A Reply to George Selgin
by Jeffrey Rogers Hummel and David R. Henderson
George Selgin posted a thoughtful critique of our Cato Briefing Paper on "Greenspan's Monetary Policy in Retrospect." His criticisms, rather than shaking our confidence in our evaluation, have strengthened our conviction that Greenspan's monetary policy is widely misunderstood by both detractors and apologists.
In some respects our disagreements with Selgin are not as severe as he believes. He devotes an early section of his critique to establishing the relationship between low interest rates during 2002 to 2004 and the growth of risky subprime mortgages. But we never denied that relationship and, indeed, think it was as important as he does. What we disagree about is the primary cause of the low interest rates during that period.
Nor does our paper argue, as Selgin asserts, that"Greenspan's Fed was innocent of any role in encouraging the housing boom [emphasis ours]." Indeed, we state:"Particularly alarming is the way the lender-of-last-resort function has been expanding the moral-hazard safety net and mispricing risk, a trend to which Greenspan no doubt contributed." We not only accept that the Federal Reserve's contributions to moral hazard may have been a major factor in the housing boom, but also stipulate at the end of our paper:"Minor blips in total reserves under Greenspan may have played some poorly understood role in any of these three events [the recessions of 1990 and 2001 and the current one]." Our disagreement is with the widespread and extravagant accusations that"easy Al" was conducting an exceptionally expansionary monetary policy after 2001.
Greenspan and Velocity
Before returning to the subject of the current crisis and its ultimate causes, let us take up Selgin's most serious challenge. Our Cato Briefing Paper argues that as a result of significant deregulation, the Fed lost most of its lingering control over the short-run movements of the broader monetary aggregates (M2, MZM, and M3), whose behavior therefore approximated what Selgin himself, along with Lawrence H. White and Steven Horwitz, contend would happen under a system of completely free, unregulated banking. Fluctuations in the money supply approximately responded to and offset fluctuations in money velocity, helping to stabilize the macroeconomy. Selgin correctly notes that a stable MV necessarily implies a stable nominal Gross Domestic Product (Py), as expressed in the equation of exchange (MV = Py). He then questions the actual stability of nominal GDP under Greenspan, buttressing his reservations with a graph (Chart 4 in his critique) showing the growth rate of final sales of domestic goods (nominal GDP minus changes in private inventories) from the recession of 2001 to the present.
Nominal GDP, however, is simply the product of real output and the price level. A decline in volatility of both output and inflation, therefore, strongly suggests that the volatility of nominal GDP has also declined. We never claimed that Greenspan's policies were perfect, and in fact our paper explicitly criticized them for being"too discretionary." After all, the Greenspan era did encompass two minor recessions, a decade apart. Yet surely Selgin does not wish to dispute the noticeable dampening of both business cycles and inflation variability during the Greenspan era, a dampening that macroeconomists now refer to as"The Great Moderation." Consider the follow two graphs.
ZOOMFigure 1 (from the Federal Reserve Bank of San Francisco Economic Letter 2008-6, February 15, 2008) depicts the striking decrease after 1987, during the Greenspan era, in the volatility of real GDP (as measured by a five-year moving average of the growth rate's quarterly variance). Over the past twenty-five years, beginning in 1983, the U.S. economy has been in recession (as defined by the National Bureau of Economic Research) a mere 5 percent of the time, as compared with 22 percent of the time over the previous twenty-five years.
Figure 2
ZOOM
Figure 2 shows how this decrease affected the growth rate of quarterly nominal GDP (year-to-year annual rates). Notice that the major deviations of nominal GDP from trend after 1987 are during the three recessions (1990, 2001, and 2008) and are far less severe than previously. During the nineteen years from December 1986 to December 2005, nominal GDP growth averaged 5.6 percent, never varying from that in either direction by more than 2.5 percentage points. While this is not absolutely constant growth, it is about as close to constant as the U.S. economy has ever recorded.
