I'm just a humble labour & corpgov wonk, so most of the time the kinds of things that interest me are buried in the business pages. But it feels like, as with the financial crisis, questions about who controls businesses and how, and in whose interests, they should be run are going to become big political issues once more.
As I've made clear previously I have *zero* knowledge of Covid-19 beyond what anyone else can read. All I can speak with any sense about is how I see it impacting the areas that I do know about. So here are a few quick thoughts about the direction of travel.
First up, we should expect a very sharp turn away from normal expectations across the market. Dividends are already being suspended by many companies as they admit they can't accurately forecast what will happen in the months ahead. Buybacks are not going to have a good crisis, and many are being suspended. Some boards have already announced that directors are cutting their own pay, more will surely follow. At work we've been trying to get companies to embrace common sense on this.
Second, some companies will get it badly wrong, and this may do them serious damage. EasyJet is taking a lot of flak for proceeding with a £174m dividend payment even as it seeks state support (it claims it is compelled to make the payment, which is something I need to check out). Ryanair spent millions on a share buyback during March even as it cancelled hundreds of flights. One report I read yesterday said it is cutting staff pay by 50% (O'Leary is taking the same pay cut, but he doesn't really need the money does he?). Getting this stuff wrong in an environment where 'license to operate' is very much in the state's hands may have long-term implications. Directors will get booted out, companies' reputations will be in tatters. It could be terminal in some cases.
Third, there will be even more pressure for a shift to a more stakeholder-oriented governance model. This argument has gained a lot of ground over the past few years, so as the Covid-19 crisis hits it's one of the ideas that are "lying around", as Milton Friedman put it. You can see already in proposals from people in both the UK and the US that a change in the nature of businesses most directly affected is already being put forward. I can imagine a consensus forming very quickly that bailed-out companies must protect employment, cut executive pay (and no bonuses, LTIPs etc, obvs), stop dividends and buybacks and (maybe?) bring employee representatives into the boardroom. This would be a very clear shift away from operating in the interests of shareholders, even 'enlightened' shareholder value. And it would be very hard to unpick afterwards. What happens to bailed out companies may well start to affect those further away from the epicentre.
Fourth, I think (and hope) the crisis will lead to demands to change our employment model. All the 'flexibility' in zero hours contracts and gig work has been exposed as coming with huge downside risk for employees. But it also makes capitalism vulnerable. Workers without sick pay aren't going to self isolate. It's a lot of lower paid workers whose jobs can't be done from a laptop at home who going to keep things moving. In contrast many of us will reflect that we have bullshit jobs. Nothing is inevitable about a change, but I hope that, once we come out of the other side, unions build on some of the excellent work they have done so far and push for a new employment settlement.
Fifth, surely we're going to change our views on which organisations should and should not be privately owned. I have no idea where the line will end up being drawn but it won't be in the same place it is now.
Finally, the experience of a prolonged health crisis might serve to reset the discussion about pay. We're already seeing medical staff risking, and in some cases losing, their lives to protect us all. They, like most of us, do their job without any expectation of getting any bonus for it. So what makes executive directors so special? If the complex pay model we have for public companies - with all its targets, vesting dates, endless pages of reporting, and wasted investor research and engagement time - is part of a governance model that itself is out of date, let's junk it for good.
PS - https://www.youtube.com/watch?v=09LTT0xwdfw
Saturday, 21 March 2020
Saturday, 14 March 2020
In defence of the Behavioural Insights Team
I've seen something a bit odd on twitter over the past few days - people from various bits of the Left having a dig at the involvement of the Behavioural Insights Team (BIT) in the government's response to the Coronavirus outbreak. I've no dog in this fight, but what I've found particularly odd is the way this criticism has bled into a general attack on using 'behavioural science' to inform policy.
I know zero about epidemiology, so have nothing of value to say whatsoever about whether the BIT should be involved or not (that doesn't seem to stop a lot of people having an opinion though...). But I do bristle a bit when I see people on the Left attack the idea of looking at things like social psychology and behavioural economics when formulating policy. I also find it jarring when people who apparently have no science background describe this as 'pseudoscience'.
