How gloriously rich AI will make us
Well, assuming that aritifical intelligence is going to add any value at all that is.
Last week, Eliezer Yudkowsky, co-founder of an AI lab, made a startling discovery: AI models can copy each other far more easily than anyone previously thought, and the larceny is almost impossible to stop.
“If you allow any sufficiently wide-ranging access to your AI model,” he explained, “you're giving away your business crown jewels to competitors that can then nearly clone your model without all the hard work you did.”
The result? “You no longer have a competitive moat,” he says, referencing industry jargon for a competitive advantage that’s hard to beat.
Assume that some new technology actually is going to add some value. There are two places that value add can go. Either to consumers in the consumer surplus - the amount we would be willing to pay but don’t have to - or to the capitalists in profits.
The defining point over which happens is how easy is it to copy that first innovation? If it’s difficult then that’s Warren Buffet’s moat against competition coming and eating your profits. If it’s easy then competition does eat all the profits and the value add ends up with consumers.
The easier AI is to copy then the more of the value will accrue to consumers - the easier it is to copy the richer it will make us and not the capitalists.
Which is nice, isn’t it? Also, it means we don’t need to tax the capitalists who make the AIs because we’ll already be doing that through the competition. Which is also nice - we leave them be to make us richer that is.
If it's good for you, it's good for me
The Adam Smith Institute enjoyed some successes in the Budget yesterday. With ‘full expensing’ at one point coined the ‘best policy you’ve never heard of’ by our Senior Fellow Sam Bowman. Now I’m not sure it deserves this moniker at all - as the Chancellor gave us ‘full expensing’, which means firms can write off (some) their investment expenditure.
Alongside Abolishing the Factory Tax, the ASI had another, more understated policy win. This has garnered less public buzz, but will bring about outsized value at the cost of a piece of low-hanging fruit: medicinal reciprocity with trusted countries.
The Chancellor told us he will implement Sir Patrick’s interim recommendations for Life Sciences reform. One policy recommendation concerns the Medicines and Healthcare products Regulatory Agency (MHRA) which will now explore: ‘partnerships with trusted international agencies [...] to provide simple, rapid approvals for medicines and technologies that have received their approval from 2024.’ (Spring Budget, 2023: 67). In short - adopting the best drug and medical device regulatory practices from around the world.
Charter Cities CEO Mark Lutter wrote an ASI paper advocating for such a policy. This called for allowing the use of drugs and medical devices that have been approved by the regulatory agencies of other developed countries. This could include the Food and Drug Administration (FDA) of the USA, and Japan’s Pharmaceutical and Medical Devices Agency. If Japan or the US approve a drug, it will be automatically approved for prescription here in the UK.
Drug reform of this kind means that countries can streamline the process of getting new medications approved, saving time and resources, and lowering the transaction costs of bringing new potential life-saving drugs to market. That money once earmarked for burdensome regulatory processes, now becomes unlocked for direct life sciences research and development (one would hope). It could also incentivise further innovation, by encouraging collaboration between the pharmaceutical companies of each country within the reciprocity web.
It’s such an easy win that one can only wonder: why has it taken the UK so long post-Brexit to implement? This piece of low-hanging fruit has been hitting us in the face every time we walk into the Garden of Potential Policies for years until yesterday we finally decided to pick it off its branch! Why didn’t we do this pre-Brexit? Because we couldn’t.
Before Brexit (remember that time?), the UK’s membership to the European Union (EU) meant that we were a part of a centralised system for the approval and regulation of medical drugs through the European Medicines Agency (EMA). We were required to follow the EU's regulatory framework and could not unilaterally reciprocate drugs with countries outside the EU without adhering to the EMA's standards and processes.
Even outside of its own multiple-state-lined walls, the EU has its own set of agreements with third countries (like the US and Japan) for the mutual recognition of drug approvals and regulatory standards. These agreements are negotiated at the EU level, and individual member states, including the UK before Brexit, still did not have the authority to establish separate agreements with non-EU countries. This centralisation of regulatory processes within the EU was designed to maintain consistency and quality control for drug approvals across all member states, but could nevertheless prove sclerotic.
By recognizing each other's regulatory processes and sharing information, the UK can follow the lead of trusted international partners and expedite the availability of new treatments, promote innovation, and ultimately improve patient outcomes. As the world continues to face new healthcare challenges, the importance of international collaboration and reciprocity in the pharmaceutical industry cannot be understated.
