WEBVTT - Why Treasuries Became Risky Again 0:00:00.130 --> 0:00:02.910 Hey, Odd Lots listeners, the Odd Lots tour continues and 0:00:02.970 --> 0:00:04.330 our next stop is in Chicago. 0:00:04.570 --> 0:00:04.930 That's right. 0:00:04.990 --> 0:00:07.500 Joe and I will be at the City Winery Chicago 0:00:07.620 --> 0:00:11.140 on October 15th for a live Odd Lots recording. Tickets 0:00:11.300 --> 0:00:14.720 are on sale now at Bloomberg.com forward slash Odd Lots. 0:00:15.020 --> 0:00:17.460 And of course, a special thank you to Barclays for 0:00:17.520 --> 0:00:19.040 supporting Odd Lots Live. 0:00:19.280 --> 0:00:23.000 So that's October 15th at City Winery in Chicago. Get 0:00:23.060 --> 0:00:23.959 your tickets now. 0:00:26.880 --> 0:00:29.730 Bloomberg Audio Studios. Podcasts. 0:00:30.050 --> 0:00:31.410 Radio. News. 0:00:42.680 --> 0:00:45.620 Hello and welcome to another episode of the Oddbots Podcast. 0:00:45.680 --> 0:00:46.540 I'm Tracy Alloway. 0:00:46.680 --> 0:00:47.860 And I'm Joe Wiesenthal. 0:00:48.220 --> 0:00:51.260 Joe, we're back in New York. Yeah. But the Jackson 0:00:51.300 --> 0:00:52.620 Hole episodes continue. 0:00:53.820 --> 0:00:55.750 That's true. You know what, by the way, oh my 0:00:57.690 --> 0:01:01.350 The 30-year rate, by the way, it's just straight up 0:01:01.410 --> 0:01:04.890 the last few weeks. But it's right around, as of 0:01:04.910 --> 0:01:08.780 the time that we're recording this, it's at 5.592. It 0:01:09.220 --> 0:01:12.380 has now hit the highest level since 2002. Wow. So 0:01:12.800 --> 0:01:14.420 a few weeks ago, we were looking at the yield 0:01:14.480 --> 0:01:16.750 curve and saying all these things. Oh, now highest level 0:01:16.790 --> 0:01:16.990 since 2007. 0:01:16.970 --> 0:01:17.110 Yeah. 0:01:18.940 --> 0:01:21.920 And then it was 2004 was the other year, and 0:01:21.980 --> 0:01:25.000 now we're getting higher since 2002. It really is both 0:01:25.080 --> 0:01:27.100 extraordinary speed and scale. 0:01:27.380 --> 0:01:27.580 Yeah. 0:01:27.670 --> 0:01:30.569 And the other thing that happened since Jackson Hole is 0:01:30.690 --> 0:01:31.630 we had a Fed rate hike. 0:01:31.650 --> 0:01:32.350 We had a Fed rate. 0:01:33.590 --> 0:01:36.770 And yet yields continue to soar. And so there's an 0:01:36.870 --> 0:01:39.550 open question over whether or not that rate hike is 0:01:39.590 --> 0:01:43.000 having its intended effect in terms of dampening down financial 0:01:43.040 --> 0:01:47.810 conditions and things like that. Yeah. Another big thing in 0:01:47.830 --> 0:01:50.340 the background, I feel like I have to list all 0:01:50.400 --> 0:01:54.600 these mega trends, but the other big thing is there's 0:01:54.620 --> 0:01:59.760 been this ongoing debate over Fed's communication style, the new 0:01:59.800 --> 0:02:03.920 communication framework, what does credibility mean when a central bank 0:02:03.980 --> 0:02:07.510 is in an inflation-fighting environment and things like that. We 0:02:07.530 --> 0:02:09.190 have to talk about all of this. Yeah. 0:02:09.610 --> 0:02:14.440 When you said another thing in the backdrop, I was like, 0:02:14.480 --> 0:02:16.560 which of the many things are you going to choose here? 0:02:16.620 --> 0:02:19.579 Because you could have said, oh, you know, this is 0:02:20.060 --> 0:02:23.780 against the backdrop of large deficits, against the backdrop of 0:02:24.120 --> 0:02:28.170 AI spending and the fact that big hyperscalers are spending 0:02:28.250 --> 0:02:30.970 it and borrowing at rates that are like the size 0:02:31.010 --> 0:02:34.650 of like pretty big nations, you could have said, in 0:02:34.669 --> 0:02:39.310 the backdrop of longstanding questions about U.S. global hegemony and 0:02:39.330 --> 0:02:42.470 the link between war and the dollar and so forth, 0:02:42.530 --> 0:02:45.710 et cetera. And so there are many backdrops. There are 0:02:45.790 --> 0:02:46.330 infinite backdrops. 0:02:46.350 --> 0:02:47.950 I could have gone on for a while, but I 0:02:47.990 --> 0:02:49.870 chose not to. No, there's so much going on at 0:02:49.889 --> 0:02:50.250 the moment. 