WEBVTT 00:00:00.000 --> 00:00:02.399 Every single investment decision you ever face, 00:00:02.520 --> 00:00:04.559 I mean, whether you're just picking a single 00:00:04.559 --> 00:00:07.540 stock for your own portfolio or, you know, evaluating 00:00:07.540 --> 00:00:10.560 a huge corporate acquisition, it all requires 00:00:10.560 --> 00:00:13.740 you to make one fundamental calculation. It's 00:00:13.740 --> 00:00:16.379 that tradeoff between risk and return. Exactly. 00:00:16.480 --> 00:00:18.739 We all instinctively know it, right? If you want 00:00:18.739 --> 00:00:21.199 a shot at a higher return, you have to accept 00:00:21.199 --> 00:00:24.899 higher risk. But the question that really launched 00:00:24.899 --> 00:00:29.019 modern financial economics was. How do we objectively 00:00:29.019 --> 00:00:31.160 price that risk? Right. How do you put a number 00:00:31.160 --> 00:00:33.619 on it? Exactly. How do you precisely quantify 00:00:33.619 --> 00:00:36.619 how much extra return is required for taking 00:00:36.619 --> 00:00:39.299 on, say, one more unit of market volatility? 00:00:39.380 --> 00:00:41.979 And that core problem, that search for the objective 00:00:41.979 --> 00:00:44.240 price of risk, well, that's the entire mission 00:00:44.240 --> 00:00:47.200 of the capital asset pricing model, or CAPM, 00:00:47.200 --> 00:00:49.429 as everyone calls it. And that is our deep dive 00:00:49.429 --> 00:00:51.950 today. We are getting into a formula that is, 00:00:52.070 --> 00:00:54.850 and this isn't an exaggeration, truly foundational. 00:00:55.109 --> 00:00:58.049 I mean, CAPM is the cornerstone model that corporations 00:00:58.049 --> 00:01:01.109 use to figure out their cost of equity. And that 00:01:01.109 --> 00:01:03.310 portfolio managers use to justify their investment 00:01:03.310 --> 00:01:06.150 choices, or in many cases, fail to justify them. 00:01:06.230 --> 00:01:08.890 Right. It's this incredible intellectual structure, 00:01:08.969 --> 00:01:11.189 and it's all built on the idea of a perfectly 00:01:11.189 --> 00:01:14.409 rational investor. Fundamentally, what CAPM is, 00:01:14.609 --> 00:01:19.620 it's a tool. It's designed to determine the theoretically 00:01:19.620 --> 00:01:22.560 appropriate required rate of return for any given 00:01:22.560 --> 00:01:24.560 asset. Okay. The rate you should be demanding. 00:01:24.939 --> 00:01:27.840 Precisely. And crucially, and this is a point 00:01:27.840 --> 00:01:29.959 we'll come back to again and again, it assumes 00:01:29.959 --> 00:01:32.299 that asset is being added to an already well 00:01:32.299 --> 00:01:34.620 -diversified portfolio. That's a huge assumption. 00:01:34.980 --> 00:01:38.079 A huge one. And what makes it so powerful and 00:01:38.079 --> 00:01:41.549 so popular, really, is its simplicity. It gives 00:01:41.549 --> 00:01:44.230 you a mathematically derived cost of equity capital, 00:01:44.329 --> 00:01:46.950 and it's determined by just one single factor. 00:01:47.090 --> 00:01:49.730 And that factor is the asset's exposure to non 00:01:49.730 --> 00:01:51.909 -diversifiable risk. The thing that the entire 00:01:51.909 --> 00:01:54.849 financial industry just calls beta. And here 00:01:54.849 --> 00:01:56.510 is the paradox. I mean, we have to acknowledge 00:01:56.510 --> 00:01:58.189 this right at the start because it's what makes 00:01:58.189 --> 00:02:01.670 this dive so fascinating for any learner. CAPM 00:02:01.670 --> 00:02:05.950 is ubiquitous. It's simple. It's elegant. It's 00:02:05.950 --> 00:02:07.769 taught in every single business school on the 00:02:07.769 --> 00:02:10.930 planet. Yet. As our sources show, decades and 00:02:10.930 --> 00:02:13.629 decades of empirical tests have proven that the 00:02:13.629 --> 00:02:17.189 model, well, in its purest form, it often fails 00:02:17.189 --> 00:02:19.909 to predict real world returns. It really does. 00:02:20.110 --> 00:02:22.349 It is the formula that failed, but somehow still 00:02:22.349 --> 00:02:24.669 rules the world of finance. And that's its enduring 00:02:24.669 --> 00:02:27.370 legacy. To understand why it still rules, you 00:02:27.370 --> 00:02:30.229 really have to appreciate the purity of its theory. 00:02:30.629 --> 00:02:33.389 The model itself was built upon the shoulders 00:02:33.389 --> 00:02:35.969 of Harry Markowitz's earlier work. On modern 00:02:35.969 --> 00:02:38.710 portfolio theory. Exactly. The key architects 00:02:38.710 --> 00:02:41.349 who really developed CEPM as we know it today 00:02:41.349 --> 00:02:44.479 were Jack Traynor. William F. Sharp, John Lintner 00:02:44.479 --> 00:02:47.520 and Jan Mawson. And they were all working independently 00:02:47.520 --> 00:02:50.599 in the early 1960s. It was just this rapid intellectual 00:02:50.599 --> 00:02:53.099 explosion, wasn't it? All centered around like 00:02:53.099 --> 00:02:56.659 1961 to 1966. And of course, that work secured 00:02:56.659 --> 00:02:59.400 its permanent place in history when Sharp, along 00:02:59.400 --> 00:03:01.159 with Markowitz and Merton Miller, received the 00:03:01.159 --> 00:03:04.199 1990 Nobel Memorial Prize in Economic Sciences. 00:03:04.300 --> 00:03:06.620 Of a landmark achievement, no question. A landmark, 00:03:06.780 --> 00:03:09.159 even if its actual performance in the real world 00:03:09.159 --> 00:03:12.560 is, well, it's questionable. OK, so. Let's unpack 00:03:12.560 --> 00:03:15.259 this concept of risk because CAPM doesn't view 00:03:15.259 --> 00:03:18.560 all risk equally. Before we even touch the formula, 00:03:18.699 --> 00:03:20.960 we have to establish which risks we actually 00:03:20.960 --> 00:03:23.680 care about and which ones we should just, you 00:03:23.680 --> 00:03:26.599 know, get rid of. That distinction is the true 00:03:26.599 --> 00:03:29.300 genius of modern portfolio theory, which, as 00:03:29.300 --> 00:03:32.400 you said, CAPM inherited. We're dealing with 00:03:32.400 --> 00:03:34.939 the two faces of risk that any investor is going 00:03:34.939 --> 00:03:36.879 to encounter. Let's start with the face we can 00:03:36.879 --> 00:03:40.159 actually control, unsystematic risk. Right. This 00:03:40.159 --> 00:03:42.860 is often called idiosyncratic risk or maybe more 00:03:42.860 --> 00:03:45.759 helpfully diversifiable risk. The key characteristic 00:03:45.759 --> 00:03:47.719 here is that it's unique to a specific company 00:03:47.719 --> 00:03:51.259 or a specific asset. So like a scandal. A perfect 00:03:51.259 --> 00:03:53.539 example. If a pharmaceutical company fails a 00:03:53.539 --> 00:03:56.460 phase three drug trial or a CEO gets caught in 00:03:56.460 --> 00:03:58.819 an accounting scandal, that event really only 00:03:58.819 --> 00:04:01.020 affects that security. It's just asset specific 00:04:01.020 --> 00:04:02.879 noise. And then there's the other phase, the 00:04:02.879 --> 00:04:04.780 risk we can't escape, which is systematic risk. 00:04:04.919 --> 00:04:07.990 That's the big one. non -diversifiable risk or 00:04:07.990 --> 00:04:10.650 just market risk. This is the macroeconomic risk 00:04:10.650 --> 00:04:12.509 that's common to all securities, just to varying 00:04:12.509 --> 00:04:14.289 degrees. OK, so what are we talking about here? 00:04:14.530 --> 00:04:17.089 Think of the big drivers, inflation spiking, 00:04:17.209 --> 00:04:19.990 the Federal Reserve hiking interest rates, a 00:04:19.990 --> 00:04:23.370 sudden global recession, a major geopolitical 00:04:23.370 --> 00:04:26.569 shock. If the overall economy contracts, almost 00:04:26.569 --> 00:04:30.069 every single stock drops. You just can't diversify 00:04:30.069 --> 00:04:32.410 away from that kind of decline within that market. 00:04:32.720 --> 00:04:34.779 And this leads us directly to the great mandate 00:04:34.779 --> 00:04:37.199 of CAPM, right? The power of diversification. 