Thursday, August 25, 2011

Speak Softly But Carry a Big Stick, Dr. Bernanke.

Jackson Hole is risk on the hoof.

This may not be as clear or precise as the St. Louis Fed’s article cited later in this discussion implies that economics can be. That’s because Goodhart’s Law is the important consideration when analyzing “Quantitative Easing.” In economics, Goodhart’s Law states that for policy purposes one can target an economic phenomenon as measured by a particular indicator. However, when one does that, the indicator will lose the information content that would qualify it to play such a role. By targeting the indicator, the policy kills its information content. It no longer conveys the same information about the economic phenomena that one wishes to target.

That is the explanation for the statement: “The issue with quantitative easing is that now they [the Fed] are playing with the shape of the yield curve …since they are buying across multiple maturities. That’s dangerous, and they [the Fed] know it.” ( See: “Something worth thinking about,” November 29, 2010). The posting on August 6, 2011, “US Triple-A Rating And Popular Fiction,” was more explicit about the danger the Fed chose to accept when it initiated quantitative easing. The posting stated: “…basically, when central banks buy bonds across the maturity spectrum and flood the banking system with liquidity, the yield curve and interbank lending rates lose their information content.”

Thus, the yield curve’s shape, one indicator analyzed in “Large Scale Asset Purchases Had Large International Effects” (Christopher J. Nelly, Working Paper from the St. Louis Federal Reserve Bank’s Research Division), may not be indicating what the article concludes. “Large asset purchases” is Fed speak for quantitative easing. There are no doubts quantitative easing influenced foreign exchange rates. Foreign governments started complaining about the weak dollar policy even before it began. However, it is extremely questionable whether the two events the article analyzes provide any guidance about the impact of future similar events.

Operation Twist and the struggle to move to a bills-only open market operation after the Treasury / Fed accord of the 1950’s illustrate the problem the Fed currently faces. Once they take one action that influences the shape of the yield curve, they can’t rely on the shape of the yield curve as a source of guidance for future policy.

Further, my contention is that the traditional post-Keynesian (IS-LM) framework becomes unreliable. Like most macroeconomics, it relies too heavily on comparative statics, and it just pretends to represent time. A dynamic stochastic general equilibrium model becomes a more realistic representation (approximation) of the economic and financial market responses. Variances can no longer be assumed to be constant. The entire discussion shifts. Basically, the economic and financial market responses become path dependent.

Path dependence would explain Vincent R. Reinhart’s comments in the Bloomberg Op Ed, “Markets Go From Nightmare to Bad Dream.” He makes the statement: “In fact, a relatively simple story line runs through recent events. The problem for policy makers, importantly including Fed Chairman Ben S. Bernanke, is that this narrative is consistent with two starkly different paths for the global economy going forward.” He chooses a different explanation for recent events. He notes: “The dizzying price gyrations -- across asset classes and indifferent to national borders -- might be taken as evidence that the market has lost its mooring to fundamentals.” Losing one’s moorings is exactly what path dependency would imply is happening. The dizzying price gyrations determine the future paths the economy could follow. They set a range of attainable future results.

This may appear to be a conclusion with no actionable implications, but it is not. To illustrate again using the St. Louis Fed’s working paper, the article notes: “The increase in bank reserves reflects strong desire for safe, liquid assets that the simple portfolio balance model is ill-equipped to model with its focus on the mean and covariance of asset returns. Therefore the benchmark portfolio balance model does not directly model the market for bank reserves.” One would be hard pressed to fabricate an explanation for current market behavior that didn’t acknowledge an increase in demand for safe, liquid assets. The increase in demand for safe, liquid assets is a logical outcome of path dependence. That is the explanation for the headlines noted in “Default: Won’t Happen. Downgrade: Fear Not; Our Best Generals Are In the Field!”

An increase in demand for safe, liquid assets is the rational response when covariances are policy dependent. While the jury is still out, one can at least make a coherent defense of the proposition that liquidity-driven asset price cycles should create covariances that converge over an extraordinarily large set of asset classes and national borders. Since the covariances are in question, it undermines other risk management techniques. The net result is that an increase in liquidity creates a greater demand for liquidity. Further, it can’t be ignored since the safe, liquid asset influences the entire yield curve and very nearly every asset’s appropriate relative price.

One can use a Portfolio Management Model (PMM) to illustrate this although more than portfolio adjustment is at work. PMM would suggest that portfolio risk is a function of portfolio composition AND ONE’S LEVEL OF CONFIDENCE IN THE ASSET COVARIANCES.

Assuming away the confidence issue, which may be the more important factor, an increase in a safe, liquid asset could be offset by increasing more risky assets. To some extent, that’s what the Fed is trying to do. But, note that if done en masse, any over-adjustment will make the risk frontier less convex during the adjustment process, but more convex once prices adjust. It introduces the risk of making the risk frontier concave. At a minimum, it risks creating a risk frontier that is not smoothly continuous. A risk frontier that isn’t smoothly and continuously convex introduces multiple equilibriums on all or some risk frontiers. At the extreme with concavity, the two extremes could both be optimums. Personally, the change in levels of confidence seems the more important issue, but they aren’t required for the entire portfolio balance model to be explosive.

