1 Introduction and Literature Review

The relationship between the size of government and the underlying growth rate of an economy has obviously not been a new topic. This debate has been conducted in various fields during the decades, each time under the prism of the contemporary socio-economic conditions. Post World War II left-right politics, various theories on political economy, evolution of econometric studies, and even further, political philosophy have been a few of the areas in which this dialogue has taken place.

Nevertheless, this debate seems to be returning back to the forefront. Earlier research finds a negative correlation between government size and growth. In his study, Cameron (1982) does find a negative correlation between the average percentage of government expenditure and the average rate of growth in real GDP ; still, the chapter concludes that increased spending does not necessarily lead to stagflation, that is, a situation in which inflation is high, economic growth slows, and unemployment remains persistently high. Similarly, Landau (1983), after expanding the dataset with education, energy consumption , and other dummy variables, also finds a negative correlation. Analogous results can be found in other research such as Marlow (1986), Barro (1991), Engen and Skinner (1992), Hansson and Henrekson (1994), and Grier (1997).

It would be a surprise not to find work that rejects the hypothesis of a strong negative correlation between government spending and growth; Mendoza, Milesi-Ferreti, and Asea (1997) is one. However, such findings have been largely questioned due to the statistical methods used or the initial assumptions made. The rationale on which criticism is based is what is more widely known as “Wagner’s law ”. The German economist Adolph Wagner (1835–1917) observed that, in the early industrialized economies, public expenditure was rising constantly as the gross national product was also growing. More recently, Easterly and Rebelo (1993) do show that there is a strong positive relationship between government size and per capita income , but this mainly refers to lower levels of income and does not hold for the highest levels of income.

More contemporary studies continue to explore further aspects of this correlation and offer more insight regarding cross-country differentiation and appropriateness of variables used. Fölster and Henrekson (2000) find a negative correlation, both in the case of richer countries, as well as when this small sample is expanded with non-OECD countries. What is more interesting is that results are more robust when government expenditure is used as an independent variable instead of total tax revenue .

Dar and Amir Khalkhali (2002), when they examined 19 OECD countries over the 1971–1999 period, concluded that, among others, total factor productivity growth is negatively correlated to the size of government; private-sector efficiency, fewer centrally imposed policy distortions, and the crowding-out effectFootnote 1 are some of the main arguments when trying to explain these results.

Romero-Avila and Strauch (2008) move one step ahead, breaking down government spending . They find that government consumption and transfers tend to affect GDP per capita in a negative manner, whereas government spending has a positive effect when it takes the form of investment , highlighting the distorting effects of taxation on the accumulation of private physical capital. This is a differentiation similar to the work of Barro (1990), who distinguishes four categories of public finances: productive vs. non-productive spending (if expenditure is contributing to growth or not), and distorting vs. non-distorting taxation (if taxation is affecting the investment decision).

Looking at a wide range of research on the fields of development economics, one can find that government spending on infrastructure, support for R&D in the form of subsidies, setting a minimum wage, and/or funding for schooling can all have a positive longer-term effect on growth. This is more evident in the emerging and developing world, where this form of spending pushes out the Production-Possibility Frontier.

In Afonso et al. (2005), the authors conclude that “big governments” tend to perform less efficiently compared to “small governments”. Again, going one step forward, Afonso and Furceri (2008) find that it is not only the level of the underlying variable which measures the size of government but its volatility, as well, that tends to have a negative effect on growth.

More specifically, indirect taxes, social contributions , government, subsidies, and government investment have a negative effect on growth; moreover, the higher the volatility, the lower the underlying growth in the economy. Only in a subset (EU countries) of the sample (OECD) do transfers have a positive and significant effect on growth.

Bergh and Karlsson (2010) added another factor in the equation: economic freedom and globalization. They do not deviate from other studies and confirm a statistically strong negative correlation between government size and its effect on growth. However, they argue that countries characterized by a high degree of openness and sound economic policies can use these features to alleviate the aforementioned negative correlation. Although the negative relationship still holds, countries scoring high in the Konjunkturforschungsstelle (KOF) Globalization index or the Fraser Institute’s Economic Freedom index could end up with a weaker negative correlation. Hence, policies to lighten the burden of a big government upon the growth rate of the country could be achieved indirectly by increasing this country’s openness to trade.

