橡树资本 Oaktree

wrote a column ·

"Howard Max's memo-Cognitive Illusion", September 2022

Memo: to Oak Capital (OaktreeCapital) customers

Poster: Howard Marks (HowardMarks)

[subject: cognitive illusion]

Since writing my first memo in February 1993, predicting the value of where the rain comes from (The Value of Projections, or Where'd All This Rain Come From), I have been saying that I will "Forecast”leave。

In the years since then, I have explained in detail why I am not interested in forecasting-some of my favorite quotes in the following chapters echo my disdain for forecasting-but I have never written another memo to explain why it is so difficult to make useful macro predictions. that's why I have this memo.

A thought-provoking thing

There are two kinds of prophets in the world: those who do not know the future, and those who do not know what they do not know.

-- John Kenneth Galbraith

Shortly after the final touch-up memo dared to I Beg to Differ, I attended a luncheon with some experienced investors and outsiders. This is not a social event, but an opportunity for people present to exchange views on the investment environment.

During this period, the host asked a series of questions: how do you expect inflation to develop? Will there be a recession, and if so, how serious is the situation? How will the conflict between Russia and Ukraine end? What do you think of the future situation in Taiwan? What is the likely impact of the US elections in 2022 and 2024? In this regard, I have heard all kinds of views.

Readers who have followed my memo for a long time should be able to imagine what I thought at the time: "No one in this room is an expert in foreign affairs or politics." No one here has a particularly deep insight into these topics, and certainly no more than the ordinary people who read the news this morning. " The ideas conveyed, even on economic issues, do not seem to be more convincing than others, and I absolutely believeNo one can improve the investment results, which is the key.。

It was at that lunch that I started thinking about writing another memo about the uselessness of the macro outlook. Soon after, I found some additional material-a book, an article from Bloomberg Opinion, and a newspaper article-all of which supported my argument (or mine).Confirmation deviation"--that is, people tend to accept and believe information and arguments that can prove their previous views. That luncheon, along with the material, inspired the theme of the memo: many reasons why predictions are rarely beneficial.

In order to get something useful, whether in manufacturing, academia, or even in the field of art, there must be aReliableTo convert the required input into the desired output In short, the problem is that I don't think a process can consistently combine a large number of variables related to the economy and financial markets (InputInto useful macro predictions (Output）。

Model

The greatest enemy of cognition is not ignorance, but the illusion of cognition.

-- Daniel Burstein

In about my first decade at First National City Bank, there was a word that was popular at the time but I haven't heard for a long time now: econometrics. Specifically, it refers to the practice of looking for relevance in economic data to produce effective forecasts. Or in short, econometrics studies how to build mathematical models of the economy. In the 1970s, econometrics were hot, but I think they are no longer popular. I think that means their model doesn't work.

Whether the model is complex or scribbled simple, mathematically based or intuitive, forecasters have no choice but to judge based on the model. By definition, the model consists of assumptions: "if A happens, then B will happen." let me put it another way,The model states the relationship and response.. But if we are willing to adopt the output of the model, we must be convinced that the model is reliable. But when I thought of modeling the economy, my first reaction was how complicated it would be.

For example, the United States has a population of about 330 million. With the exception of the very young and some very old, the rest are participants in the economy. As a result, there are hundreds of millions of consumers, as well as millions of workers, producers and middlemen (many meet multiple categories). To predict the path of economic development, it is necessary to predict the behavior of these people-if not each participant, at least the total number of groups. The real simulation of the US economy must deal with billions of interactions or nodes, including interactions with suppliers, customers, and other market participants around the world. Is it possible to do this?

For example, can you predict what consumers will do in the following situationsBehavior(1) if they get an extra dollar of income (what is the "marginal propensity to consume"? (II) if energy prices rise, squeezing other categories in the household budget; (III) if the price of one commodity rises relative to other commodities (will there be a "substitution effect"? ); and (4) what if the geopolitical arena is disturbed by events on other continents?

Obviously, this complexity requires the frequent use of simplified assumptions.

For example, if you can assume that if B is not better or cheaper (or both), consumers will not buy B instead of A, then modeling will be easier. It is also helpful if it is assumed that the cost of producing X is not less than that of Y, then the producer will not price X less than Y. But despite the higher price of B (and even because of this), consumers are still attracted by B's brand effect, what will be the result? What if X is produced and developed by entrepreneurs who are willing to lose money for several years to gain market share? Is it possible for the model to predict decisions that consumers are willing to spend more and entrepreneurs are willing to make less money (or even lose money)?