00:04:37.519 --> 00:04:40.959 The model assumes the investor is rational. And 00:04:40.959 --> 00:04:43.939 a rational investor should never, ever take on 00:04:43.939 --> 00:04:47.000 diversifiable risk. Why? Because the model states 00:04:47.000 --> 00:04:49.519 very clearly that only non -diversifiable risk 00:04:49.519 --> 00:04:51.879 is rewarded with a premium return. You don't 00:04:51.879 --> 00:04:53.779 get paid for holding on to company -specific 00:04:53.779 --> 00:04:56.459 bad luck. That's the critical insight. It's so 00:04:56.459 --> 00:04:59.139 important. When you assemble a portfolio, the 00:04:59.139 --> 00:05:02.319 specific risks of those individual assets, they're 00:05:02.319 --> 00:05:04.800 designed to average out or cancel each other 00:05:04.800 --> 00:05:06.899 out. So the good news from one company offsets 00:05:06.899 --> 00:05:09.709 the bad news from another. Exactly. Therefore, 00:05:09.949 --> 00:05:13.209 the required return on any single asset is only 00:05:13.209 --> 00:05:15.649 linked to its contribution to the overall portfolio 00:05:15.649 --> 00:05:19.790 riskiness, not its own unique standalone risk. 00:05:20.310 --> 00:05:23.209 That standalone risk becomes irrelevant once 00:05:23.209 --> 00:05:26.089 you diversify. So practically speaking, what 00:05:26.089 --> 00:05:28.629 does it take to get diversified enough to satisfy 00:05:28.629 --> 00:05:31.389 this model? I mean, do you, the investor, need 00:05:31.389 --> 00:05:34.209 to own 5000 different stocks from all over the 00:05:34.209 --> 00:05:36.839 world? Well, our sources suggest that, you know, 00:05:36.860 --> 00:05:38.819 in large developed markets like the U .S. or 00:05:38.819 --> 00:05:41.100 the U .K., the bulk of the benefits are actually 00:05:41.100 --> 00:05:43.639 achieved pretty quickly. A portfolio of about, 00:05:43.660 --> 00:05:47.199 say, 30 to 40 randomly selected non -correlated 00:05:47.199 --> 00:05:49.839 securities is usually sufficient. 30 to 40 stocks. 00:05:50.060 --> 00:05:51.899 That's it. That's usually enough to essentially 00:05:51.899 --> 00:05:54.439 eliminate the vast majority of that unsystematic 00:05:54.439 --> 00:05:57.259 risk. At that point, your risk exposure is primarily 00:05:57.259 --> 00:05:59.720 the systematic market -wide kind. Okay, but there's 00:05:59.720 --> 00:06:01.740 a catch, I assume. There's a big caveat, yes. 00:06:01.959 --> 00:06:04.759 In less developed emerging markets, asset volatilities 00:06:04.759 --> 00:06:06.459 are much higher and the correlations between 00:06:06.459 --> 00:06:09.660 them can be, well, they can be strange. You'd 00:06:09.660 --> 00:06:11.699 likely need a significantly larger number of 00:06:11.699 --> 00:06:13.680 assets to achieve that same level of protection. 00:06:14.100 --> 00:06:16.180 Okay, so once we've assumed the investor has 00:06:16.180 --> 00:06:19.180 successfully diversified away all that idiosyncratic 00:06:19.180 --> 00:06:23.040 risk, we're left with only systematic risk. And 00:06:23.040 --> 00:06:24.759 that's where we need a measure. We need a ruler. 00:06:24.959 --> 00:06:27.860 A ruler. And that measure is beta. I always see 00:06:27.860 --> 00:06:31.420 beta as like a financial Geiger counter. It measures 00:06:31.420 --> 00:06:34.319 an asset's unique exposure to that remaining 00:06:34.319 --> 00:06:37.279 inescapable market risk. That's a perfect way 00:06:37.279 --> 00:06:39.420 to put it. Beta is the one and only variable 00:06:39.420 --> 00:06:43.319 CIPM uses to justify why different assets should 00:06:43.319 --> 00:06:45.740 have different required returns. It measures 00:06:45.740 --> 00:06:48.379 an asset's sensitivity to that non -diversifiable 00:06:48.379 --> 00:06:50.750 market movement. Okay. Technically speaking, 00:06:50.910 --> 00:06:52.930 it quantifies the sensitivity of the expected 00:06:52.930 --> 00:06:55.829 excess asset returns to the expected excess market 00:06:55.829 --> 00:06:58.089 returns. Let's lay out the benchmarks then, because 00:06:58.089 --> 00:07:00.810 beta numbers are basically a universal shorthand 00:07:00.810 --> 00:07:03.350 in finance. What's the baseline? The baseline, 00:07:03.529 --> 00:07:06.709 by definition, is a beta of 1. The market as 00:07:06.709 --> 00:07:08.850 a whole, which we usually proxy with a broad 00:07:08.850 --> 00:07:11.509 index like the S &P 500, has a beta of exactly 00:07:11.509 --> 00:07:14.430 1. So an asset with a beta of 1 just moves perfectly 00:07:14.430 --> 00:07:16.490 in sync with the market. That's the idea. So 00:07:16.490 --> 00:07:19.899 if the S &P 500 goes up 10 % in a year... A stock 00:07:19.899 --> 00:07:21.980 with a beta of one is also expected to go up 00:07:21.980 --> 00:07:24.639 10%, all else being equal. What about the higher 00:07:24.639 --> 00:07:28.040 risk assets? Well, if an asset has a beta greater 00:07:28.040 --> 00:07:31.740 than one, let's say 1 .5, it means the asset 00:07:31.740 --> 00:07:34.060 is fundamentally leveraged to market movements. 00:07:34.319 --> 00:07:36.579 What kind of stocks are we talking about? Think 00:07:36.579 --> 00:07:39.100 aggressive growth stocks, cyclical tech companies, 00:07:39.379 --> 00:07:43.230 luxury consumer goods, things people buy. When 00:07:43.230 --> 00:07:45.589 the economy is booming, if the market rises 10 00:07:45.589 --> 00:07:48.750 percent, that stock with a 1 .5 beta might rise 00:07:48.750 --> 00:07:51.910 15 percent. It carries more systematic risk exposure 00:07:51.910 --> 00:07:54.810 and therefore it's expected to generate a higher 00:07:54.810 --> 00:07:57.750 return to compensate the investor. And then conversely, 00:07:57.769 --> 00:08:00.350 you have the defensive lower risk plays. Exactly. 00:08:00.769 --> 00:08:02.910 Assets with a beta less than one indicate they 00:08:02.910 --> 00:08:05.110 are less sensitive than the market average. So 00:08:05.110 --> 00:08:07.129 beta of 0 .7, for instance. And these would be? 00:08:07.269 --> 00:08:09.370 These are your classic defensive stocks utilities. 00:08:09.970 --> 00:08:12.250 Establish consumer staples like toothpaste companies 00:08:12.250 --> 00:08:14.990 or big telecom providers. If the market drops 00:08:14.990 --> 00:08:18.329 10%, these stocks might only drop 7%. They're 00:08:18.329 --> 00:08:20.410 associated with lower systematic risk and therefore 00:08:20.410 --> 00:08:22.829 imply a lower required return. Okay, to really 00:08:22.829 --> 00:08:24.709 appreciate beta, I think we have to look at the 00:08:24.709 --> 00:08:27.850 math behind it. You defined it using covariance 00:08:27.850 --> 00:08:29.970 and variance. And for the learner, that can sound 00:08:29.970 --> 00:08:32.190 a little dense. So let's look at the formula 00:08:32.190 --> 00:08:34.470 and then maybe translate it into simple movement. 00:08:34.860 --> 00:08:38.519 Sure. The formula is the covariance of the assets 00:08:38.519 --> 00:08:41.539 return and the markets return divided by the 00:08:41.539 --> 00:08:44.000 variance of the markets return. And what's fascinating 00:08:44.000 --> 00:08:46.600 here is what that ratio is actually doing. Right. 00:08:46.659 --> 00:08:49.580 The numerator covariance that just measures how 00:08:49.580 --> 00:08:51.580 much the assets returns and the markets returns 00:08:51.580 --> 00:08:54.320 move together. Are they highly correlated? Do 00:08:54.320 --> 00:08:57.580 they surge and fall in unison? And the denominator, 00:08:57.779 --> 00:09:00.120 the variance of the market return, is simply 00:09:00.120 --> 00:09:02.659 a measure of the market's total volatility. It's 00:09:02.659 --> 00:09:04.820 just its own internal movement. So we're taking 00:09:04.820 --> 00:09:07.860 the degree to which our specific asset is sensitive 00:09:07.860 --> 00:09:09.919 to the market's ups and downs, and we're dividing 00:09:09.919 --> 00:09:12.759 it by the market's own total movement. Precisely. 00:09:12.759 --> 00:09:15.039 And by doing that division, beta gives you a 00:09:15.039 --> 00:09:18.279 clear standardized ratio. If a stock's movements, 00:09:18.559 --> 00:09:21.580 its covariance, are twice as large as the market's 00:09:21.580 --> 00:09:23.759 own movements, its variance, then that stock 00:09:23.759 --> 00:09:26.440 has a beta of two. It's just a simple ratio. 00:09:26.700 --> 00:09:29.860 It strips away all the noise and tells us exactly 00:09:29.860 --> 00:09:33.159 how sensitive that stock is to the big macroeconomic 00:09:33.159 --> 00:09:36.860 tides. It gives us a really robust measure of 00:09:36.860 --> 00:09:39.940 its systematic risk contribution. So with beta 00:09:39.940 --> 00:09:42.460 established as the one and only driver of price 00:09:42.460 --> 00:09:46.340 risk, we can now assemble. the standard CAPM 00:09:46.340 --> 00:09:49.200 framework. This is the central equation, the 00:09:49.200 --> 00:09:51.179 one that connects that systematic risk directly 00:09:51.179 --> 00:09:53.620 to the expected return an investor has to demand. 