In light of these theories, Hillenbrand (2023) has documented a surprising pattern: interest rate changes in a narrow window of time around the Federal Reserve's monetary policy meetings capture virtually the entire trend decline in interest rates over the past few decades.^[5] The Federal Open Market Committee (FOMC) is the committee of the Federal Reserve System (Fed) responsible for implementing monetary policy. The FOMC consists of 12 voting members: the seven members of the Board of Governors, the President of the Federal Reserve Bank of New York, and four of the remaining 11 Reserve Bank presidents, who serve on a rotating schedule with one-year terms. Decisions on monetary policy are made by the FOMC based on a majority vote. Since 1981, the FOMC has typically held eight scheduled meetings per year. The majority of monetary policy decisions since 1994 have been taken at these scheduled meetings, while a few decisions have been taken at unscheduled meetings (usually in the form of teleconferences). Prior to 1994, these unscheduled meetings were more common. He arrived at this conclusion by studying only the interest rate changes that occur around monetary policy meetings and comparing the sum of these changes with the interest rate changes that occur on all other days of the year.^[6] Greenlaw, Hamilton, Harris and West (2018) used the same approach to study the effects of the Fed's asset purchases after 2008. In Figure 2, we replicate his results for the United States and extend the sample period to March 2024.^[7] The results are very similar if we instead study the corresponding US Treasury Inflation-Protected Securities.

The method of calculating these, so to speak, hypothetical interest rate changes may require some further explanation. We can start by noting that between June 1989 and March 2024 there are about 9000 working days. Thus, interest rates since 1989 change for different reasons on each of these days. Together, all changes in interest rates add up to the trend observed in Figure 1. During these 9000 days, the Fed has held about 300 meetings to decide on monetary policy in the United States.^[8] 277 of these meetings were scheduled and 26 unscheduled. Figure 6 shows the distribution of daily changes in a 10-year US government bond on Fed days compared to all other days. The distributions are similar but the interest rate changes on Fed days have a slightly higher variance with a longer ‘tail’ to the left. Surprisingly, if one adds up the interest rate changes that occur only during these 300 days - as well as the day before and the day after - the sum is roughly equal to the sum of all 9000 days of interest rate changes (blue line versus light blue line in Figure 2). Thus, the entire trend decline in interest rates occurs in the roughly 10 per cent of days surrounding the Fed's meetings.^[9] The decline in the ten-year rate is roughly equally distributed across the three days. If we had only summarised the interest rate changes on Fed days, the total decline in interest rates is about 2.8 percentage points. The corresponding figures for the day before and the day after are -2.5 and -2 percentage points respectively. But not only that. In addition, we observe that the majority of the trend decline in interest rates after 2005 is in connection with Fed meetings which take place in the last month of each quarter, i.e. in March, June, September and December.^[10] See Figure 7 in the Appendix. During these months, the Fed publishes a document called the "Summary of Economic Projections" (SEP). This summary is an overview of the FOMC members' forecasts of economic developments. It contains forecasts for growth in gross domestic product, unemployment, inflation, and interest rates over both short and long time horizons. The four months are also important in bond and derivatives markets. For example, they mark the end of financial quarters, which often leads to increased trading activity. Investors and companies make quarterly reconciliations of their portfolios and balance sheets, which may include buying or selling bonds and derivatives to balance risk or hedge profits. Moreover, many standardised derivative contracts, such as futures and options, have maturity dates that fall in these months. This increases the number of trading transactions as market participants roll their existing positions into new contracts or liquidate positions to realise gains or losses. Thus, interest rate changes around all Fed meetings do not contribute equally to the decline in interest rates.

Figure 2. Accumulated changes, 10-year government bond yield, USA Percentage points

The documented pattern is surprising in several ways.

First, it is surprising that the trend decline in interest rates occurs around Fed meetings. One would think that the changes in interest rates on these occasions are mostly due to the decisions the Fed takes on its monetary policy.^[11] See, for instance, Hanson and Stein (2015) for a discussion. However, as mentioned above, many theories have suggested that the actual causes of the downturn are not directly linked to monetary policy. Interest rates started to fall in the 1980s when inflation was in double digits. The subsequent decline in the first part of the 1980s was probably due to a fall in long-term inflation expectations.^[12] See, for instance, Cieslak and Povala, (2015). Economists have mostly seen the decline in interest rates over the following three decades as a real phenomenon. For example, estimates of the natural interest rate – the real interest rate at which monetary policy is neither expansionary nor contractionary – have declined noticeably.^[13] See for example Laubach and Williams (2003), Bauer and Rudebusch, (2020) and Armelius, Solberger, Spånberg and Österholm (2024). While there is considerable uncertainty about the level of the natural rate, declines in market measures of real interest rates have presented a similar picture.

Second, it is surprising that long-term interest rates change so much and so regularly in the context of Fed meetings. From a theoretical perspective, we would not have expected it because the Fed only controls the nominal short-term interest rate (the overnight rate). Monetary policy can only affect real interest rates temporarily through price and wage rigidity. Therefore, we would expect the impact on long-term interest rates and the real interest rate to be only temporary. Moreover, the real economic effects of monetary policy do not appear to be very long-lasting.^[14] See, for example, Ramey (2016). Thus, it is surprising that prices of long-term bonds, especially long-term real bonds, exhibit such large and systematic movements around Fed meetings.

Third, we note a change in the pattern after 2021. The rapid rise in interest rates no longer occurs around Fed meetings to the same extent. However, as we are focusing on trends, it is difficult to say much about the long-term pattern after only a few years. In any case, it is clear that the red line in Figure 2, which is constructed by taking into account only interest rate changes that did not occur around Fed meetings, has increased in a way not seen since 1989.