Friday, August 7, 2015

Monetary Policy: Part 4 - Solving the Mystery of Money Growth

As monetary policy became the focus of many in the bond markets in the 1970's more and more attention was directed to the weekly release of the M1 Money Supply each Thursday after the markets closed. (M1 consisted then of currency and demand deposits.) An unexpectedly large increase, on the order of a few billion dollars, often tanked the bond market, whereas an unexpected drop of a billion or so would generate a rally. Not everyone became a monetarist, but most market participants paid attention to the money numbers at the time.

The Fed's Operating Procedure Then and Now

When the money supply rose above a band the Fed considered prudent, it would tighten policy eventually. The way it did this was to intervene in the fed funds market, draining funds unexpectedly. That action would convince fed funds dealers that they needed to conduct their transactions at a new, somewhat higher, interest rate level. They would soon learn, from further Fed interventions, whether that new level was a quarter, a half, or a full percent, higher than the previous level.

The Fed was operating under the theory described in a previous post, i.e., that raising short rates would cause all rates to increase, thereby eventually slowing the economy and, along with that slowing, the demand for money would also slow. As also described in that post, this was a long and uncertain process resulting in many pauses along the way and usually culminating in an overshoot as the Fed found itself tightening during what would eventually be determined to be the start of a recession.

Today, the Fed still operates under the same theory but finds itself in a bind because zero percent on fed funds doesn't seem to be low enough to generate the economic recovery they anticipated. Thus, we got QE1 through QE3 and the Fed attempted, successfully, to lower the interest rate on longer maturities by taking direct action.

The Fed Funds Market Operating Procedures

Following my realization that Effective Reserves was a better indicator of the Fed's actual policy than was the M1 money supply, I started paying closer attention to other weekly monetary events. One of these was the behavior of the fed funds rate on Wednesdays. Each week, every bank in the system had to settle its books with the Fed at the end of trading on Wednesday. That is, their final average reserve balance for the period was calculated at close of business Wednesday. That was matched against their required balance as determined by their deposits for the previous week, ending Wednesday. That is, required reserves lagged deposits by one week.

There was one additional feature in the system. If a bank was either long or short on its required reserve balance on one Wednesday, it could carry forward the balance against the next week's requirement. So, for example, if it was short $5 million at the end of week one, it could make up the shortfall by ensuring that it held an extra $5 million at the end of week two.

That particular feature meant that when there was a surplus of fed funds in the market for whatever reason, the fed funds rate would come under some downward pressure on Wednesday afternoon, but not undue pressure. After all, any extra balance could be applied against next week's reserve requirement, so why sell funds cheaply? However, if that surplus remained in the market at the end of week number two, the banks as a system simply didn't need them, so they sold them at any price because they couldn't carry their excess forward for another week.

The result was that a surplus of excess reserves in the banking system would drive the fed funds rate to exceptionally low levels on the second week, sometimes as low as 1 or 2 percent even when rates were far higher during the remainder of the week.

Another Personal Discovery: The Real Source of Money Growth

On Thursday afternoon, the Fed would announce last week's M1 number and the current week's excess reserve figure. I soon noticed that the excess was usually abnormally high, on the order of $400 to $600 million, during week's when the fed funds rate plunged on Wednesday afternoon. And because the process really began a week earlier, it was also unusually high the preceding week.

Over a period of a few months I began to notice a pattern to the occasional unexpected money bursts that would occur. When two consecutive weeks of a high excess reserve position occurred, then the M1 number released a week later would show a significant increase. Also, quite often, the excess reserve position would again be unusually high and would have been presaged by another weak fed funds market the day earlier, on Wednesday.

So, I started to wonder. Is the money supply somehow responding almost immediately to the extra reserves in the banking system? Eventually, I decided that was the case. Somehow, the system was circulating the excess in a way to generate a rapid increase in demand deposits. And since during those consecutive weeks when the banking system as a whole was in an excess reserve position, every funds desk manager was desperately trying to get the reserve position back down, it seemed a reasonable conclusion that somehow those rapidly circulating reserves found their way into customer demand balances, perhaps via overnight loans too cheap to pass up. Regardless, that was the pattern.

Some Evidence That Was Indeed What Was Happening

I wish I had it at hand. I had it and later discarded it, so I'll describe it instead. There was one particular year, either 1977 or 1978, I believe, during which over 100% of the M1 money growth that year could be accounted for by adding up the increases in M1 that were announced during the specific periods I described above. That is, there was a three week period of high excess reserves in the system accompanied by an immediate burst in M1. As I recall there were five, perhaps six, such periods in the entire year. More than 100% of the year's growth in M1 were associated with those periods. The rest of the year, consisting of about 2/3 of the weeks in the year, M1 actually fell.

If someone chooses to investigate the above claim, all that need be done is go back to the weekly reports of excess reserves and M1 issued at that time. Why at that time? Because seasonal adjustments smooth out historical data. You'll need to look at the original data. The most likely year was 1978.

A New Theory: The Hot Potato Theory of Money Growth

Eventually I concluded that excess reserved do indeed feed rapidly into demand deposits. I believe this happens because the managers on the funds desks are selling reserves with such abandon on the relevant weeks that the money simply gets lost in the system. Not really, of course, but loans get made, particularly overnight and very short term loans, and those loans get deposited in another bank, whose funds desk manager then discovers them, sells the resultant excess to another bank, another loan is made, and the process continues until the end of trading is called. When the accounting is made, someone ended up with the hot potato.

Mistakes Then Followed Mistakes

Interestingly, the Fed's ongoing use of a different theory of money growth, i.e., that interest rates affect the economy which only later affects money growth, kept them from correcting the erroneous infusion of excess reserves. After a money burst, the Fed would recalculate everyone's new required reserve balances and would then supply those reserves accordingly, instead of acting as would be indicated here, and simply take them back out again. Had they done so, the process likely would have reversed itself, with loans being taken off the books just as fast as they went on in the excess weeks, and M1 would have dropped to its previous lower level.

Instead, by supplying the newly-required, higher, level of reserves, the Fed effectively locked in its earlier mistake and the burst in M1 became permanent.

The next post will summarize the implications for Fed policy had this all been realized back in 1980. Unfortunately, subsequent events overran these discoveries, rendering them all but irrelevant today.

Continue to Monetary Policy: Part 5 - The Perfect Monetary Policy

No comments: