Official Government Statistics? The Fed Don’t Need No Stinkin’ Government Statistics!
by Paul Kasriel, The Econtrarian
One of the “casualties” of the federal government shutdown is the suspension of official federal government statistics measuring the performance of the U.S. economy. Numerous economists have commented how this lack of economic information will make the Federal Reserve’s management of monetary policy all the more difficult. I respectfully disagree. I believe that the lack of official federal government economic statistics might actually enhance the Fed’s management of monetary policy. Let me explain.
Milton Friedman, may he rest in peace, used to argue that the pursuit of discretionary policy actions by the Fed actually increases the amplitude of a business cycle. In other words, discretionary monetary policy tends to reinforce booms and busts. Increasing the amplitude of a business cycle, of course, is not the intent of discretionary Fed policy, but an unfortunate outcome because of insufficient knowledge on the part of the Fed and, for that matter, all mortals. The Fed lacks sufficient knowledge with respect to the lagged effects of a policy action. Sometimes a policy action might have its maximum effect on the economy in three months; sometimes in six months. Even if it knew with certainty when a policy action would have its maximum effect on the economy, the Fed often knows too late when such action would be called for. Most of the economic data, official federal government or otherwise, the Fed uses to assess the performance of the economy are available only with a lag, typically a lag of a month or a quarter. And then all of the important official federal government economic data get revised, often significantly revised and often revised over and over again for several years thereafter.
An example of the revisions to key measure of economic performance, the quarter-to-quarter annualized growth in real GDP, is shown in Chart 1. Shown are two estimates of annualized growth in real GDP for each of the four quarters in 2007. The red bars represent real GDP growth as estimated in January 2008. The blue bars represent real GDP growth for these same 2007 four quarters as estimated in September 2013. In each of the first three quarters of 2007, the estimates of real GDP growth available to the Fed in January 2008 were higher than the estimates available currently. The largest overestimate of real GDP growth occurred for Q3:2007, where the January 2008 estimate was 219 basis points above the current estimate. Mind you, that January 2008 estimate of Q3:2007 real GDP growth was already the third revision of the October 2007 first guess by the Bureau of Economic Analysis of Q3:2007 real GDP growth. To the Fed’s credit, the majority of FOMC members in January 2008 based their policy decisions on the mounting dysfunctional behavior of the financial markets rather than ephemeral coincident indicators such as real GDP growth. But this example highlights how misleading “real-time” federal government economic statistics can be.
Even if the economic data provided by the federal government were more timely and subject to less revision, they still might not justify discretionary Fed monetary policy. Bear in mind, monetary policy affects the demand for goods, services and assets – not the supply. Negative supply shocks to the economy, such as a foreign oil embargo, will reduce the production or supply of real goods and services. If the Fed were to respond to weaker real output resulting from a negative supply shock by taking actions to stimulate aggregate demand, higher inflation would be the ultimate result, not higher real output. Sometimes, it is difficult to disentangle the effects on real output from simultaneous supply and demand shocks. But regardless of whether a decline in output occurs because of a negative demand shock or negative supply shock, the political pressure usually mounts on the Fed to “do something”.
Instead of pursuing discretionary monetary policies, Milton Friedman suggested that the Fed adopt a rules-based approach to policy. Friedman favored a rule for some steady growth in the nation’s money supply. He did not believe that a steady rate of growth in the money supply would eliminate business cycle booms and busts. He did believe, however, that such a rule, if adhered to, would dampen the amplitude of cycle booms and busts. My interpretation of Friedman’s rule is for the Fed first to do no harm.
As much as I respect and admire the scholarly economic research conducted by Milton Friedman, I would humbly offer a “tweak” to his monetary rule. Rather than the Fed pursuing a policy resulting in some steady rate of growth in the money supply, I would suggest that the Fed attempt to produce a steady rate of growth in the sum of the credit it creates and the credit created by depository institutions, i.e., commercial banks, savings associations and credit unions. If there are any regular readers of my commentaries – heck, if there are any readers of my commentary – you might remember that I call this credit sum “thin-air” credit. Why the term “thin-air” credit? When the Fed purchases securities or makes a loan, it pays for these asset acquisitions with funds created figuratively out of thin air. Similarly, in a fractional reserve requirement environment, when the depository institution system adds loans and securities to its assets, it “pays” for these asset acquisitions with funds created figuratively out of thin air.
Friedman’s concept of the money supply is a subset of the total liabilities of the depository institution system. My concept of depository institution system thin-air credit is a subset of total assets of the depository institution system. Because assets equal liabilities plus net worth, our two concepts are similar so long as the ratio of money supply liabilities to depository institution credit remains relatively stable. Chart 2 indicates that this ratio has been anything but stable over the post-WWII period. Checkable deposits, savings deposits and small time deposits of depository institutions, the bulk of the M-2 money supply, as a percentage of loans and securities on the books of depository institutions has ranged from a high of 101.4% in Q1:1952 to a low of 54.6% in Q4:2007.