In this chapter, we bound our research on the relationship between the size of government and the underlying growth rate to the case of Greece. This country has already signed three memoranda of understanding and is already taking additional measures as a prelude for what could end up being a fourth one. It differentiates considerably from other cases in the European periphery and has lagged significantly in returning sustainably to growth rates . One can argue that it illustrates policy mistakes and one-size-fits-all approaches of correcting imbalances in the European Union, as well as inefficiencies commonly seen in the rest of the European south; and all this at a time when doubts over the coherence of the European Union have been rising, especially after the UK referendum and the triggering of Article 50 for exit from the European Union. We follow a two-part empirical methodology, first considering variations of the national income identity and, second, the temporal persistence of different explanatory variables on economic growth. We are mainly interested in examining the negative relationship between taxes and growth, at a time when the disposable income of the Greek household is shrinking, and the government’s budget constraints do not allow for fiscal expansion.

The rest of the chapter is structured as follows: in Sect. 14.2, we describe the data and variables used; in Sect. 14.3, we discuss our empirical methodology; in Sect. 14.4, we discuss the first round of our results and introduce the notion of the budget constraint ; in Sect. 14.5, we focus on policy implications; finally, in Sect. 14.6, we offer some concluding remarks and extensions of the current research.

2 Data and Variables

The source for all our data is the National Statistical Service of Greece, either directly from the official site or downloaded from Bloomberg, for maximum cross-variable availability and consistency. Our data is quarterly, spans a time frame from Q1 1999 to Q2 2016 and is based on the non-financial accounts of the general government.

We start by using two different measures of GDP as the dependent variable:

  1. 1.

    The year-on-year change of the nominal GDP, non-seasonally adjustedFootnote 2, in percentage terms % (variable name: GDPYOY).

  2. 2.

    The quarter-on-quarter change of the nominal GDP, non-seasonally adjusted, in percentage terms % (variable name: GDPQOQ).

Based on previous literature and our own initial conjectures, the set of explanatory variables includes the following:

  1. 1.

    The quarter-on-quarter change, the year-on-year change, and the share of GDP of the Gross Capital Formation, non-seasonally adjusted, in percentage terms % (variables’ name: CFQOQ, CFYOY, CFGDP, respectively).

  2. 2.

    The quarter-on-quarter change, the year-on-year change, and the share of GDP of the Gross Final Consumption Expenditure, non-seasonally adjusted, in percentage terms % (variables’ name: CONSQOQ, CONSYOY, CONSGDP, respectively); this covers both the general government final consumption and the private/household final consumption.

  3. 3.

    The Trade Balance (Exports minus Imports ) as a share of GDP, non-seasonally adjusted, in percentage terms % (variables’ name: TBGDP).

  4. 4.

    The Unemployment Rate, in percentage terms % (variable name: UNEMPGDP).

  5. 5.

    The Total Revenue, Total Expenditure , and the General Government Balance (Total Revenue minus Total Expenditure ), as a share of GDP, in percentage terms % (variables name: TRGDP, TEGDP, PBGDP, respectively).

  6. 6.

    The quarter-on-quarter change, the year-on-year change, and the share of GDP of the Subsidies Payable, non-seasonally adjusted, in percentage terms % (variables name: SUBQOQ, SUBYOY, SUBGDP, respectively).

  7. 7.

    The quarter-on-quarter change, the year-on-year change, and the share of GDP of the Value-Added Tax Receivable, non-seasonally adjusted, in percentage terms % (variables’ name: VATQOQ, VATYOY, VATGDP, respectively).

  8. 8.

    The quarter-on-quarter change and the year-on-year change of the Retail Sales, non-seasonally adjusted, in percentage terms % (variables’ name: RETQOQ, RETYOY, respectively).

  9. 9.

    The quarter-on-quarter change and the year-on-year change of the Industrial Production, non-seasonally adjusted, in percentage terms % (variables’ name: IPQOQ, IPIYOY, respectively).

The use of the aforementioned variables and their respective lags is based on two assumptions. First, we start by utilizing standard Keynesian macroeconomic theory: the gross domestic product (GDP) is a way to measure a nation’s production (method of total value of all goods and services sold to final users). Then, we proceed and insert variables which are exogenously set by the governments but are expected to affect or alter consumption and, finally, growth. We will statistically test our hypotheses and examine whether our expected thesis, on the negative relationship between government size /higher taxes, and lower growth, holds or not.

