Sustainability of Public Social Spending: Asymmetric Effects and Financialization

Abstract

We investigate the sustainability of the asymmetric public social spending (PSS)–financialization relationship in the Eurozone over the period of 1995q1–2023q4. We follow the theoretical endogenous nexus of PSS with the financial fragility hypothesis (FFH) and finance-led growth regime; the nonlinear autoregressive distributed lag (NARDL) model and cointegration are applied for this purpose. The analysis suggests the following: (1) The selected determinants of the three stages of the FFH affect dependent PSS asymmetrically in the long run (as well as in the short run, sometimes); meanwhile, more often than not, significantly larger effects tended to be negative changes. (2) The asymmetric shocks of explanatories gently increase PSS in many cases but also decrease it strongly in others. (3) The “automatic stabilizer” role of PSS is proven, whereas the contrary is not rejected; that is, PSS was also used as a “counter-automatic stabilizer” tool. (4) This leads to “ratchet effects”; the direction of these effects is unclear, but it seems to decline over time. (5) The financialization of the PSS phenomenon is revealed and discussed using relevant economic interpretations for certain determinants, such as credits to nonfinancial corporations, relative profitability, domestic borrowing from abroad, and the snowball effect; all of these have long-term effects on PSS, comprising negative changes, with asymmetric dynamics towards a new equilibrium at a horizon of between 4 and 16 quarters. Policy implications are related to the sustainability of PSS through the control of the economy’s financialization. We contribute to the literature by analyzing—for the first time as far as we know—the endogenous nonlinear long- and short-run dynamics of PSS based on a comprehensive model of the FFH and the finance-led growth regime.

1. Introduction

There is a great difference between the concepts of economic growth and development. Solow’s [1] seminal growth model refers to the mechanisms of quantitative increases in the real output of an economy; meanwhile, development, in addition to growth, refers mainly to the quality of life of citizens, not only as consumers but primarily as people [2]. The 17 UN Sustainable Development Goals (SDGs) have set the tone for current global progress in the sense of promoting prosperity while protecting the planet. Thus, social cohesion, with a reduction not only in income inequalities but also in the pursuit of equal opportunities for all, is a key feature of development. Another feature is the political goal of controlling the failures of either the market or the government, such as “democracy”—alongside freedom, equality, and justice—that presupposes the greatest possible participation of citizens in social and political decision-making. Freedom of speech and expression, with an emphasis on the freedom of the mass media, is another objective of economic development. In any case, health, education, security, and the public, social, or common good are also considered essential elements for an economy to be described as “developed”.

Therefore, the concept of economic development presupposes social cohesion—an important means of achieving the social policy of a given country. Effective social security policy and the presence of an effective welfare state are, therefore, not only scientifically interesting but also politically useful for reducing a number of household risks and needs. This is more necessary for Western societies, as Beck [3] originally explained. The individualization of social inequality and the commercialization of any risk, with the structural upheaval in the “employment market” instead of the labor one, increases the need for modern social policies. For instance, this has been shown in the surprisingly long-term dynamism of East Asian capitalisms, which build upon a proactive state policy [4,5].

In economics, a crucial transformation of contemporary Western economies concerns the institution of the “financial system”, which has prevailed over the entire economy, contributing to the “financialization” phenomenon: “The increasing roles of financial motives, financial markets, financial actors, and financial institutions, in the operation of the domestic and international economies” [6] stands as the most acceptable definition. Financialization appears in four versions: (1) as another form of capital accumulation, the financial one, which drives competition mainly in the financial markets; (2) as shareholder value, which radically changes the management of companies, since managers and shareholders tend to have the same interests; (3) as everyday finance, where the citizens should become investors; and (4) as the change in the calculating practices of the credit rating agencies dominated by the structured finance ratings approach [7,8]. However, it should be noted that the root of financialization is already known from Keynes since 1933 [9] and it concerns the non-neutrality of money in the economy. Given that we believe in the unity of economics like many others (e.g., [10]), we use the two macro-theories of financialization to explain the possibilities of sustainable social policy. These are the financial fragility hypothesis (FFH), especially in the Eurozone case [11,12,13], and the finance-led growth regime [14].

Moreover, the change in the nature and functioning of governments dominated by the international capital market (globalization) further limits social policy because “public interventions are needed not at the national but at the international level (nowadays) and captured by the massive power of financial sector lobbying” [15]. Thus, the sustainability of social policy necessarily depends on the attractiveness of economies to international investors [16]. In total, in Western economies, the public finances that include social security are still determined by the “logic” of the global financial system.

In addition, although recent studies have examined the role of public social spending (PSS) as an automatic stabilizer or even as something with ratchet effects, these are not grounded in a comprehensive theoretical body [17,18]. Thus, this study tries to fill this gap through the literature on financialization [19,20]. The theoretical background refers to both the financial fragility hypothesis (FFH) (which includes Minsky’s [11,21,22,23,24,25] financial instability hypothesis (FIH)—as it has been extended by Arestin and Glickman [12] for open economies (eFIH)—and the Eurozone fragility hypothesis (since the sample country (Greece) belongs to the core of the European Union (EU) that of the Eurozone or EMU) [13,26,27,28], and the “finance-led growth regime” [14]. Minsky’s famous expression that “stability breeds instability” expresses the endogenous nature of business cycles or even crises; here, the PSS transfer payments are explained—Minsky’s quote is the theoretical crown of the study.

It has already been shown that Greek data are consistent with the financialization process—indeed endogenously—of both the whole economy and its public sector especially [29,30]. The research question (RQ) of this study is as follows: “Do some indicative determinants of our financialization theory (combination of FFH and the finance-led growth regime) have asymmetric effects on the Greek PSS in the long term?” This is important: if this is indeed the case (as we have seen), then the economic policy mix (either European or national (Greek)) should act countercyclically to the endogenous economic cycles that are now asymmetrically magnified by financialization; in addition, monetary policy makers will have to intensify their efforts to maintain (permanent) the stability of the financial system. Thus, the identification of the models (1) is based on the mentioned theory, i.e., the FFH and “finance-led growth regime”, and (2) sets the research hypotheses (RHs) to be tested, i.e., RH1: “First-stage FFH determinants (hedge financing structures), such as the relative profitability of the financial versus nonfinancial sector ($rgoss12s11$) or credit to the nonfinancial sector ($credtnfgdp$), have asymmetric and long-run effects on PSS”; RH2: “Second-stage FFH determinants (hyper-speculative financing structures), such as the long-run loans to the domestic sector from abroad ($s1ltls2t$) or the Athens stock exchange general share price index ($asegspi$), have asymmetric and long-run effects on PSS”; RH3: “Third-stage FFH determinants (Ponzi and de facto Eurozone fragility), such as the limitation of public borrowing (especially the snowball effect, $rgsnowb$) or gross external debt of the country ($extdbtgdp$), have asymmetric and long-run effects on PSS”.

Based on the properties of the sample series, the estimations were performed using the nonlinear autoregressive distributed lag (NARDL) method. The main conclusions of the study are as follows: the effect of PSS determinants—as indicated by FFH—is asymmetric and often greater under recession-compatible conditions than it is at expansion; meanwhile, the duration of the asymmetric adjustment varies from fast (up to 4 quarters) to slow (even as long as 16 quarters). In other words, the main conclusion of the study is that its research question turned out to be valid since none of the three developed research hypotheses could be rejected.

Our findings are partially consistent with the relevant literature [17,18]. Policy implications relate to the attention that European policy makers should pay to all RH1-3 determinants, whose negative changes proved to have a larger effect on the PSS, both in magnitude and in their adjustment time to a new equilibrium point.

We contribute to the financialization literature by analyzing—for the first time as far as we know—the endogenous, nonlinear long- and short-run dynamics of the PSS, based on a comprehensive model comprising the FFH and the finance-led growth regime. Our analysis is generalizable to any Eurozone member country because the imperfect and fragile financial structure of EMU is the same for all; so, all can suffer the consequences of these EUR area characteristics. For instance, a sovereign liquidity crisis can be transformed by the capital markets into a solvency crisis and eventually force the member country into default. Member countries issue debt on the EUR that they do not control; meanwhile, there is no centralized fiscal budget and the ECB has no mandate of last resort for lenders. Therefore, their fiscal policies, including the social one—the sustainability of which we are examining here—are subject to the above-mentioned common constraints. Thus, generalization is a matter of the governments’ credibility with the markets, ensured through the control mechanisms that were put in place during the EMU debt crisis of the 2010s [31].

