Crowding Out or Ricardian Behaviour? Evidence from South Africa

Abstract

This paper examines whether government debt financing crowds out private consumption in South Africa or whether household behaviour is consistent with Ricardian equivalence. Using quarterly data from 1960Q1 to 2025Q1, the study employs a Bayesian time-varying parameter framework that accommodates non-stationarity, structural change, and evolving fiscal transmission mechanisms, and is complemented by a Markov-switching Bayesian VAR as a robustness check. All variables are expressed relative to GDP to avoid scale effects, and inference is based on posterior distributions. The results reveal pronounced state dependence in the debt–consumption relationship. In earlier decades, increases in the debt-to-GDP ratio are associated with statistically meaningful declines in the private consumption share, consistent with crowding-out or precautionary behaviour under weaker fiscal credibility. Over time, however, this negative association weakens and converges toward neutrality, with post-2010 estimates indicating no significant effect of debt on consumption. Conditioning on fiscal stance and financial conditions shows that debt does not exert an independent influence on consumption once government expenditure, tax revenue, and interest rates are taken into account. A constant-parameter Bayesian benchmark masks these dynamics, producing an average effect close to zero. Evidence from a Markov-switching Bayesian VAR similarly finds no persistent regime-specific crowding-out effects. Overall, the findings suggest that observed debt–consumption linkages in South Africa operate primarily through broader fiscal and macroeconomic conditions rather than debt accumulation itself, highlighting the importance of fiscal credibility and policy composition.

1. Introduction

The relationship between fiscal policy and private consumption remains one of the most contested issues in macroeconomic theory and empirical analysis. At the centre of this debate lies the Ricardian Equivalence Hypothesis (REH), which posits that government debt-financed fiscal expansions do not affect aggregate demand because forward-looking households internalise future tax liabilities and adjust their consumption accordingly (Barro, 1974, 1989). Under this framework, fiscal deficits and public debt are neutral with respect to private consumption, as households increase saving in anticipation of future taxation. In contrast, the traditional Keynesian perspective predicts that debt-financed fiscal policy can influence private consumption through income, liquidity, and interest-rate channels, potentially leading to crowding-out effects if higher borrowing costs or expectations of fiscal tightening reduce household spending (İkiz, 2020; Ofori-Abebrese & Pickson, 2018; Isah et al., 2022).

Theoretically, rising public debt may signal deferred taxation or reduced future public services, prompting precautionary saving and restrained consumption. Empirical evidence suggests that fiscal expansions often generate positive, but incomplete, saving responses rather than full consumption increases (European Central Bank, 2021). While REH predicts complete offset under ideal conditions, recent research finds that consumption responses are typically partial and sensitive to household balance sheets, credit constraints, and uncertainty (Alpanda et al., 2024). Cross-country evidence further confirms that Ricardian behaviour varies across institutional and macroeconomic contexts (Morina et al., 2024). While the theoretical foundations of this debate are well established, empirical evidence remains mixed, particularly in emerging market economies where structural constraints, liquidity limitations, and fiscal credibility issues complicate household responses to fiscal policy. In the case of South Africa, the question of whether public debt influences private consumption is especially pertinent (Morina et al., 2024; Gogas et al., 2014). The country has experienced a prolonged period of rising public debt, increasing debt-service costs, and persistent macroeconomic challenges, including low growth, high unemployment, and fiscal imbalances (Ranchhod & Finn, 2017; IMF, 2017, 2025a, 2025b; World Bank, 2018). These developments have intensified policy discussions around fiscal sustainability and the effectiveness of fiscal consolidation strategies, making it crucial to understand how households respond to debt accumulation over time.

In economies characterised by high unemployment, liquidity constraints, and imperfect credit markets—features consistent with South Africa’s recent experience—non-Ricardian responses may dominate. Public borrowing can influence private consumption through income effects, expectations, precautionary saving, and financing conditions. Fiscal effectiveness depends critically on the prevalence of liquidity-constrained or “hand-to-mouth” households (Guo, 2023). Empirical studies document partial offsets rather than full neutrality (European Central Bank, 2021) and emphasise the importance of household balance sheets in shaping fiscal transmission (Alpanda et al., 2024). In certain contexts, debt-financed fiscal interventions may even crowd in consumption by relaxing binding constraints, as suggested by evidence on South Africa’s Social Relief of Distress grant during COVID-19 (Bhorat et al., 2017; Sardoni, 2021; Bassier et al., 2022; Bhorat & Köhler, 2025). Overall, theory permits multiple transmission channels, and country-specific evidence remains ambiguous (Buthelezi, 2023; IMF, 2025c).

A growing empirical literature has examined the relationship between fiscal policy and macroeconomic outcomes in South Africa using a range of methodologies, including vector autoregression (VAR), structural VAR (SVAR), autoregressive distributed lag (ARDL) models, and local projection techniques (e.g., Jooste et al., 2013; Kemp, 2020; Du Rand et al., 2023; Mbaleki, 2024). These studies provide valuable insights into fiscal multipliers and transmission mechanisms, but typically rely on constant-parameter frameworks that assume stable relationships over time. Such an assumption may be restrictive in an economy characterised by structural change, evolving fiscal institutions, and shifts in policy credibility. More recent contributions in the broader literature emphasise that fiscal transmission is inherently time-varying and state-dependent, with the impact of government debt on private behaviour differing across macroeconomic regimes, levels of fiscal credibility, and financial conditions. In particular, the effectiveness of fiscal policy may depend on whether households perceive government debt as a signal of future taxation, fiscal stress, or macroeconomic instability. These considerations are especially relevant in the South African context, where fiscal dynamics have evolved significantly over the past six decades, reflecting changes in economic structure, policy regimes, and global financial conditions (SARB, 2021; National Treasury, 2025). However, existing studies largely assume stable fiscal transmission mechanisms, thereby overlooking the possibility that the debt–consumption relationship evolves over time in response to changing fiscal credibility and macroeconomic conditions.

Despite these advances, there remains a limited empirical understanding of how the debt–consumption relationship evolves over time in South Africa. Existing studies largely focus on average effects, thereby obscuring potential heterogeneity across different periods and macroeconomic environments. Moreover, few studies explicitly model the possibility that the relationship between public debt and private consumption may shift gradually or exhibit regime-dependent behaviour. This represents a critical gap in the literature, particularly given the importance of fiscal credibility and macroeconomic stability in shaping household expectations.

