Does fiscal consolidation hurt economic growth? Empirical evidence from Spanish regions

This article brings empirical evidence on the effect of fiscal consolidation in decentralized countries. The focus on Spain is justified by three reasons. First, it is one of the OECD countries most affected by the Great recession in terms of both GDP and public deficit. Second, Spain is one of the most decentralized countries in the world. Third, compliance with fiscal consolidation targets has been very diverse across regions. Using both time series econometrics and the Synthetic Control Method approach (SCM), we show that compliance with fiscal targets at the regional level has not involved lower GDP growth rates in the short-run. Openness and economic integration of regional economies involve that fiscal multipliers tend to fade. Hence, while a fiscal stimulus would not work on this scale, the opposite is also true: the potentially negative demand effects of a stronger regional fiscal consolidation strategy would be exported to other regions. JEL H74 R11 H62


Motivation
The Great Recession revived the long-standing debate between Keynesians and "supply-side" scholars (Andrés and Doménech, 2013), and it reopened the discussion on the relevance of fiscal multipliers and the effect of fiscal consolidation on GDP growth in the short-run. Researchers are still divided between those who affirm that fiscal consolidation can end up producing expansive effects (Giavazzi and Pagano, 1990;Ardagna, 1998 and2010), especially when based on expenditure cuts (Alesina et al., 2019), and those who maintain that immediate fiscal austerity is counterproductive for economies (Blanchard and Cottarelli, 2010;Blanchard and Leigh, 2013;De Grauwe and Ji, 2013;Fatàs and Summers, 2018).
A middle position outlines the importance of the initial conditions (economic and fiscal), claiming that the anticipated consolidation is better than a progressive form if the starting value of the debt/GDP ratio is huge (Nickel and Tudyka, 2013) or if the period of consolidation effort follows a financial crisis (Buti and Pench, 2012). Moreover, the cyclical situation of the economy is relevant: fiscal multipliers are significantly higher in turbulent times (Auerbach and Gorodnichenko, 2012;De Mello, 2013;Warmedinger et al., 2015;Hernández de Cos and Moral-Benito, 2016;Attinasi and Klemm, 2016).
As a consequence of this diversity, opposite estimates of fiscal multipliers are reported. Spilimbergo et al. (2009) show fiscal multipliers ranging between −1.5 and 5.2, with Corsetti et al. (2012) reducing these magnitudes to −0.7 and 2.3, and Gechert and Wil (2012) to −1.3 and 2.8.
The divergence shows the difficulty of isolating the effects of fiscal stimulus policies from those of other factors affecting the economy: the exchange rate, monetary policy, health of public finances, availability of bank lending, and so on. In order to control by those factors, the analysis of highly decentralized countries can provide further insights into this empirical question. Comparing the performance of regions that are subject to the same macroeconomic conditions but committed to fiscal consolidation to different degrees makes the analysis much easier than comparing countries. Furthermore, the debate on the effect of decentralization on fiscal stability of countries is still open (Lago-Peñas et al, 2019). Looking at the growth effects of regional fiscal consolidations may also help to shed additional light on this debate. This is the second target of this article.
Insofar as this empirical analysis requires strong expenditure decentralization, a regional capacity to enter into debt and a substantial shock affecting public finances and involving the need for fiscal consolidation, Spain is the best candidate among the OECD members.
In short, our analysis is based on the short-run GDP dynamics of the regional leader in the compliance with fiscal consolidation targets (Galicia) and compares the observed GDP growth rates with what would have happened with softer regional fiscal consolidation. Two complementary analyses are carried out. In Section 3, we perform a time-series econometric analysis and compare the observed GDP growth rates with a forecast. In Section 4, we rely on the more sophisticated synthetic control method (SCM) approach to make the comparison. Section 5 concludes.

The fiscal adjustment in Spain and the case of Galicia
Spain is one of the most decentralized countries in the world . In terms of the share of sub-national (both local and regional) in total public expenditure, it ranks fifth in the OECD. 1 If the attention focuses on the regional tier, Spain tops the ranking for the European Union (EU), with figures similar to those of Canada and Switzerland. Moreover, the impact of the Great Recession on the deficit and debt in Spain has been huge. Public debt rose from 36% of the GDP in 2007 to 100% (Lago-Peñas, 2017). While the economic crisis struck the GDP growth rates of all regions ( Figure 1) and the deficit expanded in all cases, there are significant differences in the implementation of budget consolidation and thus the meeting of the fiscal targets agreed with the European Commission.

