In The Productivity Paradox
David Rotman writes:
Productivity growth in most of the world’s rich countries has been dismal since around 2004. Especially vexing is the sluggish pace of what economists call total factor productivity—the part that accounts for the contributions of innovation and technology. In a time of Facebook, smartphones, self-driving cars, and computers that can beat a person at just about any board game, how can the key economic measure of technological progress be so pathetic? Economists have tagged this the “productivity paradox.”
Some argue that it’s because today’s technologies are not nearly as impressive as we think. The leading proponent of that view, Northwestern University economist Robert Gordon, contends that compared with breakthroughs like indoor plumbing and the electric motor, today’s advances are small and of limited economic benefit. Others think productivity is in fact increasing but we simply don’t know how to measure things like the value delivered by Google and Facebook, particularly when many of the benefits are “free.”
My view is that IT is only one of the factors driving the decrease of productivity in the general economy, but that there are some areas of the economy in which IT is greatly increasing productivity. An explanation is below the fold.
The original productivity paradox was described by Erik Brynjolfsson in 1993's The productivity paradox of information technology
One of the core issues for economists in the past decade has been the productivity slowdown that began in the early 1970s. Even after accounting for factors such as the oil price shocks, most researchers find that there is an unexplained residual drop in productivity as compared with the first half of the post-war period. The sharp drop in productivity roughly coincided with the rapid increase in the use of IT ... Although recent productivity growth has rebounded somewhat, especially in manufacturing, the overall negative correlation between economy-wide productivity and the advent of computers is behind many of the arguments that IT has not helped US productivity or even that IT investments have been counter-productive.
In The Puzzle of the US Productivity Slowdown
Timothy Taylor runs through a number of possible explanations put forward by the Congressional Budget Office, and the CBO's explanation for why they don't apply:
- Is the productivity slowdown a matter of measurement issues?
- Is the productivity slowdown a result of slower growth feeding back to reduced productivity growth?
- Is it a result of less human capital for US workers, either as a result of less experience on the job or reduced growth in education?
- Is the problem one of overregulation?
- Is the scientific potential for long-term innovation declining?
The CBO's skepticism of last of these is based on this observation:
no evidence exists of an abrupt change around 2005 connected to such developments.
But as I discussed in Falling Research Productivity
, based on Scott Alexander's Considerations on Cost Disease
and Are Ideas Getting Harder to Find?
by Nicholas Bloom et al
, it is certainly the case that R&D productivity is falling. The simple explanation is in this comment
Kelvin Stott's 2-part series Pharma's broken business model: Part 1: An industry on the brink of terminal decline and Part 2: Scraping the barrel in drug discovery
uses a simple economic model to show that the Internal Rate of Return
(IRR) of Pharma companies is already less than their cost of capital,
and will become negative in 2020. Stott shows that this is a consequence
of the Law of Diminishing Returns; because the most promising research
avenues (i.e. the ones promising the greatest return) are pursued first,
the returns on a research dollar decrease with time.
It is likely that the slowdown around 2004 was due to a combination of factors, none large in isolation, combining to exceed a critical level. One of them was very probably IT, because the notorious failure rate of large IT projects was driving up the cost while driving down the benefits. These pie charts, showing that the odds of success decay rapidly with size, are based on a study of over 50,000 software projects over 8 years by the Standish Group.
But there are clearly some areas of the economy where IT has greatly improved productivity. One recent example is documented in a report
from blockchain analytics firm Chainalysis
(hat tip to Technology Review's The Download
We took a look at hacks that target cryptocurrency organizations such as exchanges. These hacks involve large thefts, often stealing tens or even hundreds of millions of dollars directly from exchanges. Hacking dwarfs all other forms of crypto crime, and it is dominated by two prominent, professional hacking groups. Together, these two groups are responsible for stealing around $1 billion to date, at least 60% of all publicly reported hacks. And given the potential rewards, there’s no question hacking will continue; it is the most lucrative of all crypto crimes.