To address these gaps, this paper adopts a flexible empirical framework that allows the debt–consumption relationship to evolve over time. Specifically, the study examines whether government debt crowds out private consumption in South Africa or whether household behaviour is consistent with Ricardian equivalence, using a Bayesian time-varying parameter (TVP) framework. Unlike conventional models that impose parameter stability, the TVP approach allows the effect of government debt on private consumption to evolve over time in a flexible and data-driven manner. This is particularly important in the South African context, where structural breaks, policy shifts, and macroeconomic shocks are likely to have altered fiscal transmission mechanisms.

South Africa, therefore, provides an instructive setting to evaluate whether households behave in a Ricardian or non-Ricardian manner. The economy has been repeatedly exposed to adverse shocks, fiscal stress, and macro-financial volatility, shaping expectations and intertemporal decisions. Official assessments highlight the role of policy credibility, inflation expectations, and macro-financial risks associated with elevated debt and rising debt-service burdens (IMF, 2024; SARB, 2023). Persistent expenditure pressures and revenue underperformance have kept debt sustainability at the forefront of policy debate (National Treasury, 2023, 2024). Evidence suggests that in such environments, fiscal stress may dampen private demand through expectations and interest-rate channels rather than generate strong Keynesian responses (Buthelezi, 2023; Burger & Calitz, 2021).

The empirical model is grounded in an intertemporal consumption framework, where private consumption depends on fiscal variables, financial conditions, and expectations about future policy. Government debt is included as the key variable of interest, capturing the intertemporal fiscal burden faced by households, while government expenditure, tax revenue, and interest rates are incorporated as control variables reflecting fiscal stance and macro-financial conditions. By focusing on the consumption channel, the analysis directly tests the core implication of the Ricardian Equivalence Hypothesis, which concerns the response of household consumption to government debt rather than investment behaviour.

The findings reveal that the relationship between government debt and private consumption in South Africa is strongly time-varying and state-dependent. In earlier decades, increases in the debt-to-GDP ratio are associated with statistically significant declines in private consumption, consistent with crowding-out effects under conditions of weaker fiscal credibility and constrained financial markets. Over time, however, this negative relationship weakens and converges toward neutrality, with post-2010 estimates indicating no statistically significant effect of debt on consumption. These results suggest that improvements in fiscal credibility and financial development may have altered the way households respond to government debt, lending support to a more nuanced interpretation of Ricardian behaviour.

To further examine the robustness of these findings, the paper complements the TVP analysis with a Markov-Switching Bayesian VAR (MS-BVAR) model, which captures discrete regime changes in fiscal transmission. The MS-BVAR results confirm the absence of persistent regime-specific crowding-out effects, reinforcing the conclusion that the debt–consumption relationship is both time-varying and context-dependent rather than structurally fixed.

This paper makes three key contributions. First, it provides new evidence on the debt–consumption relationship in South Africa using a flexible empirical framework that accommodates structural change and evolving fiscal dynamics. Second, it demonstrates the importance of accounting for time variation when evaluating fiscal transmission mechanisms, showing that conclusions based on constant-parameter models may be misleading. Third, it offers policy-relevant insights into the role of fiscal credibility in shaping household behaviour, with implications for the design and communication of fiscal consolidation strategies.

The remainder of the paper is structured as follows. Section 2 reviews the relevant theoretical and empirical literature on Ricardian equivalence and fiscal transmission. Section 3 outlines the data and empirical methodology, including the Bayesian time-varying parameter model and the Markov-switching VAR framework. Section 4 presents the empirical results and discusses their implications. Section 5 concludes.

2. Literature Review

This section reviews the theoretical and empirical literature on the relationship between government debt financing and private consumption, with particular emphasis on the Ricardian Equivalence Hypothesis and its applicability across different macroeconomic and institutional contexts.

The discussion is structured in two parts. Section 2.1 presents the principal theoretical frameworks linking debt financing to consumption behaviour, outlining the conditions under which Ricardian neutrality is expected to hold and the mechanisms through which departures from neutrality may emerge. Section 2.2 examines the empirical evidence from advanced and emerging economies, including Africa and South Africa, highlighting the mixed, heterogeneous, and context-specific nature of existing findings.

2.1. Theoretical Perspectives on Debt Financing and Consumption

Macroeconomic theory has long emphasised the relationship between private consumption and government debt financing. The debate traditionally contrasts two opposing perspectives: non-Ricardian (Keynesian) consumer behaviour and the Ricardian Equivalence Hypothesis (REH). Formally articulated by Barro following Ricardo’s original insight, REH posits that debt-financed fiscal expansions need not stimulate aggregate demand. Under stringent assumptions—including infinitely lived or altruistically linked households, perfect capital markets, lump-sum taxation, and rational expectations—government debt is viewed as deferred taxation. Households internalise the government’s intertemporal budget constraint and adjust their savings to offset fiscal deficits, leaving private consumption unchanged. In this framework, public debt does not constitute net wealth, and the timing of taxation is irrelevant for consumption decisions.

Relaxing these assumptions yields non-Ricardian outcomes. When households face liquidity constraints, finite planning horizons, distortionary taxation, or uncertainty about future fiscal adjustments, government debt can influence consumption through several channels. Debt-financed spending may crowd out private consumption via higher real interest rates or expectations of future fiscal tightening. In settings of weak fiscal credibility, rising debt may heighten macroeconomic uncertainty and induce precautionary saving. Alternatively, if households underestimate future tax liabilities or fail to internalise intertemporal fiscal constraints, public debt may be perceived as net wealth, generating short-run consumption increases.

Contemporary macroeconomic theory increasingly recognises that the validity of Ricardian equivalence is state-dependent. Fiscal transmission mechanisms are shaped by macroeconomic conditions, institutional credibility, and interactions with monetary policy. Recent theoretical contributions emphasise that consumption responses depend on perceptions of fiscal sustainability, expectations regarding debt stabilisation, and the prevailing interest–growth differential. These insights suggest that a constant, time-invariant relationship between debt and consumption is unlikely to hold over extended periods marked by structural and policy change.

2.2. Empirical Literature

Empirical investigations of the Ricardian Equivalence Hypothesis have intensified following the sharp rise in public debt after the Global Financial Crisis and the COVID-19 pandemic. Although Ricardian equivalence predicts that government debt issuance should not affect private consumption because households are forward-looking, international evidence remains mixed and largely inconclusive, particularly in emerging and developing economies.

