Why So Many People Got This Year's Economy Wrong

0:00You're being sold an AI future where you're obsolete or irrelevant. That vision is wrong. At Palantir, they're building AI that helps workers and unlocks their full potential. American workers are our nation's greatest strength. AI shouldn't eliminate them. It should elevate them. Palantir is here to tell their stories. From factories to hospitals, AI is freeing people from drudgery, letting them do what humans do best. Create. Solve. Build.

0:36So, let me get this straight. Your company has data here, there, and everywhere. But your AI can't use the data because it's here, there, and everywhere? Seems like something's missing. Every business has unique data. IBM helps your AI access your data wherever it lives to change how you do business. Let's create smarter business, IBM.

1:14Hello and welcome to another episode of the All Thoughts Podcast. I'm Tracy Allaway. And I'm Joe Weisenthal. Joe, it's nearly the end of 2023. It's been a wild ride. What an incredible year. I mean, the stat that just like jumps out to me is 41.24%. As of right now, we're recording this December 14th. That is the annual gain of the NASDAQ, 23.1 % on the S &P, mortgage rates below 7%. Good times. It looks like good times when I look at the screen. Even though you're talking about the rally, it sounds like you haven't escaped the year unscathed because of your voice. It sounds like you've had a rough 2023.

1:56It's been a good 2023, but my voice is not in great shape at this moment. Joe was singing last night at his first show. Yes, that is my excuse that I was out late at a bar singing country music. But here I am, and I'm very excited to talk about the bizarre, weird, unexpected year that 2023 was. Yes. So if you recall this time last year in 2022, it seemed like the consensus going into the year was we were going to have a recession. You know, people were talking about a hard landing, this idea that the Fed was going to have to keep hiking rates, and that that was eventually going to have to bring down employment and we were going to see a slowdown in the economy.

2:38And yet here we are and it hasn't materialized. This time last year, there was just like so much pessimism and the market was really like in a rough shape in November, December last year. There was this view. It's like, you know what? Things are going to things are in rough shape, but it doesn't matter. Inflation is still so high. The Fed has to keep pressing. The Fed has to keep hiking. It was pretty grim. And then somehow like this is the big thing that I think people are going to be talking about for years, which is how did we have like the biggest rate hike cycle ever or one of them? without more slowing in the economic activity?

3:14And how did inflation come down from where it was at its peak in the middle of 2022 without more weakening in the labor market? Yes. And I should note, it wasn't just economists and analysts who were very pessimistic on the economy going into 2023. We had a lot of real world people, for lack of a better term, who also thought that things were going to slow down. For instance, you had, I think, the conference board survey of CEOs, like almost 100 percent, were predicting a recession in the U.S. All the sentiment surveys, as we've been discussing on the show, have been coming. Totally. Very negative.

3:52Well, up until recently, were coming in very negative. So, you know, this wasn't just an economist problem. But I think we should go over the year and we should review, like, what exactly happened that surprised a lot of people? Well, you said this, I think, in our recent episode that we did with Jan Hatsias, And I totally agree. I mean, I feel like the last few years will be one of these periods in economics that people are going to be writing Ph.D. papers on for 50 years. Right. Kind of like the Great, you know, the Great Depression or other periods that are these sort of Holy Grails or Rosetta Stones for understanding how the economy works.

4:28There's going to be so much debate and work and academic research and relitigating debates, et cetera, about like what happened over the last four years. I really hope we still have all thoughts in 2073 and we'll just do episodes on like what happened. My voice will sound better then. I hope so. OK. All right. Why don't we just get to it? We are going to be speaking with Anna Wong. She is the chief U.S. economist for Bloomberg Economics. It's the first time we've ever had her on the show, which is kind of surprising. Anna, thank you so much for coming on All Thoughts. Happy to be here, Tracy. So this time last year, why don't you walk us through what exactly Bloomberg Economics was expecting?