So, thanks to IT, two small groups mounted heists yielding "around $1B to date". This level of productivity would have been impossible before the advent of IT. Crypto crime is but one small example of the extraordinary productivity of IT-enabled criminals. Others include:
- I have written before about the immensely profitable business of ransomware. Two years ago, discussing the losses from Internet crime, Quinn Norton wrote:
The predictions for this year from some analysis is that we’ll hit seventy‐five billion in ransomware alone by the end of the year.
These are total losses; the fraction realized by the ransomware gangs is much less, probably only a few percent. But that's still a level of productivity impossible without IT.
- Probably even more profitable is advertising click fraud. The Association of National Advertisers wrote about 2017:
The third annual Bot Baseline Report reveals that the economic losses
due to bot fraud are estimated to reach $6.5 billion globally in 2017.
This is down 10 percent from the $7.2 billion reported in last year's
study. The fraud decline is particularly impressive recognizing that
this is occurring when digital advertising spending is expected to
increase by 10 percent or more.
That's $6.5B/year revenue for the click fraudsters. Before IT, criminal gangs grossing $6.5B/year would have had much lower margins. Drug smugglers, for example, would have to spend on planes, boats, staff, and bribes, not to mention raw materials. None of these are needed for click fraud.
- But these amounts are small change compared to Wall Street's ill-gotten gains in the Global Financial Crisis:
To begin with, a number of big hedge funds figured it out. Unlike investment banks, however, they couldn't make serious money by securitising loans and selling CDOs (collateralised debt obligations), so they had to wait until the bubble was about to burst and make their money from the collapse. And this they did. Major hedge funds including Magnetar, Tricadia, Harbinger Capital, George Soros, and John Paulson made billions of dollars each by betting against mortgage securities as the bubble ended, and all of them worked closely with Wall Street in order to do so.
The CDOs and even the underlying mortgages depended upon IT systems such as that operated by Mortgage Electronic Recording Systems (MERS):
It is the company created and owned by all of the big banks to process title to property in the U.S. Approximately 60% of the nation’s residential mortgages are recorded in the name of MERS.
These criminal schemes would have been impossible without the IT systems on Wall Street. The Financial Crisis Cost Every American $70,000, Fed Study Says. The US population is around 325 million, so the cost to Americans alone was around twenty-three trillion dollars. The proportion that ended up with the perpetrators was small, but still vastly exceeds the proceeds of crypto-crime, ransomware, and click-fraud.
MERS is a shell corporation with no employees, but thousands of officers.
MERS, the banks and the mainstream financial press all say that it was simply to save fees by digitizing mortgage electronic.
But as Ellen Brown notes, there is in reality a very different reason that the big banks created MERS:
The rating agencies required that the conduit be “bankruptcy remote,” which meant it could hold title to nothing ….
Some crime, typically sex and drugs, is included in some countries' GDP computation
, but the US
has no plans “for now” to start counting illegal sex and drugs. “We need to look at the issue more closely to see what data are available before any decision could be made,” said Jeannine Aversa, chief of public affairs and outreach at the United States Bureau of Economic Analysis, in a statement. “We haven’t done any research yet, so we don’t know how much this would add to the U.S. economy as measured by G.D.P.”
So, what we have is a small number of people (the denominator) generating a large amount of income (the numerator), which implies high productivity. But since their income is not
included in the numerator but they are
counted in the denominator, the effect is to reduce
GDP and thus reported productivity.
In The market for cyber-insurance is growing
, The Economist
writes that companies are increasingly incurring losses from cyber-crime:
Such mishaps are feeding a fast-growing market for specialist cyber-insurance. Solid numbers are in short supply, but Munich Re, a reinsurer, reckons that a market that wrote $4bn of premiums in 2018 could be writing $8bn-9bn by 2020. Rob Smart of Mactavish, a firm that works with big British insurers, says that “almost all” the firms’ clients have inquired about cyber-insurance in the past couple of years.
and the expenditures incurred in recovering from attacks, are
included in the GDP figures, so at least a little of the effect
of Internet criminality increases GDP. But, as The Economist
makes clear, cyber-insurance is not a panacea. For example
working out who was behind a particular hack has already made the news. Mondelez, an American food company hit by the NotPetya malware, is suing Zurich, a big insurance firm, for refusing to pay out under a general insurance policy. Zurich cites an exclusion clause for losses related to war, on the ground that the NotPetya attack is thought to have been carried out by Russia.
