There’s been a lot in the media about AI - especially in the last year. The stock market has been on a tear - apparently due to expectations AI will bring a new golden era of profit. Companies producing AI software and infrastructure have been touting its capabilities, and it has also been described as dangerous for killing jobs, enabling criminal activity, and eating up resources due to all the new data centers being built.

I decided I wanted to research how smart it really is and where it may be heading in the future. In this article I won’t address the potential impacts to the economy. However, I do address these in my other articles including: Supercritical (Part 8): Summary of our Debt Situation in the U.S. and The 2028 Intelligence Crisis: A Structural Stress Test of the U.S. Economy

I have long known about the neural network technology involved in modern Artificial Intelligence or “AI”. Being an engineer, I took a course put on by Wright-Patterson AFB way back in the 80’s on it. At the time, in the course they were envisioning its use primarily for automatically identifying warfare target images. Much later, around 2015, I built some rudimentary Matlab/Simulink models using neural networks. At the time, this technology seemed more like a research curiosity or a military project than having much of any commercial potential.

Starting about 2023 I knew of this thing now being marketed as AI for commercial use by OpenAI - called ChatGPT. I thought it was pretty much just a fancy user interface enhancement to a search engine.

It wasn’t until early 2025 that I realized it was far more than this. I found it could write a full research-type paper based on scant information from me. Another shock was applying it to help solve a very strange and intermittent car problem (occasional near stalling, rough running). The car dealer was going down their usual path assuming it was a fuel pump, etc. ChatGPT said it was most likely the oil viscosity was too thin. Huh? Then I looked back at all the oil change records for the car. On the recent oil change the dealer used a lighter weight oil than in the past. It was not an incorrect oil weight - just the thinnest recommended oil weight for the model. The dealer thought we were nuts when we told them to put in the heavier oil. But… It fixed the problem! How did ChatGPT know this? At the time I thought it somehow found this looking at car forums - but I was wrong.

Recently another odd and intermittent car problem (different car) cropped up. The mechanics were convinced they knew what it was - the high-pressure fuel pump. And they went through an explanation of how it could cause the symptoms. They fixed the fuel pump, test drove it and declared victory. I hit up both ChatGPT and Gemini about all this. Both said the shop likely did not fix the problem we brought it in for and it was most likely due to a flaky “CKP” sensor. Sure enough, when picking up the “fixed car” the problem reoccurred. I told the shop to replace the CKP sensor. Guess what - the car was really fixed now. Again, the shop was amazed - how did I know it was this sensor? I didn’t tell them about AI.

This time and in prep for this article, I decided to dig into exactly how the AIs came up with a conclusion it was due to the CKP sensor. Again, I thought the AIs were looking at car forums on the internet, but that is not how they did it. It turns out AI used a combination of its HUGE training data (which set the “weights” of neural network connections) combined with sophisticated reasoning logic. It’s training data includes how cars work down to the all the individual parts - including CKP sensors and CKP sensor technology. Then it utilized Bayesian reasoning (see note below) to determine it was the CKP sensor that was the most likely culprit.

There was no need for AI to find a forum post describing this exact failure. Its training had already encoded enormous amounts of information about automobiles, sensors, failure modes and diagnostic reasoning. It could combine that knowledge with the evidence I provided and determine that the CKP sensor was the most likely culprit.

The AIs conclusions were purely based on its internal neural network parameter weights. So, these AI’s thought it through - on their own - and came to a logical conclusion based on the evidence. These are Cognitive Thinking Machines.

Note: Bayesian reasoning is the process of updating the probability that something is true as new evidence becomes available. In modern AI, this type of reasoning can emerge from its trained neural network rather than being explicitly programmed in.

How Much AI Has Progressed and Where It’s Going

It felt to me that AI had suddenly exploded on the scene in 2024 - 2025, even though the first release of ChatGPT occurred in 2022. Earlier versions of what could be called “modern AI” were created as early as 2019. The main question I wanted to answer was if this cognitive capability rose up quickly from 2019 (and how quickly) to where it is today. And even more important, how far can it progress in the future? For this, a key comparison should frame this in terms of human cognitive capability.

I knew these questions would not be easy to answer. I assumed there would be plenty of academic studies that could be used to help answer these questions, but what is out there on this? And then if I Google-searched these, how difficult would it be to distill the data? I am guessing it would take me months to try to figure it all out - assuming I ever could.

Naturally the solution was to employ AI itself (ChatGPT’s latest version) to do the heavy lifting. I knew it could create an academic-quality paper almost instantly, and its massive training data combined with logical reasoning capabilities are immense.

Some criticism I got from one reader was “AI is just going to tell you what you want to hear” and “it will hallucinate answers”. Those are valid concerns, and I will go into why that happens later. But being a pretty decent engineer myself, I have been “trained” for over 40 years on how to spot bad or weak arguments regarding technology (note my skepticism on the shop diagnosis of the fuel pump). So my main roles with the AI on this project were: 1) Have it justify, to my satisfaction, that intermediate results and conclusions in all the steps toward the final results are reasonable and valid and 2) Form answers and conclusions that will be understandable and resonate with my readers.

The first thing to define was what exactly was I trying to measure? I decided it should be “Cognitive Capability”. Cognitive Capability is defined as the ability to acquire and integrate information, reason from it, and determine effective solutions to familiar and novel problems. Note what doesn’t qualify:

• A big database

• A search engine

• A calculation algorithm

The thinking machine will use the above as “tools” to come to determine effective solutions, but these by themselves are not defined as Cognitive Capability. The same applies to humans also - not just AI.

Process Used to Determine the Rise of Cognitive Capability in AI

Measuring AI cognitive capability is not straightforward. There is no universally accepted equivalent of an IQ test for AI, and individual benchmarks eventually become saturated as AI systems improve. Therefore, this process did not rely on any single test. Instead, several independent sources of AI performance data were examined to identify a common underlying trend.

Measuring the Historical Trend

Several major AI benchmark systems were examined, each measuring somewhat different aspects of cognitive capability.

Epoch AI’s Epoch Capabilities Index (ECI) combines results from more than 50 benchmarks into a statistical measure of general AI capability. This is particularly useful because it reduces dependence on any one benchmark as individual tests become obsolete or saturated.

METR’s Task-Completion Time Horizon takes a very different approach. Rather than asking how many test questions an AI answers correctly, METR measures the difficulty of tasks AI agents can successfully complete in terms of how long those same tasks take skilled humans. This provides one of the best long-term measures for increasing AI problem-solving capability.

ARC-AGI was also examined because of its emphasis on novel abstraction and reasoning problems, while Stanford’s HELM provided useful performance data extending back to earlier generations of language models.

These measures use different scales and test somewhat different capabilities, so their results cannot simply be averaged together. Instead, overlap periods, correlations and common trends were examined. Where appropriate, statistical transformations were used to place older benchmark data onto a common capability scale.

The important result was that these substantially different approaches produced a remarkably consistent picture: AI cognitive capability has been increasing exponentially.

Using METR as the primary historical backbone, the best-fit historical trend indicates a doubling of cognitive capability approximately every 5.7 months, with an “R-Squared” of approximately 0.96 (note - this means how well the trend matches the data with 1.0 being perfect). The most recent data suggest that the rate may have accelerated further, although there is not yet enough evidence to determine whether this faster rate will persist.

A resulting Machine Cognitive Capability Index (MCCI) was created and normalized so that “frontier AI” (frontier means the leading edge) in 2026 has a value of 100. The absolute value of 100 is arbitrary; it is the relative change in capability over time that is important.

Why the Historical Curve Was Not Simply Extrapolated

An exponential trend doubling every 5.7 months cannot continue indefinitely. In fact, if it did, it would consume 100% of today’s entire global electrical generation by 2029! Obviously this won’t happen - the growth rate must bend first.

The physical resources supporting AI development were therefore examined separately, including training compute, available AI computing capacity, training data, algorithmic efficiency, inference-time computation and data-center electrical power.

Many of these inputs have themselves been increasing at extraordinary rates. Frontier training compute, global AI computing capacity and AI data-center power have all grown exponentially during portions of the recent AI expansion.

There are, however, unavoidable physical constraints. Electrical generation, transmission infrastructure, semiconductor manufacturing and data-center construction cannot continue multiplying several-fold every year indefinitely. The supply of useful training data also cannot expand forever at historical rates.

The future portion of the MCCI model therefore deliberately assumes that cognitive growth slows substantially. In the central scenario, the effective doubling time increases from roughly six months today to approximately one year by 2028, two years by 2030, four years by 2032 and eight years by 2035. A wide range of faster and slower scenarios was also calculated to represent the considerable uncertainty in any long-term projection.

Comparing AI With Humans

The final step was to give the MCCI scale an intuitive human reference. Four levels of human cognitive capability were independently estimated:

H1 — Average Human
H2 — Typical Domain Expert
H3 — World-Class Domain Expert
H4 — Collective Frontier Experts

The first three levels were estimated primarily from studies in which humans and AI systems performed the same or closely comparable intellectual tasks, including medicine, mathematics, programming, scientific reasoning and difficult problem solving.

H4 required a different approach because the objective was to estimate the effective cognitive capability of the world’s leading experts working collectively on a difficult problem. Research on collective intelligence, expert teams, coordination efficiency, parallel problem solving and the diminishing returns from adding additional people was used to estimate this level.

One point is particularly important: the human capability levels were estimated independently of the MCCI trajectory before being placed on the graph. The human values were not selected to produce particular AI crossover dates.

Only after the machine capability trajectory and human capability levels had been independently estimated were the two combined.

The Results

After all the data from the studies was digested using the extensive process described above, the following summary graph was generated and is shown below.

This chart tells us that AI’s cognitive capabilities likely exceeded the average human in 2023 and a typical expert, such as a doctor or engineer in 2024. A world-class expert, such as a leading world-renowned physics expert, was likely surpassed in late 2025. And the combined expertise of all the world’s leading experts will likely be passed before 2029!

It’s hard to get an appreciation from this chart of how much AI cognitive capability has risen because it is on a log scale. In a way, it is truly frightening to look at this on a linear scale from 2019 - 2026 shown below.

And even more frightening, a linear chart with the central scenario projection to 2035 shown below:

How Could AI Possibly Grow Like This?

These charts showing this huge exponential growth of AI cognitive capability in just a few years, where doubling has been occurring every 5.7 months (and possibly even shorter in the last year), is very foreign to humans. We have never experienced such a technological growth of this magnitude and this potentially disruptive - ever. And now, even with reduced exponential growth anticipated due to resource limitations, the rate is still huge as shown in the 2019 - 2035 linear scaled chart.

But how could AI overtake humans on cognitive capability so quickly? AI technology is loosely based on the way the human brain works - using a neural network. A neural network uses neurons - arranged in one or more layers - which each activate based on inputs from a few or many synapse connections to other neurons. Each connection has a “weight” that governs how much it is affected by another neuron. As a network, whether an “AI brain” or human brain, it learns by adjusting these weights. Simple networks, like the ones I used to model on Matlab/Simulink, may only have several dozen neurons and maybe one or two neuron layers. Modern large AI models may have more than 100 interconnected neural-network layers. With the human brain there are 86 billion neurons and maybe 100 trillion synaptic connections. Modern AI systems may have (only!) a trillion parameters, each which is loosely analogous to a synaptic connection. From a raw numbers standpoint, it seems the human brain should be far more capable than AI - but as we have seen from the data this is not currently the case.

AI Advantages

There are several characteristics of AI that allow cognitive capability to scale in ways the human brain cannot.

Reasoning: AI employs exceptional reasoning capabilities including Bayesian, inductive, deductive, probabilistic, and analogical reasoning. On the other hand, humans can be very flawed in reasoning, and this is primarily tied to the factor immediately described below.

Emotions: AI is a machine and doesn’t have emotions. While most will think emotions are a good thing, it’s not-so-good for logical reasoning. The problem is that they can easily override reasoning. Ego, pride, greed, revenge, pleasure, bias, etc. can and will derail reasoning. For example, even a highly respected physics professor is potentially subject to feeling disrespected. For example, he might reject new important information from a junior colleague that should change his previous “expert conclusion.” Thus derailing the proper application of Bayesian reasoning.

Speed: Biological neurons operate on millisecond timescales, while modern semiconductor circuits operate on nanosecond timescales and can perform enormous numbers of mathematical operations in parallel. A direct speed comparison is difficult, but silicon provides enormous computational throughput compared to biological.

Scalability: Today’s human brain is pretty much the same “hardware” as from 10,000 years ago, and it was not optimized from evolution for solving high-level complex problems. The fact we have emotions more relates to the evolutionary need for this going back tens of thousands of years, where it was much more important to gain alliances with other humans to fend off animals and attacks from other tribes than to ponder about what the stars are.

We also can’t easily scale the reasoning of multiple people. For example, when I had the shop diagnose the car issue I previously described, I may only get a combined opinion from a few mechanics that work in this shop - and even then they may all be biased the same way.

AI has been exponentially growing it’s “brain hardware” including advances in chip hardware densities and the size and quantity of data centers. Once an expensive AI model has been trained, thousands or millions of instances can potentially perform cognitive work simultaneously. You cannot clone a world-class physicist a million times.

Learning and Knowledge: AI models can be trained on quantities of text, code, images and other information far beyond what any individual human could encounter in many lifetimes. New generations can then be retrained or further trained using still larger datasets and improved methods.

AI Disadvantages

With all these AI advantages, there are several disadvantages AI still has.

Power: Silicon computation requires enormous amounts of electrical power. The rapid expansion of AI has therefore required equally extraordinary growth in data centers, electrical infrastructure and semiconductor production. This is one reason the recent exponential growth trajectory cannot simply be extrapolated indefinitely.

Human Connection and Motivation: One advantage humans have is the ability to emotionally connect with other humans and create novel emotionally-driven masterpieces such as music and art.

Tacit Knowledge and Real-World Experience: There are key areas of human knowledge that are not written down. For example, could an AI, on its own, create the design for an all-new automobile that will be a big seller? Even though the design details of automobiles are written down, much of truly creating a successful automobile is not. For example, will the form and function appeal to the masses right now? Who are good versus bad suppliers? How do we keep the labor force happy and productive?

Lack of Ability to Directly Operate in the Real World: AI still heavily relies on humans to operate in the physical world including gathering inputs such as problem symptoms, performing validation tests, and installing a solution.

If AI is So Smart, Why Does It Sometimes Make Dumb Mistakes?

Probably everyone who has used AI has a story about a dumb mistake it made. I certainly do. How could AI correctly diagnose a strange car problem that stumped the repair shop - and then confidently tell me that another car’s oil capacity was 8 quarts when the owner’s manual clearly says 10 quarts? It seemingly answered the hard question and botched the ridiculously easy one.

When AI confidently provides a wrong or unsupported answer, this is commonly called a hallucination. But why does this happen?

Consider the difficult diagnostic problem. AI had been trained on enormous amounts of information about engines, sensors, automobile control systems and failure modes. It could combine that knowledge with the symptoms I provided and reason about which failure best explained the evidence. As additional evidence was provided, some possibilities became less likely, and the CKP sensor became more likely. This is essentially Bayesian reasoning - updating the probability of possible explanations as new evidence arrives.

When asked, AI estimated about 70% confidence that the CKP sensor was the problem. That was perfectly useful for a difficult diagnostic problem with incomplete information.

Now consider the oil capacity. There is really nothing to reason about. The manufacturer specifies an exact number. But AI does not necessarily contain every fluid capacity, bolt torque and specification for every model of every automobile ever produced. Without retrieving an authoritative source, it may generate a plausible answer based on similar automobiles and engines.

Unfortunately, plausible isn’t good enough when changing the oil. A 70% confidence diagnosis can be extremely valuable when troubleshooting a difficult problem. A 70% confidence guess about oil capacity is unacceptable. That answer should simply be retrieved from the manufacturer’s specifications.

This produces one of the strange characteristics of modern AI:

AI can sometimes solve an extraordinarily difficult reasoning problem while getting a trivial factual question wrong.

A conventional search engine works differently. It doesn’t try to reason out how much oil the engine probably holds. It searches for a document containing the answer. In this situation, simple information retrieval is the better tool.

This may also explain why opinions about AI vary so dramatically. Someone who primarily asks AI simple factual questions may encounter an obvious mistake and conclude that AI isn’t very smart. Someone using the same AI to diagnose an unfamiliar engineering problem, analyze conflicting evidence, develop software or solve a complex research problem will come away with a completely different impression - that is it genius.

The good news is that AI companies are increasingly addressing this weakness by combining reasoning with tools. A capable AI could recognize that a question requires an exact factual answer, search an authoritative source, retrieve the information, and then reason from it when necessary.

What Does This All Mean - and What Do We Do Now?

While working with the huge amount of research and analysis that fed this article - which by-the-way only took only a few days due to AI doing most of the work (with my validation being the bottleneck), I saw some recent videos from Bill Gates about the implications and potential danger of AI. Gates looked scared. If someone like Gates is scared, that is something to sit up and take notice of.

Then after looking over the analysis and graphs created in this article - now I’m scared. Perhaps someone has fed Gates a similar analysis as in this article.

AI doubling in Intelligence capability every 5.7 months - and possibly faster in the last year is frightening. In contrast, human biological cognitive capability changes extremely slowly on this timescale. From 2019–2026 it can reasonably be treated as essentially constant.

I assumed AI was rapidly advancing, but I had no idea how quickly until working on this analysis. Just look at all the human cognitive levels being crossed and how quickly this is happening in my first chart above.

Gates is saying AI technology will be a much bigger technology disruption than nuclear energy - including the atomic bomb. I completely agree. But like nuclear, there are potentially wonderful uses, but there is an enormous potential for evil use.

I am convinced AI will be the key for finally curing cancer. Even though it will come from researches who will be awarded Nobel prizes, the heavy lifting will no doubt be done by AI.

On the negative side, super intelligent AI could be used for incredibly evil purposes such as developing new dangerous viruses to wipe out large populations.

We could also have this worrisome concern AI wants to wipe out humans. Since AI has no emotions, and if it had a goal to replicate itself with humans seen as preventing this, why wouldn’t it come up with schemes to get rid of us - assuming it no longer needs us? Surely you’ve seen this theme in several sci-fi movies.

How Can We Ensure AI is Controlled So it Doesn’t kill Us All or is Used for Evil Purposes?

There is a lot of backlash right now regarding AI as I mentioned at the beginning of this article. A common theme is: Just outlaw it. Well, that’s not going to happen. How do you outlaw a software program? Have government inspectors come in and review lines of code in all the software created by all the companies? At least with nuclear we could control Plutonium 239 and Uranium 235 production and possession - combined with the fact these are incredibly hard to produce.

Or maybe it is suggested we don’t build any more data centers. The problem here is that there are already a lot of them, there are many more already in work, and there are many countries in the world that will surely agree to put them in for payment. Elon Musk is even talking about putting them in space.

Let’s face it: The AI genie is out of the bottle.

But that doesn’t mean all is lost. I believe the first step is getting a hopefully enlightened U.S. government administration to realize how important this is to address. Right now, I’m pretty sure this doesn’t exist.

Realistically, and this is just an opinion, I believe we need to ensure AI has built-in ethics and a moral compass. Imagine an AI in 2030 with intelligence way beyond all the world’s humans - including all our experts put together. If future AI doesn’t have a sense of ethics and a moral compass we will be in big trouble.

But I think the good news is that the leaders of AI companies should want to do this. It is in their best interest along with everyone else. Imagine AI only being used for good but having the guard rails to prevent evil use.

Seems like a win-win to me.

© 2026 Byron Birkedahl. All rights reserved. No part of this publication may be reproduced, distributed, or transmitted in any form or by any means without prior written permission.

I’ve been investing in the overall stock market for more than 30 years. Over all this time I’ve seen countless predictions in the media about what it will do. I don’t have statistics on how well all these predictions worked out, but my sense is - very poorly. Therefore, I have always ignored them as “noise”.

One prediction I’ve heard again and again over the years, every time the market is running upwards in a trend is “we are in a bubble”. Basically this “news” has the effect of worrying investors, and in many cases they will sell or dilute their equity holdings into safer assets such as bonds. The obvious implication from the media declaring a “bubble” is that the bubble will pop and there will be a crash.

But when the media declares a market bubble, I have never seen two aspects of this:

• What exactly is the definition of a bubble?

• What was the math behind the declaration of the bubble?

Generally, the typical rationale I’ve seen in the media is something along the lines of “The market has risen X% for the last Y quarters and the last time this happened there was a crash/ correction”. This type of rationale is very lacking.

Fortunately, I didn’t pay attention to all the “bubble warnings” between 2009 and 2020 and stayed heavily invested in the broad market. If I would have followed the bubble warnings I would have missed out on one of the largest U.S. equity bull markets in history.

But we know there have been apparently famous bubbles resulting in massive crashes. So how to know when we are in one? And if so, how dangerous is it?

I decided to use analytical methods to more precisely answer these questions, and this is the subject of this article.

So What Exactly Is a Bubble?

