Government investment effect:

an empirical assessment for the euro area countries

Matteo Deleidi            Francesca Iafrate        Enrico Sergio Levrero

Keywords: Fiscal Multipliers; Government Investment; Local Projections; Euro area

JEL codes: H50, H54, E60

Abstract

Until the recent Great Recession, monetary policy was regarded as the primary tool for business cycle stabilization. In particular, the preference for monetary policy was justified by scepticism regarding the impact of discretionary fiscal policy on economic activity and by emphasising that a discretionary monetary policy could be enforced more quickly than a fiscal policy decision, as a consequence of the long implementation lags of the latter.

However, in recent years, a series of elements has determined a renewed interest in the effectiveness of fiscal policy and therefore the magnitude of fiscal multipliers. One of these is related to doubts of the International Monetary Fund (IMF) concerning the previously elaborated growth projection and the effective magnitude of fiscal multipliers. Indeed, in the World Economic Outlook of October 2012 (IMF, 2012), it is claimed that a fiscal consolidation of 1% of GDP has determined an output growth that is lower than the expected one. In this regard, Blanchard and Leigh (2013) have suggested that the errors on growth projections should derive from a value of fiscal multipliers that is higher than the one previously assumed. More specifically, they affirm that the fiscal multipliers have ranged from 0.7 to 1.5 since the Great Recession, whereas they were estimated as being approximately equal to 0.5 until 2009. Grounded on this aspect, this debate has recently gained momentum among international institutions and academic scholars.

Since the recent literature has developed alternative approaches to inferring the effects of public expenditures on GDP, this paper starts by reviewing the recent literature on fiscal multipliers. To do this, we distinguish between different methods according to the model class applied and identify two main ways used to estimate fiscal multipliers. The first one is a theoretical approach, based on simulations built within the Dynamic Stochastic General Equilibrium models, such as Real Business Cycle and New-Keynesian models; the second one employs econometric techniques and relies on a structural VAR and Local Projections approach.

This paper also aims to provide a clear picture of the impact of government investment on the GDP level in selected euro area countries. For this reason, it shows our own estimates of the government investment multiplier for the euro area counties. We focus on government investment because few contributions on fiscal multipliers have distinguished between the effects of the single components of government expenditure (Perotti, 2004b; Auerbach and Gorodnichenko, 2012). Moreover, international institutions, such as the IMF and the European Commission (EC), have recognized its strategic role in supporting economic growth both in the short and the long run, especially in this historical period characterized by the Zero Lower Bond (ZLB). Indeed, the IMF (2014) suggests that government investment boosts economic activity through the fiscal multiplier and a virtual crowding-in effect on private investment in the short run, and an enlarging effect on the productivity capacity in the long run. Similarly, the European Commission (EC) highlights that public investments in infrastructure are one of the main policy levers for supporting economic growth by proposing a public investment plan in 2014 to get out of the economic stagnation which started at the beginning of 2008.

For the purposes of this paper, we implement an econometric analysis based on advanced panel techniques; we rely on a sample of 11 euro area countries (Austria, Belgium, Finland, France, Germany, Ireland, Italy, Luxemburg, Netherlands, Portugal, and Spain) and consider the 1970-2016 period. Unfortunately, high-quality detailed quarterly fiscal data are not available for many European countries and we therefore attempt to provide a clear picture of the size of the investment fiscal multiplier for the Eurozone countries by using yearly macroeconomic data provided by the OECD’s Economic Outlook database and the World Development Indicator (WDI) of the World Bank.

In order to understand whether public investment shocks generate permanent, persistent effects on the level of economic activity, we apply the Local Projections method (Jorda, 2005). As clearly explained in Auerbach and Gorodnichenko (2017) and Dell’Erba et al. (2018), it is a natural alternative to SVAR models for obtaining impulse response functions (IRFs).

We use a dynamic two-way fixed effects model to estimate the effects of a government investment shocks on GDP by applying the LPs method. Following Auerbach and Gorodonikencho (2017), who were among the first to apply this method, the estimated model has the form shown in equation (1)

(1)

where  and  index countries and time, respectively;  is the countries fixed effects and  is the time fixed effects.  represents the rate of growth of output between  and , defined as. The coefficient  is the response of at horizon  to the shock at time , thus IRFs are built directly from the  coefficients. In practice, by applying the LPs approach, single regressions in which the variable of interest is considered in each time  following the realization of the shock at time  are estimated. In the spirit of Riera-Crichton et al. (2015) and Owyang et al. (2013), we introduce a set of control variables to clear the coefficients  from their dynamic effects. In particular, we consider  and  which are the GDP growth rate and government spending rate of growth at time, respectively. Additionally,  is a vector which contains the real effective exchange rate (REER) and the long-term interest rate which is included to control for the stance of monetary policy (Auerbach and Gorodnichenko, 2017).

