Advisor Perspectives welcomes guest contributions. The views presented here do not necessarily represent those of Advisor Perspectives.
The Fed’s FOMC finally raised its target rate of interest by 0.25% at the end of last year. But by the time it met again on January 27, widespread public controversy had emerged over whether it had made an historic mistake. Driving that purported mistake were systemic problems that prevented the Fed obtaining reliable and timely data upon which to base its decisions.
In its January 27 meeting, the Fed’s FOMC monetary policy committee was once again confronted with a big gap between its expectations and economic reality. Late last year, the Fed said economic performance was “solid” and consumer and capital spending were likely to increase. Early in winter, the surface might have looked solid, like a sheet of ice on water, but when the Fed members walked out on that surface it broke, and they found themselves in icy waters.
The economy had turned downward.
The Fed’s choice of the word “solid” was inappropriate. There were evident signs of slowdown in retail sales and capital spending plans, and the inventory-to-sales ratio was extraordinarily high. Productivity was declining. Family incomes lacked significant growth.
Continuing weakness in oil did mean cheaper gasoline and fuels for households, but healthcare costs were rising last year by more than the gains from lower fuel costs. In 2016 healthcare insurance premiums have been raised by more than 20% in many cases, and deductibles (out of pocket expenses of the insured) had sharply increased. The quality of new jobs tabulated was rapidly weakening as part-time jobs were filled by workers unable to secure full time jobs. More and more workers were becoming “independent contractors” in part-time jobs with varying, limited hours of work. A better unemployment rate was illusory; it was the result of a decline in the labor participation rate (meaning a growing share of the working-age population stopped looking for work and “permanently” dropped from the official calculation of the total labor force).
Contrary to upbeat official reports of job gains, declining real-time government data on payroll taxes were showing a labor market decline throughout December.
The Fed had warned repeatedly that its decisions would be “data dependent.” But the fundamental problem with the Fed’s “data dependent” guidance to markets is that the data the Fed uses is obsolete; it is based on economic activity recorded in several months past. It relies on data compiled by the Bureau of Labor Statistics (BLS) and the Bureau of Economic Analysis (BEA) in the Department of Commerce. The GDP data is compiled from data covering the previous quarter and published a month afterwards. The GDP growth rate is revised 30 days later, and again after another 30 days incorporating additional data. Moreover, all of that data is subjected to “smoothing” with seasonal adjustments based on experience during the same seasons in previous years.
In other words, the Fed’s decisions are made on the basis of economic activity that took place at least four months earlier and was revised to include additional adjustments over the succeeding two months and adjusted with experience from previous years.
Because new businesses are starting up and old businesses are shutting down each month, the BLS does not have timely data on jobs lost due to shutdowns or hiring generated by newly emerging firms. The BLS therefore uses sample surveys, ignoring firms going out of business and reliant on lagged indicators of new hiring by new firms. It assumes that “births” and “deaths” of businesses are roughly equivalent (which is not true, as shutdowns exceed startups lately). Revised quarterly, an add-on number is applied to the non-farm payroll (NFP) calculation for new jobs provided by new firms based on experience during the same quarter in the previous year.
Extrapolation of a linear move perceived in the most recent quarter from the same quarter a year earlier means that the birth-death adjustment is not able to identify a turn in the business cycle. Because new small and medium-sized firms generate a large portion of the total of new jobs, the birth-death adjustment estimate each month is a substantial number. In December 2015 the birth-death adjustment accounted for much of the published “job gains” of 292,000. If real-time falling payroll tax deductions had been taken into account instead, the estimate of new jobs created would have been much smaller – closer to zero.
Comparing data released on any specific date and then looking at official “revisions” made periodically over the subsequent five years will reveal more serious reasons to doubt BEA data. Such revisions are not published until a five-year expiration.
BEA's Official and Changing View of First Quarter 2008 GDP
|
Reported Growth Rate
|
Report Date
|
|
+0.6%
|
April 30, 2008
|
|
+1.0%
|
June 26, 2008
|
|
-0.7%
|
July 31, 2009
|
|
-1.8%
|
July 29, 2011
|
|
-2.7%
|
July 31, 2013
|
Looking back, the National Bureau of Economic Research (NBER) determines in retrospect when an officially defined recession begins and when it ends. The NBER determined that a recession began in December 2007 and did not end until June 2009. During that recession, real GDP declined in the first, third and fourth quarters of 2008 and in the first quarter of 2009. In other words, the official U.S. data available to the FOMC in Q1, 2008 showed positive growth. But five years later, following four additional “revisions,” it was shown that the U.S. economy in Q1, 2008 had already entered a period of significant negative growth.
The Fed staff prepares public economic forecasts in an effort to provide guidance to financial markets about what the FOMC members foresee. Looking back, these public forecasts have proven to be significantly overoptimistic for many years. Market analysts have become accustomed to overoptimistic forecasts by the Fed, often suspecting that the Fed is trying to “cheerlead” economic confidence.