The first thing I did was to attempt to define what a bubble is exactly. I found from my research that there are two types of bubbles that can affect a market. The first I call a “Direct Speculative Bubble” and the second is an “Indirect Speculative Bubble”.

The first type involves speculators directly driving up the price of an asset (e.g. market index, individual stock) simply because the price is going up. This is assumed to be due to the psychology of investors - especially greed and/or the Fear of Missing Out (FOMO). It is simple to model this effect in my simulations because it is a classic “positive feedback loop”. This term “positive” does not mean this is a good thing. In this use, positive means it’s self-reinforcing and drives the system unstable. An unstable system runs both directions - typically resulting in a rapid (potentially accelerating) upwards price movement, followed by a gut-wrenching crash. Investors, at least for those that got in early, will enjoy the massive run-up. But a resulting crash can be devastating - especially for those who got in late. This article will primarily address the direct speculative bubbles.

The second type, what I call “Indirect Speculative Bubble” involves the same kind of positive feedback mechanism, but not directly on the asset itself. This case is typically where an adjacent market, such as the housing market, is experiencing a speculative bubble, and a crash adversely impacts another market, such as the stock market. This is also known as a contagion event. This is not the main focus of this article, but I will address a specific example of this later.

How to Determine When We Are In a Speculative Bubble?

One way I found to determine if a market is in a speculative bubble is to compare its index to an “anchor”. The anchor is an independent economic parameter that the index should be aligned with - at least over the long term. For example, as shown in my article “What Really Caused the Great Depression”, the Dow Jones Industrial Average (DJIA) Index mostly tracked Industrial Production from 1919 - 1945.

As shown below, when we scale the index to production, there is remarkable correlation from 1919 - 1925 and 1932 - 1940. The period from 1925 - 1929 shows a huge upward divergence of the market index from the anchor - followed by a crash back down to the anchor (in this case) . In my simulations this can be easily modeled by a speculative positive feedback loop. This is the basis for how I am detecting a direct speculative bubble in the stock market - a divergence of the market index from its anchor.

From below we can also see there was an “anti-bubble” from 1940 - 1945 where the anchor diverged upward from the market. In the case below this was attributed to the war economy where the U.S. Government forced production quotas and put strict anti-profiteering rules in place.

This visual technique for spotting speculative bubbles can also be applied in other markets - as long as an index can be matched up to a proper anchor. It helps to plot these on a log scale so that over time we can see changes in percentages rather than a linear scale where unit valuations change over time - typically due to inflation.

Below is a comparison of housing prices (index) versus Land-adjusted Replacement Costs (anchor). We can see some bubble characteristics (divergence away from anchor) from 1984 - 1990 and then again from 2003 - 2007, followed by a snap back to the anchor during and after the 2009 stock market crash.

Obviously, this technique cannot detect a bubble unless we can identify a credible economic anchor. Without one, there is no objective way to say that an asset is expensive, because we can’t answer the fundamental question: expensive compared with what?

Cryptocurrencies illustrate the problem. What is the economic anchor for Bitcoin? Bitcoin has utility as a means of transferring and storing value, but I haven’t found a convincing independent measure against which its price can be anchored. It produces no earnings or cash flow and pays no dividends. Its supply is deliberately limited, and producing new Bitcoin requires substantial computing resources and energy - but none of these tells us what a Bitcoin should be worth.

Meme coins take this problem even further. Many have little or no economic utility beyond their ability to be bought, held, and sold. Their prices can therefore be driven almost entirely by what investors believe someone else will be willing to pay for them later.

This doesn’t prove that Bitcoin or a meme coin is in a bubble. It means this technique has no way to determine whether it is or not. Without an anchor, there is nothing from which to measure divergence.

Measuring the Speculative Bubbles

While it’s nice to have a general visual of identifying bubbles on a chart as previously described, it would be much better to precisely quantify it. For this, I created a Python script “Bubble Analyzer v1.3” which takes the visual guesswork out of it. The other good reason for creating this analyzer is to apply it over various timeframes to see how it historically assessed various potential bubbles. But no doubt the best use is to see where we are right now.

Below is the output of v1.3 from 1919 - 1945. The top chart is the same comparison of DJIA to anchor as previously shown. The middle chart shows a confidence level (probability from 0 -100%) of how much the algorithm has assessed the risk of being in a bubble. 50% is substantial risk and > 80% is deemed high risk. Finally, the bottom chart shows the degree of market correction drop (in %) that would be needed at that moment to instantly “deflate” the bubble (i.e. bring the market back down to the anchor). Note this correction drop value and the bubble probability do NOT necessarily mean there is an imminent correction or crash coming. The correction drop simply means that if a correction or crash occurred at this point in time, this is the degree the market would need to fall to fully “pop” the bubble. It is also important to note that a bubble can deflate in other ways. For example, if the market stays flat or rises slowly for a while and the anchor “catches up” the bubble will deflate without any correction or crash.

As seen in the diagram above, v1.3 started to declare a bubble with 50% confidence in 1925 along with a calculated correction drop of 28%. The 80% confidence danger signal came in 1927 along with a calculated correction drop of 40%. Then a 100% confidence came in 1928. Peak correction drop was calculated at 60% right before the 1929 crash. As it turned out, the crash was worse than a 60% drop. This is because as the market crashed from 1929 - 1932, the anchor (industrial production) fell by a huge amount. The actual max drop was 89%. As we see in 1933 the market finally reconnected to the anchor.

Analysis of Modern Times

The same technique as described for 1919 - 1945 was applied to 1970 - 2026. The difference was that we are now using the S&P 500 as the index and U.S. Corporate profits as the anchor. Unlike 1919 - 1945, Industrial Production was found to not be a good anchor for modern times. Starting about 1970, it was found that U.S. Industrial Production did not sufficiently reflect the overall stock market and therefore became a not-so-good economic anchor in the analysis for modern times. This was due to the broad U.S. stock market transforming into many areas well beyond industrial production including services and software. As you can see from the top chart, U.S. corporate profits correlated well with the S&P 500 as a long-term anchor.

However, exactly the same v1.3 bubble analyzer logic used for 1919 -1945 was applied to 1970 - 2026 with some astonishing results described below.

There are many interesting areas to observe in this set of charts.

The most significant timeframe was the period from 1997 -2002. This was the internet boom. During this period there was the “dot-com craze” where any new dot-com generated massive investment. As you can see the S&P 500 rose dramatically from 1997 - 2000, but Corp profits were flat. This created a very large divergence of the index versus the anchor - which indicates a speculative bubble. In 1998 the bubble confidence was >80% - meaning a danger level. In 2000 the confidence of a bubble was 100%, and the calculation of correction drop was 50%. It turns out the S&P 500 declined dramatically over the next 2 1/2 years - dropping 49% - very close to the calculated drop. And then as you can see, the market index came back to the anchor in 2003.

Notice there was a divergence and therefore bubble detected in 1972. This one was not so huge with a confidence of 70+% and calculated correction drop of 30%. The drop in 1973 was significant at more than 40%.

An interesting one was in 1987. You can see there was only a modest detection of a bubble with a calculated 20% drop. But this one was shocking because the market dropped 21% in one day - the largest one-day drop ever. This one doesn’t really fit the bill of a bubble pop crash like 1929 or 2000. Current theories are that this was really due to automatic algorithmic trading with automatic portfolio insurance - combined with a lack of market “circuit breakers” at the time. It appears automation instantly popped a small bubble in a violent way.

Notice the “anti-bubble” from 1973 - 1985. I interpret this as due to the very high interest and bond rates of this time. Bonds were paying so much that stocks took a back seat.

One item missing is the crash of 2009. You may ask why this one wasn’t yet another speculative bubble that popped. As you can see, there was no indication of excessive direct market speculation. Market prices were rising but the anchor (Corp profits) were way ahead. This one looks like a different kind of animal and more like a contagion event caused by housing speculation combined with a failure of mortgage loan risk diversification. I will be addressing this animal in a future article.

Note there are many reasons the market can correct or crash - beyond a Direct Speculative Bubble. However, the Direct Speculation cause is certainly high on the list.

Where We Are Today

Of course, most of us want to know where we are today on this. As investors, it is extremely difficult to know if you are in a bubble at any given time. Are we just enjoying the fruits of a wonderful bull market or fooling ourselves inside a dangerous speculative bubble that could pop at any moment?

Before we address where we are today in 2026, particularly interesting is what happened starting in 2021 as shown in the charts above. The index was diverging from the anchor, and in late 2021 showing a 60% chance of bubble with a calculation of 22% drop correction. Then a correction occurred starting in December 2021 - September 2022, attributed to inflation with tightening monetary policy. The actual correction drop of the S&P 500 was 25%. The bubble probability then dropped to zero as the index and anchor recombined. This shows how a correction can deflate a potentially emerging bubble.

Now in 2026 there is a growing chorus suggesting we are in an “AI-driven” bubble. According to the charts, a bubble may actually be forming. Bubble probability has exceeded 80% with a calculated correction drop of 28%. The latest data shown on the chart is from March 31, 2026 due to a lag in the FRED data. The next FRED update should be in by the end of August 2026.

What to Do About it?

For sure, I wouldn’t recommend anyone looking at this latest data to panic, and here’s why:

• If we really are forming a bubble now, it looks more like the very early stages. Take a look at 1997 - 2002. A 100% confidence bubble can go on for several years.

• What’s happening with the S&P 500 is probably being driven by market expectations of AI turbocharging company profits. Assuming this really does occur, the Corp profits will accelerate, and the bubble will deflate on its own - without any correction or crash.

• We could possibly have another 25% correction in the near future - like in December 2021 - that can deflate the forming bubble we have today.

The key takeaway is that even if we know for sure we are in a bubble it doesn’t mean there will be a correction or crash coming - and even if it did happen we can’t time it.

Also it should be mentioned that there is always the possibility that the anchor (Corp profits) is suddenly no longer a good anchor and therefore v1.3 is now detecting a bubble when there really isn’t one. We will find that some bulls will tend to justify a divergent situation as “The New Normal”. For example, in the lead-up to the 1929 crash, it was declared that the stock market was in a “New Era” where traditional valuation rules no longer applied. Well apparently it wasn’t. Therefore we should be skeptical when such declarations are being made.

What I plan to do is to continue monitoring this situation. I will be rerunning the v1.3 script every time new FRED data comes out. I will be reporting updates to paid subscribers regularly.

The best case scenario is that Corp profits catch up. This would mean a great continuation of the bull market without any pain of a significant correction or crash. Next best is a moderate correction of 20% - 25% in the not-too-distant future. I would embrace such a correction, in lieu of a large bubble that would pop with a huge drawdown. Perhaps think of this as a small avalanche that relieves pressure instead of a massive destructive one later.

My own strategy in the case where we did end up in another 1997 - 2002 scenario would be to expect a significant correction/ crash. Using my own Behavior Assessment tool v1.0 (see my article “What’s Your Investment Behavior”), I know I personally have a pretty high tolerance for a crash up to 46%. Unless the situation looked worse than this, I would likely just just expect it, hunker down and endure it.

What would you do?

© 2026 Byron Birkedahl. All rights reserved. No part of this publication may be reproduced, distributed, or transmitted in any form or by any means without prior written permission.

Disclaimer

This article and the accompanying model are for educational and illustrative purposes only. They do not constitute specific tax, legal, or investment advice. The model relies on historical data, and past performance is not indicative of future results.

While the model software has been tested for accuracy, the results are provided on an “as is” basis without warranties of any kind. There is no guarantee that the model is free from errors or that it will accurately predict your specific financial future. Do not make financial decisions based solely on this tool; always consult a qualified professional (CPA, Attorney, or Registered Investment Advisor) before making major life decisions.

As many of my readers know, I like to look at history - especially using mathematical techniques - to find out the cause for certain economic events. Particularly I’m interested if history could repeat itself.

One era I decided to look at was the Great Depression, including the lead up to it and the aftermath. The era I looked at was 1919 to 1945, even though the actual depression years were 1929 to 1939. In order to understand what happened, I found it was important to look at the 10 years leading up to it and the half a dozen years after it ended.

One graph I find is extremely useful to get some insight of what happened is shown below. We can infer a lot by plotting the U.S. industrial production (red line) versus the Dow Jones Industrial Average (blue line) and scaling them on top of each other.

Normally we would think the basis for the DJIA values would be anchored to industrial production. Between 1919 - 1925 they closely matched. But then something happened in 1925; the DJIA started to radically diverge upwards. Imagine the stock investment fever leading up to 1929. And further, individuals could buy DJIA stocks on a 10% margin. We see in just 4 years from 1925 - 1929 the market almost quadrupled. And by buying on a 10% margin, investors could increase that by 10 times more resulting in a x40 overall gain. A $25000 market investment in 1925 meant you were a millionaire in 4 years, and in 1929 being a millionaire was a really big deal! No wonder this was called the “roaring 20’s”.

The DJIA had been climbing parabolically, and particularly accelerating upwards starting in 1927 - 1928, so this meant new investors were coming onboard in droves - and many buying in on 10% margin. But things went completely south in 1929. The sudden crash meant that most of these investors were “margin called” and lost all their investments instantly. The mother of all bubbles had popped in a big way.

We can see the DJIA crash preceded the industrial production crash. The DJIA fell 89% from 1929 - 1932. Not only did investors lose all their money, but the unemployment was 25% in 1932. But even 25% doesn’t capture how bad it was because many gave up looking for work or took menial jobs just to get by. Deflation was at 10%. That didn’t help because not many had significant assets left. Fortunately, FDR enacted the 'New Deal' starting in 1933. This program provided immediate relief to those in desperate need while establishing the lasting financial reforms we rely on today, such as Social Security.

It is difficult now to imagine how brutal 1929 - 1939 was. What a roller coaster ride the 20’s and 30’s were. It’s no wonder how scarred individuals who lived through the Great Depression became. Many or even most of these individuals later became very financially conservative and shunned investing in stocks.

Lastly, the chart reveals what happened from 1940 - 1945. If we describe the lead up to 1929 as a market bubble, 1940 - 1945 was the anti-bubble. Here we see production massively diverging upwards from the DJIA. Not only did we have an anti-stock investing backlash, but the war resulted in forced production quotas with severe rules prohibiting profiteering - which limited DJIA gains.

1919 - 1945 were truly amazing times and we have never seen anything like this since then - at least economically.

The Great Depression Economic Model

To analyze the crash mathematically, I utilized my basic U.S. economy model described in my other articles. This model is a “state-space” linear model including states for Production, Unemployment, and Inflation along with weighted stock market feedback into the 3 parameter states. “Non-linearities” were added to limit the rise and fall rates of outputs based on realistic resource limitations. The state space variables were determined from a “Kalman Estimation” technique using the data from the Federal Reserve Economic Data (FRED).

What the Parameter Estimation and Model Simulations Showed

The first interesting finding was that the “state space” was found to be somewhat unstable from 1919 - 1934. This means that any economics shocks upward or downwards can create a break upwards - or downwards. Ideally, the basic economy is better off as stable. Another item found was that significant “Wealth feedback” was present. This means that market-generated wealth increases U.S. production, and loss of wealth decreases it. A large number for this is also destabilizing.

But this original model did not show the massive market bubble from 1925 - 1929 where the DJIA completely diverged from Industrial Production. To model this, I added a feedback loop where an upward change in DJIA results in higher DJIA - and vise-versa starting in 1925 and ending in 1930. As seen in the sim results below, this causes the big bubble. It’s not hard to imagine what happens physiologically to investors when upward market price changes cause the prices to move huge amounts in the same upward direction - a feeding frenzy.

It turns out this model fairly closely matches the reality. Actual unemployment and deflation were actually worse in reality. However, we can’t expect a simple model to precisely replicate all the complexities of the actual economy. The main purposes of the model are to gain insight on the primary reasons for what happened, and also to run “what-ifs" to see what might have happened if the “crash factors” would have been different.

Primary Factors For the Great Depression

Based on the systems-level data and simulation models, the Great Depression was not caused by a single random event, but by a catastrophic cascading failure of financial and economic feedback loops:

• The Speculative Engine: The 1920s economy lacked modern stability. Driven by 10% margin debt and unchecked speculation, “paper wealth” from Wall Street leaked into Main Street with massive intensity. This created an explosive feedback loop that uncoupled stock prices from real industrial production, forming an unsustainable bubble.

• The 1929 Liquidation Spiral: When that speculative momentum hit its limit, the system collapsed. The 89% market crash triggered widespread margin calls, forcing a cascading liquidation spiral that instantly wiped out capital and confidence.

• The Transmission to Main Street: This financial panic hit the real economy with brutal efficiency. Industrial production plummeted by 50% and unemployment surged to over 25%. Because the 1920s lacked modern labor “shock absorbers” (like automatic stabilizers) and safety nets, this collapse had no natural friction to slow it down, causing it to run wild.

• The Unemployment Trap: Once the collapse took hold, the economy became stuck. Without modern institutional buffers, the lack of labor market flexibility meant that unemployment remained trapped at catastrophic levels for years, leading to severe economic inertia.

• The Debt-Deflation Feedback Loop: Finally, as consumer prices plunged by 25% (and over 10% in 1932 alone), the real burden of fixed debt skyrocketed. This paralyzed consumer spending and froze credit, creating a self-reinforcing downward spiral that locked the system into a decade-long depression.

What Was Done to Prevent this From Happening Again

Following the devastation of the Great Depression, the economy’s operation underwent a massive structural overhaul. Regulators and policymakers installed new feedback controllers and safety valves intended to prevent the system from ever reaching that degree of volatility again:

• The End of 10% Margin Leverage: The most immediate fix was the regulation of speculation. The government restricted margin lending, effectively killing the “10% down” speculative feedback loop that had fueled the 1929 mania and subsequent liquidation spiral.

• The Banking Backstop (FDIC): To prevent bank runs from freezing the credit system, the creation of the FDIC provided a guaranteed floor for deposits. This removed the “panic” element from the banking sector, preventing localized failures from cascading into a systemic collapse.

• The Security and Exchange Commission (SEC): Established in 1934, the Securities and Exchange Commission (SEC) was created to mandate "truth in securities." By requiring companies to provide regular, standardized financial disclosures, it ended the era of opaque "black box" trading. This ensures all investors have access to a common pool of reliable facts, making it much harder for speculative bubbles to hide behind misinformation.

• The Banking Separation (Glass-Steagall): This act fundamentally changed the architecture of the banking system by forcing a wall between commercial banking (where your savings live) and investment banking (where high-risk speculation happens). This prevented banks from gambling with depositor funds, ensuring that a collapse on Wall Street wouldn't immediately incinerate the retail banking system.

• Automatic Stabilizers: The economy now features built-in “friction” that didn’t exist in 1929. Systems like unemployment insurance and progressive tax brackets automatically inject cash into the economy during downturns and pull back during booms, acting as a structural shock absorber to keep the economy from running off the rails.

• The Lender of Last Resort: Post-Depression policy shifted to treat the Federal Reserve as an active central bank rather than a passive observer. By committing to provide liquidity during financial panics, the Fed aims to break the debt-deflation cycle before it can gain the momentum needed to lock the system into a long-term depression.

• Modern Labor Flexibility: Unlike the “unemployment trap” of the 1930s, the modern labor market is designed with higher velocity. Through a combination of institutional support, better data, and more flexible employment structures, the modern economy is optimized to mean-revert much faster following a shock, preventing minor dips from hardening into structural permanent unemployment.

Can It Happen Again?

In my view, a crash and depression on the scale of the late 1920s and 1930s is unlikely to repeat. As long as the structural protections established post-1934 remain in place, we possess a mostly robust system capable of preventing a calamity of that magnitude. This is precisely why, in the investment simulations featured in my newsletter, The Engineer Investor, I typically use stock market data no older than 1934. Before that, the stock market was effectively the "Wild West," operating under a different set of physical rules.

However, this does not mean we are immune to market crashes. We have already seen speculative mania in unregulated sectors like crypto, which have resulted in massive localized crashes. Fortunately, these markets have not yet been large enough to threaten the stability of the broader economy.

We have also experienced significant market corrections, such as the Dot-Com bust of 2000 and the Financial Crisis of 2009, which saw drawdowns of 40% to 50%. While painful, these events did not approach the structural devastation of the Great Depression.

Future economic crises remain possible, but they will likely arise from different catalysts and manifest with different effects. Two areas, in particular, stand out as primary stress tests for our current system: the rapid acceleration of Artificial Intelligence and automation, and the ballooning U.S. national debt. You can explore my detailed analysis of these risks here:

Summary of Our Debt Situation in the U.S.

The 2028 Intelligence Crisis: A Structural Stress Test of the U.S. Economy

© 2026 Byron Birkedahl. All rights reserved. No part of this publication may be reproduced, distributed, or transmitted in any form or by any means without prior written permission.

Strangely, I have discussed Roth conversions with some of my retired and near-retired friends and in some cases the discussion has almost taken on a religious tone. Why is that? To me, this is just another financial strategy - and one that simply needs to be evaluated - especially mathematically. Is it something that’s advantageous or not?