In addition, the model must predict the behavior of each group of participants in the economy in a variety of environments. But there are many unpredictable factors. For example, consumers may behave in one way at one time and in a different way at another similar moment. Given the large number of variables involved, it seems impossible for two "similar" moments to occur in exactly the same way, and we are unlikely to see economic participants exhibit the same behavior. In addition, the behavior of the participants will be affected by their psychology (or should I say their emotions? ), and their psychology may be affected by qualitative, non-economic development. How are these modeled? How can an economic model be comprehensive enough to deal with situations that have never been encountered before, or situations that have not occurred in modern times (that is, in comparable situations)? This is another example of how models cannot simply replicate something as complex as the economy. Of course, one of the typical examples is the COVID-19 epidemic. It shut down most of the world's economies, subverted consumer behaviour and spurred massive government bailouts. Which aspect of the existing model can predict the impact of the epidemic? Yes, the world experienced an epidemic in 1918, but the situation was so different (there was no iPhone, Zoom calls, etc.) that the economic situation at that time was almost incomparable to that of 2020.

In addition to factors such as complexity and psychological fluctuations and dynamic processes that are difficult to capture, it is also necessary to take into account the limitations of trying to predict things that cannot be expected to remain unchanged. Shortly after I started writing this memo, I received a wonderful weekly magazine from Morgan Housel. One of the articles describes many observations in other areas related to our economy and investment. The following two are borrowed from the field of statistics, and I think they are related to the discussion of economic models and forecasts ("Little Ways the World Works", Morgan Housel,Collaborative Fund,2022 July 20, 2000):

Stationarity: this is an assumption that the main factors that affect the system do not change over time, which assumes that history can be used as a guide for future statistics. If you want to know how high a dam will be built, look at flood data from the past 100 years and assume that it will be the same in the next 100 years. Stationarity is a wonderful, science-based concept that is valid until it fails. It is the main driving force of important events in economy and politics. [but in our world,] "things that have never happened before are happening all the time," said Scott Sagan, a professor at Stanford University. Cromwell's Law: never say that something will not happen. Even if there is only a 1/1000000000 chance that something will come true, and you will interact with billions of things in your life, you will almost certainly experience some shocking accidents and should always be open to the possibility that incredible things will become reality.

Cromwell's Law: never say that something will not happen. Even if there is only a 1/1000000000 chance that something will come true, and you will interact with billions of things in your life, you will almost certainly experience some shocking accidents. and should always be open to the possibility of incredible things becoming a reality.

Stationarity may be a reasonable assumption in the field of physical science.

For example, because of the law of gravitation, objects can always fall at the same acceleration under given atmospheric conditions. The result is always like this, and always will be. But in our field, few processes are smooth, especially considering psychological, emotional, and human behavior, and they change over time.

Take the relationship between unemployment and inflation as an example. For the past 60 years or so, economists have relied on the Phillips curve, which holds that wage inflation will rise as unemployment falls, because employees gain bargaining power when there are fewer unemployed workers. and can successfully negotiate higher wages.

For decades, the 5.5% unemployment rate was also thought to indicate "full employment." But the unemployment rate fell below 5.5 per cent in March 2015 (and reached a 50-year low of 3.5 per cent in September 2019) but did not rise significantly until 2021.Phillips curveThe important relationships described have been applied to various economic models established over the decades, but it does not seem to apply for most of the past decade.

Cromwell's law is equally important.Unlike physical science there are very few things that absolutely must or cannot happen in markets and economics. So in Mastering the Market Cycle, I list seven terms that investors should remove from their vocabulary: "never", "always", "never", "will" and "must".

But if these words really have to be discarded, so must the idea of building models that can reliably predict the macro future. In other words, in our field,Almost nothing is immutable.， The unpredictability of behavior is my favorite topic.。

Famous physicist Richard Feynman (Richard Feynman) once said: "Imagine how difficult physics would be if electrons had a sense."the rules of physics are reliable precisely because electrons always do what they are supposed to do. They will never forget to perform their duties. They never resist, they never strike, they never innovate, they never act in the opposite way.But none of this applies to participants in the economy, and it is because they are not applicable that the behavior of participants is unpredictable.If the behavior of participants is unpredictable, how to model the operation of the economy? We are talking about the future, and there is no way to predict the future without making assumptions. Minor errors in economic environment assumptions and subtle changes in the behaviour of participants can cause serious problems. As mathematician and meteorologist Edward Lorenz famously said, "A Brazilian butterfly flapping its wings can trigger a tornado in Texas." (historian Neil Ferguson (Niall Ferguson) mentioned this in an article discussed below. )

To sum up, can we think that the economic model is reliable? Can the model replicate reality? Can it describe the behavior of millions of participants and their interactions? Is the process of trying to model reliable? Can these processes be simplified to mathematics? Can mathematics capture the qualitative nuances of people and their behaviors? Can the model predict changes in consumer preferences, changes in corporate behavior and the response of participants to innovation? In other words, can we trust the output of the model?

