Data

Since 2021, 80 Percent of Population Growth in the US Has Been from International Migration

Jeremy HorpedahlLeave a comment

The following chart shows cumulative population growth in the US since 2010, from two sources: the natural population growth (birth minus deaths) and international migration:

In total, the US population has increased by about 30 million people since 2010. Cumulatively, about 55 percent of the growth has been from international migration, but there are two distinct periods within this 15-year timeframe. From 2010 to 2020, about 60 percent of the population growth was from natural population change, both cumulatively and in most of those years. From 2021 forward, 70-90 percent of the growth has been from international migration.

The flip in 2021 happens because both factors changed. First, the natural rate of population growth slowed dramatically, with just 146,000 people added to the population, compared with close to 1 million or more before the COVID pandemic. The decline in population growth is a result of gradually slowing birth rates, but also skyrocketing death rates in 2021-2022: about 3.4 million deaths per year, compared with about 2.8 million pre-pandemic. Second, international migration picked up dramatically, from around half a million people in 2019-2020, to an average of over 2 million per year from 2022-2024.

Note: the years in this data run from July to June, so when it says 2025 in the chart, this means from 7/1/2024 to 6/30/2025. Thus, we don’t yet have a full year of data under Trump. But even with the half year under Trump, which includes 6-7 months under Biden when the border policy was already being reversed, the latest year of data from Census suggests the US still had a net international migration of almost 1.3 million people. That’s half the number from 2024, but still well above pre-pandemic numbers. Keep in mind that these are estimates, subject to change, and estimating changes in the illegal immigrant population is often very difficult to do accurately. But these are probably the best estimates that we have right now.

What does the future hold? Of course, any future projection has to make assumptions about how both the birth rate and immigration rate will change over the coming years. But a recent estimate from CBO suggests that by around 2032-2033 the natural rate of population growth will essentially hit zero, and that by the early 2050s it will be so negative as to completely offset the projected immigration. In other words, total population growth could essentially be zero in the US by 2055 or so. That’s 30 years in the future, so take it with a grain of salt, as any small change in immigration, births, or deaths could throw that projection way off. But it seems like a fairly likely scenario.

Regulatory Burden By Presidential Administration

Zachary Bartsch1 Comment

During president Trump’s first term in office, he made a bunch of waves (as he’s wont to do). His more educated supporters said that he engaged in substantial deregulation of telecommunications, which got a lot of press. There was a quiet contingent of educated voters who were relatively silently supportive on Trump’s regulatory policy, even if his character was indefensible or his other policy was less desirable.

But was Trump a great deregulator? Or was it one of those cases when we say that he regulated *less* than his fellow executives? The George Washington University Regulatory Studies Center can help shed some light with their data. Specifically, they have calculated the number of ‘economically significant’ regulations passed during each month of each president going back through Ronald Reagan’s term. What counts as ‘economically significant’? The definition has changed over time. But, generally, ‘economically significant’ regulations:

  1. “Have an annual [adverse] effect on the economy of $100 million or more
  2. Or, adversely affect in a material way the economy, a sector of the economy, productivity, competition, jobs, the environment, public health or safety, or State, local, or tribal governments or communities.”

The only exception to this is between April 6, 2023 and January 20, 2025 when the threshold was raised to $200 million.

The Data

The graph below-left shows the number of economically significant regulations for each president since the start of his term, through July of 2025. It’s reproduced from the link above except that I appended Trump’s second term onto his first term. What does the graph tell us? There doesn’t seem to be much of a difference between republicans and democrats. Rather, it seems that, generally, the number of economically significant regulations increases over time. Importantly, the below lines are cumulative by president. So each year’s regulations each cost $100m annually and that’s on top of the existing ones already in place. So, regulatory costs generally rise, with the caveat that we don’t see the relief provided by small or rescinded regulations (for that matter, we don’t see small regulatory burdens here either). Something else that the below graph tells us is that presidents tend to accelerate their economically significant regulations prior to leaving office. Reagan was the only exception to this pattern and he *slowed* the number of regulations as the end of his term approached.

