The Kraken Wakes, or Japan Revisited

The Days Ahead:

• New home sales and GDP update  

This Week:

• We look at some major economic indicators that are revised, wrong or just puzzling
• They’re revised over time but can give a distorted view about what’s going on
• Unemployment claims in Massachusetts are a mess
• Japanese equities are up and the head of the stock exchange wants to keep it that way
• Japan did some clever things to bring more people into the workforce
• We do a sprint around the commercial real estate market
• Many headlines are wrong on it
• Debt ceiling update: getting there definitely, maybe, probably

Revisions, Errors, Changes and Puzzles.

Some of the biggest data points that economists, market watchers, and money managers use are often wrong. These are the big numbers that everyone looks at and for which the market awaits impatiently. The big five are payrolls, inflation, GDP, the Fed announcements and consumer confidence.

Then there is a run of secondary reports like all the regional Fed surveys (such as the manufacturing surveys from the NY Fed, Philadelphia, Richmond and Chicago districts), retail sales, housing starts, claims, retail sales, durable goods, job openings (JOLTS) industrial production, producer prices and balance of trade.

Finally, there are numerous other data sources that the market tends not to look at. We’d include the Senior Loan Officer Opinion Survey (SLOOS), the Fed’s Beige Book, the Employment Cost Index, and auto production.

Now, the big five tend to always move markets. After all they cover what the economy’s doing, how many people are working, how people feel about the future and what the Fed thinks of it all. Sometimes, an item from the second list can become very important. Housing is often a first-tier data point but if the economy is going sideways and there’s neither a boom nor a crash going on in housing, then it tends to get pushed into the second tier. JOLTS is almost a tier one data point but that’s mainly because the last two Fed chairs have, and continue to, talk about it as if it’s very important (it’s really not because they’re two months old when published).

Anyway, lots of data and news come out. The trouble is some of it is wrong. The wrong data tends to fall into three categories: the revisions, the errors and the puzzles.

First up are the revisions. The chart shows job growth for the last five years, which includes the 22 million job losses in March and April 2020 and the long climb back.

From January 2020 to April of this year, there were a total of 26.1 million jobs gained and 22.2 million jobs lost, for a net gain of 3.9 million. Initially, it was reported as 23.5 million jobs gained and 21.3 million lost for a net gain of 2.2 million. That’s a 1.7 million gap.

The jobs numbers are revised many times over many months before the final number comes out. The initial reports show very different numbers. This next chart shows the difference from the first headline number until the final number some months later.

A bar below the midline is a negative revision, meaning the final number was worse than the initial report, and a bar above the line means a positive revision. The average revision was 185,000 with the average “up” revision at 195,000 and the average “down” revision at 176,000. Even if we omit the early 2020 numbers, given that we saw 21 million job losses in a month, the average up revision remains unchanged and the down revision falls to 133,000. Over the 40 months to April this year, revisions have been around 50% either side of the initial (the one we all paid attention to) number.

It was particularly hard in mid-2021, when the job market appeared to add 2.3 million jobs but it actually added nearly 4 million. In other words, the economy was running hotter than it looked. If the Fed knew then what it knew in early 2022, it may have avoided seeing inflation as temporary and started the tightening cycle earlier and more slowly.

This is nothing new. The BLS is clear that the margin of error for the new jobs numbers is 130,000. The reasons have to do with late reporting, incomplete returns and straight-up errors. But the biggest cause is the birth-death model. If a company with 200 employees in the BLS data base closes, or “dies”, the survey simply reports that the company let 200 people go and the BLS marks down that as 200 job losses.

But what might have happened was that a new company (a birth) started a few miles away and hired 150 of those employees. It’s a new company, so isn’t on the BLS radar. They’ll eventually catch up, but it will take some months. It’s at that point that the job number is revised from 200 job losses to 50.

So, when you hear that the new payroll numbers “beat” expectations, remember that the number first printed at 150,000 may end up as 20,000 or 280,000. No one will know for a few months. It’s kind of annoying that the biggest headline number in the economic calendar is subject to the biggest revisions but that’s where we are.

The second category is errors. This happened recently when we saw the weekly unemployment claims number jump to 264,000 or 20,000 more than expected. Claims is usually a reliable number as it merely counts the people who sign up for unemployment. Unless there’s a hurricane or flood, there’s very little delay between someone losing their job and applying for benefits.

Last week we saw this happen in Massachusetts.

