Viewpoints

For a third straight month, the initial estimate of state labor markets for June had only 5 states sieeing statistically significant increases in payrolls, none appreciably numerically large. New York had the highest absolute gain (28,100) and Alabama had a .9% gain. Two states had statistically significant declines, with Indiana losing 13,900 jobs and Vermont recording a 1.4 percent drop (the other 5 New England states had insignificant declines).

11 states had statistically significant drops in unemployment from May to June (the same number as in May). Maryland’s .4 percentage point decline was the highest. Yet again, Nevada’s unemployment rate stayed the highest in the nation at an unchanged 5.4 percent. In another repeat from May, no other state had a rate more than a point higher than the national 3.7 percent, though DC’s was 5.1 percent. Alabama, Maine, Maryland, Montana, Nebraska, New Hampshire, North Dakota, South Dakota, Utah, Vermont, and Wisconsin all have rates more than a point lower than the nation, with New Hampshire and South Dakota both at 1.8 percent. California, Texas, Illinois, Washington, and Delaware (along with DC) are the states other than Nevada with rates at or above 4 percent.

Puerto Rico’s unemployment rate stayed at 6.1 percent. The job count on the island moved below 950,000. The bulk of the 8,200 drop was in the public sector.

The economy remains resilient and core measures of inflation are stubbornly high in the face of the most aggressive Fed tightening in decades. This resilience is partly a result of government stimulus programs during the COVID-19 pandemic. These programs generated a stockpile of excess savings that has continued to support household spending through rising inflation and higher interest rates. The Fed needs to counter that fiscal expansion in its fight against inflation.

Despite all the layoffs and furloughs, household income jumped during the pandemic due to the massive fiscal stimulus. Chart 1 shows the actual level of disposable personal income through the pandemic compared to trend. The income supports totaled more than $2 trillion in 2020-21. As the chart shows, the rise in income supports matched disposable personal income almost dollar for dollar.

State real GDP growth in 2023:1 ranged from North Dakota’s 12.4 percent annual rate to 0.1 percent in Rhode Island and Alabama. Growth was generally strongest in agricultural regions (though estimating the growth of farm output in the first quarter is always problematic). Northeastern states grew more slowly, in some instances in part due to weakness in agriculture, in others losses in manufacturing.

State personal income growth rates ranged from Maine’s 11.4 percent to Indiana’s -1.0 percent. As always, erratic swings in transfer payments account for much of the variation by state. Net earnings (employee compensation plus proprietors’ income) growth was strong in agricultural states, presumably due to a rise in farm income associated with the increase in agricultural output.

There is an economist in Chicago who has been known to see a silver lining behind every cloud. For example, after some natural disaster that resulted in hundreds of millions of dollars of damage to structures, this economist had been known to say that on the bright side, think of the rebuilding activity that will take place. By this logic, if the federal government wanted to increase the pace of economic activity, it could call on the US Air Force to carpet bomb some selected suburb, giving the residents plenty of notice to vacate the their premises with their irreplaceable possessions. (You might want to Google Bastiat’s “broken window fallacy” for the nonsense of this). I bring this up because after the debt-ceiling increase/extension legislation was signed into law on June 3, 2023, analysts, who unlike the aforementioned Chicago economist, see a cloud behind every silver lining. Even before the debt-ceiling bill hit the desk of President Biden, these nattering nabobs of negativity (you youngsters can Google this phrase) were saying that the rebuilding of Treasury balances at the Fed would suck liquidity out of the financial system, which, in turn, would cause all sorts of unspecified problems in the financial markets.

All else the same, it is true that an increase in Treasury deposits at the Fed would drain reserves from the banking system. But all else has not been the same in this case. Let’s go to Chart 1. Plotted in Chart 1 are the four-week billions of dollars changes in reserve balances held at the Fed by depository institutions (the blue bars), Treasury deposits at the Fed (the red line) and reverse repurchase agreements (RRPs) with the Fed (the green line). The last data points plotted are for the week ended June 28, 2023. In the four weeks ended June 28, Treasury deposits at the Fed increased by $360 billion, which, all else the same, would have drained that amount of reserves from the banking system. Yet, in the four weeks ended June 28, reserves at the Fed contracted by only $29 billion (the last blue bar plotted). Evidently, all else was not the same. One major factor that was not the same was the amount of RRPs executed with the Fed. An increase in RRPs drains reserves from the banking system; a decrease in RRPs adds reserves. In the four weeks ended June 28, RRPs executed with the Fed declined by $344 billion, which offset all but $16 billion of the reserves drained via the increase in Treasury balances at the Fed.

