The Dual Track US Economy and Why Interest Rates Should Come Down

The United States economy appears schizophrenic. GDP growth in excess of 3% with real time monitoring from the Federal Reserve Board of Atlanta placing it near 4% at the same time job growth appears flat and companies announcing large layoffs. How can we explain simultaneous robust GDP growth and flat job growth?

Thirty years ago in the United States, in the early days of the internet hitting the mainstream, a curious puzzle stymied economists. The United States was enjoying a period of unprecedented economic growth. From 1996 through 2000, economic growth never dropped beneath 4% and approached 5% which for a country the size and income level represented an astounding growth rate.

However, what puzzled economists even more was the lack of pricing pressures in the economy. United States inflation hit bottom at 1.6% in 1998 when real GDP growth hit an amazing 4.5%. Befuddled economist openly wondered what allowed such economic growth but simultaneous did not put upward pressure on prices stoking inflation?

Economists think in terms of simplified models. For models of economic growth this means something along the lines of people, capital, and a third ambiguous term to capture variables like education or technology with all factors of production remaining static in the short term. Translated that means the number of people in an economy is effectively fixed at a specific point and time as is the education, technology, and capital an economy can access.

This view of the economy, while reasonable, also reveals specific limitations. The model thinking while helping capture the fundamentals of the economy intellectually drifts to a static understanding. It struggles to deal with and incorporate rapid change into thinking about the economy. Though reasonable most of the time, as most things do not cause rapid change, this thinking can also prove a prison when unable to grapple with change.

This method of thinking also suffers from linear conception of an economy. Since technological growth is assumed to remain constant at 2% a year, we grow the economy by either adding more people or capital or both which place non-linear rates increasing pressure on prices. In the mid to late 1990’s economists, holding the rate of technological change and productivity growth constant, failed to understand why robust economic growth depending on increased investment and more labor inputs failed to stoke inflation.

The Chairman of the Federal Reserve Board at the time Alan Greenspan arrived at a conclusion which drew widespread and fevered debate. Chairman Greenspan argued that the technological boom from the internet reduced price pressures in aggregate at the macroeconomic level. In the middle of the internet boom in 1996, Chairman Greenspan wrote:

Yet, with all the extraordinary technological advances of the past couple of decades, why have our recent productivity data failed to register any improvement...first, insofar as recent productivity growth is concerned, I have a serious question about the quality of the data that we employ to measure output in today's economy. I shall come back to that issue shortly. Second, like the major technological advances of earlier periods, it will take time for our newest innovations to work their way into the nation's infrastructure in a productive manner.

Chairman Greenspan argued that the internet increased productivity that allowed higher economic growth from the same number of people and capital. The increased productivity not only drove a higher level of economic growth but it allowed a lower interest rate to maintain low steady price growth.

Let me demonstrate how productivity allows for a lower interest rate even with higher economic growth. Let’s use a simple example where we 100 people and 100 units of capital combine to output 100 widgets. If we assume that total productivity grows at 2% a year, if we have the same number of people and same number of machines next year, output widget output can rise to 102. However, now assume there is a technology shock that allows productivity to grow at 5% annually. Now those same 100 people and 100 machines can produce 105 widgets. As workers became more productive their wages grew faster with higher output but because they became more productive their wages grow in line with their real output not stoking inflation.

So why does this matter in 2025? We see a similar phenomenon playing out in the 2025 United States economy. A structurally higher level of GDP driven by higher productivity reduces the short term demand for labor. It’s impact on prices over the long run will likely restrain prices in the aggregate with higher productivity and lower interest rates, it will drive sectoral inflation such as in electricity generation and transmission.

If we start simple since, ChatGPT was introduced in November 2022, the data supports the idea the United States has enjoyed a structural uptick in labor productivity. It must be noted, that due to covid induced wild oscillations, I am omitting that period from the analysis but pre-covid labor was ranging from 1-2% but post ChatGPT has ranged typically from 2% to even at times above 3%. Given Chairman Greenspan’s musing about official data failing to capture productivity boosts, nor can we rule out we are missing productivity enhancing changes now.

