This got way too long, like my initial efforts at H money. Maybe just read the bullets, check out quotes at the top of the piece and then look at the second last chart, stolen from Bernanke, who totally burns Taylor and should probably have dropped the mic.
- John Taylor has been claiming for the past several years that the housing and credit bubble of the mid-2000s could have been prevented had the Fed followed a Taylor Rule or similar rules-based approach to policy. But the evidence is strongly against that claim, as I document in this post.
- Former Fed chairs Greenspan and Bernanke are probably right when they argue (effectively) that the fundamental displacement behind the bubble was the development of surplus savings in developing Asia and OPEC, which were recycled, in part by central banks, into US capital markets.
- Those capital inflows pushed up the dollar and created both a disinflationary pulse and a prolonged drag on aggregate demand through the trade sector. The Fed responded by delivering a lower path for the fed funds rate than would have otherwise been appropriate, in order to crowd demand into the domestic sectors and offset the direct price effects of the stronger dollar.
- This policy “worked”, in the sense that growth remained firm during the mid-2000s, the unemployment rate moved steadily lower, and core inflation did not undershoot the Fed’s long-term objective of 2%.
- Because the policy “worked”, it allowed a wedge to develop between depressed interest rates and continued favorable overall macroeconomic performance. This predictably put upward pressure on domestic asset prices, particularly housing, which eventually morphed into the bubble. Briefly, then, the bubble was caused by the Fed ratifying the effects of the capital inflows.
- For a while, housing and its bubble were the crowding in. The housing boom allowed the household sector willingly to run the large financial deficit that was required to sustain moderate aggregate demand growth, despite the drag associated with the foreign capital inflows. To the extent, however, that sustaining the household sector’s financial deficit ultimately relied on bubble dynamics, the process was unsustainable, as we learned the hard way.
- Former Fed Chair Greenspan is probably wrong when he claims that the Fed was helpless to stop the bubble and that it was imposed on the US by foreign capital inflows and the yield “conundrum.” There existed at the time a funds path that would have stopped the housing bubble cold, although at the cost of higher unemployment and an inflation undershoot during the mid to late 2000s.
- Instead of delivering that policy, the Fed allowed the housing bubble to get traction by following the underlying logic of the Taylor Rule. I am going to call that underlying logic the Taylor Principle, taking a minor liberty with its formal definition in economics. * The Principle, as I conceive it here, urges monetary policy makers systematically to lean into disinflationary and contractionary forces irrespective of their cause, i.e. without discretion.
- This is not to criticize the behavior of the Fed or the application of the Taylor Principle during the mid-2000s. Monetary policy cannot fix all problems and perfect foresight is an impossible standard. But this experience puts the lie to the idea that the bubble could have been prevented by the application of Taylor. They applied Taylor and we got a huge bubble.
- It is the underlying logic the Taylor Rule and not its specific parameterization that matters most in this context, as John Taylor has repeatedly emphasized. Still, the Fed seems to have followed a fairly standard, “balanced”, parameterization during the mid-2000s. This further undermines John Taylor’s claim for his Rule, particularly given that he explicitly endorsed that parameterization in 1999.
- The idea that bubbles can easily be warded off by tweaking monetary policy seems both silly and dangerous. It is dangerous because somebody might try to apply it. It is dangerous also because it distracts attention from other initiatives, such as regulation, that might actually succeed at stopping bubbles.
- I suspect this distraction may, in many cases, be intentional – because there seems to be a tight overlap between people who resist sensible regulation and those who insist that monetary policy fight bubbles. But I don’t assume that applies to Taylor and do not elaborate on it here. I may do so in a follow-up post.
I agree with their overall evaluation of Alan Greenspan that “when the score is toted up, we think he has legitimate claim to be the greatest central banker who ever lived.”
