TONY NASH: Good afternoon, good morning. This is Tony Nash. I'm the CEO of Complete Intelligence, and I'm guest hosting this "Real Vision Live" talk with Grant Wilson, the head of Asia-Pacific at Exante Data.
Today we're talking about coronavirus and the impact specifically on China and the Chinese economy, but also other parts of the world as well. I'm coming to you from Houston, Texas, and Grant is in Sydney, Australia.
Good morning, Grant.
GRANT WILSON: Hi, Tony, yes, bright and early here.
TONY NASH: Great. Grant, can you tell me a little bit about Exante, and then I'll talk a little bit about Complete Intelligence?
GRANT WILSON: Sure, yeah, Exante is based in New York founded by Jens Nordvig 4 or 5 years ago, macro advisory with a very heavy emphasis on data analytics, full spectrum of data analytics. Obviously at the moment, we're looking very closely at what's happening with coronavirus, but how that's playing through into economic statistics, real-time statistics, open-source statistics, capital-flow statistics, economic statistics, and so on and so forth.
So a heavy emphasis on data, but perhaps just at the top, Tony, I think given what we're about to talk about, let's not lose sight that beneath that data, there are real people, and the situation on the ground at the moment in China is obviously deeply concerning.
TONY NASH: That's right. That's right. Just a quick bit on Complete Intelligence. We're an artificial intelligence platform. We use publicly available data, enterprise data, and market data to help people make decisions about markets, costs, and revenues.
So let's just jump right into it, Grant. You know, with the coronavirus, you know, we hear a lot about it, and I think a lot of people assume they know what it is, but can you just give us a very quick overview of what this is, and, also, what is the data telling you? Why is it different this time?
GRANT WILSON: Yeah, sure. Again, there's been so much information. So one of the words that sort of come out recently is it's an infodemic, and it's fair. You know, there's a flood of information at the moment. It's also true that, like, news volumes have spiked on coronavirus, and by the way, it's now being referred to as COVID. This is a new development overnight. So maybe we run with that today.
But, yeah, there's been an overwhelming amount of media attention in the past couple of weeks. I think that's conditioned by the recent heritage. Coronavirus is a family, if you will, of viruses. It has a sequence, in this particular case, which resembles some things we've seen before, most notably SARS, 2003, and MERS, you know, 5, 6 years later, which was not so much in the news because it was sort of very localized within the Middle East, but actually had a much higher fatality rate than the either SARS or COVID today.
TONY NASH: So, as you know, we've all got this information overflow with COVID or way too much. So how do you tell good data from bad, right?
GRANT WILSON: Yeah.
TONY NASH: Where every amateur is saying, oh, you can't trust Chinese data, but at some level, we have to at least trust it directionally, I would think. So how do you tell good data from bad and what data do you use?
GRANT WILSON: Yeah, it's the right question, obviously, particularly for us as a data-driven advisory firm. I think the starting point would be let's think of two sources, two primary sources at the moment. The first is the Chinese statistics, which is being published by the National Health Commission, on time, pretty much every day around 8:00 or 9:00 PM, New York.
They're collecting the data from the provinces, aggregating that, and then publishing in Chinese a series of texts, which can be translated and then backed into a spreadsheet, and you can do pretty much what you want with it. So that's the sort of primary data source. We can talk about the veracity of that in a moment, if you like.
But it's true. The market has taken some guidance from that. That data set will form the historical record of this outbreak. So let's not lose sight of that.
A second set of data, which we think has credibility, at least some level, is the academic sort of virus modeling. So there's a lot of effort, and academic effort, that's gone into this branch of modeling over the last 20 years or even previous to that. So these are very sophisticated mathematical models, which try to give you a sense of the core characteristics of COVID in this instance.
So the most early stage estimates started to come through mid-January, so Jan 17, Jan 20, some schools, Hong Kong University, Imperial College of London, started publishing papers, and these papers are peer reviewed, OK? So they're not just people sitting around with a spreadsheet with a stochastic model trying to figure out how big the outbreak is. There's a lot of discipline that goes into the publishing process before you're actually allowed to print something.
So there's probably 5 to 10 of these models that have, you know, into the public domain, and, obviously, the issue here is that there's a disjuncture, you know, between the historical record and what these models are suggesting, and these models are pretty much based upon the number of infections that we saw globally at a given date, where they occurred, and then using sort of traffic maps, particularly air traffic, to try to estimate how big the outbreak was, how fast it might be spreading, on a pre-intervention basis.
And then at the back end of that, now, we've got a public health intervention of historic magnitude trying to re-estimate the parameters and build in some form of estimate of what we've got. Now these models have got very, very wide confidence intervals. That's part of the discipline, but what's telling is that some-- to the extent that they were published early, and now to the extent that they've been updated, there is a disjuncture, if you will, between the official record and the academic community.