As a first remark, (Tables 14.1 and 14.2) the post-crisis period—hence, the years from 2010 to 2016—has been characterized by a negative annual GDP growth rate (−4.493), down from a 1.854% in the full sample, extending from 1999 to 2016. It is the period in which most macroeconomic variables such as investment , consumption , retail sales , industrial production are all collapsing, and unemployment is rising to an average of roughly 22%, while proceeds from VAT as a percentage of GDP remain roughly stable, implying lower proceeds in absolute levels. We are interested in testing if there is any dependence among these variables and, especially, between the ones acting as a proxy to the size of the government—that is, subsidies and the VAT —and the GDP growth rate.

Table 14.1 Full sample descriptive statistics 1999–2016
Full size table
Table 14.2 Post-Crisis Descriptive Statistics 2010–2016
Full size table

3 Methodology

Our empirical methodology is broken into two parts, rather standard but highly illustrative on the results we obtained. The first part considers variations of the national income identity , and the second part considers the temporal persistence of different explanatory variables on economic growth. To this end, consider the following regression specification:

(14.1)

where y t is the appropriate measure of economic growth as the dependent variable, x t is a (K x 1) vector of explanatory variables, β is the (K x 1) vector of parameters, and u t is the regression error term. We assume, and subsequently test, that the regression error passes all standard assumptions. We also assume that some or all of the explanatory variables in the vector x t are endogenous, a standard assumption when working with the set of macroeconomic variables that enter into the national income identity . The parameters of the model are then estimated by instrumental variables (IV), the choice of instruments being confined to a subset of the first four lags of all K explanatory variables we have available. Being aware of the problems that pertain to IV estimation, we take particular care to balance the number of instruments with appropriate specification tests on the validity of the instruments used and, obviously, economic intuition. In particular, we validate the need for the use of IV by applying the Hausman test on the consistency of least squares estimates, and the Sargan test of over-identification and the validity of the chosen instruments in each model. Finally, note that a static model such as the one in Eq. (14.1) can be considered as an “equilibrium” or long-term model, and the interpretation of each parameter estimates should be made as such.

We then consider a simpler framework, in a time series-like context, where we examine the individual total effect—over time—of some explanatory variables on economic growth. We perform this second step in our analysis to validate the inference from the analysis of the first part and to further illustrate the significance of our findings, namely the negative influence of government size and higher taxes on growth. We, thus, consider the following regression model:

(14.2)

where now x tj is one of the components of the vector x t . We are interested in the long-term impact of the explanatory variable which is defined as follows:

(14.3)

that is—scaled by the persistence of economic growth—the sum of the parameters of the lags of the explanatory variable. Note that since all our variables are measured in the same scale (%), we can compare the magnitudes of the estimated long-term impact coefficients and assess the potential priorities on the way a growth-conducive policy might be implemented.

We further discuss our approach in the following sections.

4 Discussion of Results and the Budget Constraint

4.1 Discussion of Results

We start with the expenditure-based approach of GDP which is obtained by summing up household consumption , investment , government spending , and net exports . Thus, we get the standard national income identity :

(14.4)

where C t is private consumption or consumer spending, G t is government spending , I t is investment or business spending, and (Xt − Mt) is the trade balance of exports minus imports .

As a starting point, we are interested in just verifying whether the annual changes in the variables at the right-hand side of the equation affect the GDP growth rate. Indeed, the expenditure-based approach of GDP holds in the case of Greece (Table 14.3, Model 1). All coefficients have a positive sign, in accordance with standard underlying theory, and are all statistically significant, at least at the 10% significance level. At the same time, the model has high explanatory power, again, to be expected, as we are regressing income on its components on an identity. However, this starting point is essential in visualizing, assessing, and discussing what (is now well known that) drives growth in the Greek economy, essentially consumption . From this very basic illustrative model, we proceed to a number of other models, where we want to examine the impact of different explanatory variables on the drivers of economic growth and the size of the government.

