How long? How variable?
It is Thursday, yet it is Fed Day (I think you can thank the US elections for that). I write this before the election outcome is known so I don’t know if the election will have any influence on the decision taken by the Fed today. But whether it has or not, one thing is clear, whatever the Fed does today will not be felt in the real economy for quite some time. The famous long and variable lags are one of the reasons why monetary policy is so hard. But how long and how variable are these lags?
New research published in the Journal of Monetary Economics disaggregated Personal Consumption Expenditure to the finest degree and estimated the lag between a 100bps interest rate hike by the Fed and the price reaction of individual goods and services. The charts below show the results for headline inflation (PCE) and core inflation. Note something? Inflation declines only after about 40 months!
Inflation reaction to a 100bps rate hike by the Fed
Source: Aruoba and Drechsel (2024)
Yes, inflation tends to decline a little in the first three to ten months after rate hikes but then starts to rise again as the rate cuts stimulate demand. The full impact of rate hikes is only felt about three to four years later. If this pattern of declining and then accelerating inflation sounds familiar, it is because I wrote about this effect a month ago and it is a key reason why we should not declare victory in our fight against inflation too early.
The intense granularity of the paper can also give a hint as to why inflation reacts so slowly. It is not that interest rate hikes don’t work. Instead, the variable lags tend to partially eliminate the impact of rate hikes on the economy. The chart below shows the reaction of inflation in different parts of the economy.
Inflation reaction to a 100bps rate hike by the Fed for different goods and services
Source: Aruoba and Drechsel (2024). Note: NPISH = Non-profit institutions serving households.
Note that in the immediate aftermath of rate hikes, inflation for durable goods keeps rising(!) while inflation for nondurable goods starts to drop almost immediately. After about a year, inflation for durable goods also starts its decline, but that is when service inflation tends to rise.
Service inflation is much more driven by wage inflation than goods inflation and I think what we see here is the effect of rising wage inflation in reaction to higher consumer price inflation. Just when rate cuts are about to unfold their full impact on prices of goods, service prices tend to accelerate, negating some or all of the impact on overall inflation.
This also means that the lag for monetary policy is changing all the time as the economy changes. Today, the US economy is much more driven by services than in the 1970s or 1980s, which means that the impact of rate hikes is felt later in the economy than in the past.
I am not sure, but this may have contributed to the surprising resilience of US economic growth in the aftermath of the latest rate hikes. An economy that is more driven by services (such as the US or the UK) will likely be more resilient and at the same time, inflation will be harder to manage and control than in an economy that has a larger manufacturing sector (think of Germany or Korea).
PS: I need to mention that the authors of the study as well as Jonathan Wright comment that the exact length of the impact can vary a lot depending on how you specify the model that measures the delay. So don’t take the 40-month lag too literally. It may well be 30 months or 50 months. But it seems much longer than the 12 months or so that is commonly used as a standard lag between rate hikes and their impact on the real economy.
Super inteteresting. And then, "It is not that interest rate hikes don’t work. Instead, the variable lags tend to partially eliminate the impact of rate hikes on the economy." ... I also understand this to mean that, because when you have an interest rate raise, which will take effect ca. 40 months later, but then already 25 months later you have a cut again, there is so much noise of lagging and overlapping effects of raises and cuts that it's hard to actually link an interest raise/cut to an outcome?
Exciting. The answer would be even more exciting: what is happening in the current cycle of interest rate cuts? If it takes longer to send signals again because of the service society, it would actually be almost logical that large interest rate increases/decreases of several percent would hardly be tradable any more - then, something else will be sent in these multi-year time frames anyway. Or is the interest rate path down to be judged differently? Thanks!