I don´t think Mark needs a lengthy introduction since he´s been a long time commenter in market monetarists blogs, in addition to taking the discussion to famous Keynesian blogs such as Mark Thomas´ Economist View. Recently Tyler Cowen did a post singling out a comment he made to one of my posts.
Richard C. Koo, Chief Economist of the Nomura Research Institute and of Balance Sheet Recession fame recently wrote a paper entitled “Central Banks in Balance Sheet Recessions: A Search for Correct Response.”
In this post I want to respond specifically to his remarks on the U.S. Great Depression:
“If these unborrowed funds are left unattended, the economy enters a deflationary spiral as it continuously looses aggregate demand equivalent to the saved but unborrowed amounts. This process, now known as balance sheet recession, will continue until the private sector either repairs its balance sheet or becomes too poor to save (i.e., the economy enters a depression). Although that may sound outlandish at first, it was precisely this deflational spiral from private sector deleveraging that resulted in a loss of 46 percent of GDP in the US from 1929 to 1933 during the Great Depression.”
“From the perspective of central banks, the importance of deficit spending by the government is multiplied by the fact that it is also indispensible in maintaining money supply from shrinking when the private sector is minimizing debt. This comes from the fact that money supply, which is a liability of the banking system, starts shrinking when the private sector as a whole starts paying down debt. This is because banks are unable to lend out the money paid back to them by the deleveraging borrowers when the entire private sector is deleveraging at the same time. During the Great Depression, the U.S. money supply shrunk by over 30 percent from 1929 to 1933 mostly for this reason (85 percent due to deleveraging, 15 percent due to bank failures and withdraws related to failures.)”
There are examples where money supply has grown despite total nominal credit market debt shrinking (one such example is the U.S. in 1946) so Koo’s claims to the contrary fly in the face of empirical reality. But what I think troubles me most about this passage is the assertion that private sector deleveraging was responsible for the decline of the money supply during the Great Depression. Between October 1929 and April 1933, the sum of commercial bank deposits and currency held by the public, which is the definition of M2 money supply used by Milton Friedman and Anna J. Schwartz in “A Monetary History of the United States, 1867-1960,” declined by 38.2%. This occurred despite the fact that the monetary base, as measured by St. Louis Source Base increased by 14.9%.
In other words the money multiplier declined sharply. Friedman and Schwartz describe in detail how this came about through the series of bank panics that started in October 1930 and which largely came to a conclusion in March 1933 with the inauguration of Franklin Delano Roosevelt. During this time about 4000 banks failed, or approximately one fifth of all banks. Bank failures had a direct effect on money supply of course when they rendered their deposits worthless. But they also had an indirect effect in that worried depositors began to withdraw their accounts causing the ratio of currency held by the public to deposits to rise. This standard formula relating money supply (M), the monetary base less bank vault cash (MB), the currency ratio (c) and the reserve ratio (r) helps explain what happened to the money multiplier:
The following graph shows the currency in circulation ratio in blue, the reserve ratio in red and the currency held by the public ratio in green:
The currency held by the public ratio rose from 8.6% in October 1929 to 21.2% in April 1933, after peaking at 22.5% the previous month. The reserve ratio rose from 5.4% to 8.5%. A simple exercise in arithmetic reveals that the increase in the currency ratio accounts for 86.9% of the decline in the monetary multiplier, with the increase in the reserve ratio of course accounting for the remainder. According to “Banking and Monetary Statistics 1914-1941” excess reserves rose from 1.8% to 18.2% of all reserves during this period. The increase was symptomatic of a banking industry fearful of runs on deposits, but it accounts for less than half of the increase in the reserve ratio. Inspection of the data reveals that the currency ratio was fairly constant until the bank panics began.
And so you don’t think this is just a story about increasing ratios, here is a graph of the quantities of the components of the monetary base:
Currency in circulation is in blue, reserves in red, currency held by the public is in green and the monetary base including bank vault cash is in orange. Bank vault cash was more or less stable during this time period except for a temporary surge in March 1933 that coincided with a one month surge in the monetary base. Despite the increase in the reserve ratio the quantity of reserves actually fell from October 1929 to April 1933. But currency held by the public rose from $3.594 billion in October 1930, the month that the bank panics started, to $5.588 billion in February 1933, or by 55.7%.
I don’t know from where Koo gets the 85% figure for the proportion of the decrease in the money supply which can be blamed on deleveraging. It is very close to the proportion of the decline in money multiplier accounted for by the rise in the currency ratio, which Koo essentially credits for part if not most of the remaining 15%. But more importantly, the proximate cause was the bank panics, which in turn were caused by bank failures, which in turn were caused by firms and individuals being unable to pay their debts due to the poor state of the economy. If one wants to call the inability to pay debts deleveraging I suppose that may be technically true, but it conceals the actual mechanism by which the money supply contracted, and suggests that deleveraging always causes a reduction in money supply which is empirically false.
