1 Introduction

The issue of public debt is at the centre of the Italian policy debate. The debt increase is considered to be the result of unproductive expenses and anaemic growth entangling the country, especially after the negative shock of the financial crisis, in a self-fulfilling process of declining output, increasing interest rates and unsustainable public finances.

Until the end of the 1960s, the sustained rates of growth together with high inflation eroding real returns limited the concerns about the sustainability of public debt. However, at the beginning of the 1970s, with the fall of the Bretton Woods agreements, the entry into the European Monetary System (EMS) and the divorce of the Bank of Italy from the Treasury that set up the separation between fiscal and monetary policy, real returns on debt started to outstrip GDP growth rates, thus giving rise to an increasing trend in the debt–GDP ratio (Magazzino and Intralighi, 2015). This relation was reinforced after the adoption of the common currency and further strengthened by the financial crisis, pushing the dynamics of public debt down a dangerous path. The centralised monetary policy, together with the restrictive fiscal measures implemented at the national level to meet European parameters, made Italy an emblematic case of unsuccessful fiscal consolidation and a threat to the survival of the European Monetary Union (Bonasia & Canale, 2015; Canale et al., 2021).

Together with the increase in debt, a decline in the labour share of income occurred. Indeed, as a result of a gradual change in production relations (Piketty, 2014, 2015), this phenomenon affected most of the advanced economies and was related to the long-run dynamics of functional income distribution, from which a decrease in the amount of national income going to labour is observable, instead moving in favour of an increase in profits and financial rents (Stockhammer & Onaran, 2012; IMF, 2007; ILO, 2011; Stiglitz, 2016; for the Eurozone Arpaia et al., 2009).

In Italy, the wage share—after a stepwise increase after the Second World War and a rather stable trend during the 1960s—started to decline in the middle of the 1970s, registering the reduced role of workers inside the production relations. These changes occurred within the framework of an ever-increasing willingness of the country to implement the shared policy prescriptions of reducing inflation and increasing market flexibility.

Working in this context, the aim of the present paper is to empirically investigate the relation between public debt and the labour component of functional income distribution in Italy in the period ranging from 1960 to 2019, with the objective of detecting the eventual presence of a relationship between the increasing debt burden and decreasing wage share of national income. The underlying hypothesis is that an increasing amount of debt implies a redistribution of resources in favour of wealthy government debt holders in the form of capital revenues or interest payments (Michl, 2006; You and Dutt, 1996; Salti, 2015).

However, because debt is an instrument for financing public expenditure, the negative relationship with wage share poses the question of the effect of debt-financed expansionary fiscal policy on labour income. Therefore, to deepen this analysis, empirical estimates distinguishing between primary public balance and interest payments are implemented to individually evaluate the effects of these two components feeding public debt on aggregate wages.

The objective is not to investigate the determinants of wage share in Italy (Torrini, 2015; Gabbuti, 2018), but rather to examine if the path of debt and the other components of public accounts are entangled in a long run relationship with the share of GDP going to labour. It cannot be denied that increasing debt—in the presence of constraints—reduces resources to be spent in increasing employment and implementing active labour market policies (Bertola, 2010), nor that the direction of causality could reverse as a decreasing wage share reduces aggregate demand and growth, therefore giving rise to the unsustainability of the debt (Stockhammer & Onaran, 2012, 2013). However, this multifaceted point of view suggests that the relation between debt and functional income distribution should be examined in light of the efficacy of fiscal policy.

Our data sample covers a rather long period, during which a wide range of economic and sociological transformations occurred in Italy; it spans from 1960 to 2019, allowing for an application of a cointegration methodology—the dynamic ordinary least square (DOLS)—that can evaluate the existence of a connection, even in the presence of a reduced number of explanatory variables. The dependent variable is the wage share calculated, accounting for both employees and autonomous workers. The main explanatory variable is public debt, subsequently disaggregated into debt service, which measures the amount of interest to be paid annually, and the primary balance, which is given by the difference between government revenues and expenditure, translating into a change in indebtedness. These explanatory variables are entered separately into the empirical model to evaluate the individual effect of each component of public accounts on the functional income distribution.