00:09:54.120 --> 00:09:56.779 The standard CAPM formula for the expected return 00:09:56.779 --> 00:10:00.879 on an asset is the risk -free rate plus beta 00:10:00.879 --> 00:10:03.179 times the market return minus the risk -free 00:10:03.179 --> 00:10:05.059 rate. Okay, let's walk through this piece by 00:10:05.059 --> 00:10:07.360 piece because every component is important. Everything 00:10:07.360 --> 00:10:09.240 is what we're solving for, right? The expected 00:10:09.240 --> 00:10:11.559 or required return on our asset. This is the 00:10:11.559 --> 00:10:14.120 minimum return an investor demands to even consider 00:10:14.120 --> 00:10:16.320 holding that asset. And we begin with true dollars, 00:10:16.480 --> 00:10:19.340 the risk -free rate. This represents the absolute 00:10:19.340 --> 00:10:21.419 floor of returns. It's the reward you get just 00:10:21.419 --> 00:10:23.899 for committing capital over time with absolutely 00:10:23.899 --> 00:10:26.320 zero systematic risk. And in the real world, 00:10:26.360 --> 00:10:28.700 what do we use for that? Typically, this is proxied 00:10:28.700 --> 00:10:30.659 by the return on short -term government debt, 00:10:30.740 --> 00:10:32.679 like a U .S. Treasury bill. Okay, so that's our 00:10:32.679 --> 00:10:34.720 baseline. Then we define the engine of the market, 00:10:34.840 --> 00:10:37.480 ERM as well as the expected return of the market 00:10:37.480 --> 00:10:41.039 portfolio. What do we generally expect the whole 00:10:41.039 --> 00:10:43.970 market to deliver? And that term in the parentheses, 00:10:44.250 --> 00:10:47.309 the market return minus the risk -free rate, 00:10:47.450 --> 00:10:50.990 that is absolutely crucial. This is the market 00:10:50.990 --> 00:10:53.370 premium. Or sometimes called the market risk 00:10:53.370 --> 00:10:56.070 premium. Right. It's the generalized reward the 00:10:56.070 --> 00:10:58.710 market offers just for accepting average market 00:10:58.710 --> 00:11:01.809 risk over and above that risk -free floor. If 00:11:01.809 --> 00:11:04.330 you had a beta of 1, this entire amount would 00:11:04.330 --> 00:11:07.019 be your reward for taking on risk. But because 00:11:07.019 --> 00:11:09.440 we have individual assets with different sensitivities, 00:11:09.500 --> 00:11:12.659 we scale that market premium using the asset's 00:11:12.659 --> 00:11:16.179 unique beta. So if your beta is 0 .5, you only 00:11:16.179 --> 00:11:18.539 get to claim half of that market premium. And 00:11:18.539 --> 00:11:20.500 if your beta is 2, you claim double the market 00:11:20.500 --> 00:11:23.240 premium. It's perfectly proportional. The logic 00:11:23.240 --> 00:11:25.639 is that the asset's reward is... perfectly calibrated 00:11:25.639 --> 00:11:27.679 by its unique sensitivity. And we can express 00:11:27.679 --> 00:11:29.960 this even more clearly if we rearrange the equation 00:11:29.960 --> 00:11:32.679 a bit. Okay. So if we just subtract the risk 00:11:32.679 --> 00:11:35.039 -free rate from both sides. We get what's called 00:11:35.039 --> 00:11:38.000 the risk premium form. It states that the individual 00:11:38.000 --> 00:11:41.159 risk premium an asset delivers, that's its return 00:11:41.159 --> 00:11:44.299 above the risk -free rate, must be exactly equal 00:11:44.299 --> 00:11:47.399 to the market premium scaled by beta. Which just 00:11:47.399 --> 00:11:49.940 confirms that central idea. It confirms the central 00:11:49.940 --> 00:11:53.629 theoretical thesis. The reward for risk has to 00:11:53.629 --> 00:11:56.309 be perfectly linear and proportional to systematic 00:11:56.309 --> 00:11:59.289 risk across every single asset in the market. 00:11:59.450 --> 00:12:01.870 OK, so if the formula is the brain, then the 00:12:01.870 --> 00:12:05.409 security market line, or SML, is the visual body 00:12:05.409 --> 00:12:08.450 of CAPM. It's the graph that analysts use to 00:12:08.450 --> 00:12:10.470 translate all this theoretical math into practical 00:12:10.470 --> 00:12:12.750 investment decisions. It's where you can spot 00:12:12.750 --> 00:12:15.289 potential market errors. The SML is just a graph 00:12:15.289 --> 00:12:17.940 of the CAPM formula. It's that simple. On the 00:12:17.940 --> 00:12:20.500 x -axis, we plot risk and remember only systematic 00:12:20.500 --> 00:12:23.139 risk measured by beta. And on the y -axis, we 00:12:23.139 --> 00:12:24.960 plot the expected return. And if you look at 00:12:24.960 --> 00:12:27.019 that line, you can immediately see the structure 00:12:27.019 --> 00:12:29.480 of the formula. The intercept where beta is zero, 00:12:29.539 --> 00:12:32.000 that has to be the nominal risk -free rate, $4. 00:12:32.080 --> 00:12:34.139 That's the reward floor for taking no market 00:12:34.139 --> 00:12:37.440 risk. And the slope of that line, how steep it 00:12:37.440 --> 00:12:40.320 is, that represents the market risk premium. 00:12:40.899 --> 00:12:43.340 A steeper slope just means the market is demanding 00:12:43.340 --> 00:12:46.379 a higher compensation for taking on risk. And 00:12:46.379 --> 00:12:49.139 this line, in theory, defines perfect market 00:12:49.139 --> 00:12:51.840 equilibrium. And this visualization is where 00:12:51.840 --> 00:12:53.779 the rubber meets the road for anyone interested 00:12:53.779 --> 00:12:55.919 in active management, isn't it? Because in a 00:12:55.919 --> 00:12:58.960 theoretical CPM world, every single security 00:12:58.960 --> 00:13:01.360 should plot directly on the SML. They should 00:13:01.360 --> 00:13:04.139 all line up perfectly. But in reality, they don't. 00:13:04.139 --> 00:13:06.580 So if we plot an asset's expected return in its 00:13:06.580 --> 00:13:10.120 beta and we find it sits above the SML, CAPM 00:13:10.120 --> 00:13:12.509 tells us that that asset is undervalued. Why 00:13:12.509 --> 00:13:14.549 undervalued? Because it means that for the level 00:13:14.549 --> 00:13:17.330 of systematic risk it carries, its beta, it is 00:13:17.330 --> 00:13:19.809 currently offering a return that is greater than 00:13:19.809 --> 00:13:21.850 what the market theoretically requires. The market 00:13:21.850 --> 00:13:23.870 is basically giving you a bonus return for that 00:13:23.870 --> 00:13:26.210 level of risk. So if an asset is offering a 12 00:13:26.210 --> 00:13:28.809 % return for a risk level where the SML says 00:13:28.809 --> 00:13:31.309 it should only offer 10%, an active investor 00:13:31.309 --> 00:13:33.629 would see that 2 % difference as a clear buying 00:13:33.629 --> 00:13:36.649 signal. Exactly. The expectation is that demand 00:13:36.649 --> 00:13:39.179 for that asset will push its price up. Which 00:13:39.179 --> 00:13:41.559 in turn drives the expected future return back 00:13:41.559 --> 00:13:45.419 down until it lands on the SML. Right. And conversely, 00:13:45.419 --> 00:13:49.200 if an asset plots below the SML, it's overvalued. 00:13:49.279 --> 00:13:51.600 You are accepting a return that is less than 00:13:51.600 --> 00:13:53.820 what the market demands for that specific amount 00:13:53.820 --> 00:13:56.860 of systematic risk. The expectation is that the 00:13:56.860 --> 00:13:59.539 asset's price will fall, which naturally drives 00:13:59.539 --> 00:14:02.200 its future expected returns back up toward that 00:14:02.200 --> 00:14:05.029 equilibrium line. That deviation, that vertical 00:14:05.029 --> 00:14:07.529 distance between where the asset's actual return 00:14:07.529 --> 00:14:10.009 is and where the theoretical SML says it should 00:14:10.009 --> 00:14:12.370 be, is so important that it earned its own famous 00:14:12.370 --> 00:14:15.309 measure, Jensen's alpha. Jensen's alpha, or just 00:14:15.309 --> 00:14:17.909 alpha, is the formal measurement of that abnormal 00:14:17.909 --> 00:14:20.570 return relative to the SML. It's essentially 00:14:20.570 --> 00:14:23.169 a mathematical way of measuring the error term 00:14:23.169 --> 00:14:26.149 in the model. Or, for a portfolio manager, it's 00:14:26.149 --> 00:14:28.409 a measure of their skill. And we can incorporate 00:14:28.409 --> 00:14:30.850 alpha right into the CAPM equation as a sort 00:14:30.850 --> 00:14:34.039 of residual term, can't we? We can. A positive 00:14:34.039 --> 00:14:36.320 alpha means the asset has outperformed the return 00:14:36.320 --> 00:14:39.220 expected based purely on its systematic risk 00:14:39.220 --> 00:14:42.000 and the market premium. It suggests the asset 00:14:42.000 --> 00:14:44.500 is undervalued or maybe the manager found a true 00:14:44.500 --> 00:14:47.110 market inefficiency. This is the return that 00:14:47.110 --> 00:14:49.649 cannot be explained by beta alone. And generating 00:14:49.649 --> 00:14:52.769 positive alpha is what active fund managers are 00:14:52.769 --> 00:14:54.809 paid, you know, massive sums of money to do. 00:14:54.889 --> 00:14:57.409 If a fund manager generates a 15 percent return 00:14:57.409 --> 00:15:00.049 in a market that gave 10 percent and their funds 00:15:00.049 --> 00:15:02.450 beta was exactly one, they have 5 percent alpha. 