Here is Koo on the initial recovery from the Great Depression:
“Unfortunately there was a period in economics profession, from late 1980s to early 2000s, where many noted academics tried to rewrite the history by arguing that it was monetary and not fiscal policy that allowed the US economy to recover from the Great Depression. They made this argument based on the fact that the US money supply increased significantly from 1933 to 1936. However, none of these academics bothered to look at what was on the asset side of banks’ balance sheets.
The asset side of banks’ balance sheet clearly indicates that it was lending to the government that grew during this period (Exhibit 10). The lending to the private sector did not grow at all during this period because the sector was still repairing its balance sheets. And the government was borrowing because the Roosevelt Administration needed to finance its New Deal fiscal stimulus. In other words, it was Roosevelt’s fiscal stimulus that increased both the GDP and money supply after 1933.”
Exhibit 10 displays the balance sheets of Federal Reserve System member banks. The data comes from “Banking and Monetary Statistics 1914-1941.” What Koo terms “credit extended to the private sector” consists of loans and investments in private domestic securities. “Credit extended to the public sector” consists of investments in federal, state and local government securities. Terming this last category “credit extended” makes it sound like the banks were directly loaning money to the various levels of government, when in fact they were simply investing in government securities. The recession didn’t start until May 1937 so running the exhibit to June 1936 misses nearly a full year of the expansion. And it’s worth noting that not all commercial banks were Federal Reserve members, but as of June 1937, Federal Reserve member banks held 84.5% of all deposits at commercial banks, and hence accounted for the vast majority of Friedman and Schwartz’s M2 measure of money supply. And in any case, there isn’t sufficiently detailed information on non-member commercial bank investments to include their assets in Koo’s categories.
The following graph shows nominal GDP (NGDP), Friedman and Schwartz’s M2 measure of money supply, credit market debt (CMD) and the amount of such debt held by Federal Reserve System member banks subdivided by sector at year’s end. The aggregate credit market debt data comes from “Bicentennial Edition: Historical Statistics of the United States, Colonial Times to 1970” and the figures for the Federal Reserve System member bank assets come from the same source as Koo’s.
The key point is there was no shortage of debt to be placed on the asset side of Federal Reserve member bank balance sheets. Member banks’ share of credit market debt declined from 18.3% in 1929 to 14.8% in 1933 before increasing to 17.4% in 1937, so it would seem, in line with our previous reasoning, the decline in member bank balance sheets was more due to the shortage of deposits than the shortage of credit market debt.
Extending Koo’s diagram to the final year of the expansion reveals that member banks increased their holding of private debt from 1933 to 1937 almost as much as their holdings of public debt. Furthermore, the rate of increase in general government debt during the contraction from 1929 to 1933 (34.8%) was almost as great as its rate of increase during the expansion from 1933 to 1937 (36.2%). And finally, the rate of increase in nominal GDP was so great during 1933-37 that credit market debt as a percent of GDP declined in every single sector (i.e. household, business and financial), including the government sector, which fell from 72.0% to 60.2% of GDP.
We don’t have detailed sector balance sheets for these years, but according to “Studies in the National Balance Sheet” by Raymond W. Goldsmith and Robert E. Lipsey, private sector assets fell from $923.4 billion to $665.0 billion between 1929 and 1933, or by $258.4 billion. Private sector financial assets fell from $540.9 billion to $389.9 billion or by $151.0 billion, accounting for 58.4% of the decrease in private sector asset value. Private sector holdings of stocks fell from $186.7 billion to $101.5 billion, or by $85.2 billion, accounting for 56.4% of the decline in private financial asset value, and nearly a third of the total decline in private sector assets.
We hear a lot in the media today about the effect of QE on the stock market. Well, in an act that was essentially the QE of its day, FDR took the country off the gold standard in April 1933 and allowed the price of gold to rise from $20.67 an ounce to $35.00 an ounce by January 1934. This price was high enough to attract a large gold inflow from abroad which the Treasury monetized by issuing gold certificates to the Federal Reserve. The monetary base, which had been rising at roughly a 6% annual rate since around mid-1930, skyrocketed upward, reaching a maximum year-on-year rate of increase of 23.0% in February 1935. In turn, the stock market, which reflected an economy starved for aggregate demand (AD), soared as well. The Dow Jones Industrial Average (DJIA) more than tripled from March 1933 to November 1936, before the decision to start sterilizing gold inflows in December 1936 meant prematurely removing the punch bowl, leading to the 1937 Recession:
In any case, we don’t have detailed information on sector balance sheets from 1937, but given that it was the decline in the value of stocks that did the most damage to the private sector’s balance sheet during the contraction, it’s not hard to see how this stock market rally went a long way to repairing that damage.