In “Default: Won’t Happen. Downgrade: Fear Not; Our Best Generals Are In the Field!” two questions were posed and answered: “Does it create a potential liquidity trap? Does it focus reserve management on the asset that is the current source of systemic risk? Yes, to both.” Using a liquidity injection runs the risk of creating a situation where safe, liquid assets are a larger and larger component of the optimum portfolio. Further, there is no guarantee that the increase in expected returns to risky assets can be raised fast enough to offset the increased expected return from a low risk asset. Under that scenario, the net result would be an optimum portfolio composed of just the safe, liquid asset. In the common vernacular, the market participants conclude that not losing is a higher return than any potential positive return. Heard that lately? If not, get the beans out of your ears.

The idea that an “Operation Twist” type of maneuver can avoid this problem is appealing. If it could be done instantaneously (i.e., if comparative statics were all that was required), it would be very appealing indeed. However, the adjustment process necessitates a period where covariances will break down. Why people assume that the covariance matrix is stable escapes this observer. Stability of the covariance matrix seems like a particularly silly assumption when policies that will destabilize it are the current rage.

In “Default, Downgrade, Lions, And Tigers And Bears; Oh My!” there’s a passing comment: “One has to admit the idea of a risk-free rate of return was truly a beautiful fiction. It allowed a massive financial edifice that spit out certainties at an amazing pace. Never mind that they were fiction.” The St. Louis Fed’s article is a good illustration. It’s an excellent analysis if one remembers: “Beware the risk-free return.”

A disclosure is in order: The St. Louis Fed’s and NY Fed’s research websites are among The Hedged Economist’s most heavily used favorites, which is in no way a slight of the other Fed bank’s research. At some point, The Hedged Economist will add links to favorites. More than likely the Fed will be represented by one link, but clearly anyone seriously interested in the economy should be familiar with every Federal Reserve Bank’s research focus.

1 comment:


    The issue of the Fed manipulating the yield curve is of interest to me. Does in invalidate the signals or send wrong signals? An inverted yield curve preceded the current recession. A few writers have said to watch for an inverted yield curve before a second dip. The inverted yield curve tells stakeholders that there is more risk in the short run than in the long run. Or, it tells them that they will be compensated as if there is more risk in the short run than in the long run. If you are actually being compensated for more risk in the short run, won't stakeholders act accordingly? I'm inclined to think they would. So, if the Fed manufactures an inverted yield curve (not just a flattish one) why wouldn't it elicit behavior appropriate for a recession?

    The IS-LM framework was John Hicks' interpretation of the general theory, and is arguably Hicksian, not Keynesian, and not Post-Keynesian. It came out of the Keynes-Hicks debates, not from the Keynesians. I knew Hicks. He was a genius with differential equations, especially with the time value of money. (two books "Value and Time" and "Capital and Time" neither contain calculus) He was concerned with getting the Austrian considerations into our analysis as best he could, but found static analysis to be an easy way to draw things. His narrative, especially footnotes and appendices, addressed all of the dynamic issues very thoroughly.

    At first as the Fed expanded its balance sheet the banks simply increased their reserves because of the schizophrenic demands of Fed verses non Fed regulators. Over the past few months M2 has been expanding at about a 15% annual rate. That's new, and people are overlooking it.


    Interesting comments. Thank you. You raise some important points. My comments / observations address them in reserve order.

    You follow the regulatory differences more closely than I do. But, no doubt the patchwork of foreign regulators and multiple U.S. regulators explains a lot. Perhaps we are at a point where regulatory details explain more about bank behavior than market forces. However, the Feds balance sheet expansion had impacts beyond bank reserves. The behavior seemed anomalous without the consideration of the Feds action on asset correlations. That was the reason behind the posting.

    It’s embarrassing that so often you have to correct errors in attribution. My excuse is that although I read Hick “Capital and Growth” and probably one or both the works you cite, my most recent review of LM-IS was in a macro text written in 1984 or ’85. It presented it as the post-Keynesian / neo-Keynesian macro framework. Thank you for cleaning up the reference. It didn’t include much on Hick’s efforts to address the dynamics. Rightly or wrongly, my recollection of Hick’s dynamics was more about alternative growth paths than adjustment paths. I have to review it again.

    Your discussion of the yield curve is on target. You stress returns while focusing on variances leads to the same conclusion. The reason for taking the later approach is summarized by the statement: “An increase in demand for safe, liquid assets is the rational response when covariances are policy dependent…. one can at least make a coherent defense of the proposition that liquidity-driven asset price cycles should create covariances that converge over an extraordinarily large set of asset classes and national borders. Since the covariances are in question, it undermines other risk management techniques.” This explains why the response can’t be avoided by diversification. It also explains why the response should surface long before the curve inverts. We may be there now.