The empirical analysis generates outcomes showing a negative relationship between public debt and wage share. However, when disentangling the components feeding of debt into primary balance and interest payments, the results appear to be more articulated. A negative (positive) primary balance or excess in government expenditure with respect to revenues increases (decreases) the wage share, hence signalling that an increase in debt generated by expansionary fiscal policy is not negative for the wage share. When considering the debt service, instead, the estimates reveal that it is the financial rent that is detrimental to the labour component in the functional income distribution.

With the aim of interpreting the results in light of the institutional changes that occurred in Italy during the period under examination, as a robustness check, the sample has been divided into three time spans: 1) 1960–1980, during which monetary policy was subordinated to fiscal policy activity and there were no concerns about public debt management; 2) 1981–2000, when the separation of fiscal and monetary policy achieved with the divorce of the Bank of Italy from the Treasury—implemented according to the evolution of the economic theory occurred from the beginning of the 1970s—transformed the issue of policy coordination and paved the way to the process of the financialisation of sovereign debt management (Fastenrath et al., 2017). This process was further reinforced in the subsequent period of 3) 2001–2019, when adherence to the common currency added institutional constraints to the connections between the government and private financial markets (Mosley, 2004). Despite the three subsamples contain a limited number of observations, they help support the result obtained for the entire considered period. It was when interest rates became a necessary remuneration for the allocation of public bonds to private markets that the negative connection between public debt and share of GDP devoted to labour was triggered.

The conclusions reached in the present paper are comparable with those in Canale and Liotti (2021), who analysed 11 Eurozone countries from 1999 to 2019 a time when, however, the institutional context remained unchanged. This focus on the Italian case allows us to further investigate the connection and, by taking into account a longer period, deepen the analysis regarding the role of financial markets in functional income distribution. As far as we know, this is the first attempt to empirically investigate the link of public debt and its components with wage share in Italy.

The rest of the paper is organised as follows: the next section provides a brief recall of the theoretical background behind the relation between public debt and expenditure on income distribution, here taking into account the institutional transformations that occurred in advanced economies and the Eurozone. Section 3 briefly recalls the historical and economic context in Italy, taking into account the changes that shaped its policy strategies. Section 4 contains the empirical analysis and is divided into three subsect. 4.1 describes the data and methodology, Subsects. 4.2 and 4.3 present the results about the impact of the entire public debt and of the other two public finance components on wage share, respectively, while also taking into account different time spans. Finally, Sect. 5 draws conclusions and derives policy implications.

2 Public debt, fiscal policy and labour income in the literature

The financialisation of economics is widely acknowledged as affecting income shares by moving resources from low and middle incomes towards higher incomes (Kohler et al., 2019). This process is amplified by the domination of policy under the concerns that the financial sector gives rise to a long-run decrease in the wage share in favour of profits and financial rents (Lapavitsas, 2013). Furthermore, the decrease in wage share in the presence of a noncooperative monetary policy reduces aggregate demand and growth, leading to problems with the unsustainability of private and public debt (Stockhammer & Onaran, 2012, 2013).

The financialisation process began with the paradigm shift that affected economic disciplines in the 1970s. This shift put the market at the centre stage of economic investigations and progressively reduced the role of economic policy in defining macroeconomic dynamics. The result was a separation between monetary policy, which was assigned the role of reaching the objective of price stability, and fiscal policy, whose task was limited to guaranteeing the correct functioning of the market (Canale & Mirdala, 2019). This policy framework was a cornerstone of ‘financial capitalism’, triggering governments to reduce taxes with the aim of supporting private activity and increasing public debt to preserve the social security system (Preunkert, 2017), hence giving rise to an ever-increasing public debt. Governments in advanced economies were forced to turn to the private market to finance fiscal policy, which gave rise to both common and country-specific dynamics, depending on the general macroeconomic conditions (Fastenrath et al., 2017). The government–financial market relation was further consolidated with the birth of the Eurozone, whose institutional settings made the dependence of fiscal policy on private markets binding. However, although it sometimes happens that balanced budgets guarantee better financing conditions and a counterpart for fiscal consolidation (Mosley, 2004), it often occurs, instead, that self-fulfilling processes of increasing interest rates and decreasing growth are triggered (Canale et al., 2021). Government security markets—for better or for worse—transmit sovereign risk to the holders of government debt, affect the balance sheet of financial institutions and interact awkwardly with macroeconomic policies (Hoogduin et al., 2011).