0:02:50.570 --> 0:02:51.950 But it all relates to rates. 0:02:52.110 --> 0:02:54.350 It all relates to rates. And we do, in fact, 0:02:54.490 --> 0:02:57.610 have the perfect guest, someone who can opine and share 0:02:57.650 --> 0:03:00.620 her research on many of these things. We're going to 0:03:00.639 --> 0:03:03.859 be speaking with Carolyn Pfluger. She is an associate professor 0:03:04.020 --> 0:03:07.280 at the University of Chicago Harris School of Public Policy, 0:03:07.400 --> 0:03:10.540 as well as a visiting scholar over at the Chicago Fed. 0:03:11.020 --> 0:03:15.019 And she presented a very famous paper at Jackson Hole 0:03:15.040 --> 0:03:17.520 a couple years ago. We didn't get a chance to 0:03:17.560 --> 0:03:19.760 talk to her when we were in Jackson Hole, but 0:03:19.800 --> 0:03:22.149 we're going to make it up right now. So Carolyn, 0:03:22.250 --> 0:03:23.720 thank you so much for coming on All Thoughts. 0:03:24.000 --> 0:03:25.810 Thank you for having me. It's really great to be here. 0:03:26.450 --> 0:03:30.440 Why don't you give us the sort of, 3,000-foot overview 0:03:30.620 --> 0:03:32.700 of your research. What's your specialty here? 0:03:33.250 --> 0:03:38.330 So most broadly, I'm interested in how the macroeconomy, monetary policy, 0:03:38.430 --> 0:03:42.290 and inflation are reflected in bond markets and are in 0:03:42.350 --> 0:03:45.230 turn shaped by them. I think that's a pretty interesting 0:03:45.270 --> 0:03:46.310 topic because. 0:03:46.410 --> 0:03:46.910 We do too. 0:03:48.110 --> 0:03:51.370 You know, to me, the macroeconomy, inflation, unemployment, those are 0:03:51.430 --> 0:03:54.570 the big questions. They're not easy to answer, but that 0:03:54.610 --> 0:03:57.330 makes it all the more intriguing to me. And of course, 0:03:57.370 --> 0:04:01.270 financial markets are unique in that they're sophisticated, they're forward-looking. 0:04:01.310 --> 0:04:03.050 So a lot of these things get priced in, and 0:04:03.070 --> 0:04:04.430 that's what I find very interesting. 0:04:04.730 --> 0:04:07.910 You said that you think about the way various things 0:04:07.950 --> 0:04:11.340 get reflected in the bond market, but then also how 0:04:11.400 --> 0:04:15.260 the bond market then reflects back onto, quote, the real 0:04:15.320 --> 0:04:18.560 economy or the real world or whatever. That part doesn't 0:04:18.600 --> 0:04:22.020 get discussed as much, it seems like for the most part, people, 0:04:22.279 --> 0:04:24.520 when they think about the link between the real economy 0:04:24.580 --> 0:04:27.380 and the bond market, It seems like they mostly make, 0:04:27.600 --> 0:04:30.020 you know, it goes in one direction. So it's interesting 0:04:30.040 --> 0:04:31.740 to me that you said reflected back. 0:04:32.520 --> 0:04:32.760 Yes. 0:04:32.900 --> 0:04:36.299 So that brings me to my research on the policy 0:04:36.360 --> 0:04:40.520 reaction function. Now, what is a policy reaction function? Just 0:04:40.600 --> 0:04:45.680 as a quick definition. Most simply, that's what markets or 0:04:46.260 --> 0:04:49.610 observers more broadly expect the Fed to do in response 0:04:49.640 --> 0:04:54.830 to economic conditions. So as a numerical example, if forecasters 0:04:54.890 --> 0:04:59.090 expect markets 2% inflation, they might forecast a 4% policy rate. 0:04:59.970 --> 0:05:05.640 If they expect 4% inflation, they might expect a 6% 0:05:05.640 --> 0:05:09.000 policy rate. So that would be a one-for-one adjustment, a 0:05:09.160 --> 0:05:16.180 perceived policy reaction coefficient of one. And that matters. This 0:05:16.240 --> 0:05:21.220 policy reaction function matters for at least two reasons. And 0:05:21.360 --> 0:05:24.260 one of them comes back to your question, this reflection back. 0:05:24.810 --> 0:05:29.010 which is that if the policy reaction function is well understood, 0:05:29.790 --> 0:05:33.390 then that means that as data comes out, unemployment, inflation, 0:05:33.440 --> 0:05:36.020 and so on, the policy rates will move in the 0:05:36.060 --> 0:05:39.260 direction that the Fed intended and thereby help with monetary 0:05:39.300 --> 0:05:42.419 policy transmission or maybe even speed it up before the 0:05:42.460 --> 0:05:43.640 next FOMC meeting. 