I have written a number posts (here here here, and here) over the past few years citing an article by one of my favorite UCLA luminaries, Jack Hirshleifer, of the fabled UCLA economics department of the 1950s, 1960s, 1970s and 1980s. Like everything Hirshleifer wrote, the article, “The Private and Social Value of Information and the Reward to Inventive Activity,” published in 1971 in the American Economic Review, is deeply insightful, carefully reasoned, and lucidly explained, reflecting the author’s comprehensive mastery of the whole body of neoclassical microeconomic theory.
Hirshleifer’s article grew out of a whole literature inspired by two of Hayek’s most important articles “Economics and Knowledge” in 1937 and “The Use of Knowledge in Society” in 1945. Both articles were concerned with the fact that, contrary to the assumptions in textbook treatments, economic agents don’t have complete information about all the characteristics of the goods being traded and about the prices at which those goods are available. Hayek was aiming to show that markets are characteristically capable of transmitting information held by some agents in a condensed form to make it usable by other agents. That role is performed by prices. It is prices that provide both information and incentives to economic agents to formulate and tailor their plans, and if necessary, to readjust those plans in response to changed conditions. Agents need not know what those underlying changes are; they need only observe, and act on, the price changes that result from those changes.
Hayek’s argument, though profoundly insightful, was not totally convincing in demonstrating the superiority of the pure “free market,” for three reasons.
First, economic agents base decisions, as Hayek himself was among the first to understand, not just on actual current prices, but also on expected future prices. Although traders sometimes – but usually don’t — know what the current price of something is, one can only guess – not know — what the price of that thing will be in the future. So, the work of providing the information individuals need to make good economic decisions cannot be accomplished – even in principle – just by the adjustment of prices in current markets. People also need enough information to make good guesses – form correct expectations — about future prices.
Second, economic agents don’t automatically know all prices. The assumption that every trader knows exactly what prices are before executing plans to buy and sell is true, if at all, only in highly organized markets where prices are publicly posted and traders can always buy and sell at the posted price. In most other markets, transactors must devote time and effort to find out what prices are and to find out the characteristics of the goods that they are interested in buying. It takes effort or search or advertising or some other, more or less costly, discovery method for economic agents to find out what current prices are and what characteristics those goods have. If agents aren’t fully informed even about current prices, they don’t necessarily make good decisions.
Libertarians, free marketeers, and other Hayek acolytes often like to credit Hayek with having solved or having shown how “the market” solves “the knowledge problem,” a problem that Hayek definitively showed a central-planning regime to be incapable of solving. But the solution at best is only partial, and certainly not robust, because markets never transmit all available relevant information. That’s because markets transmit only information about costs and valuations known to private individuals, but there is a lot of information about public or social valuations and costs that is not known to private individuals and rarely if ever gets fed into, or is transmitted by, the price system — valuations of public goods and the social costs of pollution for example.
Third, a lot of information is not obtained or transmitted unless it is acquired, and acquiring information is costly. Economic agents must search for relevant information about the goods and services that they are interested in obtaining and about the prices at which those goods and services are available. Moreover, agents often engage in transactions with counterparties in which one side has an information advantage over the other. When traders have an information advantage over their counterparties, the opportunity for one party to take advantage of the inferior information of the counterparty may make it impossible for the two parties to reach mutually acceptable terms, because a party who realizes that the counterparty has an information advantage may be unwilling to risk being taken advantage of. Sometimes these problems can be surmounted by creative contractual arrangements or legal interventions, but often they can’t.
To recognize the limitations of Hayek’s insight is not to minimize its importance, either in its own right or as a stimulus to further research. Important early contributions (all published between 1961 and 1970) by Stigler (“The Economics of Information”) Ozga (“Imperfect Markets through Lack of Knowledge”), Arrow (“Economic Welfare and the Allocation of Resources for Invention”), Demsetz (“Information and Efficiency: Another Viewpoint”) and Alchian (“Information Costs, Pricing, and Resource Unemployment”) all analyzed the problem of incomplete and limited information and the incentives for acquiring information, the institutions and market arrangements that arise to cope with limited information and the implications for economic efficiency of these limitations and incentives. They can all be traced directly or indirectly to Hayek’s early contributions. Among the important results that seem to follow from these early papers was that the inability of those discovering or creating new knowledge to appropriate the net benefits accruing from the knowledge implied that the incentive to create new knowledge is less than optimal owing to their inability to claim full property rights over new knowledge through patents or other forms of intellectual property.