In fact, debt is the instrument used to finance an additional public deficit and increases as a result of both net expenditure and interest to be paid on it. Its interaction with relevant macroeconomic variables convinced several authors that whatever its origin, countries with a low level of debt experience better growth performance. The need for sound public finance is linked to the long-term macroeconomic impact of unsustainable expenses and the supposed existence of a threshold above which outstanding government debt is associated with decreasing growth rates (Reinhart & Rogoff, 2010; Checherita and Rother, 2012). This relation may occur even in the short run because the increase in sovereign long-term bond yields can negatively affect private savings, public investments and total factor productivity (Oguro & Sato, 2014). The higher the debt, the higher the amount of interest to be paid (Berti et al., 2013) and the lower the resources to allocate in sustaining the growth process.

Moreover, the negative impact of public debt on growth also leads to a reduction in the public resources available to smooth the differences among individuals (Arawatari & Ono, 2017; Giambattista & Pennings, 2017; Preunkert, 2017). Debt sustainability implies that a fraction of tax revenues from workers must be used to pay interest to the government debt that is not homogeneously owed across the population (Michl, 2006). Public debt, therefore, also distributes wealth unequally, and this effect is mainly driven by after-tax wages, even in the presence of a highly progressive tax system or when real interest rates are low (Chatzouz, 2014). This occurs all the more when the amount of debt held abroad is relatively high (Salti, 2015). The negative effects operate when after-tax labour income rises more slowly than interest payments (You and Dutt, 1996). The consequence is that wage-share dynamics are often associated with similar trends in income inequality (Gabbuti, 2018). All of these arguments contribute to the widespread belief that public debt is a negative for income distribution.

However, following the Keynesian literature, public debt may contribute to finance aggregate demand and, via the multiplier equation, to employment and aggregate equilibrium income. It cannot be excluded, therefore, that debt-financed deficit spending for countercyclical purposes could be beneficial for employment and, thus, for the sum of wages paid in the country (Michl, 2006). Based on these models, when growth is determined by aggregate demand rather than by the supply of resources and income is distributed between workers who earn wages and capitalists who receive profit and interest income, the outcomes in terms of wage shares depend on the effect exerted by debt-financed fiscal policy on economic growth (Dutt, 1996).

In this vein, many authors have underlined the importance of debt-financed fiscal expansion in reaching output and employment target levels, especially during declining macroeconomic conditions (Blanchard, 2019; Buchner, 2020). When the increase of public expenditure appears to be necessary to sustain employment and growth (Blanchard & Leigh, 2013; Christiano et al., 2011; DeLong & Summers, 2012; Krugman, 2013), especially under a fixed exchange rate regime or a common currency, as in the case of the Eurozone (Ilzetzki et al., 2013), the increase in public debt has positive effects on employment and aggregate equilibrium income.

These conclusions suggest that an increase of public debt can also have positive effects on employment and, therefore, on wage share, signalling that it could be useful to separate primary public deficit from debt service or, in other words, to investigate the connection through the lens of fiscal policy effectiveness.

3 Public debt and wage share in Italy: a historical perspective

Between the end of the Second World War and present day, Italy has undergone numerous changes that have affected both its production structure and management of economic policy. The 1950s started a period of sustained growth driven by increasing employment and wages. The social climate of this period was characterised by increasing levels of workers’ political and contractual strength, which peaked in the 1970s, bringing about improvements in general living conditions (Pugliese, 2015). On the side of economic policy, the fiscal authorities were committed to supporting aggregate demand, while monetary policy, in a context of sustained growth and low inflation, had the main task of financing public deficit and containing the increase in interest rates. Regarding the impact of the international context, the adherence to the Bretton Woods agreements and limited capital mobility greatly contributed to the high degree of fiscal policy effectiveness.

After the fall of the Bretton Woods agreements, the national and international scenario completely changed. An increase in employment in high-productivity sectors together with an increase in the average rate of unemployment occurred. An increase in wages in the first half of the 1970s was attributed to the fight between insiders and outsiders. The resulting increase of prices together with higher inflation from the international crisis caused a shift in the economic policy strategy (see the previous paragraph). The Bank of Italy began gaining independence from fiscal policy and became less accommodative (Tabellini, 1988). In 1979, Italy adhered to the exchange rate mechanism (ERM) of the EMS, and in 1981, monetary policy independence from the fiscal authority was reached with the ‘divorce’ of the Bank of Italy from the Treasury (Favero & Spinelli, 1999), giving rise in Italy to the phenomenon of the financialisation of sovereign debt management (Fastenrath et al., 2017). The latter implied that the central bank was no longer obliged to be a residual buyer at the government’s bonds auctions (Daniele et al., 2017). Public debt had to be sold on the private market unless the Bank of Italy decided otherwise, hence giving rise to complex issues of coordination between fiscal and monetary policy. From then on, the rate of employment suffered ups and downs, while wages underwent progressive reductions, generating a decline in the share of income going to labour.

