Last weekend, while the world was paying attention to the performance of athletes at the Paris Olympics, the July employment statistics were released in the United States. The unemployment rate in this employment statistics exceeded market expectations, and the "Sahm Rule ," a signal of an economic recession, was lit, which became a hot topic.
The Sarm rule is calculated based on the unemployment rate proposed by former Federal Reserve Board member Claudia Sarm. Specifically, it states that if the value (percentage) exceeds 0.5, there is a high possibility of the economy falling into a recession.
In this employment report, the unemployment rate for July 2024 was announced to be 4.3%, with the therm rule percentage being 0.53, the first time it has exceeded 0.5 since COVID-19.
The Therm Rule percentage is calculated from the monthly unemployment rate using the following definition:
① Average unemployment rate for the past three months (three-month moving average)
②The minimum of the three-month moving average over the past 12 months
Percentage (therm rule) = ① - ② (if ≧ 0.5 the therm rule signal will light up)
In this blog, we will calculate this percentage using monthly unemployment rate data in a JMP data table and visualize the trend in this value using a graph.
Function to calculate moving average in JMP
The therm rule percentage is calculated as follows:
- Calculate the "3-Month Average" from the "Unemployment Rate" (①)
- Calculate "Minimum of 3-Month Average in Prior 12 Months" from "3-Month Moving Average"
- Subtract ② from ① (Percentage (Sahm Rule))
As an example, to calculate the percentage for the most recent month of July 2024, follow these steps:
- 3-month moving average: (4 + 4.1 + 4.3) ÷ 3 = 4.13
- Minimum of 3-month moving average (yellow part above) over the past 12 months = 3.6 (July 2023)
- Percentage: 4.13 - 3.6 = 0.53
Let's calculate it using the JMP formula.
To calculate the three-month moving average of i., use the Col Moving Average() function to calculate the moving average and specify the following formula:
There are many arguments, but by clicking the ? icon at the top of the formula editor, help will appear and you can check the meaning of each argument.
The numeric arguments 1, 2, 0, and 1 have the following meanings:
| argument |
値 |
meaning |
| weighting |
1 |
When calculating the moving average, the weight is set to 1 (no weighting is given to past values). |
| before |
2 |
The target is up to two points in time (the average of the current point, the point one point in time, and the point two points in time) |
| after |
0 |
Values after the current time are not included |
| partial window is missing |
1 |
If you set it to 1, the data that has values up to two points in time will be included (in the example, the values in the first and second rows will be missing values). If you set it to 0, the first row will contain the current value, and the second row will contain the average of the current value and the value one point in time before. |
The minimum value of the three-month moving average for the past 12 months in ⅱ uses the Lag(x, n) function, where n returns the value of column x n points prior. Here, we are calculating the minimum value among the three-month moving averages from 1 point prior to 12 points prior.
Finally, subtract the value found in ii from the value found in i.
Changes in unemployment rate and therm rule since 2000
Now that we have the values, we can use Graph Builder to plot the unemployment rate (top of the graph) and the therm rule percentage (bottom of the graph) over time.
The trend in the unemployment rate is shown by the orange broken line, and although there is a slight upward trend, looking at the smooth line (red dotted line), it is difficult to say that the trend is still upward. If the unemployment rate continues to remain flat or rise in the future, the smooth line will also show an upward trend.
Looking at the therm rule percentages (blue bar graph), we can see that there have been three large peaks in the past 2000s.The main causes of these peaks are the IT bubble around 2001, the Lehman shock around 2008, and the coronavirus shock of 2020.
Excluding the coronavirus shock, in the most recent period (2023 to July 2024), it appears that a peak is beginning to form, similar to those seen during the IT bubble and the Lehman shock.
Will the economy really enter a recession? Incidentally, if the unemployment rate announced next month (August 2024) is 3.9%, the percentage will be 0.47, which is below 0.5.
It seems unlikely that the unemployment rate will suddenly drop by that much, but it would be a good idea to continue to keep an eye on the Sarm rule, which is an economic signal.
Data source: Unemployment Rate (UNRATE) | FRED | St. Louis Fed (stlouisfed.org)
by Naohiro Masukawa (JMP Japan)
Naohiro Masukawa - JMP User Community
Last weekend, the employment report for August was released, showing the unemployment rate at 4.2%, down 0.1% from 4.3% in July, but the therm rule calculation gave it a rate of 0.57, still above 0.5.
