In economics, as in other sciences, we need universal theories, and we need to understand how they apply in specific contexts. This is quite distinct from the ideas of theoretical and empirical work or basic and applied science.
When we talk about theories of human behaviour (whether in economics, psychology, political science, etc.) we don’t often consider the domain of those theories. The domain is of some importance, as it tells us how widely the theory can be generalised, and in some way, at what level the theory explains things. I can imagine three domains of human behaviour:
1. Contextual behaviour that applies only to some humans, or only some of the time. Those theories may be constrained by situation, culture, or any number of other aspects that can either change in the short or in the long run. For example, the crowding out of intrinsic motivation by extrinsic rewards (famously in some Israeli nursery schools, late pickups increased rather than decreased when a fine was introduced, presumably parents were able to absolve their guilt at coming late by paying the fine).[1]
2. Universal human behaviour, which applies to all humans, but not to other life forms or intelligences. There would be no reason to believe that intelligent aliens we might encounter would exhibit these behaviours. Recoiling from pain might be a good example.
3. Universal behaviour, which applies to all intelligent beings. These are “laws of nature”, and no intelligent being can violate them. A good example from economics is the Theory of Comparative Advantage. Although its applicability to nations is not fully resolved, it is clear that as applied to individuals it is an elegant, surprising, and powerful theory which must apply to all intelligent beings.[2]
One interesting thing to note about these three categories is that strictly speaking, the second category doesn’t exist. If we can imagine aliens not following some behaviour that humans tend to exhibit, then we must also be able to imagine some humans not exhibiting that behaviour. We know of humans who don’t recoil from pain for example (a condition known as congenital insensitivity to pain).[3]
In economics in particular these distinctions appear to be mostly lost. It sometimes seems to me that economists working on different domains argue with one another about what is true when an explicit acknowledgement of the domain would solve the disagreement.
Perhaps there is an unspoken belief that we should be mostly looking for universal theories. That, the crowding out theory of motivation is no good if it isn’t universal. This is a serious mistake. Having discovered that it is not a universal theory we ought to improve our clarity on which specific scenarios it applies in and make that knowledge widely available. When designing incentives, we have to consider these ideas.
The opposite error would be to focus only on the specifics. To give up on trying to discover universal theories. To imagine that all that economics can be is many pieces of specific knowledge applied in many different areas that do not cohere together to form larger ideas.
Behavioural economics seems to focus on domain 1. However, it also sometimes provides interesting insights in domain 3. For example, Herbert Simon’s concept of bounded rationality considers the limited information that all intelligent beings have and that gathering and processing more information is a cost that must be weighed off against the benefits. Bounded rationality is often considered a human shortcoming, but the genius of the idea is that it is in fact a universal aspect of intelligence. There must always be a tradeoff between the speed and quality of a decision based on the information processing resources invested. Unfortunately, most behavioural economics seems uninterested in these kinds of general insights.
The current trend toward empirical economics may inadvertently shift the rest of economics towards domain 1 as well. This seems to be a mistake to me. We need universal theories of economics. And then, we need to understand how they apply to particular conditions in particular societies. These are two separate, very difficult problems. We will not solve them by ignoring the distinction.
Gneezy & Rustichini (2000). There are various competing claims for what exactly was going on there. A recent follow-up, testing various hypotheses in the lab suggests the reaction depends significantly on the personality of the individual (Kornhauser & Tontrup, 2020). Cancer research would also almost exclusively fall into domain 1. Although we could discover some universal theories of cancer, most current research is particularly focused on the specific contexts in which cancer arises and how it can be dealt with.
Also consider altruism. We might say this belongs in the second category. Humans seem to have built-in altruism. They may not always be fully altruistic, but they do appear to be altruistic some of the time, in (somewhat) predictable ways. We can certainly imagine intelligent machines, if not intelligent aliens not being altruistic at all (we may be wrong about that). Obviously, part of the reason we can imagine that is that some humans don’t appear to ever behave altruistically. So, there are exceptions. Altruism isn’t a universal feature of intelligence (at least we don’t think so now, we could be wrong of course).
However, altruism can’t really be considered a universal feature of human behaviour as (apart from individual exceptions mentioned above), human behaviour is completely flexible. If there is no universal constraint, local constraints are shackles we can learn to discard.
Here is a metaphor: Chlorophyll is a near universal method for converting light energy to chemical energy on earth. However, there are alternative processes and there are contexts (emission spectrum on another star) that may make another process dominant on another planet. In that sense chlorophyll isn’t the universally dominant method of photosynthesis even though it appears that way on earth.
Current Trends in Economics and their Shortcoming
In economics, as in other sciences, we need universal theories, and we need to understand how they apply in specific contexts. This is quite distinct from the ideas of theoretical and empirical work or basic and applied science.