Table 14.3 Model estimates, Eq. (14.1), full sample 1999–2016
Full size table

In an attempt to, thus, capture additional information, we proceed with adding the unemployment rate and the industrial production as two additional explanatory variables (Table 14.3, Model 2). Confirming the previous model, positive yearly changes in capital formation and consumption still lead to higher GDP growth rates. Consumption continues to hold the highest coefficient among all. Furthermore, the higher the trade balance as a percentage of GDP, the higher the effect on the growth rate . An interesting point needs to be made at this stage. In our dataset, the trade balance —either in absolute values (TB) or as a share of GDP (TBGDP)—is negative; hence, Greece has been running a trade deficit , with imports surpassing exports in almost all years in our sample. Therefore, the positive sign of the coefficient means that the higher the trade deficit is (imports surpass exports ), the lower the GDP growth rate . This is also consistent with existing literature. Furthermore, contrary to our ex ante expectations, the unemployment rate and annual changes in industrial production do not seem to explain annual changes in the GDP growth rate , as they are not statistically significant and do not survive the related significance tests.

We next move on to the inclusion of the two variables we use as proxies for government size , the subsidies, and the VAT variables. However, we find that these do not have a statistically significant effect on the annual GDP growth rate . We use these two variables as an indicative set of proxies to government size or the government’s means of “interfering” in the economy. Their inclusion in the first two models has resulted in a reduction of the overall explanatory power of the right-hand side variables. This result is not entirely unexpected as annual changes tend to be affected by the general state of the macroeconomy and not by the faster-moving (and more volatile) evolution of subsidies and taxes. Thus, we next model, consider, and discuss quarterly relationships of all the underlying variables, in an attempt to capture the faster-moving information contained in the quarter-to-quarter changes while keeping the same explanatory power and underlying economic intuition.

Looking at our next set of results (Table 14.3, Model 3), we can see that the original set of explanatory variables—capital formation, consumption , and trade balance —remains statistically significant. Adding the unemployment rate as a supplementary explanatory variable, now becomes statistically significant, bearing a negative sign, as implied by the underlying theory. Assuming that unemployment falls to 20% from 25%, quarterly GDP growth rate will benefit by roughly 1.0%. Most importantly, consumption remains as the explanatory variable bearing the highest coefficient; bearing a coefficient of 0.843, a 5% rise in consumption would boost growth by roughly 4.2%—a rather significant outcome. On the other hand, the new sign of the coefficient of the capital formation variable, now, seems to be changing from positive to negative. This strikes a bit at odd as a positive sign of the coefficient would look more appropriate according to literature: positive changes in capital formation are expected to lead to a rising GDP growth rate . A possible explanation for this negative sign might be a confounding effect, where the significance and sign of the role of capital formation is assumed by the other variables plus the rising unemployment during the sample period.

By replacing the capital formation variable with industrial production as a proxy, we end up with a model that has mildly higher explanatory power (Table 14.3, Model 4); most importantly, all signs now look in harmony with the underlying theory and earlier findings. In order to provide an example, a 5% quarterly increase in industrial production would boost GDP by roughly 1.9%; similarly, achieving a trade surplus of 5% against GDP would result to a growth rate of 3.5%. We need to highlight that a decline of 5 percentage points in the rate of unemployment would boost growth by roughly 1.8%, while a 5% increase in total consumption would also boost growth by an estimated 3.8%. A pattern is starting to take shape, which is consumption playing a determinant role with respect to growth.

We need to point out, as noted in passing before, that consumption tends to bear higher-magnitude coefficient estimates and, hence, we feel additional focus should be given on consumption expenditure in building a more comprehensive model. The reason is that, as defined by the National Statistical Service of Greece and envisioned in this chapter too, it includes both private/household and general government consumption . It is worth noting that this variable has risen from an already stunning 85% of GDP towards 90–92% of GDP in late quarters (Fig. 14.1). Even more interestingly, the ratio of household-to-government consumption is highly skewed towards the former with a ratio of roughly 3:1. It is clear, and by now well understood, that the Greek economy has adopted a consumption-driven model all these years. This means that total consumption , and, in particular, private consumption can boost or derail growth, subject to economic conditions, and the current productive structure and capacity of the Greek economy. Both these states of the world, boost and bust of growth, have appeared in Greece throughout the years from 1980 and on, and in that exact order.