That grand impossibility of planning the economy
An interesting comment here from the Office for Budget Responsibility:
In an interview with The Telegraph, Mr Miles conceded that the OBR's central forecasts published on Wednesday were “almost guaranteed to be wrong”, adding that the watchdog had faced accusations of being both “ridiculously optimistic” and a “bunch of pessimists”.
The problem here is not with the people - the Treasury. NIESR, the IFS, varied bank models and so on are all as bad - but really is in our stars. It is not possible to model something as complex as the economy to the sort of level of detail people are trying to.
This is of course Hayek but Hayek applied to macroeconomics rather than the details of microeconomic policy like trying to determine what the price of bananas should be.
It’s the implication of this fact that is so important. Given that the model is impossible then attempts at management by the model will not work either. The best that can be done is some vague and broadbrush idea - try not to collapse things, nor set off an inflationary spiral - and the rest of it has to be left to take care of itself.
That detailed demand management so beloved of Keynesians simply does not work. Not just for political reasons - those sunny days when deficits should be reduced never do seem to arrive - nor worries about the theoretical linkages of the models. But because we simply cannot produce a model reliable enough to allow us to make forecasts and thus tinker.
We’re actually rather left with the idea that macroeconomics as a management tool isn’t very useful. Which is fine by us of course. We’ve long said that if we get the nuts and bolts of incentives and price setting mechanisms - markets - right then the macroeconomy will be emergent from that. She’ll be right when she arrives cast in all her raiment too.
That is, it’s not so much don’t meddle as you’ve not got a clue what you’re doing so don’t meddle. Which, come to think of it it, is useful advice for so much of life, not just political control of the economy.
The Candle Makers are back
M. Bastiat and the Candle Makers are back in the news:
From lawsuits to IT hacks, the creative industries are deploying a range of tactics to protect their jobs and original work from automation
There are calls for changes to copyright in order to protect those jobs and incomes. Which is, from the societal point of view, entirely the wrong answer. For it is to fail to understand why we have copyright in the first place.
Innovation (which is covered by both patents and copyright) is difficult and therefore expensive to do. Once done it’s easy to copy. Therefore artistic works are, in an economic sense, public goods. The ease of copying means that they are non-rivalrous and non-excludable. We can all have a print of a Vetriani, a copy of Pratchett and hum along to a popular beat combo once that difficult work of creation has been done.
The public goods argument is that given this then less effort than we would like will be put into the original creation. Given how difficult it is to profit from having created less creation will happen. So, we deliberately construct these copyrights in order to make such things excludable and so enable artists to generate an income from the artifical scarcity.
There is no natural right here, this is a purely invented and for good enough reason attempt to solve the public goods problem.
Now we’ve the AIs able to copy out unlimited amounts of tosh. We have solved our supply problem. We do not need to protect the supplies of tosh in order for there to be supplies of it. We have solved the public goods problem not by placing restrictions upon supply but by entirely abolishing said restrictions upon supply.
Super. The public goods nature of the creation of what is a useful description of vast amounts of modern art, literature and so on - that tosh - is now solved. We can relax copyright therefore. Indeed, should. Restrictions which are past their sensibility date should be relaxed. Artists can go back to starving in their garrets and the rest of us have an unlimited amount of creativity to enjoy.
Protecting the jobs of artists was never and is not the point at all. Protecting the incentive to produce what we wish to consume was. Now that AI - as we’re being told is true by the very complaint we’re getting from the candle makers - automates and therefore makes free the creation of what we wish to consume then the correct response is to abolish copyright.
Fun how a proper appreciation of basic economics illuminate matters, no? Draws back the curtains on the correct policy response?
For the avoidance of doubt, this writer makes a living scribbling about financial markets. Exactly the area where such ‘bots and AIs have been making in roads - reporting company results and so on - this past decade. Yes, this is a welcome to my world argument.
Yes, of course productivity is flatlining
Just one tiny quote:
with productivity flatlining for 15 years
Vast amounts of deliberate policy are being deployed to insist that productivity flatlines. This being something that folk just aren’t getting - or being told.
This is not, by the way, some complaint about diversity officers who do nothing useful, council clipboard wielders who positively prevent anything from happening. These exist but we’re not on one of our dyspepsia kicks here, this is a simple matter of definitions and arithmetic.