A central theme in the literature is that the strict assumptions underlying Ricardian equivalence, such as perfect capital markets, lump sum taxation, infinite horizons, operative intergenerational altruism, and strong policy credibility, rarely hold in practice. When these assumptions are relaxed, non-Ricardian outcomes often emerge. Evidence from advanced economies generally supports partial rather than complete Ricardian equivalence. For example, Bohn (1998) shows that although households adjust savings in response to fiscal imbalances, the response is insufficient to fully offset government dissaving. Similarly, recent European Central Bank studies, such as Coenen et al. (2023), find that fiscal expansions increase household saving, but by less than unity, implying incomplete offsets. These findings suggest that Ricardian behaviour may hold only in a limited sense, with uncertainty, credit constraints, and household heterogeneity shaping transmission.

Empirical investigations of the Ricardian Equivalence Hypothesis (REH) have produced mixed evidence across different periods and methodologies. Early contributions, such as Kormendi (1983), provide partial support for Ricardian behaviour, suggesting that private consumption may respond to government fiscal actions in a manner consistent with forward-looking expectations. However, subsequent studies have challenged this view. For instance, Evans (1993) finds little evidence supporting Ricardian equivalence using historical U.S. data, concluding that consumers do not fully offset government borrowing through increased savings. Similarly, Seater (1993), in a comprehensive survey of the literature, highlights that empirical support for REH is highly sensitive to model specification and data choice, with no clear consensus emerging. From a fiscal sustainability perspective, Bohn (1998) demonstrates that government debt dynamics are systematically related to fiscal policy behaviour, reinforcing the importance of debt as a key variable in intertemporal analysis, although not providing direct support for strict Ricardian neutrality. More recent studies continue to report deviations from the Ricardian proposition. For example, Bajo-Rubio et al. (2010) provide evidence from Spain indicating that fiscal deficits and debt dynamics affect private sector behaviour, suggesting only limited support for Ricardian neutrality. Furthermore, Minea and Villieu (2012) show that persistent fiscal deficits can influence economic growth and private sector responses, highlighting that the assumptions underlying Ricardian equivalence may not hold in the presence of fiscal imbalances.

Structural approaches further challenge strict Ricardian neutrality. Using a heterogeneous agent New Keynesian framework, Eichenbaum et al. (2025) show that households largely fail to internalise future tax liabilities when responding to fiscal transfers, exhibiting high marginal propensities to consume. Their results contradict the core prediction of Ricardian equivalence. Likewise, Al Mamun et al. (2025), using a Markov switching VAR for Central and Eastern European economies, finds that consumption responses to public debt depend on fiscal regimes, undermining the notion of a stable Ricardian relationship across time and macroeconomic conditions.

Evidence from sub-Saharan Africa similarly highlights state dependence, liquidity constraints, and policy credibility as key determinants of fiscal transmission. Panel evidence indicates that fiscal effects vary across the business cycle and are sensitive to macroeconomic slack conditions (Woldu, 2023). Ofori-Abebrese and Pickson (2018) argue that uncertainty and liquidity constraints weaken Ricardian offsets, while Omodero (2019) reports that fiscal deficits significantly influence private consumption, suggesting crowding out effects inconsistent with Ricardian neutrality. Ghosh et al. (2013) further show that rising public debt, particularly in countries with weak fiscal institutions, alters private sector expectations and behaviour. Taken together, this evidence suggests that strict Ricardian equivalence is unlikely to hold in fiscally vulnerable and institutionally constrained economies.

Recent empirical evidence continues to provide mixed support for the Ricardian Equivalence Hypothesis, particularly in developing and emerging economies. For instance, Banday and Aneja (2019) examine the validity of Ricardian equivalence in China using time-series techniques and find no evidence supporting strict Ricardian neutrality. Similarly, Ofori-Abebrese and Pickson (2018), employing a panel ARDL approach for Sub-Saharan African countries, report that fiscal variables significantly influence private consumption, thereby rejecting the Ricardian proposition. In a cross-country setting, Afonso and Jalles (2019) provide evidence of partial Ricardian behaviour, suggesting that households only partially offset government fiscal actions. More recently, Isah et al. (2022) argued that Ricardian Equivalence is valid in the case of Nigeria when the strict assumption of RET is maintained, but is insignificant when the major assumption of RET is relaxed.

In the South African context, existing research has examined fiscal policy, debt dynamics, and macroeconomic performance, although relatively few studies directly assess the debt financing channel into private consumption. Tendengu et al. (2022) employ ARDL and nonlinear ARDL models to show that fiscal variables exhibit both short-run and long-run effects, often with asymmetric dynamics. Maluleke et al. (2023) document asymmetric effects of public investment on private investment. On the household side, Mutezo (2014) finds that liquidity and balance sheet conditions significantly influence consumption behaviour, indicating that credit constraints play an important role in shaping fiscal transmission in South Africa.

Overall, the empirical literature offers limited support for strict Ricardian equivalence, especially outside advanced economies. Instead, findings point to partial, state-dependent, and regime-contingent consumption responses. In emerging and developing economies, departures from Ricardian behaviour appear more pronounced due to liquidity constraints, financial market imperfections, fiscal credibility concerns, and heightened macroeconomic uncertainty. Importantly, recent contributions emphasise that the relationship between public debt and private consumption is not stable over time but varies across fiscal regimes, phases of the business cycle, and institutional environments.

Despite these advances, much of the existing evidence, particularly for Africa and emerging markets, relies on constant parameter models that assume stable fiscal transmission over long samples. This assumption is problematic in environments characterised by structural change and evolving fiscal institutions. Similarly, few studies clearly distinguish between the effects of debt accumulation itself and the broader fiscal stance and financial conditions that accompany rising debt, making it difficult to determine whether observed crowding out effects reflect debt itself or correlated macroeconomic factors.

This study addresses these gaps by providing a comprehensive assessment of the relationship between public debt and private consumption in South Africa using a Bayesian time-varying parameter framework applied to a long quarterly dataset spanning multiple fiscal regimes. By allowing the impact of government debt on private consumption to evolve endogenously, the analysis captures regime dependence and structural change that static models cannot detect. In addition, by conditioning explicitly on government expenditure, tax revenue, and interest rate conditions, the study disentangles the independent effect of debt from the broader fiscal environment. In doing so, it contributes new evidence to the Ricardian equivalence debate and provides policy-relevant insights into how fiscal credibility and macroeconomic conditions shape household consumption behaviour in an emerging market setting.