0:05:43.660 --> 0:05:46.400 So this is the market doing, quote, the Fed's work 0:05:46.520 --> 0:05:46.880 for it. 0:05:47.060 --> 0:05:47.620 Correct. 0:05:47.660 --> 0:05:47.860 Yes. 0:05:47.880 --> 0:05:51.960 But you say only if the reaction function is understood, right? 0:05:52.830 --> 0:05:56.789 Correct. That's why the understanding is important, right? So if 0:05:56.830 --> 0:06:01.490 there is, you know, whatever the intended policy reaction function is, 0:06:01.510 --> 0:06:05.770 you know, if it's well understood, markets will adjust in 0:06:05.810 --> 0:06:09.839 response to inflation news coming out and unemployment news coming 0:06:09.940 --> 0:06:10.660 out and so on. 0:06:11.360 --> 0:06:15.420 How are you actually measuring how the market incorporates its 0:06:15.500 --> 0:06:17.880 future expectations of Fed policy? Because there are a number 0:06:17.920 --> 0:06:20.289 of ways that you could do it, you know, some 0:06:20.380 --> 0:06:22.510 obvious ones, but what are you looking at specifically? 0:06:22.870 --> 0:06:25.770 Yes, that's a great question. I mean, I think the 0:06:25.790 --> 0:06:28.770 main point is that a policy reaction function can and 0:06:28.810 --> 0:06:33.930 should be measured. And what my co-authors and I have done, 0:06:33.950 --> 0:06:37.190 I would think of is actually the most obvious way 0:06:37.250 --> 0:06:39.310 of doing it, which is, I think, always a good 0:06:39.470 --> 0:06:43.040 starting point. So this is joint work with my co-authors, 0:06:43.130 --> 0:06:45.880 Michael Bauer at the San Francisco Fed and Adi Sundaram 0:06:46.010 --> 0:06:49.720 at Harvard Business School. So we took two approaches. The 0:06:49.760 --> 0:06:55.060 first one is based on forecasts. We didn't run our 0:06:55.120 --> 0:06:57.820 own surveys because we wanted to know, you know, how 0:06:57.860 --> 0:07:02.849 have these perceptions changed over time, historically, over decades. And, 0:07:03.110 --> 0:07:06.630 you know, a policy reaction function is quite similar to 0:07:06.770 --> 0:07:11.050 a 1993 type Taylor rule, but there are important differences. 0:07:11.870 --> 0:07:16.760 The first difference is that it's forward-looking. So it's about expectations. 0:07:17.200 --> 0:07:21.000 The second difference is that it's a little bit shorter run. 0:07:21.040 --> 0:07:23.680 It's about the next couple of quarters, at most the 0:07:23.700 --> 0:07:29.220 next couple of years. And traditional monetary policy rules are 0:07:29.600 --> 0:07:33.530 used to describe historical data in a backward-looking manner. What 0:07:33.550 --> 0:07:37.530 we said is, hang on, rather than using this historical data, 0:07:38.330 --> 0:07:44.210 let's just use forecasts, survey data, interest rate forecasts. And 0:07:44.230 --> 0:07:47.130 it turns out that every month, There is a lot 0:07:47.150 --> 0:07:49.970 of richness in terms of the forecasts that are being 0:07:50.010 --> 0:07:53.560 produced for the policy rate. And so we use this 0:07:53.680 --> 0:07:57.400 data from the blue chip of financial forecasters, which asks, 0:07:58.000 --> 0:08:02.000 what is your forecast for the Fed funds rate? And 0:08:02.060 --> 0:08:05.960 what are the underlying assumptions that you used for inflation 0:08:06.430 --> 0:08:09.370 and output to come up with this policy rate forecast? 