Thursday, April 21, 2011

Further Evidence that the Private Sector Fully Controls the Money Supply and QE Doesn't "Work" as Advertised

In October, I wrote a post titled How the Loan/Bond Choice Helps the Private Sector Self-Determine the Money Supply — AND Yet Another Reason QE is a Non-Event for the Economy. To give a brief summary, it described my thoughts on the inter-relational dynamics between debt, money supply, and Quantitative Easing, as follows:
  1. There are two high level ways that borrowing occurs in the private sector — bank loans, and all other forms of borrowing, i.e., bond issuance, securitization, peer to peer lending, etc.
  2. When there is an "excess" of short duration assets (primarily money) held by the private sector, the holders of those assets will be eager to lend it out, and thus acquire a higher yielding asset. The money supply remains unchanged.
  3. When there is no "excess" of money to lend (i.e., portfolio preferences are satisfied with the current levels), then banks will have a higher propensity than non-bank lenders to fulfill the economy's current borrowing needs, because they can lend an [almost] unlimited amount, independent of their level of money/reserves. (Their only limit on lending to worthy borrowers, and it is temporary, is how much equity capital they can raise). Bank lending increases the money supply, so from a macro perspective, banks could be considered the private sector's lender of last resort.
  4. By choosing the relative proportion of the two type of borrowing, the private sector is able to choose its portfolio mix of long duration assets and short duration assets (money), independently of the actions of the federal government, and [mostly] independently from the desired level of private sector borrowing!
  5. When the Federal Reserve conducts Quantitative Easing, it buys long duration assets (treasury bonds, etc) out of the private sector, and gives the private sector short duration assets (money balances) instead.
  6. If the private sector is not happy with this new portfolio mix resulting from QE, it likely has the power to "undo" the change over time via shifts in the proportion of bank loans versus other types of lending that it uses!
If the balance sheet impacts of bank lending, non-bank lending, and quantitative easing are not familiar to you, please play with the Macroeconomic Balance Sheet Visualizer — it is a graphical web-based tool, now with a step-by-step walk through mode.

Possible Evidence from a Post-Keynesian Expert that this Theory may be Correct

In the comments of my previous post, commenter Ramanan helpfully linked to a PDF from Marc Lavoie, one of the leading Circuitists. Circuitists (also known as horizontalists) are one "school" of theory within Post-Keynesian economics, and they share most of the same concepts as Modern Monetary Theorists. In section 9.3.1, Lavoie seems to describe the same dynamic that I have attempted to describe. Here is an excerpt:
"...for apparently the demand for money and the supply of credit are determined by two independent mechanisms. In the Lavoie and Godley (2001-2002) model for instance, the demand for credit, at the end of the period, depends on the part of investment expenditures which has not been financed by retained earnings and new equity issues..."
"...the decision by households to hold on to more or less money balances has an equivalent compensatory impact on the loans that remain outstanding on the production side."
While Lavoie only mentions "equity issues", I have seen evidence elsewhere that he uses that term as shorthand to describe any non-bank borrowing mechanism employed by a firm, i.e., his reference to "equity issues" is supposed to also encompass bond issues and some other types of liabilities.

For reasons that are unclear to me, some MMT authors repeatedly claim that QE does not add to the broad money supply. While it is true that if the primary dealers sell their own treasuries to the Fed, then only base money supply is affected, this scenario is too limited given the size of QE to date. As I understand it, primary dealers frequently act as an intermediaries to facilitate QE buying assets from the larger private sector. But if the private sector can react relatively quickly enough via the mechanisms described here to "undo" the money supply changes, then the money supply data won't show any bulge in broad money supply resulting from QE! (And of course there are other factors impacting the money supply at the same time, primarily a desired deleveraging within the private sector, so it is probably not possible to disentangle these different dynamics when looking at the data.)

Evidence in the Recent Data for the QE in the United States

In the last post on this topic, I showed a graph of Japan's bank and non-bank borrowing. While inconclusive, it suggests that Japan's Quantitative Easing from 2001-2006 may have caused a relative decrease in bank-based borrowing as compared to non-bank borrowing, which could add to evidence of the theories above.

Below is a graph of US borrowing from Q2 2004 to Q1 2011.

Effects of US Quantitative Easing on Bank Lending Compared to Total Borrowing:

(click to enlarge)

The red line shows the annualized percentage change in bank loans and leases, by quarter. The green line shows the annualized percentage change in total private sector debt (from the Z.1 report, which is not yet available for Q1 2011). The blue line shows the annualized percentage change in non-financial private sector debt (since the private sector's debt has many layers that could overwhelm the other trends, I thought it worth separating financials out.) I have not subtracted bank loans out of the total debt data, so the shape of the blue and green lines is slightly impacted by the shape of the red line (though the absolute amount of bank debt is around $7 trillion versus around $41 trillion for total private sector debt, so crossover impact is not huge).

Notice how bank loans declined substantially faster than total debt after the first round of Quantitative Easing started! The gap between the red line and the blue line is probably the most relevant. Despite tracking closely to each other until late 2008, they diverged significantly starting in 2009, perhaps as the private sector favored non-bank lending on a relative basis over bank lending, in order to "shed" the excess money supply imposed by quantitative easing!

