It's Decidedly Not the Math. It's Always People

It's Decidedly Not the Math. It's Always People
AP Photo/Kin Cheung, File
Story Stream
recent articles

The government of Great Britain had sunk more than £17,000 in Charles Babbage’s project by 1832 and wanted to see what had been accomplished with the funds. Babbage was a very clever fellow with a wide range of interests and capabilities, a true Renaissance polymath. He had been one of the founders of England’s Analytical Society in 1812, an organization intended to induce exciting new mathematical concepts across the Channel from Europe.

Working with another body he helped to create in 1820, the Royal Astronomical Society, Babbage and another man were tasked with improving vital navigation tables. Tedious and long, the necessary calculations were assigned to separate clerks, done twice so as to be able to better spot errors in the derived results.

As expected, the pair found many, too many, leading Mr. Babbage to exclaim, “I wish to God these calculations had been executed by steam.” It wasn’t the first time he had dreamed about such a way to reach solutions; in his autobiography the man claimed to have had a dream in 1812 or 13, sleeping overtop a table of logarithms and lazily hoping one day to have them all “calculated by machinery.”

In 1822, he set out to do just that. Charles wrote the Astronomical Society to inform them of his idea, which the society enthusiastically endorsed, and then an article published in Brewster’s Journal of Science, titled, On the Theoretical Principles of the Machinery for Calculating Tables.

“I have contrived methods by which type shall be set up by the machine in the order determined by the calculation. The arrangements are such that... there shall not exist the possibility of error in any printed copy of tables computed by this engine.”

By 1823, he had had an interview with the Chancellor of the Exchequer and by the next year had been advanced £1,500 from the government to work on his mechanical opus. Expecting initially to take three years, eight years and thousands of pounds later Parliament and the Royal Society, to which Babbage also belonged and which had infused more than its enthusiasm, demanded a demonstration.

He called it his “difference engine” and showed it to MPs, two of whom, according to Babbage, had wondered if it possessed borderline magical properties.

“Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?”

His unspoken response, Charles later recalled, would be classified today as irreverent snark. “I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question.”

As comedian Ron White says, you can’t fix stupid (but you can drink while watching it).

Charles Babbage’s difference engine didn’t amount to all that much, the project shut down not long after its appearance at Parliament. Undeterred, he would go on to create a new version, the Analytical Engine, which is credited as the modern forerunner of the digital computer. Genius takes time; stupidity, as we all witness, is eternal.

While the machines have become more sophisticated the understanding surrounding them often has failed to keep up. Bestowed mystical properties because of the computational power they do possess, computers and computer programs are treated as modern Oracles. Believed to be pure science, who would argue with the machine?

George Fueschel had been working for IBM in the late 1950’s, teaching the company’s customers in New York about the formidable new 305 RAMAC. As amazingly capable as it was for the time, it was still a crude device not all that much more advanced (relative to where computers are today) than George Babbage’s Analytical Engine.

Errors were common. Not because of the calculations the 305 performed or produced, but in a more powerful fundamental concept: if you pour garbage into any computer you should expect only garbage to come flying out of it. Its capacity limited to performing the duties George Babbage had envisioned and nothing more.

Fueschel is credited with coining the phrase Garbage In/Garbage Out, or GIGO.

There’s no God from the machine.

Thus, if flawed human understanding of utterly complex systems and topics is used to create even the most elegant means to calculate the results from a set of problems, what will come out of those calculations is a flawed human understanding of utterly complex systems and topics. The numbers might be beautiful and magnificent, and still downright meaningless and often misleading.

But that’s not the world we’ve been pushed into. Instead, we are the two guys from the British Parliament who have been purposefully led to believe a complex series of regressions is the equivalent of mathematical alchemy. No matter how flawed our conception of this mortal coil we live in, the numbers are ceded their unearthly weight regardless.

Math has taken over without anyone having stopped to consider Charles Babbage’s question; and Ron White’s truism. If you are intellectually talented enough to create the most advanced statistical model, very much like a computer program, that just means you’re really good at statistics (or programming).

If you are stupid on the subject of, say, the economy, or modeling the spread of a pandemic, then Stupid In/Stupid Out. Thus, you can’t fix either stupid or stupid’s models.