Here is where Hirshleifer’s paper enters the picture. Is more information always better? It would certainly seem that more of any good is better than less. But how valuable is new information? And are the incentives to create or discover new information aligned with the value of that information? Hayek’s discussion implicitly assumed that the amount of information in existence is a given stock, at least in the aggregate. How can the information that already exists be optimally used? Markets help us make use of the information that already exists. But the problem addressed by Hirshleifer was whether the incentives to discover and create new information call forth the optimal investment of time, effort and resources to make new discoveries and create new knowledge.
Instead of focusing on the incentives to search for information about existing opportunities, Hirshleifer analyzed the incentives to learn about uncertain resource endowments and about the productivity of those resources.
This paper deals with an entirely different aspect of the economics of information. We here revert to the textbook assumption that markets are perfect and costless. The individual is always fully acquainted with the supply-demand offers of all potential traders, and an equilibrium integrating all individuals’ supply-demand offers is attained instantaneously. Individuals are unsure only about the size of their own commodity endowments and/or about the returns attainable from their own productive investments. They are subject to technological uncertainty rather than market uncertainty.
Technological uncertainty brings immediately to mind the economics of research and invention. The traditional position been that the excess of the social over the private value of new technological knowledge leads to underinvestment in inventive activity. The main reason is that information, viewed as a product, is only imperfectly appropriable by its discoverer. But this paper will show that there is a hitherto unrecognized force operating in opposite direction. What has been scarcely appreciated in the literature, if recognized at all, is the distributive aspect of access to superior information. It will be seen below how this advantage provides a motivation for the private acquisition and dissemination of technological information that is quite apart from – and may even exist in the absence – of any social usefulness of that information. (p. 561)
The key insight motivating Hirshleifer was that privately held knowledge enables someone possessing that knowledge to anticipate future price movements once the privately held information becomes public. If you can anticipate a future price movement that no one else can, you can confidently trade with others who don’t know what you know, and then wait for the profit to roll in when the less well-informed acquire the knowledge that you have. By assumption the newly obtained knowledge doesn’t affect the quantity of goods available to be traded, so acquiring new knowledge or information provides no social benefit. In a pure-exchange model, newly discovered knowledge provides no net social benefit; it only enables better-informed traders to anticipate price movements that less well-informed traders don’t see coming. Any gains from new knowledge are exactly matched by the losses suffered by those without that knowledge. Hirshleifer called the kind of knowledge that enables one to anticipate future price movements “foreknowledge,” which he distinguished from actual discovery .
The type of information represented by foreknowledge is exemplified by ability to successfully predict tomorrow’s (or next year’s) weather. Here we have a stochastic situation: with particular probabilities the future weather might be hot or cold, rainy or dry, etc. But whatever does actually occur will, in due time, be evident to all: the only aspect of information that may be of advantage is prior knowledge as to what will happen. Discovery, in contrast, is correct recognition of something that is hidden from view. Examples include the determination of the properties of materials, of physical laws, even of mathematical attributes (e.g., the millionth digit in the decimal expansion of “π”). The essential point is that in such cases nature will not automatically reveal the information; only human action can extract it. (562)
Hirshleifer’s result, though derived in the context of a pure-exchange economy, is very powerful, implying that any expenditure of resources devoted to finding out new information that enables the first possessor of the information to predict price changes and reap profits from trading is unambiguously wasteful by reducing total consumption of the community.
[T]he community as a whole obtains no benefit, under pure exchange, from either the acquisition or the dissemination (by resale or otherwise) of private foreknowledge. . . .
[T]he expenditure of real resources for the production of technological information is socially wasteful in pure exchange, as the expenditure of resources for an increase in the quantity of money by mining gold is wasteful, and for essentially the same reason. Just as a smaller quantity of money serves monetary functions as well as a larger, the price level adjusting correspondingly, so a larger amount of foreknowledge serves no social purpose under pure exchange that the smaller amount did not. (pp. 565-66)
Relaxing the assumption that there is no production does not alter the conclusion, because the kind of information that is discovered, even if it did lead to efficient production decisions that increase the output of goods whose prices rise sooner as a result of the new information than they would have otherwise. But if the foreknowledge is privately obtained, the private incentive is to use that information by trading with another, less-well-informed, trader, at a price the other trader would not agree to if he weren’t at an information disadvantage. The private incentive to use foreknowledge that might cause a change in production decisions is not to use the information to alter production decisions but to use it to trade with, and profit from, those with inferior knowledge.
