Archive for January 2023


To QE or not to QE

Over at Mercatus, Andrew Levin and Bill Nelson have an excellent essay on the fiscal cost of the Fed’s QE4 program. There isn’t much that I disagree with, but I’d frame a few of the issues slightly differently. The following reflects my own views, which largely but not entirely overlap with the views of Levin and Nelson:

1. According to Levin and Nelson, the Fed’s QE4 program is ultimately expected to cost taxpayers about $800 billion. This reflects the fact that the bonds that were purchased have sharply declined in value due to rising interest rates.

2. These losses cannot easily be avoided by holding the bonds to maturity and refusing to pay interest on bank reserves. Doing so would cause inflation to explode. The loss is real.

3. However, there is a sense in which these losses are illusionary. The Fed is part of the consolidated balance sheet of the federal government, and the Fed’s holdings of T-bonds are a liability of the Treasury. The Treasury gains when T-bond prices decline. So in one sense the gains and losses net out to zero. Even the fact that some of the bonds are MBSs doesn’t really change that fact.

4. But in a counterfactual sense, the $800 billion loss is real. If the Fed had not purchased these bonds, the Treasury would have profited handsomely from rising interest rates reducing the market value of its liabilities. The Fed took away that profit, and thus effectively cost the Treasury about $800 billion.

5. This loss is similar to the loss that would have been incurred if the Fed had not done QE, but Treasury officials had decided to issue lots of T-bills instead of long-term T-bonds right before a period of sharply rising interest rates. Ultimately, it’s a loss from choosing the wrong maturity structure for federal liabilities—too much short-term debt (or bank reserves) during a period of rising rates.

6. If markets are efficient, then the expected profit or loss from QE is rather small, ex ante. But if the central bank has inside information on its future policy, it may be able to beat the market. Surprisingly, I don’t believe the Fed does have inside information on future Fed policy.

7. The Fed profited somewhat from the first three QE programs as rates remained lower than expected, and lost much more from QE4 as rates rose more than expected.

8. The impact of QE on the fiscal position of the government should also account for changes in the macroeconomy that result from QE. This is an exceedingly complex issue.

9. Two policy reforms would allow the Fed to achieve its macroeconomic objectives with a much smaller balance sheet, and hence much less risk of large losses on its bond portfolio. One reform would be to return regulation to the pre-2008 system, where commercial banks held very small reserve balances. Much of the recent growth in reserve balances has been driven by regulatory decisions of dubious value. Second, a regime of NGDP level targeting would lead to a higher equilibrium nominal interest rate during recessions, and reduce the amount of time that the economy is stuck at the zero lower bound.

10. If policy remains dysfunctional, then QE may be the lesser of evils. It’s better to take on some financial risk and stabilize the macroeconomy, rather than avoid risk and allow deep recessions. The fiscal cost of recessions is much larger than the risk associated with plausible QE programs.

11. However, the actual QE programs have not been optimal. The first three QE programs were too small; in retrospect the policy should have been more expansionary. The final program (QE4) was associated with a monetary policy that was far too expansionary. Thus QE4 was not a “price worth paying” for sound monetary policy, it was a costly program that made monetary policy even more unsound. My own view is that the initial purchases (in 2020) were appropriate, but the program should have been unwound in 2021, before the sharp increase in interest rates. This would have resulted in much smaller losses.

12. Where QE actually is necessary to achieve macro goals, it is worth doing despite the financial risks.

13. If Congress plans to investigate these losses, it needs to consider a wide range of issues. How does Congress feel about giving the Treasury wide discretion over the maturity structure of federal debt? These decisions have huge fiscal consequences and are extremely risky. How does Congress feel about regulators pressuring banks to hold massive reserve balances, even though alternative liquid assets such as T-bills are equally effective at reducing default risk? The inquiry should not merely focus on the Fed’s decision to do QE4, there are much more important questions at stake.

14. Congress needs to also think about how it views a large Fed balance sheet. My own work suggests that in the long run the size of central bank balance sheets (as a share of GDP) is strongly (negatively) correlated with the trend rate of growth in NGDP. Countries with low trend rates of NGDP growth (Japan, Switzerland, etc.) tend to have very large central bank balance sheets. Countries with relatively high trend rates of NGDP growth (Australia) tend to have small central bank balance sheets. This reflects the fact that NGDP growth drives the equilibrium nominal interest rate, and demand for base money rises sharply at near zero rates. This is the “socialism or inflation” dilemma that most conservatives refuse to address.