5:09How did you expect this particular year to turn out? Yeah, so a year ago, or even more like a year and a half ago, we first put out our recession model. Because at the time, the Fed started hiking rates. And there was just a lot of curiosity about how this would end. And our goal then was to have a more precise timing of the probability of recession as opposed to just giving a vague, what is the 12th month ahead of recession? And there's a cottage industry of these models out there. And so our model was to put out some kind of numbers on each month, probability in July, August. And over that period, since this model first came out in spring of 2022.

5:57And at that time, the model is actually foreseeing a recession, a high probability of recession, only starting towards the end of 2023 and even in early 2024. And over the course of last year and this year, that model evolves in terms of the timing of when that trigger is pushed. And all of the calls of that model has been that the recession would begin in the second half of 2023. And so coming into this year, we thought that the recession that is so widely expected would be towards the end of 2023. Backing up for a second, what goes into a recession model? What does that even mean? Like, how do you build or construct a model and something like that?

6:45Yeah, there is a variety of models, right? So, for example, our model, we took 13 indicators, typical indicators that had some track record in identifying recessions in the past. And many of them overlap with the LEIs from conference support. So it's yield curve spread sentiment models. And so any models that have those two things, yield curve spreads and sentiment, would tell you there's a high probability of recession. Even the LEI has over 90 % probability of recession. But another type of models that one uses is just thinking about the probability that NBER would date a recession. And NBER told the public that they usually look at six monthly indicators.

7:33And so one could perceivably be also building a model based on that. But of course, when we make a call for a recession, that is only a very small part of all our inputs. We tend to look at things in Bloomberg economics. In my team, we tend to look at things in three ways. We approach a question in three ways. And if those three ways all say the same thing, then we make the call. And so for a recession call last year, we also what really influenced our view is the range of other theoretical models. So the model recession probability models I just described are just empirical models, which has no theoretical frameworks, right?

8:16But economists, of course, in their tool set, we have general equilibrium models. We have large scales model. We have state-of-the-art models. And so we used a model on the terminal, which is called SHOCK, which mimics— Yeah, I was looking at this earlier. It's pretty cool, actually. What's it called? SHOCK. SHOCK. go on the terminal. Right. And that model mimics the Fed's own in-house general collaborate model, FURBIS. And that model would suggest that the lags of monetary rate hikes, at least on labor market, should be about 18 to 24 months, which is about the standard of what has been found in economic literature.

9:00And then another state-of-the-art model we used is a model that central bankers have been discussing a lot last year. This is based on a cutting-edge economic paper, and that was the paper that tipped a lot of central bankers off into thinking about shorter legs of monetary policy. Which paper was this? It's a paper by Bauer and Swanson, and that was the paper that found that, in fact, the legs of monetary policy are much shorter. So we also looked at that model, And what those models found, especially that Bauer and Swanson, the very cutting edge model, is that, yes, it's true that the lags of monetary policy are shorter.

9:44For example, for industrial production, we already have seen IP decline for over a year. And in fact, the decline of industrial production is almost like exactly matched the contour of that model. And so that model also says that inflation now responds faster to Fed's rate hike than, you know, back in the times of Milton Friedman. But the one area which the model says that two areas actually that says that the lags of freight hikes still have yet to really hit the peak is labor market and also credit market. And I think those two are precisely the area where we have not seen much adjustment. And that is why we don't have most people at least don't think we have a recession yet this year.

10:37That's really interesting, especially putting my former credit market reporter hat on the credit side of it. And I do have a pet theory right now that I think we've talked about, which is that the sheer size of the private credit market might be making a difference here. Like if you have this bundle of money that actually doesn't seem to be that rate sensitive in the current environment, maybe it's propping up parts of the market. But talk to us about why the credit market might not be as efficient at transmitting rate hikes as it once was. Yeah, I think this might be the surprise. The surprises in the credit market might be a surprise of 2024, which is, well, I think my pet theory of why credit market hasn't adjusted yet in 2023 is that corporates have locked in low interest rates.

