Credit, inflation, and the supply side: a theory and global evidence from a new LRAS-leaning barometer

The dual nature of credit remains one of the most enduring paradoxes in macroeconomics. On one hand, financial deepening is celebrated as the engine of modern growth, a Schumpeterian force (Schumpeter, 1983) that mobilizes savings, facilitates intertemporal smoothing, and funds the capital accumulation necessary for development (King and Levine, 1993; Levine, 2005). On the other hand, rapid credit expansion is frequently identified as a predictor of financial fragility, asset bubbles, and deep recessions (Schularick and Taylor, 2012; Mian and Sufi, 2014). This dichotomy of credit as a creator of wealth versus a harbinger of crisis has generated two large but largely separate literature: the growth literature emphasizes long-run convergence, while the macro-finance literature focuses on leverage cycles, economic bubbles, and short-run volatility.

This paper bridges these perspectives by revisiting a fundamental yet under-explored question: under what conditions does credit expansion shift the Long-Run Aggregate Supply (LRAS) curve positively, and when does it merely act as a demand stimulus?

While classical growth theory posits that savings determine the capital stock (Solow, 1956), modern monetary systems allow banks to create purchasing power (Werner, 2016; McLeay et al., 2014). The macroeconomic consequence of this credit creation depends on its allocation. If new purchasing power is directed toward productive investment, it expands capacity while exerting disinflationary pressure. However, if credit flows into consumption, speculative asset markets, or low-productivity construction, it generates excess demand without a commensurate increase in output, resulting in inflation or asset price bubbles (Turner, 2015). This mechanism is amplified in an open global economy where demand shocks can be absorbed through imports. Despite the theoretical clarity of this distinction, empirical work has struggled to disentangle these channels globally. Most studies aggregate credit into a single variable, the credit-to-GDP ratio, thereby obscuring the heterogeneity of credit's real effects.

We address this limitation by proposing a theoretical framework and a new empirical metric to separate the supply-side and demand-side effects of credit shocks. We argue that the quality of a credit boom is observable in the dynamic covariance of prices and quantities. A supply-oriented credit regime should display a strong investment response and muted (or negative) inflation response, whereas a demand-oriented regime is characterized by rising prices and weak capital formation.

To operationalize this intuition, we introduce the LRAS-Leaning Barometer (LLB). The LLB is defined as the ratio of the cumulative multi-year response of real investment (Gross Fixed Capital Formation) to the cumulative response of inflation following a credit impulse. This measure provides a continuous diagnostic of credit quality. Unlike static debt thresholds that warn only of quantity risks, the LLB captures the allocative efficiency of the financial sector. A positive LLB signals that the financial system is successfully relaxing supply constraints; a negative LLB indicates that credit creation is outpacing the economy's productive absorptive capacity.

We apply this framework to a panel of 266 advanced and developing economies spanning 1995–2024. Using local projection methods (Jordà, 2005) to trace the impulse responses of investment and inflation, we document three stylized facts.

First, we challenge the conventional assumption that rapid credit expansion is inherently inflationary. Globally, the pass-through from credit impulses to consumer price inflation is statistically weak in the short run and modest in the medium run. In capital-scarce regions such as Sub-Saharan Africa and parts of Eastern Europe, positive credit shocks are associated with a decline in inflation. This supports the “working capital” hypothesis: in developing economies, credit relaxes supply constraints and reduces the unit cost of production.

Second, we document substantial heterogeneity in the conversion of finance into capital. While the average credit multiplier for investment is positive, the efficiency of this transmission varies systematically across regions. Developing Asia and Western Europe display the highest conversion rates of credit into capital formation. In contrast, in Latin America and the Middle East, credit shocks often dissipate into price pressures or consumption rather than durable capital accumulation. These patterns suggest that financial depth alone is insufficient for growth; the institutional capacity to allocate credit productively remains the key constraint.

Third, we identify a global “credit-regime cycle.” During the 2000s the global average LLB was positive, indicating that financial deepening largely supported real capacity expansion. After 2010, however, the LLB turned negative, suggesting that recent credit cycles have become increasingly decoupled from the real economy and more closely associated with nominal demand or asset-price dynamics. This pattern points to a structural shift in the global financial environment in which the marginal unit of debt yields diminishing real returns.