3. Materials and Methods

This section outlines the data, variable construction, and empirical methodology used to investigate the relationship between government debt financing and private consumption in South Africa. Section 3.1 describes the data sources, sample period, and the construction of the principal variables included in the analysis. Section 3.2 presents the econometric framework and identification strategy, explaining the rationale for adopting a Bayesian time-varying parameter approach that accommodates non-stationarity, structural change, and evolving fiscal transmission mechanisms. Together, these elements provide the foundation for the empirical results presented in Section 4.

3.1. Data and Variable Construction

The empirical analysis employs quarterly South African data covering the period 1960Q1 to 2025Q1, sourced from the South African Reserve Bank. The dependent variable is private consumption expenditure as a share of GDP, denoted CE, which captures household consumption behaviour in relative terms and aligns directly with empirical tests of Ricardian equivalence. Government debt is measured by the gross public debt to GDP ratio, denoted DEBT, reflecting the stock of outstanding fiscal obligations relative to the size of the economy. This measure is particularly relevant in the South African context, where debt sustainability and fiscal credibility have become central policy concerns.

To account for the broader fiscal and macroeconomic environment influencing consumption behaviour, the model includes government expenditure as a share of GDP, denoted GE, tax revenue as a share of GDP, denoted TR, and the short-term interest rate, denoted IR.

In the South African context, short-term interest rates are commonly used to capture monetary policy stance and borrowing conditions, which directly influence consumption through credit channels and intertemporal substitution.

Furthermore, the time-varying parameter framework allows the effect of interest rates to evolve over time, thereby capturing both short- and longer-term influences.

Similarly, Ricardian equivalence concerns how households respond to accumulated government liabilities rather than short-term fiscal flows. Debt captures the intertemporal fiscal burden and is therefore more appropriate for analysing consumption responses based on expectations of future taxation.

These variables capture the principal channels through which fiscal policy and financial conditions shape household decisions, including public and private sector substitution effects, expectations of future taxation, and intertemporal consumption and saving trade-offs.

All variables are expressed as ratios or percentages relative to GDP, except for the interest rate, in order to minimise scale effects and reduce the risk of spurious correlations associated with trending nominal aggregates. Descriptive statistics are presented in

Table 1. Descriptive Statistics.

Variable	Mean	Std. Dev.	Min	Max	Skewness	Kurtosis
CE (Private Consumption, % of GDP)	4.54	2.36	2.08	11.6	0.89	3.12
DEBT (% of GDP)	51.02	80.04	0.82	348.91	1.75	5.89
GE (% of GDP)	15.3	3.54	7.9	23.2	0.41	2.78
TR (% of GDP)	100	1072	-14	17338	8.92	92.14
IR (%)	10.5	3.54	4.75	17.8	0.22	2.35

Note: CE = Private consumption; DEBT = Government debt; GE = Government expenditure; TR = Tax revenue; IR = Interest rate. All variables are expressed in percentages unless otherwise stated.

, while trend patterns are illustrated in Figure 1.

3.2. Econometric Framework and Identification Strategy

The empirical model employed in this study is grounded in Ricardian Equivalence framework, originally formalised by Barro (1974). The theory argued that rational forward-looking households maximise intertemporal utility subject to a lifetime budget constraint, taking into account the government’s intertemporal budget constraint.

Consider a representative household that maximises lifetime utility:

$$U={\sum }_{t=0}^{\infty }{\beta }^{t}u\left(C_t\right)$$

(1)

Subject to the intertemporal budget constraint:

$$A_{t+1}=\left(1+r\right)\left(A_t+Y_t-T_t-C_t\right)$$

where $C_t$ denotes consumption, $Y_t$ is the income, $A_t$ assets and r is the interest rate. The government finances its expenditure through taxes and borrowing, subject to the intertemporal budget constraint:

$$B_t={\sum }_{t=0}^{\infty }{\displaystyle \frac{T_t-G_t}{{\left(1+r\right)}^{s-t}}}$$

(2)

where $B_t$ represents the public debt and $G_t$ government expenditure.

Under Ricardian Equivalence, households internalise the government’s budget constraint, recognising that current debt-financed expenditure implies higher future taxes. Consequently, consumption depends on the present value of taxes rather than their timing, implying that fiscal deficits and debt accumulation do not affect aggregate consumption.

However, in the presence of liquidity constraints, finite horizons, or uncertainty, this neutrality result may not hold. In such cases, government debt can influence consumption through expectations, interest-rate effects, and income channels, potentially leading to crowding-out behaviour.

In addition to the Ricardian channel, fiscal policy may also affect the economy through a crowding-out mechanism operating via the interest rate. Following the intuition of Carlson and Spencer (1975), an increase in government borrowing raises the demand for loanable funds, which may lead to higher interest rates and a reduction in private investment. Formally, private investment can be expressed as a decreasing function of the interest rate, $I_t=I{r}_t$ with $\delta I/\delta r<0$. In this setting, government debt may influence aggregate demand either through intertemporal consumption smoothing consistent with Ricardian equivalence or through interest-rate effects that crowd out private sector activity. This dual mechanism underpins the empirical investigation conducted in this study.

However, in contrast to models such as Carlson and Spencer (1975), which majorly emphasise investment-based crowding-out through capital markets, the present framework focuses on the consumption channel, which is central to the Ricardian Equivalence Hypothesis.

This analytical framework, incorporating both Ricardian and crowding-out channels, provides the theoretical foundation for the empirical specification, in which private consumption is modelled as a function of fiscal and macroeconomic variables. Specifically, the empirical model can be expressed as

$$C_t=f\left({DEBT}_t,{GE}_t,{TR}_t,{IR}_t\right)$$

where public debt (${DEBT}_t$) captures the intertemporal fiscal burden faced by households, government expenditure (${GE}_t$) and tax revenue (${TR}_t$) reflect the fiscal stance, and the interest rate (${IR}_t$) captures financial conditions.

The primary objective of this paper is to determine whether government debt financing crowds out private consumption or whether household behaviour is consistent with Ricardian equivalence. Conventional constant coefficient time series models assume a stable and invariant relationship between debt and consumption across extended historical samples. Such an assumption is unlikely to be appropriate in the South African context, given substantial changes in fiscal institutions, macroeconomic regimes, and financial market development over the sample period.

The empirical strategy, therefore, adopts a Bayesian state space framework with time-varying parameters, allowing the effect of government debt on private consumption to evolve gradually over time. This approach offers several advantages. First, it accommodates persistent non-stationarity and structural change without relying on excessive differencing or extensive pre-testing procedures. Second, it captures regime-dependent fiscal transmission mechanisms that reflect the historical evolution of South Africa’s fiscal and macroeconomic environment. Third, it permits an assessment of whether Ricardian-type behaviour emerges endogenously over time rather than being imposed as a maintained assumption.

