than one may assume on the surface. It’s not just that we both like economics, that we were both born at the vanguard of our respective generations (you the Boomers, me the Millennials), or that our fathers were men of science and mothers teachers. Being amongst the first of our generations grants us some perspective on what the mass of humanity which is poised to follow will do, how they may act, spend, and impact the economy. Both of us came of age in a decade of great economic turbulence, and a growing cynicism on the efficacy of public institutions.

Most importantly, you witnessed the largest and richest generation in history come of age, join the labour force, form households, and have children of their own. I am witnessing, at the least, the largest generation — that of your children — in history come of age. The reason I believe this to be significant is that the quantity of the population in this age group (let’s call them the young age cohort) is strongly correlated with inflation. When there is a surfeit of young, household-forming aged people, there has been proportionately heightened inflation.

We each lived through the deepest recession since the Great Depression at roughly the same age, with our slightly younger generational siblings (aged 20-24) suffering an unemployment rate above 16%. The cost to my generation from the past recession is already high. On top of very high unemployment, we have experienced delayed entry into the labour force, deferred forming our own households, and the development of all facets of our lives retarded by the economy running so far below potential. Fixing this output gap, I imagine, is the top priority of the Federal Reserve right now, and there is no doubt that we are making headway. The record 17.2% unemployment of the aforementioned aged 20-24 group has declined to 11.1%.

I write you not only to congratulate you on your appointment, but also warn you of a coming inflation pressure, and petition you to raise the inflation target. No, I am not talking about the contentious unconventional policies the Fed adopted to combat the Great Financial Crisis. The Fed has demonstrated its capability and facility in managing them, and have evidenced their ability to wind them down as appropriate. Instead, I refer to a demographic argument that inflation is strongly related to the size of the young age cohort.

Screen Shot 2014-02-01 at 10.12.16 PM

Figure 1: Black: year-over-year inflation (fraction) Orange: Model estimate based on population growth of 16-24 years age cohort

You acknowledged the Federal Reserve should not necessarily over-emphasize price stability over unemployment in 1995 [1] by stating, “to me, a wise and humane policy is occasionally to let inflation rise even when inflation is running above target.” The scenario to which you directly referred were the oil supply shocks of the 70’s, but inflation also ran well above 2% in the 80’s – averaging 5.6%. Even after the supply shock ended and energy inflation’s impact on prices was over, inflation averaged 3.6% from 1983 through the end of 1989.

There are, I’m sure you will agree, many things you can attribute as factors to inflation. At its most basic level, inflation simply means that demand is advancing into a upwardly sloping supply curve. But that definition struggles to really pictorialize what that actually means. What seems to make the most sense to me is anchoring to the cycle of life which has been carried over for generations: we are born, we are children, we become educated young adults, we begin our careers, we leave the nest to form our own households, we have children of our own, we work to provide for our families, we work to save for our retirement, and we retire. Each of these steps represent a milestone in our lives, and to a great degree, the likeness of age we do each one of these things is contrasted with our parents. Thus, the demand for the goods and services at each of these milestones represent a demand which is more influenced by our aging than by price. That is to say, the prices of goods and services related to physical maturation is relatively inelastic.

Screen Shot 2014-02-02 at 11.53.54 PM

Figure 2: Year-over-year change in population aged 16-24, thousands

If much of demand grows by the schedule of maturing young adults, the same cannot so readily be said of supply. The role of demography in the corporate sector is largely contained within marketing departments, and far away from capital allocators who focus on much shorter-term objectives. Young adults find themselves with tougher prospects to break into the job market than any other age cohort [2]. Their inexperience makes them initially less desirable hires (a firm would generally prefer to hire one worker at twice the productivity than two at half the productivity), and the business sector eventually suffers (like it does now) from a deficit of experienced peak-aged workers.

This forms the basis of our demographic trap. Real goods and services are supplied by the production of real capital and labour. The configuration of output — the planning, training of people, purchase of machines — operates at a lag to the decisions of executives, and executives are hesitant to new capital expenditures before they see new demand.