The clear limitation of the paper is that we studied the total magnitude of PSS (the European System of Accounts (ESA-2010, § 4.83) defines the (PSS) “Social contributions and benefits (D.6) as: social benefits are transfers to households, in cash or in kind, intended to relieve them from the financial burden of a number of risks or needs [(a) sickness; (b) invalidity, disability; (c) occupational accident or disease; (d) old age; (e) survivors; (f) maternity; (g) family; (h) promotion of employment; (i) unemployment; (j) housing; (k) education; (l) general neediness] made through collectively organized schemes, or outside such schemes by government units and Nonprofit Institutions Serving Households (NPISHs); they include payments from general government to producers which individually benefit households and which are made in the context of social risks or needs”.); therefore, we do not know which specific components of their asymmetric shocks are involved.

The structure of the study is as follows: In the second section, we discuss the theoretical underpinnings of the financial fragility hypothesis and the finance-led growth regime; here, we document that these theoretical underpinnings explain public social spending—that is why we have used them as the bases of this study’s research hypotheses. In the same second section, we also describe the research design, modeling, and data analyzed in the study. The third section presents the empirical analysis and its results, offering relevant interpretations; meanwhile, in the fourth section, we present the policy implications. The fifth section concludes the study.

2. Materials and Methods

2.1. Brief Theoretical Background

Through a holistic, interdisciplinary approach, we develop a theoretical interpretation of public social spending (PSS) which places it within the financial fragility hypothesis (FFH) and the finance-led growth regime. We synthesized theoretical schemes of endogenous economic fluctuations (FIHs), extended with the inherent instability of the globalized financial system (eFIH), and magnified by the vulnerability of the incomplete Eurozone, for a small member country like Greece.

Our theory stems from the so-called heterodox of “financialization”, but with older roots, which means that it is not a reversible disease of our modern economies, i.e., a structural problem cannot be solved by de-financialization. Only a course toward democracy as universal freedom could resolve the problem [32]. We have empirically found that it is, rather, what Keynes since 1933 [9] and Javidanrad among others [10] have long been saying about the “monetary economy”; in the latter, private investments due to risky choices—as a result of the reversal of present values due to the maturation of innovations—is no longer directed to the market of goods and production but to the market of money and capital, which turns them into counter-productive values. The “monetary gains” prove that these investments are more profitable and clearly less risky than their long-term productive counterparts, where capital becomes “scarce” and expensive. The mechanism of credit–debt reconstruction and the active role of the private money of commercial banks (the opposite of the “passive role of money” of mainstream economics, which axiomatically leads to the exogenous control of the money supply by the central bank; the monetarist view (Friedman, 1953) of the operation of the quantitative theory of money—with the quantity of liquid assets circulating in the economy (M) as the short-term cause with the price level (P) as the outcome—is interpreted in the opposite direction by heterodox economists (endogenous money supply) especially explain why the capital accumulation of the nonfinancial sector is mainly financial and not real (industrial or, more generally, the goods market). From the same perspective, however, it is the form of financialization known as “value for the shareholder” that has changed the management of large corporations; the nonfinancial ones are now mainly competing not on the markets for goods and services but mainly on the international capital market [14,19]. Also, “everyday finance” and the “transformation in computing practices and the financial culture of credit rating agencies” complete this complex form of financialization [7,8]; ultimately, it moves the economy away from the profitability of firms through the production of goods and services to profitability through credit lending speculation.

The famous Minsky quote “stability breeds instability” is at the core of the theoretical foundation of the present study [11]. Every (capitalist) economy endogenously establishes financial structures that are susceptible to crises [24]; economic performance is determined by the way in which companies of the nonfinancial sector finance their gross fixed capital formation or investments.

Minsky has called the first stage of this mechanism “hedge finance”; it appears in the upward part of the business cycle, the recovery phase, when firms fully cover the interest and principal of their debts from their operating income; corporate profitability and easy banking credit and finance characterize this first stage of Minsky’s [21] financial instability hypothesis (FIH). However, reversals in present value relations take place when the increase in demand—funded by speculative finances—raises interest rates, wages, and the prices of material, so that profit margins, and thus the ability to validate the past, erode [11]. He emphasized the endogenous nature of the fragile or crisis-prone financial structures of the economy, based on the sequence of the three stages of his FIH [21]. Thus, the reversals of present values cause the failure of some investments (whose innovations have matured) so that the respective borrowers are unable to repay part (at best) of their loans; then, if banks still believe in the business of their clients, they may refinance the principal while they steadily receive interest; Minsky calls this second stage of indebtedness and inflated asset prices the “speculative-financing structure”. It should be underlined that financial leverage seems to be crucial in Minsky’s thought for the liquidity, solvency, and—in the end—the bankruptcy of any firm. In the last third stage (that of “Ponzi finance”), the borrower, firm, sector, or the entire economy cannot repay even the interest from its operations, so they go bankrupt.

The latter can occur when the market witnesses many insolvent borrowers when the authorities implement a reverse (unexpected) economic policy, or—as in the Eurozone’s case—when the incomplete and fragile architecture of the EMU is revealed. One can see that this occurred in the absence of a centralized EMU budget, which resulted in a situation where the Eurozone could not perform the recycling of surpluses. The opposite of this would involve the issuing of national debts on a currency that no member country controls—this would be a sui generis of European unification arrangements [13]. The fragility of the EMU could be understood in accordance with the mandate of the ECB, which cannot act as a lender of last resort; hence, any member of the EMU cannot give a 100% guarantee to its bondholders that it will have the necessary liquidity to pay them out at maturity [13,28,33,34]. Is it not absurd to abruptly change the programming of the lives of the citizens of an entire country because some people are able to speculate in the global capital market? Who gave them this ability other than the authors of the European treaties? Economists study—out of a desire to understand—the conditions of operation in free markets, not unwise ones.

Therefore, because of the Eurozone’s fragility, we use the term “hyper-speculative financing structure” to refer to the second stage of the FFH in the cases of EMU member countries [30]. We added the prefix “hyper” to Arestis and Glickman’s [12] “super speculative financing” proposal, which in turn modifies Minsky’s [25] original idea of “speculative financing” for the open economy; the intent here is to express the inherent architectural vulnerability of the Eurozone. This is justified by the fact that the risk of a liquidity crisis or even bankruptcy is sharply increased in these cases.

Arestis and Glickman [12] extended the Minskyan FIH [25,35,36] to include the open economies (we label this eFIH). They analyzed the role of residents’ borrowing in foreign exchange, and they distinguished three potential scenarios: (i) a crisis that is domestic (d) in origin but impacts its external (e) situation (they labeled this the “d to e crisis”); (ii) a crisis that is external (e) in origin but impacts its domestic (d) situation (they labeled this the “e to d crisis”); and (iii) crisis-intensifying interactions between (i) and (ii) [30].

The above theoretical analysis of the FFH is clearly summarized in Figure 1.

Boyer [14] has explained in depth in a seminal work for the “finance-led growth regime” of our economies that it is based on the “global financial regime”, the “shareholder value as a new form of competition and governance”, and the “highly reactive wage-labor nexus”. It is important to remember major Boyer’s [14] mechanisms of the economic circuit, starting and ending in the “financial system”; after the relevant license from Taylor and Francis (5954670743084/23 January 2025), which is shown in Figure 2. The mechanisms under the symbols—arrows E, F, and G—pass through the “credibility of government”, e.g., firstly, “tax system favorable to most-mobile factors” (E arrow), “productive investment”, “effective demand”, and “profits”, which, combined, lead to a rise in “stock market prices”; meanwhile, via “monetary policy, financial market stabilizer”, this mechanism is completed for the “financial system”, and so forth. Secondly, “limitations of public borrowing” (arrow F), which put pressures on declining “public expenditures”, favor the crowding-out effects in “effective demand”, etc., as is the case in the previous mechanism, E. Thirdly, “monetary policy as financial market stabilizer” (arrow G) returns directly to the starting point, the “financial system”, and so forth.

Thus, based on the aforementioned theory and given the credibility of the idea that the government’s actions should reflect the international financial system (Figure 2), and within the inherent fragility of the EMU (Figure 1), we set up the following three research hypotheses:

RH1: 

“First-stage FFH determinants (hedge financing structures), such as the relative profitability of the financial versus nonfinancial sector ($rgoss12s11$) or credit to the nonfinancial sector ($credtnfgdp$), have asymmetric and long-run effects on PSS”.

RH2: 

“Second-stage FFH determinants (hyper-speculative financing structures), such as the long-term loans to the domestic sector from abroad ($s1ltls2t$) or the Athens stock exchange general share price index ($asegspi$), have asymmetric and long-run effects on PSS”.

RH3: 

“Third-stage FFH determinants (Ponzi and de facto Eurozone fragility), such as the limitation of public borrowing (especially the snowball effect, $rgsnowb$) or gross external debt of the country ($extdbtgdp$), have asymmetric and long-run effects on PSS”.

Through these three research hypotheses, the research question (RQ: Do some indicative determinants of our financialization theory (combination of FFH and the finance-led growth regime) have asymmetric effects on the Greek PSS in the long term?) will be tested.

2.2. Methodology and Modeling

As should have been made clear from the above, we follow a holistic approach that accepts that the social policy (expressed indicatively here through the PSS) is not autonomous from the entire national socioeconomic one within the EMU. Furthermore, the latter has its own political limits too. Therefore, a purely economic analysis is really complex in the absence of an interdisciplinary approach.

Methodologically, the identification of our empirical models for the three research hypotheses (RHs) is based on the financialization theory that was briefly presented in the previous section; this will be tested via the nonlinear autoregressive distributed lag (NARDL) econometrics. Our reference work is that of Shin, Yu, and Greenwood-Nimmo [37], who extended the autoregressive distributed lag (ARDL) approach (For a survey of relevant methods, see Cho et al. [38]) that was popularized by Pesaran, Shin, and Smith [39]. Their NARDL (p, q), a fully dynamic cointegration framework, allows for weak endogeneity to be present in the regressors and/or serially correlated errors; meanwhile, it can model relationships that exhibit combined long- and short-run asymmetries. Their starting point is the following NARDL (p, q) model (1):

$$y_t={\displaystyle {\sum }_{j=1}^p}{\phi }_j{y}_{t-j}+{\displaystyle {\sum }_{j=0}^q}({\theta }_j^{+\prime }{x}_{t-j}^{+}+{\theta }_j^{-\prime }{x}_{t-j}^{-})+{\epsilon }_t$$

(1)

where $x_t$ is a k × 1 vector of multiple regressors, defined such that $x_t=x_0+x_t^{+}+x_t^{-}, {\phi }_j$ is the autoregressive parameter, ${\theta }_j^{+},{\theta }_j^{-}$ are the asymmetric distributed-lag parameters, and ${\epsilon }_t~iid\left(0,{\sigma }_{\epsilon }^2\right)$.

Following Pesaran et al. [39], it is more convenient to rewrite Equation (1) in an error correction form, as follows:

$$\begin{gathered} {\Delta y}_t=\rho y_{t-1}+{\theta }^{{+}^{\prime }}{x}_{t-1}^{+}+{\theta }^{{-}^{\prime }}{x}_{t-1}^{-}+{\displaystyle \sum _{j=1}^{p-1}}{\gamma }_j\Delta y_{t-j}+{\displaystyle \sum _{j=0}^{q-1}}({\phi }_j^{{+}^{\prime }}\Delta x_{t-j}^{+}+{\phi }_j^{{-}^{\prime }}{\Delta x}_{t-j}^{-})+{\epsilon }_t \\ =\rho {\xi }_{t-1}+{\displaystyle \sum _{j=1}^{p-1}}{\gamma }_j\Delta y_{t-j}+{\displaystyle \sum _{j=0}^{q-1}}({\phi }_j^{{+}^{\prime }}\Delta x_{t-j}^{+}+{\phi }_j^{{-}^{\prime }}{\Delta x}_{t-j}^{-})+{\epsilon }_t \end{gathered}$$

(2)

where $\rho ={\displaystyle {\sum }_{j=1}^p}{\phi }_j-1$, ${\xi }_t=y_t-{\beta }^{+\prime }{x}_t^{+}-{\beta }^{-\prime }{x}_t^{-}$ is the nonlinear error correction term—here, ${\beta }^{+}=-{\displaystyle \frac{{\theta }^{+}}{\rho }}$ and ${\beta }^{-}=-{\displaystyle \frac{{\theta }^{-}}{\rho }}$ are the associated asymmetric long-run parameters; ${\gamma }_j=-{\displaystyle {\sum }_{i=j+1}^p}{\phi }_i, j=1,\dots ,p-1,$ ${\theta }^{+}={\displaystyle {\sum }_{j=0}^q}{\theta }_j^{+}$, ${\theta }^{-}={\displaystyle {\sum }_{j=0}^q}{\theta }_j^{-}$, ${\phi }_0^{+}={\theta }_0^{+}$, ${\phi }_j^{+}=-{\displaystyle {\sum }_{i=j+1}^q}{\theta }_j^{+},j=1,\dots ,q-1,$ ${\phi }_0^{-}={\theta }_0^{-}$, ${\phi }_j^{-}=-{\displaystyle {\sum }_{i=j+1}^q}{\theta }_j^{-},j=1,\dots ,q-1.$

To address the “discretionary fiscal policy” variable (discrpol), we note that this was constructed through the methodology of Fatás and Mihov [40], and it was used as a proxy for unobserved government expenditures, such as the “interest paid for the public debt” or “countercyclical additional payments”. This implies that, in general, and theoretically, it should be exogenous. This comes from the residuals of the ordinary least squares (OLS) approach, estimated using Equation (3):

$${\mathrm{\Delta }lgovcons}_t={\mathrm{\Delta }lgovcons}_{t-1}+{\mathrm{\Delta }lgdp}_t+{inflr}_t+{inflr}_t^2+{fuelspiyy}_t+{\epsilon }_t$$

(3)

where ${\mathrm{\Delta }lgovcons}_t$ is the growth rate (yoy) of the government consumption expenditures; ${\mathrm{\Delta }lgdp}_t$ is the growth rate of the nominal gross domestic product (GDP); ${inflr}_t$ is the inflation rate, as measured using the general consumer price index; ${inflr}_t^2$ is the squared inflation; ${fuelspiyy}_t$ is the growth rate of the fuel price index (yoy).

Thus, based on the theoretical relationship of the PSS within the three stages of the FFH (see Figure 1), we have identified three respective models in estimating Equation (2) with NARDL (p, q); these are distinguished by their explanatory variables (${\mathit{x}}_t$) and deterministic terms (constant, trend, time-dummies, etc.). Thus, the dependent of the PSS or $y_t(=scbtotgdp)$ has been regressed onto the representative determinants of each one of the FFH stages: for the first stage, “hedge financing”, the vector of explanatory include ${\mathit{x}1}_t=(grgdpcu, invgdp,rgoss12s11,credtnfgdp)$; for the second stage, “hyper-speculative financing”, we have chosen ${\mathit{x}2}_t=(s1ltls2t, nlbs13gdp,nfidebtoutsgdp,asegspi)$; meanwhile, for the third stage, “Ponzi schemes and Eurozone’s fragility”, the vector was ${\mathit{x}3}_t=(extdbtgdp, rgsnowb,discrpol,nplttgl)$; a small sample, only partially applied for the first model/stage, is presented in the following Equations (4a)–(4c):

$${\mathrm{\Delta }y}_t=\rho {\xi }_{t-1}+{\displaystyle \sum _{j=1}^{p-1}}{\gamma }_j\mathrm{\Delta }{y}_{t-j}+{\displaystyle \sum _{j=0}^{q-1}}({\phi }_j^{{+}^{\prime }}\mathrm{\Delta }{(grgdpcu, invgdp, rgoss12s11, credtnfgdp)}_{t-j}^{+}+{\phi }_j^{{-}^{\prime }}{\mathrm{\Delta }x}_{t-j}^{-})+{\epsilon }_t$$

(4a)

$${\mathrm{\Delta }y}_t=\rho {\xi }_{t-1}+{\displaystyle \sum _{j=1}^{p-1}}{\gamma }_j\mathrm{\Delta }{y}_{t-j}+{\displaystyle \sum _{j=0}^{q-1}}({\phi }_j^{{+}^{\prime }}\mathrm{\Delta }{(s1ltls2t,nlbs13gdp,nfidebtoutsgdp,asegspi)}_{t-j}^{+}+{\phi }_j^{{-}^{\prime }}{\mathrm{\Delta }x}_{t-j}^{-})+{\epsilon }_t$$

(4b)

$${\mathrm{\Delta }y}_t=\rho {\xi }_{t-1}+\sum _{j=1}^{p-1}{\gamma }_j\mathrm{\Delta }{y}_{t-j}+\sum _{j=0}^{q-1}({\phi }_j^{{+}^{\prime }}\mathrm{\Delta }{(extdbtgdp, rgsnowb, discrpol, nplttgl)}_{t-j}^{+}+{\phi }_j^{{-}^{\prime }}{\mathrm{\Delta }x}_{t-j}^{-})+{\epsilon }_t$$

(4c)

The definitions of all variables and sources are given in

Table A1. Definitions, labels, and sources of variables.

