Imagine a financial report in a certain currency, indicating the following numbers (and below them, indicating the amount after conversion to USD, from the same report).
1,394,278
(USD46,743)225,283
(USD7,553)111,518
(USD3,739)62,938
(USD2,110)38,777
(USD1,300)
As you can see, the conversion rate seems to have been off slightly each time:
1,394,278/46,743 = 29.8285946559
225,283/7,553 = 29.8269561764
111,518/3,739 = 29.825621824
62,938/2,110 = 29.828436019
38,777/1,300 = 29.8284615385
However, an important detail is that the numbers are all listed in 1000s (they may thus have been rounded), meaning that they need to me multiplied by one thousand to find their real order of magnitude.
My question is: how much can the conversion rate estimate (see how we have calculated it above) differ between the estimates to be able to discover whether or not the source has actually used multiple conversion rates on the actual un-rounded numbers?