While economists and business leaders alike will argue over arcane measurement methodologies and definitions of productivity, we can see the impact of this productivity in much more tangible ways.

Take the professional and business services sector in the United States comprised of accountants, lawyers, and consultants a field that would seem one of the most able to benefit from the introduction of large language models. Total employment in professional and business services peaked in early 2023 at just under 23 million but since then has fallen slightly to 22.5 million. Typically falling employment in a sector heralds dark times. However, we do not see that.

In the fourth quarter of 2022 when ChatGPT rolled out, total nominal output came in at $5.4 trillion but in the second quarter of this year clocks in above $6.1 trillion. Coupled with the decline in total employment this means in two and half years output per worker is up nearly 15%. In real terms, output per worker in professional and business services is up 6.5% in total which would be expected to close in a three year growth just above 2.5% real growth since ChatGPT was introduced.

While it may be easy to dismiss this as unique to sectors poised to adopt LLMs such as those responsible for making spreadsheets, we see a similar pattern across a wide variety of sectors. Total retail employment peaked in February 2023 and is basically unchanged since, with a small decline during that time. However, sector adjusted nominal output is up 10.7% per worker and real output per worker is up 4.6%.

In transit and ground transportation, not a sector one would think would be early adopters of LLMs, we see a not entirely similar pattern but producing the same story. Since the end of the fourth quarter 2022, real growth in transit and ground transportation is up 46% but employment is only up 12.5% for an increase in output per worker of 30% in real terms.

Whether due to post-covid stabilization or the release of ChatGPT, we are clearly seeing a change in the labor market some time in the first half of 2023 that started slowing hiring and increasing productivity.

There are a couple of important empirical implications. First, given what appears to be a structural shift upwards in total productivity, this seems to imply a lower natural interest rate due to the higher productivity. Just as Chairman Greenspan argued thirty years ago that higher productivity would allow interest rates to remain lower than before the technology shock because output would rise without demand leading supply.

In fact, if we accept the Federal Reserve data that tariffs have added approximately .3-.5% to inflation, that would imply that the United States is now, absent the tariffs, just slightly above its target inflation rate while simultaneously enjoy robust economic growth. Given that even the Federal Reserve admits a material amount of the post-covid inflation was driven by supply problems, approximately half of the inflation according to some Federal Reserve research, there is a strong argument to make that higher interest rates are exacerbating inflationary pressures rather than reducing them. Take electricity generation and transmission which are long term investments feeding LLM voracious energy appetites. Reducing electricity prices would benefit from additional capacity which would be aided by reducing interest rates. High interest rates are a solution to cure excess demand, but they exacerbate it if the problem is lack of supply. Keeping interest rates high will not alter the growth in demand for LLMs.

Second, while there are clear trends in many industries enjoying strong productivity growth, there are also industries not seeing strong productivity growth. While there is a general trend that the lack of productivity occurs more in tangible good or services, it is decidedly not a uniform generalization that can be made as noted with transit and ground transportation. This divergence between industries seeing sustained productivity growth and those with minimal acceleration drives the two track economy in the United States. There does not seem to be a universal explanation across all the industries but the clear lumpy patterns explains why some parts of the economy remain robust while others seem flat.

Third, if we are in the early stages of a structural shift in productivity, and I would say that is quite likely, the forecast in the short to mid-term should be for restrained growth in employment levels and shifts in skills needed in the work place. The oft used cliché teach them to code may not be entirely accurate but teach them new skills will be very accurate. From insurance adjusters using drones to manage claims faster and from a distance to mechanics being able to diagnose problems and repair them faster, the shift in technology requires new skills even when fulfilling the same basic occupation. One of the explanations for the lumpy change in productivity across industries comes from technological adoption and domain expertise by the technology manufacturers.

The United States is in the early to middle stages of escalating productivity growth. This will impact the demand for labor, the neutral interest rate, the skills workers need, and what we need from government. We need to recognize you cannot solve supply shortage inflation with higher interest rate.