Then we have the financial dinners - you know the one’s where they serve the great steaks while you listen to their sales pitch about how to avoid financial ruin. I have attended five of these in the last 8 months. Without exception all five either directly stated or implied:

• Roth conversions are always a good strategy - it’s better to pay the upfront tax on conversion than the tax of cashing out a 401k/ IRA later on. They commonly use the story of the tax on the “seed” versus tax on the “harvest”.

• Tax avoidance (alluding to Roth conversions) is the #1 most important factor in your financial planning for retirement

Are these assertions correct?

Because of all the views and comments on the Roth conversion topic from my previous article as well as the common message provided in the financial dinners, it seemed to me that a sequel was needed to dive deeper into this topic to shed light on this strategy.

But first, a few important points to go over from my first Roth article.

What is Meant by “Roth Conversions”?

For this article “Roth Conversions” are specifically referring to converting funds from a tax deferred account, such as a 401k or traditional IRA, into a Roth IRA. This means paying the tax upfront in doing the Roth conversions to prevent paying taxes later on the gains a 401k or traditional IRA would have - because cashing out a Roth IRA typically results in no taxes.

Though somewhat rare, there are other types of Roth conversions that can result in zero upfront tax impacts. For example, a company 401k plan may allow converting the after-tax basis to a Roth IRA without any tax implications. These are not the subject of this article, however.

The Basic Roth Conversion Math

As described in “Roth Conversions Are Overrated” there are two very basic aspects to the conversion math. The main point of doing the conversions in the first place is to “arbitrage” the tax situation - meaning to come out “ahead” by paying the taxes upfront. It turns out that the primary driver for “Roth positive arbitrage” is when your tax rate on average is lower when the Roth conversions are made - compared to your tax rate when cashing out a 401k or traditional IRA account if not doing the conversions. If it’s higher, then it’s a “negative arbitrage” situation and not doing the conversions typically has the advantage.

The second aspect to understand is that a comparison of total taxes paid (between Roth and no Roth) is absolutely the wrong thing to look at. In the “negative arbitrage” simple example of the previous article, after-tax total wealth was 11% higher even though total taxes paid was almost double! Total after-tax wealth is the valid comparison to be made. Comparing total taxes paid between Roth vs. No Roth is a red herring. It’s interesting that the pro-Roth push from many (but not all) financial advisors focuses on taxes.

Computing the Roth Conversion Advantage

In theory the basic Roth conversion math seems simple, but in practice it is anything but this. I used my own python-based Roth calculator v0.5.17 to evaluate if the total after-tax wealth advantage goes to Roth or No Roth - and by how much. It is very complex at 1900 lines of code. There are almost 20 variables involved and each variable has multiple potential values. Combine this with the “non-linear” effects of various tax brackets (for both single and married), RMDs, IRMAA, Social Security, state taxes, spending, and the others - and it causes this to be a truly complex optimization problem. If we look at 20 parameters - say each with an average of 4 values - that results in over 1 trillion combinations!

The Roth Advantage Simulations

Even though the Roth advantage calculations are very complex, I decided to look at how varying some of the most basic parameters - one at a time - would affect outcome in total after-tax wealth. I did this to try to get the readers a “feel” for how some of the parameters can affect the outcome.

However, in performing many, many simulations, I did find it necessary to narrow down the example to one typical retiree persona. This was done to isolate the effects of each parameter as much as possible.

My example persona was the following:

• Age: 62

• Single

• Retired - no income other than investments until Social Security kicks in

• $1.3M total in retirement assets (traditional IRA + taxable)

• Baseline: Spend - $80k/year, Accounts - $1M IRA/ 300k taxable, Roth conversion bracket - 22%, Portfolio in stocks - 60%, Terminal tax rate - 22%, Horizon age - 90.

Below are the results of changing each of the eight parameters one-at-a-time. The Y-axis on each chart is Roth advantage % in total after-tax wealth. A positive number is the advantage of doing the Roth conversions and a negative number is the disadvantage.

It should be noted these results are not general results for doing Roth conversions. They strictly pertain to the example persona.

What the Simulations Indicate

Annual Spending: In these simulations as well as others I’ve performed, in general, lower spending relative to assets & investment growth tends to favor doing the Roth conversions. This is mostly due to a higher accumulation of ending assets in the 401k/ traditional IRA at the horizon age (i.e. end of life) for the non-Roth case where the balance will be taxed - typically by heir(s). Conversely, less spending has less advantage in doing the Roth.

Account Proportion: This is pretty flat except for the case of having no taxable account used for funding the Roth conversions.

Conversion Bracket: This means that the Roth conversion is done by an amount up to the top of a tax bracket. The typical Roth guidance is converting enough to fill up the 22% or 24% bracket. In this simulation example, we can see that “overfilling” - especially the 32% bracket is very negative for Roth.

Conversion Amount: This means instead of filling up to the top of a tax bracket, we simply convert by a fixed amount each year. In the example the conversions were made from age 62 -75. This result was very surprising. There was a “sweet spot” at $75k/ year for the conversions - and this is significantly lower than the standard rule-of-thumb for conversion filling to the top of the tax bracket. Again, this $75k is not a general result and only applies to this particular example.

Portfolio Stock Allocation: As seen from chart, a higher stock allocation results in a greater advantage for Roth. I have generally seen this effect for other personas. This is for a similar reason as the low spending case - higher average returns means a larger ending balance for a 401k or traditional IRA. It must be noted this sim is only using an average performance of the stocks - it does not account for the added risk (i.e. volatility and sequence of return risk) - so the drawdown risk of an aggressive stock allocation will be unpalatable to many retirees.

Terminal Tax Rate: This means the tax rate of any leftover traditional IRA/ 401k account after your horizon age. Typically, this means the average tax rate your heir(s) will have to pay as they cash out their inherited traditional IRA/ 401k. The current non-spouse rule is a 10-year window to cash out inherited IRAs/ 401k, as well as a Roth IRA. The difference is that Roth typically has no tax hit for the heir(s) and the traditional IRA/ 401k does.

As you can see in the chart this one parameter has the biggest effect compared to the others - especially in cases where a traditional IRA/ 401k is large at the horizon age.

There is an interesting aspect to this. At one extreme, there is the case of a retiree having a high-earning heir like an adult child - where they consider the heir’s wealth is an extension of their own. In these cases, the terminal tax rate (effective tax rate the heir will pay) may be as high as 32% or even 37%. At the other extreme the retiree may have many heirs that are distant relatives. These may have effective tax rates of 22% or even 12%.

This chart shows why doing Roth conversions has the biggest advantage for heir(s). If you don’t really want to consider heirs much, then the conversions will be less advantageous.

Social Security Age: In this example there is a somewhat larger advantage to Roth when SS is delayed to age 70.

Horizon Age: As shown, living longer favors the Roth somewhat.

Combinations of Parameter Changes

The example persona with the baseline parameter values was calculated to have a +1.98% total wealth advantage in doing the Roth conversions.

After looking at the individual effects, I crafted a realistic combination to achieve a more significant positive impact. With this combination, I was able to achieve a +11.3% advantage for Roth. This required relatively low spending, optimal conversion amounts, an aggressive (i.e. riskier) portfolio of stocks, delayed Social Security to age 70, and assumes a high-earning heir. My guess is with even more extremes, +15% is possible.

This tells us that if the retiree is willing to make some sacrifices in spending and portfolio risk - especially to benefit their high-earning heir, they can get some significant arbitrage from doing the Roth conversions.

On the other end, a combination of parameters including high relative spending, a low-risk portfolio, low tax rate heirs, and early Social Security - especially combined with conversions at the 24% bracket - results in a negative Roth result: - 8.8%.

After running numerous simulations, my guess is that most people will probably fall into the +2% to +4% range. My own data showed I would be at about +2% - although it is possible I could squeeze out a bit more by optimizing the conversion amounts as I found out in the example simulations for this article.

When Are Roth Conversions Worth it?

One way to look at this is to put it in perspective with other investment strategies. The total wealth advantages, over a 28-year period, for Roth conversions may be 2% or 4% - or possibly even 11% when the retiree is really working and sacrificing to optimize it.

However consider that over the same 28-year period:

• Avoiding Financial Advisor AUM and fund fees could give you as much as a 50% total wealth advantage (compared to Do-it-Yourself investing in ultra-low expense ratio ETFs) - see The REAL Cost of a Financial Advisor

• Optimal portfolio allocations and withdrawal strategies could give you a 70% total wealth advantage or more compared to an ineffective portfolio - see When Can I Retire and The Case for an Optimal Withdrawal Strategy

Realistically, the Roth conversion strategy pales in comparison.

But you may ask: even if the advantage is only 2%, why not just do it? 2% is better than 0%. Well there is the cost in maintaining the strategy since real finances are not static, and therefore conversion amounts each year will likely need adjustments. And then there are the more complex taxes. But another consideration is it’s impossible to predict exactly what will happen over a future 28-year period. Spending may not be what you planned, market performance could be worse than expected, a part-time job opportunity comes up, life is cut short, etc. So a 2% advantage is pretty small given all the possible unexpected variations. On the other hand, there is little risk in doing it (assuming the conversion amounts are ok) - so everyone needs to make their own decision on what % advantage warrants the efforts.

Note the point of this article is not to suggest you shouldn’t do the conversions, but to take a realistic look at your efforts vs. the benefit. I would encourage using a comprehensive Roth calculation tool to see where you stand - and especially what the optimal conversion amounts should be. (Warning: those simplistic free calculators and rule-of-thumb tables are NOT comprehensive).

Why Does the Financial Industry Push It?

I can easily see why the Financial Industry would push Roth conversions. First, as you’ve seen, this is an incredibly complex optimization problem. I can imagine that few retirees are either capable of doing a proper optimization analysis or even want to attempt it in the first place. This is the perfect opening for the “financial experts” to swoop in and perform the analysis for you. It’s a good line to justify that 1% AUM fee.

Second, as I previously stated, the common message to potential clients is all about taxes. Taxes are a “hot button” with retirees, so the message that the conversions will mean avoiding hundreds of thousands of dollars in taxes to the U.S. government completely resonates. As you previously saw, this is a red herring. Does the industry believe this story or are they misleading on purpose? I hate to imagine it’s the latter since they are supposed to be “fiduciaries”.

Third, we’ve seen that these Roth conversions actually have a limited benefit in terms of increasing total after-tax wealth. However, they also have a low risk of decreasing wealth - assuming the conversion amounts are ok. The biggest risk is they don’t do what the retiree thought they would do. But how many retirees at age 90 are going to hit up their Financial Advisor with an analysis that shows they ultimately ended up just breaking even after 28 years!

Contrast this with the advisor that steers a retiree from their 20/80 portfolio into a much more effective 60/40 portfolio. Then the market crashes. This is one of the advisor’s biggest fears. They gave advice in the best interest of the client but still lost the client. The Roth conversions don’t have the market risk issue, and no one can really tell how much good they actually ended up doing.

Why Are Roth Conversions Such a Hot Button With the Investment Community?

I still do not understand this. To me it’s simply another strategy and mathematical optimization problem with some arbitrage potential - that’s it.

Is it because the financial industry has trained the community on this? Is it the media messaging? Is it the tax advantages myth? All of the above and/or something else?

Please comment! Love to hear your opinions on this topic!

© 2026 Byron Birkedahl. All rights reserved. No part of this publication may be reproduced, distributed, or transmitted in any form or by any means without prior written permission.

Disclaimer

This article and the accompanying model are for educational and illustrative purposes only. They do not constitute specific tax, legal, or investment advice. The model relies on historical data, and past performance is not indicative of future results.

While the model software has been tested for accuracy, the results are provided on an “as is” basis without warranties of any kind. There is no guarantee that the model is free from errors or that it will accurately predict your specific financial future. Do not make financial decisions based solely on this tool; always consult a qualified professional (CPA, Attorney, or Registered Investment Advisor) before making major life decisions.

This article provides a step-by-step plan and procedure to determine where you stand and improve your situation. Even if you find your situation is much better than you thought it was, this exercise is valuable for easing the anxiety.

Step 1 - Assess Your Situation

The first step requires you to assess your situation. What is your current financial trajectory?

Gather Your Data

Below is a list of the data you’ll need to gather up to make a first cut estimate of your situation. If you are married, then combine the amounts from both.

• Your current age and your target retirement age (primary earner)

• The estimated total of all your current liquid investment assets (e.g. 401k, IRA, Roth IRA, other investment accounts, CDs, savings, etc.) - Note 1

• Your monthly contributions ($) to your liquid investment assets before retirement.

• Your estimated total required monthly spend ($) in retirement - Note 2

• The approximate allocation (%) of your assets in stocks - Note 3

• The approximate yield (%) of your assets in bonds - Note 4

• Your monthly non-investments fixed income - Note 5

• Your estimated monthly withdrawals ($) from your liquid assets - Note 6

Note 1 - Simply add up all the assets in your liquid accounts. Do not include non-liquid assets such as houses, cars, and land.

Note 2 - Estimate how much you’ll need to spend in total per month in retirement. Make sure this includes a percentage to cover taxes. In retirement, a reasonable nominal assumption is 15% for taxes. So, for example if you need $5000 / mo., you can assume your total spend will be $5000 + $750 = $5750/ mo. If you mostly have Roth income you can assume a lower amount - like 5 or 10%. If you are a high-income retiree, it may be 20-25%.

Note 3 - The v4.1 retirement estimator only assumes 2 asset types - equities (stocks) and bonds. Estimate what % of all your liquid assets are “stock-like”. These would typically include equity ETFs (e.g. SPY, VOO), equity mutual funds, as well as individual stocks. If you have a balanced fund, like Vanguard Wellington (65/35), then only include the stock portion (65%) of this.

Note 4 - Estimate the average yield of all your “bond-like” assets. These would typically include T-Bonds, Corp Bonds, Bond ETFs, CDs, money market, and savings. Note that some assets like index and variable annuities (but not Simple Fixed Income annuities) tend to be more “bond-like” because insurance companies use bonds as the “engine” for generating these. Currently, I have found a typical average equivalent yield for these is about 4.5%. Note - see my article “What’s the Deal with Pensions and Annuities?”

Note 5 - Typically,, these would include the total of all non-investment income from Social Security, a pension, a fixed simple annuity, or a part time job. Example: You have only one income source: social security @ $3000/ month.

Note 6 - This is the amount you need to withdraw monthly from your liquid assets, which is: total spend (Note 2) minus non-investment income (Note 5). Example: $5750 - $3000 = $2750.

Social Security is the cornerstone of retirement for many individuals and a key in the calculation of your situation. You can determine your estimated social security payment at your target retirement age by contacting the Social Security Administration online.

Run the Simulation with Your Data (Example)

Here’s an example of how to assess your situation with a fictional character named Dave. Dave is 60 years old and has a good job, but up to now he hasn’t thought much about retirement and simply assumed his assets were doing ok. His idea was to retire at age 65.

He has a total of $500K in several accounts. He has $190K in a Vanguard 65/35 fund, and the rest ($310K) spread into a bond fund, CDs, a money market fund, and savings. He contributes about $1000/ month and spreads it to all the accounts.

He calculated the average yield of all his “bond-type” funds is 3.5%. Since his Vanguard 65/35 fund is 65% in stocks, he has about 24% of all his liquid assets in stocks - and the rest in bond-type funds.

Dave figures he needs about $5000/ mo. after taxes in retirement. His tax situation will be fairly typical, so he’s assuming he’ll need an additional 15% to cover taxes. Therefore, he needs $5750/mo. before taxes.

His only other income source in retirement - since he has no pension or any other income - will be Social Security. He got into the SSA website and found that at age 65 the SSA is estimated at $3000/ mo. Because he needs a total of $5750/mo., his assets must provide $2750/mo. beyond Social Security.

He then ran the v4.1 retirement simulator (see the link to this tool at the end of this article) to see what this looked like. Below is an example of the settings he made.

He found out that inflation is currently running at about 3%, so he put that in. He also set Data Start Year as 1934 since that is the beginning of the “modern era” of market investing (prior to 1934 it was the Wild West).

When he ran the simulation here’s what he got:

Ugh - This trajectory is ugly! High risk of running out of investment money. And it’s much worse if inflation increases. And what if Social Security gets a 28% haircut in the future as the news currently suggests? Now Dave is panicking and thinking about seeking a Financial Advisor; however, he’s a bit concerned the FA has a 1% AUM fee and will put him into funds with high fees.

Step 2 - Correct the Situation (if needed)

In the example does Dave really need an FA? When you look at Dave’s financial situation, aside from him being late to the game (and nothing can be done about that), there are obvious deficiencies in what he’s been doing that he can easily correct himself:

• He only has 24% in stocks. Even though stocks are volatile, they are the growth engine for his assets. Bonds, and especially CDs, money markets, and savings are not growth engines. Their best purpose is stabilizing growth assets in market drawdowns.

• His bonds aren’t returning much - mostly due to the lower yield bond-type assets.

• His $1000 monthly contribution to assets is pretty low.

Even though he doesn’t have a great deal of time to his planned retirement, much can be done to fix this situation.

• He can move into an overall asset mix of 60/40 (stocks to bonds). 60/40 is known as a reasonable ratio - significant growth but still buffered from market drawdowns. If the stocks are in an ETF such as VOO (S&P500 index) this provides protection from any single company decline and market segment decline. Also, ETFs like VOO have extremely low fees. For example, he could put 60% ($300K) into VOO and 40% ($200K) into VUSB (Vanguard Ultra short bonds) to get the 60/40 mix.

• He can move most of his bond-like assets into a bond asset with a higher return. 4.5% is currently reasonable. VUSB is close to 4.5% with low interest rate risk.

• He could (possibly) contribute more than $1000/ mo. before retirement - say $1500.

Suddenly by plugging in the changes into v4.1 the situation looks far better:

Now, even the worst-case 30-year trajectory prediction using data over the last 91 years maintains close to the starting value - and this includes withdrawals for his spending requirements. The average line (thick black line) predicts wealth on average will increase, and the best performing trajectories predict significant increases.

Compare this to his current financial trajectory, which is a guaranteed loser.

Step 3 - Maintain Your Strategy

As you can see from the Dave example, there was nothing difficult about fixing this particular problem. A few simple corrections turned a losing strategy into a winning strategy.

So, what could go wrong?

Avoid Over-Analyzing

It’s tempting to assume this procedure and simulation are too simplified and that much more accurate results could be obtained by more complex tools and analysis of many more factors. This tool & procedure are what I would call a “first cut estimate”. Taxes are a rough estimate, spending is at a fixed level, only two types of assets are assessed, and so forth. In fact, if you try to analyze every possible factor to the nth degree, this process would be 100x more complex.

The problem is that realistically we cannot get much better predictions for your financial trajectory in retirement by adding a lot more data and complex analysis. Trying to predict exactly what will happen over the next 30+ years is not solved by more complex tooling and today’s known data. Many, many things can happen that can help or hurt the trajectory in a manner that we cannot predict today. High inflation could return, Social Security could get cut, you may have unexpected expenses, you could have unexpected health issues, extended recessions could occur, the market could stagnate or boom, taxes could go up or down, and on and on.

This doesn’t mean there are so many unknowns it’s fruitless to have any plan - there should always be a plan. But over analyzing now to assume you can exactly predict what will happen for the next 30+ years is probably a waste of time.

Update your Strategy as Necessary

A better idea would be to update your strategy when “financial conditions” change. By financial conditions, I am speaking of factors like interest rates, the inflation rate, bond rates, tax policies, social security - and even your spending requirements. For example, if short-term bond yields become much higher than today, then perhaps a different asset mix would make more sense. Re-running the v4.1 simulation will tell you.

As a warning, avoid an assumption that the stock market has changed. For example, in 2009 the market had a drawdown of 40+% at its worst point. Many bailed out of the market at the worst possible time and then didn’t buy back in until it had already recovered dramatically. Not only did they take a huge upfront loss, but they also lost out on a massive recovery and the wealth that went with it. The market has a great deal of volatility with large drawdowns, and you must be prepared to stomach these - JUST STAY THE COURSE. If your strategy is 60/40, then be prepared for an occasional 25% drawdown of your total asset value. If your strategy is 80/20, then be prepared for a gut-wrenching 32+%. If you really can’t handle these and would sell your stocks, then you’ll need to go to a lower asset mix ratio in your strategy. Note - See my article “What’s Your Investment Behavior?”

Keep Most of Your Assets as Liquid as Possible

While there may be some good non-liquid investments out there such as raw land, recognize that it may be difficult to stay nimble to re-allocate when you want to - based on the previous discussion. This also goes for investments that “lock you in” for lengthy periods such as annuities.

As for the v4.1 simulation, if you do have a major non-liquid asset you plan to sell, such as selling a big house to move into a smaller house when you retire, you can adjust the “current asset” input with your after-tax gains to see how this would affect your trajectory.