Take the Green Party’s longstanding insistence:
“Figures from our Green MEPs show that investment in renewable energy would create far more jobs than nuclear power. These jobs would be far more secure in the future as they would come from many investment projects at many different scales, rather than the tiny number of huge deals involved in building new nuclear plants.
“Meeting our 2020 wind targets alone would create up to 130,000 jobs and the opportunities in solar power are massive, with more than 100,000 potential jobs installing solar panels across the UK. We are lagging behind our European neighbours already. In Germany there are 250,000 jobs in renewable energy but here we have 26,000 at best.
This has the unusual advantage, among Green Party policy insistences, of actually being true. Renewables produce more jobs per TWhr of electricity produced than fossil or nuclear. Therefore installing solar or wind instead of fossil or nuclear lowers productivity.
Productivity is defined - defined note - as output at market prices divided by labour hours to gain that output. So, a TWhr of electricity is worth whatever it is and if we now have more jobs producing that TWhr of electricity then labour productivity in electricity production has declined.
That’s simply a definitional truth.
Whole economy productivity is calculated by GDP divided by labour hours. That’s just what it is. So, we move to more labour heavy methods of producing GDP then productivity declines. That’s just maths.
As we’ve been known to say, jobs are a cost, not a benefit. But the point here is not to make some right wing, or free market, or we’rebeingpaidbythefossilfuelcapitaliststosayit point, rather this is all simply definitional.
To switch back the other way for a moment. Much of modern economic policy is concerned with externalities. Those CO2 emissions, those pollutions, the unfairnesses in life which are not well dealt with by pure market economies. Some of those dealings with we agree with, others we don’t. But again, that’s not our point here.
Externalities are, by definition, those things which do not appear in market prices. They do not appear in GDP therefore - which is calculated at market prices. So, the more we deal with externalities then the lower productivity will go.
For, by our standard and mutually agreed definition of productivity - again, labour hours in, output out at market prices - we are carrying the costs of dealing with those externalities in the labour hours put in without measuring the output from them in that market price measure of output.
Again, this is not a political point, it’s purely definitional.
The entire climate change problem is that standard economic accounting doesn’t include the costs of boiling Gaia. So, now we do the work - create those jobs in solar - to not boil Gaia. But standard economic accounting doesn’t include the benefits of not boiling Gaia but does the costs of not doing so. That’s just what an externality means.
So, as we address externalities then productivity declines. Not because we should not address externalities. Not because climate change is, or is not, some WEF cooked up nonsense. Or even, a catastrophic threat to the very existence of our species.
We are addressing externalities. We are, when we do so, measuring the labour costs of our doing so and not the output of our doing so. Therefore productivity declines as we address externalities.
QED.
Again, just to beat up on the dead equine again, this does not mean we should not address externalities. It illuminates precisely nothing in the debate about whether to have solar or nuclear.
What it does do is enlighten us as to one of the reasons why productivity numbers are flatlining. We’re addressing non-market costs - those externalities - so when we measure productivity by market output we’re not capturing the benefits of those actions while we are the costs.
This is just definitions and maths - not even maths, it’s arithmetic.
By addressing externalities we reduce measured productivity. Which makes the commonality between those shouting that we must address externalities and those whining about capitalism’s failures as productivity flatlines, well, it’s cakeism, isn’t it?
Obvious solution is obvious
We’re told of another environmental problem that we simply must do something about.
The world is facing an imminent water crisis, with demand expected to outstrip the supply of fresh water by 40% by the end of this decade, experts have said on the eve of a crucial UN water summit.
One of the joys here is that one part of the insistences about what must be done will actually solve the problem as defined:
Mariana Mazzucato, an economist and professor at University College London, also a lead author, added: “We need a much more proactive, and ambitious, common good approach. We have to put justice and equity at the centre of this, it’s not just a technological or finance problem.”
No, that’s not it. In fact that’s exactly wrong - such a surprise from that source. The answer in the actual report is entirely the opposite of a common good approach.
As with Garrett Hardin and the standard approach to commons tragedies, the answer is to stop treating the scarce resource as a commons. As The Guardian says:
…pricing water properly…
As the report says:
….We must cease underpricing water. Proper pricing…
That’s all that actually is required. As with those stories of California farmers using $300 worth of water to produce $200 worth of alfalfa proper pricing removes uneconomic uses of water entirely.
