Broken Symmetry

The Ensemble of Parametric Oscillators Model of the Economy

Markets can be modeled as ensembles of parametric oscillators.  The parametric oscillator model is the simplest model that is useful in understanding dynamic market prices.  For non-physicist readers, you've made a parametric oscillator whenever you've pumped your legs on a swing to change your frequency of oscillation.  If you've ever had somebody push you, then you've made an amplified parametric oscillator, which is equivalent to a market hooked up to a time-varying external money supply.

Supply can be modeled as an ensemble of oscillators, one for each person.  The cumulative frequency distribution of the supply ensemble is equivalent to the aggregate supply available to a market within a window of time.  Demand can be modeled as an ensemble of oscillators, one for each person.  The cumulative frequency distribution of the demand ensemble is equivalent to the aggregate demand available to a market within a window of time.  See here.  Elasticity is a function of the fatness of the frequency distributions at the half-maximum to their peaks.  The distributions will be poissonian in shape.

Both cumulative distribution functions can be parametrized in terms of the opportunity cost of any scarce resource within an economy, or in terms of a currency that does not vary fast with respect to other currencies within the size of the time window.  (Doesn't that explain why we use currency rather than bartering?)

The "temperature" of these ensembles (i.e., the shape of the distribution for a given amount of capital when scarcity and size of the ensemble are fixed) will be a function of the capital available.  Similarly, other changes in the cumulative frequency distributions of supply and demand will be a function of capital (energy), scarcity (volume), and the size of the ensemble (pressure).  If the changes are made slowly with respect to the time windows within which the distributions are measured, then convexities in the function of frequency with respect to increasing capital, decreasing scarcity, and increasing ensemble size may be observed.  Certain ranges of capital, scarcity, and size of the population will be characterized by certain types of structures.  In other words, as capital, scarcity, and size of population are tuned through different ranges, spontaneously ordered structures for the allocation of capital and resources throughout the ensembles will emerge.  Thus, the parametric oscillator model is consistent with a thermodynamics of institutional design.

Thermodynamics gives us no insight into how and when change will occur.  But the parametric amplifier model also permits an insight into market dynamics.  According to this model, the ensemble of supply oscillators  couples nonlinearly to the ensemble of demand oscillators.  Mathematically, the mechanism for coupling is analogous to a damping force on each ensemble that is, in part, a function of the frequency distribution for the other ensemble.  In other words, the oscillations of the two ensembles don't simply add or subtract from one-another.  They can multiply or divide one-another.

In practice, the coupling mechanism might be provided by anything that causes the frequencies of the ensembles to multiply rather than add, such as transactions costs or liquidity constraints that do not vary linearly with the quantity of goods exchanged.  Study of models of the coupling mechanism will be one of the most fruitful areas of research for econometricians.  For the coupling mechanism is not simply a function of the frequency of the supply and demand ensembles of the market in question.  Rather, it is a function of the frequency distribution for any supply or demand ensemble with non-trivial cross-elasticity with the supply and demand ensembles for the market in question.  The coupling mechanism, including the phenomenon of cross-elasticity, is the dynamic mechanism that describes how and when phase transitions will occur.

Note that variations in external money supply would be a source of capital to the supply or demand ensembles that should be considered separate from the coupling mechanism.  Thus, an increase in external money supply might give rise to parametric amplification.  Variations in external money supply add many complications to understanding the dynamics of parametric oscillators.  Having a Taylor rule that describes how the external money supply varies in time makes the model easier to solve.

Parametric oscillators exhibit many interesting dynamics.  One is the phenomenon of parametric resonance, whereby the ensembles may become synchronized in phase.  Phase synchronization is an implicit or explicit characteristic observable in all markets.  Another is the phenomenon of parametric instability.  Price bubbles can form when the resonance peak (or peaks) are too high-frequency to be sustainable.

For the Hayekians out there, given constant resources and population, as capital is removed from the system, spontaneous symmetry breaking will result in new spontaneous ordering of capital, resources, and population within the market.  In other words, holding two out of three of capital, resources, or population fixed, and minimizing the other variable will lead to more spontaneous order within society.

As an end note, the wave equation necessary to the parametic oscillator model will not apply over longer time scales.  Wave equations are second-order in time.  For very large time windows, dissipative forces will have more noticeable effects, and a heat equation (like the Schrodinger equation) will provide a better approximation of dynamics.  The difference in observable dynamics at different time-scales is part of why microeconomics and macroeconomics are not readily joined in econometric theory.

Does the long tail result from a power law?

No. Wired's Chris Anderson published an article and later book on the "long tail" that characterizes demand for information.

The hypothesis advanced by Anderson and others is that this long tail results from a power law in demand -- i.e., that the slope of the demand curve declines algebraically rather than exponentially, even at high levels of quantity.  Others have noted how this would be a major coup in economics, since the law of demand from economics predicts a faster than algebraic decay.

Let me suggest how this paradox might be resolved.  In fact, the distribution of consumption for information is not governed by a power law.  Rather, the distribution is poissonian in shape.  The poisson distribution can be approximated with power laws when the mean frequency of events is close to zero.  But unlike the power law distribution, the poisson distribution approaches zero at zero frequency.

The effect of these differences is that the cumulative distribution function -- i.e., the aggregate demand, still has a well-defined peak in quantity.  The fat-head is finite.

This mistake is easy to forgive since people confuse local trends with power laws all the time.  I may blog more at some point about how the notion that a power law applies to demand for information is misleading.

On why we should add the Fundamental Hypothesis of Periodicity to Rational Expectations Theory

John F. Muth simultaneously blessed and cursed the field of dynamic econometrics when he published his "Rational Expectations and the Theory of Price Movements" in 1961.  For while the fundamental rational hypothesis is useful for identifying equilibrium measures of price and quantity given static supply and demand curves, economists have always known that supply and demand curves are not static.  Hence, we should only expect rational expectations theory to provide accurate forecasts of market prices when changes to the supply and demand curves are much slower than the time between transactions.  Happily enough for many economists, this condition happens often to obtain in the publicly traded markets.  Most of the time.