Definitions	Labels	Sources
Greece: social contributions and benefits paid (ratio to GDP).	scbtotgdp	Non-financial transactions (ESA2010); social contributions and benefits—total economy (paid, current prices, EUR (ratio to GDP); Eurostat via Refinitiv-LSEG.
Greece: gross domestic product, market prices, annualized rate, and current prices.	grgdpcu	Quarterly national accounts; copyright OECD via Refinitiv-LSEG.
Greece: gross fixed capital formation to GDP ratio and current prices.	invgdp	Authors’ calculations based on Hellenic Statistical Authority data.
Greece: gross operating surplus (GOS) of non-financial firms (S.11).	goss11gdp	Hellenic Statistical Authority and Refinitiv-LSEG/OECD data.
Greece: gross operating surplus of financial firms (S.12).	goss12gdp	Hellenic Statistical Authority and Refinitiv-LSEG/OECD data via Refinitiv.
Greece: ratio of the GOS of the financial sector (S.12) to that of the nonfinancial sector (S.11).	rgoss12s11	Authors’ calculations based on Hellenic Statistical Authority and OECD data via Refinitiv-LSEG.
Greece: credit to the non-financial corporations as a ratio of the GDP.	cdtnfgdp	Authors’ calculations based on Bank of International Settlements via Refinitiv -LSEG.
Greece: discretionary policy of the expenditures of the general government (Fatás and Mihov [40]).	discrpol	Authors’ calculations based on OECD via Refinitiv-LSEG and Hellenic Statistical Authority data.
Greece: money supply M3 growth rate.	m3gr	Authors’ calculations based on Bank of Greece data.
Greece: the ratio of long-term (LT) loans of residents (S1) from non-residents (S2) to total LT loans of residents (S2/F42).	s1ltloanss2tot	Authors’ calculations based on Bank of Greece data.
Greece: ratio of net lending (+)/net borrowing (−) (B.9) of the general government (S.13) to GDP (B.1g).	nlbs13gdp	Authors’ calculations based on Hellenic Statistical Authority data.
Greece: nonfinancial corporations debt outstanding to GDP.	nfidebtoutsgdp	Authors’ calculations based on European Central Bank via Refinitiv-LSEG data.
Greece: Athens Stock Exchange (ASE) general share price index.	asegspi	Authors’ calculations based on Bank of Greece data.
Greece: gross external debt position to GDP.	extdbtgdp	Authors’ calculations based on Bank of Greece data.
Greece: spread between 10-year government bond yield (r) and the growth rate of nominal GDP (g) [expressing the snowball effect].	rgsnowb	Authors’ calculations based on main economic indicators, copyright OECD via Refinitiv-LSEG.
Greece: nonperforming loans to total gross loans and financial soundness indices.	nplttgl	Authors’ calculations based on International Monetary Fund (IMF) via Refinitiv-LSEG.
Greece: money supply M3 outstanding amounts, (EUR Mill.) to GDP.	m3outsgdp	Authors’ calculations based on Bank of Greece data. OECD via Refinitiv-LSEG for the GDP.

of Appendix A, but for convenience, we repeat them here too. Equation (4a) (Model 1/Table 3): $grgdpcu$ = the growth rate of the nominal GDP; $invgdp$ = the private investments to GDP; $rgoss12s11$ = the ratio of the gross operating surplus of the financial sector (s12) to that of the nonfinancial sector (s11); $credtnfgdp$ = the credit to the nonfinancial sector as a ratio to GDP; the deterministic terms used other than the constant are $discrpol$ = discretionary fiscal policy and $m3gr$ = the growth rate of M3 measure of the money supply. Equation (4b) (Model 2/Table 3): $s1ltls2t$ = the ratio of long-term (LT) loans of residents (S1) from non-residents (S2) to total LT loans of residents; $nlbs13gdp$ = the ratio of net lending (+)/net borrowing (−) of the general government (S.13) to GDP; $nfidebtoutsgdp$ = the outstanding debt of the nonfinancial sector to GDP; $asegspi$ = Athens stock exchange general share price index; the deterministic terms used other than the constant are ${d200811}_t$ = dummy for taking account of the GFC-2008 and Eurozone’s one, having the value 1 when t = 2008q1-2011q4 and having the value 0 otherwise. Equation (4c) (Model 3/Table 3): $extdbtgdp$ = gross external debt to GDP; $rgsnowb$ = spread between 10 years government bond yield (r) and the growth rate of nominal GDP (g); $discrpol$ = discretionary fiscal policy; $npltgl$ = nonperforming loans to total gross loans; $m3outsgdp$ = The ratio of money supply M3 outstanding amounts to GDP; the deterministic terms used other than the constant are ${d201015}_t$ = dummy for taking account of the GFC-2008 and Eurozone’s one, having the value 1 when t = 2010q1-2015q1 and having the value 0 otherwise.

To establish a reference point, we report in Table 2 the estimation results for the restricted symmetric ARDL (p, q) regression, as found in the following Equation (5), with the same specification as already mentioned for the final estimated NARDL (p, q), reported in Table 3.

$${\Delta y}_t=\rho y_{t-1}+\theta x_{t-1}+{\displaystyle \sum _{i=1}^{p-1}}{\gamma }_i\Delta y_{t-i}+{\displaystyle \sum _{i=1}^{q-1}}{\pi }_i\Delta x_{t-i}+u_t$$

(5)

The dataset of the present study concerns Greece, an EMU member country; this is a paradigmatic case of SWEAP (South and West Euro Area Periphery, i.e., Greece, Italy, Spain, Portugal, and Ireland) members, which suffered a lot during the Eurozone debt crisis in 2010–2018 as a consequence of the GFC-2008.

It seems that the European Commission’s authorities or the Greek ones have been reassured by standard economic theory and research that, in general, they should not take measures to reduce their external deficits (e.g., [41]). This was based on the mistaken assumption that higher expected rates of the returns of private investments (mainly on real estate, which was not taken into account) in poorer but fully opened EMU countries (like Greece or Portugal) would be able to normalize the increased current account deficits. However, it turned out in the end that the heterodox financialization literature could be better in its explanations of the endogenous nexus between FFH and PSS [29,30]. It should be noted that the FFH includes De Grauwe’s Eurozone’s fragility hypothesis [13].

3. Results and Discussion

Summary descriptive statistics of the variables used in the empirical analysis of the PSS-FFH relationship are provided in

Table 1. Summary statistics.

Variable	Obs	Mean	Std. Dev.	Min	Max
scbtotgdp	97	0.428	0.046	0.343	0.517
grgdpcu	112	0.011	0.078	−0.146	0.174
invgdp	113	0.180	0.061	0.085	0.294
rgoss12s11	97	0.122	0.061	0.000	0.286
credtnfgdp	112	0.528	0.138	0.292	0.716
discrpol	109	0.000	0.036	−0.109	0.092
m3gr	108	0.012	0.033	−0.113	0.099
s1ltls2t	102	0.466	0.182	0.150	0.779
nlbs13gdp	97	−0.065	0.059	−0.307	0.062
nfidebtoutsgdp	97	0.587	0.100	0.366	0.725
asegspi	114	1859.685	1328.195	542.120	5667.600
extdbtgdp	74	8.098	2.559	3.400	12.607
rgsnowb	103	0.051	0.060	−0.067	0.255
nplttgl	57	23.626	14.816	2.360	47.200
m3outsgdp	108	3.762	0.550	2.675	5.295

Notes: The definitions and sources of variables are given in Table A1 of Appendix A.

(the statistics and estimations provided in Table 1,

Table 2. Dynamic linear estimation (ARDL) of the PSS-FFH relationship.