Fig. 14.1
figure 1

Total Consumption as a share of GDP

Source: National Statistical Service of Greece

Full size image

Thus, and as an intermediate step, we estimate a regression with total consumption expenditure as the dependent variable; as explanatory variables, we include the unemployment rate , the subsidies, and the VAT as shares of GDP, and the quarterly change of retail sales . Taking into consideration that a significant part of economic activity comes from private consumption , via this step, we are trying to isolate and remove from the picture any direct effect from government consumption . Nevertheless, we do stay focused on the issue of measuring the size of the government indirectly, by incorporating the subsidies and the VAT variables. To be more precise, we do not outright measure the size of the government in the economy using the traditional ways, such as total revenue or total expenditure to name a couple; we are mainly interested in how the government may affect economic activity by altering some of the tools it has in its discretion.

We end up with a very illustrative model for consumption expenditure (Table 14.4), with anticipated signs and interpretations. A rising rate of unemployment (negative sign estimate) and a rising VAT (negative sign estimate) hurt total consumption expenditure , whereas subsidies on products payable (positive sign estimate) support consumption; it is not surprising that retail sales —as a proxy to private consumption —have a positive sign, boosting total expenditure . All coefficients are statistically significant either at the 1% or the 5% level of significance. Offering a more qualitative point of view, consumption is presenting a comparatively huge elasticity to changes in subsidies (38.177) and the VAT (−9.463). The two aforementioned explanatory variables outpace by far the magnitude of both unemployment (−1.866) and retail sales (0.233). At this point, we have suggestive clues that are supportive with regard to the importance of taxation and subsidies as two additional determinants of the behavior and decision-making process of Greek households when consuming.

Table 14.4 Model estimates for consumption growth
Full size table

Given the results from Table 14.4, we next proceed and plug the consumption determinants into the equation for growth, returning our focus on the remaining two models of Table 14.3. The new results (Table 14.3, Model 5) show now that all variables—except for the capital formation and retail sales —are statistically significant at least at the 5% significance level. On top of this, the signs of the coefficients match the ex ante ones from economic theory. Subsidies and an improving trade balance tend to increase growth, as they have a positive sign; on the other side, rising unemployment and increases in VAT as a percentage of GDP tend to hurt growth, as both bear a negative sign coefficient. Contrary to previous models, the capital formation variable no more looks problematic, as the coefficient sign now matches the one suggested by macroeconomic theory (positive); albeit, it is not statistically significant. Moreover, retail sales do not seem to survive significance tests, despite initial expectations. Once more, we are taking a qualitative look, trying to identify if anything stands out in our results, and so is the case; indeed, the magnitude of the subsidies and VAT coefficients arises as quite large, both in absolute terms as well as compared to those of the other explanatory variables. Most importantly, we have managed to break down the effect of—the wider measure of—consumption and identify which parts of it have set it high in the list of explanatory variables in previous models: taxation and subsidies.

It looks like a final adjustment on our model at the government side is needed, in order to account for budget constraints and how they potentially affect our model. At this stage, the means and magnitude of government interference in an open economy is starting to take shape. Subsidies and the VAT are just two out of a whole set of economic variables, which are being set exogenously by the state, but they feed through the economy and, finally, affect how various agents act in the real economic environment. We discuss these issues next.

4.2 The Budget Constraint

The history of the Greek economy, since 2009, is fairly well known. Capital formation and exports /imports have not been factors capable of changing the picture in boosting growth. On the contrary, the former has collapsed, both in absolute terms as well as relatively to GDP, whereas the country continues to run a trade deficit , albeit shrinking (Figs. 14.2 and 14.3). A few factors that have contributed to this development can be identified: a series of national elections, one referendum voting , and an overall fragile political environment have driven political risk to ultra-high levels and investment to ultra-low levels. The imposition of capital controls back in June 2015 amplified this trend.

