Bernanke´s GSG hypothesis: A cop-out

Bernanke is at it again. In a speech last friday and in a new paper, he gives new life to the global savings glut hypothesis (GSG) that he first put forth in 2005 as an explanation for Greenspan´s “Conundrum” (the fact that after mid 2004, with the Fed increasing the policy rate (FF), long term rates “stayed put”). This time he extends his original arguments by considering the type of assets desired by the emerging economies (the savers) as they recycled their dollar reserves into US investments.

Indirectly, this discussion, which has been going on for several years, smacks of an effort to absolve the Fed from responsibility for the financial crisis which erupted with the bursting of the house bubble. For many (see this post by David Beckworth), bad monetary policy in 2002-04 (“rates too low for too long”) is a major factor behind the housing boom and ensuing crisis.

My take is very different. First, I don´t think “low rates” were connected to “savings glut”; second, I don´t think monetary policy was “easy” in 2002-04 and third, I don´t think these two factors had much influence on the housing price boom (and it may not even right to classify it as a “bubble”).

It certainly has been fashionable, especially given the GSG hypothesis, to fault Asians in general, the Chinese in particular and an assortment of other developing and emerging economies, not only for those previous problems but also for present day ones!

I like to think 1997/98 was the moment when “it all started”. The major economic event of those final millennium years was the Asian crisis, quickly followed by the Russian and Brazilian ones. As the figures show, coinciding with those events, there developed a large gap between savings and investments in a broad set of countries.

But this state of affairs did not fall out of the blue. It was the successful result of the policies and actions recommended by the IMF and others (see, for example, Martin Feldstein´s “A Self-Help Guide for Emerging Markets”, Foreign Affairs March/April 1999) following those events.

On the other side of the “adjustment equation” was the US, the only major economy growing robustly at the time (Japan was in the midst of its long slump and Germany/Europe was wobbly). This state of affairs – a growing current account deficit in the US to counterbalance a growing surplus in emerging markets – was baptized by Michael Dooley, Peter Garber and David Folkerts-Landau in 2003 as “Bretton Woods II”. To call this an “international imbalance” sounds strange since it was the “balancing” mechanism.

If an unintended consequence of the adjustment process was fraud (from lack of supervision) in addition to government giving the wrong incentives by “encouraging homeownership to all”, that is a wholly separate issue. Krugman suggests that the crisis is the Revenge of the Glut. I prefer to ascribe it to “payment due for sins of omission, false testimonials and political greed”.

Whilst in Asia and emerging markets resources were being diverted from the non tradable to the tradable sector, in the US the opposite movement had to occur. In 1997, in addition to all the “incentives to homeownership”, the US government abolished capital gains taxes on homes sold after two years. Given the need to transfer resources to the non tradable sector, this opportunity was too good to be ignored. So, not surprisingly a “housing boom” ensued.

The figure below shows national house prices (Case-Shiller). It is clear that the rise in prices starts in 1997, long before the “too low for too long” rates were in place. But the national house price index masks very different price behavior in different regions, states and metro areas. Compare, for example, Texas and California, or Nevada and Arizona. Population in Texas grew almost double the rate in California but house prices in Texas behaved very differently. Both Arizona and Nevada experienced strong population increases but again, while house prices in Nevada shoot up after 2002 and then drop like a stone after 2006, prices in Arizona behave much more “smoothly”. One important reason for these very different price behaviors is the presence of “zoning laws” or land development restrictions. California has them in abundance, while land development in Texas is mostly uninhibited. Nevada is a good test case. With most of the land federally owned, until the early 2000s new development permits were easily obtained, so prices were “well-behaved”. That changes in 2002, following pressure from environmentalists. It´s a no brainer: prices go up when supply restrictions become strong. 

In the midst of all these events, what was happening to monetary policy? In 1997 the “Great Moderation” was in full swing. The graph below characterizes this fact by showing that nominal spending was evolving along a stable level growth path. Maybe tricked by financial fall-outs of the Asian/Russian crises, in 1998 the Fed allows spending to grow “excessively” (i.e. monetary policy becomes “easy”). A correction begins in late 1999, but there is an “overshoot” and monetary policy becomes “tight”. By the end of 2002 nominal spending (aggregate demand) is 2% below trend.

