Sep 21, 2017

Weekend Links - 9/21/17

QUOTE OF THE WEEK

…“more” is not always the best way to “Great”.  Fritz 

GRAPHIC OF THE WEEK




2050 here would be age 65 retirement 
Gordon Irlam AAcalc.com   
Rather than a rule of thumb it is also possible to compute an optimal variable withdrawal strategy if it is assumed annual returns are from a known distribution that is independent over time by using the Stochastic Dynamic Programming (SDP) algorithm ... To compute the variable withdrawal strategy using SDP it is only necessary to treat the withdrawal amount as a dimension to be optimized over analogous to an asset allocation dimension. That is working backwards by age, for each portfolio size, we consider every asset allocation and withdrawal amount and pick the best one.


RETIREMENT FINANCE AND PLANNING

Much of financial planning focuses on retirement. But what if your client isn’t planning to retire? 

The purpose of this research is to develop a multivariate PDF for asset returns that is suitable for quantitative retirement plans. The model fits any set of returns, however the curse of dimensionality will limit the number of securities. We propose a multivariate mixture having fixed mixture marginals using normal components. The model is motivated by the claim that a lognormal PDF is virtually indistinguishable from a mixture of normals. Whereas the lognormal PDF is intractable with regard to weighted sums, the normal mixture is not. The lognormal PDF is only justifiable when short-term returns are iid and the PDF is CLT-compatible for the given sample size. A typical retiree could endure several market crashes and we should not expect the historical sample to represent all possible extremes. We can stress test a retirement plan by subjecting it to a return PDF that has been fit on the historical sample seeded with black swan events. The normal or lognormal PDF are unhelpful in this regard as neither can accommodate such outliers.  [commentary in note [1] ]
  
Safe Withdrawal Rates – Part 20: More Thoughts on Equity Glidepaths, ERN
Historical simulations show that an equity glidepath is useful when the CAPE is high at the commencement of retirement. As it is today! If the CAPE is below 20, glidepaths are of no use and an aggressive static equity allocation (close to 100%!!!) has performed best in historical simulations! … Monte Carlo simulations miss important elements of real-world data, i.e., mean reversion of equity valuations and changing asset return correlations. Hence, glidepaths that were calibrated to do well in Monte Carlo simulations (Kitces and Pfau) tend to do poorly in historical simulations. Unless we believe that the past observed dynamics of equity returns no longer apply in the future, we should disregard the Kitces/Pfau glidepaths because they’d likely perform worse than even most static asset allocations.  



 [commentary: duh!]


MARKETS AND INVESTING


Swedroe's war on Dividends
[he is either correct, in which case all of this warring is unnecessary or he is incorrect in which case he is misleading.  I think he is a little of both.  From a traditional finance theory / Miller Modigliani perspective (and some empirical evidence to boot) he is correct and dividend-focused strategies don't matter or maybe have some distinct disadvantages.  On the other hand I believe there are trading situations, multiperiod decumulation scenarios, and some utility arguments that at least would give me pause before throwing div strategies out the window.  I have posted on this before… I like dividends for reasons that are not in Miller (or Swedroe)]

…every individual is unique and neither expected utility theory nor prospect theory appropriately capture the diversity in risk tolerance. This paper seeks to make Kahneman-Tversky’s research on prospect theory/behavioral economics, and their value function practical and user-friendly, thus improving investment decision making. 


ALTERNATIVE RISK

After examining all the signals, the authors find only a handlful of trading strategies that are “anomalous” and most of these strategies make no economic sense! Now the authors do assume (through their tests), that the Fama and French 5-factor model plus momentum explain the cross-section of stock returns (so all the classic characteristics we all argue about are controlled for in the study), but the author’s main contribution is that there is little to no evidence for additional anomalies. "If we properly account for the statistical properties of the data-generating process and use the FDP approach, we are left with a handful of exceptional investment opportunities. If we adopt an all-together conservative approach and control FDP at γ= 1% (i.e., we accept one per cent of lucky discovery among all discoveries on average or in our sample), we reject all the two million strategies." 

the correlations between the active returns of individual systematic managers are very low, comparable to those between discretionary managers. We present empirical evidence of systematic and discretionary managers’ performance and risk, concluding that neither group has been inherently better than the other one and that they have historically been good complements. 

