Do Tariffs Decrease Prices?

Zachary Bartsch2 Comments

Much of what economics has to say about tariffs comes from microeconomic theory. But it’s mostly sectoral in nature. Trade theory has some insights. But the effects on the whole of an economy are either small, specific to undiversified economies, or make representative agent assumptions that avoid much detail. Given that the economics profession has repeatedly said that the Trump tariffs would contribute to inflation, it seems like we should look at the historical evidence.

Lay of the Land

Economists say things like ‘competition drives prices closer to marginal cost’. Whether the competitor lives abroad is irrelevant. More foreign competition means lower prices at home. But that’s a partial equilibrium story. It’s true for a particular type of good or sector. What happens to prices in the larger economy in seemingly unrelated industries? The vanilla thinking that it depends on various elasticities.

I think that the typical economist has a fuzzy idea that the general price level will be higher relative to personal incomes in some sort of real-wages and economic growth mental model. I don’t think that they’re wrong. But that model is a long-run model. As we’ve discovered, people want to know about inflation this month and this year, not the impact on real wages over a five-year period.

Part of the answer is technical. If domestic import prices go up, then we’ll sensibly see lower quantities purchased. The magnitude depends on the availability of substitutes. But what should happen to total import spending? Rarely do we talk about the expenditure elasticity of prices. Rarely do we get a simple ‘price shock’ in a subsector. It’s unclear that total spending on imports, such as on coffee, would rise or fall – not to mention the explicit tax increase. It’s possible that consumers spend more on imports due to higher prices, or less due to newly attractive substitutes. The reason that spending matters is that it drives prices in other parts of the economy.

For example, I argued previously that tariffs reduce dollars sent abroad (regardless of domestic consumer spending inclusive of tariffs) and that fewer dollars will return as asset purchases. I further argued that uncertainty makes our assets less attractive. That puts downward pressure on our asset prices. However, assets don’t show up in the CPI.

According to the above discussion, it’s unclear whether tariffs have a supply or demand impact on the economy. The microeconomics says that it’s a supply-side shock. But the domestic spending implications are a big question mark.

What is a Tariff Shock?

That’s the title of a recent working paper from the Federal Reserve Bank of San Francisco. It’s a fun paper and I won’t review the entirety. They start by summarizing historical documents and interpreting the motivation of tariffs going back to 1870. They argue that tariffs are generally not endogenous to good or bad moments in a business cycle and they’re usually perceived as permanent. The authors create an index  to measure tariff rates.

Here’s the fun part. They run an annual VAR of unemployment, inflation, and their measure of tariffs. Unemployment in negatively correlated with output and reflects the real side of the economy. Along with inflation, we have the axes of the Aggregate Supply & Aggregate Demand model. Tariffs provide the shock – but to supply or demand?.  Below are the IRF results:

Continue reading →

Macroeconomic Policy In a Nutshell

James BaileyLeave a comment

What I’m telling my Intro Macro students on the last day of class, since we weren’t able to get through every chapter in the textbook:

A few of you might end up working in economic policy, or in highly macro-sensitive businesses like finance. For you, I recommend taking followup classes like Intermediate Macroeconomics or Money and Banking so you can understand the details. For everyone else, here are the very basics:

  1. In the long run, economic growth is what matters most. The difference between 2% and 3% real GDP growth per capita sounds small in a given year, but over your lifetime it is the difference between your country becoming 5 times better off vs 10 times better off.
  2. How to increase long-run economic growth? This is complicated and mostly not driven by traditional macroeconomic policy, but rather by having good culture, institutions, microeconomic policy, and luck.
  3. In the shorter run, you want to avoid recessions and bursts of inflation.
  4. High inflation means too many dollars chasing too few goods. To fix it, the federal government and the central bank need to stop printing so much money (the details can get very complicated here, but if we’re talking moderately high inflation like 5% the solution is probably the central bank raising interest rates, and if we’re talking very high inflation like 50% the solution is probably a big cut to government spending).
  5. If there is a recession (which will look to you like a big sudden increase in layoffs and bankruptcies), the solution is probably to reverse everything in the previous point. The government should make money ‘easier’ via the central bank lowering interest rates while the federal government spends more and taxes less.
  6. If you don’t take more economics classes, you will likely hear about macro issues mainly through the news media and social media. You should be aware of their two main biases: negativity bias and political bias.
    • Negativity Bias: If It Bleeds, It Leads on the news. Partly this is because bad news tends to happen suddenly while good news happens slowly, so it doesn’t seem like news; partly it just seems to be what people want from the news and from social media.
    • Political Bias: People tend to seek out news and social media sources that match their current preferences. These sources can be misleading in consistent ways for ideological reasons, or in varying ways based on whether the political party they like is currently in power.
  7. There are different ways to measure each key macroeconomic variable. Think through them now and make a principled decision about which ones you think are the best measures, and track those. Otherwise, your media ecosystem will cherry-pick for you whichever measures currently make the economy look either the best or the worst, depending on what their biases or incentives dictate.
  8. There are good ways to keep learning about economics outside of formal courses and textbooks, I list a few here.

Google’s TPU Chips Threaten Nvidia’s Dominance in AI Computing

Scott BuchananLeave a comment

Here is a three-year chart of stock prices for Nvidia (NVDA), Alphabet/Google (GOOG), and the generic QQQ tech stock composite:

NVDA has been spectacular. If you had $20k in NVDA three years ago, it would have turned into nearly $200k. Sweet. Meanwhile, GOOG poked along at the general pace of QQQ.  Until…around Sept 1 (yellow line), GOOG started to pull away from QQQ, and has not looked back.

And in the past two months, GOOG stock has stomped all over NVDA, as shown in the six-month chart below. The two stocks were neck and neck in early October, then GOOG has surged way ahead. In the past month, GOOG is up sharply (red arrow), while NVDA is down significantly:

What is going on? It seems that the market is buying the narrative that Google’s Tensor Processing Unit (TPU) chips are a competitive threat to Nvidia’s GPUs. Last week, we published a tutorial on the technical details here. Briefly, Google’s TPUs are hardwired to perform key AI calculations, whereas Nvidia’s GPUs are more general-purpose. For a range of AI processing, the TPUs are faster and much more energy-efficient than the GPUs.

The greater flexibility of the Nvidia GPUs, and the programming community’s familiarity with Nvidia’s CUDA programming language, still gives Nvidia a bit of an edge in the AI training phase. But much of that edge fades for the inference (application) usages for AI. For the past few years, the big AI wannabes have focused madly on model training. But there must be a shift to inference (practical implementation) soon, for AI models to actually make money.

All this is a big potential headache for Nvidia. Because of their quasi-monopoly on AI compute, they have been able to charge a huge 75% gross profit margin on their chips. Their customers are naturally not thrilled with this, and have been making some efforts to devise alternatives. But it seems like Google, thanks to a big head start in this area, and very deep pockets, has actually equaled or even beaten Nvidia at its own game.

This explains much of the recent disparity in stock movements. It should be noted, however, that for a quirky business reason, Google is unlikely in the near term to displace Nvidia as the main go-to for AI compute power. The reason is this: most AI compute power is implemented in huge data/cloud centers. And Google is one of the three main cloud vendors, along with Microsoft and Amazon, with IBM and Oracle trailing behind. So, for Google to supply Microsoft and Amazon with its chips and accompanying know-how would be to enable its competitors to compete more strongly.

Also, AI users like say OpenAI would be reluctant to commit to usage in a Google-owned facility using Google chips, since then the user would be somewhat locked in and held hostage, since it would be expensive to switch to a different data center if Google tried to raise prices. On contrast, a user can readily move to a different data center for a better deal, if all the centers are using Nvidia chips.

For the present, then, Google is using its TPU technology primarily in-house. The company has a huge suite of AI-adjacent business lines, so its TPU capability does give it genuine advantages there. Reportedly, soul-searching continues in the Google C-suite about how to more broadly monetize its TPUs. It seems likely that they will find a way. 

As usual, nothing here constitutes advice to buy or sell any security.