Koo also warns about the damage that QE will do when it inevitably must be withdrawn. Although not directly dealing with the Great Depression, I think the years following the Great Depression reveal much about why his concerns are terribly ill placed:
“A QE is removed when the private sector is back borrowing money (i.e., maximizing profits) and the central bank is worried that it has to remove excess reserves in the system in order to avoid a runaway inflation. At this juncture, money multiplier is significantly in a positive range. This means it does matter whether the needed removal is $200 billion or $2 trillion, with a larger number having a much bigger impact on interest rates, especially bond yields. Put differently, those central banks who implemented limited or no QE would be much more relaxed than those who implemented massive or “unlimited” QE when the private sector is ready to borrow money again.
So far the only successful removal of QE was the one engineered by the Bank of Japan in 2006, when its first-ever quantitative easing in history was ended after five years. This removal went without a hitch because the QE was all at the short end of the market. The removal of reserves at the short end under zero interest rates had very little impact on the rest of the yield curve. For example, the yield on 10-year JGBs jumped by about 40 basis points right after the announcement of the end of quantitative easing, but the yield returned to the original level after a few months.
This time, however, the US and UK central banks are in the long end of the market. This means the removal of QE will have much larger effect at the long end of the yield curve, with equally larger impact on the economy just when the economy is regaining its health and willingness to borrow.”
First of all I’m not at all convinced that the Japanese withdrawal of QE, which involved shrinking the monetary base by about 20% from January to September 2006, was successful. After all, much like the unwise U.S. decision to sterilize gold inflows in December 1936, this was soon followed by a very serious recession. Japanese real GDP (RGDP) plunged at a 4.8% annual rate in 2008Q2, and at a 4.0% annual rate in 2008Q3, resulting in consecutive sharp declines in quarterly RGDP before the U.S. economy ever did the same.
More importantly, Koo seems unaware of the magnitude of the quantitative easing that took place during the Great Depression and World War II. The following graph shows the monetary base (St. Louis Source Base) and reserves (not adjusted for changes in reserve requirements) as a percent of GDP from 1919 to present:
The monetary base was 17.7% of GDP as of 2013Q1. But it reached 22.8% of GDP in 1946Q1 and was greater than its current level during 1939-1941 and again from 1945-47. Reserves peaked at 10.7% of GDP in 2011Q3. But they reached 13.6% of GDP in 1940Q2 and 1940Q3. They were above 10.7% of GDP during much of 1939-41. And the return to the levels they had been before the Great Depression took many years. Reserves did not fall to the 2.2% of GDP level they had been in 1929Q3 until 1972Q3, and the monetary base did not fall to 6.5% of GDP until 1975Q4.
Koo points out that the Fed, along with the BOE, is in the long end of the market. The Federal Reserve currently holds $487 billion in Treasury bonds (of maturity over 10 years) out of assets of $3,314 billion. Total holdings of Treasury securities are $1,882 billion. Thus Treasury bonds are 14.7% of all Federal Reserve assets and 23.8% of all Treasury security holdings. Treasury bond holdings were $457 billion at the end of the first quarter which was about 2.9% of that quarter’s GDP of $16,004.5 billion at an annual rate.
According to “U.S. Monetary Policy and Financial Markets” by Ann-Marie Meulendyke, Treasury bonds were 65.1% of all Treasury securities held in 1941, and total holdings of Treasury bonds reached $10,977.2 million in 1948, which was 23.8% of the monetary base and 4.3% of annual GDP. True, Treasury security issuance was dominated by longer term securities back then, but that doesn’t take away from the fact that long term Treasury securities were a larger proportion of both Federal Reserve total assets and all Treasury security holdings than they are today.
However, much of this is moot if one considers that the last time the Federal Reserve spent a protracted period at the zero lower bound the money multiplier did not fly out of control the moment it was necessary to raise interest rates. Here’s a graph of the Friedman and Schwartz M2 money multiplier and the yield on short term (3 to 6 month) U.S. securities from 1925-70:
The money multiplier reached a low of 2.52 in November 1940 when short term interest rates were 0.003%. After the Fed un-pegged the short term rates from the 0.375% rate that had been set from mid-1942 through mid-1947, they were allowed to rise above 1% in August 1948 for the first time since March 1932. But the money multiplier remained close to three, rising only very slowly even as short term interest rates rose high enough by 1953 to resume their more familiar cyclical volatility.
Moreover there was never any need to exit from the more than six-fold increase in the monetary base that took place in 1932-48. The most that the monetary base decreased was from $48.413 billion in December 1948 to $42.960 billion in April 1950, or by 11.3%. And even that decrease probably had only minimal negative consequences because of the expected decline in real output following World War II.
So given the history of the last time the Federal Reserve spent an extended period of time at the zero lower bound, as well as the availability of newer monetary policy tools (e.g. interest on reserves), I seriously doubt that an exit from QE will be necessary, much less desirable.