Unemployment claims in Massachusetts climbed from 15,000 to 35,000 (the blue line) in two weeks (the graph shows approved not filed claims so it’s a smaller number) and Massachusetts share (green line) of all claims jumped to a record high of 12%. There are 3.7 million people who work in Massachusetts or 2.2% of the total national workforce. What happened to make unemployment jump from 3% to 9.3%?

Fraud. It doesn’t take much to submit a fraudulent unemployment claim. It’s a classic phishing tactic and only requires a name, an employer and some guesses about tax identity. On the Massachusetts state website they report it as:

Criminal enterprises using stolen personal information from earlier national data breaches have been attempting to file fraudulent unemployment claims through the DUA system. This is part of a national unemployment fraud scheme.”

Massachusetts saw 2.5 million claims attempts for unemployment in nine months in 2020, which was equivalent to two out of three employees! Many of the fraudulent claims were uncovered quickly but the scale of the problem was unprecedented.

Connecticut just saw another big increase in fraudulent claims and the total national bill may be as high as $80 billion. California thinks that 35% of all claims are fraudulent, although a lot are caught before they’re paid. Luckily a new bill was signed by Congress to deter and recover such fraud. It’s called the “Protecting Taxpayers and Victims of Unemployment Fraud Act” and does pretty much what it says and by going after false claimants for 10 years, which is up from three years. It was very bi-partisan and signed by Congressman George Santos, which he may regret.

Anyway, the Massachusetts number will surely be revised down in the claims reports this and next week. But it’s a big error that caught the market’s attention.

The final category is the plain unknown. Recently the Fed published its annual revisions to the country’s industrial production. One of the successes of the post Covid-19 recovery was in the Defense and Space Equipment industry and in March it (the red line) looked like this:

That’s some strong growth. The index reached 135 and was up 27% from its pre-Covid-19 level. It’s now at 114 and below its pre-Covid-19 high.

What happened? How could a large $740 billion industry be revised down from a 27% gain to a 2% loss? The short answer is we don’t know. From the Fed:  

Relative to earlier reports, the index for defense and space equipment now records noticeably slower growth over the 2020–22 period.”

Which leaves one better informed but none the wiser. I mean, haven’t we been sending Abrams, Himars and Javelins over to where they’re needed and hasn’t defense been a no-go area for budget cuts?

One explanation is that the production location is reattributed, so steel for a tank ends up in the defense category but should be in the primary metals section. That’s common enough where source data is lagged. This may happen when, again, a steel producer ships raw steel and it only shows up in a plane or rocket a year later. But this is the defense industry and the sourcing is pretty well-known.

Anyway this is all very new and we called the Fed and asked what it thinks is going on. They believe it may be companies like Boeing and Lockheed redesignating a new military aircraft that later becomes a civilian aircraft. Or simply late deliveries of big aircraft. If that’s the case, then the production will eventually show up elsewhere. It’s likely that overall industrial production of space and defense did not fall nor is it headed for recession. It’s just that it’s miscategorized. We’ll follow up.

All this is not to complain about the data or about its accuracy. It’s just that revisions can be frequent and large. And what you thought was going slow turned out to be too fast, as in the case of jobs in mid-2021, and what you thought was going well was actually much slower, as in the case of defense. It can make forecasting a bit of a mugs game.

We’re not worried. We believe the economy is slowing but not at a rate that will cause deep, recessionary cuts in employment. Sometimes, however, you have to wait a while for the data to settle down.

The Kraken Wakes or Japan Equities on the Move

As anyone who’s worked in finance for over five years knows, the Japanese bond and stock market are where performance goes to wither. Japanese government bonds have not moved from a -0.3 to 0.4% range since 2014 and have been below 2% since 1997. Government debt has risen to 222% of GDP, compared to the U.S. at 92% (ok, 118% if you include debt held by the trust funds).

Wages have fallen 11% in the last 25 years and purchasing power has only been maintained because inflation has been flat since 1993. Prices fell for much of that period and there were only occasional bouts of 1% to 2% inflation when the government imposed a VAT or sales tax.

On top of the general economic malaise, Japan is one of the first countries to go through what Europe and China are about to go through in the next 20 years, a greying population. Japan’s total population peaked around 127 million in 2000 and is now 124 million. The number of people in Japan aged 15 to 74 fell 9% since 2000 and the 65 and over age group went from 18% of the population in 2000 to 30% in 2023.

As we mentioned a few months ago, we think there are some changes underway in Japan as the new Governor of the Bank of Japan starts to grapple low rates, deflation and low growth problems. There are no quick fixes when the central bank owns $4.2 trillion of government bonds or 104% of GDP. The bond holdings are around 50% of all outstanding government bonds (the Fed owns around 16% in the U.S.) and 80% of all bonds with a 5- to 10-Year maturity. That means that 80% of all intermediate government bonds simply do not trade!

