Monday, January 20, 2014

Personal Savings Rate and the Balance of Payments

Personal Savings Rate & Current Account Balance. Source: FRED
One of the most common--and yawn-worthy--criticisms of my December post outlining why the rise of labor bargaining power was going to be very negative for corporate profits (my estimated CAGR for the next 20y is 1.65% nominal on the upper end) was that corporate profits would be able to survive an increase in real wages because nominal demand would expand at a greater-than or equal-to rate as wages. When confronted with the fact that this would necessitate a lower savings rate, an anonymous reader stated that the savings rate would decline because, "it's demographic." This is, of course, total nonsense. Although the baby-boomers will indeed draw down their stock of savings as they retire, these are most likely to be offset by rising stock of savings of the echo boom and millennials resulting from higher employment and real wages.

Arguing for a decline in the personal savings rate means--by definition--an increase in the rate of growth of one or more of the following in excess of the rate of growth in personal income:
  • Corporate savings (profits - dividends)
  • Government savings (budget balance)
  • Foreign savings (current account deficit)
The first, as I indicated in my December post, is not happening.The second one, a contracting deficit is currently taking place as a result of the labor recovery and the last is actually in contraction. It is this last one I want to talk about here.

Although the causality is bidirectional, the savings rate is intimately tied to the balance of payments by the accounting entity CA= NS-NI (Current Account Surplus =  National Savings - National Investment). As such, a revisiting of the low savings rates associated with the 2000s would need to be coincident with a much much larger current account deficit and tighter fiscal policy. The main driver in the 2000s for this were the so-called "Clinton surplus" followed by the large current account deficit.

Source EIA
As increases in petroleum production cause imports to decline, the portion of the trade balance associated with petroleum declines. Additionally, the ongoing rebalancing of the Chinese economy towards consumption (as well as the ongoing RMB revaluation) will continue to drag on China's current account surplus--and, in my opinion, turn it into a deficit--making the return of 6% current account deficits highly unlikely.

There is quite a few reasons to be bullish on increases in nominal final demand, but a decline in the personal savings rate is not one of them. The gap in quality between the mean and median household balance sheets (wealth skew) has probably peaked and is headed nowhere but down. As I have said before, the accumulated stock of corporate sector savings is about to be transferred to the household and public sectors.

Friday, December 13, 2013

Even more evidence of shifts in labor markets

The following charts are part of December 13's Bloomberg Economics Brief. The top chart is Bloomberg's Economic Surprise Index. As you can probably  notice, the labor market has been surprising significantly to the upside lately. On the bottom, is changes in nominal wages for the month of November by selected industry. The red line represents the Fed's symmetrical target for inflation while the yellow highlight indicates value of the year-over-year percent change in the PCE deflator.  

While wage growth leaves lots to be desired, all but two industries, utilities and Leisure & Hospitality, showed wage increases in excess of PCE deflator increases. This is good news, these are real wage increases and more evidence of labor market tightening, even if very marginal. This is my 3d post in 2 weeks about this and you better get used to them, because there's probably going to be a lot more of them. The tide is changing, and labor is on its way up!

2013-12-12 -- More on labor bargaining power
2013-12-05 -- Profit margins, tax receipts, and labor demand curves

Thursday, December 12, 2013

More on labor bargaining power

Last week I wrote about how a number highly reliable indicators are indicating the bargaining power of labor is increasing and that it will lead to tighter labor markets and wage inflation.

On Nov 26, The New York Times reported that NYU will once again recognize the graduate assistants' union. The voting was scheduled to take place this week, but a search of press releases shows no results yet as  to the outcome. Excerpt below, emphasis mine.
 Some 1,200 graduate assistants at N.Y.U. and the Polytechnic Institute of N.Y.U. in Brooklyn are scheduled to vote on Dec. 10 and 11 on whether to join the union...
An issue that had long prevented agreement between N.Y.U. and the union was whether graduate research assistants in the natural or physical sciences could be included in the union. The university argued that the research those assistants did was an essential part of their academic training, and should not be viewed as employment. In their statement, the union and N.Y.U. acknowledged that they had not resolved their differences over whether the 275 graduate research assistants in the so-called hard sciences had bargaining rights. They will therefore not be included in the union.
NYU's claim that the work (they called it "work") their research assistants do is part of their training, they are acknowledging it is, indeed, work. As with the recent unpaid intern cases, some will argue that this is not valuable enough to be compensated and the benefit of it mostly accrues to the employee/intern/assistant but, if their labor is truly of so little use, maybe they should be free to apply those resources elsewhere. Like in the intern cases, the labor slack as a result of the financial crisis created a large number of people who wanted work and were willing to sacrifice wages in the hopes that experience would be worth something and make it easier to find a paying job in the future. This shift right in the labor supply curve would have led to lower clearing price for labor, the evidence of which is everywhere. Whether this was unfairly exploited or not is up for the judges to decide, but what seems self-evident to me is that the trough in the bargaining power of labor is behind us and, as a result, the labor consumer surplus--in the neoclassical sense. remember, employers are consumers of labor--is about to be redistributed.