We must correct the misallocation of water via property rights
Proper pricing does exactly that. As we know, the initial allocation of property rights doesn’t matter as long as we have a subsequent free market. Assets will, through that market magic, end up in their highest value use. Job done.
They have a whole section:
We must cease underpricing water
Quite so.
That the report contains the answer is nice but of course that’s not what’s actually going to happen. Vide that blather about turning it into a common good.
The problem is that it is a common good. As Hardin pointed out, when Marxian access results in demand higher than possible supply then there will be shortages of that common good. The answer is to allocate property rights, institute a free market in the good and thereby make water a private, not a common, good. This is the answer to fishing problems (ITQs), copyrights and patents, to farmers and their fields. It is also the correct answer to water.
After all, when farmers can make more by supplying water to cities than by spraying it on fields then that’s what they’ll do. The problem is thus solved. For the entire analysis here is that farmers currently spray too much expensive water onto fields and not enough is left over to go to cities.
It is entirely true that not all commons problems are solved by private property rights and markets. Water is one of those that is.
Full expensing isn't - not really - a tax break, It's Keynesian stimulus
We have a certain problem here when one of the major newspapers of the nation fails to grasp the basics.
Jeremy Hunt has launched a flurry of tax breaks to encourage investment by businesses after the double blow of microchip designer Arm opting for a New York stock market listing and AstraZeneca deciding to build a new factory in Dublin.
Businesses that invest in IT equipment and machinery will be able to claim back the cost by writing it off against tax on their profits, the chancellor announced in his budget on Wednesday.
Businesses always have been able to write off the costs of investment against taxes upon their profits. Not only always have been able to do so there’s no possibility of a rational tax system upon profits in which they cannot.
Called “full expensing”, it will allow firms to claim up to 100% of the cost of the investment.
This has always been true and always will be.
The only thing that changes here is when can a business take that deduction. To fail to grasp this is to be ignorant of the subject under discussion. As one of us has noted elsewhere this is not the first (or even the second) time The Guardian has made this dreadful mistake.
A profits tax is levied upon profits. Upon that gap of the revenues from being in business against the costs of being in business. Buying a computer, a lorry, building a factory, is a cost of being in business. Therefore it is subtracted from revenues in order to arrive at the profit figure which is then taxed.
On the basis that politicians prefer money to spend now - while they’re still in office - we place limitations upon how much of those investment costs any business can claim this year. Whether we call them capital allowances, allowable depreciation or whatever doesn;t change the essence. We don’t allow a business to claim all of this year’s investment costs against this year’s profits.
Except, with full expensing, we do. And that’s it. All full expensing is is the removal of the limitations we place - time limitations, nothing more - upon a business being able to put the costs of being in business against the revenues of being in business.
The effect of this is to remove the delay on claiming the costs for the business. Equally, and obviously, it reduces profits tax this year. But over the life of the investment the tax paid on the profits is the same (with very minor issues over the interest on the cashflow). All that changes is which year that tax is paid.
Which allows us to show that this is in fact Keynesian demand stimulus. Government now delays - delays only - gaining that tax revenue from corporate profits. This, assuming steady spending, widens the budget deficit. That is the very definition of Keynesian stimulus, we’re just using the business investment channel to do it rather than handing out welfare cheques.
Business always is able to expense the full cost of investment against corporate profit taxes. All that “full expensing” means is that it’s able to do it today, not next year or next decade.
We can’t help but think that the world would improve if those who wrote about the tax system for major national newspapers had even the vaguest grasp of the tax system. Full expensing affects cashflow, not tax revenue over time.
Just to explain a detail about that lithium discovery in Iran
A certain amount of excitement as Iran claims to have found 8.5 million tonnes of lithium reserves. Apparently this will make Iran very rich and will also change global geopolitics.
No. We explained all of this, at length, some time ago. This is, in anything like a useful approximation of reality, trivially unimportant.
The claim:
Ebrahim Ali Molabeigi Iran’s minister of Industry announces “the discovery of the first lithium reserve estimated to be 8.5 million tonnes of lithium carbonate equivalent (LCE) in Hamedan province signalling positive news of the possibility of other reserves in the western Iranian region”.
Just no. If the claim is “estimated” then it’s not mineral reserves. Reserves are what has been proven, not what is estimated. Resources are the estimated, vague, stuff. We can even check this:
According to Iran’s industry officials, the lithium reserve, discovered in western Hamedan Province, holds up to 8.5 million tons of the element, and considering the world’s current discovered lithium reserves, Iran could hold up to 10 percent of the total global reserves.
