Wednesday, 10 September 2014

Things Are Seldom What They Seem

I am interested in a lot of things.  I am amused by still others.  I am paid for work that I do at a small intersection of the two.

Quickly, I work in a large, global pharmaceutical company, in a department called "Health Economics and Outcomes Research (HEOR)."  It's a long title, but can (relatively) simply be put as a function whose job it is to try to quantify economically and socially the benefits (and risks) of health care interventions.

My work often touches on traditional randomised controlled trials (RCTs), which are the bread-and-butter of pharmaceutical R&D.  These are the classic studies where subjects are enrolled into trials with a proportion randomly assigned to treatment/intervention A, and a proportion randomly assigned to treatment/intervention B.  The subjects are then followed, and the clinical outcomes assessed.

The endpoints of these trials might be survival - for example, lung cancer patients treated with two competing oncological agents.  They might be heart attacks - for example, patients with CHF being treated with beta-blockers or ARBs.  

But traditional RCTs are really just part of the process by which new treatments are assessed.  Once their relative safety and efficacy are estimated, the question moves on to the practicalities of effectiveness versus efficacy (i.e. do the treatments perform in actual life-use versus strictly enrolled and controlled trials, where patients may or may not adhere, etc.), or whether the added benefit is the best allocation of resources.  

The world we live in is one of constrained resources and unlimited need, so it is natural, even perhaps laudatory, for those who must pay to consider the optimal allocation of the public purse. Is it the best use of one million euros to treat 100 cancer patients or 10,000 diabetics?

These are the sorts of questions I work with each day.  

My academic training is in mathematics and mathematical modelling; though the term "Economics" is embedded explicitly in my title, I am most assuredly not an economist, and my formal economic training is actually a single course in college.  But I am very skilled at constructing models and at evaluating the ways in which they are deployed.  And over more than two decades, I've achieved a certain level of comfort dealing with economics.

Which brings me, rather circuitously, to the current argument raging right now in France, where I live.  The data for the second trimestre (the second quarter) for the French economy have been published, and they are awful.  A record number of French are looking for jobs.  The growth in PIB (produit intérieur brut), or GDP is 0.0% and projected at best to be 0.2% or so for the remainder of 2014.  The president, François Hollande, is polling currently at 13% - a performance so poor that it is historic, and his regime are flailing about, reshuffling ministers trying to 'fix' the situation.

The US economy, while not running terrifically, is looking practically robust, if not bullish, by comparison.  Our neighbours in the UK and Germany are also doing substantially better.

So the question is asked, naturally, "Why is France performing so poorly compared to other OECD nations?"

The answers offered are manifold, but the motivations are a bit wide of the mark.

The local morning news, Direct Matin had an essay this morning talking about the possibility that there is Un Climat de Japon-isation in France, alluding of course to the more-than-a-decades-long doldrums in Japan.  (Apologies; the article is in French)  A period of low or no-growth, competitive disadvantage, and perhaps even financial deflation.  Prices indeed in France, while still high, have remained flat and in some cases in fact are falling.

If one digs into the article, some odd things emerge.

Recall, in the late 1980s and early 1990s, there was in the States a real fear that Japan would over-take the US economically.  These fears were driven by the erosion of the US auto industry, the rise of the tech companies (e.g., Sony), and the swagger of the Japanese, including some pretty ostentatious acquisitions - Pebble Beach Golf Course, for example.  The US responded with an array of activities, including a weak dollar policy, and the threat abated.

La réponse américaine s’était ordonnée autour d’une vraie guerre commerciale, dont l’arme principale avait été une politique du dollar faible, face à un yen dont on annonçait qu’il pourrait un jour menacer la suprématie du dollar. Le résultat a été l’entrée du Japon dans cette longue période de stagnation dont il ne parvient pas à sortir, malgré une politique plus nationaliste mais qui ne prend pas le chemin du succès. 
Certes, des facteurs strictement japonais ont joué : faiblesse du système financier, recul de la capacité d’innovation… Mais les Etats-Unis ont utilisé toutes leurs armes commerciales et monétaires pour se redonner un avantage vis-à-vis du Japon.

[The American response was a true commercial war, but the primary weapon was a weak dollar policy, which was threatened with replacement versus the yen as the primary world currency.  The result was that Japan entered into a long period of stagnation that it could not escape; despite a strong nationalist policy, Japan could not find its way.
Certainly, other factors played a role: a weak financial sector, loss of innovation...but the US used all weapons, both monetary and commercial to reclaim its advantage over Japan]

The author goes on to compare the situations of Japan circa 1995 and France currently.  I find that he misses the mark on a number of counts.  First, whilst the US did in fact deploy monetary policies - and these have side effects that are not 100 per cent positive (inflating the dollar, inter alias, makes all imports expensive, including commodities and raw materials like oil) - the major structural problem that the Japanese faced was not that they could not export goods to the US, but rather, that their own growth was largely fuelled by the sort of speculation in real estate that made internal consumption slow down.  Much of the "wealth" created in Japan was illusory; apartments bid up to ridiculous levels.  This phenomenon was to an extent played out in the US during the first decade of the 21st century.  

The Japanese acted by, among other things, encouraging borrowing with low, zero, or in some cases, negative interest rates.  None of it worked, as Japanese consumers simply did not respond in sufficient numbers.  