So, which variable has a closer relationship to nominal domestic purchases of currently-produced goods and services, i.e., gross domestic purchases, the M-2 money supply or the sum of Federal Reserve and depository institution credit? Plotted in Chart 3 are the quarter vs. year-ago quarter changes in nominal gross domestic purchases vs. the nominal M-2 money supply. The correlation between the two from Q1:1960 through Q2:2013 is 0.36 out of a maximum possible 1.00. In Chart 4, substituted for the changes in the M-2 money supply are changes in total thin-air credit, i.e., the sum of Federal Reserve and depository institution credit. The correlation between changes in nominal gross domestic purchases vs. nominal total thin-air credit is 0.58. So, it would appear that if the Fed were to pursue a rule of a steady rate of growth in monetary variable, total thin-air credit would be superior to the M-2 money supply.
Now, only an absolutist (moi?) would recommend that the Fed pursue a rule of promoting a steady rate of growth in thin-air credit with total disregard to the performance of the economy. But are there available indicators of aggregate economic performance that are not generated by federal government bean counters, not subject to a lot of revisions and yet comport with official federal government economic statistics after revision? You betcha. Let’s look at some of them.
The Institute of Supply Management releases on the first business day of each month its composite Purchasing Managers Index (PMI) pertaining to manufacturing firms. The purchasing managers at these firms are polled with regard to their new orders, production, employment, inventories and supplier deliveries (the speed at which suppliers fill and ship orders). Currently, PMI is an equal-weighted average of these five sub-diffusion indices. The higher the value of the PMI, the stronger is economic activity. Importantly, the not-seasonally-adjusted values of the sub-indexes are never revised.
As shown in Chart 5, from 1948 through 2012, the correlation between the percent change in revised annual average real GDP and the unrevised annual average PMI is 0.73 out of a maximum possible 1.00. So, no, the PMI will not tell the Fed that real GDP is growing at X.X%, but it will tell the Fed whether economic conditions are strengthening or weakening. And, as important, the message conveyed by the PMI will not be revised next month, next year or next century.
If the PMI does not suit you in gauging the strength and direction of economic activity, try sales of autos and light trucks. Chart 6 shows that there is a 0.73 positive correlation between the percent change in revised annual average real GDP and annual average light weight vehicle sales. I purposely used in Chart 6 the light weight motor vehicle sales data generated by the Bureau of Economic Analysis, a division of the Commerce Department, because of the length of the series. Of course, these data will not be available again until the Commerce Department re-opens for business, so to speak. But the Fed really does not need the BEA’s estimate of light weight vehicle sales because Autodata Corporation, a private automotive research firm, provides almost identical data to those of the BEA (see Chart 7). To the best of my knowledge, Autodata Corporation’s light weight vehicle sales data, which are derived directly from the very corporations who sell cars and trucks in the U.S., never get revised.
Economists were saved from the embarrassment of being wrong again in their forecasts of the monthly change in non-farm payrolls this past Friday because the government shutdown prevented the release of the September employment report. But without this monthly report generated by the federal government, how can the Fed know where it stands in meeting one of its mandates, full employment? How about using the data collected by the states enumerating how many people are collecting state unemployment insurance benefits? Chart 8 shows that there is a negative correlation of 0.66 between the percent change in the annual average of revised nonfarm payrolls and the percent change in the annual average of revised continuing state unemployment insurance claims.
Even if the federally-generated nonfarm payroll data were available, the Fed would be wise to rely more on the state-enumerated continuing claims data. Why? For starters, the unemployment insurance claims data are not derived from a sample. These data are the universe of real people collecting real unemployment benefits. The claims data contain no business birth/death adjustments or other imprecise elements that mainly serve to excite conspiracy theorists. Admittedly, because a government entity is counting the real people standing in real lines, the first count is usually inaccurate. But after about four weeks, the state governments have corrected their errors and there are no more revisions. If only that were the case with regard to revisions for the monthly Bureau of Labor Statistics non-farm payroll estimates!
Because residential real estate is such a “credit intensive” sector of the economy and because Fed monetary policy has such a large effect on credit conditions, it would be natural for the Fed to want to monitor supply and demand conditions in the housing market. But where could the Fed turn to in the absence of official federal government housing starts and new home sales data? To the National Association of Home Builders (NAHB), that’s where. Each month the NAHB surveys its members with regard to their sales of new homes. Members are asked whether sales are good or poor. If all members responded “good”, the NAHB diffusion index would be 100. If half respond “good” and the other half respond “poor”, the index number would be 50. Chart 9 shows that the correlation between monthly revised single-family housing starts and the monthly NAHB diffusion index of single-family new home sales is 0.84. Chart 10 shows that the correlation between monthly revised single-family new home sales and the monthly NAHB diffusion index of single-family new home sales is 0.76. Any revisions to the NAHB index are small and limited compared to the magnitude and frequency of revision to the other government-generated series.
In summary, the Federal Reserve has more than enough information to conduct a proper monetary policy in the absence of official federal government economic data. The Fed should back away from its attempts to micro manage the business cycle, which Greenspan popularized. There was a time when the Fed was so “neutral” in its effect on the business cycle that the average informed woman or man on the street did not know the name of the Fed chairman. If the Fed were to adopt an operating policy of achieving a steady rate of growth in nominal thin-air credit, it could return to its prior anonymity. For reality checks as to how the economy is performing, the Fed could monitor the behavior of quarterly averages of the PMI manufacturing index, car and truck sales, state unemployment insurance benefit claims and the NAHB diffusion index of new home sales. Reasonable people can debate the costs to society of the current federal government shutdown. Call me unreasonable if you want, but I do not believe society has lost much from the suspension of the releases of federal government economic statistics.