At the beginning of the 1990s, the fall of the Berlin Wall and German reunification gave rise to capital outflows that forced the Bank of Italy to raise interest rates in an attempt to preserve the exchange rate agreements and the balance of payments equilibrium. The increase in interest rates caused negative effects on internal equilibrium and a decline in output and employment. This situation triggered national policy authorities to abandon exchange rate agreements and devalue the Italian lira in 1992. In the same year, the Maastricht treaty was signed, defining the rules to adhere to the monetary union. These rules—especially the one regarding debt/GDP ratio—were often respected through the use of financial innovation that further exposed Italy to the threat of financial markets (Lane, 2016).

From 1999 to the present day, the European Central Bank (ECB) has been the monetary policy authority for all countries belonging to the European Monetary Union (EMU), while fiscal policy, at least until the recent COVID-19 pandemic, was nationally managed under severe spending constraint criteria. In 2001, the euro became the only currency in circulation. For members of the EMU, the relation between fiscal and monetary policy turned completely upside down compared with the previous 50 years as fiscal policy was subordinated to the objective of price and interest rate stability. This stability would have contributed to the convergence across Eurozone countries. An additional instrument for convergence was represented by the flexibility in the labour market as a tool capable of absorbing asymmetric shocks, here in line with the theory of the optimal currency area (Mundell, 1961). However, the different market perceptions of sovereign creditworthiness caused a persistent segmentation of Euro area capital markets, producing asymmetric effects on public debt. Virtuous countries (like Germany) benefited from these asymmetries, while ‘vicious’ ones, like Italy, despite often having a positive primary balance (see Fig. 2), remained on the ‘periphery’ of the Eurozone (van Riet, 2021). Since the divorce between the Bank of Italy and the Treasury, the public debt in Italy, despite the restrictive fiscal measures applied to comply with the Maastricht criteria for entering the monetary union first and to respect the stability and growth pact subsequently, has, on average, never stopped growing, entangling Italy in a self-fulfilling spiral of increasing interest rates and anaemic growth.

4 Empirical analysis

4.1 Data and methodology

The empirical analysis aims to verify the hypothesis according to which the increase in public debt has triggered a fall in the wage share because of the absorption of the product of work in Italy from 1960 to 2019. This relationship is strictly connected to the circumstance that an increase in debt increases the interest shares of GDP paid to the holders of public debt. However, if used to finance public expenditure by sustaining economic growth and the amount of public services, a fraction of public debt returns as an increase in the wage share. The hypotheses formulated in the present paper follow the idea that these two effects do not offset each other because the progressive financialisation of the state budget transfers the product from workers to rentiers, thus reducing the portion of the GDP assigned to labour.

Data about wage share are collected from AMECO (https://ec.europa.eu/economy_finance/ameco/user/serie/SelectSerie.cfm). The dependent variable in the reference empirical model is the wage share, which is calculated as follows:

$${\text{WS}} = \frac{{{\text{Compensation}}\;{\text{of}}\;{\text{employees }}}}{{{\text{GDP}}}} \ast \frac{{{\text{number}}\;{\text{of}}\;{\text{self}} - {\text{employed}}}}{{{\text{number}}\;{\text{of}}\;{\text{employees}}}}$$