00:15:02.669 --> 00:15:05.250 They generated excess return that the standard 00:15:05.250 --> 00:15:07.250 model couldn't predict. And a negative alpha 00:15:07.250 --> 00:15:10.399 is, of course, the dreaded opposite. underperformance. 00:15:10.700 --> 00:15:13.960 It often suggests overvaluation, or simply that 00:15:13.960 --> 00:15:16.159 the manager destroyed value relative to the risk 00:15:16.159 --> 00:15:18.559 they took on. And if an asset has a zero alpha, 00:15:18.740 --> 00:15:20.980 it's considered fairly valued and plots directly 00:15:20.980 --> 00:15:23.639 on the SML. So we've established that CAPM outputs 00:15:23.639 --> 00:15:26.419 this required return, EWAR. Let's turn to the 00:15:26.419 --> 00:15:28.899 practical application now. Why is this number 00:15:28.899 --> 00:15:30.940 so essential in corporate finance and investment 00:15:30.940 --> 00:15:33.200 banking? Because it is the required discount 00:15:33.200 --> 00:15:36.700 rate, the year use entry that CAPM generates. 00:15:37.450 --> 00:15:39.509 provides the asset appropriate required return. 00:15:39.789 --> 00:15:41.970 And this number becomes the critical component 00:15:41.970 --> 00:15:44.750 for determining a company's cost of equity. Which 00:15:44.750 --> 00:15:47.110 then feeds into the WACC, the weighted average 00:15:47.110 --> 00:15:49.970 cost of capital. Exactly. And without that rate, 00:15:50.110 --> 00:15:53.509 companies can't accurately assess whether a new 00:15:53.509 --> 00:15:56.470 project, building a new factory, launching a 00:15:56.470 --> 00:15:59.070 new product, is actually going to create shareholder 00:15:59.070 --> 00:16:01.470 value. It's a really intuitive relationship, 00:16:01.690 --> 00:16:04.110 isn't it, between risk and discounting? If I 00:16:04.110 --> 00:16:06.330 need a higher return to take on higher risk, 00:16:06.509 --> 00:16:08.950 then that higher return must translate into a 00:16:08.950 --> 00:16:11.110 higher discount rate that I apply to any future 00:16:11.110 --> 00:16:14.250 cash flows. It has to. A stock with a high beta 00:16:14.250 --> 00:16:16.789 will have a high required return, and therefore 00:16:16.789 --> 00:16:18.950 its future cash flows will be discounted at a 00:16:18.950 --> 00:16:22.009 higher rate. Less sensitive stocks with lower 00:16:22.009 --> 00:16:25.070 betas are discounted at a lower rate. This reflects 00:16:25.070 --> 00:16:27.769 a fundamental aspect of investor behavior, which 00:16:27.769 --> 00:16:30.049 is described by what economists call the concave 00:16:30.049 --> 00:16:32.169 utility function. OK, that's a bit of academic 00:16:32.169 --> 00:16:34.769 jargon. Can you break down what the concave utility 00:16:34.769 --> 00:16:36.850 function means for someone just learning this? 00:16:37.070 --> 00:16:39.990 Sure. It just means that investors are risk averse. 00:16:40.149 --> 00:16:42.929 They value the first dollar they gain much more 00:16:42.929 --> 00:16:45.570 than the hundredth dollar they gain. Therefore, 00:16:45.769 --> 00:16:48.210 to persuade a risk averse investor to take on 00:16:48.210 --> 00:16:50.629 a higher level of risk, you have to offer them 00:16:50.629 --> 00:16:52.610 a disproportionately larger amount of expected 00:16:52.610 --> 00:16:54.990 return. The more risk I take, the more you have 00:16:54.990 --> 00:16:57.509 to pay me, and at an increasing rate. At an accelerating 00:16:57.509 --> 00:17:00.090 rate, exactly. So the higher discount rate we 00:17:00.090 --> 00:17:02.889 use for high beta stocks just reflects this behavioral 00:17:02.889 --> 00:17:06.029 requirement for compensation. So once that required 00:17:06.029 --> 00:17:09.609 return, it's established via CAPM, it is plugged 00:17:09.609 --> 00:17:11.869 directly into the foundation of all fundamental 00:17:11.869 --> 00:17:15.190 valuation. calculating an asset's intrinsic value 00:17:15.190 --> 00:17:17.650 based on its expected future cash flows. This 00:17:17.650 --> 00:17:20.509 is the classic present value calculation. We 00:17:20.509 --> 00:17:22.589 take all those future cash flows we anticipate 00:17:22.589 --> 00:17:24.769 the company will generate and we shrink them 00:17:24.769 --> 00:17:27.470 back to today's dollar value using that CAPM 00:17:27.470 --> 00:17:30.069 derived rate. The formula for the intrinsic present 00:17:30.069 --> 00:17:33.269 value, or PV, is just the sum of all those discounted 00:17:33.269 --> 00:17:36.670 future cash flows. And the resulting PV is the 00:17:36.670 --> 00:17:39.170 theoretical fair price for that asset today. 00:17:39.589 --> 00:17:42.329 This gives you the most important valuation comparison 00:17:42.329 --> 00:17:46.329 you can make. If the PV you calculate is higher 00:17:46.329 --> 00:17:48.829 than the current market price, then the asset 00:17:48.829 --> 00:17:51.569 is undervalued. It's undervalued. You are essentially 00:17:51.569 --> 00:17:54.690 buying its future cash flows at a discount relative 00:17:54.690 --> 00:17:57.549 to their systematic risk. And if the current 00:17:57.549 --> 00:18:00.150 market price is higher than the intrinsic value 00:18:00.150 --> 00:18:02.650 calculated, then the asset should be considered 00:18:02.650 --> 00:18:06.069 overvalued. This is the mechanism, the nuts and 00:18:06.069 --> 00:18:10.269 bolts by which CAPM translates risk into a tangible, 00:18:10.490 --> 00:18:13.559 actionable price tag. And I should probably mention 00:18:13.559 --> 00:18:16.039 our sources note that CAPM can also calculate 00:18:16.039 --> 00:18:18.460 the asset price using an alternative structure. 00:18:18.680 --> 00:18:20.339 It's sometimes called the certainty equivalent 00:18:20.339 --> 00:18:23.119 pricing formula. OK, that sounds a little esoteric. 00:18:23.119 --> 00:18:24.759 How is that different from the standard present 00:18:24.759 --> 00:18:26.839 value calculation we just discussed? It's just 00:18:26.839 --> 00:18:29.319 another way to frame the exact same idea. Instead 00:18:29.319 --> 00:18:31.200 of adjusting the discount rate, the denominator 00:18:31.200 --> 00:18:34.160 for risk, the certainty equivalent method adjusts 00:18:34.160 --> 00:18:36.829 the expected cash flows. The numerator. How does 00:18:36.829 --> 00:18:39.470 it do that? It calculates the certainty equivalent 00:18:39.470 --> 00:18:42.430 cash flows. That's the guaranteed amount of money 00:18:42.430 --> 00:18:45.309 today that an investor would accept instead of 00:18:45.309 --> 00:18:48.210 the risky future cash flows. Then you just discount 00:18:48.210 --> 00:18:49.990 that guaranteed amount at the risk free rate. 00:18:50.109 --> 00:18:52.390 So you either adjust the cash flow for risk or 00:18:52.390 --> 00:18:55.089 you adjust the discount rate for risk. Exactly. 00:18:55.309 --> 00:18:57.950 Both the methods lead to the same result. But 00:18:57.950 --> 00:18:59.950 the certainty equivalent approach focuses on 00:18:59.950 --> 00:19:02.130 risk adjusting the cash flow itself instead of 00:19:02.130 --> 00:19:04.289 the rate. So we have the year I, the rate that 00:19:04.289 --> 00:19:06.650 CAPM demands we earn, and it's derived purely 00:19:06.650 --> 00:19:09.230 from beta. How do we turn that into an actual 00:19:09.230 --> 00:19:12.130 investment decision? Well, you have to compare 00:19:12.130 --> 00:19:15.349 that CAPM required rate of return to an independent 00:19:15.349 --> 00:19:17.589 estimate of the asset's actual return outlook. 00:19:17.910 --> 00:19:20.170 Right. CAPM can't tell you what the future cash 00:19:20.170 --> 00:19:22.009 flows will be. You have to do that work yourself. 00:19:22.190 --> 00:19:24.309 So you need separate analysis. You need fundamental 00:19:24.309 --> 00:19:26.329 analysis, maybe technical indicators, looking 00:19:26.329 --> 00:19:28.569 at PE ratios, comparing it to similar firms, 00:19:28.670 --> 00:19:30.799 using market -to -book ratios, and so on. Okay, 00:19:30.839 --> 00:19:32.480 let's put some numbers on it to make it concrete. 