Fragility and Collapse: Slowly at first, then all at once

Josh Keyes

[In italiano]

[En français]

This article is based on the notes from one of the talks I gave at the Age of Limits conference.

I have been predicting collapse for over five years now. My prediction is that the USA will collapse financially, economically and politically within the foreseeable future... and this hasn’t happened yet. And so, inevitably, I am asked the same question over and over again: “When?” And, inevitably, I answer that I don’t make predictions as to timing. This leaves my questioners dissatisfied, and so I thought that I should try to explain why it is that I don’t make predictions as to timing. I will also try to explain how one might go about creating such predictions, understanding full well that the result is highly subjective.

You see, predicting that something is going to happen is a lot easier than predicting when something will happen. Suppose you have an old bridge: the concrete is cracked, chunks of it are missing with rusty rebar showing through. An inspector declares it “structurally deficient.” This bridge is definitely going to collapse at some point, but on what date? That is something that nobody can tell you. If you push for an answer, you might hear something like this: If it doesn’t collapse within a year, then it might stay up for another two. And if it stays up that long, then it might stay up for another decade. But if it stays up for an entire decade, then it will probably collapse within a year or two of that, because, given its rate of deterioration, at that point it will be entirely unclear what is holding it up.

You see, the timing estimates are inevitably subjective and, if you will, impressionistic, but there are objective things to pay attention to: how much structure is left (given that large chunks of concrete are continuing to fall out of it and into the river below) and the rate at which it is deteriorating (measurable in chunks per month). Most people have trouble assessing such risks. There are two problems: the first is that people often think that they would be able to assess the risk more accurately if they had more data. It does not occur to them that the information they are looking for is not available simply because it does not exist. And so they incorporate more data, hoping that they are relevant, making their estimate even less accurate.

The second problem is that people assume that they are playing a game of chance, and that it’s a fair one: something Nassim Nicholas Taleb calls the “ludic fallacy.” If you drive over a structurally deficient bridge every day, it could be said that you are gambling with your life; but are you gambling, exactly? Gambling normally involves games of chance: roll of the dice, flip of the coin, unless someone is cheating. Fair games form a tiny, insignificant subset of all possible games, and they can only be played in contrived, controlled, simplified circumstances, using a specially designed apparatus that is functioning perfectly. Suppose someone tells you that he just flipped a coin 10 times and all 10 were heads? What is the probability that the next flip will be heads too? If you think 50%, then you are discounting the very high probability that the game is rigged. And this makes you a sucker.

Games played directly against nature are never fair. You could say that nature always cheats: just as you are about to win the jackpot, the casino gets hit by an asteroid. You might think that such unlikely events are not significant, but it turns out that they are: Taleb’s black swans rule the world. Really, nature doesn’t so much cheat as not give a damn about your rules. But these rules are all you have go by: a bridge is sound if it corresponds to the picture in the head of its designer. The correspondence is almost perfect when it’s new, but as it ages a noticeable divergence takes place: cracks appear and the structure decays. At some more or less arbitrary point it is declared unsafe. But there is no picture in anyone’s head of it collapsing, because, you see, it wasn’t designed to collapse; it was designed to stay up. The information as to when it will collapse does not exist. There is a trick, however: you can observe the rate of divergence; when it goes from linear to exponential (that is, it begins to double) then collapse is not far, and you might even be able to set an upper limit on how long it will take. If the number of cement chunks falling out of your bridge keeps doubling, you can compute the moment when every last piece of the bridge will be in the river, and that is your upper bound.

Still, your forecast will be subjective (or, if you like, based on your luck as a forecaster) because you are still just playing the odds. If you measure that the deterioration in your bridge is linear (say one chunk falls out per month) then you extrapolate that it will remain linear; if it is exponential (2x chunks from the previous month) then you extrapolate that it will remain exponential, and, if you are lucky, it will. But the odds of it remaining one or the other are strictly in your own mind: they are not predictable but subjective. Calling them “random” or “chaotic” doesn’t add much: the information you are looking for simply does not exist.

To summarize: it is possible to predict that something will happen with uncanny accuracy. For example, all empires eventually collapse, with no exceptions; therefore, the USA will collapse. There, I did it. But it is not possible to predict when something will happen because of the problem of missing information: we have a mental model of how something continues to exist, not of how it unexpectedly ceases to exist. However, by watching the rate of deterioration, or divergence from our mental model, we can sometimes tell when the date is drawing near. The first type of prediction—that something will collapse—is extremely useful, because it tells you how to avoid putting at risk that which you cannot afford to lose. But there are situations when you have no choice; for instance, you were born into an empire that’s about to collapse. And that is where the second type of prediction—that something will collapse real soon—comes in very handy, because it tells you that it’s time to pull your bacon out of the fire.