John Taylor, August 2005
John argues that the (Greenspan) FOMC kept interest rates much lower than prescribed by the Taylor rule during 2003-2005, and that this deviation was a major source of the housing bubble…
Ben Bernanke, August 2015
The Taylor Rule is no cure all
The main effect of the Taylor Rule and other reductionist guides to monetary policy over the past several years has been to make the hawks overconfidently wrong about the outlook for interest rates. Regardless of how these rules are parameterized, it would be dangerous to rely on them in the neighborhood of the zero bound on interest rates. And so the Fed has not.
In this post I address a related but somewhat separate issue, the claim that applying the Taylor Rule would have prevented the housing and credit bubble of the mid-2000s and thereby saved us the pain of the crash and Great Recession.
That claim dates to well before the recent federal election and achievement of full Republican control of DC. For example, John Taylor was making a stink about this issue as early as 2009, if not earlier. However, the claim has taken on more prominence recently, because of the move afoot in Congress to impose on the Fed a requirement that they at least explain their policy in terms of a rules-based approach.
Here is John Taylor in April of 2015 reviewing on his blog a debate he had just had with Ben Bernanke and others over the value of a rules-based approach. For context, Taylor presumes that policy followed the Taylor Rule sufficiently closely until the early 2000s and that there was then a break:
But then there was a setback. The Fed decided to hold the interest rate very low during 2003-2005, thereby deviating from the rules-based policy that worked well during the Great Moderation. You do not need policy rules to see the change: With the inflation rate around 2%, the federal funds rate was only 1% in 2003, compared with 5.5% in 1997 when the inflation rate was also about 2%. The results were not good. In my view this policy change brought on a search for yield, excesses in the housing market, and, along with a regulatory process which broke rules for safety and soundness, was a key factor in the financial crisis and the Great Recession.
Admittedly, that passage from Taylor’s Economics One blog does not present a fat target. Taylor lumps monetary policy in with regulatory failure and then describes the sum of the two as merely key factors in the financial crisis and Great Recession. But it seems obvious from the context – of the IMF session and the debate more broadly – that Taylor wants to imply that the Taylor Rule would have prevented the bubble and crash. He made precisely that claim, quite clearly, in 2009 and has returned to it several times since then.
Certainly former Fed Chairman Bernanke has my interpretation. Here is his take on Taylor’s post-debate summary, posted to Bernanke’s blog at Brookings:
John argues that the FOMC kept interest rates much lower than prescribed by the Taylor rule during 2003-2005, and that this deviation was a major source of the housing bubble and other financial excesses.
Based in part on a personal experience, I find Taylor’s claim here not just extremely dubious and revisionist but also jarring. I remember asserting to colleagues back at the hedge fund, I think maybe around 2004, that the Fed’s insistence on following the Taylor Principle — in the face of a contractionary / disinflationary shock from Asia – would give a fundamental lift to home prices, which markets might eventually get extrapolated into a housing bubble.
I might have been wrong to believe, as I still do, that the housing and credit bubble did not get going properly until later in the cycle, as home prices continued to rise even with rates moving higher. And I certainly failed to appreciate the credit side of that bubble and the disaster that would arise from it. But I clearly recall thinking at the time – and using the very words – that the logic of the Taylor Rule would be a cause of a bubble. So when John Taylor comes along and presents as prevention what I took in real time to be a cause, I find it jarring.
In the remainder of this post, I will explain and partly document my take on the role that monetary policy played in allowing the bubble to develop in the mid-2000s. Let me summarize the argument here, so that you may more easily follow its elaboration below.
As I and many others see it, the fundamental displacement creating the housing and credit bubble was the development of surplus savings in Asia and OPEC, which were recycled, in part by central banks, into the US capital markets. Whether they came as equity, mortgage or Treasury-related flows is largely irrelevant, and I will not get into that here.
The point is that these flows put upward pressure on the dollar and thereby created both a disinflationary pulse and chronic drag on US aggregate demand through the trade sector. The capital flows and dollar move also reflected relatively weak domestic demand in the capital exporting countries, which further exacerbated their contractionary effect on the US trade sector.