11:29Right. Everybody knows that. And also on the household side, household also had wonderful balance sheets during the pandemic. Many households paid down the debt. So deleveraged during the pandemic. But at the same time, I think this is the following areas where I don't think the market understands very well. For households balance sheets, how really how accurate are the credit scores being reflected? So I think that the credit forbearance, a lot of the debt forbearance during the pandemic had distorted the credit scores, inflated credit scores. And there are some studies that estimate that perhaps by as much as even 50 basis points.

12:15So a lot of the, you know, what currently looks like to be near prime are actually subprime. And some prime could be actually near prime. And then looking at auto delinquencies, which has risen to, you know, the level of 2010, right? And you look at who are the ones who are defaulting. They're the ones who bought cars in 2021 and 2022 when car prices were extremely high. And they are also the ones who are having a supreme and near prime credit ratings. So the question, I think, in 2024 is how many of these borrowers who have leveraged up in the past two years are in fact the credit rating, the good credit quality that they looked to be like?

13:09And when the moment that more defaults happen as prices slow, when inflation slows, what also happens is income slows, wage growth slow. And if interest rate doesn't fall as fast, so suppose that the Fed ultimately do hold higher for longer, whereas income and prices are coming down, that would mean that there would be more delinquencies. So when that moment happened, whether there will be a credit crunch versus just a normal, gradual credit slowdown depends on how the lenders are seeing the information, right? This is the famous adverse selection issue. It's like if there's a portion of people whose credit scores don't appropriately reflect their true behavior, can lenders seize who are the bad seeds, who are not?

14:03And this is like a famous asymmetry in the used car market, right? This is why it's very hard. The market for lemons, right? Right, exactly.

14:28Silicon Valley is selling you a future where you're obsolete, or worse, identical. At Palantir, they're witnessing something different and revolutionary. From re-industrializing the nation's defense base, to shipyard workers building faster, and frontline workers boosting productivity, AI is transforming work across the nation. AI is not replacing American workers or flattening them into conformity. It's unleashing what makes each one irreplaceable, their judgment, their craft, their creativity. When American workers become more powerfully themselves, they own the future. Palantir, making Americans irreplaceable.

15:13So you're telling me that the AI that's meant to make everyone's job easier to manage just adds more to manage on top of the thousands of apps the IT department already manages. Funny how that works. Any business can add AI. IBM helps you scale and manage AI to change how you do business. Let's create smarter business. IBM.

15:40What is the takeaway from the realized disinflation that we've seen? I think at one point, CPI was around 9%. Now we're basically, you can argue that in recent months, we are, by some measures, at the Fed's target, and yet unemployment is at 3.7%. And that was a set of conditions that very few people would have anticipated was even possible, in part because the standard story is, well, you need to reduce demand to ease prices. And the way you ease demand is by people, unfortunately, having to lose their jobs. What have we learned about at least what we've seen so far in this cycle? Yeah, Joe, so I would say one will have to acknowledge that it's going better than what everybody thought.

16:25At first, that it will have to be extremely painless. But at the same time, I think, as Powell said yesterday, it's too soon to declare victory on slaying inflation. And here's why. So based on various model decomposition of the drivers of inflation over the last two years, our assessment is that half of it is driven by supply and about half of it is driven by demand. But of course, it's the mixture of those two high demand while supply is hurt, which led to this explosive inflation. Right. And in terms of disinflation, we saw this year earlier this year when SVB collapsed at that time, the CPI recall it was actually falling and everybody thought inflation is not a problem.

17:16But in fact, the super core, which is Powell's preferred measure, which captures the labor intensive part of inflation, was not slowing at all. But then moving into the second half of this year, we start to see a lot of disinflation. And it actually started in June of this year. And this is where I said that, you know, the lags of monetary policy on labor market is about 18 to 24 months. So that slowdown in wage growth actually hit right about the time that all these models would suggest. And so we sort of see movement in that. But still, core inflation is still around 4 % during the summer. On the second day of the December FOMC meeting, the FOMC received a very critical data point.