Fig. 14.2
figure 2

Capital formation in absolute levels and as a share of GDP

Source: National Statistical Service of Greece

Full size image
Fig. 14.3
figure 3

Exports minus Imports as a share of GDP

Source: National Statistical Service of Greece

Full size image

So, how can a government boost growth, given the fiscal limitations that are now imposed? A lot of theories have tried to give an answer to this question, some with notable success and others with less. What seems to be a more appropriate question in this chapter is “how can a government boost growth, subject to its constraints?” In the case of Greece, we need to solve the problem backwards, hence starting with the constraints. The primary budget surplus has been one major constraint, stemming from the already signed bailout programs. Currently, fiscal loosening is not an available tool in the case of Greece. Another constraint is the currently existing model of economic activity. As highlighted previously, the growth model that Greece has been largely following during the last decades has been based on consumption , both private and public. But since primary surpluses need to be reached and funding is hard to get, public spending seems limited. At this point, it seems that a policymaker would try to solve an unsolvable equation: trying to boost GDP, under an existing consumption-based model, not only via cutting down government consumption but, also, by increasing taxes, hurting private consumption as well. A quite difficult, maybe impossible, task one would say.

Nevertheless, what looks as a constraint, might also be part of the solution. Let us, thus, look at a very informative model, (Table 14.3, Model 6). As estimated in previous regressions, growth in industrial production (positive sign) results to a growing GDP, as well. A 5% growth in industrial production would boost growth by 1.7%. Investment in capital goods has a front-loaded, immediate effect—via hiring and increasing household income—as well as a longer-term effect, by pushing out the Production-Possibility Frontier of the economy. Retail sales (positive sign) are also positive for growth but with a lower magnitude; a 5% increase in retail spending would add just 1% to GDP growth . It might currently be hard to see consumer spending by Greek households rising sufficiently. This might require reducing the rate of unemployment (negative sign) to see GDP rising, as increasing taxation and losses in disposable income have been dramatic during the current crisis. The absolute level of the unemployment coefficient is higher than 1, meaning that a reduction by 5 percentage points in the unemployment rate has a multipliable effect upon growth, increasing it by roughly 6.9%. Since unemployment is a lagging macroeconomic indicator, any positive effects from its reduction might need time to diffuse in the economy. As an alternative, the increase in disposable income could as well be a result of cutting back indirect taxes such as the VAT (negative sign), and/or increasing subsidies (positive sign), in order to generate a positive shock to the economy. The use of the word “shock” is not accidental since both explanatory variables have a rather high coefficient. For example, a very small decrease in the state revenues from indirect taxes as a percentage of GDP would act as a multiplier and give a significant boost to the GDP growth rate . Last but not least, running budget surpluses is beneficial by itself, as the estimated coefficient is carrying a positive sign (0.592). At first sight, this might strike at odds with Keynesian economics; the latter dictates that budget deficits are appropriate in order to boost growth in periods of recession whereas saving and budget surplus are best at times of growth. Nevertheless, in the case of Greece, this looks more like the case of “tiding up” public finances and cutting back non-performing areas of the public sector , and less as a growth-disruptive policy tool. What started as a policy constraint (running a budget surplus), now seems to help; no need to remind the reader that focus should be given on the additional variables in the model and the corresponding policies that would boost private consumption .

Before moving on with the discussion of our results of Model 6, we briefly switch focus on the estimates of the individual long-term effect of each individual explanatory variable upon the GDP growth rate and their relevant magnitude, based on the specifications in Eqs. (14.2) and (14.3) (Table 14.5). In the long run, industrial production does not seem to survive the significance tests, whereas all the rest do, at least at the 10% significance level. More significantly, the VAT and the subsidies’ variables present the highest long-term impact estimates.

Table 14.5 Long-Run Estimates, Eq. (14.2)
Full size table

Taking into consideration these long-run effects, and given that VAT is a much larger part of GDP compared to subsidies, we can easily deduce that, should existing fiscal policy were to change, any attempt should start by reducing the VAT . One would argue that lowering the VAT rate would result in lower tax revenues ; however, this might not be the case. “Elasticity ” is the key. Theory suggests that, when moving away from the extremes, a small reduction in an already high VAT rate might lead to such an increase in retail sales that would ultimately increase total VAT proceeds, instead, of reducing it. So far, evidence does not suggest otherwise, at least not when examining the reverse. Findings in a working paper published by the Foundation for Economic and Industrial Research (2015) show that the effectiveness of the tax collection mechanisms worsened during the first years the country entered recession, most probably due to the simultaneous rise in the unemployment and VAT rates. Tax evasion and tax avoidance have gained ground not only in the case of Greece; Portugal, Lithuania, and Spain also suffered similar symptoms after the VAT had increased.