Baseline Time-Varying Parameter Model

The baseline empirical specification is given by

$${CE}_t={\mu }_t+{\beta }_t{DEBT}_t+\gamma {GE}_t+\delta {TR}_t+\theta {IR}_t+{\epsilon }_t,$$

(3)

where ${CE}_t$ denotes the private consumption share of GDP, ${DEBT}_t$ is the government debt-to-GDP ratio, and ${GE}_t$, ${TR}_t$ and ${IR}_t$ are control variables capturing fiscal stance and financial conditions. The disturbance term ${\epsilon }_t$ is assumed to be normally distributed with constant variance.

The intercept ${\mu }_t$ follows a random-walk process:

$${\mu }_t={\mu }_{t-1}+{\eta }_t,$$

(4)

which absorbs low-frequency movements and persistent shifts in consumption behaviour arising from structural change, demographic trends, or long-run macroeconomic developments. The coefficient on government debt, ${\beta }_t$, is also modelled as a random walk:

$${\beta }_t={\beta }_{t-1}+v_t$$

(5)

This allows the impact of debt financing on private consumption to evolve gradually over time. By contrast, the coefficients on government expenditure, tax revenue, and the interest rate are initially specified as constant parameters, as these variables serve primarily as conditioning controls rather than the central focus of the analysis.

The coefficient on government debt is allowed to vary over time because the paper’s central objective is to trace the evolving effect of debt financing on private consumption, while government expenditure, tax revenue, and the interest rate are included as conditioning controls. Keeping these latter coefficients constant preserves parsimony and facilitates identification of the debt channel.

State-Space Representation and Estimation

The model is cast in a state-space form, consisting of:

• A measurement equation linking observed consumption to the explanatory variables;
• State equations governing the evolution of the time-varying parameters.

Estimation is conducted within a Bayesian framework, which allows for coherent inference on latent states and parameter uncertainty. Weakly informative priors are specified for the initial states and variance parameters to avoid over-restrictive assumptions.

Posterior inference is obtained using Markov Chain Monte Carlo (MCMC) simulation methods, which generate draws from the joint posterior distribution of the parameters and latent states. From these posterior draws, we compute the time-varying path of the debt coefficient ${\beta }_t$ along with associated credible intervals. This enables a detailed assessment of how the impact of government debt on private consumption evolves across different macroeconomic regimes.

Bayesian estimation is conducted using weakly informative priors, which regularise the estimation without imposing strong restrictions on parameter dynamics. Posterior inference focuses on the evolution of ${\beta }_t$ with particular attention to whether its credible interval excludes zero (indicating crowding-out) or overlaps zero (consistent with Ricardian neutrality).

Conditioning on Fiscal Stance and Financial Conditions

Although the baseline specification captures the overall fiscal environment facing households, increases in government debt typically occur alongside changes in public expenditure, taxation, and interest rate conditions. To identify the independent effect of debt, an alternative specification is estimated in which government expenditure, tax revenue, and the interest rate are standardised prior to estimation, while the consumption shares and the debt-to-GDP ratio remain in their original units.

This conditioning exercise serves two purposes. First, it enhances numerical stability by limiting the influence of highly volatile fiscal series, particularly tax revenue, on the estimation results. Second, it allows for a clearer interpretation of the time-varying debt coefficient as the partial effect of debt, holding fiscal stance and financial conditions constant. Under this specification, movements in the time-varying debt parameter can be interpreted as evidence supporting or rejecting Ricardian behaviour net of contemporaneous fiscal policy actions.

Comparing the baseline and conditioned models, therefore, provides insight into whether any observed crowding out effects arise from debt accumulation itself or from the broader fiscal and macroeconomic conditions that accompany rising debt.

Constant-Parameter Bayesian Benchmark

To benchmark the time-varying results against a simpler framework, the analysis also estimates a constant-parameter Bayesian regression:

$${CE}_t={\mu }_t+\beta {DEBT}_t+\gamma {GE}_t+\delta {TR}_t+\theta {IR}_t+{\epsilon }_t,$$

(6)

This specification assumes a stable relationship between debt and consumption over the full sample period and abstracts from structural change and regime dependence. The resulting estimate of $\beta$ captures the average effect of government debt on private consumption.

Estimation is also conducted within a Bayesian framework, ensuring comparability with the TVP specification. The contrast between the constant-parameter and time-varying models allows us to assess whether ignoring parameter instability leads to misleading inference regarding fiscal transmission.

The constant parameter model is not designed to compete with the time-varying framework, but rather to illustrate the limitations of static specifications. An insignificant average effect of debt would suggest that periods of crowding out and periods of neutrality offset one another over time, thereby underscoring the importance of modelling evolving fiscal transmission mechanisms.

Inference and Interpretation

Inference throughout the analysis relies on posterior distributions and credible intervals rather than conventional hypothesis testing. Crowding out effects are identified when the posterior distribution of the debt coefficient lies predominantly below zero, whereas Ricardian-type behaviour is inferred when the credible interval includes zero. This Bayesian approach is particularly appropriate for long macroeconomic samples characterised by structural change and parameter instability.

The empirical framework is, therefore, structured to distinguish between debt-induced crowding out and Ricardian neutrality in a manner that remains robust to non-stationarity, regime shifts, and evolving fiscal institutions. The empirical findings are presented and discussed in Section 4.

Robustness: Markov-Switching Bayesian VAR (MSBVAR)

A potential concern is that the time-varying estimates may reflect unmodelled regime shifts rather than gradual changes in fiscal transmission. To evaluate whether the main results are sensitive to alternative forms of nonlinear fiscal dynamics, this study implements a Markov Switching Bayesian Vector Autoregression as a robustness exercise following Brandt and Freeman (2006). Whereas the baseline analysis relies on a Bayesian time-varying parameter framework, the Markov Switching model offers a complementary approach by permitting discrete shifts in fiscal and consumption dynamics. The model allows the data to switch probabilistically between latent fiscal regimes while maintaining a parsimonious dynamic structure.