00:19:32.940 --> 00:19:36.339 Let's say CAPM tells me that based on this asset's 00:19:36.339 --> 00:19:39.940 beta, its required return is 9%. But based on 00:19:39.940 --> 00:19:41.759 my own deep dive into the company's expected 00:19:41.759 --> 00:19:44.599 growth and its dividend policy, I calculate that 00:19:44.599 --> 00:19:46.420 the current market price will actually generate 00:19:46.420 --> 00:19:49.799 an expected return of 11%. That 2 % discrepancy 00:19:49.799 --> 00:19:52.430 is the key signal. If the expected return of 00:19:52.430 --> 00:19:55.730 the asset, your 11%, is higher than what CAPM 00:19:55.730 --> 00:19:59.250 says is required, the 9%, it implies that the 00:19:59.250 --> 00:20:01.329 current market price must be too low. The asset 00:20:01.329 --> 00:20:04.309 is undervalued. It's undervalued. Assuming, of 00:20:04.309 --> 00:20:06.250 course, that the market is efficient enough to 00:20:06.250 --> 00:20:08.509 eventually correct that price, which would pull 00:20:08.509 --> 00:20:10.390 the expected return back down to the required 00:20:10.390 --> 00:20:13.789 9%. The pricing implication is simple. Find assets 00:20:13.789 --> 00:20:15.710 where the expected return exceeds the required 00:20:15.710 --> 00:20:18.119 return and buy them. So we've spent a lot of 00:20:18.119 --> 00:20:20.339 time admiring the simplicity and the utility 00:20:20.339 --> 00:20:22.980 of CAPM, but now we have to confront its intellectual 00:20:22.980 --> 00:20:25.660 foundation, because the model only works perfectly 00:20:25.660 --> 00:20:28.019 if the real world adheres to a set of extremely 00:20:28.019 --> 00:20:31.519 strict and often very unrealistic theoretical 00:20:31.519 --> 00:20:34.579 assumptions. Oh, reading this list of assumptions 00:20:34.579 --> 00:20:37.160 really makes it clear that CAPM is describing 00:20:37.160 --> 00:20:40.319 a financial utopia. It's a frictionless world 00:20:40.319 --> 00:20:43.559 where every single actor is a math genius. Let's 00:20:43.559 --> 00:20:45.660 start with the behavioral side of things. Well, 00:20:45.700 --> 00:20:47.579 the model requires investors to be perfectly 00:20:47.579 --> 00:20:50.000 rational and risk -averse. They're always trying 00:20:50.000 --> 00:20:52.140 to strictly maximize their economic utility. 00:20:52.460 --> 00:20:55.180 And quickly, they must share what are called 00:20:55.180 --> 00:20:57.920 homogenous expectations. What does that mean 00:20:57.920 --> 00:21:00.569 in plain English? It means every single investor 00:21:00.569 --> 00:21:02.910 has to agree on the probability distribution 00:21:02.910 --> 00:21:06.150 of potential returns and risks for every single 00:21:06.150 --> 00:21:08.769 asset. So if your expectation of Google's future 00:21:08.769 --> 00:21:11.690 growth differs from mine, the model starts to 00:21:11.690 --> 00:21:13.990 break down. That seems like an impossibility 00:21:13.990 --> 00:21:16.369 right there. Everyone processes information differently, 00:21:16.529 --> 00:21:18.430 but it gets even more restrictive, doesn't it? 00:21:18.490 --> 00:21:20.450 They must also have all information available 00:21:20.450 --> 00:21:22.890 at the same time, and no single investor can 00:21:22.890 --> 00:21:25.630 influence prices. They're all price takers. Then 00:21:25.630 --> 00:21:27.470 you hit the assumptions about the market structure 00:21:27.470 --> 00:21:31.069 itself. The market has to be absolutely frictionless. 00:21:31.190 --> 00:21:34.250 That means zero transaction costs, zero taxes. 00:21:34.410 --> 00:21:37.730 And assets must be perfectly divisible and perfectly 00:21:37.730 --> 00:21:40.150 liquid so you can trade fractions of a penny 00:21:40.150 --> 00:21:43.190 instantly and without any cost. Wait, let me 00:21:43.190 --> 00:21:45.250 stop you on the most famous or perhaps the most 00:21:45.250 --> 00:21:48.329 infamous assumption of them all. The ability 00:21:48.329 --> 00:21:51.329 for investors to lend and borrow unlimited amounts 00:21:51.329 --> 00:21:55.000 of money at the exact same risk -free rate. I 00:21:55.000 --> 00:21:56.579 mean, that just seems like a complete fantasy 00:21:56.579 --> 00:21:58.819 for the average person. It's a profound abstraction 00:21:58.819 --> 00:22:02.160 from reality. That one assumption, coupled with 00:22:02.160 --> 00:22:04.380 all the others, is what allows every investor 00:22:04.380 --> 00:22:06.960 to achieve an optimal, diversified portfolio 00:22:06.960 --> 00:22:09.619 on what's called the efficient frontier. It leads 00:22:09.619 --> 00:22:12.319 them all to hold the exact same market portfolio. 00:22:12.680 --> 00:22:15.180 It's necessary for the math to work. It's essential 00:22:15.180 --> 00:22:17.160 for the mathematics of the equilibrium to hold. 00:22:17.609 --> 00:22:20.549 But it is deeply, deeply disconnected from financial 00:22:20.549 --> 00:22:22.569 reality where lending rates are always higher 00:22:22.569 --> 00:22:25.349 than borrowing rates. And nobody, nobody can 00:22:25.349 --> 00:22:28.009 borrow unlimited money. So these nine core assumptions, 00:22:28.170 --> 00:22:31.069 they create this elegant theoretical construct. 00:22:31.470 --> 00:22:34.269 But it's one that demands a financial world that 00:22:34.269 --> 00:22:37.130 just doesn't exist. It's really no wonder the 00:22:37.130 --> 00:22:39.269 model started facing issues as soon as it hit 00:22:39.269 --> 00:22:42.529 real world data. Right. And because that assumption 00:22:42.529 --> 00:22:45.589 about the risk free asset. The ability to lend 00:22:45.589 --> 00:22:48.690 and borrow freely at that one rate was such a 00:22:48.690 --> 00:22:51.109 high hurdle. Fisher Black developed a really 00:22:51.109 --> 00:22:53.950 influential variation back in 1972 to get around 00:22:53.950 --> 00:22:57.730 it. This is the Black CPM or the zero beta CAPM. 00:22:57.769 --> 00:23:00.049 OK, so if the model can't rely on the existence 00:23:00.049 --> 00:23:02.809 of a perfect risk free asset that everyone can 00:23:02.809 --> 00:23:05.410 borrow against, how did Black adjust the formula? 00:23:05.650 --> 00:23:07.809 You replace the restrictive risk free rate, 10 00:23:07.809 --> 00:23:10.569 authors, with the return of a theoretical zero 00:23:10.569 --> 00:23:13.880 beta portfolio. We'll call it ERIZ. What exactly 00:23:13.880 --> 00:23:16.839 is a zero beta portfolio? It's a portfolio that's 00:23:16.839 --> 00:23:18.759 constructed from risky assets that when you combine 00:23:18.759 --> 00:23:22.000 them have a weighted average beta of zero. Therefore, 00:23:22.140 --> 00:23:23.839 it's a portfolio that has no correlation with 00:23:23.839 --> 00:23:25.539 the movements of the overall market. So it's 00:23:25.539 --> 00:23:27.180 not risk -free in the sense that it has no volatility. 00:23:27.539 --> 00:23:29.839 No, not at all, because it is made of risky assets. 00:23:30.420 --> 00:23:32.799 But its systematic risk contribution is nil. 00:23:33.069 --> 00:23:34.910 So instead of benchmarking everything against 00:23:34.910 --> 00:23:37.089 a government bond, you're benchmarking against 00:23:37.089 --> 00:23:39.549 a basket of stocks and bonds whose systematic 00:23:39.549 --> 00:23:42.009 movements just happen to cancel each other out. 00:23:42.549 --> 00:23:45.269 The Black KPM formula looks just a little bit 00:23:45.269 --> 00:23:48.430 different then. By substituting that zero beta 00:23:48.430 --> 00:23:51.250 return for the risk -free race, Black created 00:23:51.250 --> 00:23:53.769 a model that actually proved to be far more robust 00:23:53.769 --> 00:23:57.349 when you tested it empirically. It was particularly 00:23:57.349 --> 00:23:59.970 successful in addressing a major initial finding. 