_________________________
1 Furthermore, according to the Regional Authority Index, computed by Hooghe et al. (2016) (http://garymarks.web.unc.edu/data/regional-authority/), Spain was second in the last year available (2010). Only Germany shows a higher score. The sample consists of all the EU member states, all the member states of the OECD, all the Latin American countries, ten countries in Europe beyond the EU and eleven countries in the Pacific and South-East Asia.   Figures 2 and 3 show the evolution of the regional average deficit and debt over the period 2008 to 2017. The regional deficit increased substantially to exceed −3% of the GDP in 2010 and 2011. Correspondingly, public indebtedness rose from less than 7% of the GDP in 2008 to almost 25% in 2017. The figures for Galicia are significantly lower for both deficit and debt.
Lago-Peñas and Vaquero (2016) and Lago-Peñas et al. (2017) demonstrate that the evolution of deficits and fiscal consolidation efforts differed significantly across regions. Table  1 summarizes the non-compliance of fiscal targets by Spanish regional governments over the period 2009-2016. Fiscal targets include the deficit target expressed in terms of regional GDP, a constraint on total spending growth (the so called-"spending rule"), and a debt target expressed again in terms of regional GDP. 2 In spite of a drop in the total revenues over the average, 3 Galicia stands out as the region with the highest degree of fulfilment. 4 Finally, the available evidence shows that spending cuts account for the lion's share of fiscal consolidation in all the regions (MINHAP, 2012;Cantalapiedra and López, 2013). Taking into account increases in both tax rates and tax benefits, the net average effect is neutral (AIReF, 2016), with some regions in negative figures (especially Madrid and Cantabria, due to tax cuts in wealth taxes). For the case of Galicia, the net effect is positive but small. In 2016, the contribution of net tax increases to fiscal stability was equivalent to 0.2% of the regional GDP. 3 The regions where both the total revenues and the total expenditure dropped more than the regional average are Galicia, Baleares, Castilla y Leon, Extremadura, Navarra and Andalucía.  Concerning the second factor, the new conservative Galician incumbent replaced a leftist coalition and made fiscal austerity the main political message of its political campaign. Hence, the new government took advantage of the timing of the election to adapt the regional budget to the new economic scenario. Regional expenditure cuts started in 2009 and intensified in 2010. Since then, austerity has been the cornerstone of political will of the regional president and ministers, including more exigent measures than in most regions, like the cut in civil servants' wages approved in 2013 and extended until 2017. Surprisingly, and in sharp contrast to the electoral results in other regions and countries, the political support for the Galician incumbent increased in the 2012 and 2016 elections.
To analyse the effect on growth of the stronger fiscal consolidation process carried out in Galicia, two complementary methodologies are used: a standard econometric approach based on a dynamic time-series model (Section 3) and the synthetic control method originally proposed by Abadie and Gardeazabal (2003) in Section 4. In both cases, we compare two series: the observed GDP growth rate affected by cross-regional asymmetries in the intensity of fiscal consolidation since 2009 and the simulated GDP growth rate from extrapolating the structural relationship before 2009.

Empirical analysis I: A standard econometric approach
The departure point of the first econometric analysis is the following AR (2) model: In Equation (1), the variables GAL and SPAIN are inter-annual real GDP growth rates of Galicia and Spain, respectively. 5 The data for the former are from the Galician Institute of Statistics (www.ige.eu), and the data for Spain are from the National Institute of Statistics (www.ine.es). In both cases, we rely on the quarterly seasonally adjusted data provided on the websites. We use Eviews 10.5 for the econometric estimates.
The estimation period is 1996:Q1 to 2008:Q4. The selection of the starting point is based on the evidence provided by Lago-Peñas (2001), who shows that the synchrony of the Spanish and _________________________ Galician business cycles has substantially converged since the early 1990s. Therefore, the goodness of fit will tend to be significantly higher, while the sample size is large enough to perform estimates. Two lags for the variable GAL are enough to capture the dynamics. 6 To rule out specification problems, we perform two complementary tests on the specification. First, a Hausman test on the endogeneity of the variable SPAIN's p-value of the corresponding Wald test is high (p-value=0.25) and thus the null hypothesis of exogeneity is not rejected. 7 Second, a Granger causality test is performed to verify the causation order. The results in Table 2 clearly show that we can reject the null hypothesis of no causation from SPAIN to GAL (p-value<0.001) but not the opposite causation (p-value=0.88). Table 3 reports the ordinary least squares (OLS) estimate of Equation (1). The fit is good (R 2 =0.853). The null hypothesis of no autocorrelation cannot be discarded according to the Breusch-Godfrey serial test (p-value=0.12). The corresponding correlogram yields the same conclusion. Both SPAIN and the first lag of GAL, but not the second one, are highly significant variables, with coefficients around 0.6.
Our econometric results allow us to forecast the Galician GDP growth rate from 2009 to 2017. In particular, using observed data for the variable SPAIN, we perform a dynamic forecast assuming that the relationship estimated in Table 2 will hold. Figure 4 shows both series. The observed GDP growth is clearly above the forecasted growth throughout the period 2009-2014. The sign of the differential changes at the end of 2014, when the forecasted growth exceeds the observed growth. Hence, in 2015 and 2016, the Galician GDP growth rate was slower that it should be, taking into account the recovery of the Spanish economy. The situation changes again at the end of 2016 when the two series overlap. Moreover, in 2017, the observed GDP growth rate is slightly higher than that forecasted.
In sum, looking at the whole period, the observed GDP growth rate is significantly above the forecasted rate. Hence, the data do not support the existence of a price to pay for fiscal consolidation in terms of short-run GDP growth. 8