Back to Model 6, the subsidies variable might also look as a valuable tool, as its multiplier value is very high, while its share of GDP is quite small (Fig. 14.4); hence, a small increase in subsidies payable by policymakers could quite easily not derail the budget constraint (Fig. 14.5) and, at the same time, offer a disproportionately positive shock.

Fig. 14.4
figure 4

Subsidies as a share of GDP

Source: National Statistical Service of Greece

Full size image
Fig. 14.5
figure 5

General Government Balance as a share of GDP

Source: National Statistical Service of Greece

Full size image

To illustrate the above conjectures, we compute the scaled estimates of the subsidies and VAT variables in Model 6, to account for the mean differences in their magnitudes (Table 14.6). These scaled estimates clearly illustrate the points made above: a small decrease in VAT would have a disproportionately positive impact on growth, while a small increase in subsidies would aid growth, while at the same time neither of them being responsible for derailing the budget constraint.

Table 14.6 Scaled estimates from Model 6 of Table 14.3
Full size table

5 Policy Implications: “Size” Matters and So Does “Timing”

One might, erroneously, think that the consumption-based model that Greece has been following in the last decades has been considered as an “optimal” one in this chapter, since it has been taken for granted. On the contrary, we have treated the existing productive structure of the Greek economy more of a constraint, hence an exogenous variable that cannot be altered neither easily nor rapidly. Existing literature, policymakers, and many economists tend to agree on a more balanced approach, that is, a growth model that would be also based on investment and exports . We do not deviate from this general approach and do argue in favor of it, at least as a general principle. Nevertheless, the notion of timing is crucial, especially for the Greek economy, as we discuss next.

Policymakers may pursue fiscal or monetary policies in an effort to boost growth. Let us remember the IS-LM model (Fig. 14.6). The IS-LM model shows the relationship between interest rates and real output in goods and services market plus money market. The intersection of the “investment –saving” (IS) and “liquidity preference–money supply” (LM) curves identifies the “general equilibrium” in the economy. By following a looser (tighter) fiscal policy , the IS curve is moving upwards and on the right (downwards and on the left), increasing (decreasing) output and interest rates equilibria, given monetary policy (LM curve). Similarly, a looser (tighter) monetary policy is moving the LM curve downwards and on the right (upwards and on the left), increasing (decreasing) output and reducing (increasing) interest rates equilibria, given fiscal policy (IS curve). Of course, simultaneous changes can take place, targeting an equilibrium point of higher output and stable, or even lower, rates.

Fig. 14.6
figure 6

The IS/LM curves

Full size image

Quoting Lawrence Peter “Yogi” Berra, “in theory there is no difference between theory and practice; in practice there is”; hence, constraints apply and set boundaries to the availability of tools. In the case of Greece, monetary loosening is not an available option. Since the country is a member of the European Monetary Union (EMU), monetary policy is set by the European Central Bank (ECB); hence, it can be treated as an exogenous variable. Consequently, the discussion is moving on to the availability of fiscal tools in order to boost growth. The latest Greek governments have committed themselves in achieving positive primary balances and balancing out their budget constraints. This means that government total revenue (taxes, T) must exceed at least government total spending (G). Looking back at our final model, we highlight three variables related to the fiscal side: subsidies, the VAT , and the general government balance . So, the system of equations we need to solve for is the following: indirect taxes such as the VAT need to be reduced in order to boost growth, subsidies need to increase for the same reason, while respecting the budget constraint .

Schematically:

$$ {\displaystyle \begin{array}{l}\mathrm{PB}\kern0.3em{-}\kern0.28em \mathrm{and}>0,\mathrm{subject}\ \mathrm{to}\kern0.28em \mathrm{VAT}\downarrow +\mathrm{Subsidies}\kern0.1em{-}\Rightarrow \left(T-G\right){-}\kern0.28em \mathrm{and}>0,\\ {}\mathrm{subject}\ \mathrm{to}\kern0.28em \mathrm{VAT}\downarrow +\mathrm{Subsidies}\kern0.05em{-}\kern0.05em\Rightarrow G\downarrow >T\downarrow \end{array}} $$
(14.5)

