Databases Reference
In-Depth Information
“laborious” this work could be. Here, for example, is Fisher's explanation of how he
generated an interest rate table for the period 1875 to 1895 that compares India bonds
in gold and silver to the estimated and actual percentages of appreciation of gold in
silver:
The methods by which the i rst column [showing rates of interest in silver] is computed
are the same as those explained in the preceding chapter, account being taken of the
fact that the price quotations for rupee paper are not “l at,” so that no corrections for
accrued interest need be applied. For computing the second column [interest rates in
gold] a more laborious method was necessary, due to the fact that the quotations are
not continuous of the same bond. The earlier ones are for a 4% bond and the later for
a 3% bond. The buyer of a 4% bond is regarded as converting it into the 3% at the
current price in 1888, the date of maturity of the earlier bond. As no bond tables apply
to such conversions, tables of present values were used and that rate was found by trial
(and interpolation) which would make the present value of all benei ts equal to the
purchase price.
23
Fisher's reference to the absence of bond tables—and to his substitution of present-
value calculations for those tables—makes it clear that scrubbing was not the only
process that rendered the available numbers commensurate to each other. Beyond
reflecting the laborious labor necessary to make the numbers speak to each other,
the figures that appear in Fisher's tables also constitute outcomes of the algorith-
mic operations—and thus the historical, theoretical, and conventional assumptions—
represented in present-value tables and in other statistical and mathematical methods.
Fisher takes it for granted that the numbers in his tables embody the outcomes of these
operations, both because the only numbers that are useful for his tables must take this
form and because this is the form in which such figures appear in the price lists, gov-
ernment reports, and investors' guides available to him. Fisher acknowledges this when
he casually remarks that he generated the numbers that appear in various tables “by the
aid of the usual brokers' bond tables,” that he used the statistical method called averag-
ing to make other sets of data work, and that he relied on index numbers to generate
yet another table.
24
All of these methods—algorithmic, statistical, and mathematical—
were commonly used in Fisher's day, of course, and there is nothing remarkable in his
use of them. Our point is simply that even an economist who wanted to do empirical
work, who wanted to take the data available in newspapers and trade journals as
his sources, always carried over the assumptions implicit in the ways these data were