Number of observations 50
Notes: t-statistics in parenthesis. *** p-value < 0.01 Source: Authors' elaboration. The SCM analyses the consequences of a specific event or intervention in a region by comparing the observed data with what would have happened in the absence of the intervention. Insofar as it requires the creation of a counterfactual, the first step of the method is to find other similar regions that remained unaffected by the event (the comparison units) and merge them to create a synthetic region close to the case of interest. More formally, the SCM supposes that there are J+1 regions or "units", where j=1 is the "treated unit" and units j=2 to j=J+1 constitute the "donor pool" (Abadie et al., 2010: 493-505). Following Abadie et al. (2010) and Abadie et al. (2015), we assume that the sample is a balanced panel and includes a positive number of preintervention periods, T 0 , as well as a positive number of post-intervention periods, T 1 , with T=T 1 +T 0.
The synthetic control is defined as a weighted average of the units in the donor pool represented by a (J×1) vector of weights W=(w 2 ,….,w J+1 ), with 0≤w j ≤1 for j=2, …..J and w 2 +….+ w J+1 =1. Choosing a particular value for W is equivalent to choosing a synthetic control. Let X 1 be a (k×1) vector containing the values of the pre-intervention characteristics of the treated unit that we aim to match as closely as possible, and let X 0 be the k×J matrix collecting the values of the same variables for the units in the donor pool. The pre-intervention characteristics in X 1 and X 0 may include the pre-intervention values of the outcome variable.
The difference between the pre-intervention characteristics of the treated unit and a synthetic control is given by the vector X 1 −X 0 W. As in the research by Abadie et al. (2010) and Abadie et al. (2015), we select the synthetic control, W*, that minimizes the size of this difference. This can be operationalized in the following manner. For m=1, …., k, let X 1m be the value of the m−th variable for the treated unit and let X 0m be a 1×J vector containing the values of the m-th variable for the units in the donor pool. Abadie et al. (2015) choose W* as the value of W that minimizes: where v m is a weight that reflects the relative importance assigned to the m-th variables when we measure the discrepancy between X 1 and X 0 W. Of course, it is crucial that the synthetic control closely reproduces the values that the variables with large predictive power over the outcome of interest take for the treated unit. Accordingly, those variables should be assigned large v m weights. In the empirical application below, we apply a cross-validation method to choose v m . Let Y jt be the outcome of unit j at time t. In addition, let Y 1 be a (T 1 ×1) vector collecting the postintervention values of the outcome for the treated unit. That is, Y 1 = (Y 1To +1, … , Y 1T )'. Similarly, let Y 0 be a (T 1 ×J) matrix, where column j contains the post-intervention values of the outcome for unit j+1. The synthetic control estimator of the effect of the treatment is given by the comparison of post-intervention outcomes between the treated unit, which is exposed to the intervention, and the synthetic control, which is not exposed to the intervention, Y 1 -Y 0 W*. That is, for a post-intervention period t (with t≥T 0 ), the synthetic control estimation of the effect of the treatment is given by the comparison between the outcome for the treated unit and the outcome for the synthetic control in that period: The matching variables in X 0 and X 1 are meant to be predictors of post-intervention outcomes. These predictors are themselves not affected by the intervention. To choose the weights v m in Equation (2), a cross-validation technique is used, following Abadie et al. (2015); these weights minimize the root mean square prediction error: In our case, the variables used for the SCM are reported in Table 4. We include variables on demography, the labour market, human capital, physical capital (both public infrastructure and private), exports and foreign investment inflows, poverty (AROPE), competitiveness (the Regional Competitive Index or RCI computed by the European Commission) and the sectorial structure. While in most cases we use average values over the pre-treatment period, data for the year 2008 are used for some variables. 9 The sample covers the period 1995-2016 for the 17 Spanish regions. We choose the weights v m to minimize the root mean square prediction error over the validation period. Table 5 shows the computed weights, which are those used to create the "synthetic Galicia", for which the values are collected together with the corresponding actual values in Table 6.  Economics: The Open-Access, Open-Assessment E-Journal 13 (2019-42) www.economics-ejournal.org 12  Figure 6 presents the previous annual growth rates. In both cases, the value is negative, but the actual rate (−2.30%) is clearly lower in absolute terms than that corresponding to the synthetic unit (−3.91%). Figure 7 shows both the actual and the synthetic values of the variation rates of the GDP for Galicia from 1995 to 2016. While most of the actual values are lower before 2008, since then they have been higher than the synthetic estimate of the region's growth. This again supports the idea that fiscal austerity implemented to meet the deficit targets has not involved an evident price in terms of regional economic growth.
To check the robustness of the results, we replicate the analysis, constraining the donor regions to those in which the deviation from the fiscal consolidation targets has been wider (Castilla-La Mancha, Cataluña, Comunidad Valenciana and Murcia). The regional weights that minimize the root mean square prediction error (RMSPE=0.0071) over the validation period are 0.791 for Cataluña and 0.219 for Castilla-La Mancha. Figure 8 shows the new results. Over the period 2004-2012, the growth of the counterfactual of Galicia is below the actual growth. After 2013, the values are slightly higher, but the positive gap in the counterfactual in recent years does not compensate for the negative gap in 2012.

