Here is The EXACT Date of The Next Recession — Prof. Jiang Xueqin

[00:00:00] Speaker 1: something strange is happening in the research departments of the biggest banks in the world right now and almost nobody outside those buildings is talking about it. Analysts are running the numbers, building the models, and quietly landing on the same conclusion. But you will not hear a single major institution say it plainly on live television because the second a household named bank puts a specific date on a recession. It stops being an internal forecast and turns into a headline that can move trillions of dollars in a single afternoon. So instead it gets buried in research notes nobody reads except other analysts. Today we're going to pull that thread out into the open. We're going to walk through the data that appears to be converging on a specific stretch of time and more importantly we're going to look at the one mechanism that seems to be masking it. From ordinary view by the time we're done I think you'll understand why is a recession coming was never really the right question to be asking in the first place. Look on this channel we don't do fear for the sake of clicks and we don't do blind optimism either. We try to follow where the incentives actually point. And when you strip the emotional noise out of the headlines a pattern tends to emerge. So here's the argument I want to walk you through today. Three parts. First, the American economy right now is not organically strong. It appears to be propped up almost entirely by one single sector. And that sector has never in modern history carried an entire national economy on its back before. And second, underneath that support beam, the actual foundation ordinary household finances looks like it's showing stress patterns that resemble every downturn going back roughly 70 years. And third, when you line up several independent timing signals that don't normally talk to each other, they don't scatter randomly across the calendar, they cluster. And that cluster points toward a window that very few people on financial television seem willing to say out loud. Let's start with where things actually stand, because you need solid ground before we go into interpretation. As of the middle of 2026, unemployment is sitting somewhere around the low 4% range. GDP growth for the year is tracking below the long-run average. This economy has typically produced over the last two decades. And the trend line is pointed downward, not upward. Meanwhile, and this is the uncomfortable part, inflation is not behaving the way you'd expect in a cooling economy. The Fed's preferred inflation measure looks like it's drifting back in the up toward the high 2, with the core reading already pushing past 3%. That's not simply an economy losing steam. That's an economy losing steam while inflation refuses to fully die down. And that combination is about the most difficult puzzle a central bank can be handed. Because the exact tool used to bring inflation down is the same tool that chokes off growth. Economists sometimes call this condition stall speed. The plane is technically still airborne, but it's lost enough altitude that it has almost no cushion left if turbulence hits. That description is useful, but it's incomplete because stall speed by itself doesn't tell you when the wheels is what appears to be happening across the entire US economy right now. Except this time, the second credit card isn't consumer borrowing. It's something closer to a trillion dollars a year pouring into AI-related infrastructure. Data centers, data centers, advanced chips, power generation to feed all of it. And all of that spending is landing directly in the GDP figures. As investment, right now, by many measures, it's the single largest engine of American economic growth. But here's the thing to really sit with. The deferral engine isn't fixing anything underneath. It's borrowing time from a future where that investment either generates real returns or it doesn't. And if it doesn't, that bill comes due often, bigger than anyone expected. So now let's look at what that engine might be covering up. Layer 1. And this is the evidence that's fairly visible if you know where to look. Total household debt in America has climbed into territory well above pre-pandemic levels with credit card balances hovering near record highs. And tucked inside that figure is a number worth remembering. Roughly 13 cents of every dollar owed on credit cards is now more than 90 days past due. Hold on to that number for a second. The last time credit stress reached anywhere near that level was back in 2011, while the country was still crawling out of the wreckage of the last major financial crisis. In other words, some of today's household stress indicators look uncomfortably similar to the aftermath of the Great Recession. Inside an economy, the headlines keep describing as fundamentally healthy. Layer 2 is the part that doesn't usually make the evening news. Auto loan delinquencies appear to have reached their highest levels on record. Student loan delinquency has climbed past the 10% mark. The worst it's been since pandemic-era payment plazas ended. And when analysts split household debt by income bracket, a distinct pattern shows up. Some call it a K-shaped economy, the upper arm of that K. Higher income households with strong balance sheets and real assets genuinely looks fine. The lower arm is a different story. Quietly cracking, increasingly leaning on credit cards, carrying interest rates, well, above 20%, not for luxuries or vacations, but for groceries and rent. That's not discretionary spending. That's survival borrowing. And survival borrowing has a natural limit. It can't be extended forever. And this matters enormously because consumer spending