What Have We Learned about the Effectiveness of Infrastructure Investment as a Fiscal Stimulus? A Literature Review

Public Disclosure Authorized

Public Disclosure Authorized

Policy Research Working Paper

9796

What Have We Learned about the Effectiveness of Infrastructure Investment as a Fiscal Stimulus?

A Literature Review

Maria Vagliasindi

Nisan Gorgulu

Infrastructure Chief Economist Office

October 2021

Policy Research Working Paper 9796

Abstract

Since the Great Depression of the 1930s, and through the more recent Asian Crisis of 1997 and Great Recession of 2008/09, governments have experimented with Keynesian style fiscal stimulus to support employment and accelerate economic recovery. The effectiveness of these policies depends on the size of fiscal multipliers. A large body of economic literature has estimated such multipliers, with gradually increasing precision, due to econometric improvements and better ways to identify fiscal impulses. Overall, the largest multipliers are found to be associated with public investment, as opposed to other types of spending. Such public investment multipliers are typically below one in the short run, but studies with multi-year horizons suggest that values higher than unity can be attained over time. The size of multipliers is sensitive to economic conditions. During recessions, and periods of high unemployment, transfer payments appear sometimes to offer higher multipliers

than public investment. An important exception is when fiscal and monetary policies are closely coordinated and interest rates approach zero, conditions that provide the strongest evidence for the efficacy of public investment multipliers. Other institutional factors also play a crucial role in determining the size of the public investment mul- tiplier, in particular the country's absorptive capacity, and the selection of high-quality shovel ready projects. However, there is limited empirical evidence available on the magnitude of fiscal multipliers in developing country settings, or for infrastructure sectors or subsectors specifically. The few studies available suggest that certain types of green infrastructure (energy efficiency, solar energy, and so forth) may bring employment benefits in the short run, while innovative digital infrastructure may yield longer-run benefits for economic growth. The relevance of these findings to the current COVID-19 crisis is explored.

This paper is a product of the Infrastructure Chief Economist Office. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at [e-mail] and [e-mail].

The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.

Produced by the Research Support Team

What Have We Learned about the Effectiveness of Infrastructure Investment

as a Fiscal Stimulus? A Literature Review*

Maria Vagliasindi and Nisan Gorgulu**

JEL: C54, E32, E62, H20, H5, N10, Q4, R4

Keywords: Infrastructure investment, fiscal stimulus, economic growth, COVID-19, infrastructure stimulus, countercyclical policy

*Special thanks go to Luis Andres, Tito Cordella, Vivien Foster, Fernanda Ruiz-Nunez and Stephane Straub for the excellent comments and insightful suggestions on a previous draft of the paper and Fan Zhang for her initial work on this topic summarized in Foster, Vivien (2020) Addressing crisis through infrastructure Official Monetary and Financial Institutions Forum, Op-Ed.

• Maria Vagliasindi is Lead Economist in the Chief Economist Office of the Infrastructure Practice Group, and Nisan Gorgulu is Short Term Consultant in the same unit.

1. Introduction

According to the Keynesian approach, government spending can be used as a powerful stimulus, particularly during times of high unemployment. A large body of economic literature, starting from the Great Recession, has estimated fiscal multipliers first focusing on overall government spending, and then disentangling some subcomponents, such as public investments, and in few cases public infrastructure investment. The effectiveness of fiscal stimulus packages features prominently in public debates surrounding the current COVID-19 crisis, as policy makers seek to understand whether encouraging public investment/infrastructure would help to raise economic growth, increase productivity, and crowding in the private sector.

The objective of this paper is to review the academic literature on fiscal multipliers in general, as well as the efficacy of the fiscal response to previous crises in particular, so as to draw the key lessons that can be applied to the current COVID-19 crisis. Despite our focus being on investment in infrastructure for developing countries, this nonetheless leads us to a broader examination on the fiscal multiplier literature across all types of spending, much of which centers on developed countries due to limitations in data and available research.

Ramey's (2011) pioneering contribution, surveying the literature after the 2009 financial crisis, finds fiscal spending multipliers for developed economies within the 0.5 to 2 range. Subsequent contributions -- including Ramey (2019) -- refine such estimates, coming up with a narrower 0.6 to 1 range. The narrowing range of estimates are partly due to new techniques used to compute fiscal spending multipliers. In a nutshell, the encouraging news is that their value is positive but less than or equal to unity, meaning that spending raises gross domestic product (GDP), but does not stimulate additional private activity and may crowd it out.