The model is specified as

$$Y_t=C_{st}+{\sum }_{i=1}^p{A}_{i,s_t}{Y}_{t-i}+{ℇ}_{t,s_t}$$

(7)

• $Y_t$ is a vector including private consumption, government debt, and other macroeconomic variables;
• $s_t$ $\in \left\{1,2\right\}$ is an unobserved regime indicator;
• $C_{st}$ is a regime-dependent intercept vector;
• $A_{i,s_t}$ are regime-specific autoregressive coefficient matrices;
• ${ℇ}_{t,s_t}$ is a regime-dependent error term.

The regime variable $s_t$ follows a first-order Markov process, meaning that the probability of being in a given regime depends only on the previous period’s regime:

$$P\left(s_t=j/s_{t-1}=i\right)=p_{ij}i,j\in \left\{1,2\right\}$$

(8)

$$P=\left(\begin{array}{cc}{p}_{11}& p_{12}\\ p_{21}& p_{22}\end{array}\right), p_{11}+p_{12}=1, p_{21}+p_{22}=1.$$

(9)

This framework allows the economy to switch between distinct regimes—such as periods of low and high fiscal stress—thereby capturing nonlinear fiscal transmission mechanisms that may not be fully reflected in the TVP model.

In the estimated Markov Switching Bayesian Vector Autoregression, regime dependence is confined to the intercepts and the variance–covariance matrix of the shocks, while the autoregressive coefficients are constrained to remain constant across regimes. This parsimonious specification limits parameter proliferation and preserves numerical stability, while still permitting differences in mean dynamics and volatility across fiscal regimes.

Estimation is conducted under a conjugate Normal Wishart prior, which imposes shrinkage on the VAR coefficients and supports well-behaved posterior inference in a medium sample context. The model is estimated with lag order p equal to 1 to ensure parsimony and empirical stability. Higher order lag structures were explored but did not converge due to numerical instability in the regime switching likelihood optimisation. Consistent with the relevant literature, inference is, therefore, based on the parsimonious specification with p equal to 1. Posterior distributions are obtained using Gibbs sampling. An initial burn-in phase is discarded, and the remaining draws are used for inference. The model is estimated using standardised data to enhance numerical conditioning.

The combined use of a TVP model and an MS-BVAR provides a comprehensive framework for analysing fiscal transmission. The TVP model captures gradual evolution in the debt–consumption relationship, while the MS-BVAR captures discrete regime shifts. Together, these approaches allow us to distinguish between persistent structural change and episodic regime-dependent dynamics, thereby offering a richer empirical assessment of Ricardian behaviour and crowding-out effects in South Africa.

4. Results

4.1. Descriptive Statistics and Stylised Facts

Table 1 reports summary statistics for the variables used in the empirical analysis over the period 1960Q1 to 2025Q1. Private consumption is measured as a share of GDP, government debt as a percentage of GDP, government expenditure and tax revenue as shares of GDP, and the short-term interest rate in percentage terms.

First, private consumption (CE), measured as a share of GDP, exhibits moderate variability with a mean of approximately 4.54 and a standard deviation of 2.36. The relatively stable dispersion suggests that consumption dynamics, while responsive to macroeconomic conditions, do not display excessive volatility over the sample period. However, the presence of mild positive skewness indicates occasional upward deviations, likely associated with periods of fiscal expansion or income support measures.

Second, government debt (DEBT) displays substantial variability, with a mean of 51.02% of GDP and a very high standard deviation of 80.04. The wide range from below 1% to nearly 349% reflects significant structural changes in South Africa’s fiscal position over time, including periods of low indebtedness followed by sustained debt accumulation in the post-global financial crisis era. The high skewness and kurtosis further indicate that debt dynamics are characterised by episodic surges, consistent with fiscal stress episodes and macroeconomic shocks.

Third, government expenditure (GE) appears relatively stable compared to debt, with a mean of 15.3% of GDP and moderate dispersion. This suggests that while fiscal spending has evolved over time, it has done so in a more controlled and policy-driven manner, reflecting institutional constraints and budgetary frameworks.

Fourth, tax revenue (TR) exhibits extreme dispersion and pronounced non-normality, as indicated by its very high standard deviation and kurtosis. This reflects the presence of outliers and measurement issues in early periods, as well as structural shifts in revenue collection capacity and fiscal reporting. The large variation in TR underscores the importance of controlling for fiscal structure when examining consumption responses.

Finally, interest rates (IR) display moderate variability, with a mean of approximately 10.5% and relatively low dispersion compared to other variables. This suggests a degree of monetary policy stability over the sample period, although fluctuations likely correspond to inflation cycles and policy regime changes.

Overall, the descriptive statistics highlight the presence of non-linearities in the fiscal variables, particularly government debt and tax revenue. These features provide strong motivation for the use of a time-varying parameter framework, as constant-parameter models may fail to adequately capture the evolving relationships between fiscal policy and private consumption in the South African context. Fiscal variables also show considerable volatility, especially tax revenue, underscoring the importance of conditioning on fiscal stance when analysing consumption behaviour.

Figure 1 illustrates the evolution of the main variables over time and highlights three key features. First, the private consumption share follows a persistent downward trend, consistent with structural transformation and declining household expenditure shares. Second, the debt-to-GDP ratio remains relatively moderate during the earlier decades before rising sharply from the late 2000s, reflecting intensifying fiscal pressures. Third, government expenditure trends gradually upward, while the interest rate exhibits pronounced cyclical movements, particularly during episodes of monetary tightening. Tax revenue displays episodic volatility, especially in earlier decades, suggesting structural shifts in revenue mobilisation and data construction.

Taken together, the descriptive evidence supports the use of an empirical framework capable of accommodating non-stationarity, structural change, and evolving fiscal transmission mechanisms.

4.2. Time-Varying Effects of Government Debt on Private Consumption: Baseline Estimates

Figure 2 presents the baseline results from the Bayesian time-varying parameter model, estimated using unstandardised fiscal and macroeconomic controls. This specification reflects the combined fiscal environment faced by households, allowing both the intercept and the debt coefficient to evolve over time.

The findings indicate pronounced state dependence in the relationship between government debt and private consumption. In the early decades of the sample, spanning approximately the 1960s to the late 1980s, the estimated coefficient on government debt is consistently negative and statistically distinct from zero. During this period, increases in the debt-to-GDP ratio are associated with declines in the private consumption share, consistent with crowding out effects and precautionary behaviour in a context of weaker fiscal credibility and macroeconomic constraints.

From the early 1990s onward, the magnitude of the negative debt effect gradually declines. In the period after 2010, and especially after the mid 2010s, the estimated debt coefficient converges toward zero and its credible interval consistently includes zero. This pattern suggests that in the more recent fiscal environment, government debt accumulation no longer exerts a statistically meaningful negative effect on private consumption. Household behaviour during this phase appears broadly consistent with Ricardian type neutrality.

