Page 1 of 50

Demand Shocks as Productivity Shocks∗

Yan Bai

University of Rochester

Jos ́e-V ́ıctor R ́ıos-Rull

University of Minnesota, Federal Reserve Bank of Minneapolis, CAERP, CEPR, and NBER

Kjetil Storesletten

University of Oslo, Federal Reserve Bank of Minneapolis, and CEPR

∗We thank seminar attendants at the University of Pennsylvania, New York University, Bank of Canada, European

Central Bank, Federal Reserve Bank of Minneapolis, Federal Reserve Bank of New York, Institute for Fiscal Studies,

Toulouse University, and CREI-CEPR for useful comments. We are especially thankful to Gary Hansen, Ellen

McGrattan, and Larry Christiano. The views expressed herein are those of the authors and not necessarily those of

the Federal Reserve Bank of Minneapolis or the Federal Reserve System.

1

Page 2 of 50

Abstract

We provide a macroeconomic model where demand for goods has a productive role. A

search friction prevents perfect matching between potential customers and producers, and

larger demand induces more search, which in turn increases output in the economy. Con- sequently, when viewed through the lens of a standard neoclassical aggregate production

function, an increase in demand will appear as an increase in the Solow residual. We esti- mate the model using standard Bayesian techniques, allowing for business cycles being driven

by both preference shocks – which we interpret as shocks to demand – and true technology

shocks. Technology shocks account for less than 18% of the fluctuations in output and the

measured Solow residual. Our model also provides a novel theory for important macroeco- nomic variables such as the relative price of consumption and investment, Tobin’s Q, and

capacity utilization.

2

Page 3 of 50

1 Introduction

In the standard neoclassical model, output is a function of inputs such as labor and capital. There

is no explicit role for demand because (Walrasian) prices will adjust so that all produced goods

will be utilized. In reality, customers and producers must meet in order for the produced good to

be consumed, so value added depends on how well they are matched. As an example, consider

a restaurant. According to neoclassical theory, the output of a restaurant should be a function

of its employees, buildings, tables, and raw material, irrespective of market conditions. However,

the restaurant’s production takes place only when customers show up to buy meals. The more

customers who demand the restaurant’s meals, the more will be served and the larger the value

added will be. Thus, the demand for goods plays a direct role. The spirit of this example extends

to many forms of production: dentists need patients, car dealers need shoppers, all producers need

buyers.

This paper provides a theory where search for goods – which we sometimes refer to as demand

– has a productive role. The starting point is that potential customers search for producers, and

a standard matching friction prevents standard market clearing, in the neoclassical sense that all

potential productive capacity translates into actual value added. Clearly, for households and firms,

the acquisition of goods is an active process that involves costs not measured in the National

Income and Product Accounts (NIPA). Technically, we resolve the search friction by building on

the competitive search model Moen (1997). Firms post prices and customers trade off good prices

versus congestion when searching for the goods: prices are higher for goods that are easier to find.

Allowing such an explicit role for demand has direct implications for business cycle analysis,

especially for our understanding of the driving factors of business cycles. A striking consequence

of the type of demand-driven business cycle model we propose is that changes in demand will

increase output even if inputs, and the intensity with which they are used, remain constant. If

viewed through the lens of a standard neoclassical aggregate production function that ignores

demand, an increase in demand would imply an increase in total factor productivity (TFP). Thus,

a preference shock that boosts consumption will also show up as a shock to the Solow residual.

This mechanism reverses the direction of causation relative to the neoclassical model: there, a true

technology shock would increase TFP which, in turn, would increase consumption and investment.

Interestingly, aggregate data for the United States suggest that it is factors influencing consumption

that drives TFP, rather than TFP driving consumption. In particular, aggregate consumption turns

3