Among economic phenomena which have in some way been tied up with the existence of uncertainty, three classes may be distinguished: (1) those which b… - Kenneth Arrow

The economic value of information offers no great mysteries in itself. It is easy to prove that one can always do better, whether as a producer or as a consumer, by basing decisions on a signal, provided the signal and the economic variables are not independently distributed. But this remark has an implication for economic decisions; the economic agent is willing to pay for information, for signals.

To conclude, then, we must in a general theory take as our unit a social action, that is, an action involving a large proportion or the entire domain of society. At the most basic axiomatic level, individual actions play little role. The need for a system of public values then becomes evident; actions being collective or interpersonal in nature, so must the choice among them. A public or social value system is essentially a logical necessity.

In view of the magnitude of an economic system, it would take only a very small percentage of improvement in economic stability or growth to make almost any conceivable data collection worthwhile. The situation is analogous to reported results of the use of linear programming in industry; the gains are small in proportion to previous profit levels but still very much larger than the costs of the programming. No country is adequate in respect to its data. In particular the underdeveloped countries, with their ambitious programs, might well ponder whether or not the marginal productivity of investment in better economic statistics is perhaps not higher than almost any conceivable alternative; they have more need and fewer data.