At the risk of oversimplifying, but in the interest of avoiding a black-hole, I will skip discussion of dead-weight losses, the welfare state and heterodox theories of economic surplus, and simply say this: If you believe any of the following:
  • welfare initiatives like the Earned Income Tax Credit subsidize low salaries leading to a higher quantity of labor supplied at a point below the incentive-free equilibrium
  • price elasticity of supply of labor flattens significantly on the left side (positive first derivative)
  • some people will agree to work for less to build a resume or gain experience
Then, the logical conclusion is that, as labor markets recover, the lions share of the benefits will accrue to the household and public sectors. Not only on the marginal increase of production (a change in quantity or labor demanded), but on the entire level of production (a shift in the demand curve).

The corporate rent-seekers masquerading as capitalists will argue the corporate sector will simply not hire or invest if faced with rapidly declining margins or increasing costs at the margin, but the truth is that almost 3 decades where productivity gains have disproportionally accrued to capital and record-high profit margins mean that the ability of capital to bring a credible bluff is just about non-existent. This trend is going to be with us for a long, long time.

UPDATE: The NYU vote was overwhelmingly for organization and both parties expect to have completed a contract by the end of the academic year.
UPDATE-2: "Machinists reject Boeing labor contract offer"

Tuesday, December 10, 2013

Corporate Debt / Market Cap as a predictor of forward returns

Earlier tonight, Matt Busigin tweeted a link to the series "Credit Market Debt as a Percentage of the Market Value of Corporate Equities" from the FRB's Z.1 report. After some back-and-forth with some other folks, Greg Merrill suggested that high debt to market cap looked like it it would correlate positively with forward returns. Intuitively, this works because, if we assume creditors are generally somewhat rational about underwriting standards, a sharp increase in Debt / Market Value would signify a sharp decrease in the value of equity from a level at which underwriters thought it was safe to lend. Stepping back, what we would be doing is measuring the "fair value" using debt as a proxy. 

After thinking about it for a few seconds, I decided to plot our the series against annualized forward returns, shamelessly mining for the forward period that offered the best correlation. After starting with 5y (R=0.5985), 7y (R=0.7069), 10y (R=0.8411) and 11.5y (R=0.8677). After peaking at 11.5y (46 quarters) the correlation starts declining again.

The actionable value of this exercise is questionable given the small sample size but, in my opinion, probably reflects the rational behavior of lenders and CFOs regarding debt levels.  The average Debt / Market Value ratio for the sample is 57%, which is 9.1% higher that the latest (Q3) reading of 47.9%. Since the end of Q3 the Wilshire 5000 has risen 6.84%, representing a market-cap increase of $1,229 billion. We currently have no estimate for net corporate debt issuance since the end of Q3 but, if we extrapolate Q3's 2.2% increase, we get an estimate of a current 45.8% debt to market-cap ratio, which would correspond to a pretty average 7.8% annualized forward return for the next 11.5y. This expected return stands in sharp contrast to much lower expected returns obtained with similar methods by John Hussman using non-financial market cap as % of GDP and the Price/Sales ratio of the S&P 500  as predictors. I should note here that this expected return also lies significantly above my own 10y expected return of just 4.5%.

Of course, the current Debt / Market Value lies in a particularly dense zone with significant dispersion, which--when combined with the 2.45% standard error of the linear regression--should give us pause as to the reliability of any such measure. In this particular casse, the expected total holding period is 132% and the -1 and +1 sd  bands would lie at 82% and 207% respectively--not exactly a Swiss watch.