This equation considers both wages and salaries earned by employees and the labour income of the self-employed, who are not identified separately in the National Accounts system. This procedure is necessary, especially in advanced economies, where the self-employed represent a relevant share of the entire workforce and data about their wages are not available. Scaling up the average compensation of employees following this methodology is a good approximation for including self-employed labour income to the extent that they receive the same wages as employees (Gollin, 2002). This expression, however, may underestimate the value of the wage share when self-employed earnings are low, leading to an overestimation if their earnings are high (Arpaia et al., 2009). Nonetheless, this is the way the European Commission calculates the share of wages on GDP, allowing us to consider the average income gained through every kind of labour. The three main explanatory variables are 1) the debt–GDP ratio or the general government consolidated gross debt, here expressed as a share of GDP; 2) the interest payments on public debt, which is also calculated as a share of GDP; and 3) the primary balance/GDP ratio, which is obtained by using the original net lending (+) or net borrowing (-) values, excluding interest payments, for the general government. All these public account variables are retrieved from the IMF database ‘Public Finances in Modern History’, which is available at https://www.imf.org/external/datamapper/datasets/FPP. Per capita GDP, which is considered a proxy of productivity and living standards in the country, is added as an additional control variable and has been retrieved from the AMECO database.

A first look at wage share and public finance variables provides insights into the phenomenon on which our analysis is focused. The variables in the figures are presented in their actual values and time trend, obtained through the Hodrick–Prescott (HP) filter (1997), which shows the long-run dynamics deprived from specific cyclical conditions. Figure 1 shows the wage share (right axis) and public debt (left axis) in Italy from 1960 to 2019.

Fig. 1
figure 1

Source: Own elaboration on AMECO and IMF databases

Wage share (right axis) and public debt (left axis) in Italy: Actual values and time trends (1960–2019).

Full size image

As shown, from 1960 to 2019, public debt continuously increased, ranging from a value of about 30% in 1960 to a value of more than 136% in 2019. Just from the mid-1990s to the years immediately before the 2007 financial crisis, Italian public debt decreased but started to increase again in the turbulent subsequent years. Contrasting dynamics are illustrated by the wage share: starting from a value of about 68% in the 1960s, it declined to a level of about 52% in 2019. Contrary to public debt, a slight inversion of the decreasing path of wage share can be seen both in the period ranging from 1969 to 1975 and from 2000 to 2009, when the convergence process inside the monetary union seemed to be successful and before the financial crisis in the Eurozone turned into a sovereign bond crisis.

Figure 2 compares the wage share (left axis) and primary public balance (right axis) time dynamics (both actual and filtered through the HP filter) for the period under investigation. Public expenses exceeded revenues more or less from 1965 until 1990 (considering the filtered values). The wage share started declining in the period when the primary public deficit began dropping, reaching the lowest level around the period in which the primary balance reaches the highest level. Both became rather stable in the past 15 years.

Fig. 2
figure 2

Source: Own elaboration on AMECO and IMF databases

Wage share (left axis) and primary public balance (right axis) in Italy: Actual values and time trends (1960–2019).

Full size image

Finally, Fig. 3 compares the wage share with the debt service as a share of the GDP. As in the case of primary public balance, interest payments started to increase contemporaneously with the decrease in wage share. From 2000 to 1019, were rather stable.

Fig. 3
figure 3

Source: Own elaboration on AMECO and IMF databases

Wage share (left axis) and Interests on public debt (right axis) in Italy: Actual values and time trends (1960–2019).

Full size image

From a descriptive point of view, the dynamics of debt, primary public balance and debt service have followed the proposed interpretation, despite suggesting that the connection might change in different subperiods. Therefore, the next step is to search for a robust connection by means of an adequate empirical analysis.

We employ a dynamic ordinary last square (DOLS) technique (Stock and Watson, 1993) because a baseline static ordinary least square (OLS) approach would release misleading results if the variables were characterised by path dependence. To verify whether this would be the case, we run different unit roots tests (augmented Dickey–Fuller, Dickey–Fuller–GLS and Phillips–Peron) on variables at levels and at first differences: the results suggest exploiting this route. Therefore, we apply the DOLS model—suitable for use in small samples—which requires variables to have a unit root (nonstationary) in their level and be stationary when considered in their first differences, that is, variables that are integrated of order one (I(1)). This DOLS model also corrects for possible simultaneity bias among the regressors, providing an estimation of the long-run equilibria. This approach accounts for the possibility of persistence in the level of both dependent and explanatory variables occurring when years with positive changes in one variable are likely to be followed by further increases in the same variable and vice versa. To this end, the long-run regression is augmented by lead and lagged differences of the explanatory variables to control for potential endogenous feedback (Saikkonen, 1991) and serial correlation (Stock and Watson, 1993). For an application of this methodology, see Masih and Masih (1996). Finally, the estimated equations need to be cointegrated to validate the results and prove the existence of a long run relationship among the variables.