With regard to the fiscal variable,  represents the government investment shocks. It is identified by applying the Blanchard and Perotti identification strategy (Blanchard and Perotti, 2002) which has been one of the main approaches used in the econometric literature. Since such an identification approach assumes that government spending reacts with a lag to macroeconomic conditions, the structural fiscal shock is identified from a Cholesky decomposition in a VAR framework with the public investment rate of growth ordered first. To estimate the effects of public investment shocks on GDP, we proceed in two steps. First, we identify structural shocks of government investment by employing the chosen identification method and then we use them as our measure of structural shocks in the estimation of the Local Projections (Equation 1).

In addition to the linear model (Equation 1), following Riera-Crichton et al. (2015), we relax the assumption of a symmetric government investment multiplier and analyse the potential asymmetric effects on output of government investment depending on whether government is going up or down. In other words, we evaluate whether the magnitude of investment fiscal multiplier depends on the sign of public investment shocks. To allow for asymmetry in the response of GDP, the linear model, specified in Equation 1, is modified by splitting the fiscal variable into positive () and negative () values, as shown in Equation 2.

(2)

In this case, the coefficient denotes the effect of an expansionary shock at time  on  at horizon ; while  is the response of GDP  periods after a contractionary shock at time .

As the fiscal shocks are identified by employing the Blanchard and Perotti approach to the rate of growth of government investment, the estimated coefficients in both models ( and ) represent the elasticities of output to public investment. Hence, fiscal multipliers are determined by multiplying the coefficients by the mean value of . By applying this method, partial derivatives represent the euro-change in GDP () of one-euro increase in public investment ().

In order to assess the sensitivity of our results, we explore whether our findings are robust to the chosen sample period. In particular, we estimate both models by dropping the recent recession period: hence, the new sample runs over the period 1970-2007. This allows us to check whether the years of the great recession have had an impact on the estimations and on the magnitude of fiscal multipliers.

In accordance with the IMF contribution (IMF, 2014), our main findings show that government investment shocks positively affect the economic growth both in the short and in the long run, by generating a permanent and positive effect on the level of economic activity. Results obtained from linear model clearly provide an impact elasticity of GDP equal to 2.7% that a public investment multiplier of 0.84 corresponds to. Moreover, the peak of the effect on output is reached during the fifth year after the public investment shock and it remains constant in the year ahead. More specifically, the peak elasticity and multiplier are equal to 10% and 3.12, respectively. Additionally, when positive and negative shocks are considered, we find that the impulse responses are strongly asymmetric. The effects on GDP determined by increases in government investment are greater than the one estimated in the linear models, while the output effects associated with negative public investment shocks are lower. Furthermore, comparing the effects on GPD generated by positive and negative shocks, we find that the former are always larger than the latter. On impact, the elasticity of output for government investment increases is equal to 3.4% and the one for negative shocks is 2.4%. Both GDP elasticities tend to grow during the years after the shock and reach a maximum between the fifth and sixth year, although their magnitude is approximately twice larger for expansionary shocks. In particular, GDP elasticities are equal to 14% for positive public investment shocks and equal to 7.9% for contractionary ones. The strong asymmetric elasticities imply asymmetries in the investment multipliers. When government investment goes up, the fiscal multiplier is of 1.06 on impact and reaches a peak of 4.38. On the contrary, the impact and peak multipliers associated with a fall of public investment are equal to 0.75 and 2.47, respectively.

Finally, results are robust when the recent recession years are dropped. Both models produce similar results in terms of the sign and persistence of the effects. Furthermore, in this case, the fiscal investment multipliers have a higher magnitude in the first three years after the shocks than the baseline results. This allows us to conclude that the years of the Great Recession have had a positive impact on the value of fiscal multipliers.

Consistently, our findings are compatible with the presence of Keynesian effects on the output level, due to the values being extensively larger than the fiscal multiplier ones. The policy implications of our findings suggest that demand policies – especially those based on the financing of public investment plans – would stimulate GDP, allowing European countries to get out of the current stagnation.