While this shock was unfolding, the Fed retained its focus – implicitly following Taylor – on hitting its medium-term inflation and employment objectives. So it reacted the shock by attempting to crowd aggregate demand into the domestic sectors and create a little home-grown inflation pressure. The means to this end was to deliver a funds rate path that was lower than would otherwise have been appropriate, and which actually ratified the downward pressure on bond yields occasioned in the first instance by the capital inflows.
Viewed in real time, the policy “worked” at preventing: a slowdown of economic growth, an inflation under-shoot or reversal of the trend toward lower unemployment. And because the policy “worked”, it created a wedge between unusually low bond yields and macroeconomic conditions that looked fairly benign. In this environment, it made perfect sense for home prices and housing activity to rise. Indeed, they were the domestic crowding in at which the Fed was aiming.
Accordingly, for right or wrong, I did not judge the housing market as a bubble in 2003 or 2004, although it sure looked like one by late 2005. (Policy makers need to be more forward looking, obviously.) But the salient point here is that the interaction of the capital flow displacement with the Fed following Taylor is what got the housing and related credit bubble going.
Before elaborating on that basic outline, I want to emphasize four basic aspects of it:
- First, the fundamental / primary cause of the housing bubble was the development of “surplus” savings in Asia and OPEC along with foreign central banks’ decision to recycle those savings into the US capital markets.
- Second, the Fed transmitted that shock into a housing bubble by following at least the Taylor Principle and very arguably a specific parameterization of the Taylor Rule which Taylor had himself endorsed.
- Third, it does not necessarily follow from this that the Fed acted badly, if we accept the premise – as I do – that the Fed cannot fix all problems or be expected to operate with perfect foresight. But the idea that the application of Taylor would have prevented the housing bubble seems utter nonsense.
- Fourth and closely related, it is probably disingenuous of Alan Greenspan to claim that he had no control over the incipient housing bubble and that the result was imposed on him by capital flows and the yield conundrum. There exists a funds rate path that would have stopped the bubble from developing. But the Fed chose, quite understandably, not to impose it implicitly follow Taylor instead.
Capital flows impose financial imbalance
The idea that there arose a “surplus” of global savings during the early to mid 2000s and that those savings were recycled into the US is pretty well established. See here for Greenspan’s relation of capital flows to the yield conundrum, offered in testimony to Congress in 2005. And see here for Bernanke’s original elaboration of the savings glut thesis in 2005 and then his update in 2007. In 07, he went to Berlin to yell at the Germans for having become partly to blame.
I don’t have much to add to this, beyond agreeing with their basic premise – if perhaps not fully where they go with it. I think it’s fun that Bernanke caught an inkling of the potential damage from the capital flow shock back in 2005, but failed to pursue it very far. His main point then was that policy had to react to the shock and coud probably do so without major fireworks. He worried about how that might affect the “sectoral” composition of aggregate demand in the US, but did not connect that to the housing and credit bubble and the risks that posed to macro stability. Having been there to some extent myself, I assume he would wince to re-read this:
A third concern with the pattern of capital flows arises from the indirect effects of those flows on the sectoral composition of the economies that receive them. In the United States, for example, the growth in export-oriented sectors such as manufacturing has been restrained by the U.S. trade imbalance (although the recent decline in the dollar has alleviated that pressure somewhat), while sectors producing nontraded goods and services, such as home construction, have grown rapidly. To repay foreign creditors, as it must someday, the United States will need large and healthy export industries. The relative shrinkage in those industries in the presence of current account deficits–a shrinkage that may well have to be reversed in the future–imposes real costs of adjustment on firms and workers in those industries.
Ouch, that was definitely not the right sectoral issue to highlight. With the advantage of hindsight, the threat circa 2005 was not that foreigners would cut off access to foreign capital or that Americans would have trouble in aggregate paying back foreign creditors. (The US borrows in dollars after all.) The problem was that the US’s ability from a purely domestic perspective to accommodate these capital inflows was limited and perhaps required the development of a bubble, as I hope will become obvious below.