or, to put it in words, government spending needs to decline to such a level that would also allow for a reduction in taxes. Which is to come first? Budget constraints cannot relax at this stage, meaning that tax revenue cannot be reduced although its effect would be more immediate and would also support growth. Nevertheless, where there is a will, there is a way. Although not all government spending is inelastic, following the last years of government cost cutting, further cuts seem hard to implement. The answer lies in privatizations and the engagement of the private initiative. Via the utilization of a large-scale privatization scheme and public-private partnerships (PPPs), significant sums of public spending could be retracted out of the state budget, allowing for a reduction in government spending and, hence, government size . Most importantly, this would allow for a reduction in taxes against a promise to be implemented at a later stage. Having a credible plan to scale down the public sector , could provide some room by the creditors’ side for even a small and gradual but immediate reduction in tax rates, against the expectation for the completion of a privatization plan. Both lower taxes and less government spending lead to a smaller government size , with the easiest-to-implement solution (tax cuts) being front-loaded time-wise, against the slower process of privatization ; at the same time, both support consumer as well as investment sentiment and, eventually, growth itself.

By no means does this strike at odds with existing literature. The size of a government tends to have a negative effect on growth. Either measured in terms of total revenue or total expenditure , the rationale remains the same: the size of the state’s economic activity and its interference in the economy (excluding laws and regulation) need to shrink. History cannot prove otherwise. In our case study, total revenue and total expenditure as shares of GDP have risen from an average of 40% and 46.7% in the pre-crisis period (1999–2008) to 45% and 54.5% (2009–2016), respectively (Fig. 14.7); growth did not benefit much.

Fig. 14.7
figure 7

Total revenue and total expenditure as a share of GDP

Source: National Statistical Service of Greece

Full size image

This also solves the issue of timing policies appropriately. Contrary to monetary loosening, which is set exogenously, fiscal action can be taken immediately, saving valuable time. Most importantly, in the paradigm of Greece and the European periphery, in general, there is significant interconnection between fiscal and monetary policies. For example, participation to the ECB’s quantitative easing program (QE) requires some form of fiscal prudence, as set in the various bailout programs. This means that some form of monetary loosening—such as moving the LM curve via QE down-right , short-term, and long-term debt relief measures—can finally be elicited.

6 Concluding Remarks

The starting point of this chapter was to examine and, finally, verify or falsify the hypothesis that the government size —as approached by indirect taxes and subsidies—has a negative effect on growth also in the case study of Greece. Our results show that such a negative relationship does exist, confirming a large part of past research. Even when accounting for a series of macroeconomic, social, and political constraints—such as the presence of Troika, the budget constraints , the long-term underlying growth model of Greece, and political instability—the solution towards escaping this downward spiral seems to lie at the grounds of fiscal loosening. Fiscal action arises as the most efficient strategy, targeting the disposable income of the Greek household. Some improvement in the tax collection mechanisms and the fight against tax evasion could allow for lower tax rates, without the need for the government to find additional resources in order to keep total revenue unchanged and sustain a general government surplus. The latter has been a prerequisite under the signed memoranda of understanding and bailout programs. The contribution of this chapter is not that of just presenting another host of statistical models covering the areas of growth, taxation, and fiscal balance; rather, we intended to offer an additional perspective on the policies that need to be examined and implemented from the IMF , European officials, and Greek policymakers in order to avoid another default in the Euro-zone. As a first step, individualities and particularities of the underlying growth model of Greece should be identified, allowing for greater specialization of measures. Emphasis should be given not only to front-loaded reforms but, also, to the reduction in indirect taxes and a lift to consumption , at least for the short term and until the model is transformed to a more productive one. Improvement in the effectiveness of tax collection mechanisms and success against tax evasion will allow for greater equality in income distribution, lower taxation where needed most, at a time when needed most, without any derailments for the budget constraints. Currently, focusing solely on the latter and achieving high primary balances does not seem to rank high with respect to the probability of achieving a sustainable growth rate to enter a long-term development cycle. By no means does this chapter cover all aspects of the relationship between government size and growth. The next step, a natural extension to indirect taxation, is to examine income taxes , although this topic is much more complex due to constant changes in tax rates across time and across governments, indicative of the lack of a long-term policy on taxation by most officials so far. We are currently pursuing this in ongoing research.