	Model 1	Model 2	Model 3
	Profitability—Financial Stability	Risky options, Open Financial Markets, Indebtedness	Eurozone’s Fragility, Financial Instability
${\Delta scbtotgdp}_t$
Panel A. Dynamic Symmetric Estimations
${scbtotgdp}_{t-1}$	−0.39 **		−1.05 ***
Long Run coeff.
${grgdpcu}_t$	−1.78 **
${credtnfgdp}_t$	0.27 ***
${rgsnowb}_t$			0.41 ***
${discrpol}_t$			0.51 **
Short Run coeff.
${\Delta scbtotgdp}_{t-1}$	−0.42 **	−0.54 **
${\Delta scbtotgdp}_{t-2}$	−0.35 **	−0.52 ***
${\Delta scbtotgdp}_{t-3}$	−0.34 ***	−0.48 ***	−0.47 **
${\Delta grgdpcu}_t$	0.49 ***
${\Delta grgdpcu}_{t-1}$	0.26 **
${\Delta grgdpcu}_{t-2}$	0.18 ***
${\Delta rgoss12s11}_{t-1}$	−0.16 **
${\Delta nlbs13gdp}_t$		−0.09 **
${\Delta nfidebtoutsgdp}_t$		0.67 ***
${\Delta extdbtgdp}_{t-3}$			0.01 **
$\Delta {discrpol}_t$			−0.36 **
$\Delta {discrpol}_{t-1}$			−0.26 **
$\Delta {discrpol}_{t-2}$			−0.16 **
Det. terms
${discrpol}_t$	0.17 **
m3gr	0.11 *
m3outsgdp			0.04 **
d201015			−0.02 **
constant	0.10 *	0.11 **	0.25 **
Panel B. Model Diagnostics
Bounds CI test stats (H0: no level of relationship)	$t_{BDM}$ = −2.97 $F_{PSS}$ = 8.12 No rejection at 1%, inconcl. at 10 or 5%	$t_{BDM}$ = −1.46 $F_{PSS}$ = 2.42 No rejection at 10 or 5 or 1%	$t_{BDM}$ = −3.98 ** $F_{PSS}$ = 7.26 ** Inconclusive at 1%, reject at 10 or 5%
Portmanteau test up to lag 40 (H0: no serial correlation)	Prob > chi2 = 0.7377	Prob > chi2 = 0.9795	Prob > chi2 = 0.6783
Breusch–Pagan heteroskedasticity test (H0: constant variance)	Prob > chi2 = 0.0839	Prob > chi2 = 0.5369	Prob > chi2 = 0.0998
Ramsey RESET F-test (H0: model has no omitted variables)	Prob > F = 0.3095	Prob > F = 0.6165	Prob > F = 0.0318
Jarque–Berra test on normality	Prob > chi2 = 0.0738	Prob > chi2 = 0.1689	Prob > chi2 = 0.9286
Observations	90	91	51
Adj R-squared	0.8045	0.7333	0.8206

Notes: Only statistically significant estimations are presented. *** (**) [*] stands for statistically significant at 1% (5%) [10%] level or lower. The specification of these linear ARDL models has adopted the identification of the respective nonlinear (NARDL) ones presented in Table 3.

and

Table 3. Dynamic asymmetric estimation (NARDL) of the PSS-FFH relationship.

	Model 1	Model 2	Model 3
	Profitability— Financial Stability	Risky options, Financial Devel., Indebtedness	… + extreme Eurozone’s Fragility, Financial Instability
${\Delta scbtotgdp}_t$
Panel A. Dynamic Asymmetric (cointegrated) Estimations
${scbtotgdp}_{t-1}$	−1.35 ***	−1.78 ***	−4.37 ***
${grgdpcu}_{t-1}^{+}$	−1.09 ***
${grgdpcu}_{t-1}^{-}$	−1.00 ***
${rgoss12s11}_{t-1}^{-}$	−0.37 **
${credtnfgdp}_{t-1}^{-}$	0.97 **
${\Delta scbtotgdp}_{t-3}$	−0.19 *
${\Delta grgdpcu}_t^{+}$	−0.19 **
${\Delta grgdpcu}_{t-1}^{+}$	0.53 **
${\Delta grgdpcu}_{t-2}^{+}$	0.37 **
${\Delta grgdpcu}_t^{-}$	−0.39 ***
${\Delta invgdp}_{t-1}^{-}$	0.64 **
${\Delta invgdp}_{t-2}^{-}$	0.62 **
${\Delta rgoss12s11}_t^{-}$	−0.24 **
$\Delta {credtnfgdp}_{t-1}^{-}$	−0.87 **
${s1ltls2t}_{t-1}^{+}$		−0.33 **
${s1ltls2t}_{t-1}^{-}$		0.40 **
${nlbs13gdp}_{t-1}^{+}$		0.26 **
${asegspi}_{t-1}^{-}$		−0.00 ***
$\Delta {s1ltls2t}_{t-1}^{+}$		0.49 **
$\Delta {s1ltls2t}_{t-2}^{+}$		0.62 **
$\Delta {s1ltls2t}_{t-3}^{+}$		0.59 **
$\Delta {nfidebtoutsgdp}_t^{+}$		1.44 ***
$\Delta {nfidebtoutsgdp}_{t-1}^{+}$		0.64 *
$\Delta {asegspi}_{t-2}^{-}$		0.00 **
$\Delta {asegspi}_{t-3}^{-}$		0.00 **
${extgdp}_{t-1}^{+}$			0.07 **
${rgsnowb}_{t-1}^{+}$			0.79 **
${rgsnowb}_{t-1}^{-}$			2.70 ***
$\Delta {scbtotgdp}_{t-1}$			2.57 **
$\Delta {scbtotgdp}_{t-2}$			1.32 **
$\Delta {scbtotgdp}_{t-3}$			0.33 *
$\Delta {extgdp}_t^{+}$			0.02 **
$\Delta {extgdp}_{t-1}^{+}$			−0.03 *
$\Delta {extgdp}_{t-3}^{+}$			0.03 **
$\Delta {extgdp}_t^{-}$			−0.02 **
$\Delta {rgsnowb}_{t-2}^{+}$			−1.69 **
$\Delta {rgsnowb}_{t-3}^{+}$			−1.57 **
$\Delta {rgsnowb}_t^{-}$			0.88 **
$\Delta {rgsnowb}_{t-1}^{-}$			−1.37 **
$\Delta {rgsnowb}_{t-2}^{-}$			−0.46 *
$\Delta {discrpol}_{t-1}^{+}$			0.36 **
$\Delta {discrpol}_t^{-}$			0.29 **
$\Delta {discrpol}_{t-2}^{-}$			0.36 **
$\Delta {npltgl}_t^{+}$			−0.01 **
$\Delta {npltgl}_t^{-}$			0.01 **
$\Delta {npltgl}_{t-1}^{-}$			0.02 **
$\Delta {npltgl}_{t-2}^{-}$			0.02 *
$\Delta {npltgl}_{t-3}^{-}$			0.02 **
Deterministic terms:
${discrpol}_t$	0.12 *
${m3gr}_t$	0.21 **
${d200811}_t$		−0.04 **
${d201015}_t$			−0.03 **
Constant	0.40 ***	0.37 ***	2.53 **
Panel B. Asymmetry Statistics
${Lgrgdpcu}^{+}$	−0.81 ***
${Lgrgdpcu}^{-}$	0.74 ***
${Lrgoss12s11}^{-}$ persistent LR asymm.	0.27 **
${Lcredtnfgdp}^{-}$ persistent LR asymm.	−0.72 **
	$W_{LR}rgoss12s11$ = 6.45 **
	$W_{LR}credtnfgdp$ = 3.84 *
	$W_{SR}grgdpcu$ = 6.21 **
	$W_{SR}invgdp$ = 4.28 **
	$W_{SR}credtnfgdp$ = 3.87 *
${Ls1ltls2t}^{+}$ persistent LR asymm.		−0.18 **
${Ls1ltls2t}^{-}$ persistent LR asymm.		−0.23 **
${Lnlbs13gdp}^{+}$		0.14 **
${Lasegpsi}^{-}$ persistent LR asymm.		0.00 ***
		$W_{LR}s1ltls2t$ = 8.72 **
		$W_{LR}asegspi$ = 14.89 ***
		$W_{SR}s1ltls2t$ = 3.69 *
		$W_{SR}nlbs13gdp$ = 6.01 **
		$W_{SR}asegspi$ = 3.22 *
${Lextdbtgdp}^{+}$ persistent LR asymm.			0.02 **
${Lrgsnowb}^{+}$ persistent LR asymm.			0.18 **
${Lrgsnowb}^{-}$ persistent LR asymm.			−0.62 ***
			$W_{LR}extdbtgdp$ = 35.66 **
			$W_{LR}rgsnowb$ = 22.33 **
			$W_{SR}rgsnowb$ = 5.08 *
			$W_{SR}nplttgl$ = 14.02 **
Panel C. Model Diagnostics
Bounds CI test stats (H0: no level of relationship)	$t_{BDM}$ = −5.53 *** $F_{PSS}$ = 5.74 ***	$t_{BDM}$ = −5.61 *** $F_{PSS}$ = 5.72 ***	$t_{BDM}$ = −10.35 *** $F_{PSS}$ = 16.08 ***
Portmanteau test up to lag 40 (H0: no serial correlation)	Prob > chi2 = 0.3490	Prob > chi2 = 0.7452	Prob > chi2 = 0.0931
Breusch–Pagan heteroskedasticity test (H0: constant variance)	Prob > chi2 = 0.1517	Prob > chi2 = 0.5892	Prob > chi2 = 0.0969
Ramsey RESET F-test (H0: model has no omitted variables)	Prob > F = 0.2739	Prob > F = 0.1742	Prob > F = 0.4028
Jarque–Berra test on normality	Prob > chi2 = 0.0011	Prob > chi2 = 0.4676	Prob > chi2 = 0.8600
Observations [sample]	90 [2000q1-2022q2]	91 [2000q1-2022q3]	51 [2009q3-2022q1]
Adj R-squared	0.8371	0.8049	0.9838

Notes: Only significant estimations are presented. Note: ***, (**), and [*] stand for statistical significance at 1%, (5%), and [10%] levels or lower; the definitions of all variables and sources are given in Table A1 of Appendix A and just after Equation (4). ${LX}^{+}$ = long-run positive [+] effect of the expl. var. X; ${LX}^{-}$ = long-run negative [-] effect of the expl. var. X; $W_{LR}X$ = Wald F-test of long-run symmetry of the variable (X) [i.e., ${LX}^{+}={LX}^{-}$]; $W_{SR}X$ = Wald F-test of short-run symmetry of the variable (X); $t_{BDM}$ = t-stat [43]; $F_{PSS}$ = F-stat [39]; Ref. [39] tabulate the 1% critical values for k = 4 (Case III: unrestricted intercept and no trend), as follows: $t_{crit}=-4.60$; $F_{crit}=5.06$.

were carried out using STATA v. 17.0 econometric software, with an academic license for the University of West Attica, Accounting and Finance Department, MSc Public Economics and Policy). The unbalanced dataset and time–series properties were seriously taken into account when correctly specifying the long-term relationships under scrutiny.