Below-right is the same data, but the x-axis is months until leaving office. Every president since Bush-41 has accelerated their burdensome regulations during their final months in office. The timing of the acceleration corresponds to how close the preceding election was and whether the incumbent president lost. Whereas all presidents regulate more in their last 2-3 months in office, the presidents who were less likely to win re-election started regulating more starting around eight months prior to leaving office. Of course, they wouldn’t say that they expected to lose, but they sure regulated like there was no tomorrow.

What about Trump? Trump’s fewer regulations is caused by his single term. He definitely still added to the regulatory burden (among economically significant regulations, anyway). While Trump started with the fewest additional regulations since Reagan, and Biden started with the most ever initial regulations, together they earn the top prizes for most regulations added in their first term.

What if we append these regulations from end-to-end? That’s what the below chart does. We do have to be careful because the series is a measure of gross economically significant regulations and not net economically significant regulations. So, it’s possible that some rescissions dampened the below values, but this is the data that I have for the moment. While each presidential administrations increases regulation more than the prior, the good news is that the rate of change is not exponential. The line of best fit is quadratic. We’re experiencing growing regulations, but at least it’s not compound growth.

The Cost

We can estimate the costs of these economically significant regulations. It’s a rough cut, and definitely a lower bound since rescission is rare and $100 million is itself a lower bound, but we can multiply the number of regulations by $100m to get minimum annual cost. Like I said, the Biden criterion from April 2023 through January 20, 2025 changed, so those regulations get counted as $200 million instead. The change in definition means that the regulation counts underestimate the late-term Biden regulations relative to the other presidencies.

Continue reading →

Most Married Women with Children Were Working By the Late 1970s

Jeremy Horpedahl1 Comment

A recent essay by Jeffrey Tucker asks “Has Life Really Improved in Half a Century?” Specifically, Mr. Tucker is interested in measuring median income of families (he uses household income, but families are clearly what he is interested in).

Tucker grants that real median household income has increased by about 40 percent from 1984 to 2024 (if he had used family income instead, the increase is almost 50 percent). But… he says this is illusory. That’s because it now takes two incomes to achieve that median income, whereas it only took one income in the past:

“Adding another income stream to the household is a 100 percent rise in work expectations but it has yielded only a 20-plus percent rise in material income. The effective pay per hour of work for the household has fallen by 40 to 50 percent!”

(He makes a data error by saying that in 1976 real median household income was $68,000-$70,000, when it was actually $59,000 in 2024 dollars in 1976 — real income didn’t fall from 1976 to 1984!)

Continue reading →

Vaccine Variety

Zachary Bartsch4 Comments

The flu and covid-19 vaccines don’t work super well. Both vaccines permit infection and transmission at quite high rates. The benefit from these vaccines come largely from reductions in mortality or severe symptoms conditional on infection. The covid-19 vaccine is itself especially risky or ineffective depending on the age and health of the individual. Plenty of people eschew vaccines.

I live in Collier County, Florida where there have been 61 confirmed cases of measles so far this year. I have since learned that Measles is EXTREMELY contagious. It floats around the air and on items and just sort of hangs out and waits for a place to replicate. I’ve also learned that symptoms include a fever, eye irritation, possible brain swelling, severe dehydration, and a characteristic rash. The severe dehydration easily puts people in the hospital, the eye irritation can lead to permanent vision loss, and the brain swelling can be acute, or a symptom delayed by 5-6 years, which can also be fatal. I’ve also learned that having the vaccine, which is usually administered in two doses, provides about 97% immunity. The vaccine works so well, that the department of health recommends no behavioral change among the vaccinated population when there is a measles outbreak. Barring unique circumstances, measles immunity can persist for a lifetime.

Unfortunately, a large segment of the anti-vaccine mood affiliation retains the salience of the covid-19 vaccine characteristics. Other vaccines and diseases in the typical pediatric schedule are not similar. Most of these prevent infection >90% of the time (TDAP is low at 73%), prevent transmission, reduce mortality when there are breakthrough infections, are effective for years or decades, and are extremely safe for all age groups.

The risks of disease versus the corresponding vaccine are orders of magnitude away from each other. The tables below summarize the data (with sources). I did not double check the source on every single figure. If you glance below, then you’ll see why: Even if the numbers are closer by 10 or 100 times, vaccines still look really good.