To be clear, there are important critiques of a lot of this kind of stuff. There is a big question mark over the replicability of some 'findings'. Stuart Ritchie has a decent thread on this topic here. And I'm happy to defer to people who have the expertise to say that there is no meaningful role for the BIT. But it's a big jump from this to the 'pseudoscience' sort of attack I see on Twitter, and apparent distaste for the BIT on the part of some people on the Left. I think this is a bit mistaken, so I thought I'd set out a few points in defence of this kind of work.
First, one line of attack is that the BIT is a gimmicky outfit that should have died with David Cameron's gimmicky government. But actually behavioural economics had an impact on policy even under Tony Blair. If you go back to the Pensions Commission's first report there's a section on what behavioural economics says that could inform savings design. This had a direct influence on decisions like auto-enrolment.
It also had an impact on the design of NEST. For example, one of the papers on Personal Accounts (what the NEST scheme was called initially) cited various bits of behavioural economics research in the discussion about default design and number of fund choices.
Now some people on the Left don't like auto-enrolment or NEST, so for such people this is irrelevant. But if you think it's a good policy (and one developed and initiated by a Labour government) then you should at least be aware of the role that behavioural economics played in framing it. This stuff is not 'Tory' or 'Cameroon'.
Secondly, on a related point, there seems to be a bit of confusion about what the big idea is. I think that's because there isn't one. FWIW from what I've seen/read I don't really think there is a 'theory' that sits behind the work of the BIT. If you read David Halpern's book on the work of the BIT it boils down to some pretty basic principles, rather than some overarching theory. This is a page from it where he summarises the key ones:
A lot of the work of the BIT seems to be applying these kinds of principles combined with randomised testing. Big wows, as I think the young people say. "Make it easy for people to do things you want them to do" doesn't strike me as a particularly ideological position. If there's a decent attack to be made on this, surely it's more that this is 'common sense' and actually probably just good sales and marketing tricks. I think I'd still take issue with that, but at least I'd feel it at least comes somewhere near the target, as opposed to suggestions that there is some malevolent right-wing anti-regulation agenda at play.
Another reasonable attack on this kind of work is perhaps that it seems to promise technocratic 'solutions' to political problems. I can remember someone making the point when Nudge came out that it was trying to take the politics out of politics, which does feel sorta kinda fair.
But I wonder if that's mixing up policy design/implementation with policy objectives a bit. Take that 'make things easy' point, it could be used to design policy interventions with wildly different political objectives. In terms of savings design it has influenced the decision to change the default from opt in to out out. But you can reverse the point - make things harder if you don't want people to do them. Which is what I thought the Conservatives might have been doing by removing payroll deduction of union subs. (I was wrong, or, at least, my FOI did not return anything!)
Which leads onto my third and main point - I don't really get why the Left would want to attack this kind of stuff anyway? It can be used for purposes we like or don't like, but we could say that about comms, or other elements of policy delivery. But some of it surely reinforces Left positions. I first came across behavioural finance when I started reading more about financial markets (Robert Shiller etc). And it was obvious to me then that this literature was problematic to those who argue that markets are efficient. Similarly behavioural economics suggests lots of ways in which markets do not operate in the way that stylised accounts of them suggest. Going further, it was reading around this kind of stuff that lead me to critiques of performance-related pay - something a lot of people on the Left are uncomfortable with.
This is not to say that behavioural finance / economics is 'left wing'. Richard Thaler and Cass Sunstein used the godawful term 'libertarian paternalism' to describe their approach (though again that tells us something, as they were attacked for using evidence of market inefficiencies as a way to propose paternalistic policies). But I don't see the value in not looking at what we can learn from this stuff.
To establish that when X happens people do Y, and not Z, does not mean you must adopt a certain policy, or that you should rely on non-regulatory interventions. All that it means is that if you have a certain policy objective it's useful to take account of this stuff to design your interventions. I find that basically uncontroversial. It doesn't mean don't look at other information to form decisions, or that you must reject certain types of intervention.
I am sure that some of the interventions that are 'behaviourally informed' will fail, and that some of the 'findings' that underpin them will be disproved. That's all fine. When the facts change and all that. Equally some of it will prove to be robust and simply become 'economics', or whatever else. Some people seem to have forgotten how much has already entered the mainstream. (Let she who is without a copy of Thinking Fast And Slow cast the first diss). I am also sure that people will be able to find things that David Halpern and/or the BIT have said/done that people on the Left (including me) would find objectionable.