Thursday, December 5, 2013

Profit margins, tax receipts and labor demand curves

Much has been made of the record levels of corporate tax profits over the last two years. From GMO's warnings of an 1100 "fair price" for S&P 500, to Hussman's forecast of 10y of negative nominal returns, to the economics "It" girl of the moment, the Kalecky-Levy profits equation. Real median household income, in my opinion at a cyclical bottom, is back to early-to-mid 90s levels and as ZeroHedge (the best indicator of policy bear zeitgeist) reports, wages are at an all-time low relative to profits. Meanwhile, it seems like the entire Very Serious Old Guy complex is warning of mean reversion in a laundry list of ratio measures, but nobody wants to talk about whether it will be the nominator or denominator that will change. It is my intention to illustrate exactly how these measures will mean revert in a simple, common-sense way accessible to anyone with a cursory understanding of supply and demand curves and lay out what I expect to be a way to make investments guided by this thesis.

Corporate profits are high because effective tax rates are low, real wages are low, and debt to large companies is cheap, a point I've previously made. They are about to start shrinking. When? Like right now. Maybe last month, or last quarter, even. What led me to write about this is, believe it or not, is public sector employment. Employees of State, Local and Federal governments are not many, they peaked at 14.68% of the labor force (not including the census high) exactly as private payrolls hit bottom, and hit a trough in June and July 2013 at 14.01% representing 21,826,000 employees. In other words, fiscal drag added 0.67% to the unemployment rate.

But this summer we had three important developments:
  1. The labor force stopped growing
  2. The number of public sector employees stopped shrinking and may be growing
  3. Real average hourly earnings growth of non-supervisory employees accelerated to a level coincident with the last expansion
Public (blue rhs) and Private employment as % of labor force
In economic parlance, we can translate the first one as a change from a marginal pressure right on labor supply curve  (curve inching slowly right as labor force grows) to curve shifting slowly left. Ceteris paribus, and assuming a downward slopped demand curve, the consequences would be an increase in the price of labor (coincident with #3) and a slight reduction in quantity demanded at the new price. The second would represent the public sector going from a marginal impulse left on the labor demand curve, to neutral or right maybe a slight force right, which will lead to stable or increasing price of labor (coincident with #3) and an increasing quantity of labor demanded at the new price. The third observation, present since Q1 of 2013, tells us that despite the public sector drag in the first half of 2013, private sector employment growth (demand curve shifting right) has been enough to lift average wages and increase the number of units of labor demanded at the higher price.

With the private sector labor demand growing, public sector stable, or growing, and labor supply (labor force) stable or shrinking, the only plausible answer is wage growth. And, as the marginal unit of labor required to produce a marginal unit of final output goes up in price, so must marginal profit margins fall. But this is not the most interesting part.

Capital expenditures / GDP (blue, lhs)
After-tax profits / GDP (red, rhs)
The interesting part about the position we find ourselves in is that, because of the very low effectivecorporate income tax rate (~16.35% last year), and very low corporate investment rate, the marginal dollar earned by the corporate sector has very little impact on the economy, it just sits as retained earnings. Using Manufacturers' New Orders: Nondefense Capital Goods as a proxy for capital expenditure we can see that even though corporate profits as a share of GDP have increased, capital expenditures as a share of GDP have decreased, meaning the marginal propensity to invest in new capacity is low. This is because of depressed aggregate demand caused by the low labor share of income and previously mentioned low real wages.

This, finally, gets me to my much delayed points:
  1. If the marginal effective tax rate of the household sector is higher than that of the corporate sector, which we know is true because FICA on its own is 15.3% (split by employer and employee), a marginal dollar that moves from profits to wages will increase tax receipts
  2. If the marginal propensity to consume of households is larger than the marginal propensity to invest of corporates, a marginal dollar that moves from profits to wages will increase aggregate demand
  3. If the marginal propensity to consume or invest of the public sector is greater than the marginal propensity to invest of the corporate sector, the increase from #1 will increase aggregate demand
  4. Any savings by households and/or government in excess of investment will be coincident with lower profits, all else equal.*
  5. Any increase in demand for labor by the public sector in response to #1 (in the form of bigger budgets) will be a marginal pressure to the right in labor demand which, all else equal, will lead to an increase in both the price and quantity demanded at the new price of labor. 
  6. Increases in employment and household income will reduce the cyclical deficit and reliance on government assistance programs like medicaid and "food stamps." This is, once again, an increase in government savings which is negative for profits. 
I will stop here, as you are likely seeing the self-reinforcing cycle that will be triggered. Because payroll and individual income taxes make up the lions share (~ 80%) of federal tax revenue, this will lead to a very strong self-feedback loop that will ultimately pressure corporate profit margins down and real wages up, reducing both the income and wealth distribution skew (the proverbial labor/capital divide) while redistributing corporate savings to the household and public sector.