A modern economy is largely driven by consumer spending, China to the side.  

In France, the leadership are looking at various means to 'fix' the economy, including tax abatements/inducements to companies to encourage hiring.  Manuel Valls, the current primer minister, is pushing President Hollande's showcase piece of legislation, called le pacte de responsabilité. which is a set of inducements to large employers accompanied by various promises of hiring, etc.  Both the patrons of the enterprises and the core supporters of Hollande's socialst government, including the unions, have balked.

Various other fixes are proposed, including weakening the Euro, providing subsidies to 'incubators' in Paris to re-create the start of culture of the US's Silicon Valley, to adopting more pro-business policies similar to Germany's or the UK's.

A signficant problem, both with the Direct Matin analysis and the machinations of Hollande's PS colleagues is an ultimately mis-guided faith in economics and economic policies as a science.  Economics is often called the "dismal science," but I wonder if economics is indeed, a science at all.  Modern economics and modern economists are very good at creating models to describe how the economy "works."  There are now "star" economists such as Paul Krugman, Josh Barro, Joseph Stiglitz, and the current flavour of the month, Thomas Picketty.  There are also their historical forebears JM Keynes, Friedrich von Hayek, and Milton Friedman.

All of these men are, I am assured, very smart.  Their models precisely described, in often highly mathematical terms, the functioning of macroeconomics.

One "Model" for Explaining Macroeconomics

The statistician George Box once famously commented that "all models are wrong, but some models are useful."  The debates continue of whether Keynes or Friedman were "right" in their thinking, and largely because economics itself does not readily lend itself to controlled experiments in the way that physics or even medicine do.  

In an anecdote, when I was in the PhD programme at Stanford some decades ago, a classmate was researching the use of stochastic models to describe the movement of the stock market.  His models could be finely-tuned to describe what had happened.  But they were hopeless at predicting with any sort of accuracy what would happen, or even could be reasonably be validated using standard, hold-out analyses and the like.

The markets move just too erratically.  Too irrationally.

One of my passions is mathematics; I have a particular interest in abstract algebra and topology.  One of the reasons is that maths tends to be very clean; elegant even, in its beauty.  But mathematics are a sort of artificial construct, a system set up to describe the world.  I studied some model theory, but I am not terribly well-versed in the philosophy of mathematics (I've read portions of Principia Mathematica, the canonical work of Whitehead and Russell, but confess, it gets over my head quickly).  

As I understand, all of maths derives essentially from one form or another of model and measure theory.  One sets up a vocabulary (the terms to be used), a grammar (the rules by which they are put together and to function), and a mapping from the vocabulary to the "space" to be described.  From these, basic axioms are agreed to (or not), and then everything is true or is not true.  Goedel showed that not everything that is true can be proved, but for all intents and purposes, it's a closed set.  At least practically speaking.  There are not "exceptions."  There is not interpretation or competing sets of facts.

Measure theory is the basis of real (and complex) analysis, and from there, probability and statistics.  

Economics is not like that.  Once can create models, of course (see Box above), but one simply must deal with complexities and realities that cannot be modelled.  This leads to famous (infamous) scandals - the recent work of Piketty, Capital in the 21st Century, has a number of controverseys surrounding it about how his data were chosen and analysed.  

Even more generally, attempts to 'control' the economy are fraught with peril.  My nine year old, reading in a history magazine about Al Capone, asked me why the Great Depression occurred.  In fact, the Great Depression had manifold "causes," no one of which really explains it.  Similarly, attempts to right the economic ship, some proposed by the most brilliant economists at the time, failed.  Spectacularly at times.


The unhappy truth is that NO ONE REALLY CONTROLS THE ECONOMY.  Not François Hollande, or Manuel Valls, or Paul Krugman, or Barack Obama. It's an incredibly complex system, and frequently, attempts to explain or affect it fail to understand that the variables are inherently interconnected.  

In my work, I create a number of what are called "systems dynamics models," which is to say, you push down on lever A, and something happens to levers B, C, and D.  It's much like the child's game "Mousetrap" pictured above.  More often than not, how the levers interact is not known; in fact, the correlations frequently cannot be known.

And most uncomfortably of all, if pressed, most economists, if they are being honest, admit that a large portion of the economy is driven by consumer behaviour, which is inherently not rational.  Worse still, it's not predictable.  In large part, the Great Depression happened because people (in many cases, rightly) lost faith in the banks.  Bank panics are an incredibly corrosive event, far more important than what the marginal tax rates or policies on hiring and firing are.  If people lose faith in the economy, they stop buying.  If they stop buying, companies stop producing.  If companies stop producing, they stop hiring.  

Lather, rinse, repeat.

The French government think that they can fix the economy by pulling just the right strings.  They think that they can create a French Silicon Valley by making hipster lofts with slick coffee kiosks in an abandoned warehouse.  One could argue whether the current iteration of Silicon Valley is in fact innovative, but one thing is clear.  Silicon Valley exists as a result of decades of somewhat organic growth, timing, and serendipity.  It exists despite, not because of, the government.  

Our leaders benefit from creating the illusion that they are in control.  It's comforting to think that they can solve economic problems.  King Canute demonstrated the limit of the King's power when he tried to order the tidewaters of the Thames to stop advancing.

At some point, Paul Krugman's feet are going to get wet.

Post a Comment