The fact that nominal spending begins to grow back towards trend at this point is an indication that monetary policy was “correct”. After all, the trend (level) path should be the “target”.

The (in)famous 2002-04 period is graphed below. Let´s start from the end. Between June 2004 and July 2006 the FOMC raised rates from 1% to 5.25% by increments of 25 basis points in all 17 meetings. In his February/05 H-H testimony before Congress Greenspan used the term “conundrum” (Latin for enigma) to describe the fact that long term (including mortgage) rates had not followed suit. The story ends in July 2007 to exclude the crisis period that began in August with the Bank Paribas hedge funds fracas. 

Funny that there are few mentions of the “reverse conundrum” that took place between January 2001 and May 2002 (a notable exception is this Cleveland Fed Commentary of January 2002 by Altig and Nosal). While the FF rate was brought down from 6.25% in December 2000 to 1.75% in January 2002, the long rate did not move. But just at that point inflation began to fall and the 10 year bond rate decreased. Bernanke´s famous “helicopter Ben” speech in November 2002 followed by the FOMC´s statement at the May 2003 meeting  where it was said that “… In contrast, over the same period, the probability of an unwelcome substantial fall in inflation, though minor, exceeds that of a pick-up in inflation from its already low level…” helped keep long rates at this somewhat lower level (around 4%).

On the way down and than on the way up, the FF rate did not directly affect the long rate. Is this tied to the GSG hypothesis? I doubt it. In my view, all the commotion around the “conundrum” is tied to the consideration that, according to Greenspan, “it is a well accepted fact that the absence of change in the long rate at the same time that the short (FF) rate changes significantly defines a break in the historical relation between those rates”.

Can we observe this “fact” in the data? A simple way to check if long rates usually move in tandem with the FF rate is to see if the ratio of the 10 year rate to the FF rate stays close to 1 at all times.

The graphs below are illustrative of an apparent change in the relation after 1990. Between 1965 and 1990, the ratio of the 10 year rate to the FF rate never drifts for too long and keeps very close to 1. This is true for the rising inflation period (1965-80) where there are large swings in rates as well as for the falling inflation period (1982-90).

However, this is not so after 1990 despite much smaller swings in rates. During 1992-94 the ratio remained persistently and significantly above 1 and even more so during 2002-04.

More formal (regression) analysis supports this result. While for the whole 1965-90 period as well as for the shorter 1982-90 period of falling inflation, the slope coefficient is statistically significant and the monthly variation in the FF rate also explains a not insignificant portion of the variation in the long rate (15% to 20%), for the post 1990 period the slope coefficient is not statistically significant and the variation in the FF rate explains less than 1% of the monthly variation in the long rate!

It appears that if we can find the reason for this marked change in the behavior of rates we will have “solved” the “conundrum”. The figure below graphs the same variables only now for the period 1955 -64. Contrary to what was observed for 1965 -90 , the 10yr/FF ratio also shows persistent and significant departures from 1, much like the 1991 – 07 period.

What does the 1955 – 64 period have in common with the post 1990 period? Both periods are characterized by low (falling) inflation and, importantly, the absence of expectations of rising inflation. In other words, in both periods the monetary authority possesses credibility. Fifty five years ago inflation was not something that “kept people awake at night”. Less than 25 years before the Great Depression had brought deflation to the fore. More recently, credibility was obtained with great effort and many years of hard work.

But Greenspan didn´t see it or, if he did, did not believe in this very compelling evidence for “solving” the “conundrum” (best put the “blame” far away through GSG) because at the very end of the “Conundrum” chapter (Chap. 20) in his book – The Age of Turbulence –he writes:

Central bankers over the past several decades have absorbed an important principle: Price stability is the path to maximum sustainable economic growth. Many economists in fact credit central bank monetary policy as the key factor in the last decade´s reduction in inflation worldwide. I would like to believe that. I do not deny that we adjusted policy to be consonant with global disinflationary policy as they emerged. But I very much doubt that either policy actions or central bank antiinflationary credibility played the leading role in the fall of long term interest rates. That decline (and the conundrum) can be accounted for by forces other than monetary policy

Why is it that central bankers – this was the case of the Marriner Eccles Fed in the 1930´s, the Burns Fed in the 1970´s – like to put the “blame” on “nonmonetary forces”?  Go figure that!