Obviously baseline economic security matters, but…

mdmakowsky1 Comment

There’s no getting around the fact that UBI experiments are not producing the kind of results many expected, myself very much included. Now, to be clear, this is in Finland, which has a quite robust social safety net, but precise zeros from a sample of 2,000 unemployed subjects is not something that can be ignored either. If you asked me five years ago where a new UBI might, at the margin, have a zero effect, I would have picked a Nordic country, but still…

The End of Boredom: How AI Companions Might Reduce Random Violence

Joy Buchanan1 Comment

Joy writes: I read Co-Intelligence by Ethan Mollick (thanks to Samford for the free book). Most of it is old news for those of us who follow Ethan on social media and use ChatGPT. However, something that stood out to me was his mention of a study in which humans decide to give themselves a painful shock rather than sit alone in silence for 15 minutes.

Claude comments further based on my prompt:

The End of Boredom: How AI Companions Might Reduce Random Violence

Remember that study where people would rather shock themselves than sit alone with their thoughts? Ethan Mollick references it in Co-Intelligence, and it reveals something unsettling: 67% of men and 25% of women chose electric shocks over sitting quietly for just 15 minutes.

Here’s a strange thought—what if our AI-saturated future accidentally reduces certain types of violence simply by eliminating boredom?

The Violence-Boredom Connection

“Idle hands are the devil’s workshop” exists for a reason. Research has long linked boredom to risk-taking, substance abuse, and impulsive violence—the opportunistic kind that emerges from restlessness rather than planning. Young men starting fights on street corners, vandalism on summer nights, the restless energy that sometimes turns destructive—much of this stems from the unbearable weight of having nothing to do.

Enter Infinite Engagement

We’re rapidly approaching a world where boredom might become extinct. AI companions are becoming always available, infinitely patient, endlessly novel, and perfectly tailored to individual interests. Your future AI won’t just answer questions—it will debate philosophy at 3 AM, create personalized games, generate stories with you as the protagonist, or help explore any curiosity rabbit hole.

The cognitive void that once led people to shock themselves rather than think? It might simply cease to exist.

The Unexpected Benefits

Consider the implications: Young men who might have started fights out of restlessness could instead be deeply engaged in AI-assisted music production or coding projects. The same restless energy that manifests destructively could be channeled into creative collaboration.

AI companions could help process frustration before it builds to a breaking point—imagine having an infinitely patient listener during those dangerous 2 AM spirals of rage. While not replacing human connection, AI interaction might buffer the worst effects of isolation that can lead to radicalization.

The Dark Side

This isn’t utopian. An always-engaged society raises concerns: What happens to human relationships when AI companionship is always easier? Does constant stimulation atrophy our ability to self-reflect? Might we lose the creative insights that emerge from boredom?

Crucially, this would only address impulsive, boredom-driven violence. Systemic violence, ideological extremism, and deeper social problems won’t disappear because people have engaging AI companions.

A Strange New World

For the first time in history, boredom—that uncomfortable void that has driven both creativity and destruction—might become optional. The same species that chose shocks over silence is about to get exactly what it wanted: constant, meaningful stimulation.

Whether this leads to a more peaceful society or new problems we haven’t imagined remains to be seen. The question is whether we’ll recognize what we’ve lost only after it’s gone.

The experiment is already underway—we’re all participating in it.

Joy comments at the end: The AI-written essay is willing to explore downsides of AI engagement.

Visualizing the Sharpe Ratio

Zachary BartschLeave a comment

We all like high returns on our investments. We also like low volatility of those returns. Personally, I’d prefer to have a nice, steady 100% annual return year after year. But that is not the world we live in. Instead, there are a variety of returns with a diversity of volatilities. A general operating belief is that assets with higher returns tend to be associated with greater return volatility. The phrase ‘scared money don’t make money’ implies that higher returns are risky. The Sharpe ratio is a tool that helps us make sense of the risk-reward trade-off.

Let’s start with the definition.

By construction, the risk-free return is guaranteed over some time period and can be enjoyed without risk. Practically speaking, this is like holding a US treasury until maturity. We assume that the US government won’t default on its debt. Since there is no risk, the volatility of returns over the time period is zero.