Crucially, we argue that the divergence in credit outcomes reflects institutional variety. Our theoretical model introduces a structural leakage parameter governing credit allocation. In economies with strong collateral registries, effective bankruptcy procedures, and disciplined banking supervision, this leakage is low and credit finances long-duration capital projects. In weaker institutional environments, credit leaks into speculative activity or short-term demand pressures. Evidence from transition economies in Eastern Europe illustrates a catch-up dynamic in which credit acts as a powerful supply-side lever.

The contribution of this paper is threefold. First, we formalize the distinction between productive and demand-pull credit in a tractable two-period model where allocative frictions determine the macro-financial trade-off. Second, we introduce the LRAS-Leaning Barometer as a practical indicator of the supply-side orientation of credit expansions. Third, we provide a comprehensive global assessment of the inflation–investment trade-off associated with financial deepening.

The remainder of the paper proceeds as follows. Section 2 reviews the related literature. Section 3 presents the theoretical framework. Section 4 describes the data and empirical strategy. Section 5 reports the main results and robustness checks. Section 6 concludes.

Our research bridges two distinct, and often contradictory, strands of the macro-financial literature: the development view, which posits finance as a driver of long-run convergence, and the macro-stability view, which identifies credit cycles as the primary architect of financial crises. By focusing on the supply-side elasticity of credit expansions, we offer a unified framework that reconciles these diverging narratives.

First, we contribute to the foundational debate on the finance-growth nexus. A seminal body of work, anchored by King and Levine (1993) and Levine (2005), argues that financial deepening fosters growth by reducing information asymmetries, mobilizing savings, and facilitating risk management, with more recent evidence linking financial development to productivity gains through capital accumulation channels (Dwivedi, 2026). In this Schumpeterian paradigm, credit is implicitly assumed to fund productive activity. However, the post-2008 literature has challenged the linearity of this relationship. Arcand et al. (2015) and Cecchetti and Kharroubi (2012) document a “too much finance” effect, where credit growth becomes a drag on productivity beyond certain thresholds. We advance this literature by demonstrating that the non-linearity of finance is driven not merely by the volume of debt, but by its composition. Our findings suggest that the “vanishing effect” of financial deepening in advanced economies arises because credit increasingly leaks into non-productive demand channels rather than shifting the LRAS curve positively.

Second, our work intersects with the literature on credit booms, leverage cycles, and financial instability. Research in the tradition of Minsky (1986), formalized empirically by Schularick and Taylor (2012) and Jordà et al. (2013), established that rapid credit expansion is a strong predictor of financial crises, reflecting the central role of bank lending dynamics in macroeconomic fluctuations (Dalla et al., 2024). Mian and Sufi (2014) further emphasize the demand-side channel, showing how household leverage fueled by asset price appreciation leads to severe consumption collapses. We qualify these findings by introducing significant heterogeneity into the transmission mechanism. While the consensus view treats credit booms as homogeneous demand shocks, our new LLB indicator reveals that distinct credit regimes exist. We show that credit booms are not universally destabilizing; when they are successfully channelled into gross fixed capital formation as observed in parts of Asia and Eastern Europe, they result in sustainable capacity expansion rather than inflationary boom-bust cycles.

Third, we engage with the emerging Quantity Theory of Credit and the disaggregation of credit flows. Werner (2012) and Turner (2015) argue theoretically that credit for GDP transactions (investment and consumption) has fundamentally different macroeconomic implications than credit for financial transactions (asset markets and real estate). Bezemer et al. (2018) provides empirical support for this, showing that credit flows to the FIRE (Finance, Insurance, Real Estate) sector retard growth, while flows to the non-financial sector promote it, consistent with recent evidence that financial misallocation can reduce innovation and productive efficiency (Shah et al., 2025). Our paper provides the first comprehensive global test of this hypothesis using dynamic impulse responses. By constructing the LLB, we provide a continuous metric of allocative efficiency that captures the Turner (2015) distinction between productive and speculative credit without requiring granular sectoral loan data, which is often unavailable for developing economies.