Anyway, this is not a litany of Japan’s problems but some notes on what’s changing. We’ll focus on the workforce and the stock market.

The female labor force participation rate (LFPR) in Japan bottomed at 45% in 1972. As Japan recovered from the war, less women were needed or wanted to work. At that time, the U.S. female labor force participation was around 48%. In the U.S., it peaked at 61% in 2000 and is now around 57%. Japan’s went on to grow to 51% in 2000 and is now 55%. Even though the working age population of Japan fell, the number of workers increased as women entered the workforce.

Japan’s female labor force (the blue line) grew 17% in the last 15 years while the male labor force grew around 3%. Again, the number of women in the population did not grow. But the number of working women did and by a remarkable rate.

Japan’s other secret weapon to bring more people into the workforce while the population shrank, was to make it easier for “young” retirees aged 65 to 70 to work. The U.S.’ 65 and oolder participation rate was around 17% in 2010 and in Japan it was 21%. Those numbers are now 19% and 26%.

Through a combination of part-time work incentives, generous child care programs, a one-year maternity leave and setting up elder care for the over 80 age group by those in the 65 to 70 age group, Japan made a declining working age population into a larger workforce.

And then there’s the stock market. Generally, stocks are an excellent long-term investment providing one can hold on through bear markets and setbacks. Japan’s stock market looks like this:

Japan’s stock market was the place to be in the 1980s. Land values were climbing; the Japanese economy was an export powerhouse and everywhere Japan’s business methods were studied and copied. There was a reason the “Die Hard” movie renamed the 20^th Century Studios Plaza the Nakatomi Plaza for its Christmas story. Japanese stocks traded at 70 times earnings and every global fund manager and bank headed over to Tokyo.

Then it all burst.

Recently things have changed. The broad Topix stock index (we use an older one in the graph, because the Nikkei 225 has a longer track record) is one of the few major markets above its 2021 peak.

Over the last year, Japanese stocks for U.S. investors are up 7% and they’re up 11% year to date. That compares to the S&P 500 at -1.1% and 6.7%. Why? Some of it is good economic news. Japan’s first quarter GDP grew 1.6%, against 0% in the last quarter of 2022 despite a drop in net exports.

Inflation is at 3.1%, which to Japan is great news given decades of deflation. After years of low real wage growth, several high-profile companies have made wage increases outside of the shuntō or “spring wage offensive” the time when employers and unions conduct pay negotiations. Nintendo increased wages by 10%. Kuraray (chemicals), AGC (glass and electronics), Fast Retailing and Nidec (motors) have all announced between 8% to 10% wage increases. The labor unions at smaller companies recently won a 3.7% wage gain, the biggest in 30 years.

Will it last? Japanese stock markets trade at about a 20% discount to the U.S. but they have done so for years. That in itself is not a reason for confidence. But the gradual step away from deflation, zero wage growth and a revived housing market all bode well. The next step will be to see how the Bank of Japan transitions to a more normal policy, especially around its targeting of the 10-Year Government Bond yield at 0.5%.

In another good sign, Hiromi Yamaji, the head of the Japan Exchange Group, which regulates the stock market, has told companies to focus more on share prices and said that companies that trade at price to book value of below 1.0 need to start paying more attention to shareholders’ needs, like stock buy-backs and underwinding complex cross-shareholdings. About 50% of the 1,800 companies listed on the main exchange trade below book value compared to 10% in the U.S. So, things could get interesting.

We’ll know more in coming months “changes coming in Japan” has been a thing for over 30 years…it’s like Lucy and the football. Meanwhile, however, patient U.S. investors are finally seeing some outperformance.

A Real Estate Sprint

One of the fun things about working at Cerity Partners is that I get to hear experts and smart thinkers speak about every corner of the investment world. Recently Yelena Pelimskaya in our Philadelphia office presented on what’s going on in commercial real estate (CRE).

We’ve seen pretty dire headlines about CRE recently. The problems at the banks have led many to fear for the $845 billion in CRE loans from the big banks (about 6% of total assets) and the $1,960 billion in CRE loans at small banks (about 30% of assets). Will they have to take write-offs? Will the CRE market crater as rates increase and refinancings come due?

Well, no. Here are a few points.

First is that when the headlines run, they tend to be about office oversupply, working from home and buildings like 555 California Street, an empty 1970s Class B office building in San Francisco, which was valued at $160 million a year ago and now can’t find a buyer at $35 million.