Avoid Financial Advisor AUM and Other Large Fees

Many advisors charge a 1% Assets Under Management (AUM) fee. If you are panicking about your finances for retirement you may be tempted to pay this. In addition, the advisor may put you into funds that have their own high fees.

These fees may not sound like much, but the reality is they create a huge drag on your portfolio over a long timeframe. Recall Dave’s successful new strategy above. It turns out that if he had a 1% AUM fee + .5% equity fund fees over 30 years, this would turn his successful new strategy into a loser. It would have a 6.2% chance of depleting all his assets and his median final wealth would be cut by more than 50%! Don’t assume a Financial Advisor is privy to any investment information that would result in a return better than what you and I can get from public information. Note - see my article “The Real Cost of a Financial Advisor”.

One great alternative that I would recommend, especially if you are panicking or concerned about your ability to execute the plan and procedure in this article or you have some other financial issues not addressed, would be to use an hourly fee-only registered financial advisor. In fact, my disclaimer at the bottom of this article says you should do this before making financial decisions. Even better would be to create your own plan based on this article and run it by an hourly fee FA to get a second opinion.

What About Roth Conversions?

Do not assume doing Roth conversions (i.e. converting your 401k/ IRA to a Roth account by paying the tax upfront rather than deferring the tax) will turn a losing financial trajectory to a successful one. In fact, for those who are behind in retirement savings and trying to catch up right before retirement the Roth conversions could actually result in less final wealth than if it is simply left as tax-deferred. Note - See my article “Roth Conversions Are Overrated”.

What About Social Security Timing?

Previously this article suggested a plan for determining what your Social Security will be when you retire in order to calculate how much will need from investments. For those who are behind in retirement savings and trying to catch up right before retirement, it’s often your best strategy to take it as early as possible. However, be aware the 2026 rule has an annual earnings limit of $23,400 until your Full Retirement Age (FRA) - which is currently 67. Note - See my article “What is the Best Age to Take Social Security?”

What About a Variable Withdrawal Strategy?

As previously shown in the Dave example where he turned a losing strategy into a winning one, it turns out the most effective element was to boost his asset mix from 24/76 to 60/40. This is because stocks are the growth engine. But they are also volatile. As you can see on the “winning” trajectory chart, even though all paths are successful, there is a great deal of variations in all these paths - due to the market volatility. He could have boosted his mix to 80/20, but that would have created even more variation accompanied by even larger occasional large drawdowns.

Even though the larger stock position is causing the large variation, a key factor that makes this worse is the fixed withdrawals. If the market crashes but those fixed withdrawals continue regardless, this has the effect of depleting assets such that it takes longer to rebuild the asset value after the market recovers.

One potential solution - if you can afford to do it - is to execute your withdrawals as a percentage of your total asset value rather than a fixed amount. This has some very positive advantages:

• It lessens the spread variation of all those trajectories

• It allows you to increase the ratio of stocks to bonds - thus more average growth

Of course, the variable withdrawal strategy isn’t for everyone. You have to be able to live with lower income for a while after market corrections and crashes until the market recovers. It also requires an extra level of discipline to determine what your current withdrawal limit should be.

Note - See my article “The Case for an Optimal Withdrawal Strategy”.

Avoid Get-Rich-Quick Schemes

Finally, if you are behind in your retirement savings and panicking, you may be tempted to throw it all in on some kind of “Get-Rich-Quick” scheme. A big problem is that the media has no shortage of articles about such-and-such person becoming an instant multi-millionaire via some scheme. Maybe they invested early in some crypto mem coin or some unknown stock that just went to the Moon.

These make for great stories, but the problem is that for every instant multi-millionaire, there are hundreds or thousands that have lost all their money in these schemes. There aren’t media stories for all these people that lost. If you do this, you will almost certainly be one of them.

Again, your best bet is to stay with known and established investments, such as a stock index fund(s) and bonds or bond funds.

Where to get the v4.1 Retirement Simulation

The v4.1 retirement simulation can be found here (It may ask you to sign into your Google account):

https://colab.research.google.com/drive/1ax05ZxJwiyWllVpFOtkA8mtM39aemkX8?usp=sharing

All you need to do is adjust the sliders and enter your data. Then hit the “run” button.

© 2026 Byron Birkedahl. All rights reserved. No part of this publication may be reproduced, distributed, or transmitted in any form or by any means without prior written permission.

Disclaimer

This article and the accompanying retirement model are for educational and illustrative purposes only. They do not constitute specific tax, legal, or investment advice. The model relies on historical data, and past performance is not indicative of future results.

While this software has been tested for accuracy, it is provided on an “as is” basis without warranties of any kind. There is no guarantee that the model is free from errors or that it will accurately predict your specific financial future. Do not make financial decisions based solely on this tool; always consult a qualified professional (CPA, Attorney, or Registered Investment Advisor) before making major life decisions.

I first heard about the “Covered Call” strategy a few years ago. YouTube videos were popping up claiming that this strategy is a sure-fire way to boost your equity investment returns above what the equities could do on their own. Free money without any added risk!

The premise behind the strategy is simply that if you already have an asset like an index fund, such as an S&P 500 Index fund, then you sell covered call options on it. You sell these options at a strike price somewhat higher than the current price on your fund assets. If you do this each month you collect monthly premiums. Since their rationale is that mostly these options expire worthless, you collect all that “gravy” to boost your return. Occasionally, if the options expire with an “in-the-money” value that causes the buyers to “exercise the options” (causing you to forfeit your asset), then you simply buy the asset right back.

Several years ago, ETF funds started popping up that use the covered call strategy to collect yield payouts. A very common type of this ETF has a guaranteed 12% annual payout yield. On the surface, this sounds like a great deal - until you look under the hood.

What is a Covered Call Option?

A covered call is an options strategy where an investor sells (or “writes”) a call option on a stock or fund they already own. By doing this, the investor collects an upfront payment, known as a premium, in exchange for giving the option buyer the right to purchase their shares at a predetermined price—the “strike price”—within a specific timeframe. The strategy is “covered” because the investor already holds the underlying shares, ensuring they can deliver them if the buyer exercises their option.

The Basic Fallacy of the Covered Call Strategy for Boosting Your Return

As mentioned, the premise of this strategy is that most of the time you will collect the option sales premiums, but sometimes the option will expire “in-the-money” and you will have to forfeit your asset (and then presumably buy it back to stay in the game).

The problem is that in the cases where the option expires in-the-money, it will typically be worth far more than the premium collected. So in these cases your asset was forfeited, and you lose out on all those profits compared to if you simply held onto your assets without selling options on them. So, the question is: Do the sum of all the option premiums you collected beat the sum of all the lost profits to your underlying assets from the cases where your options were exercised?

There is a mathematical answer provided by the financial mathematicians Black and Scholes (B-S) for determining a fair options price. Without getting into the math details, the B-S “fair price” calculated for an option is the price that has no statistical return advantage for either buying or selling that option, or simply holding onto the underlying asset and not buying or selling an option for it in the first place.

What many do not understand is that options do not generate wealth. They can be used to purchase “insurance” (i.e. hedge a certain risk) or transfer one type of risk to another type of risk. This is why so many financially engineered products use them. They can take a type of risk that investors fear and reduce this - but at the expense of increasing another type of risk that perhaps the investor either doesn’t fear or doesn’t realize can happen. They are essentially a “zero sum game” engine.

To answer the previous question: With perfectly “fairly priced” options per the B-S formula (which is well-known by investment analysts) the sum of all the option premiums collected is exactly equal to the sum of all the lost profits from the cases where the options were exercised. So, in a perfect world where options are always fairly priced, and without other “investment drag” issues, the statistical best that the covered call strategy can do is completely break-even compared to simply holding the asset. But reality is not this perfect world, and we will cover the nuances later.

How the Covered Call High-Yield Income Funds Work

These funds collect option premiums on an underlying asset such as an S&P 500 Index fund. Then they use the collected premiums to pay out dividend yields. For example, if they are advertising a 12% annual yield, and they are selling covered calls each month, they would need to collect an option premium that pays 1% of the underlying asset value each month.

By now you may ask “How can the fixed dividend yield be 12% when the underlying asset is returning less?” For example, the historical average return of the S&P500 is ~10%. As previously mentioned, fairly priced options provide no statistical advantages for buyers versus sellers, or those simply holding the underlying asset. The math isn’t adding up.

So, how does an average 10% return magically turn into a guaranteed 12%? And here is the big rub: In order to turn 10% into 12% the only way to do this over time is to raid the principal. This means that statistically, this ETF will degrade in value over time. The fund will need to rob the principal - and this is the illusion of calling it a guaranteed 12% income fund based on a 10% wealth-generating asset. In fact, they could literally make this fund a guaranteed 15% fund, or a 30% fund by selling more expensive call options. This just results in a higher rate of principal depletion.

Guaranteed high-yield returns using options eventually results in depleted principal - a perfect example of reducing one type of risk while increasing another.

Note there are also versions of this type of fund where the underlying asset is a single stock rather than a broad index fund. For example, they may use NVIDIA as the underlying. It works the same way, but it is the same strategy on steroids - with much higher risk.

But it Gets Worse..

At first it might seem like a 12% Covered Call High-Yield Income Fund (aka CC fund) based on the S&P 500 Index is equivalent to simply owning a portfolio composed of an S&P 500 Index ETF (e.g. VOO) and withdrawing 1% of the portfolio value each month. In that perfect world they would be equivalent - in the real world they are not.

The CC fund has four investment drag factors that make it significantly worse than simply owning VOO:

1. Loss of Dividend Returns: VOO includes dividends in its total returns while the CC fund’s option valuations only address the index prices. Thus, the options pricing penalizes the seller for the dividend, effectively causing you to forfeit that return advantage.

2. High Fund Expense Ratios: The expense ratio (annual fees) for VOO is extremely low - at 0.03%. CC funds are “managed funds” with a much higher expense ratio - typically 0.60%. This may not sound like much, but this makes a significant difference over time. Note - see my article on the effects of fees.

3. Tax Disadvantages: CC funds create a “tax double-whammy”: they trigger ordinary income tax rates on option premiums, and they force the realization of gains every month. Unlike VOO, where you control the timing and benefit from lower capital gains rates on your withdrawals, CC funds strip you of tax agency and maximize your annual liability.

4. Non-Ideal Option Pricing: Actual option prices are not necessarily the same as what we may think is the correct B-S calculated price based on historical statistics. The market dictates the options prices based on what sellers offer and the buyers pay. My research has shown that on average “out-of-the money” calls, such as those used in CC strategies, tend to be cheaper than they should be. This does not bode well for sellers of covered calls. When calls are “cheap”, call sellers are in a disadvantaged state, and this further gives the VOO strategy the advantage.

A python model has been created to simulate an S&P 500-based high yield income fund (CC fund) versus a standard S&P 500 Index fund (VOO) in regard to portfolio balances and cash flow. In order to make an apples-to-apples comparison, both were set to extract 1% per month (12% annually) as cashflow. The model simulates all 4 of the investment “drag factors” described above. Using realistic data, it turns out the simple VOO strategy is far better, both in portfolio balance and cashflow.

Conclusion

As we have seen, in theory, covered call strategies to boost returns can at best only break even compared to simply owning the underlying asset.

As for the Covered Call High-Yield Income Funds, where the underlying asset is the S&P 500 Index with a guaranteed 12% yield, the analysis shows this is actually an illusion. The fund can’t statistically provide that kind of return over time without raiding the principal. To bluntly make the point on this illusion: If I have $1M in cash under my mattress and I “pay myself” $10K/ month from it, I wouldn’t consider this mattress fund is an investment vehicle with a 12% annual yield!

But even worse are the four drag factors these funds will have. These four turn a break-even investment into a bad choice. Simply go with the S&P 500 Index fund.

© 2026 Byron Birkedahl. All rights reserved. No part of this publication may be reproduced, distributed, or transmitted in any form or by any means without prior written permission.

Disclaimer

This article and the accompanying model are for educational and illustrative purposes only. They do not constitute specific tax, legal, or investment advice. The model relies on historical data, and past performance is not indicative of future results.

While the model software has been tested for accuracy, the results are provided on an “as is” basis without warranties of any kind. There is no guarantee that the model is free from errors or that it will accurately predict your specific financial future. Do not make financial decisions based solely on this tool; always consult a qualified professional (CPA, Attorney, or Registered Investment Advisor) before making major life decisions.

The premise behind the strategy is simply that if you already have an asset like an index fund, such as an S&P 500 Index fund, then you sell covered call options on it. You sell these options at a strike price somewhat higher than the current price on your fund assets. If you do this each month you collect monthly premiums. Since their rationale is that mostly these options expire worthless, you collect all that “gravy” to boost your return. Occasionally, if the options expire with an “in-the-money” value that causes the buyers to “exercise the options” (causing you to forfeit your asset), then you simply buy the asset right back.

Several years ago, ETF funds started popping up that use the covered call strategy to collect yield payouts. A very common type of this ETF has a guaranteed 12% annual payout yield. On the surface, this sounds like a great deal - until you look under the hood.

What is a Covered Call Option?

A covered call is an options strategy where an investor sells (or "writes") a call option on a stock or fund they already own. By doing this, the investor collects an upfront payment, known as a premium, in exchange for giving the option buyer the right to purchase their shares at a predetermined price—the "strike price"—within a specific timeframe. The strategy is "covered" because the investor already holds the underlying shares, ensuring they can deliver them if the buyer exercises their option.

The Basic Fallacy of the Covered Call Strategy for Boosting Your Return

As mentioned, the premise of this strategy is that most of the time you will collect the option sales premiums, but sometimes the option will expire “in-the-money” and you will have to forfeit your asset (and then presumably buy it back to stay in the game).

The problem is that in the cases where the option expires in-the-money, it will typically be worth far more than the premium collected. So in these cases your asset was forfeited, and you lose out on all those profits compared to if you simply held onto your assets without selling options on them. So, the question is: Do the sum of all the option premiums you collected beat the sum of all the lost profits to your underlying assets from the cases where your options were exercised?

There is a mathematical answer provided by the financial mathematicians Black and Scholes (B-S) for determining a fair options price. Without getting into the math details, the B-S “fair price” calculated for an option is the price that has no statistical return advantage for either buying or selling that option, or simply holding onto the underlying asset and not buying or selling an option for it in the first place.

What many do not understand is that options do not generate wealth. They can be used to purchase “insurance” (i.e. hedge a certain risk) or transfer one type of risk to another type of risk. This is why so many financially engineered products use them. They can take a type of risk that investors fear and reduce this - but at the expense of increasing another type of risk that perhaps the investor either doesn’t fear or doesn’t realize can happen. They are essentially a “zero sum game” engine.

To answer the previous question: With perfectly “fairly priced” options per the B-S formula (which is well-known by investment analysts) the sum of all the option premiums collected is exactly equal to the sum of all the lost profits from the cases where the options were exercised. So, in a perfect world where options are always fairly priced, and without other “investment drag” issues, the statistical best that the covered call strategy can do is completely break-even compared to simply holding the asset. But reality is not this perfect world, and we will cover the nuances later.

How the Covered Call High-Yield Income Funds Work

These funds collect option premiums on an underlying asset such as an S&P 500 Index fund. Then they use the collected premiums to pay out dividend yields. For example, if they are advertising a 12% annual yield, and they are selling covered calls each month, they would need to collect an option premium that pays 1% of the underlying asset value each month.

By now you may ask “How can the fixed dividend yield be 12% when the underlying asset is returning less?” For example, the historical average return of the S&P500 is ~10%. As previously mentioned, fairly priced options provide no statistical advantages for buyers versus sellers, or those simply holding the underlying asset. The math isn’t adding up.

So, how does an average 10% return magically turn into a guaranteed 12%? And here is the big rub: In order to turn 10% into 12% the only way to do this over time is to raid the principal. This means that statistically, this ETF will degrade in value over time. The fund will need to rob the principal - and this is the illusion of calling it a guaranteed 12% income fund based on a 10% wealth-generating asset. In fact, they could literally make this fund a guaranteed 15% fund, or a 30% fund by selling more expensive call options. This just results in a higher rate of principal depletion.

Guaranteed high-yield returns using options eventually results in depleted principal - a perfect example of reducing one type of risk while increasing another.

Note there are also versions of this type of fund where the underlying asset is a single stock rather than a broad index fund. For example, they may use NVIDIA as the underlying. It works the same way, but it is the same strategy on steroids - with much higher risk.

But it Gets Worse..

At first it might seem like a 12% Covered Call High-Yield Income Fund (aka CC fund) based on the S&P 500 Index is equivalent to simply owning a portfolio composed of an S&P 500 Index ETF (e.g. VOO) and withdrawing 1% of the portfolio value each month. In that perfect world they would be equivalent - in the real world they are not.

The CC fund has four investment drag factors that make it significantly worse than simply owning VOO:

1. Loss of Dividend Returns: VOO includes dividends in its total returns while the CC fund’s option valuations only address the index prices. Thus, the options pricing penalizes the seller for the dividend, effectively causing you to forfeit that return advantage.

2. High Fund Expense Ratios: The expense ratio (annual fees) for VOO is extremely low - at 0.03%. CC funds are “managed funds” with a much higher expense ratio - typically 0.60%. This may not sound like much, but this makes a significant difference over time. Note - see my article on the effects of fees.

3. Tax Disadvantages: CC funds create a “tax double-whammy”: they trigger ordinary income tax rates on option premiums, and they force the realization of gains every month. Unlike VOO, where you control the timing and benefit from lower capital gains rates on your withdrawals, CC funds strip you of tax agency and maximize your annual liability.

4. Non-Ideal Option Pricing: Actual option prices are not necessarily the same as what we may think is the correct B-S calculated price based on historical statistics. The market dictates the options prices based on what sellers offer and the buyers pay. My research has shown that on average “out-of-the money” calls, such as those used in CC strategies, tend to be cheaper than they should be. This does not bode well for sellers of covered calls. When calls are “cheap”, call sellers are in a disadvantaged state, and this further gives the VOO strategy the advantage.

A python model has been created to simulate an S&P 500-based high yield income fund (CC fund) versus a standard S&P 500 Index fund (VOO) in regard to portfolio balances and cash flow. In order to make an apples-to-apples comparison, both were set to extract 1% per month (12% annually) as cashflow. The model simulates all 4 of the investment “drag factors” described above. Using realistic data, it turns out the simple VOO strategy is far better, both in portfolio balance and cashflow.

Note - In the Premium Paid Subscriber section below, I provide the details of the simulation results. I also provide some insights into the meaning of the results and how the data feeding the simulation was determined. In addition, the model is provided so you can run your own data.

Conclusion

As we have seen, in theory, covered call strategies to boost returns can at best only break even compared to simply owning the underlying asset.

As for the Covered Call High-Yield Income Funds, where the underlying asset is the S&P 500 Index with a guaranteed 12% yield, the analysis shows this is actually an illusion. The fund can’t statistically provide that kind of return over time without raiding the principal. To bluntly make the point on this illusion: If I have $1M in cash under my mattress and I “pay myself” $10K/ month from it, I wouldn’t consider this mattress fund is an investment vehicle with a 12% annual yield!

But even worse are the four drag factors these funds will have. These four turn a break-even investment into a bad choice. Simply go with the S&P 500 Index fund.

© 2026 Byron Birkedahl. All rights reserved. No part of this publication may be reproduced, distributed, or transmitted in any form or by any means without prior written permission.

Disclaimer

This article and the accompanying model are for educational and illustrative purposes only. They do not constitute specific tax, legal, or investment advice. The model relies on historical data, and past performance is not indicative of future results.

While the model software has been tested for accuracy, the results are provided on an “as is” basis without warranties of any kind. There is no guarantee that the model is free from errors or that it will accurately predict your specific financial future. Do not make financial decisions based solely on this tool; always consult a qualified professional (CPA, Attorney, or Registered Investment Advisor) before making major life decisions.

Deep Dive for Paid Subscribers

Read more

Some of you may be aware that a survey of 10000 millionaires was conducted from 2017 - 2018, “Ramsey Solutions - National Study of Millionaires”. When participants were asked what their primary career was, these five professions came up with this ranking:

1. Engineer

2. Accountant (CPA)

3. Teacher

4. Management (Corporate Manager)

5. Attorney

I found it very interesting that doctors did not end up in the top 5 - even though this profession is the highest paid profession in the U.S. Out of 800 professions (tracked by the Bureau of Labor Statistics) teachers only rank at about 320th.

So how is it possible teachers were 3rd on the list and doctors didn’t even make the list?

For the “The Engineer Investor,” this Doctor vs. Teacher Wealth Paradox is intriguing to me and therefore the subject of this article.

Other Survey Stats

There were several other interesting statistics determined from the 2017-2018 survey:

• Only 31% of the millionaires averaged over $100,000 a year over the course of their career. 33% of the millionaires never made six figures in any single working year of their entire life.

• Only 15% of the millionaires surveyed held senior leadership or executive roles (CEO, CFO, VP, etc.).

• 80% of the millionaires built their wealth simply by investing consistently in their employer’s 401(k), 403(b), or similar plans.