FWIW, I buy Bernanke’s saving glut / capital flow shock story, but I have never been satisfied with his documentation of it. He tends just to point out that Asia, OPEC and more lately German had current account surpluses and that the US had a correspondingly large deficit.
Accounting identities do not identify the direction of causation, although country to a popular assertion these days they certainly do imply codetermination. A similar pattern of current account balances might easily have been observed had the US had a savings shortfall, as might have arisen if the housing bubble had domestic origins. But the fact that this pattern was observed at a time of low US interest rates is consistent with Bernanke and Greenspan’s interpretation, so I will leave it there.
My attempt at a contribution here is to look at some domestic reflections of those foreign capital inflows. Take a look at the top left panel of the chart above. It shows that the domestic private sector’s financial balance deteriorated from a surplus of about 2% of GDP in the mid-1990s to an average of about -2% of GDP during the late 1990s and early to mid 2000s.
A simple accounting identity defines that balance as the difference between the current account balance and fiscal balance. The fiscal balance oscillated up and down with the cycle and with policy, but on net all of the deterioration of the domestic private financial balance can be traced to the capital inflows and the resulting current account deficit.
The overall private domestic financial balance is also the sum of the financial balances in the household sector and the business sector, corporate and otherwise. So if the difference between the current account and fiscal balances is a deficit, then either the household sector, the business sector or both have to be running a deficit. Which of these ends up bearing the burden of adjustment cannot be determined by accounting identities, but reflects – among other things – how the various sectors respond to price signals and economic conditions.
Going through all the causal linkages there is beyond the context of this post, which I concede makes this post incomplete. But as it turns out, the household sector accommodated more than all** of the adjustment imposed by the foreign capital inflows and the relative passivity of fiscal policy from the early 1990s to mid-2000s.
The household sector’s financial balance deteriorated from a surplus of about 4% of GDP to a deficit of about 1% of GDP. This reflected both the downward pressure on personal savings occasioned by the asset price boom, including that in housing, as well as a surge of residential investment. That this was achieved in an environment of roughly full employment over that entire period is evidence that the required crowding in, of which the housing boom was part, “worked.”
To repeat, the story I tell above is much more descriptive than analytical and you have to bring some strong priors to this to accept it. But I think it is a fairly standard interpretation, so in the interest and time and space I will just leave it at that.
The Fed’s role
The link between monetary policy and the household sector’s large financial deficit during the bubble era is complicated and I give it short shrift here. But at bottom, the Fed had to deliver a funds path that was consistent with that large financial deficit being willingly sustained or “desired” ex ante.
Had the Fed chosen to deliver tighter policy than it ended up delivering, the housing activity would have been lower, desired savings would have been higher, and consumer spending growth substantially weaker. The resulting reduction of the household sector’s financial deficit would have been reflected in what I have chosen to now to describe as its determinant, that is the difference between the current account balance and the public sector financial balance. Weaker domestic demand would have reduced the current account deficit, and slower growth would have pushed up the fiscal deficit via the effects of automatic stabilizers.
However, the more salient point facing the Fed is that these accounting identities would have held, as they invariably do, at lower levels of national income, employment and probably inflation.
The choice the Fed faces was to a large extent that between triggering and then sustaining bubble dynamics in the household sector on the one hand and allowing the paradox of thrift to take hold on the other. It chose the former, following the logic of the Taylor Principle. The precise funds rate path that arose as part of that choice is secondary. But it stands to reason, as both Greenspan and Bernanke pointed out that things had to balance at a lower level of interest rates and bond yields than would have otherwise been appropriate.
Greenspan’s claim that the bubble was imposed on the Fed by foreign capital inflows and resulting yield “conundrum” can be assessed in the context of this fundamental choice facing the Fed during the early 2000s. While the choice Greenspan made was understandable, the story he tells is an evasion.
Greenspan perhaps had little control over interest rates at the long end of the yield curve, but he could have avoided the fundamental displacement occasioned by those inflows by raising interest rates sooner and more quickly and thus by refusing to ratify their effects on domestic financial conditions. Or to put it more briefly, he could have radically inverted the yield curve had he chosen to. That would have stopped the bubble, but at the cost of higher unemployment, an inflation overshoot, and perhaps even an “early” recession. The paradox of thrift, described above, would have resulted.
