Asset Allocation in the Real World
The
long-term return of your investment portfolio will be dominated by the asset allocation
between stocks and bonds that you select. As crucial as this question is, there
is no generally recognized solution for determining this. In the following we
will discuss the classical economy, a quantitative solution used often in the
financial industry and suggest another much simpler common-sense solution that
relies upon the intrinsics of the situation.
In
any time period, the stock market generates a lot of data. It is therefore a
fascinating topic of study for the quantitatively inclined who try to deduce
patterns or principles. But, as the Financial Crisis of 2008 illustrated,
quantitative finance beyond a simple momentum analysis is problematic. The
reason for this is embedded in engineering control, which studies how
information controls the behaviors of dynamic systems.
Control
System
The negative feedback control
system model summarizes a simple classical economy with fixed supply and demand curves.1 It
reaches equilibrium in one step:
|
* The larger the discrepancy between the amount demanded and the amount supplied, the more prices adjust to equalize both. This simple price-auction model forms the basis of classical economics, where there are no market imperfections. The classical model makes at least eight important assumptions that are almost always forgotten in economic debate
|
The
classical model results in timeless economic equilibrium, where supply=demand
at the equilibrium price. But perfect markets suffer from disturbance,
imperfect competition, externalities, social herd behavior and imperfect
information about the present and future.2 The question we now ask is what
amount of information imperfection is tolerable?
Recall
the cross-continent flight. A passenger steps onto the aircraft; it takes off,
levels out at 35,000 feet for a while and then descends. He steps out of the
aircraft, benefiting from the large number of sophisticated adaptive control
systems that keep it flying. The goal of adaptive control is to achieve an
acceptable level of performance when the parameters of the model are unknown
because the model is disturbed. Thus, in the real world, the altitude and
temperature change, storms occur, the control system encounters wear and so on.
According to Landau and Karimi (2009), an adaptive control system is a
hierarchical system consisting of a conventional negative feedback system and
an adaption loop that can appropriately change model parameters.
What
is of interest is when it is not possible to build an adaptive control
system. According to the authors:
1) It is possible to build a control
system when the model is unknown, but the disturbance model (of the
classical economy) is known.
2) It is possible to build a control
system when model is known, but the disturbance is unknown.
3) When both the model and the
disturbance are unknown, then this is a “very difficult problem” because it is
hard to distinguish the prediction error of the model from the disturbance
model error.
The
third is exactly the problem that lies at the core of quantitative finance,
because short-term markets, mainly influenced by events, do not have a
reference point if the parameters of the model of supply and demand are unknown
and if events affect the emotions of participants. Thus, to cite an example,
mathematical asset allocation by Modern Portfolio Theory requires that the
markets be placidly Gaussian, that their returns are stationary with a known
mean and variance. That is, this highly mathematical analysis of optimal
portfolios requires two reference points.
But
citing data from Kindleberger (2000 ed.), markets are subject every ten years
or less to manias, panics, and crashes. Market reference points change due to
changes in financial regime, as our article, “The Nature of Stock Market
Equilibrium" notes. (Our stock market analysis states the above model in econometric form under the financial conditions, 1968-1999.) Therefore, the precise idea of risk-return
optimization promised by Modern Portfolio Theory is not possible. Further,
because the free market is the major agent of globalization and social change,
we think the U.S. stock market can be described by error-correcting mathematics
only if the Fed's monetary policy can modulate growth. Otherwise, it will
violently error-correct.
In
general, Charles Lindblom (2001) of Yale writes, “Market systems, wisely or
foolishly, largely in effect give up the possibility of an efficient resource
allocation and pattern of production. They settle instead on inefficient
allocations improved to the limited degree the voluntary transactions make
improvement possible.” 3 Catalyzing “creative destruction,” markets
are inefficient; but they are free and cause large social changes.
Intrinsics
So,
how might our readers 4 deal with financial markets? Since markets are the
major agent of change, they likely do not have any intrinsic nature (unless effectively
modulated by institutions). The intrinsics of portfolio management are not to
be found in the market, but in the natures of the portfolio owners and their
investments.
The
first intrinsic is the simplest, the owner's age. It is common sense that older people
should take less risk and that younger people should take more. The reason for
this is not only the conventional reason, that older people have less time to
make the money back. It is also that money is accumulated over a career, and it
should be handled carefully. This is why portfolio owners should know their
investments. It makes no sense to be skilled in a career and then be careless
about the money it has earned.
The
second intrinsic, for a value investor, is the intrinsic value of the company
they are invested in. Although it is possible to calculate an “intrinsic value”
for the many companies that have positive cash flow after capital costs, there
are fewer companies that meet the second assumption of value investing – that
there is a franchise, or moat, around their businesses.
Which
leads to a discussion of why we are not replacing stocks after Mr. Market has
priced them at intrinsic value. The issue is quality.
Value
investing has been likened to trash can investing; but that doesn't always need
to be so. By “quality,” we mean credit quality, when a company has ample cash
flows and low debt. They are the kind of companies that banks want to lend to,
but which don’t need the money. As the economic cycle progresses, Mr. Market
will take these companies to fair intrinsic value, and then beyond. Since value
portfolio managers must usually stay totally invested, they have to find
low-priced value stocks. Then they can be caught in “value traps,” companies
that appear to be values by the numbers, but aren’t. In the beginning of 2008,
a noted value investor (somewhat too stridently) said that he was invested in
the financial sector (AIG, Wachovia, Bear Stearns, Freddie Mac). His stock
portfolio dropped by 55% in that year. This was not a “reversion to the mean”
in competitive markets; it was a mistake.
