(Econometrica, July 2023)
This paper revisits the New Keynesian framework, theoretically and quantitatively, in an economy with multiple sectors and input‐output linkages. Analytical expressions for the Phillips curve and welfare, derived as a function of primitives, show that the slope of all sectoral and aggregate Phillips curves is decreasing in intermediate input shares, while productivity fluctuations endogenously generate an inflation‐output tradeoff—except when inflation is measured according to the novel divine coincidence index. Consistent with the theory, the divine coincidence index provides a better fit in Phillips curve regressions than consumer prices. Monetary policy can no longer achieve the first‐best, resulting in a welfare loss of 2.9% of per‐period GDP under the constrained‐optimal policy, which increases to 3.8% when targeting consumer inflation. The constrained‐optimal policy must tolerate relative price distortions across firms and sectors in order to stabilize the output gap, and it can be implemented via a Taylor rule that targets the divine coincidence index.
(Annual Review of Economics, September 2023)
This paper reviews a framework for studying the aggregation and propagation of microeconomic shocks in general equilibrium. We discuss the determinants of aggregate measures of real economic activity, like real GDP, real domestic absorption, and aggregate productivity in both efficient and inefficient environments. We also discuss how shocks from one set of producers are transmitted to other producers through prices and quantities. The framework we provide is amenable to generalization and can be used to study any collection of producers ranging from one isolated producer, to an industry consisting of heterogeneous producers, to an entire economy. We conclude with a brief survey of some of the applied questions that can be addressed using the analytical tools presented in this review and avenues for future work.
This paper derives the cross sectional effects of monetary policy on employment and consumption in a model economy where heterogeneous agents earn and spend their income in different industries. Agents are characterized by their labor type, consumption preferences, wage rigidity and labor supply elasticity. Industries hire different bundles of labor types and capital assets, and face heterogeneous price rigidity and demand elasticity. In response to a monetary expansion, goods that have stickier prices or use supply-elastic factors (or whose inputs have sticky prices or use supply-elastic factors...) become cheaper, thereby raising the relative demand and employment of the workers who produce them. In the aggregate, the ability of producers and consumers to substitute towards sticky-price or supply-elastic goods increases monetary non-neutrality. Calibrating the model to the US economy, and interpreting labor types as different occupations, reveals significant heterogeneity in the impact response of employment to monetary policy, which varies from 0.25 (food services) to 1.1 (construction) for a 1increase in nominal GDP. Ignoring input-output linkages would reduce the cross-sectional range from 0.86 to 0.35.
The recent Covid pandemic impacted demand and supply in widely different ways across industries, while disrupting supply chains and labor markets. At the same time, governments implemented unprecedented demand stimulus packages. How much of the recent increase in inflation can be attributed to cross-sectional demand and supply disruptions versus aggregate stimulus policies? Building on a model with multiple industries and primary factors, I show that declines in relative productivity or increases in relative demand for specific sectors – in particular those which use inelastically supplied factors – worsen the inflation-output tradeoff, beyond the control of monetary policy. Moreover, these cross-sectional shocks induce different changes in relative prices compared to aggregate demand. Levering up on this observation, I show how to filter out the effect of crosssectional shocks and recover the component of inflation driven by aggregate demand. Applying my decomposition to US consumer price inflation between 2020 and 2022, I find that the rapid deflation and subsequent inflation of 2020 are almost entirely driven by supply factors, while around three quarters of the increase in the CPI since 2021 is driven by aggregate demand.
I propose a novel partial equilibrium theory of the fragmentation of production across industries, in an environment where firms have variable input requirements over time, while workers have different comparative advantage over inputs and cannot be freely reallocated across firms. Fragmenting production allows firms to allocate workers to tasks at all times according to their comparative advantage, but it requires costly trading of inputs between firms. I show that, in general equilibrium, there is a two-way relationship between technical change, fragmentation, and the wage distribution. On the one hand, reducing the cost of trading inputs between firms implies a hollowing out of the wage distribution. On the other hand, a rise in the skill premium strengthens the incentives to fragment production. The model shows that skill biased technical change not only affects the wage distribution, but also the organization of production. Moreover, the theory suggests that outsourcing – which is a form of fragmentation – could be the consequence as well as the cause of a higher skill premium.