These are frictions which resolve themselves over time. Unless policy, or another force hitherto unevident, prevents it, these frictions will also eventually resolve for Millennials. The main question is how quickly, and at what cost.

The Federal Reserve explicitly does not take responsibility for inflation from supply shocks. I will argue that its tolerance of 5.6% inflation in the 80’s indicates it implicitly distances itself from responsibility of demand shocks, too. Let us be candid and admit that there is a cost, both familial and more broadly societal to turn boys and girls into men and women. Even after they become young adults, taxpayers subsidize below market-rate student loans and colleges, businesses take on entry-level workers and apprentices, and the young adults form households largely by buying or renting at below replacement prices in areas with existing mature infrastructure and services.

The relevance this holds to the Federal Reserve is in the societal cost borne through the surplus inflation attributable to a bulging generation’s maturation. Based on the historical relationship between inflation and the size of the young age cohort, a 2% inflation target is unrealistic and incongruent with the Millennial generation, the biggest in history, forming households. It is difficult to imagine the economy running anywhere but far below its potential output if inflation is forcibly repressed under 2% while Millennials form households. Instead, an accommodative Federal Reserve will ultimately lower the average intensity and length of the elevated inflation by facilitating the creation of all the new real capital, housing stock and infrastructure required to serve the needs of the incoming generation. This real wealth will translate into greater future income, the ability to raise real interest rates, and less inflation because of the subsequent bending of the supply curve. In the same way you fight fire with fire, you can fight inflation with inflation – after all, as the saying goes, the cure for high prices is high prices. Since we have observed income gains to be sticky, the economic opportunity cost of suppressing this generation’s household formation could be optimistically measured in the trillions of dollars over our lifetimes. This is what makes you potentially the most important person in the world, most certainly in the country:  your decisions impact the standard of living of your son’s generation so greatly both contemporarily through (de-)incentivizing and facilitating (or suppressing) real capital formation into the capital factors of aggregate supply — homebuilding, infrastructure investment, factory machinery — as well as the resultant national income that real capital will generate even after you are gone.

Ms Yellen – thus I entreat you to explore raising the inflation target to a level which is more consistent with the demographic cards dealt to you. For Chairman Bernanke, perhaps 2% was an appropriate target. It was not for Chairmen Volcker or even for much of Greenspan’s tenure, and probably is not for you.

Congratulations & regards,

Matthew Busigin.

Posted in Uncategorized | 2 Comments

Estimating the impact of monetary policy normalisation on assets

A New Age of Policy Transparency

The Federal Reserve, beginning in 2011, began a new era of transparency that began with the publishing of economic and monetary policy projections from Open Market Committee members.  Along with projections for unemployment and GDP, they now include two critical charts: Appropriate timing of policy firming and Target federal funds rate.

Screen Shot 2013-08-13 at 9.19.27 PM

The Treasury curve itself represents the expectation of the Federal Funds path, plus whatever extra term premium investors may demand for holding longer dated securities. We calculate the expectation – the forward rate – by using the difference in rates at two maturities to find the expected return after investing in the shorter maturity and imputing the yield it would take to break even with the yield of simply buying the longer maturity.

Screen Shot 2013-08-21 at 12.31.13 PM

As of January 2012, the Federal Reserve explicitly set a 2% inflation target.  This combines with the consensus of the committee’s longer-run target Fed Funds rate of 4% — 2% nominal, 2% real.  While this had been assumed for a long time, despite the implementation of two quantitative easing programmes, the term premium embedded in the forwards relative to this expectation actually didn’t consistently break negative until it was announced.

Michael Woodford’s 2012 Jackson Hole paper and speech, Methods of Policy Accommodation at the Interest Rate Lower Bound, furthered the view that projection of expectations has the most impact on assets of any monetary policy tool.  The evidence post-new-Fed-era-disclosure is consistent with that conclusion.

The impact of Federal Reserve policy really is broken down into three classes: markets which the Federal Reserve is directly targeting, those that they are trading, and in those which the Fed is doing neither.

The overnight rate is under the Federal Reserve’s direct control, facilitated by the unique tools afforded to a central bank.  After January 2012, combining the Fed’s existing dominance over overnight rate with the Treasury curve being the sum of Fed Funds path expectations, the Treasury curve could also be considered directly targeted, albeit with the optionality of policy deviation potentially presenting a positive or negative term premium.