When considering the impact of public debt as a whole, the empirical model is as follows:

$$WS_t = \alpha + \beta PD_t + \sum_{k = - p}^p {\delta_k \Delta WS_{t - k} } + \sum_{k = - p}^p {\varphi_k \Delta PD_{t - k} } + \varepsilon_t$$
(1)

In Eq. (1), \(WS_t\) indicates the wage share at time t and \(PD_t\) the public debt in the same year; \(\Delta WS_{t - k}\) and \(\Delta PD_{t - k}\) refer to changes of the respective variables, with \(p\) indicating the number of lags and leads. The coefficient \(\beta\) is the DOLS parameter to be estimated to assess the magnitude of the relationship between public debt and wage share. The term α represents factors not considered in the estimates, while ε is the error term.

When separately including the two components fuelling public debt, we remove the variable public debt (PD) and include PB for the primary public balance and INT for the interest payments, respectively, both considered as a share of GDP, as in the following equation:

$$WS_t = \alpha + \beta_1 PB_t + \beta_2 INT_t + \sum_{k = - p}^p {\gamma_k \Delta WS_{t - k} } + \sum_{k = - p}^p {\delta_{1k} \Delta PB_{t - k} } + \sum_{k = - p}^p {\delta_{2k} \Delta INT_{t - k} } + \varepsilon_t$$
(2)

In Eq. (2), \(WS_t\) is the usual wage share at time t, and \(PB_t\) is the primary public balance as a share of GDP. A positive primary balance indicates that revenues exceed expenditure, while a negative primary balance indicates that the government is spending more than what it receives to buy goods and services for the Italian population. Finally,\(INT_t\) represents interest payments as a share of GDP. The empirical model in Eq. (2) allows for investigating the contemporaneous effect of public balance and debt service on wage share, therefore allowing us to take into account the possible interactions between interest rates and public balance through GDP.

4.2 Estimation results

The first step of the empirical analysis refers to the investigation of the properties of the dataset. Indeed, as asserted in the methodology, the preliminary conditions of nonstationarity at level, stationarity at first differences and the cointegration of the variables are essential to prove the goodness of the DOLS methodology chosen for the econometric analysis. Test results are presented in Table 1. The variables used in the tests and estimates are those derived from the application of the HP filter (1997) to the original data. This transformation—as suggested by the graphs in Figs. 1, 2 and 3—helps smooth the deviation from the long-run trend and to minimise the importance of several peculiar economic or institutional situations across the long-term relation of dependence.

Table 1 Unit root and cointegration tests on wage share (WS) and public debt (PD)
Full size table

Panel a in Table 1 presents unit root test results. The augmented Dickey–Fuller (ADF) test is the Dicky–Fuller test (Dickey & Fuller, 1979), including lagged differences and improved using the MacKinnon (1994) procedure in estimating the critical values. The DF-GLS test runs a unit root test on the variables’ interpolated values and is acknowledged as the most powerful one.Footnote 1 The third is the Phillips–Peron test (Phillips-Peron, 1988), which uses Newey-West (1987) standard errors to account for serial correlation. The three tests are performed, including a ‘constant’ and a ‘constant and trend’ option to account for the possibility that the feature of the variables changes with specific options. If we exclude the ‘constant’ option for the DF-GLS test for PD, the three methodologies used to perform the test accept the null hypothesis of the presence of the unit root when WS, PD, PB and INT are considered at their level, hence supporting the nonstationarity of variables at levels. When variables are considered at first differences, almost all the tests reject the null hypothesis of the presence of a unit root. The result of the Phillips–Perron test for ΔPB and ΔINT must be excluded. However, because two out of three tests reveal stationarity at first differences, the variables in in the two equations to be estimated can be considered integrated of order one (I (1)). Panel b in Table 1 reports results from the Granger causality test (Granger, 1969), assessing, in terms of time, the causality direction proposed in the previous paragraphs that the debt comes first in respect to WS. It rejects the null hypothesis that PD does not Granger cause WS while accepting that WS does not Granger cause PD. Once the properties of the variables have been set to assess the validity of the DOLS methodology, it is possible to move on to the results of the econometric estimates.