10y treasury yield minus %YoY change in CPI (blue)
%YoY change in average non-supervisory hourly wage
minus %YoY change in CPI (red)
If the past is any indication, a real increase in wages will lead to higher nominal and real rates of interest for long term securities (thanks to Matt Busigin for this one) which, if you recall earlier discussion, is one of the primary reasons for the elevated levels of profit margins. As liabilities mature and reprice at higher rates, this will be a direct hit to profit margins, especially so if the increase in the rate of financing is not only nominal but also real. Given the very low present financing rates (without even mentioning qualitative measures like easy covenants) any future liability repricing is likely to increase the cost of capital and, once again, pressure profit margins.

Grantham, Hussman, Gross and many other investment managers have expressed concerns over elevated profit margins. To my knowledge, none of them have chosen to describe exactly how and why profit margins will fall and who will benefit and how. In this analysis, it is obvious that the beneficiaries of the sectoral rebalancing will be wage-earning households and tax receipts. The losers will be the labor-intensive employers, especially those that are highly leveraged because the cost of debt capital and cost of labor will rise at a speed greater than final demand or price inflation. It should also be clear by now the role of low investment (capital formation) and high unemployment (and associated cyclical deficits and low household savings rates) have had in the final sharp impulse upwards of corporate profits during a struggling economy and that, as labor markets recover, the household and government savings rate will gradually recover as corporate savings decline. Perhaps ironically, the same lack of investment that has helped prop-up corporate savings and holding unemployment high and kept inflation low will, as real wages increase, be the cause of any future increase in inflation. As, Matt Busigin has shown before and you can see to the right, this analysis is not only theoretically sound, but also empirically true.

Net Investment / GDP
This is likely to be very, very long cycle, even if it hits small cyclical snags along the way. Wages just started growing, the public sector just stopped being a drag on employment and net investment just turned positive. To anyone who missed the historic stock market rally, it may feel like it is too late, but this is just the beginning of the real economic recovery. It is also the beginning of inflationary pressures, but it will benefit wages more than prices. If you are wondering, "what's the play?" it is to favor being exposed to credit risk backed by public sector and household incomes and avoid being exposed to credit risk being backed by corporate incomes, especially those that are labor-intensive or highly indebted. Because the Federal government is considered a credit-risk free entity, this would mean the credit risk of state and local governments (which are ultimately backed by incomes of the residents) and of households, either directly through securitized obligations or indirectly through institutions that have a large exposure to households as creditors. Using extreme examples, you would want to own a company that insures mortgages and consumer ABS securitizations as well as municipal bonds, and avoid the lower tranches of any recent vintage CLO.

Is this a doomsday sign for stocks? Maybe not. It is possible that aggregate demand growth is enough to let profits fall as GDP increases and profits rise more slowly, however, it is not likely in my opinion. Assuming a reversal to mid 2000s effective corporate tax rates of 22% after tax profits as % of GNP to a generous 7% (median is 6.10% and mean is 6.34%) and a NGNP growth rate of 6% (mean 6.11%, median 6.7%, current 3%) over 10 years the CAGR of after-tax corporate profits would not amount to more than 1.65%. But it is also not terrible. GMO likes to flaunt their 1100 "fair value" number and Hussman is partial to his price-to-sales ratio chart, but there is no reason for stocks to fall to their fair value, and every day that passes, their fair value will close in to their present value, and at an accelerating pace. This would make selling short equity a trade with a short lifespan, while other trades--especially those where time works in your favor, like being long high-grade municipal bonds, an asset class which happens to be offering remarkable value--offer much more attractive ways to play the thesis.

Shorter version
Marginal head-winds for employment and wages are turning into marginal tail winds as the economy recovers. These same factors posses self-reinforcing properties and are likely to continue to be positive impulses for, real wages, employment levels, tax receipts, and aggregate demand and negative impulses for corporate profit margins and corporate savings. Favor the liabilities of the household and public sector over those of the corporate sector.