Since an asset’s return doesn’t mean much in a vacuum, we subtract the risk-free return. The resulting ‘excess return’ or ‘risk premium’ tells us the return that’s associated with the risk of the asset. Clearly, it’s possible for this difference to be negative. That would be bad since assets bear a positive amount of risk and a negative excess return implies that there is no compensation for bearing that risk.

The standard deviation of an asset’s returns are a measure of risk. An asset might have a higher or lower value at sale or maturity. Since the future returns are unknown and can end up having any one of many values, this encapsulates the idea of risk. Risk can result in either higher or lower returns than average!

Putting all the pieces together, the excess return per risk is a measure of how much an asset compensates an investor for the riskiness of the returns. That’s the informational content of the Sharpe ratio, which we can calculate for each asset using historical information and forecasts. Once we’ve boiled down the risk and reward down to a single number, we can start to make comparisons across assets with a more critical eye.

Sometimes friends or students will discuss their great investment returns. They achieve the higher returns by adopting some amount of risk. That’s to be expected. But, invariably, they’ve adopted more risk than return! That means that their success is somewhat of a happy accident. The returns could easily have been much different, given the volatility that they bore.

Let’s get graphical.

Consider a graph in (standard deviation, return) space. In this space we can plot the ordered pair for some portfolios. The risk-free return occurs on the vertical intercept where the return is positive and the standard deviation is zero. Say that a student was thrilled with asset A’s 23.5% return and that it’s standard deviation of returns was 16%. Meanwhile, another student was happy with asset B’s 13.5% return and 5% standard deviation. With a risk-free rate of 3.5%, the Sharpe ratios are 1.25 & 2 respectively. We can plot the set of standard deviation and return pairs that would share the same constant Sharpe ratio (dotted lines). Solving for the asset return:

The above is simply a linear function relating the return and standard deviation. In particular, it says that for any constant Sharpe ratio, there is a linear relationship between possible asset returns and standard deviations. The below graph plots the two functions that are associated with the two asset Sharpe ratios. The line between the risk-free coordinate and the asset coordinate identifies all of the return-standard deviation combinations that share the same Sharpe ratio. This line is known as the iso-Sharpe Line.

With this tool in hand, we can better interpret the two student asset performances. There are a couple of ways to think about it. If asset A’s 23.5% return had been achieved with an asset that shared the Sharpe ratio of asset B, then it would have had risk that was associated with a standard deviation of only 10%. Similarly, if asset A’s volatility remained constant but enjoyed the returns of asset B’s Sharpe ratio, then its return would have been 35.5% rather than 23.5%. In short, a higher Sharpe ratio – and a steeper iso-Sharpe line – imply a bigger benefit for each unity of risk. The only problem is that a such an nice asset may not exist.

Continue reading →

Benefit Cliff Data

James Bailey1 Comment

I said years ago on my Ideas Page that we need data and research on Benefit Cliffs:

Benefits Cliffs: Implicit marginal tax rates sometimes go over 100% when you consider lost subsidies as well as higher taxes. This could be trapping many people in poverty, but we don’t have a good idea of how many, because so many of the relevant subsidies operate at the state and local level. Descriptive work cataloging where all these “benefits cliffs” are and how many people they effect would be hugely valuable. You could also study how people react to benefits cliffs using the data we do have.

But it turns out* that the Atlanta Fed has now done the big project I’d hoped some big institution would take on and put together the data on benefits cliffs. They even share it with an easy-to-use tool that lets you see how this applies to your own family. Based on your family’s location, size, ages, assets, and expenses, you can see how the amount of public assistance you are eligible for varies with your income:

Then see how your labor income plus public assistance changes how well off you are in terms of real resources as your labor income rises:

For a family like mine with 3 kids and 2 married adults in Providence, Rhode Island, it shows a benefit cliff at $67,000 per year. The family suddenly loses access to SNAP benefits as their labor income goes over $67k, making them worse off than before their raise unless their labor income goes up to at least $83,000 per year.

I’ve long been concerned that cliffs like this in poorly designed welfare programs will trap people in (or near) poverty, where they avoid taking a job, or working more hours, or going for a promotion, or getting married, in order to protect their benefits. This makes economic sense for them over a 1-year horizon but could keep them from climbing to independence and the middle-class in the longer run. You can certainly find anecdotes to this effect, but it has been hard to measure how important the problem is overall given the complex interconnections between federal, state, and local programs and family circumstances.