makes up roughly 70% of America's entire GDP. And the lower half of that K finally runs out of room to borrow, that 70% doesn't slow down smoothly. It tends to drop off a cliff. Layer 3 is the piece that ties the timing together, the yield curve. This is the gap between short-term and long-term government borrowing rates. And it inverted back in 2022, staying inverted for an unusually long stretch. Historically, an inverted yield curve has preceded essentially every US recession, going back to the 1950s. But here's the part that rarely gets explained correctly on financial television. The inversion itself isn't actually the warning signal. The real signal, historically, has been what happens after the curve uninverts, after it steepens back toward normal. That re-steepening process appears to have begun in the final months of 2025. Now do the math with me for a second. So if you take that starting point and add the 12 to 18 month lag that's shown up historically between re-steepening and recession, you land somewhere in a window stretching from late 2026 into the middle of 2027. That's not a random coincidence. Some analysts would call that a clock and one that already started ticking. Layer 4 adds a more numerical dimension to this picture. There's a widely watched leading economic index built from 10 different forward-looking components, specifically designed to anticipate future conditions rather than describe the present. And it has reportedly posted negative 6 month and 12 month growth trends for well over a year now. Historically, sustained negative readings in that index have shown up before every modern US downturn. Typically, with a lag of somewhere between 6 and 12 months before the actual contraction becomes visible in the data. Layer that timing on top of the yield curve math and the window tightens even further. Centering somewhere in the second half of 2026 through the middle of 2027 with the visible headline grabbing downturn potentially confirming itself sometime after that once the official data finally catches up to what's already happening on the ground. Now here's one more piece worth adding. Because it's what makes the AI spending story more than just a footnote to all of this. And there's a structural feature of this current investment wave that should give anyone pause. A meaningful portion of the capital flowing through the largest AI companies appears to be circular. One company invests heavily in another and that second company turns around and spends a large chunk of that money buying is computing capacity back from the first. On paper, this shows up as revenue. It gets counted in GDP but if you strip away the accounting layer and ask a simpler question. How much of this activity represents genuinely new demand coming from outside the system. Versus the same dollars simply cycling back and forth generating a growth reading each time they pass through. The picture gets murkier. We've arguably seen a version of this pattern before in a different infrastructure boom roughly a generation ago. And the lesson from that earlier cycle wasn't that the underlying technology was worthless. It clearly wasn't. And the lesson was that the financing structure built around the technology outran the actual revenue the technology could generate at the time. And when that financing structure cracked it dragged years of investment down with it. Regardless of whether the innovation itself was ultimately real and valuable. To be clear. I'm not telling you the AI story is destined to end exactly the same way. What I'm suggesting is that the deferral engine and the circular financing pattern may be two names for the same underlying risk. And treating them as separate unrelated stories is a mistake. A lot of mainstream coverage currently seems to be making. Here's why this matters. What we might be looking at isn't a fundamentally healthy economy simply carrying a temporary technology boom on top of it. It may be closer to a fragile economy being kept upright by one enormous wave of investment while the load-bearing wall underneath the American consumer is already failing a stress test in slow motion. Now, before going further, I want to give real weight to the other side of this argument. Because a fair analysis means presenting the strongest possible counter case, not a straw man. And the optimists genuinely have solid numbers behind them. Credit card debt, for instance, only makes up a relatively small slice, around 7% of total household debt in America. Liquid assets among higher income households remain historically strong by most measures. Household debt service payments as a share of income averaged across the whole population are still sitting below pre-pandemic norms. A well known recession timing rule, built around how quickly unemployment rises off its load point, currently sits comfortably below the threshold that would typically flag danger. And prediction markets, as of now, are pricing the odds of a 2026 recession in the low double digits. Not exactly a chorus of alarm bells. This is a legitimate case, and it deserves to be taken seriously. It does a reasonably good job describing conditions in 2026. What it may not fully capture is 2027 averages, by their nature, tend to smooth over and hide the K-shape. Underneath them, a recession rule built primarily around unemployment can mis-stress that's currently showing up somewhere else entirely in debt delinquency rather than job losses. Part of the reason the labour market looks stronger than it might otherwise appear is a separate and somewhat overlooked factor. A sharp slowdown in net migration, meaning fewer new workers are entering the labour force to even be counted as unemployed in the first place. Strip that particular distortion out, and the underlying