Finally, methodologically, we build upon the local projection (LP) framework of Jordà (2005). The flexibility of LPs has made them the standard for estimating state-dependent fiscal and monetary multipliers (Auerbach and Gorodnichenko, 2012; Ramey, 2016). We extend this framework to the measurement of structural supply-side shifts. By utilizing the ratio of cumulative impulse responses (investment versus inflation), we effectively estimate a dynamic sacrifice ratio for credit expansions, linking credit dynamics to broader macroeconomic outcomes associated with inflationary pressures (Nguyen, 2025). This allows us to move beyond static early-warning indicators toward a more refined diagnostic tool that assesses the real-economy trade-offs of financial deepening in real time.

To structure our empirical analysis, we develop a two-period, probabilistic general equilibrium model with heterogeneous investment technologies. This framework formalizes the mechanism through which credit misallocation generates the inflation–investment trade-off captured by the LLB. Unlike standard financial accelerator models where credit constraints distort the volume of investment (Bernanke and Gertler, 1986), we model a friction that distorts the composition of investment.

Consider a closed economy populated by a continuum of risk-neutral entrepreneurs and households. Time is discrete, t ∈ {1, 2}.

Households. Households are endowed with initial wealth W₁ and supply inelastic labor N. They choose between current consumption C₁ and savings S₁ (deposited in banks) to maximize expected utility:

where β is the discount factor. Household budget constraints are:

where r is the deposit rate and w is the wage rate.

Entrepreneurs and Technology. Entrepreneurs have no initial wealth and require external credit D_t to operate. The economy features two distinct uses for credit:

1. Productive Technology (Supply-Shifting). Entrepreneurs invest in capital K, which produces output in period 2 according to:

where A denotes total factor productivity. This technology shifts positively the LRAS curve.

1. Unproductive - Speculative Technology (Demand-Pull). Credit can also be directed toward a storage - speculative technology (e.g. real estate bidding, inventory speculation, import consumption). This yields return R_u but produces no new output capacity:

Spending on this technology contributes to period-1 aggregate demand but not to future supply.

In a frictionless Modigliani–Miller environment (Modigliani and Miller, 1958), credit flows exclusively to the highest-return technology. We introduce an allocative friction motivated by information asymmetries, weak institutions, or political lending. Banks cannot perfectly screen borrowers' intended use of funds.

Let ΔL denote an exogenous positive credit shock (a “credit impulse”). We posit a leakage parameter Λ ∈ [0, 1] governing how much of the additional credit fails to translate into productive investment. This parameter aggregates:

1. Consumption leakages ψ,

2. Misallocation to low-productivity sectors ϕ.

The leakage parameter therefore summarizes two conceptually distinct channels through which credit fails to expand productive capacity. The first channel reflects borrowing that finances current consumption or import demand, which raises short-run spending but does not contribute to capital formation. The second channel reflects speculative or low-productivity uses of credit, such as real estate bidding or asset trading, where lending reallocates existing assets rather than generating new output. For tractability, the model aggregates these channels into a single parameter Λ, capturing the overall allocative efficiency of the financial system. Empirically, the LLB therefore measures the combined macroeconomic consequences of these two forms of credit leakage rather than attempting to distinguish them separately.

Effective capital formation is:

where P₁ denotes the investment-goods price index. The remaining fraction ΛΔL constitutes excess demand:

which fuels nominal spending without expanding productive capacity.

Prices P₁ adjust flexibly to equate aggregate demand and supply in period 1.

Demand Channel. The credit shock raises nominal aggregate demand:

Because leakage-funded activity increases spending on consumption or speculative assets.

Supply Channel. The period-2 LRAS shifts only to the extent that credit funds productive capital:

Inflation Dynamics. Inflation arises from the mismatch between the nominal demand expansion and the real supply response:

Substituting the supply response yields:

Where

Is the marginal efficiency of capital.

We define the theoretical counterpart to our empirical estimator. The LRAS-Leaning Barometer measures the ratio of the real-investment elasticity to the price elasticity with respect to credit:

1. Case 1: Efficient Allocation (Λ → 0). Credit flows fully into productive capital. Supply expands, inflationary pressure is minimal, and the ratio

1. Case 2: Pure Demand Shock (Λ → 1). Credit is fully misallocated.

Inflation is maximal because demand rises while supply does not.

Proof. Differentiating (3) and (5) with respect to Λ yields ∂K/∂Λ < 0 and ∂π/∂Λ > 0. Thus the ratio decreases in Λ.