First, mortality: The data is divided by disease and age group, and provides mortality rates for both the disease and for the vaccine. The numbers are proportions, conditional on infection or vaccination. There are a lot of zeros in the vaccine mortality rates and certainly more than for the diseases. For example, a measles infection is 10,000 more lethal than the MMR vaccine which prevents it. In fact, all of those zeros in the vaccine rates reflect mortality that is so uncommon, that the estimated one out of every 10 million is just rounded up because researchers don’t think that the risk is zero.

Continue reading →

GDP Forecasts for 2025Q4

Jeremy Horpedahl1 Comment

UPDATE 2/19/2026: the last GDPNow estimate from the Atlanta Fed is 3.0% and the Kalshi markets are now predicting 2.8%. I would expect this is a slightly better range than the 3.3-3.6% from my post written on 2/18/2026.

In April 2025 I wrote about several different forecasts for GDP growth. At the time the latest GDP quarter available was 2025Q1. We’ve had two more quarters of data since then, plus a highly anticipated report for Q4 coming out this Friday. How have these different predictions done recently? Here is the updated table from that prior post:

When I wrote the post in April 2025, I said that a simple average of the Atlanta Fed and Kalshi forecasts was the best simple predictor of the actual BEA advance figure. Based on the middle quarters of 2025, I think that continued to be true: each of them was the best estimate in one quarter, and perhaps just as importantly the NY Fed and WSJ survey of economists understated GDP growth pretty significantly.

There is still one more Atlanta Fed GDPNow update coming tomorrow before we get the actual BEA data on Friday, but based on where the numbers are now, we should expect Q4 to be around 3.4-3.5% (annualized) growth rate. This would put the total 2025 calendar year growth at around 2.3% — decent, but still below 2024’s 2.8% growth.

Truth: The Strength and Weakness of AI Coding

Zachary Bartsch1 Comment

There was a seismic shift in the AI world recently. In case you didn’t know, a Claude Code update was released just before the Christmas break. It could code awesomely and had a bigger context window, which is sort of like memory and attention span. Scott Cunningham wrote a series of posts demonstrating the power of Claude Code in ways that made economists take notice. Then, ChatGPT Codex was updated and released in January as if to say ‘we are still on the frontier’. The battle between Claude Code and Codex is active as we speak.

The differentiation is becoming clearer, depending on who you talk to. Claude Code feels architectural. It designs a project or system and thrives when you hand it the blueprint and say “Design this properly.” It’s your amazingly productive partner. Codex feels like it’s for the specialist. You tell it exactly what you want. No fluff. No ornamental abstraction unless you request it.

Codex flourishes with prompts like “Refactor this function to eliminate recursion”, or “ Take this response data and apply the Bayesian Dawid-Skene method”. It does exactly that. It assumes competence on your part and does not attempt to decorate the output. It assumes that you know what you’re doing. It’s like your RA that can do amazing things if you tell it what task you want completed. Having said all of this, I’ve heard the inverse evaluations too. It probably matters a lot what the programmer brings to the table.

Both Claude Code and Codex are remarkably adept at catching code and syntax errors. That is not mysterious. Code is valid or invalid. The AI writes something, and the environment immediately reveals whether it conforms to the rules. Truth is embedded in the logical structure. When a single error appears, correction is often trivial.

When multiple errors appear, the problem becomes combinatorial. Fix A? Fix B? Change the type? Modify the loop? There are potentially infinite branching possibilities. Even then, the space is constrained. The code must run, or time out. That constraint disciplines the search. The reason these models code so well is that the code itself is the truth. So long as the logic isn’t violated, the axioms lead to the result. The AI anchors on the code to be internally consistent. The model can triangulate because the target is stable and verifiable.

AI struggles when the anchor disappears

Continue reading →

Forecasting An Eventful 2026

James BaileyLeave a comment

May you live in interesting times – apocryphal Chinese curse

In early 2025 I shared forecasts about the economy that turned out to be pretty good. This year, economic forecasts center around a boringly decent year (2.6% GDP growth, inflation below 3%, unemployment stays below 5%, no recession), though with high variance. But forecasts about politics and war foretell a turbulent year.

In the US, midterm elections have a 78% chance to flip control of the House and 35% chance to flip the Senate despite a tough map for Democrats. A midterm wave for the out-of-power party is typical in the US, given that the party in power always seems to over-play their hand and voters quickly get sick them. More surprising is that forecasters give a 44% chance that Donald Trump leaves office before his term is up, and a 16% chance that he leaves office this year. Markets give a 20% chance that he will be removed from office through the impeachment process, so the rest of the 44% would be from health issues or voluntary resignation.