0:08:10.250 --> 0:08:13.190 So in a way, it's an ideal setup because it's asking, 0:08:13.290 --> 0:08:15.750 what is the rule for the Fed that you're plugging in? 0:08:16.970 --> 0:08:22.630 And so we just took this data that's forward-looking, you know, 0:08:22.630 --> 0:08:25.970 30 to 50 forecasters, a couple of forecast horizons, up 0:08:26.030 --> 0:08:28.970 to six quarters out, and we ran the exact same 0:08:29.010 --> 0:08:33.550 regression that you would run for a monetary policy rule, but, 0:08:33.770 --> 0:08:37.819 you know, on this different data, forward-looking data. So, yeah, 0:08:37.860 --> 0:08:41.420 if you basically treat this, we can't run experiments on 0:08:41.460 --> 0:08:44.660 the macroeconomy, but if you treat this data as the 0:08:44.720 --> 0:08:49.760 best approximation of, experiment on the macroeconomy, we ran a 0:08:49.800 --> 0:08:54.339 regression of the forecasted Fed funds rate onto forecasted inflation 0:08:55.059 --> 0:09:00.700 and forecasted output relative to potential. And so the coefficients 0:09:00.760 --> 0:09:04.750 there on inflation would tell us when forecasters are forecasting 0:09:04.809 --> 0:09:08.410 high inflation, how much higher is the expected Fed funds 0:09:08.450 --> 0:09:12.870 rate and vice versa, for example. So that's the first methodology. 0:09:13.630 --> 0:09:16.980 The second one is closer to what I said about 0:09:18.320 --> 0:09:24.740 the macroeconomic announcement dates and rates moving in response. You 0:09:24.760 --> 0:09:28.000 can use that also to obtain a market perceived rule, 0:09:28.410 --> 0:09:32.010 just to say if inflation comes out higher or lower 0:09:32.050 --> 0:09:36.730 than expected. Let's say inflation comes out higher than expected 0:09:36.890 --> 0:09:40.720 and interest rates, say the two-year rate, goes up a lot. 0:09:41.790 --> 0:09:45.870 then that is also a regression that we run. You 0:09:45.910 --> 0:09:48.790 need a couple of observations, of course. It's a rolling window. 0:09:50.010 --> 0:09:52.170 But you can run this regression of how much do 0:09:52.250 --> 0:09:56.510 yields change onto what's the inflation news on that day. 0:09:57.030 --> 0:10:00.640 And that gives another way of getting a perceived reaction. function. 0:10:01.440 --> 0:10:05.340 It turns out that both of these methodologies often give 0:10:05.380 --> 0:10:08.239 very similar answers. And I love that because it makes 0:10:08.300 --> 0:10:10.380 me a lot more comfortable with either methodology. 0:10:11.270 --> 0:10:15.090 So one thing I'm curious about is, okay, you know, 0:10:15.110 --> 0:10:17.929 you look at market reactions and you get, there's one 0:10:17.990 --> 0:10:20.390 number there, right? There's a lot of different views of 0:10:20.410 --> 0:10:22.609 the market, but it's one number. Whereas if you're doing 0:10:22.650 --> 0:10:26.150 a survey, you're going to get, you can average them together, 0:10:26.190 --> 0:10:29.080 but you can see all the different views. Something I'm 0:10:29.240 --> 0:10:35.120 curious about with the surveys or anything is whether we 0:10:35.220 --> 0:10:39.460 see historical changes over time with respect to, I suppose, 0:10:39.970 --> 0:10:43.490 the spread or the dispersion of views, are there times 0:10:43.610 --> 0:10:46.490 where people in the market have, you know, there's a 0:10:46.610 --> 0:10:52.059 clear consensus of what market participants or forecasters perceive the 0:10:52.220 --> 0:10:55.120 reaction function to be? And are there times that we 0:10:55.160 --> 0:10:58.980 can point to where there is a lot of ambiguity 0:10:59.040 --> 0:11:00.860 on the part of market participants? 0:11:01.380 --> 0:11:07.870 Yes, that's interesting. So we have focused on the variation 0:11:08.070 --> 0:11:11.870 over time in the point estimate of the reaction function. 0:11:12.790 --> 0:11:17.330 There clearly is a lot of heterogeneity across forecasters. We've 0:11:17.350 --> 0:11:19.429 looked at that a little bit, but we've really focused 0:11:19.510 --> 0:11:23.240 on how does it vary over time. But as one example, 0:11:23.280 --> 0:11:29.610 when there was a lot of agreement after Late 2011 0:11:29.610 --> 0:11:31.860 was when the Fed came out and gave very clear 0:11:31.929 --> 0:11:35.120 date-based forward guidance. Interest rates will be at zero at 0:11:35.200 --> 0:11:39.140 least until mid-2013. At that point, all the Fed funds 0:11:39.160 --> 0:11:43.880 rate forecasts collapsed to zero and you'll see absolutely no dispersion. 