TONY NASH: OK, so given that disjuncture, we saw earlier this week that there was a decline in the number of suspected cases, and it wasn't a small decline. It was really notable. And you guys covered this in detail. So can you talk about that a little bit? Is that decline, I guess in your mind, kind of real, or as real as it gets? And what's behind that? You know, better testing, is it data manipulation, a little bit of both?
GRANT WILSON: So it was a very pronounced decline in the last two daily sets. The stock of suspected cases fell from about 27,000 down to about 21,000. And it was the first time I've actually seen a decline in any of these metrics. So, again, you know, the market brings things into that. There's been some commentary.
There was some local commentary, which we picked up because we're obviously collating a lot of resources, both in English media, but also in Chinese media, and it suggested that, effectively, they found a way, just in Hubei, to fast-track the diagnostic procedures. We think it was probably like a two or three day process, maybe even longer than that at inception, but as, you know, the medical procedures became a little bit more familiar and the scale of the issue became a little bit clearer, they've changed protocols in a few different ways, and we think this was one of those instances.
It seems to be just Hubei. And, again, perhaps there's some debate around whether this should be done, whether it's accurate, but it does, I think, explain the change in the stock of confirmed cases, because they're able to process people and regard them as no longer suspected. Interestingly though, there's another way to measure suspected cases, and that's based on the flow.
So the number of new people each day, which are suspected, is also released by NHC, and it's broken down into Hubei, and then all other provinces, and that measure itself has not moved very much. That's been between 3,500 and 5,000 people a day. So while the market sort of caught on to this stock measure coming over, the actual number of new suspected cases every day is still pretty-- it's come off a little bit, but it's still, yeah, tracking relatively high.
TONY NASH: OK, so but if you're saying 3,000 to 5,000 a day, and that's fairly constant, then what is the rate of transmission? Is it kind of 1 to 1 or 1 to 2 or 1 to 5, or, you know, how is that changing over time?
GRANT WILSON: Yeah, this is where we go back to the academic sort of debate, and without unpacking it too far, the way to think about these outbreaks is ultimate in terms of what's known as a reproductive number, and this is referred to R0, R-nought, OK? And in very simple terms, what we're trying to figure out here is if you're infected, Tony, and you're doing your average day to day, and you're in touch with average day-to-day people that had not been exposed to a viral outbreak before. They are not immune. They haven't been vaccinated. On average, how many people are you going to infect, OK?
Now what's interesting here is that, again, these are wide estimates, but it looks as though COVID is now going to come in perhaps 2.5 to maybe 3.5 people, OK? This is, again, just ball parking just to help your listeners understand the basic framework and the terminology.
SARS was a bit lower than that, right? So, you know, not as productive in its native state. Again, even after the fact, we don't know exactly where that was because the data at the time was even worse than what we've potentially got now, right?
But SARS is generally, ex post, thought to be somewhere closer to 2 and 2.5, OK?
TONY NASH: OK.
GRANT WILSON: So what's interesting here is that if you think about these two outbreaks as comparable, you know, we obviously understand, I think by now, that COVID does look different in terms of the outbreak, but it looks sort of dramatically different, even if we take the Chinese statistics as an absolute given, the extent to which this is broken with the SARS analog is quite startling, you know?
It takes SARS many months to sort of build up a head of steam, and this one had some pretty explosive characteristics early on. So that does give some credence, I think, to the academic modeling in the sense that there could be still a lot of stock of unreported claims. The symptoms are not as dramatic either, so, you know, potentially not sick, and, also, there's an incubation period we have to deal with, OK, because even if you contract this, there's going to be a latency period of maybe 4 to 5 days, right, and, obviously, you're not reporting to hospital, you know, being tested at this point.
So all these sorts of mechanics, you know, come into play and sort of cloud the issue, but that's the basic framework, and it's important because once we try to have a public health intervention, what you're effectively trying to do is to bring that reproductive number down, OK, and try to bring it down hard. That's the aim.
And there's two main sort of constraints, or vectors, that you're trying to sort of push down with. One is transmissibility. So this is the idea that, you know, we alert people that there's an outbreak. It's public knowledge, and then people sort of take their own steps to, you know, reduce the chances that they're one of the 2.5 people, Tony, that you might have infected, OK?
TONY NASH: So sorry about that.
GRANT WILSON: Yeah, they could wear-- they could wear a face mask. They can use antiseptic wipes. They can distance themselves from people, you know, things like this, and that should bring things down to some degree, right? We saw that with SARS.
The second is just human mobility in general, so quarantining cities, extending the spring break for Lunar New Year, obviously the flight bans. So you're trying to push down, you know, in these two ways. And, you know, the aim is to bring that R0, which is not a constant number, and it will vary country by country, potentially, down through 1, and at that point, the disease starts to fade out, the outbreak fades, and it's on a sustainable path.