When we talk about theories of human behaviour (whether in economics, psychology, political science, etc.) we don’t often consider the domain of those theories. The domain is of some importance, as it tells us how widely the theory can be generalised, and in some way, at what level the theory explains things. I can imagine three domains of human behaviour:
1. Contextual behaviour that applies only to some humans, or only some of the time. Those theories may be constrained by situation, culture, or any number of other aspects that can either change in the short or in the long run. For example, the crowding out of intrinsic motivation by extrinsic rewards (famously in some Israeli nursery schools, late pickups increased rather than decreased when a fine was introduced, presumably parents were able to absolve their guilt at coming late by paying the fine).[1]
2. Universal human behaviour, which applies to all humans, but not to other life forms or intelligences. There would be no reason to believe that intelligent aliens we might encounter would exhibit these behaviours. Recoiling from pain might be a good example.
3. Universal behaviour, which applies to all intelligent beings. These are “laws of nature”, and no intelligent being can violate them. A good example from economics is the Theory of Comparative Advantage. Although its applicability to nations is not fully resolved, it is clear that as applied to individuals it is an elegant, surprising, and powerful theory which must apply to all intelligent beings.[2]
One interesting thing to note about these three categories is that strictly speaking, the second category doesn’t exist. If we can imagine aliens not following some behaviour that humans tend to exhibit, then we must also be able to imagine some humans not exhibiting that behaviour. We know of humans who don’t recoil from pain for example (a condition known as congenital insensitivity to pain).[3]
In economics in particular these distinctions appear to be mostly lost. It sometimes seems to me that economists working on different domains argue with one another about what is true when an explicit acknowledgement of the domain would solve the disagreement.
Perhaps there is an unspoken belief that we should be mostly looking for universal theories. That, the crowding out theory of motivation is no good if it isn’t universal. This is a serious mistake. Having discovered that it is not a universal theory we ought to improve our clarity on which specific scenarios it applies in and make that knowledge widely available. When designing incentives, we have to consider these ideas.
The opposite error would be to focus only on the specifics. To give up on trying to discover universal theories. To imagine that all that economics can be is many pieces of specific knowledge applied in many different areas that do not cohere together to form larger ideas.
Behavioural economics seems to focus on domain 1. However, it also sometimes provides interesting insights in domain 3. For example, Herbert Simon’s concept of bounded rationality considers the limited information that all intelligent beings have and that gathering and processing more information is a cost that must be weighed off against the benefits. Bounded rationality is often considered a human shortcoming, but the genius of the idea is that it is in fact a universal aspect of intelligence. There must always be a tradeoff between the speed and quality of a decision based on the information processing resources invested. Unfortunately, most behavioural economics seems uninterested in these kinds of general insights.
The current trend toward empirical economics may inadvertently shift the rest of economics towards domain 1 as well. This seems to be a mistake to me. We need universal theories of economics. And then, we need to understand how they apply to particular conditions in particular societies. These are two separate, very difficult problems. We will not solve them by ignoring the distinction.
Bibliography
Gneezy, U., & Rustichini, A. (2000). A Fine Is a Price. The Journal of Legal Studies, 29(1), 1–17. https://doi.org/10.1086/468061
Kornhauser, L., Lu, Y., & Tontrup, S. (2020). Testing a fine is a price in the lab. International Review of Law and Economics, 63, 105931. https://doi.org/10.1016/J.IRLE.2020.105931
Gneezy & Rustichini (2000). There are various competing claims for what exactly was going on there. A recent follow-up, testing various hypotheses in the lab suggests the reaction depends significantly on the personality of the individual (Kornhauser & Tontrup, 2020). Cancer research would also almost exclusively fall into domain 1. Although we could discover some universal theories of cancer, most current research is particularly focused on the specific contexts in which cancer arises and how it can be dealt with.
A good example from biology is the Theory of Evolution.
Also consider altruism. We might say this belongs in the second category. Humans seem to have built-in altruism. They may not always be fully altruistic, but they do appear to be altruistic some of the time, in (somewhat) predictable ways. We can certainly imagine intelligent machines, if not intelligent aliens not being altruistic at all (we may be wrong about that). Obviously, part of the reason we can imagine that is that some humans don’t appear to ever behave altruistically. So, there are exceptions. Altruism isn’t a universal feature of intelligence (at least we don’t think so now, we could be wrong of course).
However, altruism can’t really be considered a universal feature of human behaviour as (apart from individual exceptions mentioned above), human behaviour is completely flexible. If there is no universal constraint, local constraints are shackles we can learn to discard.
Here is a metaphor: Chlorophyll is a near universal method for converting light energy to chemical energy on earth. However, there are alternative processes and there are contexts (emission spectrum on another star) that may make another process dominant on another planet. In that sense chlorophyll isn’t the universally dominant method of photosynthesis even though it appears that way on earth.