0:11:43.920 --> 0:11:46.640 So that would be an example where that collapsed. 0:11:47.120 --> 0:11:49.500 So what moves the needle when it comes to the 0:11:50.020 --> 0:11:55.110 market's policy reaction function? Is it the Fed actually doing 0:11:55.170 --> 0:11:59.610 something and raising or lowering rates? Or is it forward guidance? 0:11:59.710 --> 0:12:02.890 Is it having a press conference where they answer a 0:12:02.929 --> 0:12:06.589 question particularly well or badly? What do you observe in 0:12:06.630 --> 0:12:08.310 terms of what actually moves the needle? 0:12:08.809 --> 0:12:12.400 Great question. And that's a question that my co-authors and 0:12:12.559 --> 0:12:15.700 I thought a lot about. I think, you know, just 0:12:15.740 --> 0:12:20.160 having measurement is extremely useful, whichever approach you choose, right? 0:12:20.220 --> 0:12:24.370 So in order, picking a framework or picking a communication strategy, 0:12:24.380 --> 0:12:28.090 evaluating it, I think is very important. So now what 0:12:28.110 --> 0:12:30.890 we looked at in terms of what moves the needle 0:12:31.010 --> 0:12:34.810 on the perceived reaction function, let me take you back 0:12:34.910 --> 0:12:37.410 to 2020. It was not a good time for any 0:12:37.450 --> 0:12:41.410 of us, but so that was the time of the pandemic. 0:12:41.490 --> 0:12:47.620 Initially inflation was extremely low and in 2021 inflation then 0:12:47.679 --> 0:12:51.339 picked up and there were some of these really big 0:12:51.580 --> 0:12:56.240 positive inflation surprises. And that was also the time where 0:12:56.260 --> 0:13:00.040 there was the discussion of, you know, is inflation permanent? 0:13:00.120 --> 0:13:02.460 Is it transitory? The T word. 0:13:02.820 --> 0:13:04.320 We need to get a transitory clock. 0:13:04.340 --> 0:13:05.350 Yeah, yeah. Foreign. 0:13:05.370 --> 0:13:05.750 Yeah. 0:13:06.590 --> 0:13:09.829 The Fed initially just kept interest rates at zero. It 0:13:09.870 --> 0:13:13.290 was also the time when there was this now old 0:13:13.309 --> 0:13:16.010 new framework of average inflation targeting in 2021. 0:13:15.990 --> 0:13:16.150 Yeah. 0:13:20.300 --> 0:13:23.959 So the interest rates stayed at zero for quite a while, 0:13:24.040 --> 0:13:31.540 even though inflation was running at 5%, 6%. And the 0:13:31.590 --> 0:13:35.190 perceived reaction function that we saw at that time was 0:13:35.910 --> 0:13:41.370 extremely flat. So there was a perception that the policy 0:13:41.429 --> 0:13:47.949 rate would stay at zero pretty much irrespective of economic conditions. 0:13:48.110 --> 0:13:52.400 And you saw that, for example... In May 2021, that 0:13:52.460 --> 0:13:57.030 was a particularly positive inflation surprise. That was an inflation 0:13:57.090 --> 0:14:01.610 surprise of almost half a percent. So annualized, the surprise 0:14:01.650 --> 0:14:06.890 component was 6%. That's huge. And the two-year rate didn't move, 0:14:07.350 --> 0:14:10.470 not at all. And you also saw that in the forecasts, 0:14:10.510 --> 0:14:13.750 which were all very, very close to zero. And that 0:14:13.790 --> 0:14:17.089 was true for the forecasters who had higher inflation expectations. 0:14:17.130 --> 0:14:20.010 That was true for the ones who had lower inflation expectations. 0:14:20.440 --> 0:14:22.120 So that's a little bit the nice thing about our 0:14:22.140 --> 0:14:25.080 data set, because, of course, there was a very big 0:14:26.100 --> 0:14:29.160 good ship transitory, as Powell called it at some point 0:14:29.240 --> 0:14:32.870 at the time. A lot of people thought inflation was transitory, 0:14:33.650 --> 0:14:37.490 but not everyone. And by looking at this set of 0:14:37.550 --> 0:14:40.770 different forecasters, we can also see the ones who thought 0:14:40.790 --> 0:14:44.850 that inflation would be higher. And all of them expected 0:14:44.890 --> 0:14:45.770 zero interest rates. 