Asset
Allocation
As
we discussed in our 12/1/12 note,
the likely return of stock market is going to be around one-half that in the
past. Therefore to reduce risks, it makes sense not to give up on quality. This
means selling high quality companies like Coca-Cola, when they are fully
priced, and leaving the proceeds in cash until the next downturn. The
age-related stock allocation (100-a portfolio owner's age) ± 10% is still
suggested, but only when buying in phases during a downturn – when high quality
stocks become bargains. This policy will reduce portfolio risks, not damage the
financial system and keeps stocks from becoming overpriced.
1 We have modified the more general economic model developed by Heylighen (1997).
2
Models
are useful for quantitative or qualitative discussion.
It
is increasingly possible to accurately model complex phenomena such as the
weather or driving a car using the brute force approach of powerful computers
and lots of data. Brynjolfsson and McAfee (2011) describe how Google solved the
problem of automatic driving on populated roads: “…(it) is an enormously difficult task, and it’s not easy to build a computer that can substitute for human
perception and pattern matching in this domain. Not easy, but not impossible
either…The Google technologists succeeded not by taking any shortcuts around
the challenges (our note)…but instead by meeting them head-on. They used
the staggering amounts of data collected for Google Maps and Google Street View
to provide as much information as possible about the roads their cars were
traveling. Their vehicles also collected huge volumes of real-time data using
video, radar…gear mounted on the car; these data were fed into software that
takes into account the rules of the road, the presence, trajectory, and likely
identity of all objects in the vicinity, driving conditions, and so on. …The
Google vehicles’ only accident came when the driverless car was rear-ended by a
car driven by a human driver as it stopped at a traffic light.”
The
social sciences deal with subject matter a lot more complicated than the above;
the best that can be presently said is that quantitative methods can be useful
if conditioned upon some state. For instance, the stock market is likely to
increase x %, if Congress compromises and interest rates don’t increase.
Otherwise, qualitative economic models are useful to discuss whether an
analysis has captured the essentials.
The
control system idiom is slightly more complicated than the supply-demand curve
discussions in economic textbooks. It is, nonetheless, an efficient analysis
because it allows the analysis of quantitative system feasibility, encapsulates
for further qualitative discussion classical economics and allows
discussion of another really crucial phenomenon, positive feedback. Risk
consultant Sean Harkin writes:
|
…observation of reality must be the only road to
truth…. If GDP growth is dependent on a feedback loop of
ever-improving technological ability, and if companies do not have adequate
incentive to fund the long-term scientific research and the twelve to twenty
years of education per person needed to support this, then the state must
take a role. Likewise, if markets are prone to wild swings in asset prices,
credit, general price levels, jobs, GDP and other key variables; then
governments must regulate the behavior of key actors (especially financial
institutions) to mitigate this tendency, and must move the money supply and
their own budget deficits or surpluses in a way that counterbalances whatever
cyclicality remains-an insight that can be traced at least to Keynes’ General
Theory in the 1930s. Finally, if there is a tendency to produce far more
inequality than basic incentives require, then there must be redistribution
to counterbalance this. But we know from history that communist central
planning, however well-intentioned, leads to inefficiency and oppression. So
the state cannot take over entirely either. The only answer is a mixed
economy. Classical economics is very different in its
thinking. It assumes that behavior is relentlessly rational - putting it
flagrantly at odds with known psychology (our note: and imperfect information
about the future) - and that individual rationality leads to collective
rationality. For example, rational people would always be willing to work for
a wage that leaves some profit for the employer and it should therefore
always be possible to arrange a beneficial exchange of work for money,
ensuring that the economy never strays from equilibrium where the only
unemployed are those who are moving jobs. If there is, transiently, some
unemployment, it must because real wages are so high that firms cannot offer
additional jobs without running a loss. Rational people understand this
(including complications relating to inflation or deflation) and quickly
respond by accepting lower wages, bringing the economy back to equilibrium. Classical equilibrium theory, in essence,
emphasises only negative feedback, neglecting that, in every case, there is
also positive feedback. On unemployment, for example, significant job losses
lead to reduced spending, harming additional firms, causing more job losses
and causing prices to fall in a way that makes those who still have an income
hold back on spending because they
expect things to be even cheaper in the future. (our note: Therefore, the Fed’s
fear of a deflationary spiral.) A positive feedback loop is thus established
that drags the economy towards collapse unless the state intervenes to stop
it. Classical economics concludes that all such interventions are
unnecessary, and that pure free markets are always ideal, simply because it
focuses on negative feedback and neglects positive feedback, ignoring half
how the economy actually works (our note: That is how people interact in society).
If economists want to be more like scientists –
which they clearly do, borrowing all the maths of physics along with terms
like “equilibrium” and “elasticity” – then they should give a central role to
the simple concept of (positive) feedback.
|
3
Lindblom,
Charles; “The Market System;” Yale University Press, New Haven; 2001; p. 174.
Current
mathematical economic research extends the rational economic model to the nth
degree, assuming that people are “relentlessly” rational. In spite of this,
economists can be very expert because they are very bright people. But experts
differ on practical economic policy, because they are making different
assumptions: classical economists assume that that the social effects are only
the sum total of rational people acting in their own economic self-interest
with perfect knowledge. The conventional economic paradigm is too simple, being
but an element of social life.
4
Federal
regulation requires that financial publishers give only impersonal investment
advice. The pertinent regulation boils down to this, don’t use, “you." The
reason for this regulation is prior financial crises.