I'd just urge some caution. What are we really arguing with? If it's actually the policy objective I think we're missing the point. If it's the science I'm happy to let the academics fight it out and tell me what I should think once they are done. I just fear a baby/bathwater outcome.
I know zero about epidemiology, so have nothing of value to say whatsoever about whether the BIT should be involved or not (that doesn't seem to stop a lot of people having an opinion though...). But I do bristle a bit when I see people on the Left attack the idea of looking at things like social psychology and behavioural economics when formulating policy. I also find it jarring when people who apparently have no science background describe this as 'pseudoscience'.
To be clear, there are important critiques of a lot of this kind of stuff. There is a big question mark over the replicability of some 'findings'. Stuart Ritchie has a decent thread on this topic here. And I'm happy to defer to people who have the expertise to say that there is no meaningful role for the BIT. But it's a big jump from this to the 'pseudoscience' sort of attack I see on Twitter, and apparent distaste for the BIT on the part of some people on the Left. I think this is a bit mistaken, so I thought I'd set out a few points in defence of this kind of work.
First, one line of attack is that the BIT is a gimmicky outfit that should have died with David Cameron's gimmicky government. But actually behavioural economics had an impact on policy even under Tony Blair. If you go back to the Pensions Commission's first report there's a section on what behavioural economics says that could inform savings design. This had a direct influence on decisions like auto-enrolment.
It also had an impact on the design of NEST. For example, one of the papers on Personal Accounts (what the NEST scheme was called initially) cited various bits of behavioural economics research in the discussion about default design and number of fund choices.
Now some people on the Left don't like auto-enrolment or NEST, so for such people this is irrelevant. But if you think it's a good policy (and one developed and initiated by a Labour government) then you should at least be aware of the role that behavioural economics played in framing it. This stuff is not 'Tory' or 'Cameroon'.
Secondly, on a related point, there seems to be a bit of confusion about what the big idea is. I think that's because there isn't one. FWIW from what I've seen/read I don't really think there is a 'theory' that sits behind the work of the BIT. If you read David Halpern's book on the work of the BIT it boils down to some pretty basic principles, rather than some overarching theory. This is a page from it where he summarises the key ones:
A lot of the work of the BIT seems to be applying these kinds of principles combined with randomised testing. Big wows, as I think the young people say. "Make it easy for people to do things you want them to do" doesn't strike me as a particularly ideological position. If there's a decent attack to be made on this, surely it's more that this is 'common sense' and actually probably just good sales and marketing tricks. I think I'd still take issue with that, but at least I'd feel it at least comes somewhere near the target, as opposed to suggestions that there is some malevolent right-wing anti-regulation agenda at play.
Another reasonable attack on this kind of work is perhaps that it seems to promise technocratic 'solutions' to political problems. I can remember someone making the point when Nudge came out that it was trying to take the politics out of politics, which does feel sorta kinda fair.
But I wonder if that's mixing up policy design/implementation with policy objectives a bit. Take that 'make things easy' point, it could be used to design policy interventions with wildly different political objectives. In terms of savings design it has influenced the decision to change the default from opt in to out out. But you can reverse the point - make things harder if you don't want people to do them. Which is what I thought the Conservatives might have been doing by removing payroll deduction of union subs. (I was wrong, or, at least, my FOI did not return anything!)
Which leads onto my third and main point - I don't really get why the Left would want to attack this kind of stuff anyway? It can be used for purposes we like or don't like, but we could say that about comms, or other elements of policy delivery. But some of it surely reinforces Left positions. I first came across behavioural finance when I started reading more about financial markets (Robert Shiller etc). And it was obvious to me then that this literature was problematic to those who argue that markets are efficient. Similarly behavioural economics suggests lots of ways in which markets do not operate in the way that stylised accounts of them suggest. Going further, it was reading around this kind of stuff that lead me to critiques of performance-related pay - something a lot of people on the Left are uncomfortable with.
This is not to say that behavioural finance / economics is 'left wing'. Richard Thaler and Cass Sunstein used the godawful term 'libertarian paternalism' to describe their approach (though again that tells us something, as they were attacked for using evidence of market inefficiencies as a way to propose paternalistic policies). But I don't see the value in not looking at what we can learn from this stuff.