What makes this decline in GDP's volatility particularly striking is that it occurred during a period in which most measures of monetary velocity were doing the exact opposite. Selgin acknowledges this well-known increase in velocity's variability, as it was the primary reason that macroeconomists--including most monetarists--abandoned the money supply as a way either to gauge inflation or to target central-bank policy. But if nominal GDP was becoming more stable at the exact same time that velocity was becoming more erratic, then the equation, MV = Py, ipso facto tells us that fluctuations of the money supply must have been offsetting fluctuations in velocity, at least partially.
The offset was reflected in the continuing or sometimes increasing volatility under Greenspan of the standard money measures, the very result that made central bankers despair of monetary targeting. Figure 3 shows the year-to-year annual growth rates of M1 (monthly, not seasonally adjusted), which after 1987 bounced around between 17 percent, higher than at any time during the 1960s and 1970s, and negative 5 percent.
Figure 3
ZOOM
Figure 4 shows the post-1987 growth rates of M2 and M3 (monthly, not seasonally adjusted) ranging between 0 and 13 percent, in a pattern not at all consistent with M1. Nor in either case does there seem to be any easily discernable correlation with the period's recessions.
Figure 4
ZOOM
This is what led Milton Friedman to marvel at Greenspan's performance in a Wall Street Journal article of April 19, 2003 (cited in our Cato Briefing Paper), where he displayed the graphs in Figure 5. They show how inflation remained relatively constant despite the M2 velocity bubble associated with the dot-com boom.
Figure 5
ZOOM
The counteracting decline of M2 growth shows up in Figure 4 above. (Our original paper regretfully did contain one bit of enthusiastic hyperbole in describing this outcome, which we can now correct. We stated that the M2 velocity bubble"was perfectly offset by the declining growth rate of M2." Obviously the word"perfectly" is too strong.) Normally an increase in the money supply's volatility would be associated with an increase in the volatility of nominal GDP, not the reverse. Clearly this development cries out for explanation. Our paper challenges the mainstream view among economists, which credits superb Fed discretion. We instead attribute it to a market process unleashed by financial deregulation.
Greenspan and the Productivity Norm
Selgin's critique goes further, maintaining that"final sales and other similar measures [of Py] would have been constant," rather than just growing at a constant rate, if the money stock had in fact adjusted for velocity. While literally true in the case where MV is absolutely fixed, this higher standard overlooks any growth of the labor force. Selgin's"productivity norm" as proposed in Less Than Zero (1997) requires only constant nominal GDP per capita, or something approximating that, which means that both Py and MV should be rising roughly at the rate of population increase.
Even adjusting for the moderate 1 percent annual growth of the U.S. population during the Greenspan era, Selgin's"productivity norm" conflates two separable goals, one long run and the other short run. His long-run goal, in which secular deflation at the rate of productivity growth yields a constant nominal GDP per capita, can be distinguished from his short-run goal of a money supply that is counter-cyclical with respect to velocity. While he would like to achieve both simultaneously, the economy can experience one without the other. Given a choice between (a) long-run mild deflation accompanied by wild oscillations in nominal GDP per capita around a zero growth path or (b) long-run mild inflation accompanied by no oscillations of nominal GDP per capita around a constant positive growth path, which goal would be Selgin's macroeconomic priority?
Some Austrian School economists, following in the monetary tradition of Ludwig Mises and Murray Rothbard, utterly reject Selgin's short-run goal. They contend that there is no need for the money stock to compensate for any shifts in money demand, whether arising from velocity shocks or output growth. All such shifts, in their view, can easily be accommodated by price adjustments. Selgin's Theory of Free Banking (1988), in contrast, built on Friedrich Hayek's focus on MV to introduce a Keynesian element into Austrian analysis. He predicted that an unregulated banking system would eliminate cyclical fluctuations in velocity (or in what Keynesians call"autonomous expenditures") and implied that such was necessary to avoid macroeconomic disturbances. Despite our initial reservations, we were persuaded of the validity of at least Selgin's prediction about the operations of free banking, which, ironically, is crucial to our analysis of Greenspan's monetary policy.
