The neoclassical normative ideal, that humans are unboundedly rational, is – and has been for some time now – losing traction. At the same time, theories identifying significant practical limits on human decision-making in the real world have steadily gained momentum in the sphere of behavioural economics. Economics is no longer divorced from the reality in which it applies. Though academics at the extreme end of the spectrum contend that models of unbounded rationality should be discarded entirely because they lack any predictive power, I will argue that this is a simplistic solution to a complex problem. Whilst human behaviour cannot be modelled according to reductivist rules and methodologies, there is currently no single theory of human behaviour that is immune from criticism. The better solution is not singular; it is multi-faceted. It is to recognise that neoclassical economics is normative and indicates how we should make decisions. Conversely, bounded rationality or behavioural economics is descriptive – it shows how we do make decisions. By understanding how these theories interrelate, we can better understand how to modify our actual decision making to bring it more in line with normative decision-making.
So what is neoclassical rationality?
Neoclassical rationality (or Rational Choice Theory – RCT) is underpinned by a number of assumptions as to how a ‘rational’ person supposedly makes decisions. Rational decision-makers have well defined and consistent preferences and consider all available information when making decisions. For example, the efficient market hypothesis assumes there are rational investors that are able to incorporate all available information when making investments and therefore capitalise on unexploited profit opportunities as soon as they materialise. Rational individuals are not constrained by context. Nor are they limited by environmental factors (framing, advertising etc.), cognitive factors (memory constraints etc.) or time constraints. It does not take a Nobel Prize winner to realise that this conception of human behaviour is reductive, simplistic and approximate.
However, as I will show later, I do not believe that neoclassical rationality is entirely superfluous.
Neoclassical economics is sharply contrasted by behavioural economics, which paints a very different picture of human decision-making. Though there are a range of separate theories that have evolved in this field, I will focus on the work of Daniel Kahnman and Amos Tversky (two respected behavioural economists).
Kahnman and Tversky deconstruct human decision-making into two separate ways of thinking: system one and system two. System one is a metaphor that describes intuitive, automatic and effortless thinking. In contrast, system two describes slow, controlled and deliberate thinking. Many examples of sub-rational and even irrational decisions are the product of system one and result from the fact that people often accept their intuitive decisions without question. Excessive and undue reliance on system one leads to a range of predictable cognitive biases that translate into systematic errors in judgement and choice.
These cognitive biases are pervasive in actual human decision-making. And fundamentally, these cognitive biases stem from heuristics. Put simply, heuristics are simplifying mental short cuts that we rely on when making decisions. The human mind is not well designed to deal with uncertainty, probability and excessive complexity. Heuristics therefore enable us to simplify our subjective realities. But they come at a corresponding cost. Cognitive biases can manifest in misunderstandings of probabilities, a propensity to construct simplistic causal explanations for inherently random events and a vulnerability to make different decisions purely on the basis of how information is framed (see the Asian Disease Experiment, the Linda problem etc.).
Self-evidently, the model of human behaviour described by Kahnman and Tversky is a more accurate characterisation of human decision-making. But it is also inconsistent with the predictions yielded by RCT.
Reconciling theory with reality
So does this mean that rationality should be discarded from mainstream economics?
My answer is no.
I believe that the previous bounded rationality theories are fundamentally descriptive. They characterise how people do make decisions. But bounded rationality is not normative. It does not indicate how people should make decisions. The fact is, people can make better decisions in line with RCT. RCT represents the standard of decision-making that we should aspire.
And here is why…
- Firstly, there are enormous difficulties in incorporating bounded rationality – as a collection of insights – into a coherent economic theory. Macroeconomic models are often reasonably good approximations of reality. Attempting to incorporate the intricacies of human decision-making may improve the predictive capacity of economic models, but at a higher cost in terms of complexity. Moreover, unless bounded rationality can be analysed predictably, it may lead to ambiguous and conflicting insights when included in economic models. This would undermine the analytical clarity that is associated with assuming humans are rational.
- Secondly, RCT is based on objective axioms of probability and statistics. When a person makes decisions that violate consistency of choice, probability and transitivity they become prone to systematic errors in judgement and choice. Kahnman shows that by describing a fictional female character as passionate for ‘social justice’ and ‘outspoken’ subjects were more likely to rate her as a feminist bank-teller than a bank teller. But, feminist bank tellers are a sub-set of bank-tellers more generally; and so there is a higher probability that Linda is a bank-teller. Failing to make decisions in line with RCT can lead to objectively incorrect answers.
- Thirdly, the study of economics is concerned with aggregates. Ross Don (2013) argues that there is little difference between assuming humans are rational versus assuming humans are boundedly rational when conducting macroeconomic analysis. This is because bounded rationality, in the context of markets, is often determined by institutional and informational factors that are specific to the market in question. This varies from market to market. Consequently, there is no general theory of bounded rationality that can explain how people behave across all markets.
- Fourthly, it is possible to ‘nudge’ individuals towards making better decisions in line with RCT. Richard Thaler and Cass Sunstein (2008) refer to this policy as ‘libertarian paternalism.’ They argue that people’s decision making can be influenced by ‘choice architecture.’ In other words, policy makers can create conditions that will encourage people to make better decisions. One simple example of this is making healthy food more accessible in supermarkets or introducing plain packaging for cigarettes. By manipulating cues in the environment, policy makers can therefore guide a person’s decision making to bring it more in line with RCT. On this basis, RCT remains a useful normative standard for policymakers.
No single theory of human behaviour accurately describes our decisions in every set of circumstances. Each theory brings with it a unique set of insights into human behaviour. Each theory brings with it a corresponding set of criticisms. It is not inconceivable that these theories will one day be reconciled and combined into a coherent general theory of human behaviour. Until then, behavioural economics serves an important descriptive function. Importantly, bounded rationality can only be meaningfully understood relative to unbounded rationality (RCT). After all, bounded rationality purportedly demonstrates the fact that humans do not always act rationally. This presupposes that rationality is – and should remain – the normative ideal of human decision making.