One of the main criticisms levelled against core macroeconomic models is their reliance on a single, representative, and usually rational, agent. Fortunately, this is an increasingly inaccurate description of how models are actually used in the real world, due in no small part to a series of papers developing ways of adding realism to the agents that populate models.
Monetary policy according to Hank, by economics professors Greg Kaplan (University of Chicago), Benjamin Moll (Princeton University) and Giovanni Luca Violante (Princeton University), is not the first piece of research to introduce heterogeneous agents into a New Keynesian framework. However, the authors adopt an innovative approach that generates crucial insights into monetary policy, viewing the problem through the lens of inequality. The paper was published in a top-ranked journal, the American Economic Review, in 2018.
Representative agent New Keynesian (Rank) models produce a consumption response to monetary policy through a direct intertemporal substitution effect. A rate hike makes it costlier to bring forward future consumption by borrowing, while increasing the return on saving, and this direct effect drives almost all of the impact of policy. The alternative, an indirect effect through a change in disposable income, often from labour, hardly features.
But even to the casual observer, this reasoning may seem suspect: large segments of the population live pay cheque to pay cheque. Those with no assets can’t vary their savings in response to interest rate changes, and those with their wealth tied up in a house may be in a similar situation. The predominance of 30-year mortgages in the US desensitises many debtors to the forces of intertemporal substitution.
Sure enough, an array of empirical studies has failed to find a central role for intertemporal substitution. “There’s a huge amount of micro and macro evidence that casts doubt on that as a plausible mechanism for how monetary policy affects aggregate demand,” Kaplan tells Central Banking. He notes that central banks themselves tend to tell a story with more of an “old Keynesian flavour” – lower rates encourage firms to borrow, firms produce more goods and take on more labour, incomes rise, and that stimulates aggregate demand.
Living hand to mouth
It is this inconsistency – essential to the question of setting monetary policy – that Kaplan, Moll and Violante target. Their heterogeneous agent New Keynesian (Hank) model is built around a more nuanced view of household consumption than the Rank model. They extend a standard incomplete markets model by allowing households to hold two assets: a low-return liquid asset; and a high-return illiquid asset that is subject to transaction costs.
“This extended model has the ability to be consistent with the joint distribution of earnings, liquid wealth and illiquid wealth, as well as with the sizeable aggregate [marginal propensity to consume] out of small windfalls,” say the authors.
The paper makes an important contribution … by introducing a model to analyse monetary policy rigorously calibrated using microdata [and] highlighting the relevance of indirect effects of interest rate changes on consumption
Galo Nuño, Bank of Spain
Kaplan estimates around 10% of the US population have no assets to speak of beyond their monthly pay cheque, and therefore lives hand to mouth, while a further 30% or so have assets that are illiquid, such as a house, but little other wealth. These “wealthy hand-to-mouth” consumers never have more than one pay cheque’s worth of liquid assets in the bank, like their poorer counterparts, and behave accordingly. Hank captures this reality.
Other aspects of the model follow the “New Keynesian tradition”, as the authors put it – prices are set by monopolistically competitive producers that face nominal rigidities. The central bank follows a Taylor rule. Agents are rational.
A crucial result emerges: the direct intertemporal substitution effect is small, but the indirect effects through changes to disposable income “can be substantial”. The presence of uninsurable risk, combined with the coexistence of low- and high-yielding assets, produces a “sizeable fraction of poor and wealthy hand-to-mouth households, as in the data”. Such households respond strongly to changes in their labour income, but not to interest rates.
“The paper makes an important contribution to the literature, not only by introducing a model to analyse monetary policy rigorously calibrated using microdata, but also by highlighting the relevance of indirect effects of interest rate changes on consumption,” says Galo Nuño, head of the monetary policy unit at the Bank of Spain.
“I think the paper makes an important contribution,” adds Jonathan Heathcote, monetary adviser at the Federal Reserve Bank of Minneapolis. “It has spurred a large and growing industry of related work. It challenged our basic understanding of how monetary policy works.”
Capital in the 21st century
The two-asset approach is a key difference between Monetary policy according to Hank and similar literature. Until Kaplan, Moll and Violante presented their argument, models had focused only on liquid assets, ignoring as much as 95% of the wealth in the US economy as a result, according to the authors.
The focus on inequality is one of the Hank’s great strengths. Inequalities of wealth and income are at the forefront of political debate, and the insight the paper provides could help rehabilitate economic models, which are often seen as divorced from real-world problems. The fact that central bank actions redistribute income and wealth matters to people, and perhaps it should feature more heavily in central bank decision-making.
“Whether you like it or not, the inconvenient truth is central bank actions have distributional consequences,” says Kaplan. He adds that central banks are well aware of this – they are berated by disgruntled retirees every time they cut rates and by young debtors every time they raise rates. But for a long time, they have been able to hide behind models that “separate the question of the size of the pie from the distribution of the pie”.
“I think what our work is showing and what is becoming more accepted is that you really can’t hide from the result – those distributional consequences are there,” says Kaplan.
The growing awareness of extreme inequality in the US, Europe and elsewhere is putting pressure on central banks to adapt their way of thinking and acting. Many institutions are coming under pressure from populist politicians, some of whom have come to power on a wave of discontent, driven in part by inequality.
Kaplan, Moll and Violante show that it is possible to face up to this inconvenient truth and arrive at a more realistic way of modelling monetary policy at the same time. As Kaplan acknowledges, the model has room for extension – adding a banking sector would be a welcome development, and several researchers are working on it already. In the meantime, the authors have produced an important result that could help central banks safeguard their position at the heart of the modern economy.
The Central Banking Awards were written by Christopher Jeffery, Daniel Hinge, Dan Hardie, Rachael King, Victor Mendez-Barreira, Joel Clark, William Towning and Tristan Carlyle