To establish that when X happens people do Y, and not Z, does not mean you must adopt a certain policy, or that you should rely on non-regulatory interventions. All that it means is that if you have a certain policy objective it's useful to take account of this stuff to design your interventions. I find that basically uncontroversial. It doesn't mean don't look at other information to form decisions, or that you must reject certain types of intervention.
I am sure that some of the interventions that are 'behaviourally informed' will fail, and that some of the 'findings' that underpin them will be disproved. That's all fine. When the facts change and all that. Equally some of it will prove to be robust and simply become 'economics', or whatever else. Some people seem to have forgotten how much has already entered the mainstream. (Let she who is without a copy of Thinking Fast And Slow cast the first diss). I am also sure that people will be able to find things that David Halpern and/or the BIT have said/done that people on the Left (including me) would find objectionable.
I'd just urge some caution. What are we really arguing with? If it's actually the policy objective I think we're missing the point. If it's the science I'm happy to let the academics fight it out and tell me what I should think once they are done. I just fear a baby/bathwater outcome.
Sunday, 8 March 2020
Fluttery votes
Here's the list of Paddy Power Betfair (now Flutter) major shareholders disclosed in last year's annual report. Capital Group is reported as having 5.9%, or 4.6m shares at 5 March 2019. Of these almost 3.9% are accounted for by the EuroPacific fund.
Here's an excerpt from the EuroPacific fund's NPX form that covered the 2019 AGM. The only resolution it said it voted against was resolution 2, to approve the dividend.
And here's an excerpt from Paddy Power Betfair's AGM voting results for the May 2019 meeting showing the vote against the resolution to approve the dividend.
The small print in the annual report says that the EuroPacific Fund delegates voting rights to Capital Research and Management. Does that mean that it voted For the dividend resolution, or didn't vote? Either way surely the NPX disclosure is pretty meaningless.
Here's an excerpt from the EuroPacific fund's NPX form that covered the 2019 AGM. The only resolution it said it voted against was resolution 2, to approve the dividend.
And here's an excerpt from Paddy Power Betfair's AGM voting results for the May 2019 meeting showing the vote against the resolution to approve the dividend.
The small print in the annual report says that the EuroPacific Fund delegates voting rights to Capital Research and Management. Does that mean that it voted For the dividend resolution, or didn't vote? Either way surely the NPX disclosure is pretty meaningless.
Against transparency!
The more time I spend looking at disclosures from investors, the less sense a lot of it makes to me. I've blogged quite a bit about meaningless Stewardship Code reporting, with dozens of hedge / boutique funds using exactly the same blurb. We are about to be hit by a wave of SRDII reporting that I think will repeat the same pattern.
I really question whether there is any point to all this reporting. I doubt most of these funds will ever aspire to be active 'stewards' of companies, and that's fine, most people that invest with them understand that. I am also sceptical that anyone reads the stewardship blurbs produced by managers or uses them to base asset allocation decisions on.
There's obviously a mini industry in producing compliance statements, which is probably going to matched over time by people within consulting or auditing firms who will assess them. So to the extent that there is interaction between people over the content of these disclosures it will be between people who write blurb and people who read blurb. It will not be between people who have capital to invest and people who are paid to invest capital, which would seem to be what we were all aiming at when the Stewardship Code came in ten years ago.
So if a lot of disclosure is meaningless, it goes largely unread, the only people who pay attention to it are compliance people (and compliance service providers) and consultants, and the firms who are making the disclosures are unlikely to change anyway, what's the point? It just seems to be imposing a cost.
I really question whether there is any point to all this reporting. I doubt most of these funds will ever aspire to be active 'stewards' of companies, and that's fine, most people that invest with them understand that. I am also sceptical that anyone reads the stewardship blurbs produced by managers or uses them to base asset allocation decisions on.
There's obviously a mini industry in producing compliance statements, which is probably going to matched over time by people within consulting or auditing firms who will assess them. So to the extent that there is interaction between people over the content of these disclosures it will be between people who write blurb and people who read blurb. It will not be between people who have capital to invest and people who are paid to invest capital, which would seem to be what we were all aiming at when the Stewardship Code came in ten years ago.
So if a lot of disclosure is meaningless, it goes largely unread, the only people who pay attention to it are compliance people (and compliance service providers) and consultants, and the firms who are making the disclosures are unlikely to change anyway, what's the point? It just seems to be imposing a cost.