The validity of Selgin's long-run productivity norm is another matter. Although he offers it as policy prescription, it may not even be an accurate description of the way an ideal commodity money would operate in practice. White's Theory of Monetary Institutions (1999) lucidly elaborates on how changes in either the supply of or demand for gold as money would affect gold production and consumption in a way tending toward long-run price stability, not secular deflation. As a result, the market for gold as a commodity would dominate the long-run equilibrium path of the price level under a commodity standard. Depletion of gold sources might deliver mild deflation, but technological improvements in the extraction and recycling of gold or in the development of gold substitutes might deliver mild inflation. Recall that the introduction of the cyanide process for extracting gold helped cause a 2 percent annual inflation from 1896 until World War I.
Whatever the most desirable macroeconomic goal, whether an inflation target, a nominal GDP target such as Selgin's productivity norm, or no target whatsoever, our paper readily admits that Greenspan could have delivered"a constant price level or even secular deflation over the last two decades." He would have had to tighten up on growth of the monetary base, which would have caused a steady decrease in total reserves. This could have counteracted the declining reserve ratios of the monetary aggregates, a decline, as we explained, that resulted from a combination of financial deregulation and innovation and that was the main source of the period's 2.5 percent average inflation. But none of these long-run factors speaks to our observation that the short-run fluctuations in the money supply dampened the short-run fluctuations in velocity.
Selgin counters that financial deregulation was insufficient to achieve the result we observe. Freezing the monetary base will not work if"banks cannot issue currency and are subject to legal reserve requirements. . . . Consequently, the prevailing regime is one in which the avoidance of monetary excess or shortages calls for frequent changes to the monetary base, that is, for adherence to some more elaborate monetary rule, if not for monetary discretion [emphasis ours]." But as our article points out, reserve requirements are now virtually non-binding within the U.S. and on the road to extinction, whereas base increases after 1987 went entirely into currency (much of it going abroad), substituting for the currency that banks would have issued in the absence of legal restrictions. (See Figure 6, reprinted from our original paper.) Thus, Greenspan's unintentional freezing of total reserves approximated a freezing of the monetary base in a more deregulated world.
Figure 6
ZOOM
We detect a certain ambivalence in Selgin's objections. He condemns the freezing of total reserves for failing both to induce price deflation and to provide enough currency. Moreover, his rejection of simple monetary rules, unless accompanied by further deregulation, would also mean that he would, if consistent, oppose such other interim reforms as currency boards, which are supported by at least some advocates of free banking. Undoubtedly complete free banking, along with the elimination of the Fed and fiat money, would have yielded an even better outcome. Nonetheless, the decrease in the variability of nominal GDP during the Great Moderation is undeniable. Why an opponent of the Fed would seemingly attribute this improvement not to deregulation and free markets but to successful central-bank discretion is perplexing.
To be fair, perhaps Selgin wishes only to insist that as long as the economy has a central bank, it will always suffer some distortion of interest rates and inevitable self-reversing booms. That may very well be true. Yet does it follow that all central bank policies are equally harmful? A claim that malinvestment is constant and pervasive, but only revealed randomly with sudden depressions and recessions, of varying intensity and timing, sometimes back-to-back and sometimes separated by a decade, hardly qualifies as a complete theory of the business cycle. Unless the theory accounts for the timing and intensity of the downturns, it really has explained little.
Greenspan and the Current Crisis
Because our paper was concerned with monetary policy over the nearly two decades beginning in 1987, we made no effort to sketch out a complete explanation for the current financial crisis. It formed only a part of our analysis, although we did write an earlier op-ed on the subject for Investor's Business Daily, and one of us (Jeff) has recently discussed this question at the Liberty & Power blog. Our Cato Briefing Paper did, however, attribute"the unusually low interest rates early this decade mainly to a massive flow of savings from emerging Asian economies and elsewhere," supporting that claim with an extended footnote scratching the surface of the vast literature, both pro and con, on the global savings-glut thesis. As many, including Greenspan, have asked, how can Federal Reserve policy starting in 2001 be the primary cause of a housing boom that began in 1997 and was worldwide, with higher percentage price hikes in the U.K., the Netherlands, and Ireland than in the U.S.? This alternative hypothesis has been widely discussed among mainstream economists.