Forecasters at Kalshi predict a greater than even chance that 4 notable world leaders leave office this year:

I find this especially notable because Viktor Orban is the only one who would be removed through regularly scheduled elections. In the UK, Keir Starmer was just elected Prime Minister in 2024 and doesn’t have to face reelection until 2029; but he is so unpopular that his own Labor Party is likely to kick him out of office if local elections in May go as badly as polls indicate. If so, he would join Boris Johnson and Liz Truss as the third British PM in four years to leave office without directly losing an election. The leaders of Cuba and Iran don’t face real elections and would presumably be pushed out by a popular uprising or US military action.

Some other important world leaders will probably stay in office this year, but forecasters still think there is a significant chance they leave: Israel’s Netanyahu (49%), Ukraine’s Zelenskyy (32%), and Russia’s Putin (14%). For the latter two, this belief could be tied to the surprisingly high odds given to a ceasefire in the Russia-Ukraine war this year (45%). Orban leaving office could be tied into this, as Hungary has often vetoed EU support for Ukraine.

Myself, I find most of these market odds to be high, and I’m tempted to make the “nothing ever happens” trade and bet that everyone stays in office. But even if all these markets are 10pp high, it still implies quite an eventful year ahead. Prepare accordingly.

2025 Was A Pretty Good Year for the US Economy — But So Was 2024

Jeremy Horpedahl1 Comment

While some of the 2025 data is still coming in (such as GDP), we already have much of the core economic data to evaluate the year.

In a recent op-ed, President Trump claimed, “Just over one year ago, we were a ‘DEAD’ country. Now, we are the ‘HOTTEST’ country anywhere in the world!” Of course, every President claims they are doing great things, and Americans are almost exactly evenly divided over whether the economy was better under Biden or Trump — but this mostly just partisanism (Independents are close to evenly split, though).

So what is the truth? I have put together what I think are the best economic indicators to judge how the economy is doing. And what does it tell us? I think the fairest read is that 2025 was a pretty good year, but based on most economic data it was almost identical to 2024.

The only indicator that is clearly better is private-sector job growth in 2024. We might add S&P 500 in 2024 growth too, although some other assets such as gold have performed better in 2025. Inflation in 2025 is a tad lower, but not the massive improvement Trump suggests. This is especially the case for one of his favorite prices, gasoline. Yes, 2025 is a little lower than 2024… just like 2024 was a little lower than 2023.

And what of that greatest of all macroeconomic indicators, GDP? We don’t yet have Q4 data for GDP, which means we don’t have full-year 2025 data yet. But the growth rate of real GDP in 2024 was 2.8%, and betting markets are currently predicting 2.3% for 2025. Betting markets could be wrong! But it seems unlikely it would be much above 2.8% (those same betting markets only think there is a 4% chance it will be over 3.0%).

None of this is to say that the 2024 and 2025 economies are exactly the same. Certainly there is more uncertainty due to the shifting tariff policy, but on the other hand even with that uncertainty the economy is still performing fairly well. And my table above only includes economic outcomes, not any changes to government budgets, nor important social indicators such as crime. These are important too, but my focus in this post is only on the economic data.

It seems that in those surveys about whether the economy is better now or under Biden, it would be useful to offer an “about the same” option. Of course, in 2021-2022 inflation was much worse under Biden — but job growth was much better. A lot of this was baked in from the pandemic, 2020 monetary and fiscal stimulus, etc. Once we were back to a semi-normal economy in 2024, it was a decent year. Not blockbuster, but decent. So was 2025.

The US Homicide Rate in 2025 May Have Been the Lowest Ever

Jeremy Horpedahl1 Comment

The following chart merges two data sources to create a long-run series on homicides in the US. Based on early estimates for 2025 from Jeff Asher, the homicide rate may be as low as 4.3 murders per 100,000 population. That would be the lowest since at least 1900, and possibly the lowest US homicide rate ever since the best evidence suggests it was even higher pre-1900. The current record low was 4.4 murders per 100,000, which the US saw in 1955, 1957, and 2014.