As a reference point, we present in Table 2 the linear estimations (ARDL) of the PSS-FFH relationship; this will aid us in proceeding appropriately to the purpose of this study, which relies on their long- or short-run asymmetries and their relevant adjustment processes (cumulative dynamic multipliers).

Obviously, the linear ARDL-estimated models (Table 2) suffer from what has been called the “hidden cointegration” [42]; this is in the sense that, as has been proved from the asymmetric (N)ARDL (Table 3) values, the cointegrating relationships are rather defined between the positive and negative components of the underlying variables, as opposed to being defined as specified (in Table 2) by each other. Here, in Table 2, it is shown that the PSS-FFH relationship is not cointegrated; this is identified in the first two models, expressing the respective stages of the financial structure invoked by the theory. The third one (the 3rd stage of the PSS-FFH relationship) could be cointegrated, although, in this case, we should reject Model 3 (Table 2), as suggested through the regression equation specification error test (RESET).

Table 3 presents the estimations that are made based on the unrestricted NARDL Equation (2) or Equation (4), allowing both long- and short-run asymmetry. The first major finding of the study is that the cointegration tests are unable to reject the “H0: no level of relationship” in the restricted linear case of Table 2 (ARDL, Equation (5)); however, for both relevant bounds statistics ($t_{BDM}$, $F_{PSS}$) of Table 3, it is possible to reject H0 when long-run asymmetry is modeled appropriately within the NARDL framework.

Thus, in Table 3/Panel C, the results of the bounds CI tests (both $t_{BDM}$ and $F_{PSS}$) firmly reject the null of long-run symmetry or non-cointegrated (CI) relationships; they are defined between the positive and negative components of the underlying variables in all three models. These long-run asymmetries (nonlinear CI) are also confirmed through the relevant Wald tests (Table 3/Panel B/$W_{LR}X$).

The findings, which are interesting and useful, have confirmed the cointegrated nonlinear–asymmetric relationships for each one of the three stages of the FFH for the sample Eurozone member country (Greece). Thus, for the “hedge financing structure”, the first stage, the respective Model 1 (Table 3) shows that the effect of the relative profitability of the nonfinancial sector ($rgoss12s11$) on PSS is asymmetric in the long-run (LR) (Table 3/Panel B/$W_{LR}rgoss12s11$ = 6.45 **), not in the short run (SR), and has significantly larger negative changes (larger significant negative changes $L{rgoss12s11}^{-}=+0.27$ [0.004] compared with the insignificant positive ones $L{rgoss12s11}^{+}=-0.07$ [0.48]; p-values in brackets); meanwhile, the adjustment asymmetry (measured using the respective dynamic multipliers (“Adjustment asymmetry derives from the interaction of impact (SR) and reaction (LR) asymmetries in conjunction with the error correction coefficient” [37] (p. 15)) tends to persist over a horizon of about 8 quarters (Figure A3); analogous evidence for the ratio of the credits/GDP provided to the nonfinancial sector ($credtnfgdp$) (larger significant negative changes $L{credtnfgdp}^{-}=-0.72$ [0.002] compared with the insignificant positive ones $L{rgoss12s11}^{+}=0.15$ [0.36]; p-values in brackets) onto the PSS was found to be significant in both the LR and SR; meanwhile, the adjustment of this asymmetry also needs around 8 quarters to reach a new equilibrium (Figure A3). The SR asymmetries for the PSS relationships with the growth rate of the GDP ($grgdpcu$), private investments ($invgdp$), and credits/GDP ($credtnfgdp$) were also confirmed (Table 3/Panel B/Model 1). Thus, the first research hypothesis (RH1: “First-stage FFH determinants (hedge financing structures), such as the relative profitability of the financial versus nonfinancial sector ($rgoss12s11$) or credit to the nonfinancial sector ($credtnfgdp$), have asymmetric and long-run effects on PSS”) cannot be rejected. To address the “hyper-speculative financing”, the second stage of the FFH (Table 3/Model 2), we found that the estimated effect of LR loans for the domestic sector from abroad ($s1ltls2t$) on PSS is also asymmetric in the LR (and less but still significant in the SR too); this effect has larger negative changes in comparison to positive changes (larger significant negative changes $L{s1lttloanss2tot}^{-}=-0.23$ [0.027] compared with the significant positive changes $L{s1lttloanss2tot}^{+}=-0.18$ [0.005]; p-values in brackets); meanwhile, the adjustment asymmetry tends to persist over a horizon of about 16 quarters (Figure A4). Similar significant estimations were found for the effects of the Athens stock exchange general share price index ($asegspi$) (larger significant negative changes $L{asegpsi}^{-}=0.00$ [0.000] compared with the insignificant positive ones $L{asegspi}^{+}=0.00$ [0.18]; p-values in brackets) on the PSS. Short-run asymmetries were also found for PSS nexus with the public debt ($nlbs13gdp$). The two above-mentioned long-run determinants (asymmetric dynamic multipliers) are presented in Figure A4. Thus, the second research hypothesis (RH2: “Second-stage FFH determinants (hyper-speculative financing structures), such as the long-term loans to the domestic sector from abroad ($s1ltls2t$) or the Athens stock exchange general share price index ($asegspi$), have asymmetric and long-run effects on PSS”) cannot be rejected. Finally, for the third stage of the FFH (Ponzi and de facto Eurozone’s fragility), the estimated Model 3 has shown that the effect of the “limitation of public borrowing” (snowball effect determinant, $rgsnowb$) on PSS is also asymmetric in the long run (and less asymmetric in the SR, albeit marginally); it was proved to have 3.4 times larger negative changes than positive changes (larger significant negative changes $L{rgsnowb}^{-}=-0.62$ [0.000] compared with the positive changes $L{rgsnowb}^{+}=0.18$ [0.009]; p-values in brackets) (equivalent to easier borrowing conditions (when the growth economic rate (g) is greater than the long-term lending interest rate of the economy (r))); meanwhile, the adjustment asymmetry, although explosive for the positive changes (${rgsnowb}^{+}$), seems to be very fast in its negative changes (${rgsnowb}^{-}$) (Figure A5). Significant analogous evidence was found for the gross external debt of the country ($extdbtgdp$) on PSS, with larger positive changes (larger significant positive changes $L{extdbtgdp}^{+}=0.16$ [0.021] compared with insignificant negative changes $L{extdbtgdp}^{-}=0.003$ [0.176]; p-values in brackets); meanwhile, the adjustment asymmetry was estimated to be explosive in both its positive and negative changes. These findings are remarkable—mainly from the perspective of interpretation—because they have been determined to underline the significance of two dummies expressing the data for the GFC-2008 and Eurozone debt crisis over the period 2008–2011 for the second stage of the FFH (Table 3/Panel A/Model 2) and the data for 2010–2015 for the third stage (Table 3/Panel A/Model 3). Thus, the third research hypothesis (RH3: “Third-stage FFH determinants (Ponzi and de facto Eurozone fragility), such as the limitation of public borrowing (especially the snowball effect, $rgsnowb$) or gross external debt of the country ($extdbtgdp$), have asymmetric and long-run effects on PSS”) cannot be rejected. In total, it turned out that none of the three research hypotheses could be rejected. So, this has to be true for the research question (RQ: “Do some indicative determinants of our financialization theory (combination of FFH and the finance-led growth regime) have asymmetric effects on the Greek PSS in the long term?”) of this study.

These findings are consistent with the FFH, confirming its three stages; additionally, they are, in general, consistent with the relevant financialization literature, especially “Boyer’s [14], finance-led growth regime” [14,16,19,29,30], while being partially in line with the research published by other authors [17,18,44].