The construction of the leading indicators has many flaws, but one of the more visible and bigger ones is the three series for new manufacturing orders. The leading index includes two dollar-based new orders series and one diffusion measure. A diffusion measure captures the breath of change, not the magnitude of change. In other words, it does not distinguish between the size of the gain or decline. Yet, dollar-based series are more important in determining economic growth since the economy runs on dollars spent.

In May, the leading index fell 0.7%. The ISM new orders index posted the most significant decline (1.4%) of any indicator, overwhelming the small gain in the other two order series (0.1% and 0.2%, respectively).

Today, the Census Bureau reported that new orders for durable goods, a dollar-based series, rose 1.8% in May. And excluding the volatile transportation sector, new orders rose 0.6%. The latest data will result in an upward revision to the dollar-based series in the leading index, but not enough to wipe out the negative contribution of the ISM New orders index.

How can anyone trust the signal from the index of leading indicators when the dollar-based new orders for durable manufactured goods (excluding transportation) posted in May, their most significant gain in over a year, and yet the sum of the three orders series in the leading index is negative because of a sharp decline in diffusion-based series? The economy runs on dollars.

The year-over-year change in the All-Items CPI for May 2023 was 4.13%. My forecast is that the year-over-year change in the All-Items CPI for June 2023 will be less than 4.13%. Barring revisions, the seasonally-adjusted month-to-month percent change in the June CPI would have to be 1.19% non annualized for the year-over-year change in the June All-Items CPI to be equal to May’s 4.13%. Coincidentally (I think), the last time the month-to-month non-annualized change in the CPI was as high as 1.19% was June 2022, when it was exactly 1.19%. If the June 2023 All-Items CPI increases by 0.55% (non-annualized), the average non-annualized percent change in the CPI in the three months ended May 2023, the June 2023 year-over-year change in the CPI would slow to 3.47% vs. May’s 4.13%.

This is not economics. Rather, it is arithmetic. And it is all about that June 2022 base. (It took me a while, but I got there.) Plotted in Chart 1 are the month-to-month annualized percent changes in the All-Items CPI (the blue bars) along with the monthly observations of the year-over-year percent changes in the All-Items CPI (the red line). The June 2022 CPI increased a whopping annualized 15.22%. The June 2022 level of the CPI is the base for the June 2023 year-over-year percent change observation. With such a high June 2022 base, the bias is for a slowing in the year-over-year percent change in June 2023. The year-over-year percent changes in the All-Items CPI beyond June 2023 are not likely to slow as much because the high June 2022 base will drop out of the calculation. However, in the 11 months ended May 2023, the All-Items CPI has increased an annualized 3.17%. So, barring some negative supply shock in the remainder of 2023, the year-over-year change in the December 2023 All-Items CPI is likely to be much lower than the 6.44% for December 2022. For example if the CPI increases a non-annualized 0.3% from June through December 2023, the December 2023 year-over-year change in the All-Items CPI would be 3.59%. I believe that the monthly non-annualized changes in the All-Items CPI will, on average, be less than 0.3%. Thus, I believe that the year-over-year change in CPI as of December 2023 will be less than 3.59%.

The Federal Reserve Bank of Philadelphia’s state coincident indexes in May increased in from April in all but 3 states (Rhode Island, Minnesota, and Kentucky). Vermont and Massachusetts were again the leaders, and the only states with increases above 1 percent, but unlike in April Massachusetts took the top spot. These two were also the states with the largest 3-month increases (Massachusetts was up by close to 4 percent). However, a full 23 states had increases over this horizon of less than 1 percent (Rhode Island’s was barely positive). Over the past 12 months, Massachusetts was yet again the leader, with an increase of over 6 percent, perhaps some consolation for the Celtics’ fadeout in the NBA playoffs. Texas was a fairly distant second. Kansas and Missouri had increases of less than 1 percent.