I look forward to seeing the research that will be enabled by the full database that the Atlanta Fed has put together, and I’m updating my ideas page to reflect this.

*I found out about this database from Jeremy’s post yesterday. Mentioning it again today might seem redundant, but I didn’t want this amazing tool to get overlooked for being shared toward the bottom of a long post that is mainly about why another blogger is wrong. I do love Jeremy’s original post, it takes me back to the 2010-era glory days of the blogosphere that often featured long back-and-forth debates. Jeremy is obviously right on the numbers, but if there is value in Green’s post, it is highlighting the importance of what he calls the “Valley of Death” and what we call benefit cliffs. The valley may not be as wide as Green says it is and it may be old news to professional tax economists, but I still think it is a major problem, and one that could be fixed with smarter benefit designs if it became recognized as such.

Poverty Lines Are Hard to Define, But Wherever You Set Them Americans Are Moving Up (And The “Valley of Death” is Less Important Than You Think)

Jeremy Horpedahl8 Comments

Last week I wrote a fairly long post in response to an essay by Michael Green. His essay attempted to redefine the poverty line in the US, by his favored calculation up to $140,000 for a family of four. That $140,000 number caught fire, being covered across not only social media and blogs, but in prominent places such as CNN and the Washington Post. That $140,000 number was key to all of the headlines. It grabbed attention and it got attention. So it’s useful to devote another post this week to the topic.

And Mr. Green has written a follow-up post, so we have something new to respond to. Mr. Green has also said a lot of things on Twitter, but Twitter can be a place for testing out ideas, so I will mostly stick to what he posted on Substack as his complete thoughts. I am also called out by name in his Part 2 post, so that’s another reason to respond (even though he did not respond directly to anything I said).

Once again, I’ll have 3 areas of contention with Mr. Green:

  1. As with last week, I maintain that $140,000 is way too high for a poverty line representing the US as a whole (and Mr. Green seems to agree with this now, even though $140,000 was the headline in all of the major media coverage)
  2. There are already existing alternative measures of what he is trying to grasp (people above the official poverty line but still struggling), such as United Way’s ALICE, or using a higher threshold of the poverty rate (Census has a 200% multiple we can easily access)
  3. His idea of the “Valley of Death” is already well-covered by existing analyses of Effective Marginal Tax Rates, and tax and benefit cliffs. This isn’t to say that more attention is warranted, but Mr. Green doesn’t need to start his analysis from scratch. And this “Valley” is probably narrower than he thinks.
Continue reading →

AI Computing Tutorial: Training vs. Inference Compute Needs, and GPU vs. TPU Processors

Scott Buchanan2 Comments

A tsunami of sentiment shift is washing over Wall Street, away from Nvidia and towards Google/Alphabet. In the past month, GOOG stock is up a sizzling 12%, while NVDA plunged 13%, despite producing its usual earnings beat.  Today I will discuss some of the technical backdrop to this sentiment shift, which involves the differences between training AI models versus actually applying them to specific problems (“inference”), and significantly different processing chips. Next week I will cover the company-specific implications.

As most readers here probably know, the popular Large Language Models (LLM) that underpin the popular new AI products work by sucking in nearly all the text (and now other data) that humans have ever produced, reducing each word or form of a word to a numerical token, and grinding and grinding to discover consistent patterns among those tokens. Layers of (virtual) neural nets are used. The training process involves an insane amount of trying to predict, say, the next word in a sentence scraped from the web, evaluating why the model missed it, and feeding that information back to adjust the matrix of weights on the neural layers, until the model can predict that next word correctly. Then on to the next sentence found on the internet, to work and work until it can be predicted properly. At the end of the day, a well-trained AI chatbot can respond to Bob’s complaint about his boss with an appropriately sympathetic pseudo-human reply like, “It sounds like your boss is not treating you fairly, Bob. Tell me more about…” It bears repeating that LLMs do not actually “know” anything. All they can do is produce a statistically probably word salad in response to prompts. But they can now do that so well that they are very useful.*

This is an oversimplification, but gives the flavor of the endless forward and backward propagation and iteration that is required for model training. This training typically requires running vast banks of very high-end processors, typically housed in large, power-hungry data centers, for months at a time.

