I spent most of last week fighting with a model.
Before anyone starts googling “Nerdy Economist in Fashion Week Brawl,” I should clarify.
I was fighting with a macroeconomic model that insisted on telling me something I didn’t believe. To be precise, it was projecting that, given the recent and projected pace of U.S. economic growth, the unemployment rate would slide to 3.0% by the end of 2025.
This I don’t believe for reasons I’ll explain. But the changes in assumptions necessary to produce a more reasonable answer can tell us a lot about the likely path of economic growth, inflation, interest rates, corporate profits and the dollar over the next two years with significant implications for financial markets and investing.
The Super-Low Unemployment Solution—And Why It Is Unlikely
For decades and through a number of jobs, I have built and run a macro-economic model of the U.S. economy. It is not particularly revolutionary model, but I am rather fond of it and, like an old car you tune up and polish every weekend, I continually update and adjust it to give me a better read on the likely direction of the U.S. economy.
In this model, the growth in employment, as measured by the household survey, is a function of current and lagged real GDP growth and job openings. Job openings at the end of December were just over 9 million—still almost one and a half million higher than in their peak month before the pandemic. The U.S. economy grew at an annualized pace of 4.9% in the third quarter of last year and 3.3% in the fourth. Even if job openings continued their gradual decline to normal levels and real GDP growth slowed to a 1.9% annualized pace over the next two years—which is the pace the model predicts by summing up the components of demand—employment would grow by an average of 152,000 persons per month.
This, of course, would be a very nice result were it not for the fact that Census Bureau projections imply that America won’t produce enough working-age people to fill these jobs. Specifically, Census forecasts from November show the population aged 16 and older growing by an average of 171,000 per month over the next two years. Holding the labor force participation rate constant at January’s 62.52% would imply a monthly increase in the labor force of just 107,000, causing the number of people unemployed to fall by 46,000 per month from its current 6.124 million level. By the end of next year, this would leave us with less than 5.1 million people unemployed and an unemployment rate of 3.0%.
This, as I said, I don’t believe because of the behavior of the unemployment rate in the later years of previous cycles.
In every economic expansion in the last 40 years, there has come a point where the unemployment rate has simply stopped falling—a point at which we reach a sort of minimum practical level of unemployment. There will always be people who have just gotten fired or laid off, who just graduated college or reentered the labor market, who impulsively quit, who are in the wrong town or wrong profession or who are incapable of holding onto a job even though they are habitually looking for one.
This minimum, sometimes called the full-employment unemployment rate, has shifted over time. In the late 1980s, it appeared to be just over 5%, while by the late 1990s it was close to 4.0%. It may have risen in the first decade of this century—before the Great Financial Crisis, the lowest measured unemployment rate was 4.4%. However, it then fell further over the last decade, with unemployment hitting 3.5% both in late 2019 and early 2020 before the pandemic struck. However, in all four cases, the unemployment rate essentially stopped falling well before a recession started.
This seems to have happened again. January marked the 26th straight month in which the unemployment rate was at, or below, 4%. But in that more than two-year period, it has oscillated in a narrow range between 3.4% and 4.0% and, in January it was 3.7%. A reasonable conclusion, looking at the past two years, is that the minimum possible unemployment rate in the economy of the mid-2020s is about 3.4% and a forecast of 3.0% is just not credible.
But if this is the case, how should I tweak the assumptions in my model to keep the unemployment rate from falling much below 3.5%? There are really only three ways of getting there: Cut expected demand growth, increase expected labor force participation or assume greater productivity growth. Or, more reasonably, make some adjustment to all three.
Lessons from History
Before proceeding down this path, it is worth looking at the lessons of history from other periods when a low unemployment rate stopped falling. In particular, we can examine the years immediately preceding the recessions of 1990-91, 2001, 2007-2009 and 2020, as well as the period since the December 2021. Looking at the behavior of economic growth, the labor force participation rate and productivity over these periods produces no universal insight. However, it does provide some guidance on what happened and why.
• In the leadup to the 1990-91 recession, the economy essentially ran out of gas. Economic growth slowed and productivity growth remained anemic. However, the labor force participation rate did rise, preventing a further decline in unemployment.
• By contrast, before the 2001 recession, economic growth remained strong and there was only a marginal increase in labor force participation. However, the more widespread use of personal computers and internet technology produced a surge in productivity.