00:24:00.130 --> 00:24:02.849 Which was what? That the observed security market 00:24:02.849 --> 00:24:06.170 line in real markets was often flatter than the 00:24:06.170 --> 00:24:08.589 standard CAPM predicted it should be. Wait, why 00:24:08.589 --> 00:24:11.130 would the black CAPM lead to a flatter SML? That 00:24:11.130 --> 00:24:13.670 seems counterintuitive. Well, in the real world, 00:24:13.769 --> 00:24:16.130 the minimum rate you can earn without systematic 00:24:16.130 --> 00:24:18.230 risk, that's the zero beta portfolio return, 00:24:18.509 --> 00:24:21.130 is often higher than the theoretical risk -free 00:24:21.130 --> 00:24:23.910 rate. Because you're still taking on some unsystematic 00:24:23.910 --> 00:24:25.869 risk in that portfolio. OK, so the starting point 00:24:25.869 --> 00:24:28.210 of the line is higher. Right. And at the same 00:24:28.210 --> 00:24:31.029 time, high beta stocks often underperform what 00:24:31.029 --> 00:24:34.099 CAPM predicts. So by raising the baseline return 00:24:34.099 --> 00:24:36.359 and simultaneously acknowledging that the rewards 00:24:36.359 --> 00:24:39.059 for high beta are often lower in practice, the 00:24:39.059 --> 00:24:42.099 black KPM effectively pivots the SML, making 00:24:42.099 --> 00:24:44.680 it flatter and a much better fit for real -world 00:24:44.680 --> 00:24:47.539 observations. We've established the beauty, the 00:24:47.539 --> 00:24:50.160 elegance, and the very strict theoretical requirements 00:24:50.160 --> 00:24:53.460 of CAPM. Now we have to dive into the section 00:24:53.460 --> 00:24:56.180 that, for me, provides the most crucial insight 00:24:56.180 --> 00:24:59.269 for the learner. The undeniable evidence that 00:24:59.269 --> 00:25:01.950 despite its intellectual pedigree, the model 00:25:01.950 --> 00:25:04.210 simply does not work that well in the wild. And 00:25:04.210 --> 00:25:06.670 this failure, it's not a matter of mild academic 00:25:06.670 --> 00:25:08.690 debate. It's really the consensus view among 00:25:08.690 --> 00:25:11.049 leading financial economists. It's perhaps best 00:25:11.049 --> 00:25:13.029 summarized by Eugene Fama and Kenneth French, 00:25:13.210 --> 00:25:15.589 whose work fundamentally challenged CAPM. And 00:25:15.589 --> 00:25:17.609 what did they say? They stated in their 2004 00:25:17.609 --> 00:25:20.009 review, and this is the quote, the failure of 00:25:20.009 --> 00:25:22.710 the CAPM in empirical tests implies that most 00:25:22.710 --> 00:25:26.210 applications of the model are invalid. Wow. That 00:25:26.210 --> 00:25:28.529 is a staggering indictment for a Nobel Prize 00:25:28.529 --> 00:25:31.710 winning framework that's taught globally. It 00:25:31.710 --> 00:25:34.430 tells you right away that modern finance is built 00:25:34.430 --> 00:25:36.329 on a foundation that we all know is cracked. 00:25:36.930 --> 00:25:39.029 Let's look at some of those specific crack, the 00:25:39.029 --> 00:25:41.809 empirical failures. The most direct failure lies 00:25:41.809 --> 00:25:43.970 in that relationship between risk and return 00:25:43.970 --> 00:25:47.750 itself. CAPM predicts the relationship is linear 00:25:47.750 --> 00:25:50.730 and steep. You must take on proportionally more 00:25:50.730 --> 00:25:53.230 beta risk to get a proportionally higher return. 00:25:53.980 --> 00:25:56.599 But the data shows the S &L is often significantly 00:25:56.599 --> 00:25:59.440 flatter than predicted. And this flatness points 00:25:59.440 --> 00:26:01.980 directly to the biggest statistical anomaly that 00:26:01.980 --> 00:26:04.339 challenged CAPM's dominance, doesn't it? The 00:26:04.339 --> 00:26:06.259 low volatility anomaly. What does this pattern 00:26:06.259 --> 00:26:08.640 look like in practice? It means that if you analyze 00:26:08.640 --> 00:26:11.200 decades of market data, you find that low beta 00:26:11.200 --> 00:26:13.819 stocks, the supposed low risk defensive assets 00:26:13.819 --> 00:26:16.619 like utilities or consumer staples, often deliver 00:26:16.619 --> 00:26:19.000 higher average returns than the CAPM model says 00:26:19.000 --> 00:26:21.339 they should. They even outperform some high beta 00:26:21.339 --> 00:26:23.819 stocks. Which is a complete contradiction. It's 00:26:23.819 --> 00:26:25.180 a direct contradiction to the model's central 00:26:25.180 --> 00:26:28.079 promise that only high systematic risk gets a 00:26:28.079 --> 00:26:30.559 significant reward. But why would that happen? 00:26:30.720 --> 00:26:33.500 Why would investors accept lower returns for 00:26:33.500 --> 00:26:36.640 higher risk? If the model is wrong and low beta 00:26:36.640 --> 00:26:39.539 stocks outperform, why are people still piling 00:26:39.539 --> 00:26:42.039 into aggressive high beta tech stocks? That's 00:26:42.039 --> 00:26:44.460 the behavioral gap. Researchers have found that 00:26:44.460 --> 00:26:46.839 investors are often constrained. They may not 00:26:46.839 --> 00:26:48.839 be able to borrow money at the risk -free rate 00:26:48.839 --> 00:26:50.880 to leverage up their safe low beta portfolio. 00:26:51.500 --> 00:26:54.180 So instead, they gravitate toward high beta stocks, 00:26:54.500 --> 00:26:57.400 hoping to achieve high returns through volatility, 00:26:57.660 --> 00:27:00.900 even if those stocks are statistically overpriced 00:27:00.900 --> 00:27:03.440 relative to their true risk. It's also just optimism, 00:27:03.640 --> 00:27:05.680 right? It's optimism and the hope of massive 00:27:05.680 --> 00:27:08.740 outsized gains. The lottery ticket effect. So 00:27:08.740 --> 00:27:11.440 beta was supposed to be the single unifying factor 00:27:11.440 --> 00:27:13.559 that explained everything. But it was proven 00:27:13.559 --> 00:27:16.279 to be insufficient because it just couldn't explain 00:27:16.279 --> 00:27:18.500 historical outperformance in specific segments 00:27:18.500 --> 00:27:20.420 of the market. Right. And those are the famous 00:27:20.420 --> 00:27:23.180 size and value effects discovered primarily by 00:27:23.180 --> 00:27:26.000 Fama and French. They observed systematic historical 00:27:26.000 --> 00:27:29.240 outperformance from two specific types of stocks 00:27:29.240 --> 00:27:31.039 that. Beta just couldn't account for it. And 00:27:31.039 --> 00:27:33.579 those were? Small cap stocks companies with smaller 00:27:33.579 --> 00:27:35.779 market capitalization. That's the size effect. 00:27:36.059 --> 00:27:38.700 And value stocks, which are companies trading 00:27:38.700 --> 00:27:41.140 at low multiples of their book value, the value 00:27:41.140 --> 00:27:44.119 effect. These stocks consistently generated returns 00:27:44.119 --> 00:27:47.299 that were higher than CAPM predicted for their 00:27:47.299 --> 00:27:49.759 level of beta. And finally, there's the issue 00:27:49.759 --> 00:27:53.220 of time. The model treats beta as this reliable 00:27:53.220 --> 00:27:56.299 static constant. Yes. The convenience of the 00:27:56.299 --> 00:27:59.000 model requires treating beta as a single fixed 00:27:59.000 --> 00:28:02.339 number. But in the real world, beta is highly 00:28:02.339 --> 00:28:04.960 time varying. It changes. It changes constantly. 00:28:05.160 --> 00:28:08.700 A company in a cyclical industry, say heavy manufacturing, 00:28:09.079 --> 00:28:11.599 might have a very high beta during a recession 00:28:11.599 --> 00:28:14.500 when investor confidence is low, but a much lower 00:28:14.500 --> 00:28:16.920 beta during a boom when its earnings are stable. 00:28:17.680 --> 00:28:20.200 Relying on a historically calculated fixed beta 00:28:20.200 --> 00:28:22.819 for a valuation model just breaks down when that 00:28:22.819 --> 00:28:24.779 company's relationship to the market is changing 00:28:24.779 --> 00:28:27.319 dynamically with the economic cycle. Beyond just 00:28:27.319 --> 00:28:29.539 the statistical errors, some critics argue the 00:28:29.539 --> 00:28:31.880 entire CAPM structure is theoretically flawed 00:28:31.880 --> 00:28:33.880 from the ground up. And the most famous argument 00:28:33.880 --> 00:28:36.720 here is Rolle's critique from 1977. This is where 00:28:36.720 --> 00:28:38.759 it gets really fascinating. Rolle argued that 00:28:38.759 --> 00:28:41.599 the entire model is practically untestable. And 00:28:41.599 --> 00:28:44.000 the reason is that the anchor of the whole equation, 00:28:44.160 --> 00:28:48.279 the true market portfolio, is completely unobservable. 00:28:48.339 --> 00:28:50.539 Why is it unobservable? We use indices like the 00:28:50.539 --> 00:28:53.099 S &P 500 all the time as a proxy for the market. 00:28:53.240 --> 00:28:55.740 But that's just it. They're a proxy. The true 00:28:55.740 --> 00:28:58.220 theoretical market portfolio must include all 00:28:58.220 --> 00:29:01.700 assets held by all investors globally. Publicly 00:29:01.700 --> 00:29:04.220 traded stocks, sure, but also private equity, 00:29:04.380 --> 00:29:06.740 government bonds, real estate, precious metals, 00:29:06.900 --> 00:29:10.099 commodities, and most importantly... Human capital. 00:29:10.220 --> 00:29:12.539 Human capital. The present value of all future 00:29:12.539 --> 00:29:15.160 labor income for every person on Earth. We simply 00:29:15.160 --> 00:29:17.660 cannot construct a verifiable measure of this 00:29:17.660 --> 00:29:20.420 total wealth market portfolio. So when analysts 00:29:20.420 --> 00:29:23.099 use a proxy like the S &P 500, they aren't really 00:29:23.099 --> 00:29:25.400 testing CAPM. They're just testing whether the 00:29:25.400 --> 00:29:28.000 S &P 500 is efficient relative to itself, not 00:29:28.000 --> 00:29:31.259 relative to the true total market. Precisely. 