The remainder of the paper is structured as follows. Section 2 reviews the key findings on the size of fiscal multipliers. Section 3 provides a historical perspective on the effectiveness of fiscal stimulus during crises. Section 4 provides early evidence of stimulus packages announced during the COVID-19 pandemic. Section 5 draws preliminary policy recommendations.

2. Evidence on the Magnitude of Fiscal Multipliers

Fiscal multipliers are generally derived from the calibration of New Keynesian DSGE models, from structural macro-econometric models, and the so-called narrative method. Since the work of Fatás and Mihov (2001) and the seminal contribution by Blanchard and Perotti (2002), empirical estimates of fiscal multipliers tend to rely on vector autoregressive (VAR) models, with new contributions increasingly making use of sophisticated identification techniques to address the possible endogeneity of fiscal shocks. While the long time series available for countries such as the US allows for the use of narrative methods to identify exogenous shocks (Ramey, 2011), estimates based on shorter time series for European and developing countries still rely on less refined methods. As noted in Kraay (2012), whereas different types of government spending may have different short-run effects on output identify disaggregated multipliers is limited by imperfect data on the composition of spending. Identifying a plausible identification strategy for total spending is hard enough. Once different subcomponents of government spending are considered, separate instruments for the different types of spending would need to be considered, which makes the problem extremely challenging.

2

The narrative approach constitutes a methodological improvement upon the traditional measurement of fiscal shocks. The structural VAR methodology, which employs output elasticities of expenditure to filter out automatic stabilizers, may fail to capture exogenous policy changes correctly, because changes in expenditure are not only due to output developments and discretionary policy, but may also be attributable to asset and commodity price movements. The narrative approach instead seeks to identify exogenous fiscal shocks directly. In addition, SVAR models by providing "average" multiplier may not be able measure accurately fiscal multipliers in the case countries have been implementing major structural changes. Finally, they may not be able capture state-contingent multiplier unless they employ non linear estimations. One advantage of DSGE models is that they describe the behavior of the economy as a whole by analyzing the interaction of many microeconomic decisions. However, results of simulations tend to be sensitive to the choice of certain parameters (e.g., degree of price and wage rigidities, investment adjustment cost, and proportion of liquidity-constrained agents) and to the specific modeling assumptions, especially if the models are calibrated rather than estimated.

To illustrate how different definitions of the multiplier can lead to strikingly different estimates, Ramey (2019) computes the effects of fiscal spending shocks, with three different methods. The first one follows the approach of Blanchard and Perotti (2002), focusing on the ratio between the peak of the response of output and the impact response of fiscal spending. This method leads to a 'quasi‐multiplier' because it overlooks the role played by fiscal spending persistence. The second one follows Mountford and Uhlig (2009) proposing computing cumulative fiscal multipliers discounting the future realization of output and fiscal spending as predicted by the VAR impulse responses. The third one based on the analysis of Hall (2009) and Barro and Redlick (2011) avoids the use of the ratio between output over public spending conversion ratio to convert the estimated elasticities to dollar terms (by transforming the variables employed in the analysis to the same units before the estimation of the econometric model). This is achieved by dividing them either by past real GDP or by past real potential GDP.1

The key conclusion from Ramey's (2019) comparison is that the calculation of multipliers à la Blanchard and Perotti (2002) can induce a substantial upward bias in the figures related to this spending multiplier. Once the persistence of spending is taken into account, the multiplier becomes less than one. Getting rid of the conversion factor reduces even further the multiplier, at least conditional on the data set employed by Ramey (2019).

In sum, the more plausible lower-end estimates of the fiscal multipliers come from more data-driven time series and narrative methods, while the upper bounds are the outputs of more sophisticated calibrated models, based on new Keynesian types of model and the use of vector autoregressive (VAR) models. However, the calibration sometimes relies on strong assumptions either in the theoretical models or in the econometric analyses to identify the fiscal policy effect.

In the following sub-sections, we explore the range of multipliers, depending on the fiscal instrument used, on the "state" of the economy as well as the composition of government spending, focusing on infrastructure investments.

Within sections 2.1 and 2.2, the concept of "multiplier" used to capture the impact of fiscal stimulus packages is simply the ratio of the expected change in output (GDP) over the proposed government outlay.

1 The use of this conversion factor is problematic for two reasons. First, it is unstable over long sample periods. Second, as noted by Sims and Wolff (2018), the numerator is acyclical while the denominator is procyclical. Hence, the ratio is procyclical too, which implies that the use of a constant ratio overestimates the spending multiplier in recessions (and underestimates it in expansions).

3