Obviously, economic relations are not static, and the economy is not governed by the schematic diagram (the schematic diagram that the model tries to simulate). So, for me, the bottom line is that without violating the hypothesis, the output of the model points in the right direction most of the time. But it can't always be accurate, especially at critical moments such as inflection points. And this is when accurate prediction is the most valuable.

Input

One fact that cannot be ignored is that all your knowledge is about the past and all your decisions are about the future.

-Ian Wilson (Ian H.Wilson) (former General Electric Co executive)

After considering the incredible complexity of the economy and the need to make simplified assumptions (which will reduce the accuracy of any economic model), let's consider the input required for a model-the raw materials for manufacturing forecasts. Is the estimated input valid? Can we understand them well enough to make meaningful predictions? Or does it simply remind us of the ultimate truth about the model: "input garbage, output garbage or garbage"? Obviously, no prediction is of better quality than the input on which it is based. Here's what Neil Ferguson wrote on Bloomberg Opinion on July 17:

Consider when we ask, "has inflation peaked?" What we really want to ask about this question is not just the supply and demand of 94000 different goods, manufactured goods and services. We are also concerned about the future interest rate path set by the Fed, and apart from the much-touted "forward guidance", it is far from clear where it will go. What we are asking is how long the dollar will remain strong because it is currently driving down the price of US imports. But there are more questions to be answered.

At the same time, the above questions are also indirectly asking how long the conflict between Russia and Ukraine will last, as the chaos caused by the conflict between Russia and Ukraine has significantly exacerbated inflation in energy and food prices since February. We are asking whether oil-producing countries such as Saudi Arabia will respond to requests from Western governments to increase crude oil production. We may also have to ask ourselves what impact the latest novel coronavirus Omicron BA.5 will have on the western labour market. UK data show that BA.5 is 35 per cent more contagious than its predecessor BA.2, while BA.2 is more than 20 per cent more contagious than the original Omicron. If you want to add all these variables to your model, I wish you luck. In fact, the future path of inflation, like the future trend of the conflict between Russia and Ukraine and the path of the spread of the COVID-19 epidemic, cannot be determined.

I find Ferguson's article very relevant to the subject of this memo, so I attach a link to it here. This article puts forward a lot of important points, although I disagree in some aspects. Ferguson mentioned above, "in fact, the future path of inflation, like the future direction of the conflict between Russia and Ukraine and the spread path of the COVID-19 epidemic, are uncertain." I think accurate prediction of inflation is "less likely" than the other two problems (if it can be predicted), because accurate prediction of inflation requires correct predictions of these two events and a thousand other factors. How can anyone do all these things right?

I would like to give a brief introduction to the forecasting process mentioned in the value of Forecast:

I think, for most fund managers, the process goes like this: "I predict that the economy will do A. If A happens, the interest rate should show B. If the interest rate is B, the stock market should show C. In this environment, the best performing sector should be D, while stock E should rise the most. " Then the investment group is constructed based on this, in order to achieve the best performance in this case. But anyway, what are the chances of E? Remember, E is conditional on A, B, C and D. In the field of forecasting, a 2/3 accuracy would be a remarkable achievement. But if there is a 67% chance that each of the five forecasts will be correct, the result is that all five forecasts are correct and there is a 13% chance that stocks will perform as expected.

Predicting event E based on assumptions about A, B, C and D is what I call single-scenario prediction. In other words, if the hypothesis about A, B, C or D turns out to be wrong, then the prediction of E is unlikely to be realized. Only if all potential predictions are correct can E get consistent results such as predictions, but this is extremely rare.

If you do not consider (I) other possible outcomes of each element, (II) the possibility of other scenarios, (III) what are the prerequisites for making one of the assumptions a reality, and (iv) what is the impact on E, then no one can invest wisely.