labour market may be somewhat weaker than the headline number suggests. In other words, the optimistic case does a good job describing the surface. It may not fully describe what's happening underneath it. So, here's the part you probably came for. When you layer every timing mechanism together, rather than cherry-picking the one that supports a particular narrative, what window do you actually land on? The yield curve's re-steepening points toward late 2026 through the middle of 2027. The leading economic index's sustained contraction points toward a very similar stretch. Several independent forecasting models, when they construct a downside scenario, in which AI investment gets reassessed once companies start demanding measurable returns, rather than just growth for growth's sake. Land that reassessments specifically around 2027, with employment and growth metrics potentially deteriorating further into 2028. Independent risk analyses, modelling what a broad AI investment slowdown could do to equity markets, to put the potential impact in the tens of trillions of dollars, with the timing attached to that scenario sitting in roughly the same 2027 to 2028 range, four separate methodologies built on four separate data sets by analysts who largely don't coordinate with one another, and they appear to be converging on roughly the same 18-month stretch. It's worth remembering, somewhere between mid-2027 and early 2028, write it down if you want, come back to this later, and see whether the data actually held up. I also want to be direct about geography for a second, because global events have a way of reshaping purely financial stories. This isn't only a domestic American question. The oil price disruption are tied to tensions around the Strait of Hormuz. Earlier this year is a reminder that the inflation side of this equation isn't entirely within America's control. A disrupted shipping lane, a spike in energy costs, and suddenly a stubborn inflation reading stops being just an internal policy problem, and starts acting like a hard floor. The Federal Reserve can't easily cut interest rates beneath. Every model we've walked through so far assumes a roughly stable geopolitical backdrop. Remove that assumption, and the window doesn't necessarily push later, it could just as easily pull earlier. There's one more variable that could genuinely shift this timeline, and I want to be upfront about it rather than present the picture as fixed in stone. The Federal Reserve itself. If it decides to cut interest rates aggressively, and those cuts actually filter down to households through meaningfully cheaper borrowing costs, the runway could extend, potentially pushing any reckoning out toward late 2028 or beyond. But the Fed is walking a genuinely difficult line right now. Inflation remains above target, cut too aggressively, and there's a real risk inflation reignites. Cut too cautiously, and the already fragile stall speed economy loses what little lift it has left. As things stand, rates have been held steady, which suggests the underlying pressure is continuing to build, rather than release. So here are three specific falsifiable predictions worth putting on the record. First, at some point in 2027, it seems reasonably likely that at least one major AI-focused company will announce some kind of pullback or delay in previously committed infrastructure spending, and it will almost certainly be framed publicly as capital discipline, rather than what it may actually represent, which is the deferral engine running low on runway. Second, consumer credit delinquency data, particularly around Autor. Loans and credit cards will likely keep climbing through the back half of 2026, even while headline GDP stays technically positive, because the K-shaped divide will keep the upper half of the economy propping up the overall average. Third, and maybe the most important one, watch for what doesn't happen. Watch to see whether the Fed avoids aggressive rate cuts through the rest of 2026. That inaction, that silence, may itself be the real signal, because it would suggest the inflation side of the internal debate is currently winning, which means growth is the side absorbing the cost. Going forward, there are three things really worth keeping an eye on. First, the pace of the yield curve's re-steepening over the next couple of quarters. If it accelerates faster than expected, the recession window could pull forward into 2026 itself. Second, the delinquency data coming out of household credit reports. A sudden jump rather than a gradual climb would suggest the bottom of the K hit its wall faster than the models. Assumed. Third, the language coming out of quarterly earnings calls from the largest AI infrastructure spenders. Specifically, listen for any mention of return on investment timelines, because in corporate speak, that's often the early language of reassessment already underway. This entire line of thinking rests on a pattern that's shown up again and again throughout economic history. Every boom, every bull run, every extended expansion has had believers right up until the very end who assumed. This time, the support beam was permanent. And historically, the people who lost the most in past downturns weren't usually the ones who never got a warning. More often, they were the ones who got the warning and simply treated it as background noise. So the real question isn't whether some kind of slowdown eventually arrives. Cycles always turn and they always eventually end too. The real question is whether you'll still be describing the deferral engine as genuine. Strength, right up until the moment it runs out of runway. Or weather, watching this, you already understand that borrowed time is never quite the same thing as solid ground. Thank you.