Saturday, 7 March 2020
NMC voting turnout
Right... here's what we can see in the NMC meeting results. The first graph just sets out what NMC reports. So there is a headline total voting turnout for all meeting and at AGMs from 2015 I can also split out the insider and minority shareholder turnout figures because of the requirement to report the votes on independent NEDs twice.
In the second graph I've put in an assumed minority shareholder turnout of around 81% as this was the level it was at later on. This applies at three meetings - 2013 and 2014 AGMs, and the 2016 EGM. If I don't assume roughly the same turnout and instead assume 100% turnout of the insider bloc, then the minority shareholder turnout at the 2014 AGM would be 65%., which just didn't feel right.
UPDATE: I've added a third graph which is raw votes (rather than %) just in the years where insider votes are identifiable.
There are a few notable points. The 2016 AGM result look a bit odd. The big drop in overall turnout and insider turnout is largely Shetty not voting his shares, and this is explained in the AGM results notice. (Shetty held 47.7m at this point according to the annual report)
The very high, and unrepeated, level of minority shareholder voting is surprising. If my numbers are right it's 58.9m out 61.4m shares being voted, so only 2.5m(ish) not voted. By comparison at the previous AGM it looks 11m shares from the same free float were not voted.
Secondly, investors really don't like this company's approach remuneration. At that 2016 AGM 41.4m votes were cast against the remuneration report, so 70% of the minority shareholder vote. If we go to the 2016 EGM, 56m votes were cast against the rem policy. If we assume an 81% turnout (with more shares in issue) I reckon the vote against was around 85% of the free float. And it was over 80% against the remuneration report at the 2017 AGM too.
Thirdly, the 2013 AGM is quite interesting as it is obviously the controlling shareholders who voted the SID off the board (the vote against him 128m, shares held by controlling party disclosed in the annual report 124m) and on the same day it appointed the former EY person who became the SID.
Finally, it's terminal turnout time again. Voting levels drop as the shorting builds up.
In the second graph I've put in an assumed minority shareholder turnout of around 81% as this was the level it was at later on. This applies at three meetings - 2013 and 2014 AGMs, and the 2016 EGM. If I don't assume roughly the same turnout and instead assume 100% turnout of the insider bloc, then the minority shareholder turnout at the 2014 AGM would be 65%., which just didn't feel right.
UPDATE: I've added a third graph which is raw votes (rather than %) just in the years where insider votes are identifiable.
There are a few notable points. The 2016 AGM result look a bit odd. The big drop in overall turnout and insider turnout is largely Shetty not voting his shares, and this is explained in the AGM results notice. (Shetty held 47.7m at this point according to the annual report)
The very high, and unrepeated, level of minority shareholder voting is surprising. If my numbers are right it's 58.9m out 61.4m shares being voted, so only 2.5m(ish) not voted. By comparison at the previous AGM it looks 11m shares from the same free float were not voted.
Secondly, investors really don't like this company's approach remuneration. At that 2016 AGM 41.4m votes were cast against the remuneration report, so 70% of the minority shareholder vote. If we go to the 2016 EGM, 56m votes were cast against the rem policy. If we assume an 81% turnout (with more shares in issue) I reckon the vote against was around 85% of the free float. And it was over 80% against the remuneration report at the 2017 AGM too.
Thirdly, the 2013 AGM is quite interesting as it is obviously the controlling shareholders who voted the SID off the board (the vote against him 128m, shares held by controlling party disclosed in the annual report 124m) and on the same day it appointed the former EY person who became the SID.
Finally, it's terminal turnout time again. Voting levels drop as the shorting builds up.
How seriously do hedge funds voting disclosure?
Not seriously, I would argue. The principle that investors should disclose their voting records is pretty well established now, but we still get this cut & paste stuff justify not doing so.
Wednesday, 4 March 2020
More NMC stuff
A quick look at disclosures for one of NMC's largest shareholders, Wellington. Last Friday there was an RNS that announced that it held a bit over 10m shares, but in a note said it didn't have voting discretion for 1.35m of them (13.5%).
Then on Monday there was an RNS that replaced the one issued on Friday that announced that Wellington held 10m shares but the not now says it didn't have voting discretion for 8.7m of them (86.5%).