Ferguson's article raises an interesting question about economic modeling: what assumptions should we make about the macro environment of economic participants? This question just shows an endless cycle: in order to predict the overall performance of the economy, we need to make assumptions about consumer behavior and so on. But to predict consumer behavior, don't we need to make assumptions about the overall economic environment?

In my first memo on the epidemic, No one knows (II) (Nobody Knows II) (March 2020), I mentioned that when discussing coronaviruses, Harvard epidemiologist Mark Lipschich (Marc Lipsitch) said: (I) facts; (II) well-founded inferences drawn from analogies of other viruses, and (III) opinions or speculations. This is our standard practice when dealing with uncertain events. In economic or market forecasts, we have a great deal of history and many similar past events to infer (but there is no COVID-19 epidemic). But even if these things are used as input by a well-constructed prediction model, they are still unlikely to predict the future. They can be useful material or rubbish. To illustrate this point, people often ask me which cycle I have experienced in the past is most similar to the present.

My answer is that there are brief similarities between current developments and some past cycles, but there are no absolute similarities. In each case, the differences are huge and outweigh the similarities. Even if we can find the same previous period, to what extent should we rely on this single sample? I think the answer is not much. Investors rely on historical references (and the forecasts they make based on them) because they worry that without them, they will act blindly, but that doesn't mean they are reliable.

Unpredictable effects

Prediction creates a mirage in which the future is knowable.

-- Peter Bernstein

If we do not first determine whether our world is orderly or random, we cannot consider the rationality of the prediction.

In short, is it completely predictable, completely unpredictable, or somewhere in between?

For me, the conclusion is betweenBetween the two, but more inclined toUnpredictableSo much so that most predictions are useless. Since our world is predictable at some times and unpredictable at other times, what is the use of prediction if we can't tell when it is predictable and when it is unpredictable? I learned a new word from reading Ferguson's article: "deterministic". The Oxford Dictionary defines it as "determined by previous events or natural laws of cause and effect". The world is much easier when we deal with things according to the rules. Just like Feynman's electrons. But it is clear that the economy and markets are not governed by the laws of nature-thanks to human participation.

Previous events may be "laying the groundwork" or "tend to repeat", but they rarely happen twice in the same way. Therefore, I think the processes that make up the economy and the operation of the market are not deterministic, which means that they are unpredictable. In addition, the input is obviously unreliable.

Many are random, such as weather, earthquakes, accidents and deaths. Others involve political and geopolitical issues. Some we know, and some have not yet surfaced. In the Bloomberg Bloomberg Opinion article, Ferguson mentioned the British writer GK Chesterton (GKChesterton GKChesterton). This reminds me of Chesterton's famous quote in my revisit to Risk Revisited Again (June 2015):

The real problem in our world today is not that the world is irrational, nor is it a rational world. The most common problem is that the world is almost rational, but not entirely rational. Life is not a contradiction, but it is a trap for logicians. It looks slightly more precise and regular than it really is; its precision is obvious, but its imprecision is hidden; and its wildness is lurking.

Returning to the luncheon introduced on the first page, the host's opening remarks are roughly as follows: "in recent years, we have experienced events such as the COVID-19 epidemic, the amazingly successful Fed rescue policy, and the conflict between Russia and Ukraine. This is a very challenging environment because all of this comes all of a sudden. "

For him, I think, this means that attendees should let themselves get rid of the inaccuracy of the 2020-2022 predictions and continue to predict the future and bet on their own judgment. But my reaction is completely different: "there are many events that affect the current environment." Isn't the fact that no one can predict any of them enough to convince the people present that they should give up the prediction? "

To give another example, let's think back to the fall of 2016. There are two things that almost everyone believes: (1) Hillary Clinton will be elected president; (2) if Donald Trump is elected for some reason, the market will collapse. In spite of this, Trump won and the market soared. The past six years have had a profound impact on the economy and markets, and I believe that any traditional prediction of the 2016 election will not be correct at that time. Isn't that enough to make people believe: (1) we don't know what will happen in the future, (2) we can't understand how the market will react to what happens?

Can the forecast lead to excess?

It is not ignorance that gets us into trouble, but fallacies that seem to be correct.

-- Mark Twain

As I mentioned in my recent memo on Macro thinking (Thinking About Macro), in the 1970s, we described economists as "investment directors who never entered the market."

In other words, economists make a lot of predictions that prove them right or wrong, and then they move on to make new forecasts, but they don't track how often they are right (or they don't publish statistics). Can you imagine hiring a fund manager without reference to your track record (or if you were a fund manager, could you imagine being hired in this situation)?