Ronald Coase was a Nobel Prize-winning economist who warned the world of taking this path. Accepting his award in 1991 for work he had performed ages before, the older Coase admonished Economics right in its Temple.

“This neglect of other aspects of the system has been made easier by another feature of modern economic theory – the growing abstraction of the analysis, which does not seem to call for a detailed knowledge of the actual economic system or, at any rate, has managed to proceed without it.”

Coase was talking about econometrics, and largely its same deficiency as what had Babbage up in arms. Economists had substituted direct and full knowledge on the subject of the real economy in exchange for becoming statisticians. They can create the most breath-taking series of equations that ultimately do nothing to help you understand how the economic system really works, let alone how anything will turn out.  

Stupid Garbage In/Financial Crisis Out.

One of the names that always comes up when reviewing GFC1, the first Global Financial Crisis during 2007, 2008, and 2009, is Bill Dudley. He was the Manager of the System Open Market Account during that time, a key position which is where the presumed dirty work of monetary policy gets done at its New York branch.

Dudley was the guy who, after consulting the usual statistical simulations, told the FOMC right at the outset of the crisis, on August 7, 2007, that there wasn’t much to be worried about (below taken from the meeting transcript).

“MR. DUDLEY. We’ve done quite a bit of work trying to identify some of the funding questions surrounding Bear Stearns, Countrywide, and some of the commercial paper programs. There is some strain, but so far it looks as though nothing is really imminent in those areas.”

Two days later, TWO DAYS, the entire commercial paper market seized up completely, Countrywide became a super-spreader of liquidity contagion, and Bear Stearns was set on the road to its, and ultimately our, ruin.

Having far outdone Ben Bernanke’s earlier “subprime is contained” fiasco – though Dudley’s was uttered in private which is why it isn’t more widely mocked – the man went on to botch every single thing he had his hand in; which, since he was in charge of the Open Market Desk, meant all the things.  

Bill was treated like every other key central banker who performs in this way. He was promoted to the presidency of the Federal Reserve Bank of New York after Tim Geithner, who had been president and Dudley’s boss throughout the disaster, was promoted to Treasury Secretary. Thorough bureaucracy rather than central bank, the ability to read and stick to approved models far more in demand than the ability to safeguard the world from financial disintegration.

On the cusp of retirement from that top job in 2018, Bill Dudley finally worked up the “courage” to admit a fateful truth on his way out the door. Literally his last day, Dudley suddenly confessed:

“One of the challenges going into the financial crisis, for example, if you look at the big DSGE model—dynamic stochastic general equilibrium model—it didn’t include a finance sector. So the whole experience of what actually happened during the global financial crisis—the collapse of the financial system and that taking down the real economy—wasn’t an actual possibility within the major macro models that some economists were using to forecast the economy.”

And even then he was being purposefully cute. These weren’t models “some” Economists were using, it was every model all Economists and central bankers (redundant) would turn to at each and every twist and turn during GFC1.

One of those was the failure and forced merger of Bear Stearns in the middle of March 2008. Because these guys were allowed to get away with blaming subprime mortgages for the crisis, people don’t realize that GFC1 was really two: it had been just the first one which finally destroyed enough leverage and risk-taking to reach a near equilibrium or settled state in Bear’s untimely demise.

Policymakers, looking at their corrupt DSGE models, didn’t realize this, either.

Instead, ignoring mountains of market prices and evidence, rather than heed these warnings, they all thought they had performed wonderfully. A successful conclusion to the crisis by Spring 2008 which by then had appeared to amount to little more than a single big name not unlike LTCM in ’98, and in the economy a single quarter (Q1 ’08) with slightly negative GDP. The congratulations went on for months after Bear.

By June of 2008, because the second crisis hadn’t erupted yet and the DSGE models were all predicting a return to growth in the second half of the year, policymakers made their biggest, costliest mistake of the whole thing (from the meeting transcript):

“MR. BULLARD. Recent data on the U.S. economy have been stronger than forecast, keeping economic performance weak but avoiding a particularly sharp contraction. The worst outcomes stemming from financial market turmoil have failed to materialize thus far. There is, to be sure, still some potential for additional upheaval, depending in part on the managerial agility among key financial firms. However, the U.S. economy is now much better positioned to handle financial market turmoil than it was six months ago. This is due to the lending facilities now in place and to the environment of low interest rates that has been created.Renewed financial market turmoil, should it occur during the summer or fall, would not now be as worrisome from a systemic risk perspective.” [emphasis added]

James Bullard’s whopper is right up there with Bernanke’s and Dudley’s. Garbage In: the economy was insulated by fast-thinking central bankers, the financial system having been fortified by “lending facilities now in place.” Garbage Out: the second and more destructive half of GFC1 would strike in a matter of weeks.