The independently estimated national figures of growth over the last 3 months (.78 percent) looks roughly in line, though perhaps a bit short, of what the state figures suggest, while the corresponding 12-month result (3.65 percent) looks like it might be somewhat stronger than the state numbers.

A research paper by the staff at the Federal Reserve Bank of Dallas (1993) claims that the composite index of leading economic indicators does not provide "reliable advance information on the direction of the economy." Other studies have found that nearly half of the cyclical peak predictions in composite indexes of leading indicators were false signals.

The key takeaway from these studies, and others, is that the forecasting record of the composite index of cyclical indicators is not 100% accurate because the group of leading indicators that accurately predicted one cycle might not work in the next. In other words, as the economy moves from one business cycle to the next, the economy changes, as does the policy structure, and some indicators become obsolete and less reliable while others gain more predictive power.

Forty percent of the economic series that comprised the leading index in the 1980s and 1990s, when the Dallas Fed paper was researched, are no longer included. Change is a recurring feature of the leading economic indicators. In the 2000s, a new way of measuring the Treasury yield curve became part of the index. After the financial crisis of 2007-09, the redesigned leading economic index included a leading credit series and the ISM new orders series, removing broad money and vendor performance.

It's too early to conclude confidently that the current composite index of leading indicators sends accurate or false signals. But the performance of several indicators needs to be examined to avoid a wrong prediction.

For example, the current leading indicators index includes three new manufacturing order series. That construction is not ideal as it is best to have indicators covering a wide range of activities and sectors to avoid the intercorrelation between economic indicators.

Also, the current economic cycle has had unique features for the goods sector. Once the economy re-opened following the closures of businesses during the pandemic, there was a record surge in demand, especially for goods, driven by pent-up demand and unprecedented fiscal and monetary stimulus. Yet, firms could only respond slowly due to part shortages and supply chain bottlenecks. That forced firms to double and even triple ordering, resulting in the most significant (record) gains for manufactured consumer goods and capital goods (excluding aircraft). Now that the "ordering binge" has ended, the new orders series have reversed, especially for consumer goods posting monthly declines in five out of the past six months. That has contributed to the leading index's monthly decreases.

Yet, is removing "double-ordering" a sign of economic weakness or a technical adjustment in the orders series? New orders for consumer goods are off their record highs but remain elevated and stand 25% above pre-pandemic levels. Meanwhile, new orders for capital goods (ex-aircraft) were at record highs in April. It's worth noting that unfilled orders, an indicator in an earlier version of the composite index of leading indicators, stands at a record high. New and unfilled orders raise questions about whether the economy is transitioning to a slower growth environment or an outright recession.

Another questionable component of the current leading index is the yield curve, or the spread between the yield on the 10-year Treasury and the federal funds rate. The yield curve series was included in the leading index in the mid-2000s. But that was before the Fed embarked on quantitative easing (QE) or the purchase of long-date securities with the primary intent of keeping long bond yields lower than what otherwise would be the case.

The current inversion of the yield curve is the widest on record, negatively impacting the composite index of leading indicators. It defies logic to think that the yield curve offers similar (leading) signals when the Fed buys securities and when the Fed does not. The Fed doubled its balance sheet to over $ 8 trillion during the past three years. The yield curve was added to the composite leading index in the mid-2000s, and the Fed balance was pretty steady at $750 billion, less than one-tenth of its current size.

With QE as a new policy tool, comparing long bond yields to inflation makes more sense as it is a proxy for real interest rates and captures the intent of QE (i.e., keeping the long-term borrowing cost low). Replacing the yield curve with a proxy for real interest rates would dramatically alter the pattern of the composite of leading indicators.

One cannot use Paul Samuleson's comparison of the stock market in predicting "nine out of the past five recessions" with the track record of the composite index of leading indicators because if the current composition of the index does not accurately predict business cyclical turning points, it will be refitted or redesigned with a group of indicators that does. Unfortunately, that does not offer any hope for investors because people make decisions in "real-time" and can't wait for data revisions or a new index. History painfully shows that using one set of cyclical indicators to predict the future is fraught with failure. That's not a criticism of using leading indicators to help predict cyclical turning points. Still, things are constantly changing in the economy, requiring more than a small set of indicators to predict the future.