FISCAL STIMULUS WITH LEARNING‐BY‐DOING

• Author: Giulio Fella,Antonello D'Alessandro,Leonardo Melosi
• DOI: http://doi.org/10.1111/iere.12391
• Date: 01 August 2019
• Published date: 01 August 2019

INTERNATIONAL ECONOMIC REVIEW

Vol. 60, No. 3, August 2019 DOI: 10.1111/iere.12391

FISCAL STIMULUS WITH LEARNING-BY-DOING∗

BYANTONELLO D’ALESSANDRO,GIULIO FELLA,AND LEONARDO MELOSI1

Bank of Italy, Italy; Queen Mary University of London, U.K., CFM, U.K.,and IFS, U.K.;

Federal Reserve Bank of Chicago,U.S.A.

Using a Bayesian structural vector autoregression analysis, we document that an increase in government

purchases raises private consumption, the real wage, and total factor productivity (TFP) while reducing infla-

tion. These three facts are hard to reconcile with both neoclassical and New Keynesian models. We extend

a standard New Keynesian model to allow for skill accumulation through past work experience. An increase

in government spending increases hours and induces skill accumulation and higher measured TFP and real wages

in subsequent periods. Future marginal costs fall lowering expected inflation and, through the monetary policy

rule, the real interest rate. Consumption increases as a result.

1. INTRODUCTION

What are the effects of changes in government spending on aggregate output? What is the

mechanism by which they propagate? Though the limitations of monetary policy in the face of

the financial crisis of 2008 have sparked a renewed interest in the role of government spend-

ing, there is a general lack of consensus on the answer. This article provides an empirical and

theoretical analysis of the effects of government spending shocks on consumption, total factor

productivity (TFP), the real wage, and inflation. Our empirical analysis estimates a Bayesian

structural vector autoregression (SVAR) model on U.S. time series for the period 1954–2007 and

identifies government spending shocks using two distinct approaches, following, respectively,

Blanchard and Perotti (2002) and Auerbach and Gorodnichenko’s (2012) implementation of

Ramey (2011). In line with previous empirical studies using similar methodologies and identi-

fication strategies, we find that a positive government spending shock increases consumption,

TFP, and the real wage and reduces inflation.2

With a few exceptions discussed below, the observed responses of private consumption and

the real wage to innovations in government spending are hard to reconcile with the predictions

of existing models with intertemporally optimizing consumers. In these models, the higher

present value of taxes to finance the increase in government spending generates a negative

wealth effect that induces a fall in consumption and leisure and a decline in the real wage. This

mechanism, which is at the heart of the real business cycle (RBC) model (Baxter and King,

1993), is powerful enough for the result to extend to New Keynesian models with optimizing

consumers. Furthermore, most theoretical models imply that inflation increases, instead of falls,

in response to a fiscal expansion.

This article proposes a novel explanation that can account for the estimated joint response of

consumption, TFP, real wages, and inflation to a government spending shock. The key channel

is the interaction of skill accumulation through past work experience—the “learning-by-doing”

(LBD) mechanism, originally proposed by Chang et al. (2002) in an RBC framework—with

∗Manuscript received March 2017; revised November 2018.

1We thank Evi Pappa, the editor (Jes´

us Fern´

andez-Villaverde), and two anonymous referees for useful comments.

The views expressed in the article are those of the authors and do not necessarily reflect those of the institutions they

are affiliated with. Please address correspondence to: Giulio Fella, School of Economics and Finance, Queen Mary

University of London, Mile End Road, London E1 4NS, U.K. Phone: +44 20 7882 8823. E-mail: g.fella@qmul.ac.uk

2Section 2 discusses the existing empirical literature.

1413

C

(2019) by the Economics Department of the University of Pennsylvania and the Osaka University Institute of Social

and Economic Research Association

1414 D’ALESSANDRO,FELLA,AND MELOSI

wage and price rigidities. A positive shock to government spending increases hours and, through

the LBD mechanism, induces skill accumulation and higher measured TFP and real wages in

subsequent periods. For a large set of values of the degrees of wage and price stickiness, the

increase in future productivity lowers future marginal costs and therefore the expected rate of

inflation. Through the monetary policy rule, the fall in expected inflation results in a persistent

reduction in the policy rate and the short-term real interest rate. The associated decline in the

current long-term real interest rate increases consumption.

It is worth noting that the presence of nominal rigidities is crucial to generate the co-movement

of consumption and hours in response to a shock to government spending. The wedge (equal

to the sum of the price and wage markups) between the consumption/leisure marginal rate of

substitution and the marginal product of labor is constant in the absence of nominal rigidities.

Since an increase in hours worked reduces the marginal product of labor but increases the

marginal rate of substitution, equilibrium requires consumption to fall in the absence of nominal

rigidities.3Vice versa, the combination of LBD and nominal rigidities implies that an increase

in hours is accompanied by a fall in the wedge that can be large enough to break the negative

association between consumption and hours worked.

Though it crucially relies on nominal rigidities, our mechanism is very different from what

Dupor and Li (2015) call the “expected-inflation channel,” which is common to all New

Keynesian models. According to this mechanism, an increase in government spending results

in higher expected inflation and a lower real interest, and therefore higher consumption, if

the central bank is nonresponsive to inflation; that is, if it raises the policy rate by less than

the increase in expected inflation (Christiano et al., 2011; Eggertsson, 2011; Woodford, 2011).

Instead the transmission channel in our model works through a fall in expected inflation to

which the central bank responds by reducing the policy rate and, by the Taylor principle, the

real rate. In our framework, the consumption and output response is larger the more, instead

of less, responsive the central bank is to inflation. This prediction is in line with the findings of

Dupor and Li (2015) in the context of a structural VAR that (i) inflation responds negatively

to innovations in government spending and (ii) the response of inflation is more negative, and

that of consumption positive and larger, in periods in which the central bank responds more

aggressively to inflation. Our mechanism is also consistent with the empirical evidence in Hall

and Thapar (2018) and Jørgensen and Ravn (2018) and discussed in Subsection 2.2.

Our contribution is related to a relatively small number of papers that also identify mech-

anisms that can generate a positive response of consumption to government spending shocks.

Devereux et al. (1996) find that government spending endogenously increases TFP, and pos-

sibly consumption and the real wage, in a real business cycle model with increasing return to

specialization. Ravn et al. (2006) show that a real business cycle model in which consumers

form consumption habits on a good-by-good basis (deep-habit) implies a positive response of

consumption and the real wage by generating a fall in price markups. Ercolani and Pavoni

(2019) show that government expenditure on health care reduces precautionary saving and

increases consumption by providing insurance against health expenditure shocks. A number

of papers obtain a positive consumption response within a New Keynesian framework. Gal´

ı

et al. (2007) find that a government expenditure shock increases aggregate consumption in an

economy with a sufficiently large share of hand-to-mouth (nonoptimizing) consumers and a

sufficiently pro-cyclical, noncompetitive, real wage. Corsetti et al. (2012) show that consump-

tion responds positively to a government expenditure shock if the latter is partly financed by

future spending reductions. Monacelli and Perotti (2008) find that, if preferences display a suf-

ficiently small wealth effect on labor supply, consumption and leisure are complements, and

price rigidity implies a positive response of the real wage and consumption to a government

expenditure shock. Gnocchi et al. (2016) show how allowing for home production implies a

3This was first noted by Barro and King (1984). Bilbiie (2009) shows that, given a constant wedge, a positive co-

movement between consumption and hours in response to a government spending shock requires either consumption

or leisure to be inferior goods.