15. I favor abandoning the experiment of paying interest on bank reserves, which has resulted in much larger than necessary central bank balance sheets, needlessly exposing central banks to the risk of large financial losses. Go back to the pre-2008 corridor system.

Affluent educated Asians in America

Life expectancy in Hong Kong is 85.3, which is the highest in the world. The US comes in at 79.1, which is well down the list.

But one group within the US exceeds even Hong Kong residents in life expectancy. Asian-Americans have a life expectancy of 85.7. Of course there are many types of Asian Americans, and I suspect that life expectancy within that group varies by class. More specifically, I suspect that highly educated Asian Americans have a life expectancy that is absolutely off the charts. (Let me know if anyone has data on that question.)

I recently ran across the following graph in the Orange County Register:

Orange County doesn’t have any poor towns (Laguna Woods is a retirement village), but it is split between upper middle class and working class towns. Focus for a moment on the three biggest towns, affluent Irvine and working class Anaheim and Santa Ana. Notice that the Covid death rate is 6 times higher in the two working class towns. That’s a lot, especially, considering that the age profile is fairly similar. And while the income difference is substantial, it’s not exactly gigantic. So what’s going on?

When I first moved to the OC, I was struck by the fact that Irvine seemed like a sort of paradise. It’s a rather boring place, but everything there is kind of perfect. Irvine also stands out in that it’s roughly 50% Asian, an unusually high percentage for such a large suburb. I suspect that its demographics have something to do with its unusually low Covid death rate (quite similar to the low Covid death rates observed in . . . you guessed it . . . East Asia.)

On the other hand, Westminister is also roughly 50% Asian, and has a relatively high Covid death rate, similar to Santa Ana and Anaheim, which are heavily Hispanic. BTW, the life expectancy of America’s Hispanics is higher than for whites.

So what’s going on here? Westminister’s Asians are mostly Vietnamese while Irvine’s tend to be Chinese, but can this actually explain such a large difference? I suspect that class plays a role. Irvine’s Asians tend to be highly educated professionals, whereas Westminister is more working class and small business.

This website suggests that Asian Americans living in states like Massachusetts and New Jersey have an unusually high life expectancy, roughly 89 years. Asians in those states tend to work in fields like science and finance. The estimated life expectancy of Asian females in Massachusetts (91.6) is the highest I’ve ever seen for any group. This estimate may be biased due to various factors, but it’s also likely measuring something real. (Full disclosure, my Chinese-American wife worked in Boston’s biotech industry–so I’m very familiar with their lifestyle.)

Some of the Asian American advantage is likely due to factors such as lower rates of obesity. But I also suspect that the same cultural attitudes that allowed highly educated Asians to live longer even before Covid, contributed to higher rates of vaccination, mask wearing, social distancing, etc.

Irvine also has geographical advantages. During Covid, it was possible to dine outside almost year around. When I moved here from Boston, I found the grocery stores here to be dramatically less crowded. In Boston, you must squeeze between shoppers. At a Gelson’s in Irvine, you can shop for 30 minutes in a spacious grocery store and never come close to another person, even when going though the self checkout.

All this leads me to believe that America’s relatively low life expectancy has nothing to do with our health care system. (I believe our system is awful, but mostly because it is absurdly expensive, and is a (the?) primary cause of slow growth in median real incomes.) Indeed the high quality of our hospitals might help to explain why our Asians live even longer than Asians in East Asia.

Instead, America’s low life expectancy seems to be due to lifestyle choices.

PS. Orange County’s Covid death rate (236) is a bit below California’s 250 and well below the national figure of 339.

PPS. Westminister’s Asians lean Republican and Irvine’s lean Democratic.

PPPS. I wonder if our federal bureaucrats are working for Trump:

A federal agency may look to ban gas stoves over concern about the release of pollutants that can cause health and respiratory problems, according to a new report.

A ban on gas stoves would cause America’s Asian American community to shift over to the GOP.

I’m utterly confused

Today’s Financial Times makes a claim that makes my head spin:

Of the 45 economists surveyed between December 2 and December 5, 85 per cent project that the National Bureau of Economic Research — the arbiter of when recessions begin and end — will declare one by next year.

While most of the economists expect the coming contraction to be shortlived — with gross domestic product growth still registering a 1 per cent gain by the end of next year — more than half are bracing for the unemployment rate to rise substantially from its current 3.7 per cent level to between 5.5 per cent and 6.5 per cent.