But while office real estate may be the most talked about part of CRE, it’s only 12% of the total market. It drops to 1.7% if we include all investable assets, as in the chart below.

Second, only about 0.6% of all commercial mortgages are more than 30 days delinquent. There’s another 0.9% in foreclosure but there’s always residual value and rental renegotiations in foreclosures. The prices may fall but they’re very unlikely to take banks down with them. Even a 2% default rate on the $2,808 billion of bank CRE loans is only $56 billion against bank capital of $2,235 billion.

Third, the loan-to-value of CRE mortgages is around 52% compared to nearly 70% in 2007. Banks and lenders have imposed conservative underwriting standards for many years. That gives a margin of safety for both banks, lenders and owners.

Finally, many of the other CRE sectors, the ones barely mentioned in the headlines, are doing fine. Here’s a snapshot of how we look at the CRE market:

Multifamily rents are growing due to steady and rising demand as people opt for, or are priced out of, home ownership. The industrial sector covers everything from data centers to warehouses to light and specialty manufacturing. All home delivery businesses need facilities for the critical “last mile” and all need additional transit and storage space. Finally, retail space might seem empty and in trouble but new retail space deliveries, which means new construction or repurposed buildings, has declined every year since 2016. The supply has contracted a lot in recent years.

We’d agree that higher rates mean real estate faces some financing risks. But many parts of the CRE sector are in solid shape, with conservative leverage and opportunities for high occupancy and lease increases. It’s just not discussed that much.

Debt Ceiling

Debt ceiling negotiations continued and we’re encouraged that all the players are talking. Treasury Secretary Janet Yellen has done a great job reminding everyone that this is not a game and trying the default option under the guise of “How bad could it be?” would end very badly indeed.

Here’s a quick update on what we see happening as of Thursday May 18, 2023 (click to enlarge).

We think something will happen next week when President Biden returns from the G-7 summit. Even a kick-the-can-down-the-road for six months option would be embraced by the market.

A couple of quick points.

First is that credit default swap prices, which would protect a bond buyer against a U.S. default, have fallen. The market in these instruments is very thin so we wouldn’t read too much into them. The press talks about them as if they’re some leading, efficient market indicator. But they’re not. We see prices for a Greek sovereign bond default at only 37 basis points, or bp, meaning if you bought a Green one-year bond at 4%, you would net 3.62% after buying a CDS. The U.S. one-year CDS stands at 158 bps. Those are strange prices and they will surely settle down as soon as an agreement is made.

Second, this:

This was put together by our Los Angeles colleague, Phong Luu. It shows every Treasury bill (short-term bonds) maturing from June 1^st 2023 to December 1^st 2023. Bills maturing in June show a yield of 5.3% and 5.6%, with the two bills, in the top left, maturing just before the June 15^th tax deadline. In July, the bill rate drops to 4.7% to 5.1% and stays between 4.9% and 5.2% for the rest of the year. Basically, the Treasury bill market is worried about a June default but not after that date. Everything from then on is relatively safe.

That fits with our read. Why hold a bill maturing in a month which, if you invested $1 million, would only give you $250 more for one month than one maturing in August?

We remain of the view that there will be a deal but we face a lot of stress until we get there. One counterintuitive market move we’ve seen is that longer term Treasuries may rally if there is a default. The logic is that it would mean weaker growth and investors would look to a flight-to-quality asset in the only major reserve currency: U.S. Treasuries.

Meanwhile, third, the discharge petition process has started. It’s a 30-day process that allows a bill to go straight to the floor without a committee review. It’s been signed by 210 members of the House and would need another 8 votes to work. But it’s history of success is very low

The Bottom Line

We had two surprise manufacturing reports this week. One, the New York Fed’s Empire State Index surprised to the down side and the other, the Philadelphia index surprised to the upside. The 10-Year Treasury rate increased to the top of its recent range at 3.6%. Some of that is because the regional Fed presidents stray from the official Fed position, with one voluble president talking about one more hike for “insurance” (eh?) and another saying they should wait. The S&P 500 is up nearly 1% for the week and near the top of its recent six-month range and 20% above the October low. The Nasdaq is way over its six-month range and up 20% year to date and up 25% from its October low. International markets, especially the Eurozone and Japan remain strong performers with the Japanese market notably showing less volatility than the S&P 500, mostly for the reasons we discuss above.

Subscribe below for our investment updates.

---

Art: Stephanie Holman (b. 1967)