0:14:46.050 --> 0:14:46.200 Hmm. 0:14:46.680 --> 0:14:49.700 And that I think is quite telling about, you know, 0:14:49.720 --> 0:14:54.240 the response that was anticipated at the time. And so then, 0:14:54.280 --> 0:14:57.440 you know, towards the end of 2021, the tone started 0:14:57.460 --> 0:15:01.180 to change and March 2022 was liftoff. At first, just 0:15:01.180 --> 0:15:04.180 25 basis points. And then we saw these really, really 0:15:04.330 --> 0:15:08.690 big interest rate hikes repeatedly. And so after the Fed 0:15:08.750 --> 0:15:11.870 started to act, that's when in our data, we see 0:15:11.930 --> 0:15:16.850 that the perceived inflation response really picks up. It basically 0:15:16.890 --> 0:15:23.670 goes from zero to one between, say, early 2022 towards 0:15:23.740 --> 0:15:27.680 the end of 2023. So there was a very big 0:15:27.820 --> 0:15:33.120 increase in the perceived inflation response, but it occurred late 0:15:33.320 --> 0:15:35.580 and only after the Fed had started to act. So 0:15:35.720 --> 0:15:37.760 we call that the learning from actions channel. 0:15:54.270 --> 0:15:58.560 In a dream country that actually executes perfect monetary policy 0:15:59.000 --> 0:16:03.360 all the time, does it have a perfectly stable reaction 0:16:03.420 --> 0:16:04.560 function in central bank? 0:16:05.640 --> 0:16:09.460 That's a good question. And that goes back to, I think, 0:16:09.500 --> 0:16:13.370 the difference between a monetary policy rule and a reaction function. 0:16:13.830 --> 0:16:17.750 I don't think it has to be stable. It doesn't 0:16:17.790 --> 0:16:21.390 have to be perfectly stable over time. It also goes 0:16:21.450 --> 0:16:26.580 back to this much older discussion about rules versus discretion, right? 0:16:26.630 --> 0:16:30.500 If it was just a mechanical rule, then, you know, 0:16:31.380 --> 0:16:35.460 why are central banks needed at all? So there is, 0:16:35.660 --> 0:16:38.620 you know, conditions change. I don't think it needs to 0:16:38.660 --> 0:16:42.800 be stable. And, you know, the appropriate reaction may be 0:16:42.840 --> 0:16:46.460 different at different points in time. Actually, what we see 0:16:46.500 --> 0:16:50.070 in our data, we do a very simple exploration. We 0:16:50.110 --> 0:16:55.020 have an earlier paper in the Quarterly Journal of Economics, 0:16:55.180 --> 0:16:59.230 also with Michael and Adi, where we look at a 0:16:59.300 --> 0:17:04.609 longer time series of the perceived reaction to output. And 0:17:04.670 --> 0:17:08.050 we see that it varies over time and we try 0:17:08.109 --> 0:17:12.090 to systematize a little bit. When is this perceived reaction 0:17:12.130 --> 0:17:15.570 to output steep? When is it flat? And what we 0:17:15.710 --> 0:17:21.010 find is that it tends to be steeper during tightening cycles. 0:17:21.210 --> 0:17:24.970 So that's when Chairs Bernanke or Yellen would have said, 0:17:25.350 --> 0:17:27.970 we're going to be quite data dependent. And you see 0:17:28.020 --> 0:17:30.800 that in the data in that there is a stronger 0:17:30.859 --> 0:17:36.760 perceived response to output, whereas easings are often sudden and 0:17:37.300 --> 0:17:40.010 then not much else is expected. So that tends to 0:17:40.030 --> 0:17:44.310 be a time when the perceived policy reaction becomes flatter. 0:17:45.130 --> 0:17:50.429 So if the market's perceived reaction function moves the most 0:17:50.590 --> 0:17:54.370 when a central bank is actually doing something, so they're 0:17:54.410 --> 0:17:57.889 taking an action rather than perhaps just talking about it, 0:17:58.490 --> 0:18:01.510 what happens if a central bank wants to be gradual 0:18:02.090 --> 0:18:04.810 in terms of raising rates? Because we all remember 2022, 0:18:04.810 --> 0:18:09.510 the big liftoff, that was a pretty painful time for markets. 