That Greenspan and Bernanke after him chose not to pursue that course almost certainly reflects that they were targeting overall demand growth and its reflection in employment and inflation rather than the composition of aggregate demand or its link to bubble risks. Given that the Fed does not have perfect foresight and given further that interest rate policy is a very ineffective way to deal with bubbles, they could hardly have been expected to steer a different course.
Taylor would not and did not argue for bubble fighting
More to the point, their approach was fully consistent with the spirit and perhaps even a specific parameterization of the Taylor Rule. Taylor, after all, urges policy makers to follow a rules-based approach, where pressure on inflation and unemployment are systematically offset by interest rate changes, regardless of the cause of those pressures, i.e. without discretion. With that in mind, let’s take a look at how the Fed behaved through the prism of Taylor, both the spirit of what he urges and a couple parameterizations of his Rule.
In his original paper from 1993 advocating what eventually became known as the Taylor Rule, John Taylor was crystal clear that adhering strictly to a specific parameterization would be impossible and that central bankers should be allowed to choose, implicitly, the parameterization that they judged appropriate.
For Taylor, changing parameterizations was not discretion in disguise, but was instead a huge improvement over discretion, because it remained systematic. He actually expressed surprise that what came to be known at Taylor (1993) seems to have predicted Fed policy so well over the period he considered, which was only 1987 to 1993. The idea that Taylor had adopted a strong commitment to the 1993 version of his Rule by 19993 or even much later, then, is purely revisionist. These three passages from the 1993 paper the point pretty clearly:
An objective of the paper is to preserve the concept of such a policy rule in a policy environment where it is practically impossible to follow mechanically any particular algebraic formula that describe the policy rule.
If there is anything about which modern macroeconomics is clear however – and on which there is substantial consensus – it is that policy rules have major advantages over discretions in improving economic performance. Hence it is important to preserve the concept of a policy rule, even in an environment where it is practically impossible to follow mechanically the algebraic formulas economists write down to describe their preferred policy rules.
Although there is not a consensus about the size of the coefficients of policy rules, it is useful to consider that a representative policy rule might look like. One policy rule that captures the spirit of recent research and which is quite straightforward is: r = p + .5y + 0.5 (p-2) + 2. (Ed note: this came to be known at Taylor (1993).
Note from the first passage in particular that the policy rule is the idea, not the specific parameterization. In a follow-up paper written in 1999, Taylor reinforced that point by insisting that the coefficient describing the reaction to deviations of real income (or output) from target could be set at any non-negative value, at least in principle. And he specifically identified what would later be described as the “balanced approach”, where the coefficient on output deviations is higher, as acceptable. I will get to that further below.
So with that in mind, let’s turn to a couple pictures that look at the bubble period through the prism of the Taylor Rule, parameterized in a couple different ways. The left panel in the chart below compares the fed funds rate with the level prescribed by a rule very close to Taylor (1993). The only differences are: that inflation is measured with the core PCE deflator rather than the GDP deflator, because the Fed targetst the PCE deflator; that the output gap is measured using the CBO’s estimate of potential rather than with a time trend; and that the data are current vintage, rather than real time.
The right panel shows simply the difference between the actual fed funds rate and its Taylor (1993) – prescribed value.
Two points are immediately striking. First, and most importantly, Taylor (1993) prescribed a fairly steep decline of the fed funds rate between the late 1990s and the early to mid 2000s, during which the housing boom (bubble) really got going. Taylor (1993) respected the logic of what I am calling the Taylor Principle.
In fairness, Taylor (1993) prescribed a much shallower decline of rates than was actually delivered, as well as an earlier “renormalization.” But that brings us to the second point. The extent to which policy was “too easy” during the during early- to mid-2000s is not really that remarkable relative to other “misses”, as least when Taylor (1993) is run using revised data, which are presumably closer to “true.” For example, policy was chronically too tight, by the Rule, during the decade to 1994.