Aggregation and Redistribution in General Equilibrium
In a disaggregated economy with multiple industries, primary factors, and final users, I show how to map micro-level demand, supply, and price elasticities into macro-level elasticities that govern the evolution of factor prices and quantities. I first characterize macro demand and supply elasticities, which describe the slopes of a factor supply equation and a relative factor demand equation. I then derive general equilibrium elasticities which govern the response of endogenous variables to exogenous shocks. For each elasticity, I isolate four components corresponding to all combinations of aggregate and cross-sectional shocks and outcomes. I find that cross-sectional shocks can have aggregate effects in general equilibrium – and vice versa – whenever primary factors have heterogeneous supply elasticities or factor prices have heterogeneous passthrough into aggregate prices. I apply the framework to illustrate the aggregate and redistributive effects of several demand and supply shocks, such as monetary policy, fiscal policy, biased productivity changes, and changes in trade costs.
Aggregate and Cross-Sectional Spending Multipliers
Traditional theoretical and empirical studies focus on the aggregate spending multiplier, which relates total government spending with aggregate output. I argue that the sectoral composition and financing scheme of government spending have important cross-sectional and aggregate implications. Using a disaggregated New Keynesian model with multiple heterogeneous industries and households, I provide necessary and sufficient conditions for government spending to redistribute employment and income across households – namely, industries should use primary factors (workers and capital assets) in different proportions, and the composition of spending must be different from private demand. I then show that, whenever these conditions hold, the spending multiplier is different than in an economy with a representative household. Specifically, the output multiplier is larger whenever government spending increases the demand for “slack” industries, that is, industries which rely on elastically supplied primary factors, or face more nominal rigidities. From a quantitative point of view, I find vastly heterogeneous effect on employment across occupations given the current composition of government spending in the US (ranging from 0.3% to 2% for a 1% increase in spending). The aggregate employment multiplier also varies widely across industries, ranging from 0.4% for oil and gas extraction to 1.2% for professional services.
Cross-Sectional Identification with Asymmetric Units
When estimating the relation between macroeconomic aggregates, identification is often confounded by the policy response to changes in these aggregates. Estimating the same relations at the local level avoids the issue, because macroeconomic policies do not respond to local conditions. I show that the traditional version of this argument is valid only if all local variables respond to aggregates with equal elasticities – an assumption which cross-sectional studies always maintain, and which is violated in practice. This assumption guarantees that all the variation comes from local shocks, and it delivers a simple mapping between cross-sectional and aggregate elasticities. If the assumption is violated, part of the cross-sectional variation is driven by aggregate shocks. Relative local responses to these shocks are informative of cross-sectional variation in local parameters, rather than of the aggregate parameters of interest. I propose an alternative method to exploit cross-sectional variation in response to an identified shock, combined with a calibrated structural model, to filter out the endogenous response of macroeconomic policies and obtain unbiased estimates of some model parameters, which in turn can be mapped into the aggregate elasticities of interest using the model.
Monetary Policy in a Globalized Economy
I consider a global production network, where every country has multiple sectors that
trade in intermediate inputs both domestically and internationally. Producers in each countrysector
pair can price different shares of their output in different currencies, also conditioning
on the destination. I derive the response of local inflation and output to local and
foreign shocks to sector-level productivity, monetary policy and exchange rates. I plan on
using this framework to revisit traditional questions, such as the effect of competitive devaluations
and monetary policy spillovers, and to evaluate the extent to which an increase
in import shares (both in production and final consumption) can reduce the sensitivity of
inflation to local employment.