Even while acknowledging the theoretical elegance of the rational expectations theory, some economists have noticed that price dynamics are not always random.  Some of these -- shall we say stubborn -- economists have attempted to account for these dynamics by (in various ways) relaxing the strict requirements of the fundamental rational hypothesis.  By considering a more limited hypothesis of "bounded rationality," these economists have been able to explain how some price dynamics result from asymmetries in the information gathering and processing capabilities.  In other words, some price dynamics result because some market participants are faster or more accurate than others at gathering and processing information into a forecast of market price.  Indeed, investors like Warren Buffett, Charlie Munger, and Ken Heebner have made their shareholders wealthy much faster than those shareholders would have been had they invested in indexed mutual funds.

The evidence that rational expecations theory is flawed has been there from the beginning.  Why then, one might ask, has nobody yet introduced a better hypothesis?  The answer is about what you'd expect.  The answer is that (as John F. Muth knew well) it's a perfect mathematical mess to rebuild dynamics from any other hypothesis.  There really is not a more general rule for forecasting how people will act.

There is not a more general rule.  But there is, on my view, a rule about how people will act that is equally general.  I have dubbed it the fundamental hypothesis of periodicity.  What it posits is the following:

The activities of people change in time according to rhythms that, averaged over a population within a period of time, have a characteristic distribution in frequency.

Following Ockham, one should ask of any new hypothesis, "What observable predictions does this hypothesis permit that cannot be made from the existing theory?"  But thanks to the discussion above, the answer to Ockham should already be apparent: Unless the hypothesis of rationality is modified, another hypothesis is needed to explain price dynamics.

By clicking through the link above, you can find links to a paper excerpt I've completed, which explains how the supply and demand curves of introductory economics can be reconstructed from the frequency distributions that characterize the observable patterns of consumption and production for a population of people within a window of time.  The aggregate supply and demand curves taught in introductory economics are in fact the cumulative distribution functions that correspond to these frequency distributions.

Now for the payoff:

Nobody, to my knowledge, has suggested a way to measure time-varying supply and demand curves.  The concept of "equilibrium price," on which rational expectations theory depends in explaining price dynamics, carries within it the assumption that supply and demand will not change on a time-scale relevant to price dynamics.  As noted above, this assumption often (but not always!) obtains in the market for publicly traded securities.  But when the assumption fails, the concept of price is no longer well-defined for rational expectations theorists.  When supply and demand are not in equilibrium, price is not in equilibrium.

Thus, what the fundamental hypothesis of periodicity promises is a brave new world for dynamic price theory -- a world of price without equilibrium.

Reprise: the connection between Supply and Demand and Frequency of Production and Consumption

When Ragnar Frisch first did a Fourier analysis of macroeconomic cycles, he guessed the parameters for the frequency distributions that would give rise to the observable cycles.  But he noted:

"I believe that it will be possible by appropriate statistical methods to obtain more exact information about them."

He was right.  You can actually measure the darn things by watching the patterns of consumption and production for a given population of people within a particular window in time.  Elasticity arises from the fatness of these distributions in frequency.  Evolutions of the frequency distributions in time accounts for diminishing marginal utility.  The aggregate supply and demand curves taught in introductory economics are merely the integrated cumulative distribution functions that correspond to the underlying frequency distributions.

What causes exogenous shocks?

An important result (see especially page 92) from research by econometricians interested in Fourier Analysis in the 1930s was that the "energy" that caused business cycles was contributed by forces external to the particular market for goods or services.

Perhaps not surprisingly, that's how economists have modeled markets ever since!  Think about the subprime mortgage market, for example.  The way economists model the bubble in the subprime market is by modeling in an exogenous shock to the money supply.

What's wrong with this analysis?  Well, lots of things.  I don't know about you, but as I've been hanging around the U.S. for the last few years, and I didn't notice that it was actually raining money in any of the neighborhoods where subprime mortgages were being offered.

Now of course economists don't literally believe that it was raining money anywhere when they model market bubbles with exogenous shocks.  They know that the money came from somewhere.  In fact, they know pretty much precisely where it was coming from -- big institutional investors who put money into CDO^2s, CDOs, and MBSs.

But if they know where the money was coming from, then why can't they have a more sophisticated model of the subprime market, a model that doesn't require external forces (i.e., raining money) to understand why a bubble formed?

The answer is that they can.  By undertaking a Fourier analysis of cycles of demand and supply in the markets for subprime mortgages, MBSs, CDOs, and CDO^2s, and their substitutes and complements, economists could model the entire mortgage industry as a chain of coupled, damped oscillations.  Resonance effects appear even in the simplest versions of these models.

So why haven't they?  Well as best I can tell, almost no eonomist has bothered to learn Fourier analysis since the 1930s.  And what you don't think about, you don't remember.

Note: I'm not saying that this will be easy.  I'm just saying that there are quite a few physicists out there who would find this kind of modeling fairly simple.  And even the simple models give all sorts of useful insights on institutional design.

Both an Agency Cost and an Externality

From VentureBeat, a quote from a Cisco VP on how and why virtualization doesn't always result in power savings for companies running large data centers:

"The reason a lot of virtualization practices never show up as savings is because the facilities department, which is not part of IT, will do what they think is right: Install processes on their cooling system that add an artificial load in direct proportion to the load you saved in the virtualization process. The device they install is called a hot-gas bypass. They take the output of the cooling system and pipe it into the input. So it keeps the chillers operational."

You couldn't make this stuff up.  It goes to show that you need good engineering in both your institutional design and your capital equipment if you're going to make cost improvements.  Focussing only on one or the other won't always help.

Bleg for Information about Liquidity

Lately, I've been particularly interested in two different measurements of liquidity:

(1) The bid-ask spread in volume (i.e., the absolute value of the difference in volume between the current bid and the current ask).

(2) The marginal increase in time it takes to complete a transactions given a marginal increase in volume

Could any readers tell me where I might be able to find statistics on these measures of liquidity?

Also, is there a terminological convention for describing these quantities that I haven't discovered yet?

Why does demand for information accelerate?

Today on the Union Square Ventures blog Brad Burnham asks why information doesn't seem to exhibit the same declining marginal utility of other forms of consumption.