Table 2 presents the DOLS estimation results of Eq. (1). Besides the main explanatory variable, Model I includes a constant term and time trend. The coefficients of lagged and lead changes in variables are not reported because their inclusion is only aimed at correcting for endogeneity biases and releasing reliable estimates.Footnote 2 According to the estimation results, an increase in PD negatively affects WS (β = − 0.195***). In particular, an increase in the debt–GDP ratio by 1% causes a reduction of the wage–GDP ratio of about 0.195 percentage points. Both the constant and time trend terms are significant, implying the presence of specific institutional factors and further conditions causing the decline of the GDP’s labour share. The explanatory power of the estimated model of dependence is high at R2 = 0.98. This result depends on the inclusion of the explanatory variables of the lead and lagged values of the dependent variable.

Table 2 Wage share and public debt: DOLS estimation results (Eq. 1): 1960–2019
Full size table

Our results reveal that the increase in PD caused a decline of the WS in Italy, thus confirming the hypothesis that the financialisation of the state budget reduces the portion of national product assigned to labour.

This is the same result as when adding per capita GDP as an additional control variable, as shown by Model II in Table 2. A comparison between the two models’ coefficients reveals that the introduction of the average per capita income in the country does not alter the sign but instead increases the size of the relation of dependence. It is worth noting that per capita GDP can also be considered a proxy of productivity: the negative sign of the coefficient reveals that, on average, when productivity grows (decreases), the WS decreases (increases), signalling that, during expansionary phases, not all output growth can be attributed to labour share. The opposite occurs during contractionary periods.

Table 2 also presents the cointegration tests. The Engle–Granger test (Engle & Granger, 1987) provides controversial results: in the case of Model I, the t-statistics rejects the null hypothesis of no cointegration, while the z-statistics accepts it. This is probably because of the existence of some instability in the parameters as a result of institutional changes that occurred in 1981 and in 2000. Therefore, we have added a further check through the implementation of the Hansen (1982) instability test; for both Models I and II, it accepts the null hypothesis of cointegration.

Table 3 presents the results of the estimation of Eq. 2, which connects both primary public balance and interest on PD with labour income share. In both Models I and II, the coefficients suggest that a decrease in public deficit and a higher debt service decrease WS.

Table 3 Wage share, primary public balance and interests on public debt: DOLS estimation results (Eq. 2): 1960–2019
Full size table

However, when including GDP per capita, the coefficient of interest payments remains almost the same, while the one of the primary public balance decreases. This can be because of the different macroeconomic conditions defining the degree of efficacy of fiscal policy on employment and equilibrium income. The Engel–Granger test reveals stationarity both for Models I and II, while the Hansen instability test presents some doubts for stationarity in Model II.

4.3 Robustness checks

The estimation results presented in Sect. 4.2 provide support for the observable negative connection between PD and WS in Italy from 1960 to 2019. In the same time span, expansionary fiscal policy may have exerted negative effects on the WS, while debt service appears to have had a negative effect. The institutional changes that occurred in Italy in 1981 with the ‘divorce’ between the Bank of Italy and the Treasury, as well as with the adoption of the common currency entered into force in 2001, suggest that the sample should be divided into three subsamples to check whether the same relationships are still valid under different mechanisms of connections between monetary and fiscal policy and different contexts regarding the ‘financialisation’ of governments. These events may well be considered a structural break in the series, which may have caused changes in the effect of PD and other public finance variables on WS. To verify this hypothesis, we have performed the ‘Chow-test’ (Chow, 1960) on the entire sample by inserting two simple dummies and two interaction dummies for the main explanatory variable in 1981 and 2001. The F test (1423.82***) rejects the null hypothesis that the four coefficients are equal to zero. Therefore, we again estimate the baseline models of Eq. (1) and Eq. (2), dividing the sample into three time spans, 1960–1980, 1981–2000 and 2001–2019, respectively. This kind of robustness check should be taken cautiously because the number of observations is few. This exercise merely represents a support to the connection presented in previous paragraphs about the financialisation of government public finance and functional income distribution.

Furthermore, because of the limited number of observations, we have had to, differently from Eq. 2, introduce primary public balance and interests’ payments one at a time, limiting the possibility of comparing the results obtained in Table 3.

