 Hi everyone, it's MJ, the fellow actor, and in this video I want to talk about risk. Now I see risk as being the consequences of uncertainty and it's important that we understand that it's consequences as well as uncertainty and not just uncertainty on its own. And a nice little example is that if I was to just say flip a coin, there is uncertainty. It could be heads, it could be tails, but I don't really care what it lands on so there are no consequences. However, if I were to place a bet, say $100 on the coin landing on heads for a $200 payoff, then we still have the uncertainty of it either being heads or tails, but now there's the consequence where I could make $100 because I get $200 less than the $100 I paid for, or I lose $100 because I don't get the winnings of the coin landing on heads. So this is the whole idea of risk is that it's the consequences resulting from uncertainty and not just uncertainty on its own. What we can also do is we can look at the dimensions of risk and risk has six dimensions. The first dimension is the event. What has to happen for the risk to occur? We then have duration for how long are we exposed to this risk? We also have frequency, which is the probability or rate of the risk event occurring. We also have severity and this is the magnitude of the consequences, you know, how big or how small. We also have correlation, which is the relationship that the risk has with other risk events. And finally, we have capital and these are the reserves needed to support the risk. Now, what makes risk a little bit confusing or difficult is the fact that we can have uncertainty within each of the dimensions. For instance, the event, the cause could be uncertain. If you're ensuring somebody against death, there's many different ways to die. So there's all these various different events that could cause the risk from manifesting. Also, we have duration, the exposure of which you're bound to some risk could be a random variable on its own. Then we sometimes see that we don't actually know what the true probability or rate could be. So once again, frequency could be uncertain. Severity, this is a big one. This is the size of the consequences could be uncertain. And in actual science, we spend a lot of time going through loss distributions. And we actually create a whole distribution for severity because there's so many different values that it could take. Correlation is probably the hardest one to understand. It's the relationship between various things. And what makes correlation difficult is that it is subject to change. And this is why we introduce things like copulas to try and model correlation that can change in extreme situations. And finally, we have capital. Now capital once again is also uncertain because the amount of reserves you hold only gives you a level of confidence. And in no way can give you certainty unless you over get a massive, massive thing. But with short term insurance and things that have extreme value theory, it's impossible to have reserves that give you 100% certainty that you'll be able to support it. So this is one of the tricky things about risk is that it has these dimensions and that there's uncertainty within each of these dimensions. So when it normally comes to managing risk, this is saying that act trees are quite good at. So let's give, for example, let's look at the actuarial risk of mortality. For instance, let's look at life insurance. What life insurance can do, it's got various ways to manage or control risk looking at the dimensions. So when it comes to events, life insurance can sometimes add exclusions to control for events uncertainty. This could be they won't pay out on suicide, or they won't pay out on a nuclear attack, or they won't pay out on war. So they can try and exclude certain events just to try and control that uncertainty. Then there's a product known as term assurance, which controls the duration. So we say we will only cover your life for the next 10 years or for the next 20 years. A term assurance contract is designed to control for the duration. We then have various actuarial modeling and investigations that try and reduce the rate of uncertainty. This is where we develop life tables and we look at mortality curves and all these things is to try and understand the frequency of this risk event from occurring. Then when it comes to severity, you can control severity by giving people fixed benefits. So if somebody dies, irrespective of how they die, you could say you're going to get a million dollars. When we start branching out and we start looking at, say, short term insurance and when you crash a car, we'll cover the cost of the damages, suddenly that fixed benefit becomes uncertain again because you could crash into a Ferrari or you could just get a little scratch and you can see that the severity becomes a random variable. But with life insurance, sometimes what they do is they make the benefit fixed so that they can control for the uncertainty. When it comes to correlation, a common trick that they'll do in, say, mortality is they will also sell life annuities. And life annuities is a product where you pay someone as long as they're alive. So the company is now exposed to longevity risk. And not only are you getting diversification, but you're also getting some offsetting benefits. So if everybody starts living longer, then you get hurts on longevity, but you make it up on mortality profit. And if the opposite happens, your life insurance product might struggle, but then your life annuities will do well. So you can see how actuaries can use correlation in a strategic or way that they position their business to try and offset the benefits. And then the final thing that that actuaries and statisticians love to do is we found that by pulling multiple similar risks together, we can increase the capital confidence. And the reason why this happens is because we reduce variant. And the classic example is if I flip a coin, I don't know if it's going to be heads, and I don't know if it's going to be tails. But if I flip 100 coins, I know that most likely there's going to be between 45 and 55 heads. I can kind of create a little bit more certainty on the