top of page

Bringing Psychology & Social Sciences into Macroeconomics: Summary

Rebuilding Macroeconomics held its first annual conference at HM Treasury, bringing together perspectives from anthropology, maths, neuroscience, psychology, sociology, physics, climate science, game theory and, of course, macroeconomics. Our aim is to see what insights we can learn by looking at macroeconomics through other scholarly lenses. We are enormously grateful to the speakers, discussants and chairs who took part. All of the presentations can be found here.


While the presentations covered a wide range of issues, a deep theme in many sessions could be distilled down to radical uncertainty and our response. This is the world that citizens, businesses and policy makers inhabit. The richness of the sessions was the insights from other disciplines about what this means for our interactions, knowledge, decisions and ultimately actions and the economy. Social settings matter and our preferences are not stable, but in which direction might this lead us?


What is considered rational can vary depending on the social environment we are in. For example, Webb Keane argued social norms are a crucial ingredient for individual behaviour, within institutions and social dynamics. Ekaterina Svetlova discussed how social settings can affect our expectations such as increasing individual optimism or pessimism when we observe this from other people. This reflexivity in how we think others will behave could affect our behaviour, making us less than fully-rational.


Economics has become reliant on the concept of “rational expectations” for talking about the future. This suggests there are such things as probability assessments for every possible outcome and agents are aware of these. Yet we know from simple psychology experiments there are a range of biases in decision making, as Sam Johnson showed.


Nava Ashraf pointed to research that suggests we might update our beliefs occasionally, but are more likely to do so if our prior assumptions deem the new information useful. Humans are bad at working with probabilities, and modelling expectations to the contrary might just ignore behaviours that are crucial for understanding economic dynamics.


One issue with how expectations are modelled is that they can be affected by recent (independent) past observations. David Laibson’s work on intuitive forecasting shows that the extent to which people display rational expectations can depend on how quickly the “underlying model” mean-reverts. If the true model takes a long time to return to the mean then it becomes incredibly difficult for participants to detect any of the underlying model’s properties. People would be likely to over-react to volatility in the short-term.


An alternative suggestion for dealing with unquantifiable uncertainty was the concept of ecological rationality by Henry Brighton. Here, rationality depends on the circumstances in which a decision was taken. Heuristics are used, which are qualitative, or ad-hoc, statements about how people make decisions under uncertainty. Techniques like this are sometimes used in agent based modelling (ABM), by assigning a behavioural rule to an agent. In the face of uncertainty, ecological rationality and simple heuristics may provide a useful model of decision making.


Cars Hommes mentioned genetic algorithms, which mimic natural selection and have been applied to ABMs with heuristics. They have been shown capable of accurately replicating observed participant behaviours in experiments. These models are useful in describing seemingly irrational behaviours such as reacting differently towards positive versus negative feedback in experiments. Also, they can work on the macro (aggregate) level as well as on the micro (individual) level.


Sharon Alvarez spoke about how heuristics often matter for entrepreneurship, and is largely made off gut-feelings rather than precisely defined success or failure probabilities. Entrepreneurs are likely to have a meaningful role in the productivity shocks we see in economic data. But these innovations are cloaked in radical uncertainty, making them difficult to quantify and understand.


Social factors are also important in the production process through family firms, as Sylvia Yanagisako explained. Family firms tend to have unique behaviours, structures, objectives, and time horizons. The business dynamics of family firms is also quite different to other types of firm, from recruitment, promotions and working hours. Given that these firms contribute towards over half of US GDP, studying the choices they make could be informative in macro models when considering firm heterogeneity.


Whilst economics does not have an established and agreed way of handling radical uncertainty, there are some suggestions. Roman Frydman considered placing boundaries around model parameters, without specifying a probability distribution. This would allow risk to be captured in a model’s error terms, and uncertainty through differences in information sets. Such a framework might be able to capture aspects of market sentiment and other non-market fundamentals.


Douglas Homes considered narratives as another possible way of dealing with radical uncertainty. Monetary policy is often conducted with the goal of influencing people’s expectations – their narratives – in the economy with words. Developing a tractable framework for narratives in economic modelling could be informative for monetary policy, as well as understanding the decisions of policy makers.


Edmund Rolls showed how cognitive science can contribute towards understanding our beliefs formation. The challenge is that we do not observe beliefs and mental states very well. Noise in the brain – the random spiking of neurons – means our decisions have a chance aspect to them. Rather than being in the brain domain of ‘rationality’, they are randomly made in the rational or emotional parts of our brain.


How our emotional (often heuristic based) and rational decisions feedback into cognition might be of interest to understand policy makers, who often describe making decisions out of “instinct.” But we face a practical issue in mobilising the people and resources required for investigating these for use in macroeconomics, as Andrew Caplin discussed. The return on such investments might help us to better understand market crashes and extreme volatility.


Leonard Smith and Erica Thompson evaluated the usefulness of models. Economic models are typically linearized so they can be solved – yet linear systems have radically different behaviour to nonlinear ones, and point estimates in nonlinear systems are meaningless. This has serious implications for modelling in macroeconomics. Efforts in modelling nonlinear systems and ideas from chaos theory may have meaningful applications to economic problems. We should bear in mind Ricardo Reis’ challenge of 10 steps for persuading an economist.


Rosmarie Nagal’s presentation was on level-k rationality, which refers to the extent to which the decisions of others factor into someone’s decision making. MRI scans show differences in brain activity when people solve problems, with more activity in the brain associated with a higher order of this type of reasoning. Standard models suppose an infinite-level of rationality – a Nash Equilibrium – yet experiments show this is not the case. Incorporating heterogeneity in reasoning ability could expand traditional models to better explain observed data.


Rebuilding Macroeconomics aims to reconnect policy makers, academics and the public. We are particularly interested in interdisciplinary insights and introducing new methods to macroeconomics. The wealth of perspectives at the conference, the thoughtful discussion, positive feedback, and a growing network of people driving these ideas forward are all promising steps in the right direction.



72 views0 comments

Recent Posts

See All

Comments


bottom of page