This proposition links theory to empirics: cross-country and cross-decade variation in the estimated LLB reflects variation in allocative efficiency Λ, which captures the combined influence of consumption-driven borrowing and speculative credit allocation, shaped by institutions, financial supervision, and the prevalence of speculative credit channels.

To analyze the global elasticity of supply to credit expansion, we construct a comprehensive unbalanced panel covering 266 advanced, developing, and transition economies over the period 1995–2024. The primary source of data is the World Bank's World Development Indicators (WDI), supplemented by the IMF International Financial Statistics (IFS) for specific monetary aggregates.

The central independent variable in our analysis is the credit impulse (ci_i,t). Following the macro-finance literature (Biggs et al., 2009), we define the credit impulse as the change in the flow of new credit relative to GDP, rather than the stock of debt. Theoretical work suggests that while the level of debt (D_t) relates to stock adjustments and solvency, it is the acceleration of lending that drives aggregate demand and new capital formation.

We calculate the credit impulse as the second difference of the nominal credit stock normalized by GDP, or equivalently, the change in the credit-to-GDP ratio:

where Credit represents domestic credit provided to the private sector. A positive impulse represents a net injection of purchasing power by the banking sector in excess of GDP growth.

Our two primary dependent variables capture the trade-off between the price level and real capacity: i. Inflation (π_i,t): Measured as the annual percentage change in the Consumer Price Index. ii. Investment (I_i,t): Measured as the annual growth rate of Gross Fixed Capital Formation (GFCF) in real terms. For robustness in long-run specifications, we also utilize the level of real GFCF.

To mitigate omitted variable bias and isolate the supply-side channel, we include a vector of macroeconomic controls Z_i,t that includes: (1) Real GDP Growth to control for business cycle demand effects; (2) Trade Openness (exports plus imports as a share of GDP) to control for imported inflation and external demand shocks; (3) The Current Account Balance to control for capital flow reversals; and (4) Real Lending Rates to capture the monetary policy stance.

Given our focus on institutional variety, we also utilize the World Bank's Country Policy and Institutional Assessment (CPIA) cluster averages to proxy for Institutional Quality, specifically measuring public sector management and property rights enforcement.

Given the heterogeneity of the sample, which includes hyper-inflationary episodes (e.g. Zimbabwe, Venezuela) and post-conflict transition data, outlier management is essential to prevent localized volatility from distorting global estimates. We therefore winsorize all continuous variables at the 1st and 99th percentiles. This procedure ensures that extreme observations do not disproportionately influence the estimated credit responses.

Table 1 presents the summary statistics for the main variables used in the estimation. The sample exhibits significant variation, with inflation ranging from deflationary episodes in advanced economies to double-digit rates in developing regions, providing the necessary variance to identify the credit transmission mechanism.

Figure 1 provides a first descriptive look at the co-movement of credit, inflation, and investment in the world economy from 1995 to 2024. Three patterns stand out. First, global credit impulses are remarkably stable and mean-reverting, showing no evidence of a persistent upward trend in net new credit creation. Second, inflation exhibits a similarly muted long-run pattern, with most of the volatility concentrated in short episodes around global shocks (1998, 2008, 2021). This visually confirms that, at the aggregate global level, credit expansions are only weakly inflationary; a result that our empirical estimates will formalize later. Third, investment growth (GFCF), in contrast, displays substantial year-to-year variation, especially in emerging-market heavy years, indicating that global capital formation is much more sensitive to cyclical conditions than either credit or prices. Taken together, the figure highlights the central motivation of this paper: the relationship between credit and macroeconomic outcomes cannot be inferred from levels alone, but rather from the balance between productive and inflationary credit channels, which varies meaningfully across time, regions, and institutional environments.

We take the theoretical predictions to the data using an unbalanced panel of 266 countries. Our primary econometric tool is the Local Projection (LP) method (Jordà, 2005), which allows us to estimate the cumulative response of macroeconomic variables to credit shocks without imposing rigid restrictions on the dynamic structure of the data.