Once a model is trained (e.g., the neural net weights have been determined), to then run it (i.e., to generate responses based on human prompts) takes considerably less compute power. This is the “inference” phase of generative AI. It still takes a lot of compute to run a big program quickly, but a simpler LLM like DeepSeek can be run, with only modest time lags, on a high end PC.

GPUs Versus ASIC TPUs

Nvidia has made its fortune by taking graphical processing units (GPU) that were developed for massively parallel calculations needed for driving video displays, and adapting them to more general problem solving that could make use of rapid matrix calculations. Nvidia chips and its CUDA language have been employed for physical simulations such as seismology and molecular dynamics, and then for Bitcoin calculations. When generative AI came along, Nvidia chips and programming tools were the obvious choice for LLM computing needs. The world’s lust for AI compute is so insatiable, and Nvidia has had such a stranglehold, that the company has been able to charge an eye-watering gross profit margin of around 75% on its chips.

AI users of course are trying desperately to get compute capability without have to pay such high fees to Nvidia. It has been hard to mount a serious competitive challenge, though. Nvidia has a commanding lead in hardware and supporting software, and (unlike the Intel of years gone by) keeps forging ahead, not resting on its laurels. 

So far, no one seems to be able to compete strongly with Nvidia in GPUs. However, there is a different chip architecture, which by some measures can beat GPUs at their own game.

NVIDIA GPUs are general-purpose parallel processors with high flexibility, capable of handling a wide range of tasks from gaming to AI training, supported by a mature software ecosystem like CUDA. GPUs beat out the original computer central processing units (CPUs) for these tasks by sacrificing flexibility for the power to do parallel processing of many simple, repetitive operations. The newer “application-specific integrated circuits” (ASICs) take this specialization a step further. They can be custom hard-wired to do specific calculations, such as those required for bitcoin and now for AI. By cutting out steps used by GPUs, especially fetching data in and out of memory, ASICs can do many AI computing tasks faster and cheaper than Nvidia GPUs, and using much less electric power. That is a big plus, since AI data centers are driving up electricity prices in many parts of the country. The particular type of ASIC that is used by Google for AI is called a Tensor Processing Unit (TPU).

I found this explanation by UncoverAlpha to be enlightening:

A GPU is a “general-purpose” parallel processor, while a TPU is a “domain-specific” architecture.

The GPUs were designed for graphics. They excel at parallel processing (doing many things at once), which is great for AI. However, because they are designed to handle everything from video game textures to scientific simulations, they carry “architectural baggage.” They spend significant energy and chip area on complex tasks like caching, branch prediction, and managing independent threads.

A TPU, on the other hand, strips away all that baggage. It has no hardware for rasterization or texture mapping. Instead, it uses a unique architecture called a Systolic Array.

The “Systolic Array” is the key differentiator. In a standard CPU or GPU, the chip moves data back and forth between the memory and the computing units for every calculation. This constant shuffling creates a bottleneck (the Von Neumann bottleneck).

In a TPU’s systolic array, data flows through the chip like blood through a heart (hence “systolic”).

  1. It loads data (weights) once.
  2. It passes inputs through a massive grid of multipliers.
  3. The data is passed directly to the next unit in the array without writing back to memory.

What this means, in essence, is that a TPU, because of its systolic array, drastically reduces the number of memory reads and writes required from HBM. As a result, the TPU can spend its cycles computing rather than waiting for data.

Google has developed the most advanced ASICs for doing AI, which are now on some levels a competitive threat to Nvidia.   Some implications of this will be explored in a post next week.

*Next generation AI seeks to step beyond the LLM world of statistical word salads, and try to model cause and effect at the level of objects and agents in the real world – – see Meta AI Chief Yann LeCun Notes Limits of Large Language Models and Path Towards Artificial General Intelligence .

Standard disclaimer: Nothing here should be considered advice to buy or sell any security.