The agency mortgage market falls under the traded market class.  Instead of price targets, the Fed has set macro conditional purchase volume targets.  This serves to increase liquidity and decrease asset supply of the traded market.  The impact of the Fed’s policy on these markets can be inconsistent, however: if the Fed is simultaneously buying and conveying uncertainty about continuation of future policy, the market will react by forming an equilibrium to the average expectation of both the normal supply and demand of the asset, combined with the extra projected demand from the Fed.

Markets of all classes which the Fed does not either target or trade are still impacted by the actions the Fed takes in the markets which they do.  Much like traded markets class, the minimum return demanded by investors must be at least that of the equivalently termed risk-free investment.  More broadly, investors demand the minimum return on any asset to be greater than the equivalently termed lesser perceived risky investment.

The expected return of any asset is the weighted-average outcome of probabilities (summing 1) multiplied by their respective scenario returns.

E(R) = ∑ (probability₁ * return₁) + (probability₂ * return₂) + (probabilityⁿ * returnⁿ)

Unless market participants are willing to accept a lower probability-weighted expected return, the expected return across all assets in this class of market is the same after beta and liquidity premia.

The penultimate question remains: what will happen to asset prices when policy begins to normalise?

The Equity Market

Evidence the Treasury term premium is embedded in the S&P 500: when it's steep, hikes are already priced in

Figure 1: Evidence the Treasury term premium is embedded in the S&P 500: when it’s steep, hikes are already priced in

The mean monthly total return of the S&P 500 is 0.65%.  That return drops, but is still positive when the Fed Funds rate increases, to 0.45%. However, when the term premium between the 10y and Fed Funds rate is positive, the average return returns to 0.64%. When the term premium is above 1%, the average return is even higher than the sample mean – 0.82%!  When adding the condition that the benchmark rate falls, the mean monthly return climbs to 1.27%. This tells us something very significant to the discussion: the price of S&P 500 contains the expectation of monetary policy.  Thusly, monetary policy tightening has not negatively impacted the S&P 500 unless it happens more quickly than the market expects.   In fact, both the 10y and the equity market don’t necessarily negatively react to Fed Funds hikes.

Consider the simple equation:

EY = ERP + 10y

ERP = EY – 10y

Figure 2: ERP vs Real 10s, 2s10s and 1y SPX Realised Volatility dependent variables

Figure 2: ERP vs Real 10s, 2s10s and 1y SPX Realised Volatility dependent variables

The Equity Risk Premium, or ERP, contains the information on the spread of expectations between our benchmark asset (the 10y) and the equity market.  If the stock market assumed no growth, but the certainty of return on capital was 100%, the difference between the benchmark rate and the stock market would be minimal.  In reality, the expected difference in returns diverge with expected inflation, liquidity needs, and the expected future rate path.

The term structure, as represented by the 2s10s Treasury term spread, has contributed from 2.43% to -1.83% to ERP.

Figure 3: Decomposition of the Equity Risk Premium
Figure 3: Decomposition of the Equity Risk Premium

Let’s fill in a little more detail into the ERP equation:

ERP = (Real 10s * -0.58047) + (SPX 1y RV * 8.83836) + (2s10s * 0.85830) + (Residual Risk Premium)

A few observations: the most volatile term is the liquidity contribution, which we estimate as the sum of the SPX 1y Realised Volatility and residual error (viewed as the Residual Risk Premium). However, it is also the most mean-reverting.  The fundamental premise is that this is the true equity market risk premium, and although unknowable, will either compress, or ultimately be shifted to the other terms (real rates and the 10y). The Real 10y represents inflation expectations, and therefore the fulcrum on which the preference for fixed cash-flows rests.  The last two pieces of the equation, the term structure and 10y, represent the minimum required return, and the embedded expectations for how the discount rate will move in the future.

Using this framework, we can run some thought experiments on potential future interest rate scenarios.