A handful of economists pencilled in an even more dire outcome, with the unemployment rate either closing in on or exceeding 7 per cent.

If you go to the link, you’ll find that the RGDP forecast is for 2022:Q4 to 2023:Q4. The median forecast for the unemployment rate in December 2023 is 4.5%. Oddly, that forecast doesn’t seem to match the consensus view that unemployment will peak at about 5.5%, and that the peak will likely occur in late 2023 or early 2024.

But that’s not what concerns me. Rather, in what universe could the US experience 1% RGDP growth in 2023 and end the year with 4.5% unemployment? What am I missing?

Some context might be useful. While the 4th quarter data for 2022 won’t be available for a few weeks, it’s pretty clear that RGDP growth for 2021:Q4 to 2022:Q4 will come in right around 1%, the same figure as being forecast for 2023. But 2022 was a boom year, with an amazing 4.5 million jobs being created. The unemployment rate fell from 3.9% in December 2021 to 3.5% in December 2022. And now we are being told that RGDP growth will continue at the same pace in 2023, and yet unemployment will shoot up to 4.5%? Really?

Companies are desperately short of workers. When I traveled over the holidays, I found service to be a complete disaster at one business after another.

Just to be clear, it would not surprise me at all if the unemployment rate did rise to 4.5% in 2023. What would surprise me is if this increase occurred during a time when RGDP was increasing at a 1% rate. What am I missing? Is AI going to suddenly cause a productivity surge in 2023?

I would not be at all surprised if there were a recession in 2023. After all, recessions often occur when anti-inflation policies are implemented.

I would not be at all surprised if there were no recession in 2023. A consensus of economists is currently predicting a recession, but the consensus of economists has been consistently wrong about every single recession over the past 40 years.

I suppose I could be accused of wimping out, refusing to put my reputation on the line. OK, so here’s my almost worthless forecast. I predict that America will experience its first ever mini-recession in 2023. The first “recession” we’ve ever had where the unemployment rate rises by less than 2 percentage points from the previous cyclical low. Whether that outcome would deserve to be called a “recession” is a question I’ll leave to others (i.e., the NBER). I don’t care.

PS. Today’s jobs report is very good news. I’m surprised that stocks aren’t up even more.

What crypto bubble?

In my entire 67 years on this planet, I’ve never seen a less bubble like asset than bitcoin. I’ve watched its price rise from less than a $1 to more than $16,000. That’s pretty impressive! Thus in January 2023, I would not expect gloating from people that “always knew” crypto was a bubble.

Over at the Financial Times, Martin Sandbu has this to say:

My colleague Jemima Kelly, who always saw the crypto bubble for what it was, writes on what the year in crypto taught us.

I’m certainly no fan of crypto. I’ve never invested in any cryptocurrency. I do not have any sort of expectations for the future path of crypto. But I am interested in bubbles, and the amazing tendency of people to see bubbles where they clearly do not exist.

Just for the fun of it, I googled Jemina Kelly and crypto bubbles, and came up with a Reuters article that she wrote in August 2017:

The price of a single bitcoin hit an all-time high of above $3,500 this week, dragging up the value of hundreds of newer, smaller digital rivals in its wake. Now some investors fear a giant crypto-bubble may be about to burst.

[To be clear, that’s the view of “some investors”, not necessarily Kelly.]

But what exactly does Martin Sandbu mean when he suggests that Kelly “always saw” that crypto was a bubble? Does he mean that she saw that bitcoin was a bubble way back in 2017, when the price was $3500? As of today, it’s $16,815.

Again, I could care less what anyone thinks of any particular investment. We all have opinions. I’m certainly no expert on forecasting asset prices. What interests me is the persistent attraction that people have to the idea of bubbles. An attraction so powerful that an example where bubble warnings were made at $3500 and the price later rose to $16,815 is seemingly viewed as confirming the earlier bubble warning.


Two milestones for 2023

Two historic milestones will be reached in 2023. For the first time in many centuries, India’s population will exceed that of China. In addition, China’s population will begin declining in 2023. During the latter part of this century, India’s working age population will likely be at least twice as large as China’s. By that time, India will likely have the world’s largest economy, with the US falling back to number three.

The Economist produced a useful graphical presentation of this historic moment:

Of course, there could be surprises ahead. But in recent years, the demographic surprises in East Asia have been to the downside, with birthrates falling much faster than expected.