• It took an average of 28 years of consistent investing to hit the $1 million mark (hitting it, on average, at age 49).

• 79% received zero inheritance.

• 94% stated that they live on less than they make.

This data is starting to paint an interesting picture.

A Tale of the Tortoise and the Hare

When we look at the financial savings and investment factors between an average teacher versus an average doctor, stark differences emerge.

The Doctor

An average doctor goes through many years of expensive schooling, and typically racks up a huge amount of medical school loan debt in their 20’s. Then they go through a residency program, not earning much all the while. Perhaps at age 32, residency is over and the doctor income finally kicks in - but they still need to dig their way out of the medical school debt. Even with a relatively large savings rate, it may not be until age 37 when they are finally able to dig out of the debt hole to get their net wealth back to zero.

According to the Medscape Physician Wealth and Debt Report (2024), 60% of physicians carry a massive mortgage, 31% carry luxury car leases, 26% have month-to-month credit card debt, and 21% are still paying off school loans well into their careers. Financial planners note that doctors are heavily targeted by lenders with “Physician Mortgages” (0% down on million-plus-dollar homes) and luxury auto leases.

Doctors frequently report a “delayed gratification” backlash. Because they suffered through a decade of being broke in their 20s, the psychological need to “reward themselves” in their 30s overrides standard financial advice.

The Teacher

Contrast this with an average teacher. Even though their income is relatively low, most teachers have the “forced savings” of a significant pension benefit when they retire. In fact, for an average teacher’s income, the pension benefit may have a value of $1M or more at age 65 (in real dollars). This means that even if the teacher saves nothing, they still have a significant source of wealth at retirement.

In addition, studies on retirement plans (like those from TIAA and Vanguard) consistently show that once teachers set up a 403(b) contribution, they rarely tinker with it. They “set it and forget it,” allowing them to capture decades of uninterrupted compound interest.

A 24-year-old teacher is not expected by society (or their peers) to buy a 4,000-square-foot house or drive a leased BMW. Their peer group tends to normalize a modest, middle-class lifestyle, which completely removes the psychological pressure to overspend.

The Race

After looking at these characteristics, we can make a generalization to conclude the doctor is analogous to the fabled “Hare” and the teacher is the “Tortoise” when it comes to personal finances. The teacher is likely to be methodical, slow and steady, and able to compound wealth at an early age. The doctor gets a very delayed start, but (in theory) able to race toward the finish at a blistering pace. But who will win?

Doctor Versus Teacher Wealth Simulation (“The Race”)

In order to see how the differences between doctor and teacher wealth might evolve, I created a Doctor vs Teacher Wealth simulator. This simulator allows us to compare the wealth that a doctor versus a teacher may accumulate in their savings until retirement.

This doctor simulation assumes residency to age 32, with no savings until after this age. User inputs are provided to adjust doctor income (after residency), savings in real return (i.e. return adjusted for inflation), medical school debt, and savings rate.

The teacher simulation assumes no school debt, an accumulating typical pension value, and other savings such as 403(b) contributions starting immediately. User inputs include teacher income, savings real return (savings return adjusted for inflation), and savings rate.

The simulation optionally adjusts results for after tax results if selected (based on rough estimations of taxes for each wealth calculation). The wealth values are computed in Real Dollars (factoring out inflation). For simplicity, the income values are fixed in real dollars over the career span. In reality, income in real dollars will likely change over the career span, but the average value is chosen to demonstrate the principle.

Note this simulation is available to paid subscribers - with a link provided below, so you can run your own numbers.

Example

Below is a simulation run between a doctor and teacher where the teacher’s average career income is $65K/year and the doctor’s average career income is $250K/year. These values were chosen in this example because they are averages across the U.S. for teachers and primary care physicians (sources: NEA, Medscape).

The example shows that if the teacher and doctor have the same investment real return rates (5%) and savings rates (14%), they roughly break even at age 65 - although you can see the teacher is ahead the whole time. In this example, the doctor is massively penalized by the 15-year “delayed start”.

In changing the parameters, it should be noted that the outcome is highly dependent on the savings rates. At 14% (for both) they end up about the same as shown. At 10%, the teacher wins by a lot. At 20% the doctor wins - but not by the same margin as the teacher - at 10% savings rates. Doctors take note: It’s not easy to catch up!

Of course, the other big factor is income level. If the doctor’s average income is actually $400K/year then they win by a lot. But… if the $400K/year doctor’s savings rate is only 9% and teacher is at 14%, they are back to a tie.

Even though this example is hypothetical, it attempts to use realistic numbers and timeframes. This result may point to why teachers are #3 on the millionaire survey and doctors did not make the top 5.

Note the model is available to Paid subscribers in the original article here.

Other Tortoise and Hare Examples

Engineer

It’s not difficult for me to see why engineer was #1 in the survey’s millionaires ranking - after all that was my profession. When I look at my retired and near-retired engineer colleagues, I see they are all, without exception, very well off. Why is that?

Engineers have similar characteristics to teachers. Typically, they are methodical, long-term, and steady in their savings. They have set-it-and-forget-it 401(k) plans. The big difference is the profession pays significantly more than a teacher, but typically without a pension.

If the teacher is the “Tortoise,” the engineer is the “Turbocharged Tortoise”.

Professional Athlete (NBA, NFL)

At the other extreme we have a professional athlete such as an NBA basketball player or an NFL football player. The current average yearly salary for an NFL player is $2.8M to $3.2M. For an NBA player it’s a whopping $11.9M to $13.2M!

At first it might be assumed ex-players must all be wealthy multi-millionaires. But an NFL career only averages 3.3 years, and NBA career only 4.5 to 4.8 years. The average retirement age is just 26 years old.

It might seem hard to believe the average NFL & NBA career is so short, but the combination of brutal physical tolls and fierce competition for limited roster spots means the vast majority of players are injured or replaced by younger, cheaper talent before their first contract even expires.

Truly, the average NBA or NFL professional athlete is a “Hare” with a quick-burning high-thrust rocket pack.

While there are cases of ex-players doing financially well later in life, unfortunately the stats are not so good for the majority:

• According to a 2009 Sports Illustrated investigation, an estimated 78% of NFL players are either bankrupt or under serious financial stress within just two years of retiring

• That same SI report estimated that roughly 60% of NBA players face financial ruin within five years of leaving the league.

For many (or most) of these players a ticket to the NFL or NBA at age 21 or 22 must seem like winning the lottery. It would be difficult to imagine most - especially at this young age - would be capable of adopting a methodical, long-term, and steady financial plan.

The Moral of the Story…

This article is not about which profession is best - so hopefully it is not interpreted as such.

The moral of this story is that long term, methodical, investment-compounding by the “tortoises” can easily win out for accumulated total wealth in the end - even with a relatively low income - compared to the much higher income “hares”.

Those with high income careers, particularly when off to a late start or limited duration, should understand how much this affects their wealth in the end. Adopting some of the “tortoise” characteristics, such as avoiding excessive spending, staying away from sketchy investments & advisors, and putting solid investments on autopilot, including making significant automatic deductions to 401(k), 403(b), 457(b), IRAs, taxable accounts, etc., can make all the difference.

What’s Your Opinion?

Do you think there are other reasons the millionaire survey came out as it did? If so, please comment on this article!

© 2026 Byron Birkedahl. All rights reserved. No part of this publication may be reproduced, distributed, or transmitted in any form or by any means without prior written permission.

Disclaimer

This article and the accompanying model are for educational and illustrative purposes only. They do not constitute specific tax, legal, or investment advice. The model relies on historical data, and past performance is not indicative of future results.

While the model software has been tested for accuracy, the results are provided on an “as is” basis without warranties of any kind. There is no guarantee that the model is free from errors or that it will accurately predict your specific financial future. Do not make financial decisions based solely on this tool; always consult a qualified professional (CPA, Attorney, or Registered Investment Advisor) before making major life decisions.

1. Engineer

2. Accountant (CPA)

3. Teacher

4. Management (Corporate Manager)

5. Attorney

I found it very interesting that doctors did not end up in the top 5 - even though this profession is the highest paid profession in the U.S. Out of 800 professions (tracked by the Bureau of Labor Statistics) teachers only rank at about 320th.

So how is it possible teachers were 3rd on the list and doctors didn’t even make the list?

For the “The Engineer Investor,” this Doctor vs. Teacher Wealth Paradox is intriguing to me and therefore the subject of this article.

Other Survey Stats

There were several other interesting statistics determined from the 2017-2018 survey:

• Only 31% of the millionaires averaged over $100,000 a year over the course of their career. 33% of the millionaires never made six figures in any single working year of their entire life.

• Only 15% of the millionaires surveyed held senior leadership or executive roles (CEO, CFO, VP, etc.).

• 80% of the millionaires built their wealth simply by investing consistently in their employer’s 401(k), 403(b), or similar plans.

• It took an average of 28 years of consistent investing to hit the $1 million mark (hitting it, on average, at age 49).

• 79% received zero inheritance.

• 94% stated that they live on less than they make.

This data is starting to paint an interesting picture.

A Tale of the Tortoise and the Hare

When we look at the financial savings and investment factors between an average teacher versus an average doctor, stark differences emerge.

The Doctor

An average doctor goes through many years of expensive schooling, and typically racks up a huge amount of medical school loan debt in their 20’s. Then they go through a residency program, not earning much all the while. Perhaps at age 32, residency is over and the doctor income finally kicks in - but they still need to dig their way out of the medical school debt. Even with a relatively large savings rate, it may not be until age 37 when they are finally able to dig out of the debt hole to get their net wealth back to zero.

According to the Medscape Physician Wealth and Debt Report (2024), 60% of physicians carry a massive mortgage, 31% carry luxury car leases, 26% have month-to-month credit card debt, and 21% are still paying off school loans well into their careers. Financial planners note that doctors are heavily targeted by lenders with “Physician Mortgages” (0% down on million-plus-dollar homes) and luxury auto leases.

Doctors frequently report a “delayed gratification” backlash. Because they suffered through a decade of being broke in their 20s, the psychological need to “reward themselves” in their 30s overrides standard financial advice.

The Teacher

Contrast this with an average teacher. Even though their income is relatively low, most teachers have the “forced savings” of a significant pension benefit when they retire. In fact, for an average teacher’s income, the pension benefit may have a value of $1M or more at age 65 (in real dollars). This means that even if the teacher saves nothing, they still have a significant source of wealth at retirement.

In addition, studies on retirement plans (like those from TIAA and Vanguard) consistently show that once teachers set up a 403(b) contribution, they rarely tinker with it. They “set it and forget it,” allowing them to capture decades of uninterrupted compound interest.

A 24-year-old teacher is not expected by society (or their peers) to buy a 4,000-square-foot house or drive a leased BMW. Their peer group tends to normalize a modest, middle-class lifestyle, which completely removes the psychological pressure to overspend.

The Race

After looking at these characteristics, we can make a generalization to conclude the doctor is analogous to the fabled “Hare” and the teacher is the “Tortoise” when it comes to personal finances. The teacher is likely to be methodical, slow and steady, and able to compound wealth at an early age. The doctor gets a very delayed start, but (in theory) able to race toward the finish at a blistering pace. But who will win?

Doctor Versus Teacher Wealth Simulation (“The Race”)

In order to see how the differences between doctor and teacher wealth might evolve, I created a Doctor vs Teacher Wealth simulator. This simulator allows us to compare the wealth that a doctor versus a teacher may accumulate in their savings until retirement.

This doctor simulation assumes residency to age 32, with no savings until after this age. User inputs are provided to adjust doctor income (after residency), savings in real return (i.e. return adjusted for inflation), medical school debt, and savings rate.

The teacher simulation assumes no school debt, an accumulating typical pension value, and other savings such as 403(b) contributions starting immediately. User inputs include teacher income, savings real return (savings return adjusted for inflation), and savings rate.

The simulation optionally adjusts results for after tax results if selected (based on rough estimations of taxes for each wealth calculation). The wealth values are computed in Real Dollars (factoring out inflation). For simplicity, the income values are fixed in real dollars over the career span. In reality, income in real dollars will likely change over the career span, but the average value is chosen to demonstrate the principle.

Note this simulation is available to paid subscribers - with a link provided below, so you can run your own numbers.

Example

Below is a simulation run between a doctor and teacher where the teacher’s average career income is $65K/year and the doctor’s average career income is $250K/year. These values were chosen in this example because they are averages across the U.S. for teachers and primary care physicians (sources: NEA, Medscape).

The example shows that if the teacher and doctor have the same investment real return rates (5%) and savings rates (14%), they roughly break even at age 65 - although you can see the teacher is ahead the whole time. In this example, the doctor is massively penalized by the 15-year “delayed start”.

In changing the parameters, it should be noted that the outcome is highly dependent on the savings rates. At 14% (for both) they end up about the same as shown. At 10%, the teacher wins by a lot. At 20% the doctor wins - but not by the same margin as the teacher - at 10% savings rates. Doctors take note: It’s not easy to catch up!

Of course, the other big factor is income level. If the doctor’s average income is actually $400K/year then they win by a lot. But… if the $400K/year doctor’s savings rate is only 9% and teacher is at 14%, they are back to a tie.

Even though this example is hypothetical, it attempts to use realistic numbers and timeframes. This result may point to why teachers are #3 on the millionaire survey and doctors did not make the top 5.

Other Tortoise and Hare Examples

Engineer

It’s not difficult for me to see why engineer was #1 in the survey’s millionaires ranking - after all that was my profession. When I look at my retired and near-retired engineer colleagues, I see they are all, without exception, very well off. Why is that?

Engineers have similar characteristics to teachers. Typically, they are methodical, long-term, and steady in their savings. They have set-it-and-forget-it 401(k) plans. The big difference is the profession pays significantly more than a teacher, but typically without a pension.

If the teacher is the “Tortoise,” the engineer is the “Turbocharged Tortoise”.

Professional Athlete (NBA, NFL)

At the other extreme we have a professional athlete such as an NBA basketball player or an NFL football player. The current average yearly salary for an NFL player is $2.8M to $3.2M. For an NBA player it’s a whopping $11.9M to $13.2M!

At first it might be assumed ex-players must all be wealthy multi-millionaires. But an NFL career only averages 3.3 years, and NBA career only 4.5 to 4.8 years. The average retirement age is just 26 years old.

It might seem hard to believe the average NFL & NBA career is so short, but the combination of brutal physical tolls and fierce competition for limited roster spots means the vast majority of players are injured or replaced by younger, cheaper talent before their first contract even expires.

Truly, the average NBA or NFL professional athlete is a “Hare” with a quick-burning high-thrust rocket pack.

While there are cases of ex-players doing financially well later in life, unfortunately the stats are not so good for the majority:

• According to a 2009 Sports Illustrated investigation, an estimated 78% of NFL players are either bankrupt or under serious financial stress within just two years of retiring

• That same SI report estimated that roughly 60% of NBA players face financial ruin within five years of leaving the league.

For many (or most) of these players a ticket to the NFL or NBA at age 21 or 22 must seem like winning the lottery. It would be difficult to imagine most - especially at this young age - would be capable of adopting a methodical, long-term, and steady financial plan.

The Moral of the Story…

This article is not about which profession is best - so hopefully it is not interpreted as such.

The moral of this story is that long term, methodical, investment-compounding by the “tortoises” can easily win out for accumulated total wealth in the end - even with a relatively low income - compared to the much higher income “hares”.

Those with high income careers, particularly when off to a late start or limited duration, should understand how much this affects their wealth in the end. Adopting some of the “tortoise” characteristics, such as avoiding excessive spending, staying away from sketchy investments & advisors, and putting solid investments on autopilot, including making significant automatic deductions to 401(k), 403(b), 457(b), IRAs, taxable accounts, etc., can make all the difference.

What’s Your Opinion?

Do you think there are other reasons the millionaire survey came out as it did? If so, please comment on this article!

© 2026 Byron Birkedahl. All rights reserved. No part of this publication may be reproduced, distributed, or transmitted in any form or by any means without prior written permission.

Disclaimer

This article and the accompanying model are for educational and illustrative purposes only. They do not constitute specific tax, legal, or investment advice. The model relies on historical data, and past performance is not indicative of future results.

While the model software has been tested for accuracy, the results are provided on an “as is” basis without warranties of any kind. There is no guarantee that the model is free from errors or that it will accurately predict your specific financial future. Do not make financial decisions based solely on this tool; always consult a qualified professional (CPA, Attorney, or Registered Investment Advisor) before making major life decisions.

Doctor versus Teacher Wealth Simulator

The simulator is available to paid subscribers with the link provided below.

The simulator is very easy to use. Simply copy the link to Google Colab in your browser and hit the “play” button. Set the sliders and check box as follows:

• Teacher Income: Average yearly income over the entire career ($65000 default)

• Doctor Income: Average yearly income over the entire career ($250000 default)

• Teacher/ Doctor Real Return: The average yearly % return rate of all investments minus inflation (5% default)

• Doctor Debt: The average debt carried from medical school loans ($200000)

• Teacher/ Doctor Savings: The average percentage of gross income that goes into savings (14% default)

• Show After-Tax Wealth: Select this box to show an estimation of wealth after taxes (note: if not selected it shows before tax)

Read more

• The Repricing Wall: A national debt where roughly one-third of all marketable liabilities must mature and roll over every 12 months into today’s higher interest rates.

• The Revenue Trap: An oncoming, labor-displacing AI automation wave that optimizes corporate margins but drains the government’s primary W-2 tax engine.

• The Empty Toolkit: A systemic scale so massive that standard political talking points—including austerity, stimulus, taxing billionaires, or liquidating federal land and gold - fail to close the deficit.

• The Household Reality: A liability that translates to a $288,000 debt burden per American household - a mismatch against a median household savings buffer of just $8,000.

Because deep political polarization virtually guarantees legislative gridlock on the Senate floor, the structural policy mechanisms required to safely steer this machine - like a pass-through Deflation Shield or National AI Stock Options - will remain paralyzed. Consequently, doing nothing remains the most likely default systemic choice.

When the legacy debt buffers officially expire at the turn of the decade, the Federal Reserve will be forced to activate its final, crude lever: the Argentina-style Escape Hatch. They will artificially cap interest rates and expand the money supply to keep the Treasury solvent, unleashing a compounding inflation wildfire that methodically liquidates the purchasing power of traditional savings to burn away the debt. The default economic simulation model pins this baseline structural breakdown arriving potentially as early as 2032.

What Part 9 Will Cover

Part 9 is designed to open up the financial engineering toolkit to answer the only practical question that remains: What can you actually do about it?

In the section below, I am going to break down the foundational psychological and structural paradigm shifts required to navigate the potential for a high-inflation, interest-rate-capped economic regime. We will explore why the forty-year baseline of American wealth management - the passive, set-it-and-forget-it index portfolio - would be vulnerable to a capped-rate environment, and why standard definitions of “financial safety” are inverted.

Then, exclusively for paid subscribers, I will hand out a step-by-step execution playbook to hedge this scenario.

Thanks for reading The Engineer Investor! This post is public, so feel free to share this introductory recap with your network.

The Default Scenario and Its Assumptions

Before we look at the asset allocation framework, I want to address some comments I received on part 8.

• Does the world basically end in 2032?

• Does the math show 2032 is the actual breaking point?

• Does this mean the wealthy will do fine while everyone else is completely wiped out?

It’s easy to see why Part 8 is so alarming to readers. Economic model simulations tend to make many of us think this is an exact “crystal ball” of the future. This is not the case. There are numerous assumptions that went into the worst-case economic model (aka default scenario) from Part 7 that may or may not occur as assumed. Some of these could delay the breakpoint by a half dozen years. There is even the (small) possibility the breakpoint does not occur. Here are the main influences:

The Worldwide Bond Traders: Where is the actual point bond traders start to declare U.S. treasury notes are becoming high risk? A 150% debt ratio was assumed, but if it’s really 175% this would delay the breakpoint timeframe.

The Fed: Will the Fed let interest rates float up with inflation (causing more debt to service the debt) or try to cap the interest rates and “print money”. At what point would they do either of these?

The Political System: The worst-case model assumes this system is paralyzed to the default path of doing nothing until it’s too late. But you never know - maybe a future political system is actually competent at trying to solve this problem.

The Automation/ AI Industry: The worst-case model assumes industry is 100% driven by greed in a winner-take-all mentality. While it’s hard to argue that this wouldn’t be the standard modus operandi of U.S. companies, there are some signs that some of the big players are starting to see that is not an ideal situation for the country - along with their own interests. For example, the latest news (June 2026) has mentioned that key AI industry leaders, Elon Musk, and even Trump are suggesting that the profits from AI should be captured by the U.S. Treasury, and benefits should be offered to those displaced by AI. These are actually elements of the “Supercritical Solution” simulation model in Part 5 & 6 in showing how to get out of this mess. So, are these just talking points or would they really be implemented?

Assuming we do end up with the breakpoint in 2032 this doesn’t mean the world is coming to an end

Crossing the breakpoint does not mean society stops functioning or the economy stops spinning. Even in the default scenario, we are looking at a state change in the value of the currency, not an end to civilization.

