Everyone who’s anyone
Last week I went to the US to give a speech to a group of institutional investors gathered in Washington for the semi-annual World Bank-International Monetary Fund (IMF) meetings and to visit some institutional clients along the East Coast. The Bank-Fund meetings are a great opportunity to gauge both markets’ and policymakers’ center of gravity because everyone who is anyone is there. While meeting various contacts at a favored café cum office across from the White House, I watched as the great and good of markets and policy – even the odd Nobel laureate – strolled past or stopped for a hot drink and snack. A few even stopped by to chat.
Orienting to the consensus
Since I no longer work on a trading floor and my clients are not an unbiased sample of market views (see below), I find it harder to judge where the consensus lies. That is a problem when you are an unintentional contrarian as I often am. Sometimes, things that appear obvious to me – surely everyone sees that, right? – aren’t; more often, the consensus converges towards my views but at a rate that I can’t clearly observe. Hence, the IMF’s outlook – the definition of policymaker consensus – and the views of market participants descending on the Bank-Fund meetings help me to orient my own views relative to consensus. Here is what I learned last week from clients, prospective clients, former colleagues, and policymakers.
Even when we’re wrong, we’re right
Let’s start with the Fund’s semi-annual forecasts from the World Economic Outlook (WEO). The Fund revised up its projections for global growth from the April WEO and even from their July update to 3.2% and 3.1% in 2025 and 2026, respectively. The global upgrade was due almost entirely to improved outlooks for the two largest economies, the US and China, with US revisions offsetting downgrades to other advanced economies. But lest you think forecast revisions were due to excessive pessimism on the effects of US tariffs, the Fund’s chief economist, Pierre-Olivier Gourinchas, tells us that conclusion is “both premature and incorrect.” Bad things are still in the pipeline and there were offsetting factors this year. The Fund forecasts a sharp slowing of US growth in the second half of the year, but despite quite downcast rhetoric, even 2026 forecasts were revised higher.
Premature extrapolation
The Fund suggests that it is premature to discount negative tariff effects because their passthrough to consumers has been less than they expected and because tariffs’ long-term negative effects on productivity growth are still to come. Yet, it’s hard not to see both as post-hoc rationalizations. Having wrongly assumed greater immediate passthrough of tariffs, without clear justification, they now assume full passthrough will manifest in the second half of 2025, rather than be diffused over a longer horizon with less pronounced effects. They also fail to explain how their long-run productivity assumptions led to short-run forecast errors. Worse still, even ignoring the reasons I’ve given for why long-run productivity effects are likely to be positive rather than negative, it’s hard to take their longer-term forecasts seriously when they don’t even acknowledge the blistering pace of US capex and its acceleration, much less explain how that is consistent with falling productivity and firms mired in uncertainty.
No one saw that coming!
Similarly, their contention that offsetting positive factors are responsible for better growth makes little sense since all the ones they list were known beforethe April WEO. Among those they list were accommodative financial conditions, the artificial intelligence (AI) investment boom, Chinese exports’ front-running of tariffs, and European fiscal expansion. But financial conditions were arguably easier before their April forecasts, the AI boom has been ongoing for two years, and strong Chinese exports and Germany’s €500 billion fiscal expansion were known in March. Yet their most bizarre claim of changed circumstances is that the (well anticipated) US immigration crackdown impaired labor supply and amid a slowing in labor demand. How either – the slowing of labor supply or labor demand – would raise US growth above their forecasts remains a mystery for the reader to solve. (And of course Thematic Markets’ readers know that it’s a labor supply, not demand story.)
Our model says so
Amazingly, they even cite their own forecasts as evidence that tariffs will have a more negative effect than so far observed! Stretching credulity even further, those forecasts – just a week old – appear to ignore the data in hand. If the latest “nowcast” of third-quarter US GDP growth from the Atlanta Fed of 3.9% annualized growth is correct, US growth in the first three quarters will have averaged 2.4% and incoming data – before the shutdown – indicated the US economy is accelerating. Yet the Fund expects 2025 growth of only 2.0% in the US, which would require a sharp fourth-quarter deceleration to just 1% annualized or less, to be followed by an equally sharp reacceleration to meet their upwardly revised forecast of 2.1% in 2026! Confused? The IMF – like many other policy institutions whose views it represents – appears strongly resistant to admitting error, particularly when it would go against their policy recommendations.
Strong priors
In Bayesian statistical analysis one updates a pre-existing hypothesis with incoming data. Statisticians call the initial hypothesis a “prior,” i.e. beforeobserving data. If new observations validate the prior, it becomes more strongly held; if they contradict the prior, it becomes weaker or if strongly contradicted by the data one might even dismiss it. Priors can be based on opinions, models, earlier studies, or any combination of these. The IMF, most policy institutions and most economists relied on studies of single product or single country tariffs in very different circumstances to come to their priors that US tariffs would be more disruptive and negative for growth. Those priors likely were reinforced by political antipathy to the Trump Administration’s goals and methods. Hence, they became what a statistician would call a “strong prior.” This makes it difficult for them to adjust their view even as data contradict it.
