Household heterogeneity and the government spending multiplier
The Global Crisis of 2007-2008, during which monetary policy transmission was severely impaired, gave rise to a renewed interest in the long-neglected topic of the stabilising role of fiscal policy. Stimulus generated by a rise in government expenditures was one of several fiscal measures analysed by economists in that context. The standard textbook logic describing the impact of this policy is commonly known and relies on the feedback loop between changes in household disposable income and the resulting responses of private consumption. The power of the associated propagation mechanism, measured by the value of the government spending multiplier, was subject to a heated debate during the last decade.
As pointed out by Woodford (2011), much public discussion of this issue had been based on old-fashioned models abstracting from the dynamic aspects of macroeconomic processes, ignoring the sources of stimulus financing, where the only channel through which private consumption amplifies fiscal shocks is the one that works through household income. To address those shortcomings, Woodford used the standard New Keynesian model and derived an analytical expression for the government expenditure multiplier that parsimoniously explains the key determinants of its size. He concluded that the multiplier’s magnitude depends crucially on the monetary policy response because the most important propagation channel of fiscal shocks works through the intertemporal substitution of private consumption. The intuition underlying that mechanism is straightforward. As the central bank tightens monetary policy in response to higher inflation and a higher output gap (resulting from an increase in government purchases), then, under sticky prices, real interest rates rise and induce households to cut current spending. Consequently, government expenditures crowd out private consumption.
Intuitively, the propagation mechanism that works through intertemporal substitution implies that government purchases should have a particularly strong impact on GDP when the central bank’s policy rate is constrained by the zero lower bound. This is because, to the extent that higher fiscal purchases are associated with a rise in inflation expectations, real rates tend to drop due to government stimulus, which induces a drop in savings and a rise in households’ spending. This theoretical prediction based on the model with identical consumers (discussed in Woodford 2011, Eggertsson 2011 and Christiano et al. 2011, among others) is challenged on empirical grounds by Ramey and Zubairy (2015, 2018) who do not find convincing evidence of substantially larger multipliers during periods at the zero lower bound or constant interest rates.
Fiscal multipliers and household inequality
In parallel with the new wave of fiscal policy research, economists developed a voluminous empirical literature documenting the features of individual consumption behaviour, measuring the marginal propensity to consume and showing the heterogeneity of spending across various types of consumers. To give an idea about those patterns, Figure 1 displays the marginal propensity to consume values reported by Italian households in the 2016 Survey on Household Income and Wealth that are plotted against four different household-level categories: consumption, nominal assets, unhedged interest rate exposure and income.1
Given the observed heterogeneity of households, several important questions arise. What is the impact of households’ inequality on the propagation of fiscal spending shocks through private consumption? How different is it from the transmission mechanism in the standard textbook framework and in the New Keynesian model where consumers are assumed to be identical? What are the policy implications of those differences? To answer those questions, I use a version of the workhorse heterogeneous agent model to derive an analytical characterisation of the multiplier value in the spirit of Woodford (2011). I find that household inequality gives rise to a rich set of new channels that propagate government expenditures shocks through consumer spending, which are related to households’ balance sheets and monetary-fiscal interactions.2
Figure 1 Empirical cross-correlations in the 2016 Survey on Household Income and Wealth
Source: The 2016 Survey on Household Income and Wealth (conducted by Banca d’Italia).
Note: Values of income and consumption are standardised by their population means, values of unhedged interest rate exposure and nominal assets are standardised by mean consumption. Pairs are sorted in ascending order with respect to first coordinate and grouped in 50 bins.
In particular, in a recent paper (Kopiec 2019), I argue that household heterogeneity may have some far-reaching consequences for the effectiveness of higher government spending when the nominal interest rate is constrained by the zero lower bound. In contrast to predictions based on the model populated with identical consumers, the multiplier value in a liquidity trap may barely differ from the one in ‘normal times’ (i.e. when the central bank uses the standard Taylor rule to conduct monetary policy) when inequality is in place. The analytical characterisation I derive offers a clear explanation of this fact. First, it shows that the relative role of the intertemporal substitution channel is rather modest when households are heterogeneous. So, when it becomes shut off under the zero lower bound, its disappearance has only minor effects on the multiplier’s value.