Broken Symmetry readers would have an explanation, which is based on an understanding of the underlying dynamics of demand.

The aggregate demand curve taught in Econ 101 is actually a cumulative distribution function.  It can be produced by integrating up the frequency distribution that represents the patterns of consumption shown by a group of people within a particular window of time.

From the mathematics, we can predict that a frequency distribution with a higher mean frequency will have a "fatter head" once aggregated into a demand curve.

But will information consumption patterns exhibit an increasing mean frequency over time (averaged within the same size window)?

Yes.  This is what happens when we learn.  As we consume certain types of information, we become more sophisticated consumers, and tend to consume more and more often.

All of this deserves to be studied empirically.  For we might also predict, for example, that the frequency distribution would get narrower -- the consequence of which would be that the aggregate demand for that information would become more inelastic.

It's not hard to see that knowing how and when the demand for information is going to increase and become more inelastic would be pretty useful to someone interested in making money by providing information.  Maybe the media and entertainment industries should hire somebody to do some reasearch on the topic.  Actually, the research on Payola in the copyright area is pretty much empirically on point, although I've never seen anybody else present the theory of underlying frequency distributions.

Schumpeterian Competition

Two weeks ago I read about University of Chicago professor Lee Fennell's work on rebundling real property, and noticed that her theory sort of implies a cycle in the housing market.  This is obviously true post-subprime bubble.  But what has slowly dawned on me since then is that economists do not yet have any simple models to describe bubbles.  Reading in the Wall Street Journal about Bernanke's team, I went to their Princeton home pages and discovered that at least some of the models used in their papers require a theory of psychology.  Psychology is of course important for investors and regulators to understand; but I believe that a theory of psychology is not required for modeling market cycles.

Fluctuations in the market price for a particular good or service can be modeled in time as a relatively simple function of scarcity in an older good and its newer substitute.  (Think of how IBM PCs were replaced by Macs in the 1980s.)  The simplest mathematical model to associate with Schumpeterian creative destruction of price as a function of time and demand for the older good is the simple harmonic oscillator.

The most useful observation about market cycles that can be drawn from Schumpeter's work is that creative destruction will reverse increases in price over time -- even as supply remains steady or decreases.  When a new substitute for an existing good becomes available, the price of the old substitute will decay over time as consumers switch from old to new.  The result is growth, then decay in price, even as supply remains stable. 

To add some phenomenological richness to the simple oscillator model, one can model in variables for transactions costs, liquidity, and external money supply.  These additional variables change the simple harmonic oscillation predicted by Schumpeter into a damped, driven harmonic oscillation.  RLC circuits and tuning forks are other systems that can be modeled as damped, driven harmonic oscillators.  In my analogy to the RLC circuit, price is a voltage signal that varies in time, which is in turn a function of the aggregate demand for the older good.

A time varying price signal will behave considerably differently after taking the effects of liquidity, external money supply, and transactions costs into account.  For example, when demand (and hence price) swing rapidly, a resonance response (i.e., bubble) may occur when transactions costs are not too large and liquidity and external money are not too small.  These were the conditions that obtained in the subprime mortgage market recently.  A corollary is that the period of the cycle could have been estimated by looking only at the magnitude of liquidity and external money supply (this is the "resonant frequency" for the bubble).  Another corollary is that the "quality factor," which is equivalent to the liquidity divided by the transactions costs, tells us how big a bubble effect we might see.  And indeed the subprime market could fairly be characterized as having very strong liquidity and very low transactions costs as it approached its peak.  (We sort of blew the circuit at that point.)

Yet another useful observation that can be drawn from the model is that bundling supply without including a proportional bundling of demand or other negative price effect will lead to an unsustainable exponential growth in price.  Actually, there are many such lessons that could be drawn from Control Theory, which physicists now routinely use to design and monitor dynamic equilibriums in other systems.

Finally, I note that no physicist would attempt to design or monitor such a system without first getting comfortable with the accuracy of her tools of measurement.  In the case of markets, our tools are not working too well for the moment because the accounting rules do not make any sense in view of the actual relationship between price and value, which is a function (ultimately) of the value that everybody in a market places on a particular good, not simply the value placed on a good by the two parties to a transaction at a moment in time (which is what most people think of when we talk about "price").  Increasing transparency into what company managers see is useful and important in preventing fraud on shareholders.  But adopting rules that are consistent with reality in terms of how economic growth and decay occurs is equally important.

It's also worth mentioning that in some sense, all of these price cycles are coupled, either weakly or strongly.  Hence the entire economy could be modeled as a chain of coupled damped, driven oscillators.  This is actually a theoretical model from which quantum field theory and non-equilibrium statistical mechanics depart.  Among other useful results, these theories have permitted physicists and chemists insights into how and when to expect phase transitions in what would otherwise be considered unstable thermodynamic states.  This in turn suggests another insight into the problems economists and accountants have been having dealing with bubbles.  Like the static models of supply and demand, thermodynamics is successful in making forward-looking predictions about how a physical system will behave in response to changes in temperature, volume, and pressure.  But it still took a mathematical model of molecular dynamics to understand exactly how and when transitions would occur: thermodynamics, like static economics, provides little insight into how long unstable equilibriums will persist.

What are the Existing Best Models for Business Cycles?

Readers:

Can anybody tell me what models academic economists right now consider the best for explaining business cycles?

IP is not an Asset: Patents and Inventors Need to Stick Close

Earlier this week, Peter J. Wallison argued that conventions in fair value accounting may in part be the cause for the recent bubble markets.  Specifically, Wallison pointed to the convention (implemented under FASB 157) that requires assets to be carried at "market" values, even when those assets are not being held for trading purposes.

Almost any scientist or engineer would immediately have recognized the truth of this argument.  Our understanding of any system -- chemical, electrical, mechanical, or financial -- will be limited in part by the accuracy of our tools of measurement.  When one considers how FASB 157 required banks to report the values of MBSs, CDOs, and CDO^2s on their balance sheets far above what the banks would themselves have been willing to give away for the same assets, one understands how the financial markets quickly lost track of the intrinsic value backing the securities traded.