We suggest that the interpretation of the present study’s results could clarify the previous sentence. For the first stage of the FFH, the declining relative profitability of banks (s12) is explained through the correspondingly higher profitability of the nonfinancial sector (s11) up to 2009 (given the great crises, i.e., that of the GFC-2008 and the Eurozone debt crisis, 2010–2015)—note that the sharp dip in GDP was around 25% over the 2008–2014 period, which the authors consider to be remarkable. However, after the bankruptcy of the government around 2010–2012, the banks that had many government bonds in their portfolios, which became “junks”, also went bankrupt. Then, apart from the major recapitalization undertaken by the Greek government (about EUR 46 billion), the banks—taking advantage of the digital revolution in their transactions (transferring operating costs to their customers through e-banking)—greatly reduced staff costs (staff took part in voluntary exit programs either for retirement or to join the rest of the labor market); thus, they increased public social expenditure on pensions (in the pay-as-you-go Greek social security system, the state is the third pillar after workers and businesses), unemployment, health benefits, etc. (Figure A1 and Figure A2). Hence, the positive coefficient indicates rising PSS along with declining relative profitability of the financial sector ($rgoss12s11$) (Table 3/Panel B/Model 1); that is, we can reject the hypothesis that “there is no automatic stabilizer role for the PSS” in this relationship (AS. PSS-$rgoss12s11$). In this context, the negative LR effect of $credtnfgdp$ on PSS is asymmetric and statistically much larger in expected recessions than in expansions (where practically is found insignificant). The euphoria period of the first decade of the EUR (2000–2008), with almost no difference in the credit ratings of assets—whether one considers core EMU members or SWEAP members (South-West Euro Area Periphery, i.e., Greece, Italy, Spain, Portugal, and Ireland)—was dominated by the second era, i.e., that of the crises (2009–2022). So, the estimated LR coefficient of negative changes (${Lcredtnfgdp}^{-}$) could be attributed to the banks’ defaults; the associated drastic credit limitation (the credtnfgdp time–series (total values, not negative or positive changes) reached its max. value of 71.6% in 2011q3 (the associated value of the PSS was then 45.2%), and reached 54.1% (the 2007q4 value of the GFC-2008) in 2022q4 (the PSS here was 46.6%). Thus, a fall of about 24.4% was recorded (data sources and definitions of the variables are given in Table A1 of Appendix A)); this was much larger than the decrease in the GDP in the denominator that had already occurred since 2009. The implementation of austerity programs by the Troika since 2010 (e.g., the International Monetary Fund (IMF), the European Commission (EC), and the European Central Bank (ECB), which signed three Memorandums of Understanding (MoUs) with the government, whereby they applied austerity policies starting with huge cuts on public expenditures, including social spending) has resulted in a sharp reduction in effective demand and, of course, PSS (the dependent variable (scbtotgdp) was decreased by −21.3% during the period 2009–2014). This can explain not only the negative estimated sign of relative credits (${credtnfgdp}^{-}$) but also its high value (−0.72), which is around 2.7 times greater than the relative profitability (${rgoss12s11}^{-}$). These facts can clearly explain why one cannot reject the null “there is no-automatic stabilizer role for the PSS” in the asymmetric relationship (PSS-$credtnfgdp$); further, we could say that it worked in the opposite direction, the AS (Counter-AS. PSS-$credtnfgdp$). Hence, in this first stage of the FFH, the combined estimated effect of the negative changes in the determinants (${rgoss12s11}^{-}$, ${credtnfgdp}^{-}$) (the average gross operating surplus (GOS) during the period 1999q1–2009q4 of the nonfinancial firms (S11) was EUR 8.249 bn, while that of the financial firms (S12) was EUR 0.780 bn, forming then a ratio of 9.5% for the GOS S12/S11; the figures were fully reversed the next crises periods (GFC-2008, Eurozone debt crisis 2010–2015, COVID-19 in 2020–2022), from 2010q1 to 2023q1; here, the average GOS of S11 was reduced to EUR 6.934 bn, while that of S12 was raised to EUR 0.968 bn; this produced an increase of 46.3% in the ratio of GOS S12/S11, which increased in the crises periods to 13.9% (data drawn from our sample)—this is an aspect of the financialization of the economy “at work”) on PSS is proved to be asymmetric and leads to a strong downward ratcheting effect (this finding is opposite to that of D’Addio [18]). That is, due to the crises and the punitive austerity policies imposed, the sharp and precipitous decline in the Greek GDP was accompanied by a large reduction in the PSS, nullifying their role as automatic stabilizers.

The ratchet effect (“The ratchet effect in consumption (Duesenberry, 1949) exists when consumers do not (or are slow to) respond to a decline in income due to previously committed spending and/or being accustomed to a consumption standard”. [45] (p. 1079)), in our case for the social policy (PSS), could be interpreted as the “scaling-up policy effectiveness over time, i.e., policy ambition needs to “ratchet up” [46] (p. 479), or as [18] (p. 150) underlines, “an upward ratcheting effect exists, i.e., increases in recessions are not matched by equivalent reductions in expansions leading to increasing social spending over time”. The opposite is also true for the “downward ratcheting effect”; this is when, e.g., the PSS increases that occur in a recession are smaller than the equivalent reductions in expansions, leading to decreasing PSS over time. In addition, the negative correlation of PSS–real output growth shows that PSS economically acts as an “automatic stabilizer” [17].

Regarding the second stage, “hyper-speculative financing”, the declining financing of the economy in foreign exchange (${s1ltls2t}^{-}$) including the EUR (which—although it is the national currency for every member country of the Eurozone—expresses its main fragility through the issuing of bonds in a currency that no one but the ECB controls; thus, there is always a danger for investors that they will not be repaid if someone believes that a specific government or firm does not have the necessary amount of maturity) is equivalent to recession pressures; this is because increasingly fewer external finance sources may reflect less credible governments (according to the markets); accordingly, limitations arise in the ability of the economy to borrow. This translates to less effective demand, less output, lower profits, and a reduction in stock market prices. With no active (countercyclical) ECB monetary policy, the financial system adjusts, increasing the funding rate of the economy (Figure 2, Arrow F); this also results in less PSS. Note here that the LR effects of ${s1ltls2t}^{\pm }$ on PSS confirm its role as the “automatic stabilizer”, given that rising finance from abroad (${s1ltls2t}^{+}$) reflects the growing economy, so there is a reduced need for PSS. Due to the fact that the estimated coefficient for negative changes is larger than that for positive changes, we have evidence for a mild upward ratcheting effect in this case (PSS-$s1ltls2t$) (this time it is in line with [18]).

Boyer’s mechanism, F [14], that was applied in the previous case for (${s1ltls2t}_{t-1}^{-}$), can also be used for the remaining significant variables in the second stage, ($asegpsi$) and ($nlbs13gdp$), but this presents a mixed picture on the PSS as an automatic stabilizer (

Table 4. Towards the verdict.

	LR or SR Asymmetry on PSS	LR (+ or −) Effects	Adjustment Asymmetry (Cum. Dyn. Multipl.)	PSS Automatic Stabilizer (AS) or Counter-AS (Counter-AS), Ratchet Effect (RE)
Model 1
$rgoss12s11$	Only LR	${Lrgoss12s11}^{-}$= 0.27	Persists over 8 quarters (Figure A3)	↑${Lrgoss12s11}^{-}$→↑PSS → AS/Upward RE
$credtnfgdp$	LR and SR	${Lcredtnfgdp}^{-}$ = −0.72	Persists over 8 quarters (Figure A3)	↑${Lcredtnfgdp}^{-}$→ pol. interv. → ↓PSS → Counter-AS/Downward RE ${Lcredtnfgdp}^{-}$$> {Lrgoss12s11}^{-}$→↓PSS → Downward RE
$(grgdpcu)$	Only SR	${Lgrgdpcu}^{+}$ = −0.81 ${Lgrgdpcu}^{-}$ = +0.74	Within 4 quarters (Figure A3)	${Lgrgdpcu}^{+}$$> {Lgrgdpcu}^{-}$ →↓PSS → AS/Downward RE
Model 2
$s1ltls2t$	LR and SR	${Ls1ltls2t}^{+}$ = −0.18 ${Ls1ltls2t}^{-}$ = −0.26	Persist over 16 quarters (Figure A4)	${Ls1ltls2t}^{-}$$> {Ls1ltls2t}^{+}$ →↑PSS → AS/Upward RE
$asegspi$	LR and SR	${Lasegspi}^{-}$ = 0.00004	Persists over 16 quarters (Figure A4)	${Lasegspi}^{-}$→↑PSS → AS/Upward RE
$(nlbs13gdp)$	Only SR	${Lnlbs13gdp}^{+}$ = 0.15	Persists over 12 quarters (Figure A4)	↑${Lnlbs13gdp}^{+}$→↑PSS → Counter-AS/Upward RE
Model 3
$rgsnowb$	LR and SR	${Lrgsnowb}^{+}$ = +0.18 ${Lrgsnowb}^{-}$ = −0.62	${Lrgsnowb}^{+}$ Explos. ${Lrgsnowb}^{-}$ Within 4 quarters (Figure A5)	${Lrgsnowb}^{-}$$> {Lrgsnowb}^{+}$ →↓PSS → AS/Downward RE
$extdbtgdp$	Only LR	${Lextdbtgdp}^{+}$ = 0.016	Explosive (Figure A5)	↑${Lextdbtgdp}^{+}$→↑PSS → AS/Upward RE

Notes: Parentheses highlight determinants that have only short-run (SR) effects on PSS; these are considered to be of less importance. Note: pol. interv. = policy intervention (austerity policies implemented by Troika); explos. = explosive (asymmetric dynamic multiplier). There are many mild upward ratcheting effects on PSS, but there are also two strong downward ratcheting effects.