00:29:31.279 --> 00:29:33.940 Roll's argument was that using any proxy leads 00:29:33.940 --> 00:29:37.380 to flawed mathematical inferences. If the true 00:29:37.380 --> 00:29:40.680 market portfolio is unobservable, then any test 00:29:40.680 --> 00:29:43.859 of CPM's validity is just circular and inconclusive. 00:29:44.200 --> 00:29:47.380 The model is perfect in theory, but its key variable 00:29:47.380 --> 00:29:49.920 can't be measured in reality, which makes definitive 00:29:49.920 --> 00:29:52.819 empirical confirmation impossible. There's also 00:29:52.819 --> 00:29:55.200 a more philosophical issue with how CPM even 00:29:55.200 --> 00:29:58.200 defines risk, isn't there? Yes. CPM assumes that 00:29:58.200 --> 00:30:00.839 variant volatility is the sole measure of risk. 00:30:01.000 --> 00:30:03.900 It treats upside volatility and downside volatility 00:30:03.900 --> 00:30:06.420 as equally bad. Which doesn't make intuitive 00:30:06.420 --> 00:30:09.180 sense. Not at all. It completely ignores asymmetric 00:30:09.180 --> 00:30:11.519 downside risk. Most investors don't mind volatility 00:30:11.519 --> 00:30:13.599 when a stock is going up. What they really fear 00:30:13.599 --> 00:30:15.599 is the probability of returns falling below a 00:30:15.599 --> 00:30:17.640 critical threshold, losing their capital. And 00:30:17.640 --> 00:30:20.140 this links to the safety first approach to investing, 00:30:20.319 --> 00:30:23.539 right? It does. Many practical investors prioritize 00:30:23.539 --> 00:30:27.019 minimizing the probability of a major loss rather 00:30:27.019 --> 00:30:30.819 than just minimizing overall variance. CAPM completely 00:30:30.819 --> 00:30:33.180 fails to capture this psychological preference 00:30:33.180 --> 00:30:36.440 for avoiding large losses, focusing instead on 00:30:36.440 --> 00:30:38.740 a symmetrical statistical measure that doesn't 00:30:38.740 --> 00:30:41.299 really reflect human fear. And finally, we have 00:30:41.299 --> 00:30:43.079 to look at the factors introduced by behavioral 00:30:43.079 --> 00:30:46.099 finance and just the messy reality of markets 00:30:46.099 --> 00:30:49.180 that completely defy CPM's neat assumptions. 00:30:49.799 --> 00:30:52.500 Let's start with the ideal world versus the reality 00:30:52.500 --> 00:30:55.500 of market frictions. The assumption of a frictionless 00:30:55.500 --> 00:30:58.519 world, no taxes, no transaction costs, perfectly 00:30:58.519 --> 00:31:01.500 divisible assets is just a non -starter. Real 00:31:01.500 --> 00:31:04.339 world trading costs. bid -ask spreads, capital 00:31:04.339 --> 00:31:07.779 gains taxes, regulatory restrictions. They all 00:31:07.779 --> 00:31:10.019 prevent investors from creating and maintaining 00:31:10.019 --> 00:31:12.960 the perfectly optimized, efficient portfolios 00:31:12.960 --> 00:31:15.279 that CAPM requires. And those frictions really 00:31:15.279 --> 00:31:17.480 matter. They matter greatly, especially in small 00:31:17.480 --> 00:31:20.059 transactions. They can completely erode the possibility 00:31:20.059 --> 00:31:22.539 of the perfect arbitrage that keeps the model 00:31:22.539 --> 00:31:24.789 in equilibrium. And then we introduce actual 00:31:24.789 --> 00:31:28.069 human beings who are anything but rational. Behavioral 00:31:28.069 --> 00:31:30.049 finance is arguably the biggest intellectual 00:31:30.049 --> 00:31:33.289 challenger to CEPM because it shows that our 00:31:33.289 --> 00:31:36.670 psychological biases drive significant market 00:31:36.670 --> 00:31:40.130 inefficiencies. Absolutely. Biases like investor 00:31:40.130 --> 00:31:43.589 overconfidence, hurting behavior, or just simple 00:31:43.589 --> 00:31:46.730 emotional trading. They all cause price deviations 00:31:46.730 --> 00:31:49.950 that the linear, purely rational CEPM cannot 00:31:49.950 --> 00:31:53.250 possibly explain. If investors are driven by 00:31:53.250 --> 00:31:56.190 emotion rather than utility maximization, the 00:31:56.190 --> 00:31:58.869 entire efficient market equilibrium that CAPM 00:31:58.869 --> 00:32:01.470 predicts just falls apart. A great example of 00:32:01.470 --> 00:32:03.809 this irrationality and how people structure their 00:32:03.809 --> 00:32:06.569 portfolios is mental accounting. CAPM requires 00:32:06.569 --> 00:32:09.089 the investor to view all of their assets as one 00:32:09.089 --> 00:32:12.569 single optimized portfolio. But in reality, people 00:32:12.569 --> 00:32:14.210 silo their money, don't they? They have a safe 00:32:14.210 --> 00:32:16.470 retirement account, a speculation fund and a 00:32:16.470 --> 00:32:18.190 rainy day savings account. Right. They create 00:32:18.190 --> 00:32:20.579 them all differently. They manage the risk profile 00:32:20.579 --> 00:32:24.000 of each silo separately, which completely contradicts 00:32:24.000 --> 00:32:26.240 the fundamental assumption that they're managing 00:32:26.240 --> 00:32:29.920 one unified, risk -optimized entity to maximize 00:32:29.920 --> 00:32:32.420 their total wealth. They apply different risk 00:32:32.420 --> 00:32:34.799 tolerances to different pools of money. And this 00:32:34.799 --> 00:32:36.420 leads to something called skewness preference, 00:32:36.680 --> 00:32:38.900 which I think perfectly explains why people buy 00:32:38.900 --> 00:32:40.980 lottery tickets even though they have a terrible 00:32:40.980 --> 00:32:43.799 expected return. That's the lottery effect. Investors 00:32:43.799 --> 00:32:46.380 may willingly accept a lower expected return 00:32:46.380 --> 00:32:49.519 for assets that exhibit high positive skewness. 00:32:49.640 --> 00:32:52.220 That just means there is a small but high impact 00:32:52.220 --> 00:32:55.160 chance of an enormous rare payout. And the dream 00:32:55.160 --> 00:32:58.319 of that huge win is enough. The potential for 00:32:58.319 --> 00:33:01.700 that huge win compensates them for the statistically 00:33:01.700 --> 00:33:05.380 lower average return. CAPM, which only looks 00:33:05.380 --> 00:33:07.579 at variance, completely ignores this powerful 00:33:07.579 --> 00:33:10.460 preference, which means it systematically undervalues 00:33:10.460 --> 00:33:12.960 stocks that have lottery -like characteristics. 00:33:13.480 --> 00:33:16.680 So the incredible utility of CPM's framework, 00:33:16.900 --> 00:33:19.319 combined with its glaring empirical failures, 00:33:19.539 --> 00:33:21.960 meant that finance couldn't just abandon it. 00:33:22.019 --> 00:33:24.890 It had to try and fix it. The failures we just 00:33:24.890 --> 00:33:27.029 talked about in part five were the direct engine 00:33:27.029 --> 00:33:29.309 that drove the creation of more complex multi 00:33:29.309 --> 00:33:31.269 -factor models. They just had to acknowledge 00:33:31.269 --> 00:33:33.710 that a single measure beta was not sufficient 00:33:33.710 --> 00:33:35.990 to describe the complexity of market returns. 00:33:36.190 --> 00:33:38.569 Right. And the most famous and influential extension, 00:33:38.789 --> 00:33:40.769 which is often considered the modern academic 00:33:40.769 --> 00:33:44.150 successor to key APM, is the Fama -French three 00:33:44.150 --> 00:33:47.089 -factor model. And this model was a direct response 00:33:47.089 --> 00:33:49.289 to those empirical failures we discussed, wasn't 00:33:49.289 --> 00:33:51.589 it? Specifically, the size and value anomalies. 00:33:51.750 --> 00:33:54.349 Exactly. Instead of relying solely on market 00:33:54.349 --> 00:33:57.230 risk or beta, Fama, and French, proposed adding 00:33:57.230 --> 00:33:59.670 two new independent risk factors to the equation. 00:33:59.930 --> 00:34:02.710 What are those factors? The first factor is SMB, 00:34:02.809 --> 00:34:05.950 which stands for small minus big. It accounts 00:34:05.950 --> 00:34:08.070 for the historical premium that we've observed 00:34:08.070 --> 00:34:09.909 in small cap stocks. That's the size effect. 00:34:10.309 --> 00:34:13.650 The second factor is HML. or high minus low, 00:34:13.869 --> 00:34:16.650 which captures the premium associated with value 00:34:16.650 --> 00:34:19.590 stocks, the value effect. So the Fama French 00:34:19.590 --> 00:34:22.699 model is essentially saying Your required return 00:34:22.699 --> 00:34:25.260 is determined not just by your sensitivity to 00:34:25.260 --> 00:34:27.840 the overall market, but also by your exposure 00:34:27.840 --> 00:34:30.480 to the systematic risk associated with being 00:34:30.480 --> 00:34:32.880 a small company and being a value company. That's 00:34:32.880 --> 00:34:34.940 it. And by including these three independent 00:34:34.940 --> 00:34:37.800 sources of risk, the model significantly improved 00:34:37.800 --> 00:34:39.920 the ability to explain the cross -section of 00:34:39.920 --> 00:34:41.960 expected stock returns. It just fit the data 00:34:41.960 --> 00:34:44.420 better. And we also saw extensions designed to 00:34:44.420 --> 00:34:46.500 capture that behavioral flaw we mentioned, the 00:34:46.500 --> 00:34:49.639 skewness preference. The co -skewness CAPM directly 00:34:49.639 --> 00:34:52.199 addresses that lottery. effect. It acknowledges 00:34:52.199 --> 00:34:54.960 that human desire for the rare massive payout. 