This wisdom has direct relevance to the secondary markets for IP.  Most of the firms now in the secondary markets for IP have taken the view -- and are conducting their businesses -- as if IP were an asset.  This is because IP does bear some characteristics of an asset.  Namely, like real and personal property, IP can be protected through exclusive rights.  The analogy to property has thus come to dominate our understanding of the nature of IP.

Although accountants often treat IP as an asset, IP is not a commodity.  IP is more like equity, although it is not like other equity.  IP is a limited exclusive right to human capital (namely, to inventors' time solving a technological problem).

Maybe part of the reason that Abraham Lincoln understood the importance of patent law is because he understood that human capital cannot be owned.  The photograph shows the Emancipation Proclamation, whereby Lincoln did more for the cause of freeing human capital than many other men together have done in the course of human history.  Lincoln loved the patent system because he understood that it too could lead to more freedom.  Scientists and engineers work best free from the immediate demands of business people and customers.  The idea of a patent system carries within itself the promise of more innovation and more freedom.

POSTSCRIPT: Please note that I do not believe that inventors are literally enslaved right now.  There are obviously huge differences between the enslavement of millions of black Americans and the metaphorical enslavement of inventors who are now forced to do work other than inventing because of the broken patent system.  I do, however, believe that making people more free leads always to a multiplicity of unanticipated social benefits.

Slicing and Dicing Insurance and Property Risks

I see that Charlie Munger has come out against slicing and dicing insurance risk.

I'm going to make a simple observation based on a couple of posts from earlier this week.  The problem with slicing and dicing insurance or property is not the slicing and dicing -- which respectively serve only to amplify the positive or negative externalities of ownership.  The problem is that the source of price signals -- i.e., the individuals who decide whether to buy property or insurance -- are not generally forced to buy both.

Each individual within our society should be required to accept some measure of both the positive and negative consequences of her actions.

Charlie, if you're out there, don't you agree?

What the Founding Fathers knew about R&D that we have forgotten

As evidenced by his lecture on discoveries and inventions, Abraham Lincoln had a deep understanding of the patent system.  It is amazing how his lecture, which is now well over 150 years old, can seem so fresh today.  He and Charlie Munger have inspired me to undertake a historical review of other important lessons of the imminent dead.  Today the lesson is from Scottish enlightenment thinker Adam Smith, famous for his authorship of The Wealth of Nations. I must shamefully admit that I have thus far been unable to make it through the entirety of his treatise.  I have nonetheless been the beneficiary of the wisdom of Adam Smith through the help of editors, from whom we have the following excerpt:

To take an example, therefore, from a very trifling manufacture; but one in which the division of labour has been very often taken notice of, the trade of the pin-maker; a workman not educated to this business . . . nor acquainted with the use of the machinery employed in it (to the invention of which the same division of labour has probably given occasion), could scarce, perhaps, with his utmost industry, make one pin in a day, and certainly could not make twenty. But in the way in which this business is now carried on, not only the whole work is a peculiar trade, but it is divided into a number of branches, of which the greater part are likewise peculiar trades. One man draws out the wire, another straights it, a third cuts it, a fourth points it, a fifth grinds it at the top for receiving the head; to make the head requires two or three distinct operations; to put it on, is a peculiar business, to whiten the pins is another; it is even a trade by itself to put them into the paper; and the important business of making a pin is, in this manner, divided into about eighteen distinct operations, which, in some manufactories, are all performed by distinct hands, though in others the same man will sometimes perform two or three of them.

Adam Smith goes on and on from here about the many benefits of the "division of labour."  Although controversial in his day, the benefits of "the division of labour" are in our day a fact so well-accepted by the majority that many people seem unaware of the history of this idea.  We seem to assume it a logical consequence of any business. That it is not.  In each case in which a division of labor is successfully implemented in business, there was first an entrepreneur who saw the benefit of separating one task into two.  Henry Ford brought the magic of divisions of labor to the production of cars through the assembly line.   Most people can't imagine this, but before him others probably scoffed at the idea that something as complex as a car could ever be assembled without a single person overseeing the entire process.

Are we not still scoffers?  In the United States, we now live in an age in which most lawyers, business people, and researchers believe that R&D and early-stage product development are incapable of being done by two teams.  The fact remains, however, that the best inventors and the best startup CEOs are not often the same person.  And the best R&D and the best product development tend to occur in different environments.  We have strained for the past twenty-years in the United States to force inventors into the role of entrepreneurs, and entrepreneurs into the role of inventors.  Being a hardworking nation, we have not been entirely unsuccessful.  But how much more successful might we be were we to accept once and for all that there is an efficient division of labor between R&D and commercialization (yes, even the "commercialization" done by startups)?

The patent system is the most sophisticated and efficient means for implementing a division of labor between R&D and commercialization ever conceived by humans.  It is by cutting back at patent rights in the United States that we have inadvertently forced inventors to become entrepreneurs and entrepreneurs to become inventors.  Let us not further disintegrate the division of labor between R&D and commercialization by weakening our patent laws in 2008.  Let us recognize that good inventors and good startup CEOs are not always (or often!) the same person.  Let us "promote the progress of science and the useful arts" in the ways our Founding Fathers intended, by a division of labour between R&D and industry.

Modeling the Subprime Mortgage Market as a Closed Loop Oscillator with only Positive Feedback

Electrical engineers who design analog circuits learn an important rule of thumb: closed loop systems that provide only positive feedback to their inputs can become unstable.  In the diagram at right, when A and B are relatively good amplifiers, the multiplication of their amplifying effect at the input Vin can cause the system to blow up.

The closed loop oscillator with only positive feedback is actually also a useful way to understand what caused the subprime mortgage market bubble, and what could have been done to prevent it.

According to the newspaper accounts, something like this was happening: home buyers were bidding for price (Vin), which was being approved by assessors. Banks were then offering loans based on the assessments to the buyers, and selling the loans to loan-buyers.  Loan-buyers were then bundling the loans into CDOs (amplifier A), and then selling them to bigger loan-buyers, who were then bundling CDOs into CDO^2s (amplifier B).  The bundling permitted for larger-scale buyers of loans to get into the market, which has a multiplying effect on demand for loans, which in turn gave the banks and assessors an incentive to offer loans at higher assessed values regardless of the underlying value of the property.  There was no mechanism for negative feedback on price to enter the closed loop that buyers, assessors, banks, and institutional investors formed.  And when A*B approaches 1, as it will eventually in a system with powerful amplifiers and no negative feedback, the system blows up.