But economists and strategists do not lose their jobs because they do not publish statistics, perhaps because clients are always willing to pay for their forecasts. Are you a consumer of these predictions? Are forecasters and economists employees of your company? Or do you subscribe to their publication and invite them to give briefings, just like my previous employer? If so, do you know that everyone predicts the correct frequency? Have you found a way to strictly determine which of these predictions can be relied on and which should be ignored? Is there a way to quantify the contribution of these forecasts to your return on investment? I asked this series of questions because I have not seen or heard of any research in this area. It is inconceivable that there is a dearth of global information on whether macro forecasts will bring excess returns, especially when compared with the number of people who need such information.

Despite the lack of evidence to prove its value, macro forecasts continue. Many forecasters are part of the equity fund management team or are providing advice and forecasts to these teams. One thing we can be sure of is that actively managed stock funds have been losing market share for decades because of their poor performance, replaced by index funds and other passive investment instruments. actively managed funds now account for less than half of the US stock mutual fund market. Is it the reason that macro forecasting is not helpful to investment in nature?

As far as I know, the only thing that can find quantitative information on this issue is the performance of so-called macro hedge funds. The hedge Fund Research Group (HFR) publishes the hedge fund weighted composite index and some sub-strategy indices. The following is the long-term performance of the hedge fund weighted composite index, the macro hedging strategy index and the S & P 500 index.

* performance as of July 31, 2022. The hedge fund index shown is the weighted composite index of each fund.

In the table above, according to HFR, the average performance of hedge funds during the study period was much lower than that of the S & P 500, while the average performance of macro hedge strategy funds was much worse (especially between 2012 and 2017). Given that investors continue to entrust about $4.5 trillion to hedge fund managers, these funds must offer benefits other than returns, but it is not clear what this will be.

This seems especially true for macro hedge funds. To confirm my view of predictions, I'm going to give you a rare example of self-assessment: a seven-page feature article entitled "I was wrong" in the Sunday View column of the New York Times on July 24. In the article, eight New York Times opinion columnists made public their erroneous predictions and biased advice. The most relevant here is an article written by Paul Krugman (Paul Krugman) entitled "I misread inflation (I Was Wrong About Inflation.)" It's a confession. I took some of these excerpts and strung them together:

At the beginning of 2021, economists had a heated debate about the possible consequences of the American bailout plan. I was on the side of not worrying too much about the impact of inflation. Of course, it turned out to be a very bad decision. History cannot lead us to expect such overheated inflation. So there's something wrong with my model... One possible reason is that history is misleading. In addition, the disturbances created to adapt to the epidemic and its consequences may still play a large role. Of course, the conflict between Russia and Ukraine and the epidemic prevention and control measures in major Chinese cities have undoubtedly pushed the degree of interference to a whole new level. In any case, the whole thing has become a lesson of humility. Incredibly, the standard economic model has been working quite well since the 2008 financial crisis, and I thought there was no problem using the same model in 2021. In retrospect, I should have realized that in the new world trend that emerged after the COVID-19 epidemic, this inference itself was risky.

I admire Krugman for showing such astonishing frankness (although I have to say, I don't remember many market forecasts optimistic enough to depict the actual situation in the next decade). Krugman's explanation of his mistake is good in itself, but I don't see him mention abandoning modeling, inference or prediction in the future.

This modesty may even permeate the Federal Reserve, one of the world's largest economic forecasters, where there are more than 400 PhDs in economics. Here's what economist Gary Shilling wrote in Bloomberg Opinion on August 22:

The Fed's forward guidance has become a disaster, challenging its own credibility. Chairman Jerome Powell Powell seems to agree that the outside world should stop speculating about the Fed's views on interest rates, economic growth and inflation at different points in the future.

The fundamental problem with forward guidance is that it relies on data, which itself comes from the Fed's poor forecasts in the past. The Fed has been overly optimistic about the recovery after the Great Recession of 2007-2009. In September 2014, policy makers forecast real GDP growth of 3.40% in 2015, but were forced to keep lowering their forecasts to 2.10% by September 2015. The federal funds rate is not a market-determined interest rate, but is set and controlled by the Fed, and no one challenges the Fed's authority.