Selgin briefly alludes to John Taylor's rejection of the global savings-glut thesis, reporting correctly that global savings"declined as a share of world GNP from 25 percent in the 1970s to 21 percent between 2003 and 2005." This comes from a talk Taylor gave, "Housing and Monetary Policy," in September of 2007 (cited in our footnote). But this statistic is hardly decisive, given that it is net of any private savings covering government borrowing, that world savings has risen in absolute amount with only the world economy growing faster, that investment demand relative to world GNP also fell, and that the statistic says nothing about flows of savings between countries. Chapter 2 of the International Monetary Fund's World Economic Outlook for September 2005, which reports the statistic Taylor cites, discusses this decline at great length without rejecting the idea that a relative rise in Asian savings caused a fall in world interest rates. And everyone admits that whatever happened to world levels compared to world GNP, the period of low interest rates witnessed the unprecedented anomaly of savings flowing from less developed countries to more developed. Because the housing boom was international, this would seem to be relevant. For Selgin to invoke as definitive what is practically a throwaway line in a Taylor speech is a bit of a stretch.
For further documentation that has appeared since we wrote our footnote on the savings-glut thesis, we recommend a recent Financial Times article by Martin Wolf, "Asia's Revenge," and his new book from which it is drawn, Fixing Global Finance (although we hastily add that we do not endorse all of his conclusions). The cause of the worldwide fall in interest rates is still a debated question, and we invite readers to explore arguments on both sides. But that just reinforces the importance of evidence from the monetary measures, something that Selgin tends to dismiss. After all, to offer a monetary explanation for the current crisis, invoking some indicator of monetary policy is necessary. Austrian economists have long rejected the price level or inflation rate as the proper indicator, which is why Rothbard devoted so much of America's Great Depression (1962) to defining the appropriate money stock and documenting its increase over the 1920s.
Selgin rejects monetary measures because velocity has become erratic, and our paper agrees that he is mostly right with respect to the broader monetary aggregates. But to turn exclusively to interest rates is hardly a solution. One must have some idea of what the natural rate would have been without central bank interventions. Yet market rates are subject to so many influences from both demand and supply that no economist has been able to articulate a full explanation for their past, much less their current, levels. Forgive us if we remain skeptical of those who declare that interest rates alone tell the direction and magnitude of Fed distortions.
We do not need to become totally agnostic about Federal Reserve policy. But analyzing it requires a subtle examination of many factors, with assorted monetary measures receiving prominence. If Greenspan's policy was as unrestrained as commonly reported, you would think it would have left a monetary footprint somewhere. Our paper reported the inconvenient fact that the rate of growth of all the broader monetary aggregates was falling during the period when the Federal funds rate was unusually low. A few of our critics have replied:"So what? The decline was from earlier growth rates for M3, M2, MZM, M1, and even the base that were higher." Yes, but the Federal funds rate is not, like the price level or real output, a magnitude affected by the money supply with long and variable lags. It is the Fed's daily operating target, so the impact of an expansionary monetary policy should be instantaneous. Interest rates should fall at the exact same time that monetary growth rises. The fact that we instead observe money growth rates falling during 2002 and 2003, when the Federal funds rate was at its lowest, is a serious, unresolved problem for those who reject the global savings-glut thesis.
The attempt to blame Greenspan for the crisis actually has two variants. The simplistic version points to the low Federal funds rate from the end of 2001 to mid-2004 as evidence of a spouting monetary gusher worse than anything since before Paul Volcker took over the Fed. A quick perusal of Figures 3, 4, and 6 dramatically refutes the simplistic version. Selgin adheres to a more nuanced version, again citing the macroeconomist Taylor. According to the Taylor Rule, the Fed--after lowering the target Federal funds rate at the beginning of 2001-- should have begun raising it in the second quarter of 2002 (or a bit earlier if you change the specifications of Taylor's equation). This already reduces the dispute to technical issues of how much the Fed should have tightened and when. Moreover, it rests the case against Greenspan on a dubious tool of central bank discretion.