• In the run up to the Great Financial Crisis, productivity growth was anemic and labor force participation hardly budged. However, economic growth slowed sharply as we approached recession, both because of a downturn in home-building and surging energy prices.
• In the pre-pandemic economy, economic growth surged, partly due to the 2017 Tax Act and productivity growth was slow. However, the labor force participation rate rose despite a demographic headwind.
• Finally, in the last two years of this expansion, we have seen a moderation in economic growth and very little net gain in productivity. However, the labor force participation rate has risen, even in the face of the continued retirement of the baby boom.
Getting To A More Realistic Unemployment Rate Forecast
Given all of this, in attempting to boost the forecast of the end-2025 unemployment rate from 3.0% to something closer to 3.5%, it seems reasonable to trim expected real GDP growth, slightly boost productivity growth and assume a slightly higher labor force participation rate. It should be stressed that, in each case, this is a game of inches. However, the collective implication of these inches is important.
On real GDP growth, it may be that consumer spending will decelerate even more than expected. A lack of available workers could continue to hamper services spending. In addition, consumers have been spending at a pace that left the personal saving rate at 3.7% in December 2023—well below the 6.2% average seen in the five years before the pandemic. If we assume that consumer spending grows by 1.60% annualized over the next two years rather than the 1.74% projected in the baseline forecast, real GDP growth would be 0.1% slower, leaving the unemployment rate at 3.1%, rather than 3.0% by December 2025.
Productivity growth could also be a bit stronger than assumed in the baseline forecast. Productivity growth slumped in 2022 but has surged in recent quarters. A combination of businesses being forced to use labor efficiently in a tight labor market and the gradual rollout of AI and other technologies could allow this rebound to continue. In the baseline forecast, real GDP per worker rises at a 0.92% annual rate. It we assume that it grows at a 1.02% rate instead, the unemployment rate would rise to 3.4% by the end of 2025.
Finally, we may see a very slight increase in the labor force participation rate. It should be stressed that a flat overall participation rate already implies a significant rise in the participation rate of both the population aged 18-64 and the population aged 65+ because of the growing population share of the latter group which normally has much lower participation. Still, if we assume that the overall labor force participation rate rises from 62.52% in January to 62.62% by December 2025, the unemployment rate would end 2025 at between 3.5% and 3.6%.
The Pressures Of A Full Employment Economy
Tweaking a forecast is, of course, an uncomfortable operation. However, in the real world it is unavoidable since there aren’t enough data to capture all the non-linear nuances of the evolution of the economy at turning points and it is important to impose some constraints on a model that would otherwise produce implausible results.
That being said, even after manhandling the model to prevent the unemployment rate from falling too low, it appears the most likely path for the economy is one of a continued, very tight labor market.
If this transpires, it may well be that wage growth, which had fallen from 5.9% year-over-year in March 2022 to 4.3% by October 2023, falls much more slowly from here on out, despite job openings returning to more normal levels. If that occurs, it could slow the decline in headline inflation, although an erosion in price increases in shelter and auto insurance should still drive the key U.S. inflation gauges down to close to 2.0% year-over-year by the end of this year. Equally importantly for investors, it could make it harder for U.S. companies to increase margins after stalling out in 2023.
A continued tight labor market, along with slower progress in reducing inflation, would likely dash hopes of more aggressive Fed easing. Indeed, it is even possible that the Fed could back away from its current projection of three rate cuts in 2024 and four in 2025, if it felt it had to continually lean against the wind of a super-tight labor market. This, in turn, suggests limited scope for further declines in long-term interest rates.
Such an outcome could also support the dollar, reducing the dollar value of foreign profits for U.S. companies and the return on international investments for U.S. investors. However, the implications for U.S. equities are less clear, as the boost from a continued strong economy and rising consumer and business confidence could offset greater pressure on margins and dashed hopes for much lower interest rates.
It should be emphasized that this forecast, like all such forecasts, has no hope of being exactly correct. This is not just because of the difficulty in modeling the evolving relationships between economic variables but because of our inability to predict the shocks that can knock an economy off track. No doubt some such shocks will impact the economy over the next two years.
However, it is worth noting that, if these shocks roughly offset each other and the economy traces out something close to this path, it would be a remarkably positive one with the sum of unemployment and inflation by the end of 2025 being roughly 5.6%. This would be lower than 97% of the combined readings on these variables over the last 60 years, making this a very healthy economy for the average American family, if not quite so positive for the average American investor.
David Kelly is chief global strategist at J.P. Morgan Asset Management.