The U.S. Geological Survey, which maps raw materials deposits across the planet, estimated the world’s total lithium reserves to stand at roughly 89 million tons.
No, USGS has global mineral resources at 89 million tonnes.
Owing to continuing exploration, identified lithium resources have increased substantially worldwide and total about 89 million tons.
Some will think that this is a trivial detail. But this isn’t - this is a definitional point of vast importance. Mineral resources are where from we think, to are pretty sure, we can dig up something of interest. Mineral reserves are where we have proven this to an economic and legal standard. It costs tens to hundreds of million of dollars to convert a resource into a reserve. Mineral reserves, that is, are a human made manufacture. They also equate, largely enough, to the current stock of extant or soon to be opened mines.
There is no, none, zip, connection between the amount of reserves and the amount of resources. Yes, contrary to all those claims from the environmentalists. There is no relationship at all between the volume of one and the volume of the other.
Finding a resource of lithium - in a clay that no one has proven, as yet, can be mined economically - is something of vague interest to aficionados of the outer edges of the mining industry. It’s not a fact of any geopolitical importance, nor really of any economic.
Definitions do matter. The error here is almost certainly one of translation. That the translator - perhaps even the Minister - does not know the difference between resource and reserve so the wrong word is used in the English language announcement.
Only if we understand the terms of art in a field can we work out what is being said in and by that field. Or even, report correctly what the field is saying. Finding lithium resources at this level of proof is not something of any great importance.
When technology changes then so does everything else
Well, yes, obvious but sometimes we do need the obvious pointed out:
On Wednesday, the bank revealed it was selling some of these investments at a loss to raise cash. Its shares promptly went into freefall.
At that point “Slack and WhatsApp groups lit up across the startup scene”, according to the Wall Street Journal. A mild panic turned into a stampede in the time it took clients to open their banking apps and transfer money with a few swipes of their finger.
By the close of play on Thursday, depositors had withdrawn $42bn from SVB. That night the Federal Deposit Insurance Corporation stepped in and seized the bank before it reopened on Friday morning.
Modern technology does indeed mean that information moves faster, that people can react to that information more quickly. This does indeed change things.
Time was when a bank might go bust and people 100 miles away would hear of the bank run days, even weeks, after it had happened. There’s a considerable literature insisting that 19th century American banking out west actually worked on this very basis. Northern Rock needed that film of depositors besieging offices to become the flood. Today, smartphones, banking apps, it all happens much faster.
OK. Now what?
Well, the answer is not very much to be honest. All banking is always a confidence trick in the sense that once the confidence is gone then so is the bank. That’s just fractional reserve banking for you.
We’re still there with the original question. Is the maturity transformation we gain from FRB worth having, even at the cost of runs? If it is then, well, there we are. If it isn’t then equally so. But that is the actual question. That modern technology enables them to happen faster is an interesting observation but not the nub of the matter at all.
A little challenge for a modern monetary theory
There’s an insistence out there that banks simply don’t lend out deposits. In one sense it’s even true. Banks lend money then look for the deposits to fund such lending. After all, capital plus deposits always does equal the loan book, as every bank balance sheet ever shows. Further, banks have to make sure those books balance at the close of the business day every business day.
That the funding comes after the loan - that’s what the bank’s treasury department does - doesn’t change the fact that the deposits fund the loans.
Therefore all this insistence that banks just create the money that they lend isn’t, in any useful sense, true.
We’ve now got a test of it. Silicon Valley Bank just went bust. There was a bank run. Deposits fleeing the bank that is. But if deposits fleeing the bank make it go bust then clearly deposits are important to a bank. They serve some important function, this must be so. And the function they serve is to fund the loan book. SIVB is bust because it can’t fund the loan book because the deposits went on that rat run.
Which is something of a challenge to this modern monetary theory that deposits don’t fund loans, banks just create the money by the act of lending. If it’s true that deposits don’t matter then Silicon Valley Bank isn’t bust. It is. Therefore deposits matter.
We would be interested to hear any attempts at explaining SIVB’s failure which conforms to that modern theory that banks don’t lend out deposits. But we do think that any such is going to run up against that ugly fact that SIVB is bust. As ever, when a theory is contradicted by reality it’s the universe that wins, not the hypothesis.