After a couple of longer posts, hopefully this week’s will be a bit shorter. One of my hobbies is running. Okay, given the speed at which I run, maybe jogging would be more appropriate. I follow the motto that if you can’t run fast, run far, so ultramarathons tend to be my distance of choice.1 Plus, the food (burgers, hot dogs, and even BBQ ribs…no pie yet though) in ultras is much better.
This means that I spend a lot of time running around the neighborhood, and almost all of that time is spent listening to podcasts, audiobooks and music. Unfortunately, there are train tracks, highways, and a regional airport all near my running routes. Add in delivery trucks, regular trucks, louder cars, lawnmowers, weedeaters, dogs barking, etc. and the noise is enough to drive me crazy at times (which, unfortunately, is not a long drive). Surely, this can’t be good for you, right?
Problems Caused by Noise Pollution
Noise pollution creates problems with both physical and mental health. On the physical health side there is the obvious potential hearing damage from loud noises. This is why you often see people operating heavy equipment wearing ear protection. According to the CDC, noise above 70 decibels over a sustained time frame is enough to start causing hearing damage. How loud is 70 dB? According to this article, 70 dB is equivalent to your washing machine or dishwasher.2
However, hearing is only a small part of the problems caused by noise as the following passage from “Noise Pollution Isn’t Just Annoying — It’s Bad for Your Health” states:
Even small increases in unwanted ambient sound have significant effects. In 2011, for example, scientists studying people living near seven major European airports found that a 10-decibel increase in aircraft noise was associated with a 28 percent increase in anxiety medication use. Another study found that people living in areas with more road traffic noise were 25 percent more likely than those living in quieter neighborhoods to have symptoms of depression. Similarly, people exposed to noise pollution were found to be significantly more likely to have heart problems like atrial fibrillation compared to those unaffected by noise.
Noise can also interfere with sleep patterns, causing people to get up earlier than intended or have problems falling asleep. An article on the impact of nighttime sleep mentions that
Noise seems to lengthen stage 1 sleep and decrease both stage 3 and stage 4 sleep. It also may trigger alarm signals in your body like the hormones adrenaline and cortisol. Your heart rate and blood pressure may rise. This happens even though you don’t wake up. In effect, your body is guarding you while you rest.
Note that stage 4 sleep is often referred to as REM (Rapid Eye Movement) and is essential for processing emotions and memories.
Clearly, we have a lot of evidence that too much noise is not good for either our hearing or mental health. It also may cause enough stress to the system to impact our physical health. All of this is a big argument to occasionally unplug and get into a quieter, more relaxed environment. As fall approaches, I sense a short backpacking trip in my future.
Noise in Business and Finance
Noise also has a definition from statistical sense related to prediction. Daniel Kahneman (insert reverent bow here to one of the masters of behavioral finance) has recently written a book with Oliver Sibony and Cass Sunstein titled “Noise: A Flaw in Human Judgment”.3 If we think of error in estimating something, we can split that error into two components — bias and noise. When students learn about linear regression, they are likely to encounter the term BLUE which stands for Best Linear Unbiased Estimator (picture me having nightmare flashbacks to my econometrics class from 30 years ago). Bias refers to a consistent error. For example, if your scale was set five pounds too low, you might feel good every time you step on the scale, but the reality would be inaccurate. Fortunately, it is an easy error to fix because the solution is to just add 5 pounds to every reading. Now imagine the same scale was unbiased (on average, it gets your weight correct), but sometimes it is 5 pounds too heavy, sometimes it is 7 pounds too light. This is a bigger problem. You could step on the scale 100 times every morning and then take an average, but that seems like a tedious process.
How does noise impact investing decisions? Remember that according to traditional finance models, managers are trying to allocate capital based on the present value of all expected future cash flows. However, as discussed in the False Precision weekly update, this may require more suspension of disbelief than believing that Sylvester Stallone actually killed 115 people in the process of saving Colonel Trautman in Rambo 3.4
In a column by Olivier Dessaint, Thierry Foucault, Laurent Frésard, Adrien Matray from 2019, the authors make the point that there are three primary reasons why firms may make poor investment decisions (investing in projects with a negative net present value). They are
Management investing in projects which management “knows” to be poor investments, but are taking advantage of investors’ irrational estimates of the potential cash flows. We might think of Nikola’s prototype truck rolling downhill to create the illusion of a functional prototype as an example of this. It is clear that management knew the demonstration was an illusion, but they were hoping to use it to generate investor interest. These types of investments may fall under the “fake it until you make it” framework or progress to outright fraud.