In addition, members of the Federal Open Market Committee (FOMC) are notoriously poor at predicting what actions they will take. In 2015, their average forecast for the federal funds rate in 2016 was 0.90% and 3.30% in 2019. The actual figures are 0.38% and 2.38% respectively. To be sure, many of the events that are happening have created market uncertainty, but the Fed's forward guidance has been highly sought after and important. Recall that earlier this year, the Fed also considered inflation caused by the epidemic and friction to restart the economy after supply chain disruptions as temporary. It was not until later that the Fed found out that things were going badly and reversed course, raising interest rates and signalling further sharp increases. The Fed's erroneous forecasts led to wrong forward guidance and exacerbated volatility in financial markets.

On this question, I would like to raise one last point, that is, where are those who gain fame (and become rich) through macro-point of view? It is certainly impossible for me to know everyone in the investment community, but among the people I know or know, I think there are only a few very successful "macro investors". When there are few examples of something, as my mother once said, "the exception confirms the rule." The rule in this example is thatMacro forecasts rarely lead to excellent performance. To me, the extraordinary success story proves that this statement is a universal truth.。

Forecast demand of practitioners

Forecasts reveal the forecasters more than they reveal the future.

-- Warren Buffett

How many people can make macro predictions that are valuable most of the time? I don't think so. How many investment managers, economists and forecasters have tried? Thousands, to say the least. This raises an interesting question: why make predictions? If macro forecasts will not contribute to investment success over time, why are there so many practitioners in the investment management industry who believe in forecasts and flock to their results? I think one of the typicalReasonIt could be:

It's part of the job.

Investors always do this.

Everyone I know does this, especially my competitors.

I've been doing this all the time-I can't stop now.

If I don't do this, I won't be able to attract customers.

Since investment involves deploying capital to benefit from future events, how can you expect to do a good job without a view of those events? We need predictions, even if they are not perfect.

This summer, my son Andrew recommended me to read a very interesting book, "making mistakes (but it's not my fault): why do we make excuses for stupid beliefs, bad decisions, and hurtful actions" (Mistakes Were Made (but Not by Me): Why We Justify Foolish Beliefs, Bad Decisions, and Hurtful Acts), written by psychologists Carol Tavris and Eliot Aronson. The theme of the book is self-defense. The authors explain that "cognitive dissonance" occurs when people are faced with new evidence to question their previous positions, and when this happens, the subconscious makes them try their best to prove and defend their previous positions. Here are some selected clips:

If you have a set of beliefs that guide your practice, and you know that some of them are incorrect, you must either admit that you are wrong and change your approach, or reject new evidence. Most people, when faced directly with the evidence that they have done wrong, do not change their views or plans of action, but refute them more stubbornly. Once we recognize a belief and prove its wisdom, it is obviously hard work to change our minds. It is much easier to put new evidence into the existing framework for psychological argument for acceptance than to change the framework.

The mechanisms commonly used by people to respond to evidence that calls their beliefs into question include these (paraphrasing the author's words):

Unwilling to listen to discordant messages

Selectively remember parts of their lives and focus on those that support their views

Act with a cognitive bias so that people only see what they want to see and seek some kind of confirmation of what they already believe.

I believe that these are the factors that cause people to continue to make predictions and rely on them. In this case, what kind of expression will there be?

Regard macro forecasting as an integral part of investment

Keen to recall correct forecasts, especially those that are bold and non-market consensus

Overestimate the correct rate of prediction

Forget or play down false predictions

Do not keep records of prediction accuracy or fail to calculate the average success rate

Pay attention to the generous returns that are rewarded to accurately predicted

Emphasize that "everyone does it."

Perhaps most importantly, blame unsuccessful predictions on being blinded by random or exogenous events. (but, as I said before, this is the crux of the problem: if predictions become inaccurate so easily, why make predictions? )

Most people, even honest people with a good heart, will take positions or actions that are in line with their own interests, sometimes at the expense of others or objective truth. They are not aware of the situation themselves, but think that what they are doing is the right thing; they have also sought a lot of legitimate reasons.

As Charlie Munger (Charlie Munger) often quotes Demosthenes, "nothing is easier than to deceive yourself." Because people always believe what they want. "

I don't think forecasters are liars or charlatans. Most of them are smart intellectuals who think they are doing something useful. However, self-interest enables them to act in a certain way, while self-defense enables them to stick to their opinions in the face of evidence to the contrary. As Morgan Housel said in a recent newsletter:

The inability to predict the past has no effect on our willingness to predict the future. Certainty is so precious that we never give up on it, and if people honestly face how unpredictable the future is, most people can't get out of bed in the morning. (from Great Faith, Collaborative Fund, August 24, 2022)

On my birthday a few years ago, Richard Masson, the co-founder of Oak Tree, gave me an interesting gift that suited his style. The gift was a bound edition of the New York Times. I've been hoping for a chance to write about my favorite subtitle for the October 30, 1929 issue, and the Dow Jones Industrial average has just fallen nearly 23% in two days.

