Economics

24 – Thinking like an economist 6: The value of information

There is something unsettling about paying for information. It’s so intangible, but it can sometimes be valuable. How valuable? The answer has to be case-specific, different for each decision maker and for each type of information. It has to be forward looking, since at the time we decide to pay for information we can only anticipate its usefulness to us.

Economics focuses on decisions (see PD#18) and so economists are interested in the value of information that is used to make decisions. In this piece I will outline how one can validly put a value on information when you don’t yet know its content.

To make it tangible, we will consider the example of a farmer who is interested in a potential new land management option (applying lime) that can reduce a form of soil degradation (soil acidity). The farmer can obtain, at some cost, information about the current level of soil acidity. This information is a form of “environmental indicator”. The link between the information and the decision is that if the soil is highly acidic, it is more likely to be worthwhile applying lime. The question is, what is the value of the information obtained by measuring the environmental indicator?

In essence, the value depends on the answers to three questions.

  1. What would the farmer do without the additional information?
  2. What might the farmer do differently with the additional information?
  3. What difference does this make to payoffs?

That sounds simple enough, but putting it into practice can be complex. We will work through it step by step – please concentrate for a minute.

In general, the farmer can have a fair guess at the current level of soil acidity, based on current knowledge and preconceptions. These preconceptions might be based on past experience, long-term forecasts, information from other paddocks, something they once heard a neighbour say, or they may be based on a previous observation within the same paddock. It would be possible for the farmer to go ahead and make a best-bet decision based solely on his or her preconceptions without extra information (e.g. the best-bet decision without observing the indicator might be to apply lime). This provides the answer to question 1.

Alternatively, the farmer could measure soil acidity before making the decision. With this extra information, an improved decision may be possible. If it turns out that acidity is lower than expected, there may be no need to apply lime. To value the information, we need to identify in advance every possible level of the indicator that might be observed, and estimate their probabilities of being observed. Then, for each of those indicator levels we ask, if the farmer did observe that level, what difference would it make to his or her decision? This provides the answer to question 2.

This process involves the farmer providing subjective probabilities that acidity, once measured, will take different values within the potential range. In my experience, most farmers have an excellent intuitive feel for probabilities, and they can handle a complex question like this. We asked them to do almost exactly this in one research project, and they were remarkably good at it.

Some potential observations of soil acidity would probably change the prior best-bet decision, while others would not. Whether it changes the decision depends on factors like:

  • whether the observation is significantly different from the farmer’s preconceptions,
  • how accurately the observation can be made,
  • how applicable the observation is to the whole area for which a decision is needed,
  • whether the prior best-bet decision is finely balanced or clear cut, and
  • how well cause and effect are understood (e.g. do we know how acidity affects yields?).

Finally, to answer question 3, we take the set of answers to question 2 and estimate the difference in payoffs between the prior best-bet and the revised, better-informed decisions. We weight them by the anticipated probabilities of each indicator level, and add them up to get an overall expected value for the information. Whether it seems worthwhile to observe the indicator depends on whether these expected benefits from improving the decision outweigh the costs of the extra information.

Although the value of information is obviously case-specific, some general insights are possible, including the following.

  1. If information does not have the potential to change a management choice, it has no value, economic, social or environmental, other than perhaps its intrinsic interest value.
  2. The change in management, if it occurs, is the result of a reduction in uncertainty about the payoffs from different decision options. The reduction in uncertainty allows the decision maker to refine his or her best-bet strategy.
  3. In many cases, the value of continuing to monitor an indicator would fall over time as uncertainty is reduced. In some cases, the value of observing a sustainability indicator may be dramatically reduced after a small number of observations, potentially just one.
  4. The gross value of monitoring an indicator (the value before deducting the cost of monitoring) can never be negative. At worst, its value would be zero if there was no realistic probability of any resulting change in management.
  5. A necessary (but not sufficient) condition for the value of monitoring an indicator to be high is for the payoffs to be sensitive to management choices. In many circumstances in agriculture, payoffs are not sensitive to management choices.
  6. If productivity is very sensitive to management choices, the optimal choice may be so obvious that there is little value in collecting further information about it.
  7. If there is a high level of uncertainty about the relationship between the level of an indicator and the payoff (financial or environmental), the value of monitoring the indicator will be low since monitoring will not reduce uncertainty significantly.

These insights provide some basis for understanding why the level of monitoring of sustainability indicators by farmers has been less than advocates of the approach would like. In essence, theory predicts that special circumstances must prevail for monitoring to be worthwhile. In my experience of applying this framework, once you allow for the fact that people can often make reasonable decisions based on their existing knowledge, the value of extra information is not as high as people tend to assume.

David Pannell, The University of Western Australia

Postscript, 3 Nov 2004. Jim Walcott noted that farmers commonly keep close track of rainfall data (and I would add prices), and wondered whether this was consistent with the “special circumstances” I referred to. Absolutely. These variables have all the required characteristics for monitoring to be worthwhile: they directly affect an important decision problem (or more than one), they have big and direct impacts on the relative payoffs of different decision options, the decision choice without the information is not obvious or irrelevant, they are easy to measure reasonably accurately, and farmers have pretty much perfect understanding of cause and effect (e.g. the link between rainfall and final payoffs). By contrast, environmental indicators commonly lack several, and sometimes all, of these characteristics.

Further reading

Pannell, D.J. (2003). What is the Value of a Sustainability Indicator? Economic and Social Issues in Monitoring and Management for Sustainability. Australian Journal of Experimental Agriculture 43(3): 239-243. Final journal version (48K pdf file)

Pannell D.J. and Glenn N.A. (2000). A Framework for Economic Evaluation and Selection of Sustainability Indicators in Agriculture, Ecological Economics 33(1): 135-149. full paper (93 K)