Electrical engineers avoid these kinds of instabilities (usually!) by building negative feedback into the loop -- i.e., by introducing some circuit element that inverts or phase shifts the signal fed back into the input of the circuit so that the signal doesn't just keep growing and growing until it blows up the circuit.

Lee Fennell at the University of Chicago law school has a great suggestion for how we could have built negative feedback into the system: bifurcate on-site and off-site risks associated with homeownership, and make buyers pay separately for these two types of risk.  So for example, a home buyer might be required to buy an insurance contract at the same time they purchase their home, the premium for which is large enough to permit the insurer to pay the home buyer off for a loss in home value should the value of her home later decrease precipitously because of off-site problems in her community (such as factory closings, pollution, or crime).

Now imagine how the same cycle would have worked: the home buyer makes a bid on both the home and the insurance for off-site risk -- the law has to require her to buy both, otherwise this doesn't work.  The home assessor approves an inflated price.  That's still the assessor's incentive because of the demand for mortgages.  But the insurer then raises insurance prices in response to the assessment to offset the larger risk associated with the inflated price.  If the home buyer can pay the increased premium, it's a deal.  If not, the deal falls through, and prices tend to fall back to a level that more accurately reflects the underlying value of the property.

UPDATE: Upon reflection, I realize that the government actually does not have to require homebuyers to buy both the home and the insurance.  In fact, only the insurance-provider and mortgage-provider would have to agree.  This raises other complications (what happens when they're owned by the same entity, e.g.?), but it's useful to see at least that new legislation might not be required.

Stranded R&D

In 1980, Congress passed the Bayh-Dole Act.  Overnight with its passage, universities and government-funded R&D labs gained a comparative advantage in funding R&D.  Universities and government labs have a cost advantage in that many had already spent tens of billions of dollars setting up research labs for non-commercial purposes, including teaching and curious exploration.  Many scientists and engineers found the prestige of academia, and the increase in professional freedom it promises, a compelling offer.  The result has been a gradual shutting down of corporate R&D labs, and an expansion of industry collaboration with scientists and engineers now employed by universities and government labs.

Many people think of the Bayh-Dole Act as an unmitigated success story.  Several multi-billion dollar technology companies that are now household names (such as Genentech and Google) started in graduate school research labs.  Many inventors are happier in the more collaborative environment that academia offers.  Collaboration is an under-appreciated driver of innovation.

Unfortunately, so far universities have underperformed private benchmarks for the successful transfer of technology.  Despite spending almost an order of magnitude more on R&D (about $50 billion), the AUTM reports only about a factor of two more revenue from R&D (about $2 billion) than does IBM (about $1 billion on about $5 billion in R&D) over an overlapping period from the mid-nineties to the mid-zeroes of the present decade.  Although it is tempting to attribute the difference in returns entirely to the diversion of R&D funding into pure science (an attribution that ought to silence the Bayh-Dole critics who favor pure science), it is important to remember that there was a net inflow of the most productive researchers from industry into academia over the same period of time.  This concentration of the brightest minds of science and engineering within academia would probably have led to a faster increase in returns from R&D if there weren't something else going on.

And there is something else going on.  The costs of licensing and litigation of patents has skyrocketed over the same period of time.  The biggest reason for increasing costs has been the inelastic supply of patent lawyers relative to the exploding demand for their services.  Unlike patent prosecution, which can be done by non-lawyer patent agents and examiners, patent licensing and litigation are services that require a state bar license (and the three years of ABA-accredited law school that this usually requires).  Law firms are struggling to meet demand by increasing starting associate salaries (patent boutiques started the chain reaction in both instances over the past ten years), but the corresponding increase in associate to partner ratios at most law firms (necessary to keep profits-per-partner high and retain top partners) has led to a decline in the quality of services overall.

It is worth noting that many technology transfer offices are staffed by non-lawyer scientists and engineers with formal or informal business training.  Many of these employees are probably undervalued by the legal services market because of their lack of state bar credentials.  Seeing the value, university tech-transfer offices and other government and private firms not constrained by state bar requirements are scooping these types of employees up.  Non-lawyers will probably play a growing role in R&D funding and technology-transfer going forward, even in providing "legal" services.  The investment banks (such as Altitude Capital) and venture capital funds (such as Intellectual Ventures) that have recently entered the secondary market for patents are early signs of this trend.

The result of these macroeconomic trends in R&D is a market in which many startups and smaller companies are realizing only a fraction of the intrinsic value of their R&D.  Technology is stranded in later-stage startups and other small private companies that are not eligible for further venture capital financing, acquisition, or IPO.  Problems in the credit markets and the passage of the Sarbanes-Oxley Act have further exacerbated the problem for these companies in the acquisition and IPO arena.  In effect, the United States is piling up a vast, invisible junkyard of stranded R&D that could be socially valuable if placed into the hands of the right owners.

As the returns to investment in R&D decline, so too do the number of jobs available for researchers outside academia.  This is a problem that is vital to the health of the U.S. economy within its global environment.  If current trends continue, there will be more Ph.D. engineers living in China than in the U.S. by 2010.  The number of U.S. patents issuing to foreign entities is already nearly equal to the number of patents issuing to the U.S.  If the U.S. were to strengthen its patent system, we would be far better positioned than any other nation in the world to bring the power of market-based incentives to bear on the problem of attracting the most talented human capital -- the single most important problem we face in our long-term prospects for economic growth.

People are starting to recognize these problems.  Recently, the Brookings Institute has called for the government to setup a National Innovation Foundation.  But aren't the market-based incentives of a strong patent system a better way for the government to encourage R&D funding?  Although a handful of firms, including Intellectual Ventures, Ocean Tomo, and other new entrants are struggling to meet immediate needs, the inventors and startups most in need cannot afford to hire anyone to answer the lobbyists hired by the large corporations that are net payers of patent licenses (when forced to pay at the end of protracted litigation).

