Do Microfoundations address the Lucas Critique?

Modern macroeconomics has been endlessly criticized, not only by critics of mainstream economics, but also by many prominent mainstream economists. For example, nobel price winners such as Paul Romer, Paul Krugman, and Joseph Stiglitz have written articles in which they disparage modern macroeconomists. Even economists who have been very involved with such models such as Greg Mankiw have argued that they have had limited policy influence. Despite the criticism these models have received, what are known as DSGE macroeconomic models basically are mainstream macroeconomics at the present time: few other approaches are given much attention.

While many argue that DSGE models have had limited impact on policy, likely due to the fact they are not intuitive, such models are used by central banks and many other economic institutions worldwide. Even if the policy prescriptions of such models are not listened to, such models have an impact, if only through preventing better models from being used for policy analysis.

The move towards modern DSGE macro models started in the 1970s. One of the main reasons that is usually given for its adoption is something called the Lucas Critique. Prior to the 1970s macroeconomists were using models known as structural Keynesian models. Structural Keynesian models would typically include assumptions about aggregate behaviour, for example, they might include the assumption that people spend a constant portion of their income each year. Alternatively they might include an assumed behavioural equation that related investment to GDP, interest rates, tax rates and inflation. Structural Keynesian models were often used to attempt to understand what policies the government should follow, and to predict economic behaviour.

Robert Lucas criticized Structural Keynesian models by arguing that the behavioral equations were not constant, and would change as government policy changed. Consumers and investors would change how they acted in response to government policy, so the effect of predicted policy changes could not be predicted by models that assumed unchanging behavioral relationships. For example, people might spend a higher portion of their income as government policy changed, anticipating inflation. Even if the relationship between consumption and income was found to be empirically supported in the past it would change as the government changed its policy. This would mean that the predictions of economic models would not hold, and their policy recommendations would be less than ideal in practice.

The Lucas critique was partially responsible for initiating a paradigm shift in macroeconomics. Economic models from that point on needed to be microfounded, or based on parameters that wouldn’t vary as policy changed. For example, instead of assuming that consumers consumed a certain portion of their income, the new models assumed certain things about consumer preferences, and then derived their consumption decisions from those “deeper” parameters. Instead of assuming that consumers spend a certain portion of their income, it was assumed that consumers tried to maximize their total well-being over time. The idea was that deeper parameters would not change when policy changed, and thus the predictions of the models would be more robust to changes in policy. The Lucas critique was not the only reason for the paradigm shift in macroeconomics, but played an important role.

---

Most DSGE models exhibit a property known as Ricardian Equivalence. Ricardian Equivalence is the idea that it doesn’t matter whether a government funds its expenditures through taxation or through debt. The reason for this is that if governments fund a spending increase through debt, consumers will just save the money in anticipation of future tax increases. Because governments will eventually pay down the debt, households will save the money they recieve as a result of debt financed government spending in order to pay the future taxes.

A non-economist learning about Ricardian equivalence would likely think that the idea is absurd. In everyday life it makes no sense to save to pay future tax increases for several reasons. For one, it seems that governments rarely pay back their debts. But even if we expected the government to eventually pay back its debt, it would still not necessarily make sense to save in anticipation of future taxes, because of the manner in which taxes are imposed. It might make sense to save in order to pay future taxes if the government said that everyone owes it 10,000 dollars, or else they get thrown in jail. But in reality government taxes a percentage of income, and sometimes of wealth. The percentage usually increases as wealth or income increases. Therefore, saving in order to pay future taxes does not make sense. We can simply have no income at the time the government raises taxes and then we will owe the government nothing.

The technical way of making the above point is to say that Ricardian Equivalence requires what are known as lump sum taxes. Lump sum taxes are taxes that the government simply imposes on everyone, by saying, for example, that everyone must pay the government 10,000 dollars or go to jail. Without lump sum taxes Ricardian Equivalence does not hold in mainstream macroeconomic models. The vast majority of modern macroeconomic models assume that taxes are lump sum.

So despite being based on so called deep parameters, that are supposedly invariant to policy changes, the vast majority of mainstream macroeconomic models represent people as behaving in ways that they wouldn’t behave under current policy arrangements. The possibility that agents behaviour might change has been replaced with the certainty that models do not describe the behaviour of people under current policy arrangements accurately. Despite this,economists assume that, by using microfoundations, the Lucas Critique has been adequately addressed.

---

I do not think economists should ignore the Lucas critique entirely and go back to 1970s structural Keynesian models. Those models had other issues. But I do think that the focus on microfoundations is ill advised, especially since microfoundations clearly cannot address the problems they were supposed to solve. Rather, I think that behaviour assumptions that are not microfounded should be much more common in macroeconomics. Moving away from microfoundations would allow economists to analyse much more complicated phenomena with simpler models. For example, Stock Flow Consistent models are likely much more useful than DSGE models when it comes to addressing practical economic questions.

The Lucas critique could still be addressed by separate papers analysing in which situations policy changes are likely to change consumer behaviour. The realism of behaviour assumptions about consumption could be studied empirically and through microeconomic analysis of how likely consumer behaviour is to change when policy changes. The insights gained from the empirical analysis could then be incorporated into macroeconomic models.

---

Further reading

Fore those interested in reading more technical material on how Ricardian Equivalence is violated under current tax arrangements should read this paper. Note that this paper only finds that Ricardian Equivalence is violated under progressive taxes, but it should also be violated under income taxes. However I believe analysing that case requires abandoning the representative agent paradigm. If anyone knows of papers looking at that case I would be interested in reading them.