As I discussed in Correlated Cryptojacking
, a much bigger problem is that of correlated risk
Perhaps the biggest difficulty for insurers is that the risks posed by cyber-attacks are not independent of each other. If an oil refinery in Texas floods, that does not mean one in Paris is any more likely to do so. Insurers build that independence into their risk models, and depend upon it in their calculations of the maximum they may have to pay out in a single year. But a newly found flaw in software can make all users vulnerable simultaneously. Insurers fret that a single big attack could hit many of their clients at once. In the worst case, the value of claims might be more than they could meet. ... Whether the industry can figure out a way to deal with such “risk aggregation” is an open question. As one insider says, it “sort of breaks the whole concept of insurance a bit”.
What all this shows is that "GDP" and "productivity" are pretty silly things to measure. This is also emphasized by Adam Tooze's tweet
of this graph:
Just one more reality check on the relative performance of the advanced economies in terms of labour productivity. Nothing to choose btw Germany, US, Switzerland AND France. The difference is in hours worked/unemployment rate.
The point being the difference between GDP per hour worked
, and the more normal graphs of GDP per head
. Oh, and ignore Ireland, whose GDP is mostly made up of tax avoidance by large US companies, and Norway, whose GDP is mostly North Sea oil, neither of which involves a lot of work by their population.
Gareth Corfield's Black-hat sextortionists required: Competitive salary and dental plan reports that:
"Extortionists are promising salaries of more than a quarter of a million pounds to skilled infosec folk willing to put on a black hat, according to research outfit Digital Shadows.
Those salaries are on offer to people willing to blackmail and extort money out of "high net worth individuals" – and at the upper end of the scale have even reportedly topped £840,000."
Now, that's evidence of high productivity!
Neil Irwin's fascinating Upshot column What if All the World’s Economic Woes Are Part of the Same Problem? includes an interesting alternative explanation for slowing productivity:
"Adam Ozimek, Dante DeAntonio and Mark Zandi analyzed data on work force age and productivity at both the state and industry level, with payroll data on millions of workers. They found that the second effect seems to prevail, that an aging work force can explain a slowdown in productivity growth of between 0.3 and 0.7 percentage points per year over the last 15 years."
Paul Krugman calls it:
"the YCTAODNT (you can't teach an old dog new tricks) theory of the productivity slowdown"
Paul Krugman points out that, if robots were rapidly taking over jobs from humans, productivity would be rising rapidly, but it isn't.
Cybercrime productivity is rising rapidly, according to the FBI, as reported by Xeni Jardin in FBI: Online theft, fraud, exploitation caused losses of $2.7B globally in 2018, up from $1.4B in 2017.
Paul Krugman tweets this graph, showing productivity growth slowed in the aftermath of the global financial crisis and hasn't recovered.
Bryce Elder discusses a possible explanation for lackluster productivity growth in Why don’t computers work?:
"A paper from University of Lausanne PhD student Seda Basihos makes an interesting contribution to the debate. (Note: unreviewed preprint, there be dragons.) She argues that because of rapid obsolescence, computing is a uniquely pernicious force.
Computers are the worst thing to happen to the global economy in 150 years because . . . well, you will have probably guessed already. Every digital fix has a knack of creating three new problems. Any tweak threatens to invoke the recursive loop of pointless labour. A PC might look modular but it’s a morass of potential incompatibilities and performance bottlenecks, meaning entire corporate systems are junked whenever a software update or a withdrawal of OEM support prematurely terminates the usefulness of one part. And because of this accelerated replacement cycle, workers have to continually relearn their jobs."
Post a Comment