18:11It is the November's PPI. And so Fed staff usually can pull together a PCE inflation number very quickly the moment they have both CPI and PPI numbers in their hand. So on the second day of December FOMC meeting, that PPI number came in to suggest that actually core PCE for November is going to be very low. We estimate that it's only 0.04%, possibly, so round to zero in November. And this would mean that the six-month annualized core PCE, this is the measure that Chris Waller and a lot of Fed officials are looking and gauging the momentum of inflation, would come in at 2.0 likely in November, right at the Fed's target.

19:08So the FOMC has that data point on the second day, in the morning of the December FOMC meeting. And that explained why the summary of economic projections see downward revision of core PCE to only 3.2 % for end of 2023. Now, how do you make of this significant drop in core PCE, right? And so I would say that when this number ultimately is publicly released late in December, I think it would spark a big debate. On one side, a lot of people would say the Fed is already at target. They should be cutting rates in January even. But the second group of people, and I would be in that second group of people, would be saying that, but a lot of the disinflation in November is actually in categories that's exogenous to the Fed's rate hike.

20:06In fact, it's due to China. If you look at the downside surprises, it's actually all in categories that with high China import content. So they're in apparels, clothing, furnishings, and those actually drove almost all of the downward surprises. So what I think is happening, and this gives me some memory of what happened in 2014 and 2015, was you have global growth slow down and started by China. And then that led to commodity prices decline. And also that also when global growth slowed down occasionally, that could also lead to OPEC having trouble keeping a discipline with an OPEC. And that leads to a race to the bottom.

20:52On top of that, you have U.S. shale who's pumping suddenly a lot more. That actually was the dynamics in 2015 and 2014 that led to that collapse in oil prices. So when you have this China slowdown and what's going on with oil prices, you actually could lead to this disinflation that we're seeing right now. But the implication is also that for the part that the Fed has been focusing on, super core services, that actually doesn't look as great. So services inflation actually picked up a little bit in November in the core PCE. So I think for the Fed to declare victory too soon would be a mistake.

21:37And just looking at our outlook out to 2024, we do see the six-month annualized core PCE dipping in the first half of 2024, dropping to even 2.2 % in March, and then stabilizing at about 2.7 % in the second half of the year. And that would be that last mile of inflation, because the Fed should not be happy about 2.7 or 2.8 percent inflation. I definitely want to discuss the outlook a little bit more. But before we do, I'm glad you mentioned these exogenous factors, these exogenous price declines, because there is, you know, 2023 has turned out to be better than a lot of people expected. But there is now this vibrant debate about how much of that can be attributed to the rate hikes and, by extension, the Fed.

22:32So I guess my question is, how do you think Fed tightening has actually worked through this economy? And how much of what we've seen so far is due to monetary policy versus perhaps normalization of things like supply chains? Yeah, good question, Tracy. You know, I think the way that monetary policy has worked this year is largely as the models expected. As I was saying, if you use these state-of-the-art models, you would see industrial production has declined exactly according to that contour that the model would describe. Inflation as well. and the places where which has not been behaving as models would describe would be the credit market and labor market but even so the labor market in fact is moving in the direction that the model would describe so as you know with 18 to 24 month lag of monetary policy on labor market We should be seeing a more clear slowdown in jobs, job growth in the second half of this year.

23:44And I think we did see that. And I will also have to add that the strong non-farm payroll data over the year is likely to be overestimating the strength of the labor market. That in fact, about 40 % of the 3 million non-farm payroll gains is due to a BLS birth and death models. So if you think about whether this makes sense, so, you know, this is a year where bankruptcies has risen to the level of 2010, right? At the same time, small business are complaining that it's been very hard for them to hire. So if that's the case, how could it be that new firms could contribute to 40 % of the 3 million job gains this year?

24:29So I think that, in fact, after all the revisions are done, which will take a year or two more, then it will be clear that, in fact, in 2023, the job market did slow down significantly in the second half of this year. What do you see in the data that makes you thinking that the strength of the labor market is overstated? Because, I mean, if small businesses are still to this day, and like I said, we're recording this December 14th. Earlier in the week, we got the NFIB survey, which said that finding labor is still a challenge for many companies. It was like in the first one or two paragraphs of the report.