Let me stress again: the process of coming up with such predictions is subjective. You might reason it out, or you might base it on a certain tingling sensation in the back of your neck. Still, people like to theorize: some declare that the events in question are random, or chaotic, and then go on to formulate mathematical models of randomness and of chaos. But the timing of large-scale, “improbable” events is not random or chaotic, it is unknown. With regular, small-scale events statisticians can cheat by averaging over them. That is useful if you are selling insurance—against events you can foresee. Of course, a large-scale event can still wipe you out by putting your reinsurer/underwriter out of business. There is fire insurance, flood insurance (not so much any more; in the US it is now underwritten directly by taxpayers), but there is no collapse insurance, because there is no way to objectively estimate the risk.

Plugging in everyone’s favorite Yogi Berra quote: “Making predictions is hard, especially if they are about the future.” Well, I beg to differ: making predictions about the past is just as difficult. The USSR collapsed unexpectedly in 1991, taking the “experts” by surprise. The root cause of the collapse remains veiled in mystery; the reason for the exact timing remains a complete mystery. Expert Kremlinologists were geared up to bet on minor power shifts within the Politburo, expert economists were entirely convinced about the superiority of free market capitalism over a planned socialist economy, expert military strategists could debate the merits of the Strategic Defense Initiative (there aren’t any) but they were all blindsided when the whole Soviet thing just folded up and blew away. Similarly, most political experts in the US are confident in their estimation of the odds that Obama will or will not be reelected in November 2012; what they can’t give you is the odds that the elections won’t be held, and that nobody will get to be president. Mind you, these odds are not zero, and we can be sure that such a day will come; we just don’t know when.

Experts can make predictions only within their area of expertise. They are constitutionally incapable of predicting when their area of expertise will undergo a spontaneous existence failure. Not being an expert in any of these disciplines, I knew that the USSR was going to collapse a year or so before it did. How did I know? By watching carefully, and by realizing that things can’t go on much longer in the same direction. I am doing the same with the USA now. So, let’s watch together.

* * *

The US Federal government is currently spending about $300 billion per month. To do so, it “borrows” around $100 billion per month. The word “borrows” is in quotes, because most of that new debt is created by the Treasury and bought up by the Federal Reserve, so in essence the government just writes itself a check for $100 billion dollars every month. If this continues forever, then the US Dollar will become worthless, so a push is on to get foreign central banks to take on some of this debt as well. They can do that, of course, but, seeing as the US Dollar is on track to become worthless, they have been decreasing their holdings of US Treasuries rather than increasing them. Nobody can tell how long such a scenario can continue to unfold, so what one looks for in a situation like this is signs of desperation.

Recently there was a flurry of activity around China: Secretary of State Hillary Clinton and Secretary of the Treasury Timothy Geithner, each with a large retinue, went to China on a high-level visit, during which the news coverage in the US was dominated by reports about a blind Chinese activist who was kept under house arrest, from which he escaped to the US embassy, and was eventually allowed to leave the country and come to the US. Hardly anyone in China knows who this person is, and the Chinese official reaction to demands that he be released were, pretty much, “Okey-dokey.” (The fact that Hillary seems to have given up on wearing makeup was considered newsworthy as well.)

Why such a powerful smokescreen? What were they hiding? Well, a couple of items of interest. First, it turns out that China can now monetize US debt directly. That’s right, the ability to print US currency is now distributed between the US and China. There is a special private line between Beijing and the US Treasury, and China can buy US Treasuries without going through any market mechanisms or making the price public. Secondly, China can now directly buy US banks. Back in the good old days attempts by foreign powers to use US Treasuries to buy equity in enterprises in the US was considered as akin to an act of war; nowadays—not so much. Basically, Hillary and Timmy went to China and said: “Take our financial system, please!” What they got is the financial equivalent of a subcutaneous morphine pump: something they give to terminal cancer patients, for continuous pain control. But what if it runs dry before the patient expires? That would be painful, wouldn’t it?