Not alone
They’re not alone. I had quite strong priors pointing in the opposite direction based on my Localization thesis, analysis of the effects of Chinese industrial policy on US productivity, estimates of tariffs’ direct effects, and different interpretation of the studies on which other economists relied for their views. The difference is that incoming data have fully aligned with my expectations: strong, Localization-driven capex momentum has been reinforced by the Trump Administration policies, while the drag on consumption was less than others’ estimates would suggest and more than fully offset by tax cuts. Hence, my priors have strengthened that economic growth is both robust and accelerating, and that inflation will be a bigger problem than markets or institutions like the IMF and the Fed seem willing to admit.
The Fed’s priors and instincts
I’ve written extensively on what I perceive as the Fed’s policy errors based on their erroneous but strong priors around neutral real interest rates, the US economy’s trend rate of growth, and the stability of inflation expectations, all of which I would argue have been strongly rejected by the data both throughout the post-Covid period and in recent months. They also have shown a strong inclination to avoid surprising markets, hence the absence of pushback on market pricing of another cut at next week’s meeting suggests that it is almost baked in the cake. My only caveat to that is this week’s CPI report, which I expect to show continued acceleration. If that manifests, be on the lookout for a Nick Timiraos article in the Wall Street Journal next week walking back markets’ expectations for a cut during the Fed’s blackout window.
Policy osmosis
Unlike most of the people descending on Washington during the Bank-Fund meetings, I generally avoid meetings with policymakers. Over the last two decades I’ve found that my biggest forecast errors came from placing too much weight on their (incorrect) views. Yet it is impossible not to osmose inferences about policy amid the Bank-Fund meetings where everyone else ismeeting them.
Missing labor analysis
From those, I inferred that none of the policy institutions in Washington have done as detailed analysis of US labor markets, and in particular the effects of “solopreneurship” on employment growth, as I have. Hence, they still overestimate the role of labor demand in the slowing of nonfarm payroll growth, impairing the impact of strong real economic activity on their priors. Looking beyond next week’s meeting, that implies that the Fed’s erroneous views on neutral policy rates and the US economy’s potential growth rate may take a while to correct and that inflation will continue to be the sole restraint on their policy path.
National defense primacy
Another interesting inference that I inferred from discussions was confirmation of the primacy of national defense in determining the Trump Administration’s economic policies. This has been one of my core contentions since the start of the Trump Administration and has been critical to my forecasts of its policies and likely effects. A variety of sources confirmed what can increasingly be gleaned from the news: Treasury Secretary Scott Bessent is deeply involved in national security and defense policy if not at its center.11
Two axes of beliefs
Which brings me to perhaps the most important orientation of my views: vis-à-vis market consensus. Throughout my discussions in Washington and elsewhere, I noticed that the degree of agreement or disagreement with my views ran along two axes: closeness to my research and whether one’s investment performance was benchmarked. Clients with interactive subscriptions who regularly engage with me were closest to my views, while less-engaged or read-only subscribers were less aligned, and those unfamiliar with my work were most likely to disagree. That shouldn’t be surprising but leaves open the question of why: Are aligned investors more likely to subscribe to my research? Is my research especially persuasive? Or are my clients the “smart money” in markets? Regardless, this explains why it is harder for me to gauge consensus unless I engage non-clients.
Tethered to consensus
The other axis of agreement-disagreement was a bit more surprising and I only noticed in retrospect. While not a perfect correlation, I perceived that portfolio managers from institutions that are benchmarked, i.e. long-only investment managers whose performance is judged versus a financial index (like the S&P 500), were less likely to accept my contrarian views. In contrast, absolute return managers from hedge funds or family offices appeared more likely to find agreement in some or most of my views. I’m hesitant to make too much of this, but it suggests that benchmarking may tether managers’ views to the consensus.
Politics, politics, politics…
It was also interesting where disagreement was strongest among those who disagreed: geopolitics and the primacy of national security in the Trump Administration’s policies. My economic views received far less pushback. Even my contention that President Trump’s incentives with regards to the Fed are likely to flip next year when he appoints a new chairman were more accepted than my expectations for geopolitical alignment and their likely effects. Upon reflection, that shouldn’t be surprising. Strong economic data are facts that clearly align with my forecasts. Interpretations of policy actions and intent are less subject to falsification by data that might “update” one’s priors, and with a polarizing figure like Donald Trump, priors are more likely to be strongly held based on one’s political preferences.
Comments are available to paid subscribers only.