). This can be used in the third stage of the FFH (which is consistent with over-indebtedness and Ponzi schemes); here, (${rgsnowb}^{\pm }$) reflects the credibility of the government’s actions. In our findings, it was proved that both the negative and the positive changes of the snowball effect are consistent with the automatic stabilizer role for PSS; a strong downward ratcheting effect is detected, with the greatest difference in this study (3, 4 times) for the relevant estimated LR asymmetric coefficients (see Table 3/Panel B/Model 3) (this is an opposite finding to that of [18]).

A good reason for this divergence could be the different methodology (theoretical and empirical) used in our studies. A simple review of Figure A1 and Figure A2 (PSS (the 2020 value of the total social contributions and benefits paid (PSS) had not yet reached the level attained in 2007))—accounting for the raw data reflecting the crises (GFC-2008, Eurozone’s debt 2010–2015, and COVID-19) and the subsequent austerity policies chosen by the European governance in Greece—along with the dynamic asymmetric estimations presented in Table 3 lead us to conclude that our sample is not consistent with an always mild upward ratchet effect, as suggested by [18]. Our estimations revealed two cases of a strong downward ratcheting effect on PSS ($rgoss12s11$, $credtnfgdp$, $rgsnowb$), which can be considered to be attributed to austerity policies and the programs implemented by the Troika, which are still in force.

The financialization of the country—mainly due to financial openness, as opposed to trade [14,19,29]—is clearly attested here; the findings reflect not only the history of the Asian crisis of 1997 but also (as [47] have recently shown (including Greece in their sample)) the positive relationship between financial openness and economic policy uncertainty (EPU) risks. The latter do not only originate from EU decisions—they are mainly magnified due to the inherent fragility of the Eurozone [13].

In Table 4, we present the study’s summary of the LR and SR asymmetries, their effects on PSS, the adjustment asymmetry (for significant determinants affecting PSS), our perspective on whether the findings reveal the role of PSS as an automatic stabilizer, and whether its size reflects its ratcheting effects.

4. Policy Implications

Policy implications arise from the testing of research hypotheses that were not rejected. Their results are summarized in the preceding Table 4. If the objective of economic policy is to reduce PSS, this could be achieved (see Table 4) if the following conditions are met: (1) The reduction in credit to the nonfinancial sector is greater than its profitability (Model 1-RH1 (RH1: “First-stage FFH determinants (hedge financing structures), such as the relative profitability of the financial versus nonfinancial sector ($rgoss12s11$) or credit to the nonfinancial sector ($credtnfgdp$), have asymmetric and long-run effects on PSS”)); this is justified by the downward ratcheting effect, revealed through the combination of both of these variables. (2) The reduction in the country’s long-term external borrowing is effective, or, to a smaller degree, the improvement in official business performance—reflected in the rise in the Athens stock exchange general index—could be reached (Model 2-RH2 (RH2: “Second-stage FFH determinants (hyper-speculative financing structures), such as the long-term loans to the domestic sector from abroad ($s1ltls2t$) or the Athens stock exchange general share price index ($asegspi$), have asymmetric and long-run effects on PSS”)); the confirmed effects of both the automatic stabilizers and the downward ratchets of the PSS here justifies the previous policies. (3) The government insists on achieving output growth rates that are higher than its average borrowing rate (reducing the snowball effect) or, to a lesser extent, reduces its external borrowing needs—see policies to restructure the production model towards a knowledge society with high value added (Model 3-RH3 (RH3: “Third-stage FFH determinants (Ponzi and de facto Eurozone fragility), such as the limitation of public borrowing (especially the snowball effect, $rgsnowb$) or gross external debt of the country ($extdbtgdp$), have asymmetric and long-run effects on PSS”)); these policies are justifiable because both the operation of automatic stabilizers and the combined downward ratcheting effect have been verified.

5. Conclusions

The study affirmatively answered the research question, and none of the three research hypotheses were rejected. These addressed the potential asymmetries of the effects of FFH determinants on the public social spending (PSS) of a Eurozone member country over the period 1995q1–2023q4. So, the presented analysis suggests the following findings: (1) The selected representative determinants of each of the three stages of the FFH have asymmetric long-term (and short-term) effects on the dependent PSS; meanwhile, more often than not, the significant large effects comprised negative changes. (2) The asymmetric shocks of explanatories increase PSS gently in many cases and decrease it strongly in others. (3) Accordingly, the “automatic stabilizer” role of the PSS was proven; the contrary was not, however, rejected—that is, the PSS has also been used as a “counter-automatic stabilizer” tool. (4) This leads to “ratchet effects”, but it is unclear whether we have an upward or a downward ratcheting effect; hence, the drift of PSS is clear, where it seems to decline over time. (5) The financialization of the PSS phenomenon is revealed and discussed through the relevant economic interpretations that we have offered here for indicative determinants, including in our research hypotheses (such as the relative profitability of the financial sector to the nonfinancial sector, the credits to nonfinancial corporations (RH1), domestic borrowing from abroad, the Athens stock exchange index (RH2), and the snowball effect or gross external debt of a country (RH3)—all of these examples had long-term effects causing negative changes for PSS, with asymmetric dynamics that moved towards a new equilibrium, with a horizon between 4 and 16 quarters.

The results are partly consistent with those presented in the literature; the divergences can be attributed to different theoretical and empirical approaches.

We contribute to the financialization literature by analyzing—for the first time as far as we know—the endogenous, nonlinear, long- and short-run dynamics of the PSS; this analysis is based on a comprehensive model comprising the FFH and the finance-led growth regime. The asymmetric adjustment paths and the durations of the estimated disequilibrium revealed several fragilities, not only in the productive–competitive model of the sample country but also in the incomplete political economy of the Eurozone. The limitation of borrowing for all three sectors of the economy (nonfinancial, financial, and government) prevails in the modeling; it showed that it was in alignment with the FFH theory, in addition to revealing many aspects of the “finance-led growth regime” that has been established in the country.

In this study, policy implications arise from the testing of research hypotheses that were not rejected. The main limitation of the study lies in the use of aggregate social security benefits and contributions paid (PSS), without distinguishing among the different components (though these are presented in Figure A2). Note that it was our choice to study the specific case of only one SWEAP member state, and not to use the panel data of the whole Eurozone; this choice was made because of the significant differences in the social security systems that they use. The unification of the European social security area is one of the major issues of European social policy, given the huge demographic problem they have to tackle; this requires a significantly centralized EMU budget policy, which could correct the architectural imperfection in the EMU, enabling our societies to remain developed and to avoid establishing risk, panic, and fear as the norm.

Directions for future research could include the following: (1) a study of the individual components of the PSS, if the data become available, and (2) a comparative study with other Eurozone countries. Given European law, the question of whether national institutions have a significant influence on the PSSs of member countries could be examined.

Last but not least, the holistic approach we followed here accepts that the social policy (expressed indicatively here through the PSS) is not autonomous from the entire national socioeconomic policy within the EMU. The latter has political limits too. Therefore, a purely economic analysis is really complex without the incorporation of an interdisciplinary approach. However, especially after the global financial crisis of 2007–2009 (GFC-2008) and the failure of our dynamic stochastic general equilibrium (DSGE) modeling (based on so-called mainstream economics [48]), as well as the lessons that can be taken from Asian capitalism transformations [16], we ought to now also consider sociopolitical solutions that aid in a shift toward “representation” and, finally, “democracy”. Then, the functioning of these systems will be ultimately easier, with the priority being “freedom” (here, we mean overall freedom, i.e., individual plus social plus political, along with established society, which will own the political system) and not just the “efficiency of the markets or financialization of the systems” [32].