[A]s under the regime of pure exchange, private foreknowledge makes possible large private profit without leading to socially useful activity. The individual would have just as much incentive as under pure exchange (even more, in fact) to expend real resources in generating socially useless private information. (p. 567)
If the foreknowledge is publicly available, there would be a change in production incentives to shift production toward more valuable products. However, the private gain if the information is kept private greatly exceeds the private value of the information if the information is public. Under some circumstances, private individuals may have an incentive to publicize their private information to cause the price increases in expectation of which they have taken speculative positions. But it is primarily the gain from foreseen price changes, not the gain from more efficient production decisions, that creates the incentive to discover foreknowledge.
The key factor underlying [these] results . . . is the distributive significance of private foreknowledge. When private information fails to lead to improved productive alignments (as must necessarily be the case in a world of pure exchange, and also in a regime of production unless there is dissemination effected in the interest of speculation or resale), it is evident that the individual’s source of gain can only be at the expense of his fellows. But even where information is disseminated and does lead to improved productive commitments, the distributive transfer gain will surely be far greater than the relatively minor productive gain the individual might reap from the redirection of his own real investment commitments. (Id.)
Moreover, better-informed individuals – indeed individuals who wrongly believe themselves to be better informed — will perceive it to be in their self-interest to expend resources to disseminate the information in the expectation that the ensuing price changes would redound to their profit. The private gain expected from disseminating information far exceeds the social benefit from the prices changes once the new information is disseminated; the social benefit from the price changes resulting from the disseminated information corresponds to an improved allocation of resources, but that improvement will be very small compared to the expected private profit from anticipating the price change and trading with those that don’t anticipate it.
Hirshleifer then turns from the value of foreknowledge to the value of discovering new information about the world or about nature that makes a contribution to total social output by causing a shift of resources to more productive uses. Inasmuch as the discovery of new information about the world reveals previously unknown productive opportunities, it might be thought that the private incentive to devote resources to the discovery of technological information about productive opportunities generates substantial social benefits. But Hirshleifer shows that here, too, because the private discovery of information about the world creates private opportunities for gain by trading based on the consequent knowledge of future price changes, the private incentive to discover technological information always exceeds the social value of the discovery.
We need only consider the more general regime of production and exchange. Given private, prior, and sure information of event A [a state of the world in which a previously unknown natural relationship has been shown to exist] the individual in a world of perfect markets would not adapt his productive decisions if he were sure the information would remain private until after the close of trading. (p. 570)
Hirshleifer is saying that the discovery of a previously unknown property of the world can lead to an increase in total social output only by causing productive resources to be reallocated, but that reallocation can occur only if and when the new information is disclosed. So if someone discovers a previously unknown property of the world, the discoverer can profit from that information by anticipating the price effect likely to result once the information is disseminated and then making a speculative transaction based on the expectation of a price change. A corollary of this argument is that individuals who think that they are better informed about the world will take speculative positions based on their beliefs, but insofar as their investments in discovering properties of the world lead them to incorrect beliefs, their investments in information gathering and discovery will not be rewarded. The net social return to information gathering and discovery is thus almost certainly negative.
The obvious way of acquiring the private information in question is, of course, by performing technological research. By a now familiar argument we can show once again that the distributive advantage of private information provides an incentive for information-generating activity that may quite possibly be in excess of the social value of the information. (Id.)
Finally, Hirshliefer turns to the implications for patent policy of his analysis of the private and social value of information.
The issues involved may be clarified by distinguishing the “technological” and “pecuniary” effects of invention. The technological effects are the improvements in production functions . . . consequent upon the new idea. The pecuniary effects are the wealth shifts due to the price revaluations that take place upon release and/or utilization of the information. The pecuniary effects are purely redistributive.
For concreteness, we can think in terms of a simple cost-reducing innovation. The technological benefit to society is, roughly, the integrated area between the old and new marginal-cost curves for the preinvention level of output plus, for any additional output, and the area between the demand curve and the new marginal-cost curve. The holder of a (perpetual) patent could ideally extract, via a perfectly discriminatory fee policy, this entire technological benefit. Equivalence between the social and private benefits of innovation would thus induce the optimal amount of private inventive activity. Presumably it is reasoning of this sort that underlies the economic case for patent protection. (p. 571)
Here Hirshleifer is uncritically restating the traditional analysis for the social benefit from new technological knowledge. But the analysis overstates the benefit, by assuming incorrectly that, with no patent protection, the discovery would never be made. If the discovery would be made without patent protection, then obviously the technological benefit to society is only the area indicated over a limited time horizon, so a perpetual patent enabling the holder of the patent to extract all additional consumer and producer surplus flowing from invention in perpetuity would overcompensate the patent holder for the invention.