The US is bleeding money in other ways: wealthy individuals are moving abroad and renouncing their US citizenship in increasing numbers, like so many rats fleeing a sinking ship. A high-profile example is Eduardo Saverin, one of the founders of Facebook, who renounced his US citizenship prior to the ridiculous fiasco that was the Facebook IPO. Congress is busy drafting legislation to stop this sort of thing from happening, or at least make it a huge boondoggle from a tax perspective. There is also a provision in the works to take away people’s passports if the IRS decides that they owe more than $50k. Somebody ought to do something! Is it not possible to renounce your citizenship and buy votes in Congress at the same time? It should be... In any case, we can be sure that what is now still a trickle will turn into a flood. That is what I saw in Russia after the Soviet collapse: the former Soviet elite lost all faith in the system and tried to grab a chunk and run away with it. This pattern continues to this day: once something collapses, it tends to stay collapsed for a long time.

And why wouldn’t you want to flee like a rat, if you happen to be one of the many temporary millionaires who made a fortune in the US economy and do not wish to lose it? The US financial system is broken, and by now it is clear that it is not going to be fixed. Case in point: Jon Corzine, former Senator, former Governor of New Jersey, former head of MF Global, made some bad bets, then dipped into his customers’ accounts to cover his losses. Is he in jail? No, he is still at large and has nothing to fear. Furthermore, he is high on Obama’s campaign donor list. JP Morgan just reported a $2 billion trading loss (actually more like $8 billion). Is anything going to be done about it? Of course not! JP Morgan has a long and proud history of mismanaging risk, be it by using preposterous mathematical models (Value at Risk) or by having traders with nicknames like “the Whale” spontaneously decide that they are God and go hugely “naked long.” Since this was all done with taxpayer-backstopped funds (like other big US banks, JP Morgan is on government life support) there was some discussion as to whether the Whale was hedging, or betting, or gambling (with public funds). But nobody even knows the difference any more, and you can be sure that nobody will go to jail over this either.

And that brings us to the political system. Are the politicians even vaguely interested in reforming the financial system? No, they are too afraid of it. The financial reform legislation, such as it is, was drafted by the financial companies themselves and by their lobbyists. The politicians would be afraid to go near it, for fear of endangering their electoral campaign contributions. As long as campaign funds are flowing into their coffers, and as long as none of their banker friends ever goes to jail, they will remain unconcerned about finance. What they are increasingly paranoid about is their own physical safety. Both parties have repeatedly exhibited an unseemly amount of bipartisanship when it came to passing legislation to compromise civil liberties, to increase social controls and surveillance, and to take away their citizens’ rights. The 2013 national security budget promises to top $1 trillion. Again, the parallel with pre- and post-collapse USSR is striking: the political system there too was unreformable, hollowed out, and used for personal advantage, as a private service to the wealthy and the powerful. Criminals, such as Boris Berezovsky, ran for public office simply in order to gain the immunity from prosecution that came with it. This pattern continues to this day, especially in Ukraine: lose an election—go to jail. Get reelected—and you can use the voters who didn’t vote for you for target practice. Once a political system collapses, everyone strenuously denies that it has, but then it tends to stay collapsed for a long time.

What does tend to change rather suddenly is commerce. If you have enough financial and political shenanigans, high-level corruption and rule of law going by the wayside, daily life goes on just like before, for a while—until suddenly it doesn’t. In St. Petersburg, Russia, the difference between the summers of 1989 and 1990 was quite striking, because by the summer of 1990 commerce ground to a halt. There were empty shelves in shops, many of which were closed. People were refusing to accept money as payment. Imports dried up, and the only way to procure sought-after items like shampoo was from somebody who had traveled abroad, in exchange for jewelry or other items of value. And that occurred in spite of the fact that the USSR had a better overall business plan: theirs was: “Sell oil and gas, buy everything.” Whereas the business plan of the US has come down to: “Print money, use it to buy everything” (most consumer products, plus ¾ of the oil used for moving them and everything else around).

The imported oil is, of course, the Achilles’ heel of US commerce. The US economy was built around the principle that transportation costs don’t matter. Everything travels large distances all the time, mostly on rubber wheels, fueled by gasoline or diesel: people commute to work, drive to go shopping, taxi their children to and from various activities; goods move to stores in trucks; and the end product of all this activity—trash—gets trucked long distances as well. All of these transportation costs are no longer negligible; rather, they are fast becoming a major constraint on economic activity. The recurring pattern of the recent years is an oil price spike, followed by another round of recession. You might think that this pattern could continue ad infinitum, but then you’d just be extrapolating. More importantly, there is a reason to think that this pattern comes to a rather sudden end.