Market mispricings may cause some stocks to be overvalued relative to what management believes the stock is worth (note that no one KNOWS what the stock is worth). In this case, management may issue more shares to obtain cheaper capital and use this capital to invest. An example of this could be AMC Entertainment Holdings which raised almost $600 million in a secondary offering at $50.85 in early June, 2021. The stock was trading under $3 per share in December, 2020.
Managers can use feedback from stock price reactions to competitor announcement to help fine-tune their expectations regarding future cash flows. For example, if Netflix announces that they are going to spend an additional $17 billion on original content for 2021 and the stock price rises by 10%, this provides a signal to Amazon, Disney, etc. that investors see this spending on developing original content as a positive NPV project. This may make Netflix’s competitors more likely to increase their spending on similar activities. Alternatively, if the stock price falls by 10%, Netflix’s competitors may cut back a bit on their planned spending.
Dessiant and colleagues examine the relationship between mutual funds reducing exposure to stocks (essentially creating selling pressure) and competitor capital spending. What they find is that the stock price noise created by selling caused competitors to reduce their capital spending by an average (across all firms in the sample) of about $29 billion per year for the firms in their sample. While this is only a few percent cut in investment decisions, it is evidence of the power of noise.
Noise or Missing Variable
One idea that has intrigued me for awhile is attempting to explain the noise. Theoretically, if your explanatory model is accurate, you won’t be able to. However, this requires us to have an accurate model that captures the impacts (both joint and non-linear) of every variable that has an influence on what we are trying to explain. For example, let me go back to the topic that started today’s post — running. Let’s say that I’m trying to model the performance of an ultramarathon to predict finishing times. I need to look at
Past performance of entrants
Their age and if the past performance is trending up or down
What the course conditions are like (is it relatively flat or hilly, is it smooth or technical, etc.)
What the temperatures are like
How much training have they done for the race
And the list can go on and on. However, I’m going to miss something (or more accurately some things as there are too many potential variables to capture it all.) So, what things am I missing? Maybe one of the entrants has another important race scheduled 4 weeks out and really only has the ability to excel in one of them. If they get off to a less than ideal start, do they take a DNF? Maybe one of the entrants is going through some personal issues and is mentally not focused. All of these (and many more) factors make it impossible to perfectly explain what happened. This leads us back to narrative bias where our brain tries to fit explanations.
However, what if there are some things that our model doesn’t capture well. David Epstein discusses the concept of responders vs. non-responders in his book “The Sports Gene: Inside the Science of Extraordinary Athletic Performance”. This concept states that different people have different responses to training stimulus. Some respond quickly and then plateau, some hardly respond at all, and others respond slowly (although may have a bigger long-term impact). This same concept also applies to medical treatments where different people have different responses to medications. According to a recent article in “The Guardian”, the tendency for pharmaceutical testing to be done primarily on males has created problems after medicines have been approved.
Of the 10 prescription drugs taken off the market by the US Food and Drug Administration between 1997 and 2000 due to severe adverse effects, eight caused greater health risks in women. A 2018 study found this was a result of “serious male biases in basic, preclinical, and clinical research”.
We even see it in the recent COVID-19 pandemic where some people are asymptomatic, some have minor symptoms, some get long-haul after effects, and some end up with severe cases requiring hospitalization or even death. Some of these are related to various comorbidities, but it is not a perfect relationship. Ultimately, being able to explain/predict what is behind this “noise” may lead to a better understanding of the world around us.
Author introduces shorter post with a discussion of ultrarunning. Does this win me an award for creative uses of irony?
I imagine (obviously) that varies depending on how close you are to washing machine, but sporting events such as a football game can reach that volume and cause hearing issues in as little as 15 minutes.
This book is on my “too read” list. As I just finished “Thinking Fast and Slow”, Noise may well be next.
According to this Screenrant article, Rambo kills 493 people over the course of the Rambo series (552 if you include the kills in Vietnam as part of his history). Quite a body count! By comparison, John Wick has a “mere” 299 kills in his 3 movies.