Conclusions
Regional governments provide a convenient field to test whether fiscal consolidation involves a price to pay in terms of the short-run GDP growth rate. Given the high fiscal decentralization in Spain, the depth of the effects of the Great Depression on the public accounts at the regional level in Spain and the diversity of commitments to fiscal targets, this country is the best case study among the OECD members.
To perform the test, we focused our attention on the region showing a better performance in terms of compliance of the fiscal targets agreed with the European Commission. In particular, we compare the observed short-run GDP growth rates with two simulations. The first one is based on the estimate of a time-series model relating the GDP dynamics of Galicia and Spain over the period 1995-2008. The second one relies on the more sophisticated synthetic control method (SCM) approach. In both cases, we show that stricter fiscal consolidation has not produced a significant reduction in economic growth. Indeed, the economic performance of Galicia would have been slightly better than expected using both methodologies.
Our interpretation of the results is closely related to the traditional theory of fiscal federalism (Musgrave, 1959). According to it, openness and economic integration of regional economies involve fiscal multipliers that tend to fade. A fiscal stimulus would not work on this scale. Our results show that the opposite is also true: the potentially negative demand effects of a stronger regional fiscal consolidation strategy would be exported to other regions.
The most relevant policy implication of our results is that decentralization could indeed contribute to the compliance with fiscal consolidation plans on the national scale, reversing the usual arguments on both the common pool's problem affecting fiscal stability in decentralized countries and the challenge involved by "soft budget constraints" and bailout expectations (Goodspeed, 2017). If the negative demand effects of regional fiscal consolidation are mostly exported to other regions, one of the reasons most often argued to delay deficit cuts and compliance with fiscal targets is deactivated.
The smartest strategy for all regions is to cut deficit, getting the benefits and exporting the costs in terms of lower GDP growth rates. But if decentralization is very high and all regions play the same, fiscal consolidation at the country level will tend to be stronger than is centralized countries. In other words, while the total negative effect of fiscal consolidation on national GDP growth rate could be the same than in a centralized country, the incumbents' perceived costs of fiscal adjustments be downward biased and then, fiscal consolidation in the country as a whole could be stronger.
It is tentative to extend this interpretation to supranational areas such as the European Union. First, from an analytical standpoint, the degree of economic integration in supranational areas would be highly relevant to understand cross-national differences in fiscal multipliers. Hence, this factor should be included in the list of control variables in cross-country estimates. Second, from a policy perspective, further progress in economic integration (trade, banking, fiscal flows…) will reduce the positive effects of national fiscal policies, but it will also fade the potential negative effects on GDP growth rates of national fiscal adjustments.