00:34:55.039 --> 00:34:58.320 Exactly. This extension includes a term for co 00:34:58.320 --> 00:35:01.440 -skewness as a priced factor. It adds higher 00:35:01.440 --> 00:35:03.900 order statistical moments to the equation, which 00:35:03.900 --> 00:35:05.980 is a fancy way of saying it incorporates the 00:35:05.980 --> 00:35:09.199 idea that investors are willing to accept a statistically 00:35:09.199 --> 00:35:12.099 inferior return if the distribution of those 00:35:12.099 --> 00:35:15.199 returns is positively skewed, offering that lottery 00:35:15.199 --> 00:35:18.150 like chance. So it's trying to model the irrationality 00:35:18.150 --> 00:35:21.210 that the standard CAPM just dismissed. That's 00:35:21.210 --> 00:35:23.329 a great way to put it. And that other major theoretical 00:35:23.329 --> 00:35:26.090 extensions address the limitations CAPM imposed 00:35:26.090 --> 00:35:29.010 regarding investor time horizons and even the 00:35:29.010 --> 00:35:32.869 very definition of utility. Robert Merton's intertemporal 00:35:32.869 --> 00:35:37.349 CAPM or ICAPM is key here, isn't it? It generalizes 00:35:37.349 --> 00:35:40.030 the model beyond just a single period. The original 00:35:40.030 --> 00:35:43.070 CAPM was inherently a static single period model. 00:35:43.170 --> 00:35:44.769 Yeah. It was optimized for a short timeframe. 00:35:45.260 --> 00:35:47.699 IGCM acknowledges that investors don't just optimize 00:35:47.699 --> 00:35:49.800 their portfolio today. They save over decades 00:35:49.800 --> 00:35:51.900 and need to manage risk over multiple periods, 00:35:52.119 --> 00:35:54.380 allowing for repeated portfolio rebalancing. 00:35:54.500 --> 00:35:56.760 It seems so obvious. The intertemporal aspect 00:35:56.760 --> 00:35:58.800 just means that investors care about the uncertainty 00:35:58.800 --> 00:36:01.000 of future investment opportunities as well as 00:36:01.000 --> 00:36:03.710 their current wealth. So ICPM simply acknowledges 00:36:03.710 --> 00:36:05.750 that we don't buy a house and then sell it tomorrow, 00:36:05.989 --> 00:36:09.269 we say for 40 years, and that long time horizon 00:36:09.269 --> 00:36:12.150 fundamentally changes our view of risk. That's 00:36:12.150 --> 00:36:14.389 the practical implication. And the consumption 00:36:14.389 --> 00:36:18.570 CAPM or CCAPM takes the redefinition of utility 00:36:18.570 --> 00:36:22.010 even further. Instead of optimizing your final 00:36:22.010 --> 00:36:24.989 wealth, CCPM argues that investors really seek 00:36:24.989 --> 00:36:27.840 to maximize their consumption utility. So they 00:36:27.840 --> 00:36:30.260 don't care about their portfolio value fluctuating 00:36:30.260 --> 00:36:31.920 as much as they care about their ability to afford 00:36:31.920 --> 00:36:33.820 goods and services in the future, their lifestyle. 00:36:34.159 --> 00:36:37.559 Precisely. CCPM suggests that an asset's required 00:36:37.559 --> 00:36:40.699 return should be based on how its returns covery 00:36:40.699 --> 00:36:43.820 with aggregate consumption, not just market wealth. 00:36:44.389 --> 00:36:46.769 Assets that perform poorly when overall consumption 00:36:46.769 --> 00:36:49.329 is low, like during recessions when people are 00:36:49.329 --> 00:36:51.409 struggling, are considered much riskier because 00:36:51.409 --> 00:36:53.630 they fail you when you need them most. And therefore, 00:36:53.730 --> 00:36:55.690 they have to offer a higher return. They must 00:36:55.690 --> 00:36:58.090 offer a higher required return to compensate 00:36:58.090 --> 00:37:00.329 for that risk. And finally, we noted the issue 00:37:00.329 --> 00:37:02.409 of horizon misalignment when it comes to the 00:37:02.409 --> 00:37:04.829 risk -free rate. This is a really practical point. 00:37:05.150 --> 00:37:07.849 For long -term investors, the choice of the appropriate 00:37:07.849 --> 00:37:11.670 risk -free rate, Shia Peer, really changes. If 00:37:11.670 --> 00:37:14.099 you are saving for retirement, decades from now, 00:37:14.260 --> 00:37:16.780 a short term Treasury bill rate isn't the most 00:37:16.780 --> 00:37:19.599 relevant benchmark. So what should you use? Well, 00:37:19.679 --> 00:37:22.119 long term investors might optimally choose assets 00:37:22.119 --> 00:37:25.500 like inflation linked bonds as their true risk 00:37:25.500 --> 00:37:28.559 free asset because their primary goal is to maintain 00:37:28.559 --> 00:37:30.960 purchasing power over a very long time horizon. 00:37:31.469 --> 00:37:33.929 This means the intercept of the SML isn't fixed. 00:37:34.090 --> 00:37:36.289 It depends entirely on the investor's individual 00:37:36.289 --> 00:37:39.329 timeline and objectives. It seems like the more 00:37:39.329 --> 00:37:41.769 we learn, the more we realize that every single 00:37:41.769 --> 00:37:45.150 simple component of that original elegant CAPM 00:37:45.150 --> 00:37:48.329 formula had to be complicated, adjusted, or completely 00:37:48.329 --> 00:37:50.690 replaced just to try and align the theory with 00:37:50.690 --> 00:37:53.130 the real world. So what does this all mean for 00:37:53.130 --> 00:37:55.440 you, the learner? We've completed our deep dive 00:37:55.440 --> 00:37:57.400 into the capital asset pricing model, moving 00:37:57.400 --> 00:37:59.320 from its magnificent theoretical structure all 00:37:59.320 --> 00:38:01.260 the way to its systematic empirical breakdown. 00:38:01.539 --> 00:38:03.860 Its core purpose to define required return based 00:38:03.860 --> 00:38:06.519 solely on systematic risk or beta changed finance 00:38:06.519 --> 00:38:09.519 forever. And despite all those powerful critiques, 00:38:09.559 --> 00:38:12.760 the low volatility anomaly, the size and value 00:38:12.760 --> 00:38:16.159 effects discovered by Fama and French, and Roll's 00:38:16.159 --> 00:38:18.860 powerful argument that the true market portfolio 00:38:18.860 --> 00:38:22.980 is unobservable, CAPM remains the essential foundation. 00:38:23.590 --> 00:38:26.210 It provides the intellectual framework, the vocabulary, 00:38:26.269 --> 00:38:28.510 and the starting point for nearly all modern 00:38:28.510 --> 00:38:31.559 valuation. Its failure was constructive in a 00:38:31.559 --> 00:38:33.820 way. It was incredibly constructive. It led directly 00:38:33.820 --> 00:38:35.840 to the creation of the multi -factor models that 00:38:35.840 --> 00:38:38.300 we use today. It is the lingua franca evaluation 00:38:38.300 --> 00:38:41.280 used daily across corporate finance to justify 00:38:41.280 --> 00:38:43.760 discount rates and asset allocations. It's the 00:38:43.760 --> 00:38:46.039 simple, beautiful, but ultimately incomplete 00:38:46.039 --> 00:38:48.820 map that forced us to acknowledge the vastly 00:38:48.820 --> 00:38:51.300 more complicated terrain of real financial markets. 00:38:51.619 --> 00:38:53.639 And that leads to a final provocative thought 00:38:53.639 --> 00:38:56.809 for you to consider. The elegance of the original 00:38:56.809 --> 00:38:59.309 CAPM required us to assume that every single 00:38:59.309 --> 00:39:02.989 investor is a perfectly rational, utility -maximizing 00:39:02.989 --> 00:39:05.449 machine, given that the greatest contribution 00:39:05.449 --> 00:39:08.550 of CAPM wasn't its accuracy and prediction, but 00:39:08.550 --> 00:39:10.650 rather the fact that its failure forced us to 00:39:10.650 --> 00:39:13.150 define, measure, and then ultimately break the 00:39:13.150 --> 00:39:15.610 rules of rationality by discovering behavioral 00:39:15.610 --> 00:39:18.449 anomalies. What does that imply about the human 00:39:18.449 --> 00:39:20.710 element that we will always fail to fully model 00:39:20.710 --> 00:39:22.869 in our search for the perfect investment? That's 00:39:22.869 --> 00:39:25.070 a profound thought. The human brain, not the 00:39:25.070 --> 00:39:27.349 beta, might just be the final unpriced factor. 00:39:27.489 --> 00:39:29.130 Thank you for joining us on the Deep Dive.