amount when the number of risks start to increase. So this is how actuaries traditionally handle, say, a risk like mortality. What we're going to be looking at and what we're going to be creating Udemy courses for is, okay, how do we use this idea when it comes to say market risk, credit risk, and operational risk, and you'll see that we're going to have to take on a slightly different approaches. And interesting with operational debt is going to include systems, people, business, and it's kind of like an umbrella term for all the other risks we might possibly face. Now, a standard approach when it comes to measuring risk is to observe a population of size n for a duration t. We then count the number of risk events that occurs over duration t to that population. And then frequency is quite simply the number of risk events divided by the population size. And for each one, we can observe the size of a loss from each risk. And we can say severity is then going to, well, the average severity is going to be the sum of all losses divided by the number of risk events. And you can kind of price risk as being severity times frequency. Of course, capital is a little bit more confusing. It's not so straightforward that we would have to build something like a RERN model and maybe do a few simulations or some calculations to figure it out. But that is your standard approach to dealing with risk. If you're ever in the exam and they give you some strange risk that you don't know what to do, this is the first principle. This is the first thing that you do. Of course, like we're going to see with credit mark and operational, this kind of breaks down a little bit. And we do need various approaches for each of those risks. Now, let's maybe talk about why do we care about measuring risks? And besides the fact that they're bad, we want to measure risks so we can prepare for it. But essentially, measuring risks accurately helps us to manage it efficiently. And what I want to do is I want to show you this little table. So what we have here is we have severity on the y-axis and we have frequency on the x-axis. And let's just make it very simple for the time being. Let's say frequency can either be low or high and severity can be low and high. If we measure a risk and we see that, well, it's severity is high and it's frequency high, then most likely what we want to do is avoid that risk. If we can, just avoid it. So let's say it's like, should we do business in this other country? There's heavy walls, there's heavy pandemics, there's political instability. The frequency of something bad happening is high and the severity is going to be quite high. Then maybe the best thing to do is to say, okay, hold on. Let's maybe not enter that country for business. Let's avoid it completely. Now, what happens if the frequency is low and the severity is high? This kind of happens specifically in personal households. It's, you know, if somebody was to die, a family member, the severity is going to be high. The frequency is going to be low or having a car crash or something like that. This kind of fits in this little box over here. And what we want to do in this box is we want to transfer the risk. Sometimes we can't avoid these risks. You can't avoid not dying or sometimes if you've got a job that's quite far away, you cannot avoid not using transport, specifically maybe your own car. So what you can then do is you can transfer that risk. If the frequency is high and the severity is low, this tends to be operational risk. This is people making mistakes in the office, sending out letters before spell checking it. You know, you can see the frequency is high, but the damage done, you just look not that professional, but it's not too, too bad. What you want to do here is maybe implement controls. And so a control could be installing a spell checker or have somebody proofreading the letter before it's sent or I'm just giving a very, very trivial example, but it shows that when frequency is high and severity is low, we know, okay, this is the risk that we want to control. And then if frequency is low and severity is low, then we might say, you know what, it's worth maybe just retaining this risk and not necessarily changing our behavior because frequency and severity is low. It's not worth the cost. And that's the thing is the more accurate we can get in measuring the risk, the more efficiently we can manage it. So at a very high level, we can say, should we avoid transfer, retain or control? But if we can accurately measure our risk, then we can start asking questions like, you know, do I want to take this risk? Because what will actually be the cost of retaining it? And then let's say I want to transfer it, if I can measure my risk properly, I can say, well, how much should I pay to transfer it? You know, what is a fair premium to either buy this insurance contract or this derivative or this put call premium? You know, there's all these various things that we can ask ourselves if we can measure risk accurately, we know how much we should pay for it. Let's say we're coming to control, if we can measure risk, then we know how much should my internal control budget be. You know, if these risks maybe are they're happening quite a bit, but they're not that devastating, then we don't need to allocate such a huge budget to it. You know, we can balance the cost benefit of implementing internal controls. And then if we're retaining it, we can say, well, how much capital do I need to retain this risk? You know, what should my liquidity or cash reserves be? So this is the whole idea when it comes to, you know, risk and why we care about measuring it is because by measuring risk accurately, we can manage it efficiently and managing it efficiently means one, knowing what strategy to do, you know, transfer, avoid, retain or control, but then also knowing how much we should spend on each of those strategic actions. And that in a nutshell is what risk and why we measure risk is so important. I'll see you guys in the rest of the course. And like I said, we're going to make one for market risk, one for credit risk and one for operational risk. So I hope you guys enjoy it. Keep well. Cheers.