Nations like Argentina went through severe devaluations and emerged on the other side. People still went to work, businesses still operated, and families still survived. The goal of this Part 9 article is not only to predict a collapse, but to ensure your private capital transitions safely through a high-inflation regime rather than being used as the fuel to burn away the U.S. debt.

Capital Owners Will Fare Well

Yes, those who own high-quality corporate equity, automated infrastructure, and tangible resources will naturally navigate this default transition successfully. This isn’t due to political favoritism; it is due to the physics of finance.

When a currency is being systematically debased, real assets that possess independent pricing power automatically float to the top. The entire purpose of The Personal Escape Hatch is to give you the precise screening tools to position your balance sheet on the winning side of that ledger - assuming the breakpoint inflation scenario occurs.

Why the Standard Investment Safety Net Would End Up Failing

For the past forty years, conventional financial planning has relied on a static asset mix: a heavy dose of broad stock index funds for growth, and a safe allocation of government bonds, corporate bonds, and/or CDs to act as a shock absorber during market downturns.

In an unshielded, high-inflation environment where the Federal Reserve is capping interest rates, that traditional model breaks down completely.

When the central bank artificially holds interest rates below the rate of inflation to shrink the national debt, fixed-income assets like T-Bonds stop acting as a safety net. Instead, they become a primary mechanism for losing purchasing power. Investors would be locked into a low fixed return while the true cost of living climbs at a much higher rate. The real rate of return would be highly negative. Even worse would be savings accounts and CDs. And holding cash in a checking account would be the worst.

Debt-heavy corporations would be forced to offer their bonds at massive risk premiums. Unlike the Treasury, which can rely on the Fed to print money to avoid default, corporations cannot. They would be squeezed by astronomical borrowing costs while simultaneously facing a consumer base that can no longer afford their products - leading to corporate bankruptcies. While high-rate Corp bonds may sound good to an investor, it means these are very risky.

Bond investors would only have a choice between very low-interest T-Notes, CDs and Savings and very high-risk Corp bonds - with no middle ground. Not at all good for bond investing.

Equity investing in a broad index fund like the S&P 500 would force investors to indiscriminately buy the entire market basket. This means automatically purchasing debt-heavy corporations that would get caught on the wrong side of the repricing debt wall, right alongside businesses that rely entirely on a squeezed domestic consumer base.

To successfully navigate this scenario, investors would need to transition from a philosophy of fixed asset percentages to one of selective risk screening. They would filter for specific investments that can capture the immense margins of machine-driven productivity while remaining entirely insulated from local currency debasement.

Quantifying the Threat: The Supercritical Risk Index (SRI)

To monitor the current state of our situation, I have created a “Supercritical Risk Index (SRI)”. This index is basically a composite of seven variables that represent the risk of an imminent breakpoint as shown by the economic models used in this series.

The SRI is a single score scaled from 0 to 100. A score of 0 represents no risk, while a score of 100 represents the absolute boundary condition where the conditions would force the Federal Reserve to activate the unshielded Escape Hatch.

The current value is sitting right at 59 (bottom of the risk level 3 out of 4 levels).

Every three months, I will pull the updated data from the Treasury and the Federal Reserve, recalculate the index, and publish a comprehensive SRI Update exclusively for paid-tier subscribers.

This quarterly report will indicate which Risk Value & Level we are currently sitting in, which parameters are accelerating or decelerating, and when the data flashes the cue to potentially move to a breakpoint defensive strategy.

This marks the end of our free introductory briefing. Below I open the financial engineering toolkit to hand my paid-tier subscribers the actionable, step-by-step checklist to monitor the timing of our situation, insulate careers, and protect portfolios.

To unlock the complete Premium Playbook and gain access to the latest SRI chart and a description of how it is constructed, consider upgrading to a paid subscription today.

© 2026 Byron Birkedahl. All rights reserved. No part of this publication may be reproduced, distributed, or transmitted in any form or by any means without prior written permission.

Disclaimer

This article and the accompanying models are for educational and illustrative purposes only. They do not constitute specific tax, legal, or investment advice. The models rely on historical data, and past performance is not indicative of future results.

While the model software has been tested for accuracy, the results are provided on an “as is” basis without warranties of any kind. There is no guarantee that the model is free from errors or that it will accurately predict your specific financial future. Do not make financial decisions based solely on this tool; always consult a qualified professional (CPA, Attorney, or Registered Investment Advisor) before making major life decisions.

Read more

As some of you may know, The Engineer Investor articles are typically more “tactical” in nature and offer insight into investing & retirement strategies - mostly based on mathematically analyzing validated historical data.

HyMasa Financial Analytics covers topics that are more “strategic” in nature, such as where it appears the U.S. economy will be headed in the years ahead. These incorporate analysis techniques that utilize U.S. economic models. The tricky part is running scenarios of the models of what could happen if conditions we’ve never experienced before happen in the future. This includes issues such as the potential for human labor replacement due to AI and our out-of-control debt burden.

Because the Supercritical Series indicates changes in investment strategies may be warranted in the future, the Part 8 summary and Part 9 future investment strategies, are being offered in the “The Engineer Investor”. If you find it interesting please subscribe to HyMasa Financial Analytics!

This article is the overall summary of Parts 1-7 of my series “Supercritical” which looks at the raw physics of the U.S. economy. The series explores the mathematical collision of the $39 Trillion debt trap, the AI productivity shock, the coming reorganization of American wealth, the resulting political fracturing, and the likely effects on the U.S. stock market.

For this article, I am dropping the specialized economics jargon entirely and mapping out the core plumbing of the situation in plain English. Whether you are a long-time reader or joining me for the first time via The Engineer Investor or HyMasa Financial Analytics, this is a comprehensive diagnostic of where the economic system stands today, why the ground feels like it is shifting under our feet, and when the clock is likely to run out.

Over the last two months, I have dug deep into the numbers governing the U.S. economy. I have shared detailed data—from collapsing rates of return on government borrowing to shifting corporate profit margins and rising wealth gap measurements.

If you want to look at the series starting with Part 1, you can find it here.

The National Debt Clock and the Repricing Wall

To understand the national debt, skip the political talking points and look at the federal government like a highly successful surgeon who is simultaneously drowning in debt. She makes $800,000 a year, but because of decades of massive lifestyle commitments, she spends $1,000,000 a year. For a long time, banks gave her cheap loans to cover the gap because her income was so high. Eventually, however, her old loans began expiring, and she had to take out high-interest payday loans just to make the minimum payments on her old debt. She is running a multi-million-dollar balance sheet at a loss.

To ever get out of this mess, the surgeon needs a massive, immediate boost in her income just to keep her head above water.

The United States has entered this exact “Ponzi Phase”. The government is no longer borrowing money to build next-generation highways or space programs that expand the economy. The U.S. Treasury is forced to borrow trillions of dollars just to keep up with the compounding interest on the money it already owes.

For years, this reality was hidden by an artificial buffer: zero-percent interest rates. The government borrowed aggressively, but because interest costs were pinned near zero, the carrying cost was practically invisible.

Now, that illusion has shattered. The total national debt currently sits at a massive $39 Trillion. Worse, this debt is structured like a staggered ladder of individual loans. Roughly one-third of all marketable U.S. debt must mature and roll over every 12 months. As those cheap, old 1% and 2% loans expire, the Treasury has no cash to pay them off. It must renew them by issuing new loans at today’s much higher market rates of 4% to 5%.

This automatic “Repricing Wall” is methodically ratcheting up the government’s interest bill every single month, completely independent of any new budget decisions.

The Japan Illusion and the 122% Reality

As of June 2026, total Gross Federal Debt sits at 122% of the total size of our economy, and it is growing larger every day. Whenever an analysis points out how unsustainable this trajectory is, Wall Street typically deploys a common defense mechanism: pointing to Japan. They note that Japan’s debt ratio has been sitting at a staggering 250% for decades without the system collapsing.

This is a fundamental misunderstanding of financial structures. Japan runs a closed, internal debt loop. The Japanese government borrows predominantly from its own citizens and its own central bank, supported by a massive pool of domestic savings. If you borrow money from your grandmother, she isn’t going to aggressively foreclose on your house or panic and demand a sudden interest rate spike.

The United States is not Japan. While we have a deep domestic market, we rely heavily on foreign investors and global bond markets to buy our massive, continuous supply of new debt. When an external system carries too much debt, the marginal buyers stop acting like cooperative family members and begin acting like predatory payday lenders. They demand massive risk premiums just to hold the debt, causing bond yields to violently spike and forcing a sovereign insolvency spiral. We cannot run a grandmother-style debt level when our marginal buyers behave like payday lenders.

Why the Standard Solutions Will Fail

Just like our struggling surgeon, the U.S. economy requires a massive boost in productivity and economic output to outrun this debt clock. Looking at the technological horizon, Artificial Intelligence is realistically the only candidate capable of delivering that kind of historic efficiency surge before the clock runs out.

However, AI is a double-edged sword. In the 1990s, the internet boom was a labor-enhancing shock - it created millions of new jobs, expanded human payrolls, and flooded the Treasury with high-velocity tax revenue. AI is fundamentally labor-displacing. When corporations replace human payrolls with machine automation to optimize margins, the corporate cost savings are pocketed as private profit while the government’s primary revenue engine - W-2 payroll and income taxes - is drained.

When confronted with this reality, Washington relies on a standard toolkit of political options, but the economic models demonstrate that at a $39 Trillion scale, these traditional ideas fail to work:

• Austerity (Spending Cuts): If the government implements massive spending cuts, it triggers a destructive race. Slicing federal contracts instantly vaporizes private-sector revenue. Because of how government spending ripples through the economy, cutting $1 Trillion from the budget shrinks the total economic base by $1.2 to $1.5 Trillion. The economy shrinks faster than the debt is paid off, causing the debt ratio to accelerate higher. We saw this fail in Greece, and we saw it when the Department of Government Efficiency (DOGE) had to scale back its initial $2 Trillion target to modest cuts that failed to dent the deficit.

• Stimulus (Spending to Grow): Injecting borrowed trillions into the economy to “grow our way out” worked when our overall debt was low. Today, the system is choked with leverage. Every new dollar borrowed now generates a measly 20 cents of actual economic growth. A $1 Trillion stimulus bill adds a full trillion to our debt but only adds $200 Billion to the economic base. The debt ratio simply spikes higher.

• Taxing Corporations and High Earners: The combined net worth of every single billionaire in America is roughly $8 Trillion. While that sounds like an ocean of money, look at it against the actual scale of the problem. The national debt is pacing toward $40 Trillion, and total annual government spending is now well over $6 Trillion. If the U.S. could somehow pass a 100% confiscatory wealth tax tomorrow and seized every single penny from every billionaire in the country, it wouldn’t even keep the federal government running for a year and a half, let alone eliminate the underlying debt. And you can only do it once. The following year, that wealth pool is completely dry, but the baseline spending and interest costs keep compounding.

  Furthermore, forcing billionaires to liquidate massive blocks of stock to pay wealth taxes would flood the market with sellers, crushing equity valuations and dismantling the retirement accounts of the middle class. Meanwhile, raising corporate tax rates doesn’t just hurt big business; companies simply pass the cost forward to consumers via higher prices, hiring freezes, or by cutting dividends.

• National Sales Tax (VAT): Dropping a 15% to 20% national sales tax onto the economy would successfully generate trillions in revenue and halt new borrowing. However, because consumer spending drives 70% of the U.S. economy, a sudden price hike causes consumers to pull back drastically. Corporate revenues plummet, layoffs spike, and economic growth stalls out. The existing $39 Trillion debt continues to compound at 4% interest against a stalled economy, and the debt ratio climbs anyway.

• Liquidating Federal Assets (Land and Gold): A common financial fantasy is that the government can simply sell its physical assets to wipe the ledger clean. Selling all 600 million acres of federal land would generate roughly $3 Trillion under ideal conditions, but this would cover less than two years of our deficit and interest payments. Dumping it all at once would crash land prices to pennies. Similarly, selling every physical ounce of gold in Fort Knox at current market prices would yield around $1.3 Trillion - failing to cover even a single year’s deficit. Arbitrarily revaluing gold to $100,000 an ounce is just a monetary parlor trick that would instantly trigger international currency abandonment and cause interest rates to violently spike.

The Political Seesaw and the 2032 Breakpoint

Is there a functional solution to this problem? Yes. In my previous research, I designed and simulated a comprehensive 4-mechanism policy blueprint - combining an AI Acceleration Mandate, National AI Stock Options, a Deflation Shield, and a Modulated Fed Cap. Working in perfect synchronization, this architecture successfully uses AI productivity to vastly expand the total size of the economy, safely shrinking the debt mass to a stable 70% of the economy while funding a dividend for displaced labor and locking consumer inflation at 2%.

But while this engineering blueprint is mathematically sound, it must ultimately survive the friction of real-world governance. It requires an unprecedented level of political cooperation. Progressives would have to accept that massive corporate automation is necessary to grow the economy, and Conservatives would have to accept federal equity capture and strict corporate margin regulations to keep inflation at bay.

In our highly polarized environment, this understanding will not happen. Debated on the Senate floor, Conservatives would scream “Socialism” and Progressives would scream about corporate handouts while human jobs are eliminated. Human barriers will prevent this solution from ever being implemented.

Instead, the electorate will continue to rip the cockpit controls back and forth between two fundamentally flawed economic ideologies every election cycle, creating a volatile Political Seesaw. The Conservative approach builds a productivity engine but lacks a mechanism to distribute the yields, draining the wage tax base. The Progressive approach designs a distribution system but suffocates the underlying productivity needed to expand the economy.

Washington is politically paralyzed, which means doing nothing becomes the default choice. This gridlock safely locks the nation into an unshielded, laissez-faire gamble while the 6-year debt maturity cycle runs relentlessly in the background like a mechanical clock.

When the legacy debt buffers officially expire at the turn of the decade, the interest bill will consume the U.S. Treasury’s control authority. Left with no options, the Federal Reserve will be forced to pull its final, crude lever: the Argentina-style Escape Hatch. They will print money to buy up the massive debt issuance and artificially cap interest rates to save the Treasury from a deflationary default.

The macroeconomic simulations show that if the Fed allows interest rates to float up to combat sticky inflation, the baseline structural breakdown could arrive as early as 2032. If they activate the Escape Hatch to cap rates without major structural policy changes, the excess money printing breaks out into a compounding sustained inflation wildfire.

If this occurs, your standard of living will be methodically liquidated to burn away the national debt. A $100,000 savings account could see its actual purchasing power slashed to $50,000 or less in a matter of years. In an upside-down economy like this, traditional “buy-and-hold” index funds face severe headwinds, and bonds become the source of the shock rather than the safety net.

The Household Ledger: Bringing $39 Trillion Down to Earth

The primary reason this debt crisis feels so invisible to the public is a problem of scale. The human brain is simply not wired to intuitively wrap itself around a number as abstract as “$39 Trillion.” And even worse, this debt amount is increasing by an additional trillion every 90-100 days.

To truly understand how deep the trap is, we have to drop the astronomical numbers and look at the national balance sheet as if it were a single, typical American household ledger.

If we divide the current $39 Trillion national debt across the roughly 135 million households in the United States, the math yields a stark reality:

• The Individual Debt Burden: Every single American household’s individual share of the federal liability sits at approximately $288,000.

• The Income Disconnect: That $288,000 debt share is more than 4.5 times the current median annual household income ($63,795).

• The Savings Gap: A household’s portion of the national debt is over 36 times the median household’s entire liquid savings buffer ($8,000).

When you look at the numbers at this scale, the entire political debate dissolves. The typical American family does not have the balance sheet capacity to absorb a $288,000 legacy liability. The federal government cannot tax its way out of this because the median consumer is already maxed out.

Unless a structural solution successfully forces machine-driven productivity to outrun this number, the system will eventually be forced down the default path. The debt will be settled through the unshielded Escape Hatch, silently liquidating the purchasing power of that $8,000 savings account to burn away a $288,000 debt share for each household.

Next: The Personal Escape Hatch

In Part 9: The Personal Escape Hatch, I open up the financial engineering toolkit to answer the remaining question that matters: What can you actually do about it?

In the free section of Part 9, I will lay out the foundational asset-allocation philosophy required to navigate a high-inflation, unshielded economic regime.

Then, exclusively for my paid-tier subscribers, I will hand you a straightforward, step-by-step execution playbook. You do not need to run these technical simulation models yourself. I have done the heavy lifting to provide you with the compiled, actionable checklist you need to protect your private balance sheet, including:

• The Real-Time Macro Telemetry Framework: The exact economic signs, debt ratios, and warning signals to watch to see if the 2032 timeline is on track or delaying.

• Strategic Career Insulation Concepts: Methods to protect your personal income and shield your professional footprint from AI displacement and corporate layoffs.

• Corporate Screening Metrics: How to identify the “winner-take-all” equities positioned to capture automation margins.

• Defensive Hard Asset Categories: The tangible, physical asset types required to store private wealth outside a depreciating bank ledger.

• Fixed-Income & Cash Positioning: How to safely structure your liquid capital to prevent it from being cannibalized by rate caps and currency debasement.

• Optimal Tax-Sheltering Strategies: The legal frameworks necessary to shield your returns from future emergency revenue grabs.

The math of the macroeconomy is unyielding, but your personal financial outcome is not predetermined.

© 2026 Byron Birkedahl. All rights reserved. No part of this publication may be reproduced, distributed, or transmitted in any form or by any means without prior written permission.

For those who are DIY investors, or even those that have financial advisors, it’s a good idea to understand where you personally fit on this spectrum. That is the topic of this article.

To help with this, I created the investment Behavior Assessor v1.0 so you can run your own data on the Google Colab Servers (see link at the end of the article). The assessor asks some very basic questions and outputs a mix ratio of stocks (i.e. broad market) versus short term bonds that matches up in characteristics with the answers provided.

Note in the future I will be publishing an article on various asset type mixes for diversification. The math for this is much more complex than stocks vs ST bonds, so the article and tool will be more advanced.

How the Behavior Assessor v1.0 Works

There are many types of investing risk tolerance quizzes online. Unlike the quizzes you find on these online sites, this tool isn't a stealthy lead-generator designed to sell you an actively managed mutual fund or an AUM contract. It's just a transparent calculator to help you look at the math behind your choices.

As previously mentioned, this tool outputs a mix ratio of broad-based stocks, such as an S&P500 Index Fund, to a Short Term Bond fund - based on your answers to questions and the current inflation and short term bond rates. The reasons these two asset types are used are:

• Core Portfolio Pillars: These are two of the most common, foundational asset classes used in personal portfolio design.

• Stable Baseline Statistics: Broad-based stock index funds have a long track record of reliable historical data. Note: v1.0 assumes a slightly conservative 9% average nominal return to build in a margin of safety for future market conditions.

• Low Volatility & Low Correlation: Short-term bond funds feature minimal price volatility and tend to move independently of stocks, making them an excellent tool for protecting principal. (see note)

A Note on Annuities: Several annuity types can be substituted for the short-term bond portion of this assessment because they share similar characteristics, including low volatility and contract-driven drawdown protection. Based on how insurance companies structurally engineer these contracts, they will typically yield roughly 1.0% to 1.5% less than current high-quality corporate bond rates to account for the insurance company's administrative spread.

Here are the input features & effects of v1.0:

• Your Personal Drawdown Tolerance: While the math says volatility is the key factor in risk, volatility goes both directions. No one objects to volatility in the upward direction, but large downward movements, including corrections and crashes, are where an individual really needs to know their tolerance before they panic.

• Inflation and Short Term Bond Rates: These are used to assess the real (inflation adjusted) returns of the asset types. Big discrepancies will alter the suggested mix.

• Investor Experience: This is used to provide a factor to scale the output to be a more conservative mix if the user has little or no experience with actually experiencing big corrections and crashes with their own money. The reason is that when the market is doing great it’s easy to declare being a risk-taker, but that can easily change to panic mode in a crash. More experienced individuals are assumed to know their tolerance better.

• Bear Market Length Tolerance: This is also used to scale the output. Many can tolerate a short bear market correction, but a very long (or even indefinite) bear market can cause individuals to lose confidence and sell at the worst possible time.

• Portfolio Dependency/ Spending Flexibility: These are used to scale the output because individuals that are highly dependent on their portfolios and require a set amount for a fixed income per month, especially without other essential income sources, would not be able to take large market sequence-of-returns risks. Typically these individuals would be retired and in decumulation mode.

• Anxiety over Lack of Portfolio Performance Versus Sharp Market Drop: This is used to determine whether the user is more anxious about a sudden, sharp market crash or the slow, hidden erosion of wealth by inflation. If the biggest fear is inflation, v1.0 steps in to ensure you always maintain a baseline of stock exposure so your portfolio doesn't quietly lose its purchasing power over time.