I work up my own measure of Taylor (1993), using current vintage data, because that allows me a longer-run of simulation to compare the 1990s with 2000s, as mentioned above. But Ben Bernanke presents on his blog a more precise set of Taylor simulations that use real time data and are likely to strike you – I hope – as pretty devastating to Taylor’s main point. My value added here, if any, is to bring Bernanke’s pictures to you and to be less polite than the former Fed chair.
The left panel in the chart above shows Taylor (1993) using real time data. During the relatively short period for which a comparison is possible, Bernanke’s simulation with real-time data looks pretty similar to my own using current vintage data. I guess that means the data revisions have been fairly small relative to the variation of the data over time. As simulated by Bernanke, the prescribed tightening after 2010 comes a bit later and is a bit shallower that I simulate, I assume because measures of potential GDP have been revised down along with the output gap. But it is a minor point.
Now take a look at the right panel. It shows a real-time simulation of the Taylor-prescribed funds rate, but with the coefficient on the output gap raised from 0.5 to 1, as John Taylor was clear back in 1999 might be acceptable. Bernanke cites Yellen in describing this parameterization as the “balanced” approach.
I don’t know if it is fluke or design, and Bernanke’s blog post on this issue was not clarifying. But it sure looks like policy heading into the bubble did not deviate much from the balanced approach, which – to repeat – Taylor had said in 1999 might be acceptable.
Indeed, in a WSJ editorial in late 2005, John Taylor expressed great satisfaction with Alan Greenspan having followed his principles (rather than Taylor-93 in particular) and made this perhaps-regrettable observation:
Alan Blinder and Ricardo Reis have provided us with a comprehensive evaluation of the Greenspan era, shedding light on key policy issues and controversies. I particularly liked their behind-the-scenes review of the move toward greater transparency. And I agree with their overall evaluation of Alan Greenspan that “when the score is toted up, we think he has legitimate claim to be the greatest central banker who ever lived.”
Or not. From Economics One in December 2009.
And in a follow-up op-editorial in mid-2006, he praised Bernanke for getting off to a good start, by following the Taylor Principle, although he strongly advised him to stop giving forward guidance or claiming that the Fed knew the appropriate inflation target (!), i.e. one of the constants in any parameterization of the Taylor Rule.
I pick on Taylor because his revisionism makes him a soft target. But the broader point here is that the most prominent advocate for monetary policy rules in the US, and the inventor of the Taylor Rule, was under the impression that the Fed was following the appropriate rules-based strategy as late as 2006. The evidence is actually consistent with that interpretation.
The Fed certainly followed what I am calling loosely here the Taylor Principle, which predictably allowed the capital flow shock to trigger a housing boom, which later morphed into a bubble. And as Ben Bernanke has pointed it delivered a rates path during the mid-2000 that, intentional or otherwise, was consistent with a parameterization of the Taylor Rule that Taylor himself had specifically said was tolerable – or indistinguishable from intolerable.
This experience certainly does not imply that the Taylor Rule is well suited to stopping bubbles. So we can add this to the list of things the Taylor Rule can’t do, particularly in the current environment.
* Strictly speaking, the Taylor Principle holds that the Fed or any other central bank should lean into deviations of inflation from target by moving the policy interest rate more than one for one with those deviations. In other words, higher inflation should bring a higher real interest rate and vice versa. The Principle itself is silent on how the Fed should react to deviations of output or employment from their respective normal values. However, in this post I refer to the Taylor Principle as “broadly” urging that policy lean into deviations of both inflation and output, regardless of their source, because all practical discussions of Taylor assume it does just that.
** I do not get into the business sector financial balance here because it did not play a major role in the early to mid 2000s bubble, although it was certainly implicated in the NASDAQ bubble, which was associated with an over-investment cycle. With the business sector financial balance well controlled during the 2000, it was up to the household sector to absorb the shock associated with the capital inflows and relative passivity of fiscal policy.