*Going back to Kalecki's Profit equation we can remember that:

Corporate Saving = Profits - Dividends
Profits = Investment – Household Savings – Government Savings – Foreign Savings + Dividends

Tuesday, November 26, 2013

Social Advertising Economics

YoY ad pricing from Yahoo!'s 3Q2013 presentation
Like Facebook before them, new "social" web properties eventually have to monetize to and generate revenue. As properties seek to monetize their user base through the use of display or per-click ads, the amount of ad space being sold will increase. In economics parlance, what we would be seeing as mature high-traffic web properties seek to monetize with advertising is the advertising supply curve shift right, increasing the quantity supplied at a lower price point (ceteris paribus). Of course, one tiny website can't really move the curve, but when Facebook,  Yelp, Twitter, Tumblr, and now Instagram, are all pushing to monetize their traffic--large sites with lots of traffic and lots of growth--it matters.

What this means, in simpler terms, is that as long as the change in web/mobile traffic * the change in density of ads (ads displayed per user/visit/pageview, whatever) increases (S right) faster than the natural growth of advertising budgets * the change in share that web commands vs traditional (D right), the price of ads will fall. Judging by the double digit pageview growth rates and double digit revenue growth targets in most of these names, as well as addition of new entrants as more maturing services start monetizing, it looks a lot to me like advertising space, especially display ads, will see increasing price pressures down and, in my opinion, the market will segment into highly-targeted ads to a small sliver of the users that will command a high but shrinking premium, and the rest, which will be a margin-destroying race to the bottom. If you don't believe me, look at Yahoo!'s YoY change in ads price and ads sold and see if you can pinpoint when Facebook went public. Their price trends have deteriorated while their volume was improved, or rather their volume has improved because their prices have declined.

I think it's very likely that in the next 12 months we will find out what the price elasticity of demand for web advertising is and--particularly for display ads--I think it will be a lot more likely to surprise to the downside.

Tuesday, October 29, 2013

How Much is $1 of Earnings Worth?

Investors often treat reported earnings as the "yield" of equities due to the wide availability of reported earnings and the ubiquitous Price-to-Earnings ratio, however this can be a misleading measure of the actual economic value created for shareholders if earnings are overstated during cyclical upturns and the subsequent adjustment during downturns is reported as one-time items or direct-to-equity adjustments. I propose an alternate measure, a sum of the change in reported book value and dividends paid out as a proxy for the economic value created in a period.

Over the last 22 years, every $1 of S&P500 reported earnings has only led to $0.76 of dividends + book-value gains. If you exclude 2008, the year in which reported earnings and the alternate measure previously described differ most, every $1.00 of earnings generates only about $0.84 of economic value to investors. The sample, admittedly limited in size, was obtained using Bloomberg. Bold red line indicates a slope of 1, lighter black line is a simple OLS regression trend-line with no intercept.

Change in BV + dividends vs reported earnings

Change in BV + dividends vs reported earnings ex-2008
EDIT: A reader helpfully pointed out that I left-out the effect of buy-backs, which accrue to the price of shares but are functionally equivalent to dividends. Lacking a time-series of total buy backs, I can't correct the post to include that. Additionally, any buy-back made at a P/B multiple of > 1 would  dilute BV per share. Assuming book value is understated is a reasonable assumption for companies where marginal investment turns out to be profitable or during inflationary periods where the nominal value of capital stock is understated as a result. My apologies for these careless omissions. I will not remove the post, but I think it is worth noting that these omissions make the prior analysis basically useless.

Saturday, October 12, 2013

The short-term car rental service of the future

In the previous post, I wrote about how I envision smartphones replacing many of the functions of in-car entertainment and navigation systems. In this post I will be focusing on how that can affect automobile ownership and culture.

One of the small features that really blew my mind when I purchased my car was proximity-based keys which allow you to enter and start your car without having to ever put a key inside a lock. Not only that, but each key stores all sorts of preferences, from mirror and seat positioning and suspension settings, allowing each key holder to customize their driving experience. While I welcome this marginal innovation, it leads me to ask, “Why do we even need keys?” With the introduction of NFC and like-technologies into our smartphones, there’s no reason why keys can’t be replaced with our smartphones, the same ones which can then be docked into the car to load your personalized settings. Turning the automobile into a “thin-client” of sorts would allow any car to be your car, introducing a whole new opportunity set for the future of automobile rentals.