The simplest scenario would be estimating a full policy normalisation – a Fed Funds rate at 4% (2% real, 2% nominal) as estimated by the FOMC for the long-term rate, and a 5% 10y (1% term premium).  In this scenario, the 2y is unlikely to be less than 4.5%, leaving a 2s10s spread of 0.5%, which is consistent with a post-policy normalisation mid-economic cycle value.  At first blush, since EY = ERP + 10y, the 10y portion of the EY equation is increased by 217bps.  However, simultaneously, the 2s10s term in EY likely declines by 186bps (216.5 * 0.85830).  Therefore, if neither inflation expectations or the liquidity premium change, EY would increase around 31bps – roughly 80 S&P 500 points lower than at the time of writing (4.8% of 1,665) if there were no change in EPS.  In the context of the discussion, this is a rounding error.

It is not necessarily the case that the other two terms, liquidity premium and inflation expectations, remain unchanged during this process.  Monetary policy normalisation would likely push real rates higher, which would be on balance positive for equities (recall that the real rate term has a coefficient of -0.58, which means real rates rising compresses ERP, which would increase the value of stocks if the benchmark rate & earnings were unchanged).  The last term, the liquidity premium, is the most difficult to forecast, but also the most quickly normalising.  It is likely unknowable where this will be on a day-to-day or month-to-month basis, but of all of the terms, it is the least stationary.

Monetary policy normalisation does not seem likely to adversely impact equities as long as it is occurs no more quickly than the market has already priced in.

The Corporate Bond Market

The Baa spreads

Figure 4: The Baa spreads

The bond market is, perhaps, more straightforward.  While the impact of the term structure must be teased out with techniques like component analysis and decomposition, the rates market has a term structure that directly conveys expected future rates.

Long-term rates are reasonably well anchored.  What happens in the intermediate-term is far more volatile, however, and the impact of intermediate-term expectations adds to the volatility to the long-term bonds because of the no-arbitrage condition embedded in the forward rates.

Figure 5: 2s10s vs the future Baa-10s spread

Figure 5: Scatterplot of 2s10s vs the future Baa-10s spread

The Baa spread between both 10s and 20s have enjoyed a very close relationship over many decades. As the expectation of ZIRP’s eventual end begins to ripple through the curve, the spread has blown out, and the Baa-10s has not normalised to previous post-recession levels, ostensibly because of the abnormally low short-rates this far in the economic cycle.

Figure 6: Discrete samples of 2s10s vs Next 52w Baa-10s spread

Figure 6: Discrete samples of 2s10s vs Next 52w Baa-10s spread

The expectation of higher (lower) future rates correlates with compressing (widening) future credit spreads.  After all, why take on credit-risk now when you expect you can purchase risk-free securities at competitive rates later?

Questioning the Veracity of the Term Structure

A termed risk-free rate in the future is the Net Present Value of the risk-free rates to maturity.  This is an opportunity-cost or no-arbitrage condition:  for instance, no-one will purchase a 10y now with the expectation purchasing a 7y to rolled into a 3y, or or that time-weighted average of overnight rates, will make more money.

Figure 7: Distribution of (Fed Funds - NGDP)

Figure 7: Distribution of (Fed Funds – NGDP)

The average FF-NGDP value in expansion -2.19%.  The same value in recession is 4.26%.

The consequence is that those predicting a policy normalisation are actually either implicitly forecasting an acceleration of NGDP to somewhere near 6%, or that the Fed will tighten prematurely (which would likely lead to recession).

Since 1Q 2012, NGDP %-y/y has declined every quarter but once, and as of the time of writing, is just 2.9% – the lowest ever outside of, or leading into, recession.  The policy neutral position as priced into the curve already assumes a level of acceleration or policy error that seems unlikely at this juncture.

If the expectations for Fed hiking soften, the benchmark rate could simultaneously soften with a flatter term structure.  Ceteris paribus, both would contribute to lower rates, credit spreads and earning yields.


Markets are not impacted by monetary policy normalistion so much as the shifting expectation of normalisation.

Not only is policy normalisation expected to have a minimal impact on asset prices moving forward, but the probabilities of a slower normalisation appear underpriced.

Posted in Uncategorized | Leave a comment