What happened there then? Who voted the other 8.7m shares (over 4% of those outstanding), or did no-one? Did they only find out on Monday? Not insignificant given that Wellington looks to have been the third largest external shareholder after Capital and Hermes (though part of its holding relates to a Vanguard fund it manages).
Then on Monday there was an RNS that replaced the one issued on Friday that announced that Wellington held 10m shares but the not now says it didn't have voting discretion for 8.7m of them (86.5%).
What happened there then? Who voted the other 8.7m shares (over 4% of those outstanding), or did no-one? Did they only find out on Monday? Not insignificant given that Wellington looks to have been the third largest external shareholder after Capital and Hermes (though part of its holding relates to a Vanguard fund it manages).
Monday, 2 March 2020
NMC Health voting rights
I spent a bit of Sunday trawling through NMC annual reports and RNS announcements. If you look at the disclosed shareholders in the annual reports you can figure out the holdings of the controlling shareholders. If you then look at the AGM results you can see both the total ISC and how much was voted. And because of the requirement to report the votes on independent NEDs both with and without controlling shareholder votes you can figure out what number the company put on controlled votes at the time of the AGM.
Looking at the 2015 AGM the numbers are exactly the same - combined voting rights disclosed in the annual report matched all votes cast minus minority shareholder votes cast. Looking at the 2017 AGM it's basically the same too. At the 2016 AGM it looks like 47.5m insider shares weren't voted, which is roughly B R Shetty's disclosed shareholding in the annual report for that year. Not sure why that might have happened. (Incidentally my initial calculation, which I'll look at again, suggests the reported turnout from the free float at the 2016 AGM was 96%, which seems high).
At the 2018 and 2019 AGM results it looks like 6.7m and 6.8m insider shares weren't voted at the respective meetings. This may be related to the fact that Shetty pledged 7m shares (but not 6.7m or 6.8m) to Goldman Sachs as part of a financing deal, about which NMC has provided more information today.
Looking at the 2015 AGM the numbers are exactly the same - combined voting rights disclosed in the annual report matched all votes cast minus minority shareholder votes cast. Looking at the 2017 AGM it's basically the same too. At the 2016 AGM it looks like 47.5m insider shares weren't voted, which is roughly B R Shetty's disclosed shareholding in the annual report for that year. Not sure why that might have happened. (Incidentally my initial calculation, which I'll look at again, suggests the reported turnout from the free float at the 2016 AGM was 96%, which seems high).
At the 2018 and 2019 AGM results it looks like 6.7m and 6.8m insider shares weren't voted at the respective meetings. This may be related to the fact that Shetty pledged 7m shares (but not 6.7m or 6.8m) to Goldman Sachs as part of a financing deal, about which NMC has provided more information today.
Sunday, 1 March 2020
Star Trek: The Paper Clip Maximiser
One of the books I really enjoyed recently was The AI Does Not Hate You by Tom Chivers. It's sorta more about the people who think about AI, and how they think about it, and it was basically right up my street.
There's a bit early on where he describes of the early thought experiments relating to AI risks, which is known as the paperclip maximiser. Here's the explanation in the book:
"Imagine a human-level AI has been given an apparently harmless instruction: to make paperclips. What might it do? Well, it might start out by dimly making paperclips. It could build a small pressing machine and churn out a few dozen paperclips a minute. But it's bright enough to know that it could be more efficient than that, and if it wants to maximise the number of paperclips it can make, it's probably better not to go straight for a small press. It could instead use its materials to make a larger factory, so that it's making thousands of paperclips a minute. Still, though, if really wants to make as many paperclips as possible, it might want to improve its ability to think about how to do so, so it might want to spend some of its resources building new processors, upgrading its RAM and so on.
"You can see where this is going, presumably. The end point of the paperclip maximiser is a solar system in which every single atom has been turned into either paperclips, paperclip-manufactoring machines, computers that think about how best to manufacture paperclips. or self-replicating space probes that are hurtling out towards Proxima Centauri at a respectable fraction of the speed of light with instructions to set up a franchise there."
This is a great little explainer, which apparently originates with Eliezer Yudkowsky (by the way whose book Inadequate Equilibria is also worth a read and I may blog about later). But it also reminded me of Star Trek The Motion Picture (the first Star Trek movie).