The second round of quantitative easing is smaller in magnitude, and the data so far only covers the start of QE2. However, in Q4 2010 the gap between the red and blue lines appears to begin widening again. Will that effect continue? I would guess so, but I could ultimately prove to be wrong.

It is also not clear how big the lag effects in this process are. Also, the changes in bank loans versus total debt diverged most significantly in the middle of QE1, and the gap narrowed partially before QE1 ended. It could be that other dynamics that I am unaware of provide a more accurate explanation than what I suggest here.

Anecdotal Evidence

While it's far from scientific or conclusive, and I don't follow the details of specific financial markets, I occasionally see anecdotal evidence of the private sector's increased eagerness for non-bank lending. For example:
  • Junk Borrowers Turn Tables on Investors With Looser Terms: Credit Markets: "...debt sold this week included a condition that allows the company to call 10 percent of the bonds at 103 cents on the dollar in each of the first four years... [The borrower] is trying to lock in low interest rates while getting the flexibility to repay debt any time, as it would with a loan..."
  • Subprime Bonds Are Back
  • [Maybe] recent increases in venture capital activity? (If true, it would be an example of equity issuance rather than bond issuance, but with a similar macroeconomic effect.)
PostScript / Technical Note

In generating the chart in this post, I discovered the importance of using the percent change bank loan data as prepared by the Federal Reserve in its H.8 release whenever possible, instead of starting with the absolute levels of loans. The latter's level jumps around too much due to balance sheet consolidations, acquisitions, etc, while the percent change data appears to be adjusted to remove this effect, if I am understanding it correctly. Details are on the about page and notes page.

Tuesday, April 12, 2011

Real GDP Per Capita and Myths about Japan's Stagnation

While addressing some myths about Japan, Bill Mitchell posted some graphs of contributions to real GDP. His graphs cover the same data I graphed last August -- the contributions of Consumption, Investment, Government Spending, and Net Exports toward inflation-adjusted economic growth. (A minor difference is I separated Inventories and other Investment in my graphs). Bill also usefully added in Ireland while emphasizing the negative effects of austerity on growth. Visuals such as these make it obvious that investment falls under austerity -- it doesn't rise on a supposed swell of increased business confidence that austerity proponents often claim will be triggered!

While I've been guilty of parroting some myths on Japan myself in years past, it has been enlightening to put more effort into reviewing the data for myself. Here is a graph I have been meaning to post for quite a while:

Annual Growth of Real GDP Per Capita in the US and Japan (1980-2009)

The blue line is for Japan, the red line for the US. It is quite astonishing how close in both magnitude and direction US and Japanese growth have been, when adjusted for population changes! The main divergences are Japan's burst of higher growth in the late 1980s, and its austerity-driven recession in the late 1990s. Once you adjust real GDP growth for changes in population, what you are left with is largely productivity growth. There are other factors that affect the level of GDP (and important things like employment!) but they have less effect on year-on-year growth rates once the adjustment has occurred.

Japan's famous deleveraging primarily meant a higher savings rate than before the deleveraging (corporate rather than household, in Japan's case), with the main effect being a one time shift downward in GDP level (but not growth). After the GDP shift caused by a savings rate shift, future and ongoing GDP growth is impacted by other factors such as whether aggregate demand is sufficient to bring growth back near potential growth, but Japan may have been a success story in this area! However, its success in achieving this real per-capita GDP growth may have been more a result of the falling household savings rate over the last couple decades than of government fiscal (or monetary!) policy.

Japan's stagnation myths (some people blame too much government spending, others too little!) derive in part from two sources of confusion -- real growth versus nominal growth (Japan has a low and sometimes negative rate of inflation) and GDP growth versus per capita GDP growth (Japan has a low-to-negative population growth trend).

Of course, real GDP growth (absolute rather than per-capita) does affect valuations of financial markets and real estate, since those valuations rely on the size of future earnings streams. Japan's asset markets have famously failed to "recover." To the extent that other nations follow in Japan's demographic footsteps, there will be some downside surprises in asset market returns in the medium to long term for many advanced nations...

UPDATE (same day):

To supplement the year-on-year growth chart above, here is a graph of the actual levels of per capita real GDP indexed for Japan and the US. Japan's surging growth in the late 1980s that accompanied its stock market and real estate bubbles did put it above the US trend, but the levels converge again before year 2000.

UPDATE 5/23/2011: After reading a post by Bill Mitchell today that appeared to contradict the findings in this post, I initially wondered whether I had erred by using the FRED2 data for Japan in which the real per capita GDP data is converted to 2009 US dollars at purchasing power parity. (i.e., perhaps this incorporated exchange rate effects as well.) However, today I extracted the appropriate source data from Japan's cabinet office and generated my own graph using yen-denominated values, and the trend line is identical to the FRED2 data I used. In other words, the trend lines for per capita GDP growth for the US as measured in dollars and Japan as measured in yen still are remarkably consistent with each other, just as this post initially showed. The main divergences in the two trends are in the time periods of roughly 1987-1990 and 1997-2000.

It turns out (based on my own graphs from the source data) that Bill seems to have accidentally switched the labels on the two lines in his second graph, so his nominal and real GDP per capita lines for Japan are reversed.