The headline reads, "Banker optimism (Bankers Optimistic)" (however, the Dow fell by about 85 per cent in the next three years). Most bankers and fund managers seem to be inherently optimistic about the future. Besides, it is in their best interest because it helps them to do more business. But their optimism must have led to their predictive views and the resulting behavior.

Yes or no?

"I never think about the future-because it's coming soon."

-- Albert Einstein

Consider the following aspects of macro projections:

The number of assumptions / inputs required

Number of processes / relationships to be included

The inherent unreliability and instability of these processes, and

The role of randomness and the possibility of accidents.

The most important thing for me is that predictions cannot always be correct to the extent that they are valuable. I have mentioned it many times, but for the sake of completeness, I would like to reiterate my views on the effectiveness of macro predictions (or, more accurately, futile):

Most predictions consist of inferences about past performance.

Since macro developments usually do not deviate from previous trends, inferences are usually successful.

On this basis, most predictions are correct. However, since inference is usually expected by the price of securities

Those who are based on inference expectations do not enjoy excess benefits when the inference is established.

Occasionally, economic behavior does deviate substantially from the pattern of the past. Because this deviation is different from that of most investors

It is expected that its emergence will affect the market, which means that the accurate prediction of deviation will bring huge profits.

However, since the economy does not often deviate from past performance, few can accurately predict deviations, and

Most deviations from the predictions turned out to be wrong.

Therefore, we have (a) to infer that most of the forecasts are correct, but will not generate excess benefits, and

(II) potentially profitable deviation forecasts, which are rarely correct and therefore usually do not produce

Excess benefit.

It has been demonstrated that most forecasts do not increase returns.

At the luncheon mentioned at the beginning of this memo, people were asked about expectations such as Fed policy and how this affected their investment positions. One person replied: "I think the Fed will still be highly worried about inflation, so it will raise interest rates sharply, leading to a recession." So I chose to avoid risk. " Another said: "I expect inflation to slow in the fourth quarter and the Fed to turn to doves in January. Begin to cut interest rates and stimulate the economy. I am very optimistic about 2023. " We often hear such a saying. But it must be recognized that these people are using a single-factor model: the speaker's prediction is based on a single variable. Speaking of simplified assumptions: these forecasters implicitly believe that everything except the Fed's policy remains the same. When they need to play three-dimensional chess, they are still playing flat checkers. Aside from the impossibility of predicting Fed behaviour, the impact of inflation on such behaviour, and the market reaction to inflation, are there other important considerations? If a thousand things play a role in determining the future direction of the economy and markets, what are the other 999 things? What about the impact of wage negotiations, mid-term elections, the conflict between Russia and Ukraine and oil prices?

The truth is that people can only remember very limited things in their minds at any time. It is difficult to take a large number of factors into account, and it is even more difficult to understand how a large number of things will interact (relevance is always a real thinking problem).

Even if you somehow manage to get the right economic forecast, that's only half the success. You still need to predict how economic activity will translate into market results. This requires a completely different prediction and involves numerous variables, many of which are related to psychological factors and are therefore almost unknowable.

According to student Warren Buffett, Ben Graham once stated: "in the short term, the market is a voting machine, but in the long run, it is a weighing machine." How to predict the short-term choices of investors? Some economic forecasters have concluded that the actions announced by the Federal Reserve and the Treasury in March 2020 will save the US economy and help the economy recover. But I don't know anyone who predicted a hot bull market before the recovery began. As I said earlier, Buffett shared his views on macro forecasts with me in 2016. "for a piece of information to be used effectively, it must meet two criteria: first, it must be important, and secondly, it must be knowable."

Of course, the macro outlook is important. Now, investors seem to be taking hold of every forecaster's comments, macro events and signals of intermittent Fed tightening. Unlike when I was in the industry in the early days, it seems that macro factors are everything, and corporate development has received less attention.

But I strongly agree with Buffett that the macro future is unknowable, or at least few people know more consistently than investors, which is the key to trying to gain a cognitive advantage and make excellent investment decisions.

Buffett is clearly at the top of the list of successful investors, shunning macro forecasts and paying more attention to "micro" areas: companies, industries and securities than others.