The use of a systematic and disciplined investment and research process is an effective way of reducing the confirmation biases by making the decision process explicitly based on a set criteria that can be tested. Decisions rules can be tested against past data and reviewed against future performance. There may still be biases based on the weighting of the evidence, but a systematic process can allow for testable analysis. Systematic investing can eliminate one of the key psychological problems facing investors. [couldn't have said it better…] 

Ways to manage risk, Blue Sky Asset Management
Dynamic hedging strategies that attempt to replicate option payoffs are by far the cheapest form of 


SOCIETY AND CAPITAL


[NYC+50M is not what the rest of the world is like, certainly not my world.  This feels more like NY gossip than a rational, systematic study. It may be a worthy conversation to start no doubt but this feels a little politically fashionable and tendentious before we even get to the facts.]

All cultures are not equal. Or at least they are not equal in preparing people to be productive in an advanced economy. [article would require a micro-aggression warning at most colleges today; this story made it to the WSJ this week and is getting some serious pushback. see also this nymag post

Most younger men ended up with less because they started out earning less than their counterparts in previous years, and saw little growth in their early years. They entered the work force with lower wages and never caught up. 

We report data from double-auction experiments in China and the U.S. using groups of exclusively females, exclusively males and mixed gender participants. We find that female groups in China generate price bubbles statistically identical to those produced by exclusively male groups in both China and the U.S., all of which are significantly larger than the bubbles produced by exclusively female groups in the U.S. Our results suggest that gender differences in financial markets may be sensitive to culture. 

The general literature documents that, outside Wall Street, there is persistent gender gap with few women at the top. To climb the corporate ladder you need both outstanding performance and positive subjective evaluations by others. This paper adds to this literature because it shows asymmetries between how men and women benefit from social ties. 

A Price on Your Head, humbledollar.com
three assets with potentially significant value are our regular paycheck, our Social Security retirement benefit and any traditional employer pension we’re entitled to. 

Social proof is the idea that we look to others to figure out what the correct behavior should be. We follow narratives instead of evidence. It feels more comfortable to go along with the crowd when making tough decisions because we look at what others are doing in times of uncertainty.

The real innovation at the heart of what’s been named the “sharing economy” is the realization that such peer-to-peer matching can transform markets for services such as transportation, lodging, and general errands. Those who carelessly speculate that these platforms will change the role of private property will inevitably be disappointed—the so-called sharing economy is another way that free market forces have evolved to put that property to its best use. 







---------
[1] the paper above (Multivariate Density Modeling for Retirement Finance) was a wee bit above my pay-grade and might be geared more towards tenure committees and co-academics rather than even the most advanced practitioners.  The paper is an exercise in quantitative virtuosity. The quantitative virtuosity, however, will neither uncover a way to predict the future nor will it extract new money out of thin air so while it might move the academic needle a bit it will not change the basic retirement game in any fundamental way (yet), a game whose rules are to allocate finite resources over an unknowable future.  (see my recent hurricane post; even the best quantitative forecasts go stale really, really fast so the game is not to build a better forecaster it is to do faster cycles of analysis and adaptation). I mean, if he finds more risk, we will spend less early to account for that since there are only so many ways to cut a pie. If he finds less, we'll maybe spend a little more or add to legacy. But either way the need to re-evaluate periodically won't change in the slightest. 

The wisest things in this paper are not necessarily even the math.  The good stuff is 1) a fairly subdued consideration of stress testing retirement plans, which makes sense, and 2) an idea for seeding a return distribution with some extra black swan events. That last I like and has been on my drawing board for a while.  I did it once before with "spend shocks" with predictable results and I was thinking of doing it again with a return distribution in my simulator in a similar fashion.  Right now I either sample with replacement from an S&P distribution for equities (in addition to bonds; I get a decent historically-derived non-normal distribution this way) or I sample from a skewable (haven't found a good kurtos-able one yet) distribution that represents an alt risk premium or a third asset. I know I can either customize the distribution -- or an equivalent probability distribution vector for sampling -- in order to wreak havoc on the downside or alternatively introduce crashes as a random event as a layer over the regular process.  Either way, one gets a robust test of the downside; who worries about the upside risk anyway? (remember that simulation all by itself, if run enough times, will find some pretty bad paths that are way worse than the historical record. That is the point of simulation. I guess I'm not really sure how making it even worse than that adds a ton of value other than confirming the obvious: spending a ton early in retirement -- if you happen to have the luck of the draw later and superannuate -- is risky). 

On another note, this paper is a good example of what I was talking about in a past post. New good ideas, and there are no doubt some new good ideas in here, need to find a more efficient way to flow downstream to normal advisors and at least some advanced retail retirees. That won't happen here. The paper is too dense so I know I can't do it and the probability that retirement researchers or advanced practitioners will look at this and then digest it and publish something usable in a meaningful timeframe for me personally is doubtful.  This thing will sink into the analytic sands for a long while.


No comments:

Post a Comment