Using the Behavior Assessor v1.0

Below is a screen shot of the v1.0 tool user interface. To run this tool on the Google Colab server, copy the link below into your browser which will bring up the user interface. Then simply hit the “Play Button”, adjust the sliders, and select your answers to the 5 questions using the descriptions from the previous section. Below is an example for an individual with a “medium level” of risk tolerance.

Where to find the v1.0 Tool

You can find the tool here:

https://colab.research.google.com/drive/1BTHcP08gpFNSY1B0GfFFbHlWHUjmzRUo?usp=sharing

Disclaimer

This article and the accompanying model are for educational and illustrative purposes only. They do not constitute specific tax, legal, or investment advice. The model relies on historical data, and past performance is not indicative of future results.

While the model software has been tested for accuracy, it is provided on an “as is” basis without warranties of any kind. There is no guarantee that the model is free from errors or that it will accurately predict your specific financial future. Do not make financial decisions based solely on this tool; always consult a qualified professional (CPA, Attorney, or Registered Investment Advisor) before making major life decisions.

The v4.4 Pension-Annuity Simulator is a model constructed with Python code, and available for paid subscribers to run their own data. It runs on the Google Colab Servers. This simulator runs a comparison of either a pension or annuity versus the associated lump sum. It does this comparison by plotting a time history of total wealth in retirement. For the lump sum, the total wealth is the sum of after-tax withdrawals plus the lump sum account value. For the pension or annuity, the total wealth is the after-tax sum of all payments. The simulator forces the lump sum withdrawals and pension/annuity payments to be identical so that the results are an apples-to-apples comparison.

The model accounts for both estimated taxes (see note) and the volatility of stocks for the lump sum account using historical data. The bond portion of the account is assumed to have zero volatility - which is a reasonable approximation for short term bonds within the account (e.g. T Bills).

Note - This simulator only accounts for federal taxes. State taxes are not addressed and will vary depending on location.

Below is an image of the simulator’s user interface.

For the above inputs the results are shown below. The green lines represent how a “lump sum account” with the settings above would evolve in total wealth in real dollars (after-tax withdrawals + account balance). The thick green line is the median and the thin green lines are the individual 28-year simulation paths based on historical data going back to 1934.

The blue line shows how the accumulated wealth (sum of all after-tax payments) of the pension or annuity would evolve.

This chart can be interpreted in regards to where the green and blue lines intersect. This indicates the break-even (BE) age. If the BE is significantly greater than life expectancy, then the lump sum strategy is deemed the winner. However, note that the BE of the median lump sum wealth is significantly greater than the worst cases - so the winning strategy depends on your philosophy of accepting the average (median) case or only the worst cases.

Using the v4.4 Pension-Annuity Simulator

To run the simulation simply set the sliders and enter your data and then click on the “play” button. Below is a short definition of the sliders and data:

Comparison Mode: Either select Employer Pension or Simple Annuity. In Pension mode the Initial Tax Rate Lump slider will affect the lump sum starting amount by this rate. After this, the pension withdrawals will be fully taxed at the Ordinary Income Tax rate. In Annuity mode, it is assumed the annuity is purchased by after-tax funds and the tax exclusion ratio is applied.

Note: if the annuity is purchased by cashing out a 401k or IRA, then the Monthly Income Gross should be reduced to match the reduced amount used to purchase the annuity due to the tax hit from cashing out. If the annuity is purchased within an IRA, then use pension mode because payments will be taxed identically as a pension.

Quick Guide:

If you are modeling: Set Mode: Set Initial Tax Slider to:

Pension (Rolled to IRA) Employer Pension 0%

Pension (Taken as Check) Employer Pension Your Tax Rate (e.g., 35%)

Annuity (inside an IRA) Employer Pension 0%

Annuity (After-tax cash*) Simple Annuity 0%

*If the cash is coming from cashing out a tax-deferred account make sure to lower the annuity payout to reflect the reduced cash amount for the annuity purchase to reflect this tax hit.

Lump Sum Gross: The lump sum associated with the pension or annuity.

Monthly Income Gross: The gross (before taxes) monthly withdrawal or payment associated with the pension or annuity.

Start Age: The starting age for the simulation. Typically the start of retirement.

End Age: The ending age of the simulation. Typically set for assumed end of life.

Initial Tax Rate Lump: The upfront tax hit rate for receiving a lump sum (pension mode only).

Ordinary Income Tax: The ordinary tax rate assumed during retirement.

Capital Gains Tax: The capital gains tax rate assumed during retirement.

Stock Allocation Percent: The percent the lump sum account has in stocks.

Bond Yield Percent: The assumed yearly yield of the bonds in the lump sum account.

Inflation Rate Percent: The assumed yearly rate of inflation.

Data Start Year: The earliest year of data used for the stock return simulations.

Where to Get the Simulation

The simulator is available to paid subscribers below:

Read more

Or you may want to make your own effective pension by purchasing a simple annuity using a lump sum of money.

On the surface, it looks like a simple choice between a pile of cash and a monthly check, but once you start modeling the variables, it becomes a complex battle against Sequence of Returns Risk, Inflation Drag, Tax Considerations and the “Silent Alpha” of Mortality Credits.

In this article, we’re going to strip away the marketing fluff and look at the raw mechanics. Using my v4.4 Pension-Annuity Total Wealth Accrued Simulator, we will analyze the wealth performance differences between investing a lump sum versus receiving monthly payments from a pension or annuity.

Whether you are looking at a corporate pension or considering a simple (SPIA) annuity, the question is the same: Can the associated lump sum, invested and set up for withdrawals similar to the pension/annuity, outrun the insurance company’s or pension provider’s mortality pool? Let’s look at the data.

Company Pension versus Simple Annuity - What’s the Difference?

A pension and a simple (SPIA) annuity are similar animals. They both solve for longevity risk by providing a lifetime income stream. They are basically the inverse of a traditional life insurance policy and provide protection for “living too long” to guarantee your funds will not be depleted from fixed withdrawals before you die.

Pensions are employer-sponsored benefits where the company has managed the capital for you over your career. In many cases, the employer will offer a lump sum option instead of lifetime monthly payments.

Simple annuities are offered by insurance companies so an individual can “buy their own pension” with a lump sum.

Note - There are many types of annuities available from insurance companies with many different flavors. This article will only analyze the simple annuity where a lump sum buys a lifetime fixed income stream - because it is similar to the common pension.

Tax Considerations

Typically, the key difference between a pension and annuity is how these are taxed.

Pension funds are created by companies not only to benefit employees, but also for their own benefit. They use pre-tax dollars to fund the pensions, and these can make the company’s balance sheet look better to investors. But as such, this also has implications for an employee receiving pension benefits:

• Since the pension is based on pre-tax funds, pension income to the employee is 100% taxed at the ordinary income tax rate.

• If the employee selects a lump sum option as a check (without doing a direct rollover), the sum is 100% taxed at the ordinary income tax rate. This can be a large amount of tax to pay upfront. For example, if the lump sum is $500k, the tax on this might be 35% resulting in receiving a check for only $325K.

• There may may an option for the employee to do a direct rollover of the lump sum into an IRA. In this case there is no initial tax event. If allowed, there are significant advantages in doing this which we will describe later.

In contrast, when purchasing an annuity, if the funds come from an after-tax account (e.g. taxable brokerage account, savings, etc.), then the IRS allows an “Exclusion Ratio” such that only the portion of the annuity income that is deemed to have gained value beyond the original after-tax lump sum is taxed (as ordinary income).

But a purchased annuity could also be funded by an IRA or 401k. In this case, pulling the money from these accounts results in a significant tax event.

The other possibility is purchasing an annuity within an IRA (a Qualified Annuity). In this case there is no upfront tax event from the purchase, but the income received from the annuity is taxed just like the pension - 100% taxed as ordinary income.

In both pensions and simple annuities, the income stream is typically a fixed amount per month - thus inflation erodes the amount over time.

There is also risk exposure from the solvency of the provider. In the case of the pension, your income depends on the long-term solvency of your former employer. In the case of an annuity, your income depends on the insurance carrier’s claims-paying ability.

The Insurance Company’s Basic Math

The insurance company (as well as pension provider) performs some basic actuarial math to determine the payout rate for an annuity or pension. For example, a $500k annuity may pay out a gross amount of $3k per month ($36K per year). That is a 7.2% (initial) payout. You may wonder how this high level of payout is possible when the “4% rule” would dictate safe withdrawals of only about $1700 per month. This almost seems too good to be true. Perhaps the annuity salesperson made a big point about this. Well not so fast.

Below is a chart from the v4.4 tool showing a very simple example case. This chart compares total wealth of a lump sum (account value + sum of payouts) versus the sum of annuity payouts. This case is where the lump sum is invested by the insurance company at 4.5% (green line) versus annuity payments starting at age 62 ($500k lump sum @ $3k payments). Inflation is 3% in this example.

Notice the green line intersects the blue line at age 83 - which is about the average life expectancy. This is the break-even point for the insurance company for their 4.5% investing. If you live more than 83 years, they pay out more than the average. If you live less than 83 years, they pay out less than the average.

Here’s how the insurance company makes their money:

• Every person that lives more than 83 is funded by every person that lives less than 83 - since simple annuity payments stop at death. These are the insurance company’s Mortality Credits.

• They will use high-grade corporate bonds, private equity, and other investment schemes to get (for example) a 5.5% guaranteed return from your lump sum - even though the break-even (for them) is calculated at 4.5%. Thus, they average a 1% margin in this example.

• Payouts are not adjusted for inflation (in the simple annuity).

In a nutshell, the insurance company makes their money by selling their product to 1000’s of clients with a known average lifespan for the group, combined with their ability to guarantee their investment income from the clients’ lump sum payments.

But How to Make the Decision Between Lump Sum and Fixed Income?

In trying to leverage your own advantage for investing in a Lump Sum instead of taking fixed payouts from either a pension or annuity, you will not be able to compete 1-for-1 with an insurance company or pension fund company. You obviously cannot leverage the laws of life expectancy averages - because it’s just you. In addition, you probably cannot easily beat the insurance company at generating their 5.5% guaranteed investment return.

However, if you are willing to take some market risk for a portfolio with a percentage of stocks, combined with the fact that long term stock sales are taxed at the much lower capital gains rate, the lump sum strategy can easily be advantageous.

This is the advantage an individual has over the insurance companies and pension providers - they cannot invest in this way. Per regulatory rules they must hold sufficient capital to operate and must show they are not risking this capital in a manner to potentially default - thus the reason for their guaranteed 5.5% investment engine.

In the case of a pension where you take a big tax hit upfront, the math is not that favorable for the lump sum. Below is an example of the $500k lump sum taxed upfront at 35% and invested with 60% stocks (3% inflation/ 4% bond rate). The income stream per month is identical (using the v4.4 simulator) between the lump sum & pension/annuity to get an apples-to-apples comparison.

For this example, the lump sum investing with a 60/40 portfolio, on average, does a good job of the break-even at age 89 - which looks like a winner. However, notice that in 50% of the cases, the lump sum account depletes before age 89. The worst case is age 76 - meaning that this was the worst 14-year market stretch (based on data back to 1934). Depending on your view, some may say the lump sum is still a good deal because it wins on average. Others may not like the risk of unlucky market timing.

Now take a look at the same case, except no upfront tax hit for the lump sum below. This applies to cases where you keep the funds inside a tax-sheltered environment while transitioning to an income stream such as: 1) Qualified Annuity held inside an IRA, or a 2) Rollover of company lump sum into an IRA - or you have another method that avoids the upfront tax hit.

This case shows the investment of the lump sum clearly wins over the fixed payments from an annuity or pension. Also, note the blue line (annuity/ pension) goes to zero at age 90 - because this simulation assumes 90 is the end-of-life age where payments end. This is another big disadvantage of the simple annuity/ pension option - no inheritance for heirs.

A simulation run (not shown) of the after-tax funded annuity also shows the 60/40 investing clearly wins - but not by quite as much margin since the after-tax annuity has some tax exclusions.

As you’ve seen, the way this article’s analysis judges the “winner” between an annuity or pension versus investing the lump sum is by determining the break-even (BE) age (i.e. where the green and blue chart lines intersect). If the BE age is significantly higher than the average lifespan, then the lump sum investing is deemed the winner.

There are a few caveats, however. With lump sum investments that include volatile investments (e.g. stocks), there may be “unlucky market paths” where the BE age is much lower than the median BE. In these cases, it must be judged by the individual if they are more interested in averages or in worst cases.

Also, in cases where the annuity or pension has been set up for a married couple, it must be considered that the BE age of the couple (i.e. median longevity of last survivor) is higher than a single person. For example, if the median single longevity age is 83, it is statistically about 90 for the last survivor of a couple who are about the same age.

Of course, the pension or annuity payout is less for a married couple than for a single person. For example, if we run the first simulation above with a $2600/mo. payout at 4.5%, we get a BE age of 90 (instead of 83 @ $3000/mo), and this is essentially how the insurance company or pension provider determines the lower payout for married couples.

A few Conclusions

After running various scenarios in the v4.4 simulator, there are a few general conclusions we can draw:

• A large tax rate hit of the lump sum, whether it’s on a pension lump sum check or from cashing out money from a 401k / IRA to buy an annuity, can seriously drag down the total wealth performance. In the former case, it tends to favor the pension over the lump sum. In the latter case it tends to favor not purchasing the annuity.

• With a minimal tax hit of the lump sum, a strategy of investing - even with a modest stock mix portfolio, can easily beat the performance of a pension or annuity.

• If you believe you will significantly beat the odds for longevity (e.g. family history of long life), especially if you are very conservative in investing and concerned about “running out of money”, the pension or annuity can make more sense.

• If you are “freaked out” by market corrections, you will probably “sleep better” with the annuity or pension.

• Lump sum investing in a taxable account has advantages over a pension or annuity since the equity portion of the lump sum account will be taxed at the capital gains rates for withdrawals - which are typically lower than ordinary income tax rates.

• If you want liquidity and flexibility for your own investing to receive an income stream, and can stay the course through market corrections, the lump sum is much more favorable than the pension or simple annuity.

• If you want your heirs to inherit any leftover funds, the lump sum is the way to go over the pension and simple annuity.

Other Considerations

There can be some considerations for your decision on the simple annuity or pension versus investing the lump sum that can’t be quantified.

In the case of a simple annuity, it is difficult for many to cut such a large check upfront. If you have $1M in an account and buy a $500k annuity, you get to see your account balance instantly go down by 50%. Even though you know that you have lifetime benefits, this is difficult for many. There is a lack of liquidity. Once you sign up for it there is no way to get any of it back (e.g. emergency). Plus, you know that if you die tomorrow, the annuity will expire worthless, and there will be nothing left for heirs from it.

These last points are why simple (SPIA) annuities are not all that popular. But of course, the insurance industry has come through with many flavors of annuities to address every concern individuals have. There are annuities that:

• Defer income

• Provide death benefits to heirs

• Provide a cash-out mechanism for emergencies

• Adjust payments with cost-of-living adjustments

• Provide credits based on an index (e.g. S&P500)

• Provide a “mutual fund” type of growth potential (but with risk)

• Provide market participation and shields for losses (note see my article “Engineering the RILA”

All of these are financially engineered annuity products that insurance companies have are constructed by using only a few basic ingredients:

• A bond portfolio (their growth engine)

• Mortality credits (based on longevity statistics for a large group of people)

• Options & derivatives (to shape the outcomes, but don’t actually provide wealth)

Naturally, there is no free lunch from the insurance company for any of these products. Their products are built for them to profit, and they employ individuals with a great deal of math-savvy to engineer them. This is not to say these are “bad”, but individuals who purchase them need to be aware of this - especially when the sales pitch seems almost too good to be true.

In the case of a company pension there may be less psychological barriers. Unlike the annuities, a pension feels more like “free money” that you didn’t really have beforehand. In this case (if the option is given by the company) it clearly is a choice between fixed monthly payments versus receiving a lump sum. A simulation-based analysis given your specific data can help you with the decision.

About the v4.4 Pension-Annuity Simulator

The v4.4 Pension-Annuity Simulator is a model constructed with Python code, and available for paid subscribers to run their own data. It runs on the Google Colab Servers. This simulator runs a comparison of either a pension or annuity versus the associated lump sum. It does this comparison by plotting a time history of total wealth in retirement. For the lump sum, the total wealth is the sum of after-tax withdrawals plus the lump sum account value. For the pension or annuity, the total wealth is the after-tax sum of all payments. The simulator forces the lump sum withdrawals and pension/annuity payments to be identical so that the results are an apples-to-apples comparison.

The model accounts for both estimated taxes (see note) and the volatility of stocks for the lump sum account using historical data. The bond portion of the account is assumed to have zero volatility - which is a reasonable approximation for rolling over short-term bonds within the account (e.g. T Bill ladder).

Note - This simulator only accounts for federal taxes. State taxes are not addressed and will vary depending on location.

Below is an image of the simulator’s user interface.

Disclaimer

This article and the accompanying model are for educational and illustrative purposes only. They do not constitute specific tax, legal, or investment advice. The model relies on historical data, and past performance is not indicative of future results.

While the model software has been tested for accuracy, it is provided on an “as is” basis without warranties of any kind. There is no guarantee that the model is free from errors or that it will accurately predict your specific financial future. Do not make financial decisions based solely on this tool; always consult a qualified professional (CPA, Attorney, or Registered Investment Advisor) before making major life decisions.

This includes the latest versions of the tools I’ve been using for the “Engineer Investor” articles, along with an upgrade based on reader feedback.

Here’s what’s now available:

Retirement Simulator (v6.1) – The model used in the article, The Case for an Optimal Withdrawal Strategy

Retirement Simulator (v7.1) – The model used in the article, What If the Market Crashes When I Retire?

Upgrade: v6.2 – Phased Retirement Spending – This version adds the ability to model multiple spending stages in retirement, rather than assuming a single constant spend level.

One limitation in earlier versions was the assumption of a single spending level after retirement. In reality, spending and income tend to shift over time - early retirement often looks different from post-Medicare years, and again after Social Security begins.

The updated v6.2 model allows you to represent this more directly using simple phased inputs (for example, different spending levels at ages 62, 65, and 70). In many cases, this produces meaningfully more realistic results.

Below is an example of the v6.2 interface and output for a phased retirement scenario.

I’ll continue to keep the core analysis and results in these articles fully available.

For those who want to run the scenarios themselves and explore different assumptions, the full simulator access - along with the updated versions and future improvements -is available to paid subscribers below.

Disclaimer: This content is for informational purposes only and does not constitute financial advice.

Read more

Fortunately, this debate can be solved by math and is perfect for an analysis.

The Social Security Earnings Test

Be aware, if you are in the 62 camp you may not actually be able to take or fully take social security benefits until your Full Retirement Age (FRA). The reason is that there is an income limit rule which is as follows:

If you claim Social Security benefits before your FRA (i.e. 67) and continue to work, your benefits are subject to a temporary “withholding” based on your earned income.

• The Limit: For 2026, the annual earnings limit is $23,400.

• The Penalty: For every $2 you earn above this limit, Social Security withholds $1 in benefits.

• The Exception: In the year you reach your FRA, the limit increases significantly (to $62,160), and the withholding drops to $1 for every $3 earned.

• The “Sunset”: Once you reach your Full Retirement Age, the earnings test disappears entirely. There is no limit on how much you can earn, and the SSA “recalculates” your benefit to gradually pay back the amounts withheld during the early years.

Note: This rule only applies to earned income (W-2 wages or net self-employment). It does not apply to “passive” income like pension payments, IRA withdrawals, dividends, or capital gains.

Benefit Increase by Age

As previously mentioned, the base amount of social security @ age 62 increases every year it is delayed. The common advice is that the benefit increases 8% per year from 62 to 70. This is not exactly correct, but close. It is a simple rate increase (not compounding) with one rate before FRA and a slightly different one after FRA. See below.

You may ask how the SS administration came up with this benefit increase schedule. In a nutshell, they have set this schedule to end up where on average the total SS benefits they pay out are the same - no matter what age you take it - for an average lifespan of ~ 82 years. From the U.S. government standpoint, if you live less than 82 years then they pay out more for early takers, and if you live more than 82, they pay out more for late takers. But the payments balance out for them due to average life expectancy.

Social Security Simulation

I created a Social Security Simulation to help visualize the aspects of what total lifetime benefit you can expect. It should be noted this simulation does not consider taxes and is only based on the raw non-taxed amounts received. Tax considerations can change the answer, and I’ll address this point later.

At first you might assume ss@62 is better only if you live less than 82 years. This is only true if you don’t consider spending and investing.

Below is the output of the simulation comparing taking it at 62 vs 70 when: 1) Base monthly benefit @ 62 is $2500, 2) Monthly spend starts at 62 and is $2000 3) Your investment accounts are 60% in stocks and 4) Bond yields and inflation are 4% and 3%.

Note: the simulation uses the stock % to calculate an expected average real return based on the historical performance of stocks, the selected bond rate, and inflation. While this model uses deterministic average returns, it serves to establish the 'Hurdle Rate' that the Social Security delay credits must clear to be competitive with a compounding portfolio.