* * *

It is something of a general property of things that things build up slowly and collapse quickly. Examples of this sort abound (buildings, bridges, dams, military empires, economies, supernovae...) Counterexamples—things that appear suddenly and then slowly decay—are harder to find (mushrooms and cucumbers come to mind, but these are manifestations of an associated process of slow growth and sudden collapse, the collapse normally occurring right after the first frost). Some time ago it occurred to me that the symmetrical bell curve which is commonly used to model global oil depletion, known as the Hubbert Curve of Peak Oil theory, should actually be lopsided, like almost everything else, but I lacked the math to illustrate this point.

Eventually Prof. Ugo Bardi came through with a wonderfully simple and clear model, which he called the Seneca Effect. Unlike other models, such as the original Limits to Growth model, which, although vindicated, is too complex for most people to grasp at a sitting, the Seneca Effect is simplicity itself. This model initially includes two elements: a resource base and an economy. The rate of development of the resource base is proportional to both the size of the resource base and the size of the economy. Also, the economy decays over time at a rate proportional to its size. Set up the initial conditions, run the simulation, and you get a symmetrical bell curve. Now add a third element, which can be variously named “bureaucracy” or “pollution” or “overhead”: all the inescapable requirements or inevitable side-effects of having an economy. This element does not contribute to the rate at which the resource base is developed. It also decays at a rate proportional to its size. Divert some fraction of the resource flow to this element, run the model, and out pops a lopsided curve: rising slowly, falling swiftly: the Seneca Cliff. The larger the fraction being diverted, the more lopsided the curve:

There is one problem with this model: we don’t really know which elements of the economy are productive (in terms of contributing to the rate at which the resource base is converted into capital) and which ones are non-productive and belong in the bureaucracy/pollution/overhead bucket. When we look at the world, we see the two summed together and can’t tease them apart. With this detail hidden from view, collapse becomes hard to see in the aggregate: the people may be starving, but there is also a lot of fat bureaucrats carving up, roasting and eating each others’ ample buttocks, so it all averages out for a while longer. But you can still tease it apart based on the fact that certain things simply stop happening. The progression to watch for is: things get bigger and bigger, then suddenly stop.

An associated problem is that the fraction of resources going to bureaucracy/pollution/overhead usually starts out being reasonable (a quarter or a third or so) but the closer the economy comes to collapse, the higher this fraction becomes. We can observe this in the US: more and more resources have been allocated to bailouts, make-work “economic stimulus” projects and national security; more and more pollution (and associated costs) from offshore oil spills and from the development of marginal, dirty energy resources such as shale oil and tar sands. As the productive part of the economy begins to fail, the bureaucrats grow desperate but, being bureaucrats, all they can do is endlessly increase the bureaucratic burden, accelerating the downward slide. Most people have heard of Gorbachev’s glasnost' and perestroïka, but there was a third initiative, acceleration (uskorenie): the doomed attempt to get the moribund Soviet economy to perform better. It sent it into shock instead.

Things get bigger and bigger, then suddenly stop. Let us look at the example of US retail. Once upon a time there was local industry, which sold products through small shops. Over the course of a few decades, the industry moved to other countries, mostly to China, and the small shops were put out of business by department stores, then by malls, culminating with Walmart, which practices “slash and burn retail”: since most of what it sells is imported, it empties the local economy of money, and is then forced to close, leaving devastation in its wake. Walmart is now expanding in China, having finally realized that it doesn’t work to sell stuff in a country that doesn’t make stuff once that country is fresh out of money. In places where retail has ceased to exist, the remaining recourse is Internet shopping, thanks to UPS and FedEx. And once UPS and FedEx services become unaffordable because of rising energy prices or unavailable because of unmaintained, impassable roads and bridges, local access to imported goods is lost.

Similarly with US banking. Once upon a time there were small neighborhood banks that took in the people’s savings and then lent it out to individuals and businesses, helping the local economy grow. Over the course of a few decades, these small neighborhood banks were replaced with a few huge megabanks, which, after 2008, became, in effect, government-owned. Once the megabanks close their local branches, local access to money is lost.

Similarly with global shipping. Once upon a time there were many small ships, called tramp steamers, which were loaded and unloaded by longshoremen at local ports, using block and tackle and cargo nets. Then shipping became containerized, and moving cargo required a container port. Then the container ships became staggeringly huge. Then, as oil prices went up, they had to resort to “slow steaming” by pulling pistons out of their engines and going slower than the sailing ships of yore. Instead of point-to-point trade, these giant container ships can only operate within hub-and-spoke networks, with the spokes provided by somewhat less energy-efficient trains and far less energy-efficient long-distance trucking. These ships are now at the limit of “slow steaming.” The next step is, obviously “no steaming” at all.