Although the big picture of innovation is so large and complex that it is difficult for most people to understand, the solutions are actually simpler and easier than most would imagine.  First, the patent laws should be reformed in ways that would promote private settlements of disputes over patent infringement rather than litigation.  Some recent changes to the patent law have been beneficial in this regard, and some detrimental.  Unfortunately, the Supreme Court's recent holding in Medimmune makes it harder than ever for inventors to get to the table with large corporations without ending up in litigation.  And after Mercexchange, startups and independent inventors do not have the threat of an injunction to keep licensees at the bargaining table when those startups and inventors have failed to find funding to themselves commercialize the technology.  Neither of these by themselves is fatal.  The threat of injunction was no doubt abused by opportunistic speculators from time to time over the past few decades.  But not in decades has it been more difficult for investors in R&D to see a return through patent licensing.

Second, if the government is going to provide funding to solve these problems, that funding might best be used to lower the barriers to entry for the practice of patent law.  For scientists and engineers, especially those who understand business, the opportunity costs of wages are probably much higher than the costs of a law school education.  Public funding of scholarships for scientists and engineers who intend to study law would over time decrease the transactions costs associated with patent licensing, and gradually decrease the amount spent on litigation as it becomes easier and easier for opposite parties to reach agreement on differing valuations of a technology.

Third, institutional investors should consider allocating a larger share of their funding to hiring more employees for their technology transfer offices now, and later for investing in private equity funds that specialize in R&D investment.  The technology transfer offices are now overwhelmed by the demands on their time in many cases.  As a result, they tend to focus on the biotech and pharmaceutical inventions that are likely to provide the largest payouts, ignoring the many other areas of R&D that could nevertheless have a transformative impact on our society.  In terms of private equity investments, over the short-term the lack of licensing revenue is going to impede the returns for these funds.  But restarting the R&D engine of economic growth is going to require the public and private sectors to work together.

These are complex problems that it will take teamwork to solve.

Update: Silicon Valley never fails to disappoint in its farsightedness.  Jaisen Mathai, Michael Arrington, and Stu Phillips are all groping around the edges of the problem.

Update 2: I was recently asked whether the figures for IBM and AUTM include capital gains from equity.  The answer is no, neither do.  I have seen no evidence and have no reason to believe, however, that the AUTM should be seeing larger returns from equity on its R&D than IBM.  So the larger point about relative efficiency in technology transfer seems still to be sound.

Every Patent Affects Two Different Markets

There are two different markets relevant to every valuable patent. First, there is the market for the R&D work that results in the patent.  Prices in this market are set by the opportunity costs for the time of scientists and engineers who are capable of theorizing about and experimenting with the technology.  Second, there is the market for the claimed products or services that the R&D work opened up.  The second market is the one that everyone naturally thinks about.  In fact, our whole nation has had a blind spot for the first market for a long time because corporate R&D divisions were serving that market very well until the Bayh-Dole Act was passed in 1980.  Most universities have not been able to consistently match pre-product funding with the flow of R&D produced by their faculty.  One former R&D employee from Apple and Microsoft blames Silicon Valley, saying "Silicon Valley forgot how to do R&D."

Within the law of antitrust, one can observe how this blind spot is affecting the outcome of litigation.  The essential facilities doctrine of antitrust law has been applied in an inconsistent way because most judges don't see that there are two different markets.  R&D work almost always has natural monopoly characteristics.  But patents covering valuable R&D should not therefore trigger the essential facilities doctrine.  Lots of R&D work needs the barrier to entry that only legal exclusivity can provide in order to recover even a reasonable profit.  Yet if the same patented R&D is valuable because the claims cover a product, the production and distribution of which have natural monopoly characteristics, then applying the essential facilities doctrine to require the patent owner to make licenses available on a non-exclusive basis might make sense.

Consider two concrete examples.  Example 1:  I patent R&D claiming a new type of telephone.  The essential facilities doctrine should not be invoked to require me to license others to sell the patented telephone.  Producing telephones is not a natural monopoly because the total average costs are not always declining.  Note that this is true even though the telephone demonstrates network effects.  Example 2:  I patent R&D claiming a new type of telephone network.  The essential facilities doctrine probably should be invoked to require me to license others on a non-exclusive basis who need to use the telephone network for their own products or services to be valuable.  Building telephone networks is a natural monopoly because the total average costs are always declining.  In both cases, the crucial question is whether the second market has natural monopoly characteristics.  The first market almost always will.

In some cases in the past, when the essential facilities doctrine has been invoked against patent owners, judges have decided either to let the patent owner maintain full exclusivity or to deny any compensation.  Better would be for judges to let most patent owners do whatever they want, but require compulsory licenses in the subset of cases in which the actual product or service sold has natural monopoly characteristics.

Those with an extraordinary interest in reading more can wade through this paper.  But know that I am no longer enthralled with the idea of applying the reverse doctrine of equivalents in patent law or the Feist originality doctrine from copyright law to the same effect.  Too much potential for mischief.

The more things change...

Thousands of years ago Plato (top left) wrote that "Until philosophers are kings, or the kings and princes of this world have the spirit and power of philosophy, and political greatness and wisdom meet in one, and those commoner natures who pursue either to the exclusion of the other are compelled to stand aside, cities will never have rest from their evils, -- nor the human race, as I believe, -- and then only will this our
State have a possibility of life and behold the light of day."

Plato's student Aristotle (bottom left) understood government differently: "When several villages are united in a single complete community, large enough to be nearly or quite self-sufficing, the state comes into existence, originating in the bare needs of life, and continuing in existence for the sake of a good life. And therefore, if the earlier forms of society are natural, so is the state, for it is the end of them, and the nature of a thing is its end. ... Hence it is evident that the state is a creation of nature, and that man is by nature a political animal. And he who by nature and not by mere accident is without a state, is either a bad man or above humanity; he is like the 'Tribeless, lawless, hearthless one, ' whom Homer denounces- the natural outcast is forthwith a lover of war; he may be compared to an isolated piece at draughts."