The reduced number of observations assigns a limited validity to the results presented in Table 4. Nevertheless, the positive value of the coefficient β (0.264***) for the 1960–80 period signals that the presence of cooperative behaviour between fiscal and monetary policy gave rise to a PD dynamic that positively affected WS. Table 4 also reveals that the negative connection for the whole period is confirmed for both the 1981–2000 and 2000–2019 intervals. This reversal of the connection is accompanied by different links with the proxy of productivity (per capita GDP), which seems to positively contribute to WS in the first interval (1960–81: 2.264***) while negatively in the third one (2001–2019: − 1.137***).

Table 4 Wage share and public debt 1960–1980; 1981–2000; 2001–2019: DOLS estimation results (Eq. 1)
Full size table

The same empirical exercise has been implemented using the primary public balance and debt service. Different from Eq. (2), the two components have been added one at a time because of the reduced number of observations: to make the model work, we have chosen to leave the per capita income in the equation to be estimated to account for the general macroeconomic conditions and insert the public budget variables separately.

Table 5 reports the estimation results of the connection between primary public balance and WS for the same time spans considered in Table 4. The evidence supports the hypothesis that a rise in expenditure over revenues, if not destined to pay interest, increases the share of GDP going to labour. The intercept and trend components’ coefficients are not always significant, signalling controversial dynamics under different institutional settings. It is worth noting that the coefficient of the primary public balance decreases in the time spans 1981–2000 and 2001–2019, signalling that the efficacy of expansionary (restrictive) fiscal policy should be evaluated by taking into account the general macroeconomic conditions.

Table 5 Wage share and primary public balance 1960–1980; 1981–2000; 2001–2019: DOLS estimation results
Full size table

The last relationship to be tested is the one aiming to detect the effect that interest to be paid on PD may exert on the WS. The results are presented in Table 6.

Table 6 Wage share and interests on public debt 1960–1980; 1981–2000; 2,001,019: DOLS estimation results
Full size table

They indicate that the amount of public balance destined to repay interest on PD always decreases WS, especially during the last subperiod when the financialisation of the state budget was absorbed inside institutional rules (1960–1980: − 0.777**; 1981–2000: − 0.1706***; 2001–2019: − 1.134***), supporting the hypothesis of a negative impact of the progressive financialisation of the state budget and share of the GDP that workers receive. The usual considerations about intercept, trend and R2 apply, as in previous estimation results. It is noteworthy that in none of the cases has the cointegration test been implemented because of the very few observations available.

These results signal that PD is not always an evil for labour share but rather when it is supposed to finance additional public expenditure to sustain public investments, services and consumption improves WS. On the contrary, when it is issued to repay the interest accrued on PD issued in the past, it subtracts a share of GDP to workers in favour of rentiers’ income.

5 Conclusion

The present paper investigates the relationship between wage share and public debt in Italy from 1960 to 2019. The main findings reveal that an increase in public indebtedness reduces the portion of income going to labour, thus worsening, all other things being equal, workers’ aggregate income conditions. However, when evaluated separately, the two components affecting the rise in public debt reveal a more articulated link. In fact, while the debt service appears to be detrimental to wage share, the deficit component improves it. Therefore, it is not public debt per se that moves the national GDP in favour of rentiers. An expansionary fiscal policy financed through debt, if not accompanied by an increase in interest rates, can increase the wage share. Extending the reasoning to the polar case of restrictive fiscal policies, this empirical analysis also suggests that the strategy of reducing debt through a reduction of primary deficit worsens the labour income conditions.

These results are relevant for Italy, a country that, starting from 1991, has always had—if we exclude the year 2009—a positive primary balance. The excess of current revenues over current expenditures was required to comply with fiscal rules and induce a reduction in public debt. However, because of the increase in debt service, public debt increased, negatively affecting Italian wage share, which declined both for the increase in primary balance and increase in interest rates. The separation between the Bank of Italy and the Treasury, which was realised to comply with the changing paradigm in economics, brought to the financialisation of the sovereign debt market and made the government dependent on private markets’ sentiments. The resulting dynamic in the debt service in the period considered has been affecting negatively the share of national GDP going to labour. This result appears to be more relevant if examined inside the policy framework of the Eurozone, to which Italy belongs and in which monetary policy is centralised and fiscal policy is left to the management of individual states. This institutional framework exposes the management of fiscal policies to financial drivers and increases asymmetries between northern and southern countries, threatening, at the end, the very existence of the common currency. Therefore, public debt should be managed by accounting for the reciprocal effects of fiscal and monetary policy and considering the introduction of a common safe asset to preserve labour income.