Page:Stabilizing the dollar, Fisher, 1920.djvu/244

190 Thus, if the adjustment interval is two months and it is assumed that the effect of any adjustment were not felt until six times that interval, or a whole year, the index number would at most deviate only 6%, assuming the other conditions unchanged from the standard case.

As a matter of fact the lag is not great.

Our experience during the war and other evidence mentioned elsewhere (Chapter II, § 8 and Appendix I, § 3) show that the influence of inflation or contraction is apparently rather prompt, the lag being probably less than two months, and possibly less than one month for an index number composed of the most responsive commodities.

It is desirable that our adjustment intervals should not be too short compared with the lag, say not shorter than a quarter of the lag.

On the other hand, the adjustment interval might be taken longer than the lag. For such a case the calculations and results would be the same as where the lag is one entire period. The influence of the adjustment would then be complete some time before the following adjustment date arrived; but since no index number is calculated during the interval, our calculations would not be affected.

Ideally, i.e. to secure the greatest attainable degree of closeness to par, the adjustment interval should be as nearly equal to the lag as possible. If the interval is shorter than the lag the influence is not felt fully until another adjustment, perhaps in the opposite direction, is made. A daily adjustment would therefore not help but hurt the closeness of the approximation. If the interval is longer than the lag, the price level is left for the balance of the interval to vary uncorrected. We would be neglecting the opportunity to correct it promptly.

C. Changing the Assumption as to the "Tendency."

(a) Assumptions same as in standard case except: tendency increased from 1% to 2% per adjustment interval.