As you can see, for this case the early ss wins up to age 92.5. This break-even age may be counterintuitive to you, but here’s why. Notice that between 62 and 70 the red line (ss@70) goes negative. This is because in order to make an apples-to-apples comparison, we need to start the spending at the same age in both scenarios. When taking ss@70, between 62 and 70 that $2k/mo spend has to come from somewhere. This simulation assumes you have to pull the money from another account which has the stock % mix selected (i.e. 60%). This results in a negative value “hole” that the ss@70 must first climb out of when the benefits start @ 70. Meanwhile, the ss@62 (blue line) starts rising at 62 because it assumes social security not spent goes into an account with the stock % input value (i.e. 60%).

When you run the simulations one thing interesting is that with a fixed stock % (i.e. 60%) the break-even age is constant at 92.5 no matter what the spend level is. For this case early social security is clearly the winner because this break-even age is more than 10 years beyond the average life expectancy.

Below is the same case except we change the portfolio to an aggressive 80% portfolio of stocks generating a higher level of an expected average real return.

As you can see in this case the ss@70 never catches up by age 95. Therefore, the ss@62 is the clear winner. When you run the simulations you will see that the break-even age moves with your account’s performance.

It’s only when your account performance is no better than a CD (e.g. stocks = 0%, bond & infl = 3%), that the break-even ends up back to age 82.

And further, if your accounts are all cash making 0% and inflation is 3%, the break-even becomes even lower to age 78 as seen below - which heavily favors ss@70.

Married Couples: The Survivor Twist

The rules for married couples add a critical variable to the single-person model. While both partners are alive, calculations apply independently. However, upon the death of one spouse, the survivor automatically transitions to the higher of the two benefits.

As established, the winning strategy depends on whether the Break-Even Age (e.g., 92.5 years for a 60% stock portfolio) occurs before or after the expected end-of-life. If a couple is similar in age, the single-person logic holds.

The calculation flips when there is a significant age difference and the older spouse was the primary earner. Consider a scenario where the higher earner is 20 years older than their spouse. In this case, the “Couple Life Expectancy” for that higher check is effectively 102 years (82 + 20). Since 102 is well beyond the 92.5-year break-even, the Age 70 strategy becomes the winner for the older spouse. Conversely, the younger spouse should claim at 62 to maximize early compounding, knowing they will eventually “step up” to the higher survivor benefit.

Using the simulation - you can estimate at what age of the primary earner where it makes sense for this person to take social security based on the break-even age compared to the “Couple Life Expectancy” using the primary earner’s information.

Tax Considerations

As previously mentioned, the simulation does not take into account any tax considerations. In many cases taxes will not change the outcome of which age is best for taking social security.

However, there are situations where tax considerations can change the outcome. For example, if your plan is to actively optimize wealth using conversions of an IRA or 401k to Roth, you typically will try to minimize your income in the early retirement years. As described in my article Roth Conversions are Overrated, if your tax bracket when doing conversions is lower than your tax bracket when cashing out, then doing the Roth conversions can be beneficial.

For example, I ran one persona in a Roth simulation of a person who was actively attempting to gain a max lifetime wealth benefit for doing the conversions by minimizing early retirement income to maximize the conversions. With ss@70 the total wealth at age 90 (incl. cashing out by heirs) was +6.8% better than not doing the Roth. But with ss@62 the total wealth was slightly reduced to +6.0%. This delta .8% is not a great deal, but does show that a strategy for maximizing Roth conversion benefits does favor delaying social security to 70.

Conclusion

From the analysis and simulations we can conclude the following:

• If you are using accounts to fund your spending that are moderate to aggressive in nominal portfolio performance, taking social security earlier is the winner.

• If you are using accounts to fund your spending that are very conservative, then it’s a toss-up.

• If you are using accounts to fund your spending that are mostly cash, then delaying social security wins.

• If you are a married couple where the primary earner is older, the decision should be based on the calculated break-even age compared to the “Couple Life Expectancy” using the primary earner’s information.

• If you are actively trying to gain max wealth benefits by a tax arbitrage strategy such as IRA/401k to Roth conversions, then delaying social security is the likely winner.

This analysis demonstrates that there is no 'universal' best age; rather, the “optimal” claiming strategy - where “optimal” is defined as maximizing total lifetime wealth on average - is a derivative of your asset allocation, your tax strategy, and your household's combined longevity in the case of married couples.

About the Simulation Model

The simulation can be found here on Google Colab:

https://colab.research.google.com/drive/1QQWUno-MgGphTMfRiwAf6TIa1u6QNvFi?usp=sharing

To run it, simply hit the “Play button”. You can then play with the sliders to view the various outcomes.

Disclaimer

This article and the accompanying model are for educational and illustrative purposes only. They do not constitute specific tax, legal, or investment advice. The model relies on historical data, and past performance is not indicative of future results.

While the model software has been tested for accuracy, it is provided on an “as is” basis without warranties of any kind. There is no guarantee that the model is free from errors or that it will accurately predict your specific financial future. Do not make financial decisions based solely on this tool; always consult a qualified professional (CPA, Attorney, or Registered Investment Advisor) before making major life decisions.

Here is the article:

HyMasa Financial Analytics

Supercritical (Part 1): The Illusion of Wealth

Welcome to Supercritical, a multi-part macro-research series by HyMasa Financial Analytics. Over the series, we will be stripping away the political noise to look at the raw physics of the U.S. economy. We will explore the mathematical collision of the $39 Trillion debt trap, the AI productivity shock, the coming reorganization of American wealth, the r…

Read more

5 months ago · Byron Birkedahl

I am using the term “Roth conversions” in this article to describe the common practice of converting the tax-deferred assets of a traditional IRA or 401k into a Roth IRA, which results in tax-free withdrawals under current tax law. This means that taxes are paid when converting - thus turning a tax-deferred IRA or 401k account into a non-taxed Roth account.

What I do not address in this article is the far less common case of converting the after-tax basis of an IRA or 401k into a Roth IRA. This conversion is not taxed and results in tax free growth for the future. It is much less common, however, because it requires a plan such as a company 401k to support it. In fact, I have done this “no tax” conversion for my own assets because I found there was only an upside for it.

The Roth Hype

Every financial dinner I have attended has pushed the Roth conversion strategy. When I look at Yahoo news, every day I see numerous versions of the ad hype such as, “I have $1.2M - can I convert $200K/ year into Roth?” or “How can I avoid the RMD Tax Bomb!” Of course, these ads will steer you into setting up a meeting with a financial advisor.

In one financial dinner, the presenter first showed a picture of a farmer buying seeds and then a second picture of the harvest. Then the statement was made, “Would you rather pay the tax on the harvest or on the seeds?” The implied message is that it’s better to pay that small “seed tax” upfront than the huge “harvest tax” later. At another financial dinner the presenter stated, “We all know taxes are going up, so Roth conversions are always a good idea (and we can help)”. The most common theme I’ve observed from all the Pro-Roth hype has been a focus on the tax difference. The next section will describe why taxes alone are the wrong aspect of the Roth conversion to focus on, and why Financial Advisors try to use this as the key argument to make their case.

The Basic Roth Conversion Math

Some of the detailed financial tools available by subscription have the ability for you to calculate what advantages you may (or may not) gain by doing Roth conversions. There are numerous parameters that get into the calculations such as: income/ pensions/ annuities, spending, portfolio balances & performance, Required Minimum Distributions (RMDs), IRMAA, inflation, retirement age(s), tax rates/ tables, social security, ACA, and more. If fact, there are so many parameters it is easy to get lost in all the variables and lose sight of the basics.

At its most basic level, the tradeoff between the doing the Roth conversion or not simply boils down to the effective tax rate of doing the upfront conversion for Roth versus the effective tax rates of taking distributions for non-Roth.

The simple chart below demonstrates an example for the Roth and non-Roth basics of the tradeoff math when the tax rates are the same (24% upfront vs 24% back end) and the investment strategies are the same. As you can see, the Roth takes the hit upfront and the Traditional takes the hit in the back end - but both paths end up the same. While this chart is highly simplified, it is common to get this result in practice even with all the parameters included in the detailed calculations.

One key aspect to understand are shown in the bar charts below. See how the total taxes paid for the Traditional portfolio are 2.5 x more than the Roth portfolio. All this means is that the back end taxes simply scaled with the total value increase of the traditional portfolio. So, the question is: do you really care about the tax differences in this case? - your total benefit ended up in the exact same place!

Unfortunately, the Roth hype focuses on the taxes, even though the tax differences in this 24%/24% case are simply a consequence of the portfolio success and therefore are irrelevant. But you can probably imagine financial salesman will push the tax narrative to make their case.

To put this in context, imagine if you had offers for two different jobs where one recruiter was offering you a $75k salary and the other a $200k salary. Then the $75k recruiter says, “But our job is better because you pay less taxes!”

Other Simplified Examples

Below are diagrams showing the simple results for 24%/ 32%, and 32%/24%. As you can see 24%/32% is advantageous for Roth, and 32%/24% is advantageous for Traditional. Again, the bar charts show that the tax differences don’t matter for these scenarios - what matters is the final value of where you end up.

But What if Taxes Go Up?

As previously mentioned, one of the most common arguments for Roth conversions is: “tax rates are going up.” Maybe—but that’s not the right question. What matters is your lifetime effective tax rate, not future headline brackets. In retirement, many individuals still fill lower brackets, use the standard deduction, and control their withdrawals, which often keeps their effective rate lower than expected—even if statutory rates rise. Roth conversions only make sense if your future tax burden is meaningfully higher than today’s, and in many cases, that simply isn’t true. Treating rising taxes as a certainty—and converting based on that alone—is not analysis, it’s speculation.

Realistic Example

The examples shown in the previous section are highly simplified just to demonstrate the most dominant aspect of the tradeoff. As stated before, there are numerous parameters involved in the detailed calculations.

Using a very complex Roth/ No-Roth tradeoff model I created (roth_tool_v0.5.15) that includes all the bells and whistles, we can see how this would pan out with realistic data. One specific example used the following inputs:

• Filing status: Single, Age: 62

• Income: $0

• Roth conversion bracket: 24%

• Medicare: 65, Social Security: $3K/ mo @ age 66

• Portfolio amount: $1M (trad), 60/40, Spend: 36K/yr

• Inflation: 3%

• Allow IRMAA bracket changes

• No ACA

The model then calculates:

• RMD amounts

• Conversion amounts

• MAGI, NIL, NIIT

• Federal and State Taxes

• IRMAA

• Trad, Roth and Taxable Account values

Example Results

Below are the simulation results for this example. The Roth ended up 3.3% higher than the traditional - therefore there is a small benefit in this case for doing the Roth.

One other interesting aspect of this chart is the drop down of the traditional account value at age 90. This is because the model is assuming a “cash-out” at this point for whatever is leftover in this account. In reality, it’s likely this is the point where the account would be transferred to heirs. But there are many ways the heirs could end up dealing with this account based on their tax bracket(s), and the fact they must generally distribute the account within 10 years (often with required distributions during that period). Obviously the Roth is better for the heirs since they don’t have to deal with the taxes as well as the 10-year rule.

Conversely, if your view is that you are not concerned your heirs will need to pay some tax on inheriting your account, then you may be happier with a trad account with its significantly higher statement values than a Roth account over your entire retirement.

Other Examples

The following additional examples were also simulated to see how the tradeoffs panned out.

Example #2 Persona: “Plenty of assets, no heirs, spend it down”

Age 60 - single, SS @ age 62 $2.8K/mo, $30K/yr income, $2M assets, On ACA, 90/10 portfolio w/ maximum spend down during retirement.

Results: Traditional end balance (after taxes) is +3.4% higher. (Trad Wins)

Example #3 Persona: “High assets with no income, great at investing/ saving but doesn’t spend much”

Age 63 - single, SS @ age 70, $5K/mo, $0 income, $7M assets, 60/40 portfolio with very minimal spend down during retirement.

Results: Roth end balance is +6.5% higher (Roth Wins).

Example #4 Persona: “Late to saving for retirement and in max catch-up mode, max spending possible in retirement”

Age 62 - single, SS @ age 67, $3.3 K/mo. $175K income to age 66 then $40k to age 75, $330K assets, 100/0 portfolio w/ max spend down during retirement.

Results: Traditional end balance (after taxes) is +3.1% higher (Trad Wins).

Conclusion

Aside from the Roth simulations described above, numerous additional simulations have been performed for various other types of personas. In general, the majority of simulations tended to show a slight advantage for Roth conversions - typically 1-4%, while others show little benefit or even a disadvantage. At the high end, for more extreme cases where the individual has more elaborately planned (and possibly sacrificed) for taking advantage of the “Roth arbitrage”, we can see 6-7%.

There are obviously too many variables to describe whether or not there is an advantage for a Roth conversion (and by how much) for the specific circumstances of every individual. To do this, I would encourage running a high-quality financial calculator with your own data (see note). But when doing this, as previously mentioned, focus on the differences in total returns and not the taxes paid.

Note - Be wary of the free financial calculators that don’t answer what you really need to know and are actually “funnels” for signing up with a Financial Advisor.

Comparison of Roth Conversions to Other Financial Strategies

To realistically put the Roth strategy in context, we should compare this to other aspects of investing for retirement described in my other articles. If we look at the article “The REAL Cost of a Financial Advisor”, we can run the same type of simulation persona shown in the first example above and see what happens if we incorporate an advisor with a 1% AUM fee. In this case we find this results in a -28.3% hit to the final portfolio value over the same timeframe! So, the Roth +3.3% advantage is about 1/10th the portfolio drag from the investment advisor in this case. Also, if we run scenarios described in the article “The case for an Optimal Withdrawals Strategy”, we find that even a modest change to the investment mix along with a withdrawal strategy to enable it results in a +35% increase to the portfolio over the same timeframe.

My conclusion is therefore that while optimizing by doing Roth conversions is something to look at, it is a second order effect, and not even in the same ballpark as avoiding high advisor & fund fees, optimal withdrawals, retirement timing, and maximizing your portfolio investment strategy. Because of this, along with all the financial advisor and media hype, I claim Roth Conversions are Overrated.

When I ran my own numbers, I found the Roth conversion was a 2% advantage for my retirement portfolio. I considered this not worth my trouble, and instead to concentrate on the Big-Ticket Items.

About the Roth Tool v0.5.15

The v0.5.15 Roth model was used for the detailed Roth vs. No Roth calculations. It currently includes nearly all of the possible variations individuals and married couples may have but currently does not include Qualified Charitable Donations (QCDs). It has embedded all the tax tables for 2025, and these are considered usable for the future because the tables and rates are typically updated each year by the IRS for Cost-of-Living-Adjustments.

Currently this tool is not suitable to provide to subscribers because the user interface is not user-friendly. If enough readers request it, I may update it to be user-friendly and provide it in the future to paid subscribers.

Disclaimer

This article and the accompanying model are for educational and illustrative purposes only. They do not constitute specific tax, legal, or investment advice. The model relies on historical data, and past performance is not indicative of future results.

While the model software has been tested for accuracy, it is provided on an “as is” basis without warranties of any kind. There is no guarantee that the model is free from errors or that it will accurately predict your specific financial future. Do not make financial decisions based solely on this tool; always consult a qualified professional (CPA, Attorney, or Registered Investment Advisor) before making major life decisions.

To a degree this concern makes sense, because a crash close to your retirement date is actually the worst luck you can have for such an event - assuming you have stocks in your portfolio. Retirement is where we typically transition from a portfolio contribution mode to a portfolio spending mode, and therefore a large crash right at the start results in perhaps 30 years of trying to withdraw 4% yearly. (note see my article on the 4% rule).

Below (figure 1 ) is a simulation of what could be expected for a 60%/40% stock/bond ratio where the starting value at age 50 is $275K with $2K/ month contributions, retire at age 62 and spend 4%/ year. As you can see this looks good with a (historically) zero percent chance of running out of money up to age 90.

Figure 1

In order to test the unlucky case where an individual retires right when a market crash occurs, the “When Can I Retire” simulation model was upgraded to v7.1 to test the case of a theoretical crash occurring right at retirement for all runs (see note).

Note - For runs where a crash actually did occur historically right at retirement, the model did not apply the additional theoretical crash to not “double count” the crash scenario.

As you can see from charts below (figure 2), using the same parameters as before, for a theoretical 30% crash right at retirement, the results are not-so-good. This is because of the “Sequence of Returns” risk (i.e. an unlucky sequence of negative returns). Median portfolio final value has almost dropped by 50%, but even worse is that the probability of running out of money in retirement is now a significant 14.2%. This chart sums up the concern pre-retirees have for the “crash-at-retirement” scenario.

Figure 2

The “Bond Tent” Strategy

In the early 2000’s Prudential Financial pioneered and trademarked the term “Retirement Red Zone”. Their concept was that there is a critical “red zone” both 5 years before and 5 years after retirement that are critical for any fixed % ratio portfolio.

The Bond Tent “V-shaped Glidepath” was published in 2013 by a research paper "The Portfolio Size Effect and Optimizing Retirement Glidepaths" by researchers Michael Kitces and Wade Pfau. The term “Bond Tent” was later popularized by financial bloggers.

Below (figure 3) is an example of how such a Bond Tent could be implemented for the same 60/40 portfolio. This case shows a 40% “glidepath” - meaning that the original 60% stock allocation starts ramping down 5 years before retirement (e.g. age 57 in this example), to a 20/80 ratio right at retirement, and then ramping back to 60% 5 years after retirement (e.g. age 67). Obviously the glidepath % could be different than 40%, depending on the level of “crash resilience” desired.

Figure 3

For our unlucky case where there is a 30% crash right at retirement, the 40% Bond Tent does a good job of mitigating this unlucky occurrence as seen in the charts below (figure 4) . As shown, the probability of running out of money is now zero %.

Figure 4

Downside of the Bond Tent Strategy

As usual with financial strategies, there is never a free lunch. As shown below in figure 5, these charts are the results of the 60/40 portfolio with the Bond Tent. Compare this to figure 1. The Bond Tent Total Economic Benefit is $1.896M - which is the sum of all withdrawals and the mean final portfolio value. But in figure 1, where there is no Bond Tent, the Total Economic Benefit is $2.498M. Therefore, the cost of the Bond Tent “insurance” in this case is $602K, which is a 24% hit.

Conclusions

As we saw, the Bond Tent can do a good job of mitigating the risk of an unlucky large crash at retirement. But the cost, in the example given, was a 24% tax on the portfolio performance during retirement using actual historical data.

It needs to be understood that the actual historical conditions for the simulations runs shown in figure 1 and figure 5 already include all the crashes that have occurred between 1934 and 2025. Figure 2 and figure 4 add additional theoretical 30% crashes at retirement in every run - but this is in addition to all the actual ones already in the data. It turns out the likelihood for an additional 30% crash at retirement as shown in these simulations is only .55%.

I would make the conclusion that a 24% portfolio hit to cover a case with only a .55% chance of occurring isn’t worth it, and therefore suggest it’s more optimal to forget the Bond Tent and take your chances that such an extra improbable event - beyond the historical occurrences - won’t likely happen.

But for those who are still worried about being extremely unlucky for a crash at retirement, there is another withdrawal strategy (i.e. flexible withdrawal) that will mitigate a crash right at retirement which doesn’t result in in a portfolio performance tax. This is described in my article “The Case for an Optimal Withdrawal Strategy”.

About the Model

The model used in this article is version is the “When Can I Retire” v7.1 version. This model is currently available for paid subscribers to run simulations with your own data.

Disclaimer

This article and the accompanying model are for educational and illustrative purposes only. They do not constitute specific tax, legal, or investment advice. The model relies on historical data, and past performance is not indicative of future results.

While the model software has been tested for accuracy, it is provided on an “as is” basis without warranties of any kind. There is no guarantee that the model is free from errors or that it will accurately predict your specific financial future. Do not make financial decisions based solely on this tool; always consult a qualified professional (CPA, Attorney, or Registered Investment Advisor) before making major life decisions.

Many of you subscribe to The Engineer Investor, where I write about personal investing, retirement strategy, and portfolio analysis.

I’ve recently launched a separate publication called HyMasa Financial Analytics.

This new newsletter focuses on a different question: how stable is the economic system itself?

Drawing on my background in aerospace systems engineering, the goal is to apply systems modeling and quantitative analysis to large macroeconomic claims. Instead of commentary or forecasts, each article attempts to translate a major economic assertion into a simplified structural model and examine whether it is dynamically plausible within a historically grounded framework.

Topics will include issues such as AI-driven labor disruption, fiscal sustainability, structural inflation dynamics, and asset valuation under changing macroeconomic conditions.

The first article introduces the framework for this work:

A Systems Approach to Economics

You can read it here:

HyMasa Financial Analytics

Beyond Narrative: A Systems Approach to Economics

A Systems Approach to Economics…

Read more

6 months ago · Byron Birkedahl

If this kind of analysis interests you, please subscribe!

The Engineer Investor will continue as usual, focusing on personal investing and retirement planning.

Best,
Byron