0:18:10.310 --> 0:18:14.570 Is there just an accepted trade-off here where maybe... you 0:18:14.609 --> 0:18:17.359 have to show the market that you're serious about inflation 0:18:17.450 --> 0:18:21.230 by actually doing something. And that helps the market change 0:18:21.600 --> 0:18:25.879 its perceived market function or its perceived policy function. And 0:18:25.920 --> 0:18:30.160 then that helps with actually fighting inflation. But the offset 0:18:30.240 --> 0:18:33.639 or the trade-off is that, you know, stocks are going 0:18:33.660 --> 0:18:36.160 to go down or there might be financial stability issues 0:18:36.200 --> 0:18:36.980 or something like that. 0:18:38.220 --> 0:18:42.160 Yes, that's very interesting. That gets me to a different 0:18:43.270 --> 0:18:47.610 branch of my research, which is, um, which is about 0:18:48.590 --> 0:18:51.930 the risks in treasury bond markets. You know, I have, 0:18:52.050 --> 0:18:54.930 I've worked on that, uh, for a long time going 0:18:54.990 --> 0:19:00.240 back and a fact that you may be aware of, 0:19:00.310 --> 0:19:03.430 but many people aren't is that treasury bonds were not 0:19:03.490 --> 0:19:09.410 always safe historically. So there were periods, especially during the seventies, 0:19:09.790 --> 0:19:14.800 eighties and nineties when Treasury bonds were viewed as quite risky. 0:19:14.840 --> 0:19:17.859 That was the period of bond market vigilantes, the inflation 0:19:17.900 --> 0:19:22.020 risk premium. And the way I measure that in the 0:19:22.060 --> 0:19:30.270 data is through the co-movement between bonds and stocks. So pre-2000, 0:19:30.430 --> 0:19:33.389 treasury bonds were actually quite stock-like, if you will, in 0:19:33.410 --> 0:19:37.090 their correlation with the bond market. And that's a recent 0:19:37.250 --> 0:19:42.580 paper with John Campbell and Luis Becerra at Harvard. Now, post-2000, 0:19:42.640 --> 0:19:47.090 treasury bonds were safe in the sense that they had 0:19:47.109 --> 0:19:50.850 a negative correlation with the stock market. And then in 0:19:50.869 --> 0:19:54.010 the most recent period, these types of bond risks have 0:19:54.050 --> 0:19:57.760 gone up again. There are many ways, many reasons to 0:19:57.840 --> 0:20:01.639 care about this type of bond risk. Most simply, if 0:20:01.660 --> 0:20:06.140 you're holding a portfolio that contains bond and stocks, having 0:20:06.180 --> 0:20:10.060 a positive correlation means that there is nowhere to hide. Right. 0:20:10.650 --> 0:20:12.990 Another reason why 2022 was so painful. 0:20:13.369 --> 0:20:16.640 That's right. And I think that is connected to monetary policy. 0:20:16.740 --> 0:20:20.540 So in the research that I've done on that, you know, 0:20:20.560 --> 0:20:23.300 I've asked what has changed in these bond risks? Is 0:20:23.359 --> 0:20:27.540 it nominal? Is it real? Why has it changed? Can 0:20:27.580 --> 0:20:32.100 we attribute it to deeper forces, the nature of shocks, 0:20:32.300 --> 0:20:36.169 luck versus policy? And what does it mean for yields 0:20:36.210 --> 0:20:39.790 going forward? Or not necessarily going forward, but just what 0:20:39.830 --> 0:20:43.630 does it mean for yields? What I've found in this 0:20:43.690 --> 0:20:50.160 research is that the inflation component in nominal bonds plays 0:20:50.200 --> 0:20:55.960 an important role. That's a very natural decomposition. So the 0:20:56.180 --> 0:21:00.920 US Treasury issues regular nominal bonds, which pay $ 100 10 0:21:00.920 --> 0:21:04.070 years in the future. And then it also issues bonds 0:21:04.160 --> 0:21:09.670 that pay $ 100 increased by the inflation index 10 years 0:21:09.710 --> 0:21:11.870 in the future. And that means if you hold these 0:21:12.490 --> 0:21:15.750 inflation index bonds or tips, you don't really need to 0:21:15.830 --> 0:21:19.320 fear inflation. Whereas when you hold nominal bonds, that's of 0:21:19.380 --> 0:21:24.180 course exposed. And so if you look over time, the 0:21:25.580 --> 0:21:30.540 bond risks were so positive in the pre 2000 period. 0:21:30.640 --> 0:21:32.639 And for that, actually, I need to go to UK 0:21:32.660 --> 0:21:35.440 data because they've had inflation linked bonds for much longer. 