A couple of weeks ago I was in Paris for a couple of days. Having limited time and wanting to see as much of the city as possible during my short stay, I opted for renting a bike (velib) and riding around the city. One of the things that caught my eye as I tried to navigate roundabouts and punished my coccyx on cobblestoned streets was the presence of an electric car sharing system called Autolib. While it may not be suitable for many Americans, the success of Zipcar in dense, urban, locations is proof that there is a market for short-term car rental within close proximity to people’s homes. Improving this model by using electric cars with inductive-charging capabilities and reserved docking parking spots located throughout the city, eliminating the need to return the car to the same point of origin would remove some of the final inconveniences.

Imagine basic compact, inexpensive and utilitarian automobiles parked every couple of blocks. If you are a service subscriber, you simply walk-up to the car, use the NFC-capability of your phone to unlock it, and dock your phone on the dash to double as a navigation/entertainment display. The unique identifier of the car is recorded by the phone and submitted to a central database, and if your account is in good standing, a one-time use key is digitally delivered to your phone which allows the car to start. Your phone’s built-in navigation system includes all the car rental locations and availability of both cars and spaces is updated in real-time. A parking-assist function (much easier to implement in fully-electric cars) helps keeps dings to a minimum and internal diagnostic checks are run and uploaded at the time of docking the automobile, allowing the rental operator to easily keep track of maintenance needs and fleet location. Because your identity is linked to your phone and entry / ignition are recorded and associated to you, there is a strong disincentive to vandalize or abuse the cars, and no need to worry about someone breaking-in to steal a stereo--you’re carrying it out.

The uses for this range from short local round trips to places inaccessible by transit, to one-way trips for those nights when you may want to feel free to indulge in a few cocktails with dinner and not have to worry about driving home at the end of the night. Additionally, reducing the cost of not driving your own car out may lead to reductions in drink-driving, an additional positive externality. The logistic challenges are not trivial, but they are not great and all of the technologies listed in this post are already available today, it’s just a matter of someone putting it all together.

Mobile computing and the future of the automobile

I'm deviating from my usual focus on capital markets here to share some early thoughts on what I believe to be the future of the automobile industry.

Recently, after a decade of not owning an automobile, I purchased a car. The car comes with an in-dash entertainment/navigation system that would have likely blown my mind ten years ago but today, despite it being an award-winning top-of-the-line system, it feels nothing short of antiquated. I find myself constantly thinking, “My phone would be better at this.” The problem, in my opinion, is the mismatch between mechanical and computing life cycles. The median age of the American automobile fleet is 11.2 years according to consulting firm Polk. In a computing time-table, 11 years is an eternity. To put it in context, the first generation iPhone was released only 5 years ago, in June 2007. In that short span of time we’ve seen cellular data technology transition from GPRS/G2 to 3G to 4G to 4G LTE which represents an increase from ~150kbps effective speeds for non-EDGE 2G to 11-16mbps on the newer 4G LTE networks. In the last 4 years alone, mobile phone bandwidth has increased by a factor of 6 and screen resolution has doubled and we’ve seen the introduction of what is probably the most advanced voice-recognition technology available to the mobile consumer.

How is this relevant? Except for functions that are unlikely to change much in coming years like speakerphone capabilities or digital audio playback, the technologies at the center of modern in-car computers like data connectivity, A-GPS-aided navigation and voice recognition are evolving more rapidly than the replacement cycle of automobiles, leading to cars which have useful lives multiple times longer than the technology at the center of the user interface. Which brings us to the question; does the automobile really need an embedded computer?

As smart-phones become ubiquitous and the computing power available in them grows exponentially, it makes sense to allow our smartphones to become our on-board entertainment and navigation systems. Just like the embedded car-phone gave way to Bluetooth hands-free technology and trunk-loaded CD changers were replaced by mini jacks and Bluetooth audio, so will data connectivity and navigation / entertainment systems. It simply makes no sense to embed wireless connectivity circuitry that will be woefully out-of-date in 3 years to a machine expected to last 15 years or more. It artificially severely limits the useful life span of an automobile and non-luxury producers will soon adapt as they see the opportunity to reduce costs while touting it as an advantage for the type of consumer that can’t afford to replace a car every 3 or 4 years.

Once products like Ubuntu EDGE become a reality, there will be no need for auto manufacturers to include on-board entertainment computers anymore. Cars will feature a “dock” (wired initially, but eventually transitioning to wireless and featuring inductive-charging) and the necessary peripherals (microphone, displays, speakers, physical controls, if any). Your handset already has your music, your phone book, your address book, voice recognition, data-connectivity and A-GPS capabilities. Most importantly, your handset can (and does for many of us) back-up wirelessly and constantly to the cloud, ensuring that replacing our smartphone is as easy as buying and activating a new device and ensuring the possibility of massive data loss is minimal and the pain of transferring information to a new car (for example, a rental) is painless.

