Thursday, November 20, 2008

Donald Kohn on Excess Reserves & the 2003 Deflation Scare

This past Wednesday I was able to attend the CATO Institute's 26th Annual Monetary Policy Conference. They had a number of interesting speakers, but the keynote speaker was Federal Reserve Vice Chairman Donald L. Kohn. He gave the first talk of the conference and then took questions from the audience. His talk was an interesting one where he reevaluated whether the Fed should attempt to check asset bubbles. Here, however, I want to focus on two of the comments he made after the talk.

The first comment had to do with the policy of paying interest on excess reserves. Someone from the audience asked him if this policy was counterproductive since it discouraged banks from lending at the very time they should be encouraged to lend. Kohn's answer was that these excess reserves would still be sitting at the banks even if the Fed were not paying interest on them. I find this view hard to accept. Surely some of these excess reserves are being held because of the interest payments. Why would any bank lend excess reserves to another one when the bank is guaranteed the federal fund rate target by the Federal Reserve? If you want to see the potential downside to this policy look no further than the Fed's 1936-1937 monetary policy.

The second comment is one that Kohn shared with me directly. As I was leaving the auditorium after one of the talks I saw that Kohn was right behind me. I took advantage of this opportunity by asking him a question that went something like this: "You mentioned that the low inflation in 2003 indicated economic weakness and, thus, justified the accommodative monetary policy at that time. Couldn't one also view the low inflation in a more benign manner by interpreting it as the result of the rapid productivity gains rather than weak aggregate demand?" His answer was "No. Productivity was not growing. Moreover, unemployment was growing at the time." I appreciate him answering my question, but have to respectfully disagree with his response. Here is why: (1) data shows productivity was growing and (2) the weak labor market--called at the time the "jobless recovery"--can easily be understood as a response to rapid productivity gains. One could also make the argument that the weak labor market conditions were the result of the Fed's low interest rates creating an inordinate substitution of capital for labor.

These next three figures make my case. The first figure shows productivity--as measured by non-farm business labor productivity--did see an acceleration in its year-on-year (Y/Y) growth rate around 2003:

The figure also shows the ex-post real federal funds rate (ffr) relative to Y/Y the productivity growth rate. Assuming there were no significant changes in intertemporal preferences and population growth rates, this first figure suggests the Fed was pushing its policy rate down as the natural rate--which is a function of intertemporal preferences, population growth rate, and productivity growth rate--was increasing.

The next figure shows the Y/Y growth rate of nominal spending--measured by final nominal sales to domestic purchasers--against the nominal ffr. The year in question, 2003, is highlighted by the two dashed lines. This figure indicates nominal spending or aggregate demand was not collapsing in 2003. The ffr, on the other hand was being pushed to down to 1%.

So productivity was increasing and aggregate demand was not collapsing. What about the weak labor market? As mentioned above, one story is that the "jobless recovery" was simply the consequence of the productivity gains. Another story is that the existing low ffr was pushing the cost of capital down and encouraging an inordinate substitution of capital for labor by 2003. In either case, there is no justification for further lowering of the ffr in 2003. The figure below sheds some light on this view. It graphs real gross private domestic investment against total nonfarm employment with the year 2003 again delineated.

This figure shows a sharp increase in investment spending in 2003 while employment remained more or less flat. Firms, therefore, were building up their capital stock while avoiding new additions to labor. I suspect monetary policy was a key reason for this development. For those who are interested, I was able to ask a similar question to Ben Bernanke back when he was a Fed governor. See his response here.

Update: The ever insightful ECB points out that the figure above that uses real gross domestic private investment reflects both residential and non-residential investment. Clearly, my capital substitution story hangs on non-residential investment recovering while labor remained flat. Real non-residential investment is graphed below with nonfarm employment:

This figure shows that non-residential fixed investment was recovering in 2003 while employment remained flat. However, it did not recover as sharply as gross domestic private investment overall.


  1. "Kohn's answer was that these excess reserves would still be sitting at the banks even if the Fed were not paying interest on them. I find this view hard to accept. Surely some of these excess reserves are being held because of the interest payments."

    Kohn is correct in aggregate of course. Excess reserves for the system don't simply disappear because bank A lends to bank B. Perhaps thats what he meant.

  2. David,
    Not sure about the capital substitution argument. Your figure shows gross investment. But if I wander over to Casey Mulligan's blog, I find this graph (link)
    It shows in 2003 nonresidential investment declining as residential accelerates. And I think residential construction is fairly labor intensive. Dont you think some of that productivity gain may be due to mismeasured labor input arising from upsurge in illegal immigration to fuel the housing boom?

  3. ECB,

    Good point. I went back and graphed non-residential investment against employment and added it to the post. The St. Louis Fed data (where I got the graphs) does show non-residential investment growing in 2003, but not nearly as fast as overall gross domestic investment.