In that film there is a giant entity, which we later find out is called V'Ger, spreading through the universe, destroying a lot of stuff in its path, including Federation spaceships! Here's how Wikipedia describes the way the film ends:
"At the center of the massive ship, V'Ger is revealed to be Voyager 6, a 20th-century Earth space probe believed lost in a black hole. The damaged probe was found by an alien race of living machines that interpreted its programming as instructions to learn all that can be learned and return that information to its creator. The machines upgraded the probe to fulfill its mission, and on its journey, the probe gathered so much knowledge that it achieved sentience. Spock realizes that V'Ger lacks the ability to give itself a purpose other than its original mission; having learned what it could on its journey home, it finds its existence meaningless."
OK, so it's not quite the same thing. An important point in Tom Chivers' book is that AI risk really isn't about machines achieving sentience. However there's a big similarity, the goal - go and find out all that you can and transmit it back - gets interpreted in a very broad way, the machine is unable to go beyond the goal, and the goal ends up being achieved in a way that leads to humans getting killed.
There's a bit early on where he describes of the early thought experiments relating to AI risks, which is known as the paperclip maximiser. Here's the explanation in the book:
"Imagine a human-level AI has been given an apparently harmless instruction: to make paperclips. What might it do? Well, it might start out by dimly making paperclips. It could build a small pressing machine and churn out a few dozen paperclips a minute. But it's bright enough to know that it could be more efficient than that, and if it wants to maximise the number of paperclips it can make, it's probably better not to go straight for a small press. It could instead use its materials to make a larger factory, so that it's making thousands of paperclips a minute. Still, though, if really wants to make as many paperclips as possible, it might want to improve its ability to think about how to do so, so it might want to spend some of its resources building new processors, upgrading its RAM and so on.
"You can see where this is going, presumably. The end point of the paperclip maximiser is a solar system in which every single atom has been turned into either paperclips, paperclip-manufactoring machines, computers that think about how best to manufacture paperclips. or self-replicating space probes that are hurtling out towards Proxima Centauri at a respectable fraction of the speed of light with instructions to set up a franchise there."
This is a great little explainer, which apparently originates with Eliezer Yudkowsky (by the way whose book Inadequate Equilibria is also worth a read and I may blog about later). But it also reminded me of Star Trek The Motion Picture (the first Star Trek movie).
In that film there is a giant entity, which we later find out is called V'Ger, spreading through the universe, destroying a lot of stuff in its path, including Federation spaceships! Here's how Wikipedia describes the way the film ends:
"At the center of the massive ship, V'Ger is revealed to be Voyager 6, a 20th-century Earth space probe believed lost in a black hole. The damaged probe was found by an alien race of living machines that interpreted its programming as instructions to learn all that can be learned and return that information to its creator. The machines upgraded the probe to fulfill its mission, and on its journey, the probe gathered so much knowledge that it achieved sentience. Spock realizes that V'Ger lacks the ability to give itself a purpose other than its original mission; having learned what it could on its journey home, it finds its existence meaningless."
OK, so it's not quite the same thing. An important point in Tom Chivers' book is that AI risk really isn't about machines achieving sentience. However there's a big similarity, the goal - go and find out all that you can and transmit it back - gets interpreted in a very broad way, the machine is unable to go beyond the goal, and the goal ends up being achieved in a way that leads to humans getting killed.
TR1 tales at NMC
A few more funky bits and pieces in the entrails of NMC Health's filings...
A couple of banks facilitating stuff for clients presumably. In the Morgan Stanley case it's an equity swap. Quite interested in who would want an NMC equity swap at this point, but again it's possible that the disclosures have lagged. The TR1 showing Morgan Stanley linked positions going below the disclosure threshold was issued to the market on 27 Feb, whereas the TR1 disclosing the previous notifiable holding was issued the day after.
24 Feb (but issued on 28 Feb)
Morgan Stanley - under 5% to 5.12% (3.94% not shares)
https://www.investegate.co.uk/nmc-health-plc--nmc-/rns/holdings-in-company/202002281540585546E/
25 Feb (issued on 27 Feb)
Morgan Stanley - 5.12% (3.94% of which not shares) to under 3%
https://www.investegate.co.uk/nmc-health-plc--nmc-/rns/holding-s--in-company/202002271636474038E/
Here's another bank in the mix: Goldman Sachs, and it's a mixture of put and call options, swaps and CFDs. At one point in early January the total interest in NMC shares goes from below 1% to almost 13%, of which less than 2% represented actual shares themselves, but that represents over a quarter of the free float. There are a bunch more Goldman TR1s that I'll look at later.
