I have now run some calculations, and I can confirm that the country weighting scheme used here seems very similar, although not identical, to the one used by the OECD for the calculation of the global CPI: the weighted average thus calculated slightly overestimates the OECD aggregate by ~1.1 ppc, while it is consistent across the last 5 quarters, and it also captures the trend.
By now, we have Q2 data from 13 (out of 38) OECD countries, which represent ~72% of the OECD according to the CPI weighting scheme. Save for Germany and France, this list includes the largest OECD economies (weight > 4), as well as a number of "middle" economies (weight 2-3); notice that the majority of countries (23 out of 38) have weight < 1 (14 of which have weight < 0.5).
Here is a table with the last 6 quarters; weighted average is the average calculated as explained above, while weighted average 13 is the weighted average of the 13 countries for which we already have Q2 readings (data in 2-digit precision downloaded from the OECD data explorer):
2024-Q1
2024-Q2
2024-Q3
2024-Q4
2025-Q1
2025-Q2
weighted average
71.26
71.30
71.32
71.29
71.40
OECD aggregate
70.17
70.22
70.26
70.20
70.34
weighted average 13
71.16
71.15
71.16
71.10
71.19
71.16
NOTE: the value of 71.16 for 2025-Q2 was calculated using the US Q2 reading of 71.7, which was subsequently revised upwards to 71.8
So, trying to use weighed average 13 as an early proxy for Q2 (the point of the whole exercise...), the message here would seem to be "no big change, but probably downward"; the 71.16 reading seems to be consistent with an eventual OECD aggregate in the range of 70.17 - 70.26 (which would be rounded as either 70.2 or 70.3 in our resolution source).
So:
taking the worse end of this range for Q2, i.e. 70.17 (a 0.17 ppc drop from Q1)
assuming no further downward revision of Q1 (70.34)
noticing that, in order to get a reading of 69.9% in our resolution source, the actual underlying value (which is calculated with a precision of 5 decimal digits) should be less than 69.95 (a value of 69.951 would be rounded to 70.0)
I calculate that, in order to have an annual reading of 69.9 in our resolution source, the average of both Q3 and Q4 should be lower than 69.65.
This sounds like a huge drop (the last time we had a reading of this order was in 2022-Q4), which, although not impossible, seems improbable to me. So, I am moving the "more than or equal to 70%" bin higher.
Files
Why might you be wrong?
A scenario where both my worst-case estimate for Q2 (70.17) is wrong to the downside, and the situation in the rest of 2025 goes much worse than I anticipate.
Why do you think you're right?
I have now run some calculations, and I can confirm that the country weighting scheme used here seems very similar, although not identical, to the one used by the OECD for the calculation of the global CPI: the weighted average thus calculated slightly overestimates the OECD aggregate by ~1.1 ppc, while it is consistent across the last 5 quarters, and it also captures the trend.
By now, we have Q2 data from 13 (out of 38) OECD countries, which represent ~72% of the OECD according to the CPI weighting scheme. Save for Germany and France, this list includes the largest OECD economies (weight > 4), as well as a number of "middle" economies (weight 2-3); notice that the majority of countries (23 out of 38) have weight < 1 (14 of which have weight < 0.5).
Here is a table with the last 6 quarters; weighted average is the average calculated as explained above, while weighted average 13 is the weighted average of the 13 countries for which we already have Q2 readings (data in 2-digit precision downloaded from the OECD data explorer):
NOTE: the value of 71.16 for 2025-Q2 was calculated using the US Q2 reading of 71.7, which was subsequently revised upwards to 71.8
So, trying to use weighed average 13 as an early proxy for Q2 (the point of the whole exercise...), the message here would seem to be "no big change, but probably downward"; the 71.16 reading seems to be consistent with an eventual OECD aggregate in the range of 70.17 - 70.26 (which would be rounded as either 70.2 or 70.3 in our resolution source).
So:
I calculate that, in order to have an annual reading of 69.9 in our resolution source, the average of both Q3 and Q4 should be lower than 69.65.
This sounds like a huge drop (the last time we had a reading of this order was in 2022-Q4), which, although not impossible, seems improbable to me. So, I am moving the "more than or equal to 70%" bin higher.
Why might you be wrong?
A scenario where both my worst-case estimate for Q2 (70.17) is wrong to the downside, and the situation in the rest of 2025 goes much worse than I anticipate.