Instead of “rehashing” the past, Bernanke should write about the mistakes he made in 2008, when he let nominal spending take a dive. But he can´t very well do that because he had already written about this situation with regards to Japan in 1999. So, instead of Beckworth´s four questions for Bernanke I only have one: Why didn´t you do in the US as Fed Chairman what you said should have been done in the case of Japan as an academic analyst/researcher?

10 thoughts on “Bernanke´s GSG hypothesis: A cop-out

  1. Marcus:

    Another great post. I agree that a more important question than the ones I ask is why the Bernanke Fed did not do more in 2008-2009 to stabilize nominal spending.

    A couple of comments. On the rising home prices, yes, they do start rising in the late 1990s, but they really start accelerating in early-to-mid 2000s. It is this acceleration that is being questioned.

    On the accumulation of foreign reserves, I would make the case they were precautionary in nature early on following the 1997-1998 crisis. But by early 2000s, the mercantilist motive was more important.

    • David
      Thanks for the encouraging comment. On home prices, between 1998 and 2002, prices went up 41%. Between 2002 and the peak in late 2006 they went up 45%. So they went up about equally before and during the “low rate” period.
      On the “mercantilist motive”, I just don´t know. Beginning in 2002 there was the China factor – a strong rise in Chinese imports – that certainly “lifted” many emerging markets.

  2. Very good post, Marcus. I agree with you, so I not understand you thought we disagree!
    I mmust think a lot to answer you with some rigor.
    But for the moment I don´t see great contradiction with you and Bernanke´s speach.

  3. Until now, I use to think that GI had no role in the crisis. And I continue to think so.
    But for me it is clear the origin of the flow of funds that entry in US. Now, Bernanke doesn´t say that this was the causeof the crisis. There were great mistake in regulatio and supervision.
    On the other hand, I don´t agree with the hipothesis of the only culpability of Bernanke because his NGDP fall (Sumner) I think that other fault lines, as the colapse of some short term market, payed an important role-

  4. Luis
    I don´t know what GI means. On the last part, if not Bernanke (the Fed), who´s responsible for the huge NGDP drop? Remember that an important part of the colapse of some short term markets you mention would be a consequence of the drop in NGDP.

  5. Global Imbalances, sorry.
    Yes, but the case “Bernanke responsable of huge NGDP drop” dosen´t contradict the case for the Glut of global saving searching for a yield when interest rat were falling. That explain why so much countries without housing bubble (Germany, Sweden) have had a huge financial crisis.

  6. Luis. Funny, they call it Global Imbalance but as I argued it was the “Balancing Mechanism”! And I don´t believe, as argued, that it had a direct role in the crisis either. Maybe indirect, by establishing the conditions that allowed the “foul play” to become prevalent.

  7. yes, I don´t say it was a factor, but a condition. Just until now (that I´ve red Fault Lines) I negleted Global Imbalances. But Rajan conviced me: The origin of the flows was GI, caused in turn by the exported oriented countries and their merchantilist devaluations. These have had some curious global effects, too long to explain now.

  8. Excellent analysis. I just have two small points.

    1. I believe the Arizona housing prices may be off. Phoenix prices fell by more than 50%, and the Phoenix area has most of Arizona’s population. I am pretty sure they also had some zoning issues, just like Vegas.

    2. It seems to me that the statistical results would be sensitive to whether you chose the ratio of long rates to the ff rate, or the absolute difference. It’s not obvious why the ratio would be better, as changes in the trend inflation rate will move both rates equally, in absolute terms. That would always tend to make the ratio closer to one at higher nominal rates.

Leave a Reply to João Marcus Marinho Nunes Cancel reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.