The wider confidence intervals observed in earlier periods reflect higher estimation uncertainty due to limited initial information and greater macroeconomic instability, a common feature of time-varying parameter models.

Taken together, the baseline time-varying parameter results demonstrate that the effect of government debt financing on private consumption is not constant over time. Earlier episodes of crowding out coexist with more recent periods characterised by debt neutrality.

4.3. Conditioning on Fiscal Stance: Standardised Controls

Although the baseline specification reflects the full fiscal environment accompanying debt accumulation, debt dynamics are often closely linked to changes in government spending, taxation, and financial conditions. To identify the independent effect of debt, Figure 3 presents results from an alternative time-varying parameter specification in which government expenditure, tax revenue, and the interest rate are standardised prior to estimation, while the consumption share and the debt-to-GDP ratio remain in their original units.

Under this conditioning exercise, the estimated debt coefficient is small, stable, and statistically indistinguishable from zero throughout the sample period. In contrast to the baseline results, there is no episode in which debt displays a statistically significant negative effect on private consumption once fiscal stance and financial conditions are controlled for in standardised form.

This evidence suggests that the crowding-out effects identified in earlier decades largely reflect the broader fiscal and macroeconomic environment associated with rising debt rather than debt accumulation itself. When households account for contemporaneous spending pressures, revenue developments, and interest rate conditions, government debt appears broadly neutral in its effect on consumption behaviour. The findings, therefore, align with a Ricardian interpretation in which neutrality emerges once fiscal policy channels are appropriately controlled for.

Altogether, the results in Figure 2 and Figure 3 indicate that any apparent crowding out effects operate mainly through broader fiscal and macroeconomic conditions, while debt in isolation does not exert an independent influence on private consumption.

4.4. Constant-Parameter Bayesian Benchmark

To benchmark the time-varying results against a simpler specification,

Table 2. Estimates from a constant parameter.

Parameter	Mean	Std. Dev.	2.50%	97.50%
Debt-to-GDP (β)	0.0007	0.001	−0.0010	0.0024
Government Expenditure/GDP (γ)	−0.6460	0.027	−0.6910	−0.6020
Tax Revenue/GDP (δ)	−0.00005	0.00005	−0.00014	0.00004
Interest Rate (θ)	0.0526	0.0218	0.0174	0.0893
Residual Std. Dev. (σ)	0.959	0.0421	0.894	1.03

presents estimates from a constant parameter Bayesian regression in which the effect of government debt on private consumption is assumed to remain fixed over the entire sample period.

The posterior mean of the debt to GDP coefficient is close to zero, and the corresponding 95 per cent credible interval clearly includes zero. This indicates that, on average, government debt does not exert a statistically meaningful effect on the private consumption share when time variation is disregarded. Importantly, this finding does not contradict the time-varying results. Instead, it reflects the averaging of earlier periods marked by pronounced crowding out effects and more recent periods characterised by near neutrality.

The contrast between the constant parameter benchmark and the time-varying estimates highlights the importance of modelling evolving fiscal transmission mechanisms. A static framework conceals substantial heterogeneity in household responses to debt financing and may lead to misleading conclusions regarding the relevance of Ricardian equivalence in the South African context.

4.5. Robustness Results: Markov-Switching Bayesian VAR

To evaluate whether the time-varying relationship between government debt and private consumption identified in the main analysis reflects discrete fiscal regimes rather than gradual parameter evolution, a Markov Switching Bayesian Vector Autoregression is estimated as a robustness exercise. This framework allows the economy to switch endogenously between two latent regimes governed by a first-order Markov process, while preserving a parsimonious dynamic structure.

The model includes private consumption, government debt, government expenditure, tax revenue, and the interest rate, estimated at quarterly frequency. Consistent with the main specification, the VAR is estimated with one lag to ensure stability and to avoid over-parameterisation in a regime-switching setting. All variables are standardised prior to estimation, and extreme observations in tax revenue are winsorised to limit the influence of outliers on regime classification.

Estimation is conducted within a Bayesian framework using a conjugate Normal Wishart prior. The autoregressive coefficients are held constant across regimes, while regime dependence is introduced through the latent state process and regime-specific shock variances. Posterior inference is obtained through Gibbs sampling, with an initial burn-in period discarded to ensure convergence.

Figure 4 and Figure 5 present the smoothed and posterior probabilities for Regime 2. Both series fluctuate narrowly around 0.5 with minimal dispersion, indicating weak regime separation and no prolonged episodes in which a single regime dominates. Under the parsimonious structure adopted here, where switching occurs in intercepts and volatility only, the data provide limited evidence of sharply distinct and persistent fiscal regimes.

Regime-specific impulse responses of private consumption to a government debt shock are shown in Figure 6. In both regimes, the median responses remain close to zero and are surrounded by wide credible intervals that include zero at most horizons. Posterior probabilities of a negative consumption response are broadly balanced across regimes, indicating no systematic or persistent crowding-out effect associated with government debt. Although minor quantitative differences appear at longer horizons, these differences lack statistical precision and are not economically meaningful. The interpretation, therefore, focuses on short to medium-term dynamics.

Appendix A reports the posterior mean transition probabilities. Both regimes display persistence, reflected in a high probability of remaining in the same state, while switches occur with meaningful but less frequent probability. Overall, the Markov Switching results reinforce the central conclusion of the study: government debt does not exert a robust or regime-specific crowding out effect on private consumption in South Africa. Any apparent nonlinearities in the debt consumption relationship reflect gradual adjustments rather than discrete fiscal regimes, supporting the findings from the time-varying parameter framework.

4.6. Discussion

The empirical findings of this study can be situated within the broader macroeconomic literature on Ricardian equivalence and household consumption behaviour. The Ricardian proposition maintains that households internalise the government budget constraint, implying that deficit financing has no first-order effect on aggregate consumption because individuals anticipate future taxation and adjust saving accordingly. In Barro’s formalisation of Ricardo’s insight, rational forward-looking agents treat government bonds as equivalent to future tax liabilities, resulting in debt neutrality in consumption decisions.