This week, Judge Posner (top right) made the following remarks on problems in the current credit-market: "This would be fine if zero regulation were the social desideratum, but it is not. The correct approach is to carve down regulation to the optimal level but then finance and staff and enforce the remaining regulatory duties competently and in good faith. Judging by the number of scandals in recent years involving the regulation of health, safety, and the environment, this is not being done. And to these examples should probably be added the weak regulation of questionable mortgage practices and of rating agencies' conflicts of interest and, more basically, a failure to appreciate the gravity of the moral hazard problem in the financial industry."

As Aristotle saw things differently from Plato, so Becker (bottom right) sees the purpose for regulation differently from Posner: "It would run counter to human nature for regulators to take a skeptical attitude toward the riskiness of various assets when the market is indicating that these assets are not so risky, and when originating and holding these assets has been quite profitable. One can expect regulators to mainly follow rather than lead the market in assessing riskiness and other asset characteristics."

Rebundling Ownership and Control of Patents

On Friday I attended a panel on the "State of the Legal Profession" held at Stanford Law School.  The panel was well-designed, with a variety of perspectives from lawyers in-house, at law firms, in academia, or in public interest.  It's hard to get six lawyers to agree on anything, but it seems that everyone more or less agreed that the traditional law firm business model is not holding up well in the 21st century.

There are many reasons for this, one of which -- the inaccuracy of the billable hour as a metric for the value of legal services -- I've discussed in a previous entry.  Another reason that was obvious after hearing the panelists is the relatively inelastic supply of talent graduating from law schools every year.  That pool of talent hasn't grown much in decades, even as the demand for legal services has exploded.  Public interest advocates are seeing more and more people opt for pro se representation in court or no representation in transactions (see, e.g., subprime mortgage lending).  Law firms are raising salaries higher and higher to attract talented associates, but also increasing leverage (i.e., the ratio of associates to partners) in order to continue attracting talented partners (profits-per-partner being a key metric for partners in deciding where to work).  Meanwhile in-house counsel is facing increasing pressures from company management to keep down the costs of legal bills as global competition narrows the margins on products and services.  Another time I may elaborate on how some big law firms are beginning to resemble a Ponzi scheme in this regard.  And anyone keeping tabs on the credit market knows how ugly deleveraging can get.

So what can be done?  There's no choice for now but to get more efficient with the supply of talent that we've got.  My earlier post gave a suggestion for how more efficient incentives could reduce costs for certain kinds of transactional work.  But it didn't elaborate on how rebundling of ownership and control is capable of solving a whole class of principal-agent problems in legal services.  A similar solution could be used by many companies to reduce the costs of procuring and enforcing patents.

The basic contractual framework for accomplishing this is simple, although the practical execution is difficult, requiring human capital with a high-level of expertise in three different, (now) weakly-overlapping professional disciplines.  Instead of paying a law firm by the hour (or paying a high fixed-rate) to prosecute patents, clients could pay a much lower (or zero, or negative) fixed-rate for their work, and then give them a slice of any future royalties earned on the portfolio.  Lawyers who (a) believe in the value of their services, (b) understand the technology patented, and (c) believe in the prospective value of the market that the patents are meant to cover should be willing to accept lower rates in exchange for a slice of future profits.  The trouble is that there just aren't that many lawyers who understand law, technology, and venture capital investing.  And the ones that do (think of senior partners at big law firms) generally are too comfortable with the status quo.

Not every company is going to feel comfortable with doing things differently, especially if the current system is meeting their needs.  Nonetheless the potential is there for forward-looking clients and entrepreneurial patent lawyers to innovate on the traditional business model, and maybe even lead the way into a better model for legal services in every market.

UPDATE: Thanks to IPKat I have learned that the French bar is in the middle of a protracted struggle to keep scientists and engineers out.  Not surprising, but comforting to know that we're not the only ones with this problem given the implications that our excessive domestic regulations have for the United States in competing in a global economy.

When failure = success? Patents stimulate new ventures.

"If I had a world of my own, everything would be nonsense. Nothing would be what it is because everything would be what it isn't. And contrary-wise; what it is it wouldn't be, and what it wouldn't be, it would. You see?"

- Alice in Wonderland

The PatentlyO posts from Bessen and Meurer are taking a trip to crazy town.  Today, their post on "patent failure" focuses on the impact that the patent system has had on publicly traded companies.  The authors conclude the post with the following remark:

"Third, we find that small publicly traded firms get small positive R&D incentives from patents. This is also very likely to be true for small, non-publicly traded firms and non-profit inventors."

Yet the point of their post is that the patent system has failed to promote R&D.  Since when are large publicly-traded companies the source of innovation in our economy?  I fail to see reasoning to support their conclusion that the patent system fails because (net) it imposes a "tax" on large publicly traded companies.  If I understand them correctly, it seems that their facts and the inferences they themselves draw from those facts and inferences suggest that the patent system is a success.

If a patent "tax" that requires manufacturers to pay inventors for R&D work ends up creating more social value than the tax plus the transactions and administrative costs of maintaining it, then it's what most economists would call a success.  Bessen and Meurer observe that publicly-traded companies are net payers of this tax, while smaller, private companies are net receivers.  Without knowing the magnitude of the payments and receipts on both ends, we can't definitively say that the system is creating social benefits.  But it seems that even Bessen and Meurer are conceding that the system works in the pharmaceutical arena.  What evidence do we have that the same might not be true in the other industries they have studied -- and at a growing rate as transactions and administrative costs decline?

As a society, we should not be concerned that big companies are paying small companies for new technology so long as they're paying a fair price.   And without patents, how would smaller companies convince larger companies to pay for valuable technology that cannot be protected with trade secrets?  Would it be better for the economy if more publicly traded companies simply bought-out smaller R&D-focused companies and then fired the employees who don't have valuable know-how?  Because of the benefits of divisions of labor (and very possibly antitrust issues), I think not.   