Oh, how history does rhyme.

Over the past few weeks, the Federal Reserve has been congratulated and celebrated for, you guessed it, “lending facilities now in place.” In a particularly cruel twist of fate, these are almost entirely the same ones that did nothing to prevent or shorten GFC1.

Dudley’s 2018 confession notwithstanding, the models that had the Great “Recession” as an impossibility ten years before consider another one in the same way. They’ve not been changed or meaningfully updated. They can’t be meaningfully updated.

Charles Babbage would’ve told you why in 1832, as he did the Royal Society.

Which means we are being royally screwed three times right now by these things. First, the virus models which were wildly inaccurate as to the spread and seriousness of the coronavirus, an order of magnitude maybe even two off, and these were the basis for this insane, overwrought and near-total – if not unconstitutional then thoroughly un-American – shutdown.

In prioritizing “bending the curve” the decision was made based on econometric models which had continuously suggested the US economy was in a good place at its start. These same models that had predicted 2019 being an inflationary breakout of a boom, complete with an increasing need for more rate hikes, when last year instead ended up at best with a serious downturn, certainly a manufacturing recession, complete with a full-on liquidity warning in last September’s repo market.

And the modeled reason why the economy was thought to be A-OK in early 2020? Unforeseen rate cuts and “lending facilities now in place” in the form of “repo” operations and not-QE both instituted in the wake of that unexpected and misjudged prior repo rumble.

Repo operations that during the worst days last month went largely unused; in some cases, totally and completely devoid of any takers. The facilities were in place, though, as if that’s all there is to it.

Already Jay Powell is doing his best Ben Bernanke imitation. It was his predecessor’s predecessor who perfected the central bank’s most potent weapon – the misdirection. Bernanke was the master of the quick, thorough transition; going from “our policies have made sure nothing happens” to “our policies will make sure it doesn’t get any worse” without skipping a beat.

That’s where we are right now. Having used “lending facilities now in place” from which to watch the stock market meltdown and credit markets become bid-less arenas of fire sale liquidations, destructive gamma rather than human disease having been so rapidly dispersed to every corner of the world, we’re now supposed to believe the same things have kept it all from becoming any worse. Hurray for Jay!

And while we are intended to swallow such a message, the economic data is coming in far beyond the capacity of even adept wordsmiths to adequately style. Coronavirus shutdown, yes, but to these levels this quickly it can only be the sort of deep demand destruction more directly associated with the letters G, F, and C. The same sort, by the way, unleashed especially in the Fall of ’08 (the shocking decline in the labor force, most particularly, which has just been entirely outdone, in one month, this time around).

That’s the third way models are leaving the world exposed, unnecessarily, to the prospects for further unimaginable economic pain. Not just the short run, either.

Models being models, what about markets? These are telling you, right now, the Fed failed, will fail, and so we are at huge risk of a repeat of March. Nothing is ever inevitable, but standing against Powell’s statistics are history and common sense. A second part of GFC2, like GFC1, where “renewed financial turmoil, should it occur during the summer or fall” actually would be “as worrisome from a systemic risk perspective” if not more so.

It's not the math. As always, it’s people.

The statistical sanitation engineers said the economy was in fine shape. They also said the virus, even with heavy mitigation efforts, would quickly overwhelm the healthcare system. Putting faith in the latter and because of the former, we’re left with a gigantic mess their numbers now say will in short order become nothing more than an unpleasant memory.  

Stupid. A religious, cult-like faith in these things had already led us to one Global Financial Crisis. I am not able rightly to apprehend the kind of confusion of ideas that could provoke such widespread, accepted garbage.

Jeffrey Snider is the Head of Global Research at Alhambra Partners. 

Show comments Hide Comments