The Fed staff also prepares “Green Book” estimates to be used by the FOMC a day or more before each of its meetings. The Green Book assessments are alleged to be more current than publicly available data, permitting the FOMC an ostensibly more “real-time” perspective on the U.S. economy’s performance. These Green Book findings are not publicly disclosed for five years, when the forecasts are no longer useful to markets and are only useful to satisfy historical curiosity. My example of the disparity of what the Fed saw just after the close of Q1, 2008 and the real outcome as a result of five years of revisions suggests that the Green Books must also have been seriously obsolete.
Recently, the Atlanta Federal Reserve devised a more frequently updated estimate of GDP growth (the “Atlanta GDPNow” growth indicator), which in the past couple of years has turned out to be far more accurate in predicting BEA end-of-quarter GDP estimates. Unfortunately, even the Atlanta Fed’s attempt to generate closer-to-real-time forecasts relies on obsolete BLS and BEA collection and calculation methodologies.
Today, computers and information technology have become far more powerful and are capable of collecting and analyzing masses of real-time digitized data that exists but is dispersed widely across the economy. There is no government program aimed at collecting and correlating what has already been digitized for varying private and public purposes.
U.S. government agencies and the Federal Reserve have neither the appropriate data-science capability nor the computer power to evaluate the current performance of the economy. Many elements of such capabilities are currently in use in industry, the U.S. military and various other entities such as NASA, the National Laboratories and universities. Our Federal Reserve and government overview of our economy is prepared with mid-20th century capabilities that are a half century or more out of date.
Moreover, none of the present Fed decision makers have meaningful experience with 21st century market functionality, which is heavily reliant on high-frequency trading (HFT) in milliseconds (thousandths of a second). Federal Reserve officials primarily rely on market opening and closing data, but they do not have the means to monitor the flow of markets during a trading day in any detail. The current pace of HFT orders is approximately four milliseconds (four thousandths of a second). Firms engaging in HFT are expecting the speed to be stepped up in 2016 to less than one millisecond per bid using laser transmissions now being installed at the NYSE and soon at other exchanges and trading platforms. The internet utilized by many financial institutions and retail investors is also in transition to support orders of magnitude speed increases with the coming introduction of “Li Fi” technology, which will eventually replace WiFi with speeds more than 100-times faster (Li Fi requires light sources, without any hard wiring or radio signals).
Although the Federal Reserve has long tried to convince markets that it is “data dependent,” the reality is that it does not have the capability to determine when there has been a turning point or “inflection point” in the U.S. economy. In fact, using its present data methodologies, it cannot know if a recession has begun until many months later, and it cannot determine the depth of such a recession until years later.
This is not to suggest that the fourth quarter of 2015 was as bad as the first quarter of 2008. But the economy was likely in a dynamic transition towards lower (and potentially negative) growth in the last quarter. We may not know exactly what the actual 4Q-2015 growth rate was for months or even years. However, what can be said is that per capita incomes are barely rising, inventory-sales ratios are historically high, healthcare costs are not only rising but will surge more in 2016, new orders for durable goods are weakening, capex plans are anemic and, with world trade flatlining, exports relative to imports will continue on a negative path for some time ahead, subtracting from growth. With job growth primarily in part-time jobs and with falling availability of full-time jobs, the outlook for personal consumption looks alarmingly weak.
The FOMC remains confused about its next monetary policy move, whether to make another rate hike, hold off from further action until the economy’s performance clarifies, or begin to consider new stimulative steps in the event a recession returns.
FOMC members can see in the flattening of the yield curve that financial market participants do not believe the economy is “solid” enough to warrant four rate hikes in 2016. Instead, there is growing market doubt than any further rate increase will be feasible, either because the U.S. economy stumbles, the financial market undergoes an end-of-cycle downturn, the world economy suffers greater weakness or some combination of all. Strong demand for longer maturity Treasury securities are pushing yields down. That strong demand is not only domestic. Foreign central banks are taking increased, even record shares of Treasury debt auctions across the spectrum from two-year maturities all the way out to 10- and 30-year bonds. As for official employment data, even Fed officials are acknowledging that jobs data understate the degree of “slack” in the labor market. The so-called “unemployment rate” is no longer credible as indicator of the economy’s health.
It is therefore not surprising that markets are less and less influenced by “data dependent” public comments by various governors of the Federal Reserve Board and presidents of the Fed Regional Banks.
Dr. Harald Bernard Malmgren is a scholar and has been an ambassador, international negotiator and senior aide to U.S. Presidents John F. Kennedy, Lyndon B. Johnson, Richard Nixon and Gerald Ford[1] and to U.S. Senators Abraham A. Ribicoff and Russell B. Long, United States Senate Committee on Finance, advisor to many foreign leaders and CEOs of financial institutions and corporate businesses, and frequent author of articles and papers on global economic, political and security affairs.
Read more articles by Harald B. Malmgren