0:21:36.520 --> 0:21:39.240 But in this UK data, you see that a big 0:21:39.380 --> 0:21:42.800 chunk of these bond risks were on the inflation side. 0:21:43.760 --> 0:21:48.140 And then post 2000, the inflation component became much smaller 0:21:48.200 --> 0:21:51.679 and both the nominal and the real treasury bond risks 0:21:52.160 --> 0:21:55.720 were negative. And then most recently, actually, the increase in 0:21:55.770 --> 0:22:00.730 bond risks has some similarities and some differences compared to 0:22:00.930 --> 0:22:04.470 the 1980s. The similarity is that bonds and stocks have 0:22:04.520 --> 0:22:07.639 moved together often when we see treasury bond yields going up, 0:22:07.780 --> 0:22:12.300 the stock market tanks. But the difference is, A, it's 0:22:12.340 --> 0:22:15.100 not as big. We're not back to the 1980s yet, 0:22:15.200 --> 0:22:18.720 at least according to my data. The other difference is 0:22:18.880 --> 0:22:22.530 that a good chunk of it most recently has been 0:22:23.250 --> 0:22:25.990 in the index bonds. So it's not necessarily about the 0:22:26.030 --> 0:22:30.570 inflation component in the same way. So I think to 0:22:30.630 --> 0:22:35.430 understand the change from the pre-2000 to post-2000, an important 0:22:35.530 --> 0:22:41.290 component is the inflation part. Why is that? Well, in 0:22:41.310 --> 0:22:45.700 the 1980s, stagflation was the thing that everyone was talking about. 0:22:46.240 --> 0:22:48.720 If you have a stagflation and you hold nominal bonds, 0:22:48.840 --> 0:22:52.619 that's terrible because they become worth less. The recession is 0:22:52.660 --> 0:22:55.260 terrible for stocks, so then they move together and bonds 0:22:55.320 --> 0:22:59.939 are risky. Post-2000, recessions tended to be more of the 0:22:59.980 --> 0:23:04.609 demand variety, which means also that recessions tended to be 0:23:04.660 --> 0:23:10.369 lower inflation, and so nominal bonds benefit from low inflation. 0:23:11.530 --> 0:23:15.050 And so then in some of my work with John 0:23:15.090 --> 0:23:18.490 and Luis, we have shown that this can explain why 0:23:18.820 --> 0:23:23.340 treasury bond risks changed around 2000. What the interesting question 0:23:23.380 --> 0:23:26.090 is that still remains is, Is, of course, you know, 0:23:26.150 --> 0:23:27.959 is it luck? You know, is this just if we 0:23:28.200 --> 0:23:30.699 experience supply shocks, is everything going to go back to 0:23:30.720 --> 0:23:34.960 the 1980s? Or is there also a degree of policy 0:23:35.020 --> 0:23:38.790 to it? So I have then, you know, lots of 0:23:38.890 --> 0:23:42.270 models and so on. But, you know, the punchline is 0:23:42.470 --> 0:23:47.290 that it really requires a perfect storm to go back 0:23:47.410 --> 0:23:52.619 to the 1980s risky bond markets. It requires the inflationary shocks. 0:23:53.060 --> 0:23:56.899 Let's say supply shocks. The typical one would be oil 0:23:56.960 --> 0:24:01.290 price shocks, or it could also be fiscal sort of 0:24:01.350 --> 0:24:06.490 lack of credibility, inflation expectations that start moving. So you 0:24:06.530 --> 0:24:09.409 need these type of inflationary shocks and you need a 0:24:09.450 --> 0:24:15.409 Fed that's willing or forced to accept a recession when 0:24:15.450 --> 0:24:18.649 these shocks happen. And I think that has been a 0:24:18.670 --> 0:24:23.590 little bit different in the past five years, uh, from, 0:24:23.770 --> 0:24:26.970 from the 1980s in the sense that, uh, the fed 0:24:27.030 --> 0:24:30.380 has been able to move more gradually. So that's the big, 0:24:30.470 --> 0:24:33.180 that's really, I think of it as two big differences 0:24:33.240 --> 0:24:37.840 that happened around 2000. One is the supply shocks changed 0:24:37.900 --> 0:24:42.679 to demand shocks, but of course, monetary policy also gained 0:24:42.740 --> 0:24:45.060 a lot of credibility. And the way that shows up 0:24:45.220 --> 0:24:48.360 in my data is as a more gradual inertial monetary 0:24:48.400 --> 0:24:52.949 policy rule. And that has the advantage that when supply 0:24:52.990 --> 0:24:56.520 shocks happen, a more gradual rule may be able to 0:24:57.230 --> 0:25:00.200