Specifically, R&D often requires a very different environment from the one that most publicly-traded companies can provide, both socially and financially.  First, almost by definition most publicly-traded companies have a vested interest in an older technology that provides their revenue stream.  Think of Micrsoft and its OS and productivity software.  They were and are at a disadvantage to Google (which has no legacy products or services) in moving into the online services business.  Second, private companies have the ability to attract young, top-quality engineers and scientists through options compensation, which publicly-traded companies have trouble competing with.  Google has been having this problem in Silicon Valley over the past few years as top-quality engineers are passing on Google for new ventures like VMware.

Nobody puts patents in a corner!

"Many of life's failures are people who did not realize how close they were to success when they gave up."

-Thomas Edison

As regular readers know, the author is well aware of some imperfections in the current patent system in the United States.  In that regard, he was very pleased to read about what looks like some good work done by some economists on the patent system.  In particular, these economists have studied and offered many insights into the ways in which the patent system is not functioning perfectly as a system of property rights.

What may be missing from a backward-looking analysis, however, are some recent changes in the economic landscape.  The biggest of these changes is the Internet.  As I have pointed out in an earlier post, the emergence of the Internet and digital media technology is having a dramatic effect on the overall costs of having a patent system.  As the transactions costs drop, markets emerge -- even markets in imperfectly defined property rights.  Sometimes the past is not prologue.

[Note to Bessen and Meurer: I would love to review your new book if you'd like to send me a copy.]

Can the law firm business model be fixed?

The billable hour is a terrible metric for either the value of client services or of associate contributions thereto -- its two primary uses.  The trouble is that there seem to be few alternatives.  The reasons are obvious enough in each case.  Clients are good at knowing whether they're in legal trouble or not, but bad at knowing whether their outside counsel is working efficiently to keep them out of legal trouble.  Partners are better at knowing whether associates are working efficiently -- they know whether they're getting the work they need and also what they're billing the client for that work.  Unfortunately, partners have a perverse incentive to bill more rather than less for associate (and partner) time -- their own bottom line is tied to the billable hour.

To most stakeholders (client, partner, associate, employees), this looks like an train wreck.  In other parts of the economy, principal-agent problems of this scale could not persist.  But the lack of client transparency into the quality of legal services seems to be keeping the business model on life support.  Can the business model be fixed?

The practice of law is a diverse and varied pursuit.  There cannot be a solution that will work for every field.  To wit, there are some areas of law (such as plaintiff class action work) where alternative fee structures are the norm.  Thus, improvements to the business model will probably be area-of-law specific.

I'll focus on corporate transactions.  Here is my humble suggestion for how companies might get more efficient work from outside corporate counsel for much transactional work: bundle insurance and legal services.

Part (not all) of why corporate management pay for D&O/E&O insurance and transactional lawyers is for a similar economic purpose: to hedge against the downside risk of future lawsuits.  There are, of course, a host of other reasons why managers pay for good outside counsel, such as information about the market.  Outside counsel that specializes in a particular type of transaction may be far more knowledgeable about what is going on in a particularly industry than managers who may be entering for the first time.

Nonetheless, the downside risk of a lawsuit is something that good transactional lawyers will seek to avoid by drafting and refining transactional documents with an eye toward future plaintiffs.  And many corporations take a "belt-and-suspenders" approach by paying for both insurance and extra hours on the proxy statement.

Why not bundle D&O/E&O insurance with transactional legal fees in the following way: inform outside counsel that their pay will be based on an hourly fee that reflects the fair value of the information and services aside from the hedge against lawsuits, then pay a bonus at the end of the year based on the difference between what the insurance premiums are and what they would have been had the insurance policies paid out.

The benefit here is that insurance companies are good at quantifying the value of downside risk avoided.  By bundling the costs in the books, company management could make a more accurate assessment of that part of the value of their transactional services.

Externalities and Semi-closed Platforms

Jason Calacanis warns companies away from the Facebook platform because of the uncertainties associated with how Facebook will share revenue with third-party application developers.

I don't disagree with many of the points he makes.  There is considerable uncertainty at this point as to how exactly revenue will be shared, what metrics will be used, and so on.  And there is no doubt that the companies that can successfully generate their own user base will monetize those users at higher margins than the companies that must rely on Facebook for customer acquisition and distribution.  That's the deal.

Rather I want to point out that Jason seems to ignore an important class of potential economic benefits for users that can be achieved by a semi-closed platform.  These benefts may enlarge the economic pie enough so that the deal is still better than what companies could achieve by monetizing at higher margins (with a smaller user base).

As I have eluded to in an earlier post, Facebook as a semi-closed platform promises to solve some of the collective action problems associated with user-generated content on the Internet.  Facebook is mostly spam free because it is a semi-closed platform.  An analogy is that Facebook:books what the open Internet:magazines.  We can expect to see a better valuable content/page ratio (and hence more users and more time-spent) on Facebook because it is a semi-closed platform.

VCs and NDAs

Today I'm attending the 14th Annual Private Equity CFO Conference in Redwood Shores, California. The speaker right now, a lawyer from Wilson Sonsini, is talking about whether VC firms should sign NDAs.

Just over six months ago I was defending a VC who had been named in a counterclaim brought by a defendant that my client was suing for patent infringement. I was amazed back then that a startup company that might be in need of VC funding would take such a risk -- who's going to meet with a company that might just sue for trade secret theft later? But here I am listening to a speaker explain that more and more VCs are breaking the industry rule against signing NDAs. What to make of this?

The legal issues are taking a back seat to the economic realities of the market. There is too much VC money chasing too few good new companies. Lower tier VC funds are trading more favorable legal terms (such as willingness to sign NDAs) in order to get access to the best deals.

But this just shows how some VCs aren't thinking clearly enough. Signing an NDA makes little or no economic sense for either the VC or the startup. Startups benefit from both the regular capital (i.e., $) and human capital that VCs offer. That human capital includes the network of potential recruits, other portfolio companies, and other VCs (with whom they may syndicate later financings). A VC who signs an NDA could be hamstrung from adding this value.

On the other hand, the VC firms who have this kind of human capital will also be the firms that are least likely to sign such NDAs. So maybe not much harm is done.

The takeaway for startups: ask if the VC will sign an NDA. If the answer is YES, then stay away. They're not going to contribute much besides money to helping your company succeed.