25:11That sounds like a robust, tight labor market. So what are you seeing in the data that makes you think that there is this deceleration going on? Yeah, several things. So I think the difference between our views, our assessment of the labor market and other soft landing, really staunch soft landers view of the labor market really differs only in how much weight we place in different labor market indicators. For example, job openings. So that has been very high throughout this year. And in fact, for most of this year, it was still over 1.8 vacancies for every unemployed. But we put very little weight in that data because, first of all, a lot of HR recruiters have been fired over the turn of last year.

26:06And so logically, you would ask yourself, if so, if most of the layoffs late in 2022 and early 2022 is in the recruitment industry, then who are the people, you know, looking for jobs, recruiting people? And second, we do rely on anecdotes. And in turning points of an economy, anecdotes are extremely useful because they don't get revised. And the Fed also relies a lot on Facebook and anecdotes during turning points as well. But so we thought that the jolts was just overstating. But what gives us more confidence are the price measures, so wage growth. And throughout the year, we have seen wage growth measures coming down softer and softer, even as you see these headline numbers being very strong in terms of payroll gains and job openings.

27:02And I do believe that price measures are better collected and less susceptible to revisions because you just collect a price data, right? whereas with counting the number of jobs you need to you need to consider are you appropriately taking account of all the failed firms because firms who are going bankrupt would not be answering surveys of how many people they hired and uh um so so you know so that's why we we put a lot more um focus on price measures which suggest to us that in fact the labor market is softening more And also we look at a range of recession rules that are that are based on unemployment.

27:51And so, yeah, Bloomberg economics has its own recession rule. I didn't realize that. Yes. So a couple weeks ago, I have a piece with Bill Dudley and we consider 28 recession rules that are based on unemployment. And the idea is that unemployment is a very good indicator because, number one, it doesn't really get revised. And number two, it also unemployment, a job is the most important variable in the economy that determines income consumption saving patterns. So that's why I would focus on unemployment. And so an unemployment rate is just the inflows of people into the unemployed state minus the people escaping the unemployed state divided by the labor force.

28:37That is the unemployment rate. But even within unemployed states, there is a lot of different categories. The most commonly used unemployment rate is the U3 rate, the rate that we know about. But there are also U1, which measures the number of people who have been unemployed for 15 months or longer. There's also U2 rate, which measures people who are laid off and who are temporary workers who finish their temporary stint. And so our 28 rules basically look at these three unemployment rates as well as just flows, inflows and outflows. And in fact, back in January 2008, Janet Yellen was discussing the state of the labor market in the FOMC meeting then.

29:28and she was the San Francisco Fed president then. And she talked about unemployment flows. And so when you see a lot more people flowing into the unemployed state but having a hard time getting out of it, that is how you get a swelling of the unemployment rate. And you don't necessarily need layoffs to get a higher flow of unemployed, right? It could be re-entrance into the labor market or new entrance into the labor market. And in fact, in the 1990, 2001, 2007 recession, the initial increase in unemployment rate is actually due to entrance, new entrance and re-entrance, not layoffs. So the most accurate rule that we have found is the unemployment inflows and outflows indicator that whenever the six month moving average of unemployed inflows exceed outflows is when you are about two months after a recession began.

30:30And I think the intuition there is just that usually the beginning of a recession began with people just finding it harder to find a job, not because there's layoffs, but because there are less people quitting, less turnover. So it's harder to find a job. And only after this, you know, the stagnant state goes on for a while in labor market when firms decided because of the low attrition, they have to lay off people. And that's where you get all the layoffs. That's the increase in U2 rate. So based on this rule, it suggests that we are likely already in this state of downturn. And that is why we think that, you know, a year from now, or even a year and a half from now, if NBER were to time the beginning of a recession, I think October could be a candidate.