No one has written a pithier exposé of the Taylor pseudo-Rule than Roger Garrison, in correspondence with Liberty Fund that he shared with one of us: It"is not really a rule at all. Taylor, in effect turns an 'is' into an 'ought' by (a) describing the Fed's past actions in terms of its responses to changes in unemployment and inflation and then (b) telling the Fed how to make sure that its future responses are consistent with its past responses: 'If you like what you've done so far, here's how to keep doing it.'" Those who appeal to the Taylor Rule as proof of loose Fed policy after 2001 almost never mention that the same rule condemns Fed policy during the mid-1990s for being too tight, that is, for holding the Federal funds rate up too high by almost as many basis points as it later supposedly pushed the rate down.
As our paper fully concedes, the monetary measure that we believe has become the most informative--total reserves--is the one measure whose growth rate does indeed show"a slight uptick into 2003, when interest rates were down." Greenspan's policies thus may have been slightly more expansionary after 2001, making a minor contribution to a housing boom already in progress. We can quibble over how large that minor contribution was. But to make Greenspan's policies the sole or even primary cause of the current financial upheaval, international in scope, is just placing far too much weight on what, at worst, was a small discretionary misstep.
Conclusion
Selgin concludes his critique of our Greenspan retrospective with a question:"Why two anti-central bank libertarians would bother to undertake such a defense"? It has an interesting answer. The genesis of our paper dates to long before the current financial turmoil. Jeff wrote the first draft back in June 2004, while Greenspan, still chairman of the Fed, was being hailed as a financial maestro. Everyone recognized that the U.S. macroeconomy had become more stable after 1987, with respect to both inflation and recessions, and our motivation was to figure out why. As libertarians, we were suspicious of the dominant explanation that credited Greenspan's wise discretion. Only the onrush of recent events has converted what began as an unequivocal rejection of the need for any government monetary manipulation into a qualified defense of Greenspan's record.
And let us be clear: our defense is qualified. To use an analogy: as much as we loath the Drug Enforcement Agency and everything it does, we would be wary of any hypothesis that blamed the DEA for the explosion of subprime mortgages. If we then published an article about the weakness of the evidence linking the DEA to the subprime crisis, it would certainly not entail any diminution in our ardor for the DEA's abolition. Or to alter the analogy slightly, imagine a scenario in which marijuana, psychedelics, and cocaine are all fully legalized, but the DEA still enforces laws against opiates and other drugs. If we were to celebrate this limited legalization and its beneficial effects, it would hardly constitute an endorsement of the DEA's remaining activities in the war on drugs.
No one yet knows how the current financial turmoil will play out. There is still a chance it will generate a recession as mild as those in 1990 and 2001, despite everything the government has done to make matters worse. Or it could become as severe as the recession of 1982, signaling an end to the Great Moderation. Either way, the macroeconomic stability of the Greenspan era remains a puzzle. Some of our critics have suggested completely non-monetary explanations based upon the dramatic transformations in the American economy over the last two decades. For instance, the Internet and other innovations may have so significantly reduced transaction costs in the labor market that the high unemployment during recessions has been mitigated for reasons unrelated to monetary shocks. Greenspan himself, in his memoirs, gave much credit to globalization for the economy's two-decades increased resilience.
But none of these non-monetary explanations can account for the second piece of the puzzle. Why was the Great Moderation accompanied with greater instability of the broader monetary measures and their velocities? That instability should have been jolting the economy with frequent monetary shocks throughout the twenty years of the Greenspan era (unless you retreat into a theoretical fairyland of complete monetary neutrality, where nothing that happens to the money supply has any real impact). So far only two alternatives have been put forward to explain both pieces of the paradox: intentional monetary discretion or unintended monetary rules. Despite Selgin's criticisms, we still opt for our invisible-hand alternative.
As our paper indicated, we believe that government fiat money is untenable in the long run. The current economic crisis confirms our belief, particularly the under-explored role played by fickle capital flows under a fiat regime, as suggested by the global savings-glut thesis. Minimizing the discretion of the Fed by chaining it to a monetary rule, so as to formalize what Greenspan inadvertently did for the most part during his tenure, is only a short-run improvement as long as the Fed exists. The ultimate solution is free banking, among whose foremost champions is George Selgin.
[Cross-posted at EconLog.]