An old post by Arya Gaduh of Nalar Ekonomi caught my attention. It is about Indonesia’s latest poverty rates published by Statistics Indonesia (BPS): It reports that the number of Indonesian poor declines by 2.13 million, or about 1.2 percentage points (from 17.75% to 16.58%).
The fall is not statistically significant, however, as Arya correctly suggests.
A back of envelope calculation gives me a standard error of the difference in the poverty rates of about 2 percentage points. It is so large so that it may overwhelm the estimated difference.
It means we cannot really say whether the poverty rate last year (17.75%) is really different from the rate this year (16.58%).
In that sense, the BPS’s press release is rather misleading. It reports that poverty rate declines, but it does not tell us the standard error of the fall.
It is as if BPS gets the number of the poor from the entire Indonesia’s population, not from a small sample of it only.
Of course, I am glad to hear that poverty rate falls. But, at the same time, BPS should acknowledge that there is too much noise in the data. That we do not really have convincing evidence that poverty rate has actually declined.