8 Jan
Goldman Sachs - 12.94% (11.17% not shares)
https://www.investegate.co.uk/nmc-health-plc--nmc-/rns/holdings-in-company/202001151000018746Z/
That stuff drifts down again to just under 9% before disappearing under 1% again on 17 January. So this all happened within 10 days.
https://www.investegate.co.uk/nmc-health-plc--nmc-/rns/holdings-in-company/202001211716315043A/
Finally, you can see a bit of stock lending activity out there too. For example, TR1s issued by Norges Bank show it winding down the amount of stock on loan (plus a jump in overall holding) from the back end of 2019. I've stuck a few of these into a chart.
Blackrock also issued a bunch of TR1s as its position went up and down, and there is a bit of disclosed stock lending in there. Here's a disclosure showing 1.6% lending near end November 2019.
https://www.investegate.co.uk/nmc-health-plc--nmc-/rns/holding-s--in-company/201911281100020114V/
According to ShortTracker, the total public short in NMC hit about 6% at max, and was at 5% at the time of the meeting in December that approved the company's buyback and its remuneration policy. Just the Norges and Blackrock disclosures show 3.75% between them out on loan around that time.
Incidentally, the voting turnout for that meeting was reported as 85% - with almost 178m out of 208m shares voted. Turnout at the AGM last June was actually a touch higher - 87% and 182m shares - that strikes me as pretty good in general and therefore not one of the cases I've found previously of shorted stocks seeing falling turnout.
A couple of banks facilitating stuff for clients presumably. In the Morgan Stanley case it's an equity swap. Quite interested in who would want an NMC equity swap at this point, but again it's possible that the disclosures have lagged. The TR1 showing Morgan Stanley linked positions going below the disclosure threshold was issued to the market on 27 Feb, whereas the TR1 disclosing the previous notifiable holding was issued the day after.
24 Feb (but issued on 28 Feb)
Morgan Stanley - under 5% to 5.12% (3.94% not shares)
https://www.investegate.co.uk/nmc-health-plc--nmc-/rns/holdings-in-company/202002281540585546E/
25 Feb (issued on 27 Feb)
Morgan Stanley - 5.12% (3.94% of which not shares) to under 3%
https://www.investegate.co.uk/nmc-health-plc--nmc-/rns/holding-s--in-company/202002271636474038E/
Here's another bank in the mix: Goldman Sachs, and it's a mixture of put and call options, swaps and CFDs. At one point in early January the total interest in NMC shares goes from below 1% to almost 13%, of which less than 2% represented actual shares themselves, but that represents over a quarter of the free float. There are a bunch more Goldman TR1s that I'll look at later.
8 Jan
Goldman Sachs - 12.94% (11.17% not shares)
https://www.investegate.co.uk/nmc-health-plc--nmc-/rns/holdings-in-company/202001151000018746Z/
That stuff drifts down again to just under 9% before disappearing under 1% again on 17 January. So this all happened within 10 days.
https://www.investegate.co.uk/nmc-health-plc--nmc-/rns/holdings-in-company/202001211716315043A/
Finally, you can see a bit of stock lending activity out there too. For example, TR1s issued by Norges Bank show it winding down the amount of stock on loan (plus a jump in overall holding) from the back end of 2019. I've stuck a few of these into a chart.
Blackrock also issued a bunch of TR1s as its position went up and down, and there is a bit of disclosed stock lending in there. Here's a disclosure showing 1.6% lending near end November 2019.
https://www.investegate.co.uk/nmc-health-plc--nmc-/rns/holding-s--in-company/201911281100020114V/
According to ShortTracker, the total public short in NMC hit about 6% at max, and was at 5% at the time of the meeting in December that approved the company's buyback and its remuneration policy. Just the Norges and Blackrock disclosures show 3.75% between them out on loan around that time.
Incidentally, the voting turnout for that meeting was reported as 85% - with almost 178m out of 208m shares voted. Turnout at the AGM last June was actually a touch higher - 87% and 182m shares - that strikes me as pretty good in general and therefore not one of the cases I've found previously of shorted stocks seeing falling turnout.
Subscribe to:
Posts (Atom)