What on earth is an article I wrote in 2001 called Alpha? What's It All About, Alpha?) The concepts of "agnosticism" school and "agnosticism" school are introduced in the memo of "agnosticism", which are elaborated in "We and them" (Us and Them) in 2004. At the end of the current memo, I will insert some of what I wrote in the latter about these two genres:

Over the years, most of the investors I have met belong to the "agnosticism" school. This was the case when I analyzed stocks during the 1968-1978 period, even when I moved to non-mainstream investments but still worked for equity-centric investment management companies between 1978 and 1995.

It is easy to identify members of the "agnosticism" school:

They believe that understanding the future direction of the economy, interest rates, markets and widely watched mainstream stocks is essential to the success of investment.

They are confident that this can be achieved.

They know they can do it.

They know that a lot of people are trying to do this, but they think either (1) everyone can.

To succeed at the same time, either (b) only a few people can do it, but they are one of them.

They are willing to invest according to their views on the future.

They are also happy to share their views with others, although correct predictions should be worth thousands of dollars, and no one will give them away for free.

They rarely look back on the past and seriously review their achievements as predictors.

"self-confidence" is the key word to describe the members of the school. On the other hand, for the school of agnosticism, the word, especially when looking at the macro future, should be "Caution”。

Its believers usually think that it is impossible to predict the future; it is not necessary to predict the future; the correct goal should be to do their best to invest on the basis of admitting that they do not have this knowledge.

As a member of the "agnosticism" genre, you can comment on the future (and perhaps someone else will take notes). You may be sought after and regarded as an ideal dinner guest. Especially when the stock market is rising. If you join the "agnosticism" school, the result will be more complicated. You will soon get tired of expressing "agnosticism" to friends and strangers.

Before long, even relatives will no longer ask you about the trend of the market. You will never enjoy the 1/1000 surprise moment when the prediction comes true, nor will you enjoy the joy of publishing your picture in the Wall Street Daily. On the other hand, you are also exempt from forecasting errors and from losses based on investments based on excessive confidence in the future.

But how do you think it feels when potential customers ask you about your investment prospects and you have to say "I don't know"? For me, the best bottom line criterion for which school comes from the famous saying of the late Amos Tversky, a behavioral scientist at Stanford University: "To realize that you may not know that something is terrible, but what is even more frightening is to realize that, in general, the world is run by people who firmly believe that they know exactly what is happening.。”

In the investment management business, macro forecasts are made, shared upon request, and entrusted to invest for clients, which is, of course,Standard practice. It seems customary for fund managers to believe in forecasts, especially their own. As mentioned above, it seems out of place not to do so. But are their beliefs realistic? I'd love to hear what you have to say.

Years ago, a well-respected sell-side economist (an old acquaintance of mine at Citi) called me: "you changed my life," he said. "I don't make predictions anymore.. Instead, I'm just telling people what happened today and what I think might have an impact on the future.Life has been better ever since.。”

Can I help you achieve the same state of happiness?

September 8, 2022

This material expresses the author's views as of the date shown, which may change at any time without notice. Oak has no responsibility or obligation to update the information contained in this material. In addition, Oaktree does not state that past investment performance is an indicator of future performance, and you are not allowed to make such assumptions. In fact, whenever there is an opportunity to make a profit, there is also the possibility of loss.

This material is for reference only and may not be used for any other purpose. The information contained in this material does not constitute or be construed as an offer to provide consulting services or to sell or solicit the purchase of any securities or related financial instruments in any jurisdiction. Some of the information contained in this material relating to economic trends and performance is based on or drawn from information provided by independent third-party sources. Oaktree Capital Management, L.P. ("Oak") believes that the source of the information obtained is reliable, but it cannot guarantee the accuracy of the information, nor does it independently verify the accuracy or completeness of the information or the assumptions made based on it.

All or any part of this material (including the information contained in this material) may not be copied, reproduced, republished or published in any way without Oak's prior written consent. To the extent that this material or some of its contents are in Chinese, these Chinese translations are for reference only, and if there is any difference between the Chinese translation and the English version, the English version shall prevail. Oaktree Capital (Hong Kong) Limited can provide the English version of this material. Oak, its affiliates or any of their respective managers, partners, employees, affiliates, shareholders or agents (I) shall not be liable for any inaccuracies, errors or omissions in the translation of this material, and (ii) shall not be obliged to notify any recipient of any inaccuracies, errors or omissions in the translation.

932K Views