Friday, July 26, 2013

Muni Madness (redux)

If you follow me on twitter, you've seen this and can probably skip it.

In what seems like a yearly event now, with 3 consecutive years of this game, tax-exempt municipal bonds are once again cheap. It could be the overall bond sell-off, it could be the Detroit scandal, it could be bond fund outflows; I don't know what the cause is, but I do know the bonds are cheap.

Frequent readers know we are generally fans of quantitative methods, so expect some numbers.

In the last 5 years, our preferred proxy for the tax-exempt AAA market has disconnected itself the ratio range that was common until the global financial crisis. Due to the tax-exempt nature of muni bonds, benchmark yields were often quoted as a percentage of treasury yields instead of a spread, but this has become an increasingly inaccurate way to look at yields. We consider two options:
  1. The market underwent a structural change by which the old measures became invalid. An example of a similar event would be the '87 crash after which options began exhibiting (steeper) volatility smiles. In this case we would consider the structural change to be either a transition from rate-based to an credit-based market, a change in the way liquidity is priced post-GFC or a combination of the two.
  2. That the "fair-value" ratio is itself a function of interest rates and tax-exempt bonds exhibit negative convexity as yields approach a floor, a relationship which had not been visible until the introduction of ZIRP brought treasury yields down far enough for the relationship top be visible.
 Although we agree with the assertions that, market-wide, the tax-exempt market has transitioned from rate to credit, we believe the AAA benchmark is exempt from this because of its information insensitive nature (for more on this see Slapped in the Face by the Invisible Hand by Gary Gorton) and while we generally agree about the repricing of liquidity post-GFC, we believe we have the ability to control for that (more later). A review of the data from 2001 to present day as well as our expectation of behavior by participants makes us believe the second option, that the "fair" ratio is dynamic, is the likely explanation.

Although we do not put much weight on the automatically generated regression, we believe it to suffice as evidence that tax-exempt bonds tend to exhibit negative convexity, possibly from an absolute rate floor. Below you can observe 2001-present and June 2009-present time series with accompanying overlaid histograms showing ratio changes over time and the distribution of values within those time periods. Please notice the increase in ratios as yields fell to new lows as the adoption of forward guidance and quantitative policy by the Fed lowered 10y rates to new post-WWII lows and the apparent reversion of this trend after the secular low was printed in June 2012.

Continuing this general line of thinking, we explore a linear relationship between the AAA 10y tax-exempt yield and the 10y treasury yield and find strong evidence of a nearly constant  sensitivity of municipal yields to treasury yields of 0.587. 

We expand this simple example into a multiple regression incorporating the MOVE index and the VIX index as a measure of treasury-implied volatility and a proxy for CDX.IG respectively. This allows us to internalize other credit-insensitive spread products into the model, namely MBS (for more information read about MBS replication) and high-grade corporate bonds. 

To test for structural changes pre- and post GFC-, we slice the data into 4 time periods:
  • pre-GFC:  prior to January 1st, 2008
  • GFC: January 1st, 2008 - June 1st, 2009
  • post-GFC: subsequent to June 1st, 2009
  • Complete: all data
The results are reproduced below:

      Min        1Q    Median        3Q       Max 
-0.260117 -0.066022 -0.004478  0.067822  0.227960 

                          Estimate Std. Error t value Pr(>|t|)    
(Intercept)              0.8584473  0.0255899   33.55   <2e-16 ***
pre_crisis_data$USGG10YR 0.6018459  0.0049374  121.89   <2e-16 ***
pre_crisis_data$MOVE     0.0019368  0.0001343   14.42   <2e-16 ***
pre_crisis_data$VIX      0.0064424  0.0004559   14.13   <2e-16 ***
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 

Residual standard error: 0.09118 on 1719 degrees of freedom
Multiple R-squared: 0.9001, Adjusted R-squared:   0.9 
F-statistic:  5165 on 3 and 1719 DF,  p-value: < 2.2e-16 

     Min       1Q   Median       3Q      Max 
-0.66890 -0.17289  0.05204  0.15769  0.59407 