Similarly with medicine. Once upon a time there were family doctors—general practitioners who made house calls, and neighborhood clinics. Eventually these were replaced by megahospitals and giant medical centers staffed with specialists, which, over time, became unaffordable for the general population. The US is currently spending over 17% of its GDP on medical care—an amount that is exorbitant and unsustainable. Once this spending is curtailed, many of the megahospitals will be forced to close. The population will, for a time, still have access to WebMD and to mail-order drugs, and, in case of serious illness or emergency, medical evacuation will remain an option for those still be able to afford it.

The state of the communications infrastructure in the US makes a particularly interesting case. The US is now behind most developed nations in access to the Internet. Many people in rural parts of the US must rely on their cell phones for Internet access, putting the US on par with such countries as Cambodia, Vietnam, Indonesia and the Philippines. However, cell phone service is far more expensive in the US than in any of these countries. Given that most products and services are now available mainly through the Internet, and that the Internet requires a steady supply of electricity, the state of the electrical grid in the US presents an even more interesting case. It is a severely overworked network of aging transmission lines and transformer farms, some dating back to the 1950s.

There is over 100 nuclear power plants, which are growing old and dangerous, but their service lives are being artificially extended through re-licensing. There are no plans, and no money, to dismantle them and to sequester the high-level radioactive waste at a geologically stable underground location. If deprived of both grid power and diesel fuel for an extended period of time, these plants melt down à la Fukushima Daiichi. It bears mentioning that a nuclear disaster, such as Chernobyl, is a particularly potent ingredient in precipitating a political collapse. Since what is keeping a series of such disasters from happening is the electric grid, followed by diesel, let us examine each of these in turn.

With regard to the electric grid, the incidence of major power outages has recently been seen doubling every year. Yes, we are committing the inductive fallacy by simply extrapolating this trend into the future, but, given what is at stake, dare we not extrapolate? At the very least, we would need to hear a very good reason why we shouldn’t. The incidence of major power outages can only double so many times before it’s time to start handing out potassium iodide tablets and before wig prices shoot through the roof.

Unless, of course, the diesel generators can be kept running continuously for the 15-20 years it would take to shut down, de-fuel and decommission all the nuclear reactors and empty the nuclear waste storage ponds. Countries that lack a reliable electric grid tend to rely on diesel generators. There is currently a lot of pressure on diesel supplies, especially since Japan took all of their nuclear generation capacity off-line following the Fukushima Daiichi disaster, with high diesel prices and spot shortages in many countries. Observing the increased incidence of power outages and price spikes, many companies in the US have installed emergency diesel generators, and are now finding that they run them even when grid power is available, whenever requested to do so by the power company.

Not much of anything continues to operate in the US once the electric grid is down. Earlier this year a central part of Boston where I was working at the time (Back Bay) went dark because of a transformer fire. For almost an entire week every business in the area was shut down. Without power, there is no heat or hot water, there is no pumped water, or, more frighteningly, no pumped sewage, there is no air conditioning (which is fatal, through heat stroke, in places such as Atlanta, Georgia, which often have 100% humidity coupled with above-body-temperature summer ambient temperatures). Security systems and point of sale systems stop functioning. Cell phones and laptops cannot be charged. Highway and subway tunnels flood and bridges do not open to let shipping traffic through—such as barges loaded with diesel. Can we be sure that diesel will continue to be supplied to all active nuclear power plants even as everything else falls apart?

This is usually the point in my talks when somebody in the audience pipes up to say: “This is all doom and gloom, isn’t it?” To which I say, “For you, maybe, if you don’t have any other plan except to wait for everything to somehow magically fix itself.” You see, building something that works takes a lot of time and effort. Things stop working in a hurry, but making a replacement takes time, resources, and, most importantly, stability. This can only be done ahead of time, and doing so takes practice (by which I mean learning from one’s own plentiful mistakes). If you wait until that last moment when, in a spasm of horror, you suddenly think to yourself “Oh shit, Dmitry was right!” then indeed Doom and Gloom will be your charming new bunkmates. But if you start your collapse early and get it over with quickly, then your chances of surviving this are quite likely to substantially exceed zero.

And so, please don’t ask me “When?”—do your own thinking! I’ve given you the tools you need to come to your own conclusions, based on which you may be able to start your collapse early and get it over with quickly.