Nor does Hirshleifer mention the tendency of patents to increase the costs of invention, research and development owing to the royalties subsequent inventors would have to pay existing patent holders for infringing inventions even if those inventions were, or would have been, discovered with no knowledge of the patented invention. While rewarding some inventions and inventors, patent protection penalizes or blocks subsequent inventions and inventors. Inventions are outputs, but they are also inputs. If the use of past inventions is made more costly by new inventors, it is not clear that the net result will be an increase in the rate of invention.
Moreover, the knowledge that a patented invention may block or penalize a new invention that infringes on an existing patent or a patent that issues before a new invention is introduced, may in some cases cause an overinvestment in research as inventors race to gain the sole right to an invention, in order to avoid being excluded while gaining the right to exclude others.
Hirshleifer does mention some reasons why maximally rewarding patent holders for their inventions may lead to suboptimal results, but fails to acknowledge that the conventional assessment of the social gain from new invention is substantially overstated or patents may well have a negative effect on inventive activity in fields in which patent holders have gained the right to exclude potentially infringing inventions even if the infringing inventions would have been made without the knowledge publicly disclosed by the patent holders in their patent applications.
On the other side are the recognized disadvantages of patents: the social costs of the administrative-judicial process, the possible anti-competitve impact, and restriction of output due to the marginal burden of patent fees. As a second-best kind of judgment, some degree of patent protection has seemed a reasonable compromise among the objectives sought.
Of course, that judgment about the social utility of patents is not universally accepted, and authorities from Arnold Plant, to Fritz Machlup, and most recently Michele Boldrin and David Levine have been extremely skeptical of the arguments in favor of patent protection, copyright and other forms of intellectual property.
However, Hirshleifer advances a different counter-argument against patent protection based on his distinction between the private and social gains derived from information.
But recognition of the unique position of the innovator for forecasting and consequently capturing portions of the pecuniary effects – the wealth transfers due to price revaluation – may put matters in a different light. The “ideal” case of the perfectly discriminating patent holder earning the entire technological benefit is no longer so ideal. (pp. 571-72)
Of course, as I have pointed out, the ‘“ideal” case’ never was ideal.
For the same inventor is in a position to reap speculative profits, too; counting these as well, he would clearly be overcompensated. (p. 572)
Hirshleifer goes on to recognize that the capacity to profit from speculative activity may be beyond the capacity or the ken of many inventors.
Given the inconceivably vast number of potential contingencies and the costs of establishing markets, the prospective speculator will find it costly or even impossible ot purchase neutrality from “irrelevant” risks. Eli Whitney [inventor of the cotton gin who obtained one of the first US patents for his invention in 1794] could not be sure that his gin would make cotton prices fall: while a considerable force would clearly be acting in that direction, a multitude of other contingencies might also have possibly affected the price of cotton. Such “uninsurable” risks gravely limit the speculation feasible with any degree of prudence. (Id.)
HIrshleifer concludes that there is no compelling case either for or against patent protection, because the standard discussion of the case for patent protection has not taken into consideration the potential profit that inventors can gain by speculating on the anticipated price effects of their patents. Of course the argument that inventors are unlikely to be adept at making such speculative plays is a serious argument, we have also seen the rise of patent trolls that buy up patent rights from inventors and then file lawsuits against suspected infringers. In a world without patent protection, it is entirely possible that patent trolls would reinvent themselves as patent speculators, buying up information about new inventions from inventors and using that information to engage in speculative activity based on that information. By acquiring a portfolio of patents such invention speculators could pool the risks of speculation over their entire portfolio, enabling them to speculate more effectively than any single inventor could on his own invention. Hirshleifer concludes as follows:
Even though practical considerations limit the effective scale and consequent impact of speculation and/or resale [but perhaps not as much as Hirshleifer thought], the gains thus achievable eliminate any a priori anticipation of underinvestment in the generation of new technological knowledge. (p. 574)
And I reiterate one last time that Hirshleifer arrived at his non-endorsement of patent protection even while accepting the overstated estimate of the social value of inventions and neglecting the tendency of patents to increase the cost of inventive activity.