TONY NASH: OK, so we hear things like the face masks aren't necessarily effective in stopping transmission. I even heard something yesterday about the pollution rate can make the lungs more susceptible to, you know, all this other stuff. How much of that are you-- are you guys paying attention to within your data and your analysis?
GRANT WILSON: Yeah, I think this is when you get into the field of conjecture, you know?
TONY NASH: Right.
GRANT WILSON: We try to avoid that at Exante, like, we do really--
TONY NASH: OK, that's fair.
GRANT WILSON: --try to ground--
TONY NASH: Yeah, that's fair.
GRANT WILSON: --our observations both on this outbreak, but also on the economic transmission. Now there's another point to make, which is that one of the differentiating features between COVID and SARS potentially is that there is a risk that COVID can be passed on pre-symptoms, OK? So this is referred to as asymptomatic infection.
So, here, the idea is that you might be in three or four or day five, and you don't get coughing, you're not getting having a runny nose and a sore throat and the sorts of symptoms which would cause you to self-quarantine yourself. In the case of SARS, that's what happened, and at that point, you're infectious, but you knew you were sick, and you did something about it, right?
TONY NASH: Right.
GRANT WILSON: In this particular instance, the risk is that there are reports in some early stage medical reports as well coming through, which support the idea that potentially COVID can be passed on asymptomatically. You are more contagious later, but you could be contagious early, and if that's the case, the first mode of intervention, namely trying to reduce transmissibility, could be impeded because it doesn't actually-- if you don't know you're sick, it might not be wearing the face mask. You might not be socially distancing.
So that's sort of the risk scenario that R0, you know, starts at a slightly higher point for COVID, and then it's sort of sticky, if you will, like even with these dramatic interventions because the interventions have been historically profound in this particular instance, but we just don't get the collapse R0 that we're targeting.
So that's the big open question, and it's a very pertinent question just now because, as you know, China is trying to return to work, OK? We even saw President Xi today, which was a pretty important signal with his face mask in Beijing.
So what we don't know yet, and what we're really, really interested in monitoring with all of these high-frequency data that we have over the next couple of weeks is, as we get a resumption in human mobility, and it won't go back from where it is right now 20% to 100%, but if it moves up to say 50% and 60%, the real question is going to be as people come back to work, potentially to school in a couple of weeks' time, do those suspected cases start to rise again, do infection rates start to rise again?
So if that happens, then we're going to have a really serious problem on our hands because the public health initiative and priority will be running straight into the economic priority.
TONY NASH: Sure.
GRANT WILSON: I hope at the moment, and what the market is buying into I think, is that you can sort of have your cake and eat it, right?
TONY NASH: Right.
GRANT WILSON: And maybe we will. But it's clearly a watching brief-- clearly a watching brief at the moment.
TONY NASH: Yeah, so before we get into the economic effects, which I'm really interested in, my real question is kind of how do we know that we know that we're entering the late phases of this? You know, it's-- I know that maybe that's the magic question, but what--
GRANT WILSON: It sound like a Rumsfeldian question, actually.
TONY NASH: What things will will we start to see that we know that we know that we're in the last, you know, the last inning of this?
GRANT WILSON: Yeah, you seem to be channeling Donald Rumsfeld with the knows and the knows.
TONY NASH: OK, sure.
GRANT WILSON: That's fine. And, listen, the reality is we don't. We don't.
TONY NASH: OK.
GRANT WILSON: I think, you know, this is different parts of this sort of the political economic structure here that we can sort of think about. The Chinese government will have their own data and publish data as well, on-the-ground data, you know, trying to figure out what's going on.
Then you got the international community with a different set of data, potentially, led by the WHO, obviously. Then we've got the media reporting, and as I said, we've got some, you know, the academic community as well. I think, you know, the World Health Organization in particular I think will probably lag, you know, in terms of giving us an all-clear. And in a sense, a precondition for that will actually be the expert community, I think.
So the sorts of modeling that's being done, although it's got these wide variances, it would be very reassuring, very reassuring, ultimately, to the WHO as well if some of these academic teams started to come out publicly and say, actually, you know, we do think it's peaking, like, that would be really, really instructive.
But in terms of the data itself, listen, it's a real problem, Tony. It genuinely is. And without getting sort of too far off pace here, you know, if you had a nuclear accident, you know, for example, in Ukraine, for example, you can use a Geiger counter and stop that, for example, and figure out how many sieverts, and you sort of know how much ionizing radiation, you know, has been emitted and how big a problem Chernobyl might be, right? You can sort of figure it out mathematically.
In this particular instance, you can't look, you know, from the outside and know with certainty. As I said, we can estimate, but at the end of the day, what we know and don't know, there's still a wide gap there.
TONY NASH: Right. Well--