Empirical evidence, however, has been far less conclusive, particularly outside the restrictive assumptions of the canonical model. A substantial body of research argues that observed consumption responses depend on expectations, financial market imperfections, and heterogeneity in marginal propensities to consume. Micro-level estimates of consumption responses to fiscal transfers often suggest sizeable effects, yet aggregation to the macro level may introduce biases that overstate overall consumption multipliers. Once corrected for specification and aggregation effects, macro consumption responses tend to be considerably smaller. This literature supports a more nuanced interpretation in which the aggregate impact of debt-financed fiscal interventions depends critically on expectations formation and structural characteristics of the economy.

Recent empirical contributions further highlight the limitations of strict Ricardian assumptions in real-world settings. Survey and experimental evidence indicate that households frequently fail to fully incorporate future fiscal liabilities into present consumption decisions, reflecting partial or weak Ricardian behaviour driven by informational frictions or inattention. Even when explicitly informed about future tax implications, many households do not significantly adjust their spending responses, suggesting that behavioural factors may weaken the neutrality proposition.

In the South African context, empirical analysis of fiscal policy and household consumption remains relatively limited, which enhances the relevance of the present study. Existing macro time series evidence using standard econometric approaches has produced mixed conclusions regarding debt neutrality and crowding out. Studies in emerging markets frequently find limited support for strict Ricardian equivalence and identify negative effects of fiscal imbalances on private consumption, often attributing these outcomes to liquidity constraints, finite planning horizons, and financial market imperfections.

The results reported here integrate and extend these strands of evidence. In the baseline time varying parameter specification, the significant negative debt effects observed in earlier decades are consistent with periods of heightened macroeconomic uncertainty and fiscal stress, during which households may have adopted precautionary behaviour. By contrast, the convergence toward neutrality in the period after 2010 suggests that as fiscal credibility evolves and macro-financial structures mature, the independent effect of debt on consumption diminishes. This pattern aligns with the broader empirical observation that consumption responses to fiscal variables are state dependent rather than governed by a single stable relationship.

Moreover, the conditioning results demonstrate that once fiscal stance and financial conditions are explicitly controlled for, the estimated effect of debt becomes statistically indistinguishable from zero. This finding is consistent with the argument that consumption responses should be interpreted within the broader fiscal and macroeconomic environment rather than attributed solely to debt accumulation.

Taken together, the South African evidence supports the view that Ricardian behaviour is contextual and regime dependent, shaped by expectations, institutional credibility, and macroeconomic conditions. While strict Ricardian neutrality remains an important theoretical benchmark, the results presented here suggest that observed consumption behaviour reflects a more flexible and state-dependent adjustment process that bridges the gap between theoretical neutrality and empirical reality.

5. Conclusions

This paper examined whether government debt financing crowds out private consumption in South Africa or whether household behaviour is consistent with Ricardian equivalence. Using quarterly data from 1960Q1 to 2025Q1, the analysis employed a Bayesian time-varying parameter framework designed to accommodate non-stationarity, structural change, and evolving fiscal transmission mechanisms. The empirical strategy combined baseline time-varying estimates, a conditioning exercise to isolate the independent debt channel, and a constant parameter Bayesian benchmark.

Three main conclusions emerge. First, the relationship between government debt and private consumption is strongly state-dependent. The baseline time varying results show that increases in the debt to GDP ratio were associated with significant reductions in the private consumption share during earlier decades, consistent with crowding out or precautionary behaviour in periods marked by weaker fiscal credibility and macroeconomic instability. Over time, however, this negative relationship weakens and becomes statistically indistinguishable from zero in the period after 2010, indicating a shift toward behaviour broadly consistent with Ricardian neutrality.

Second, once fiscal stance and financial conditions are explicitly controlled for by standardising government expenditure, tax revenue, and the interest rate, the estimated effect of government debt is consistently small and statistically insignificant across the sample. This suggests that earlier crowding-out effects largely reflect the broader fiscal and macroeconomic environment accompanying debt accumulation rather than debt itself. Household consumption appears more responsive to spending dynamics, revenue performance, and financing conditions than to the level of public debt in isolation.

Third, the constant parameter benchmark underscores the importance of allowing for time variation. A static specification produces an average debt effect close to zero, thereby masking substantial heterogeneity across historical periods. This illustrates how constant coefficient models may yield inconclusive or misleading conclusions regarding Ricardian equivalence when applied to long samples characterised by structural change.

To address the possibility that the time-varying results reflect discrete regime shifts rather than gradual evolution, the paper estimated a two-regime Markov Switching Bayesian VAR as a robustness check. The regime probabilities display weak separation, with no sustained dominance of either regime, and regime-specific impulse responses do not reveal a systematic or statistically robust crowding-out effect of government debt on private consumption. These findings indicate that the main results are not driven by discrete fiscal regimes, but rather by gradual adjustments in fiscal conditions and stable consumption behaviour, thereby reinforcing the validity of the time-varying framework.

The findings carry several implications for fiscal policy in South Africa. First, concerns about debt mechanically crowding out private consumption should be treated with caution. The evidence suggests that in the current fiscal environment, debt accumulation does not automatically depress consumption. Household responses depend on the broader policy mix and macro-financial context in which debt evolves. Fiscal consolidation focused solely on reducing the debt ratio may, therefore, have limited effects on consumption unless supported by credible signals regarding expenditure priorities, revenue strategy, and financing conditions.

Second, the results highlight the importance of fiscal credibility and policy composition. When debt accumulation is perceived as part of an unsustainable or uncertain trajectory, households are more likely to reduce consumption defensively. When fiscal policy is embedded within a credible medium-term framework, debt appears largely neutral. Strengthening fiscal institutions, improving transparency, and anchoring expectations through coherent medium-term strategies may, therefore, be more effective in supporting consumption than debt reduction alone.

Third, the analysis emphasises the interaction between fiscal and monetary policy. Interest rate conditions play a significant role in shaping household decisions. Policies that stabilise financing conditions and limit the transmission of fiscal stress to borrowing costs may help mitigate adverse consumption responses even when debt levels are elevated.

Although this study provides new evidence on the evolving nature of debt and consumption dynamics in South Africa, several extensions merit further research. Future work could incorporate household-level data to examine heterogeneity across income groups or liquidity-constrained households, or model expectations using survey-based measures. Allowing for richer nonlinear dynamics or interactions with growth and labour market conditions may also deepen understanding of the macroeconomic consequences of debt financing.

In conclusion, whether government debt crowds out private consumption in South Africa cannot be summarised by a single time-invariant coefficient. The impact of debt financing depends on fiscal credibility, macroeconomic conditions, and the broader policy environment. These factors should remain central to both empirical analysis and fiscal policy design.