There are the two other propagation mechanisms that are heavily affected by the liquidity trap, but they broadly offset each other. The first is related to the cost of servicing public debt – in ‘normal times’, nominal rates rise in response to stimulus, which makes it more costly for the government to issue new debt. This automatically leads to a larger budget deficit that has to be eventually funded with an increase in taxes which, in turn, reduces household consumption. As the cost of public debt issuance remains unaffected during fiscal expansion under the zero lower bound, this negative pressure on private spending disappears, which raises the multiplier’s value.
The second effect is related to household balance sheets – as already mentioned, in normal times fiscal expansion is accompanied by higher nominal interest rates, which has some important redistributive consequences. On the one hand, it increases the cost of taking loans by households with a relatively high willingness to spend (which corresponds to the negative value of unhedged interest rate exposure) and therefore forces them to reduce consumption. On the other hand, households with positive savings (or, equivalently, with positive unhedged interest rate exposure) benefit from higher interest earnings, which boosts their consumption. As the bottom left panel in Figure 1 shows, the distribution of households across the values of unhedged interest rate exposure is skewed towards positive values and the average marginal propensity to consume is similar in both sub-groups. Therefore, the second effect that works through households’ balance sheets imposes an upward pressure on the multiplier’s value in normal times, which is eliminated under the zero lower bound scenario.
It turns out that both the first and the second effect are of opposite signs and have a similar effect, which means that their net impact on aggregate consumption spending is almost zero. This, coupled with the abovementioned low importance of intertemporal substitution, explains why the multiplier’s value in normal times and in a liquidity trap barely differs when households are unequal, in contrast to conventional wisdom based on the representative agent framework.
Auclert, A (2019), “Monetary policy and the redistribution channel”, American Economic Review 109(6): 2333-2367.
Auclert, A, M Rognlie and L Straub (2018), “The intertemporal Keynesian cross”, NBER working paper 25020.
Brinca, P, M Faria-e-Castro, M Ferreira and H Holter (2019), “The nonlinear effects of fiscal policy”, Federal Bank of St. Louis working paper 2019-015A.
Brinca, P, H Holter, P Krusell and L Malafry (2016), “Fiscal multipliers in the 21st century”, Journal of Monetary Economics 77(C): 53-69.
Challe, E and X Ragot (2011), “Fiscal policy in a tractable liquidity constrained economy”, Economic Journal 121(551): 273-317.
Christiano, L, M Eichenbaum and S Rebelo (2011), “When is the government spending multiplier large?”, Journal of Political Economy 119(1): 78 -121.
Eggertsson, G (2011), “What fiscal policy is effective at zero interest rates?”, NBER Macroeconomics Annual 25: 59-112.
Ferriere, A and G Navarro (2018), “The heterogeneous effects of government spending: It’s all about taxes”, working paper.
Hagedorn, M, I Manovskii and K Mitman (2019), “The fiscal multiplier”, NBER working paper 25571.
Kopiec, P (2019), “Household heterogeneity and the value of government spending multiplier”, National Bank of Poland working paper 321.
Ramey, V A and S Zubairy (2015), “Government spending multipliers in good times and in bad: Evidence from US historical data”, VoxEU.org, 23 January.
Ramey, V A and S Zubairy (2018), “Government spending multipliers in good times and in bad: Evidence from US historical data”, Journal of Political Economy 126(2): 850-901.
Ramey, V A (2019), “Ten years after the financial crisis: What have we learned from the renaissance in fiscal research?”, Journal of Economic Perspectives 33(2): 89-114.
Woodford, M (2011), “Simple analytics of the government expenditure multiplier”, American Economic Journal: Macroeconomics 3(1): 1-35.
 Unhedged interest rate exposure is defined by Auclert (2019) as: “the difference between all maturing assets and liabilities at a point in time” and measures households’ balance sheet exposure to interest rate changes.
 The analytical decomposition exercise conducted in my work is similar to the one presented in Auclert (2019) in his seminal paper on the redistribution channel of monetary policy. In a related work, Auclert et al. (2018) derive an elegant multiplier’s formula in the model populated by unequal consumers under pegged real interest rates and highlight the importance of the disposable income channel for the propagation of higher government spending, which echoes the conclusions from the standard textbook framework.