                     Estimate Std. Error t value Pr(>|t|)    
(Intercept)          1.770108   0.113410  15.608   <2e-16 ***
crisis_data$USGG10YR 0.294021   0.027904  10.537   <2e-16 ***
crisis_data$MOVE     0.005621   0.000594   9.462   <2e-16 ***
crisis_data$VIX      0.003370   0.001531   2.201   0.0284 *  
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 

Residual standard error: 0.2308 on 349 degrees of freedom
Multiple R-squared: 0.517, Adjusted R-squared: 0.5128

     Min       1Q   Median       3Q      Max 
-0.30228 -0.08212 -0.00518  0.06629  0.47018 
                          Estimate Std. Error t value Pr(>|t|)    
(Intercept)               0.981564   0.016827  58.331   <2e-16 ***
post_crisis_data$USGG10YR 0.519255   0.006456  80.426   <2e-16 ***
post_crisis_data$MOVE     0.002171   0.000249   8.718   <2e-16 ***
post_crisis_data$VIX      0.006691   0.000766   8.735   <2e-16 ***
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 

Residual standard error: 0.1195 on 1034 degrees of freedom
Multiple R-squared: 0.9322, Adjusted R-squared: 0.932 
F-statistic:  4736 on 3 and 1034 DF,  p-value: < 2.2e-16 
F-statistic: 124.5 on 3 and 349 DF,  p-value: < 2.2e-16
     Min       1Q   Median       3Q      Max 
-0.43382 -0.06802 -0.00074  0.07707  0.38103 

                   Estimate Std. Error t value Pr(>|t|)    
(Intercept)       0.8175504  0.0107706   75.91   <2e-16 ***
all_data$USGG10YR 0.6160607  0.0021784  282.81   <2e-16 ***
all_data$MOVE     0.0014400  0.0001251   11.51   <2e-16 ***
all_data$VIX      0.0068161  0.0004370   15.60   <2e-16 ***
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 

Residual standard error: 0.1128 on 2757 degrees of freedom
Multiple R-squared: 0.9726, Adjusted R-squared: 0.9726 
F-statistic: 3.268e+04 on 3 and 2757 DF,  p-value: < 2.2e-16 

As you can see, outside the GFC-period, all of the variables are statistically significant, and the models exhibit high R^2s. Although there is mild colinearity between VIX and MOVE, the correlation is limited and the results are not invalidated. A test for heteroskadicity reveals no significant skerodasticity outside the November, 2008 credit meltdown.

The following table displays the correlation matrix between the residuals of various estimation methods:

               all   pre-gfc  post-gfc   non-gfc
all      1.0000000 0.9957032 0.8588698 1.0000000
pre-gfc  0.9957032 1.0000000 0.8949263 0.9957032
post-gfc 0.8588698 0.8949263 1.0000000 0.8588698
non-gfc  1.0000000 0.9957032 0.8588698 1.0000000 
Based on the very high correlations as well as a review of the residual time series, we don't believe there was a significant structural change after the GFC and believe the benefits of using a strictly post-GFC distribution are outweighed by the lower sample size. For the following illustrations, we use the model regressed from the full sample. A higher residual value represents a lower than predicted yield for the MMA 10y benchmark and a lower value represents a "cheaper" MMA 10y. The first chart illustrates the last year, where the "basis" is the residual of the regression, in percentage points. The second is a time series of residuals over the entire sample.


We think high-grade tax-exempt bonds are cheap. The yields we are seeing right now are usually associated with higher rate and/or higher volatility periods. Additionally, given the apparent turn in interest rates, we see the lower rate-sensitivity and accompanying negative convexity as fundamentally attractive features in a rising-rate environment. The loss in "upside" from the negative-convexity is more than compensated, in our opinion, by higher pre- and post-tax yields, and there remains significant price-upside if yields normalize to predicted levels in a sideways-rates market. We believe ample liquidity and accomodative monetary policy limits the risks of a 2008-like tail scenario, despite what a certain blonde talking-head (who, incidentally, triggered the last sell-off of a similar magnitude in early 2011) with no P&L to speak-of may be saying on TV.  Depending on how much zerohedge you read, your mileage may vary.

Tomorrow we will attempt to revisit the topic from a different angle, comparing tax-exempt credit to corporate credit to illustrate the difference in priced-in losses and then, sometime this weekend, we hope to touch on some vehicles the average investor can use to express a view on the sector.