We estimate the following specification for horizon h = 1, …, 5:

where X_i,t denotes the dependent variable of interest (either the log of the Consumer Price Index or log Gross Fixed Capital Formation). CI_i,t represents the credit impulse, defined as the change in the credit-to-GDP ratio. The vector Z_i,t includes a robust set of control variables capturing contemporaneous macroeconomic conditions, including lagged GDP growth, the current account balance, trade openness, institutional quality, and lending rates. Country (α_i) and year (γ_t) fixed effects control for time-invariant heterogeneity and global common shocks.

A potential concern in macro-financial studies is that credit expansion may respond to anticipated economic growth, which could introduce reverse causality. Several features of our empirical design mitigate this issue. First, the credit impulse measures the acceleration of lending rather than the level of credit, capturing unexpected shifts in credit creation rather than gradual balance sheet adjustments. Second, the local projection framework estimates forward responses of macroeconomic variables to contemporaneous credit impulses while controlling for lagged macroeconomic conditions, including GDP growth and external balances. Finally, the inclusion of country and year fixed effects absorbs persistent institutional characteristics and common global shocks that may simultaneously influence credit supply and economic activity. Taken together, these elements help isolate the macroeconomic responses to credit impulses rather than the endogenous expansion of credit driven by expected growth.

From these estimates, we construct the LRAS-Leaning Barometer as the ratio of the cumulative investment response to the cumulative inflation response over a three-year horizon:

This ratio provides a scalar summary of the credit regime. A value significantly greater than zero implies that credit shocks primarily feed into capital formation. A value near zero or negative implies that credit shocks dissipate into price increases or have negligible real effects.

This section presents the core empirical findings. We proceed in three steps to map the global distribution of credit outcomes. First, we utilize local projections to trace the dynamic transmission of credit impulses into inflation and real investment for the full global sample. Second, we disaggregate these effects by income group and geographic region to identify structural heterogeneity in the credit-supply nexus. Finally, we calculate the LLB across decades, documenting the secular shift in the allocative efficiency of global finance.

We begin by testing the central tension of our theoretical framework: does the average credit shock manifest primarily as a demand shifter (raising prices) or a supply shifter (raising capacity)?

Table 2 reports the impulse response functions (IRFs) for CPI inflation over a five-year horizon. The results provide striking evidence against the purely monetarist view that credit expansion is intrinsically inflationary. The coefficient on the credit impulse (ci) is statistically indistinguishable from zero on impact (h = 1) and remains insignificant through the second year. It is only in year three that we observe a statistically significant pass-through, with a coefficient of 0.041.

Economically, this magnitude is trivial: a one standard deviation increase in the credit impulse raises inflation by merely 0.04% points three years later, before the effect dissipates. This sluggish transmission suggests that, on a global average, the immediate liquidity injection is largely absorbed by the real economy or asset markets rather than goods prices.

In contrast, the real economy response is robust. Table 3 presents the response of Gross Fixed Capital Formation (GFCF) growth. While the impact effect is noisy, reflecting the time-to-build lags inherent in capital projects, the cumulative response is economically substantial. The coefficient turns positive and quantitatively large by year five. This establishes our first stylized fact: unconditionally, the global credit cycle leans toward the supply side. However, these global averages mask a profound bifurcation between advanced and developing economies, to which we now turn.

Figure 2 summarizes the dynamic responses of inflation and investment to a one-period credit impulse using local projections. The left panel shows that the effect of credit on inflation is quantitatively small and statistically indistinguishable from zero at all horizons: the point estimates remain close to zero and the confidence intervals consistently span both sides of the axis. This provides visual confirmation that, at the global level, credit expansions are not inherently inflationary. In contrast, the right panel reveals a markedly different pattern for investment. Although short-run responses fluctuate, the medium-run effects become increasingly positive, and by horizons four and five the investment response is sizeable and economically meaningful. The confidence intervals widen due to cross-country heterogeneity, but the upward-sloping profile indicates that credit-financed investment takes time to materialize, consistent with capital adjustment frictions and multi-year project cycles. Together, these two panels highlight the core mechanism underlying the LRAS-Leaning Barometer: credit tends to exert only weak short-run demand pressure on prices, while its supply-side effects accumulate gradually and dominate at medium horizons in many economies.

In Table 4, we formally test for structural breaks between income groups. The results reveal a stark dichotomy in the function of credit.

For developing economies, the credit impulse coefficient for inflation is negative and statistically significant (−0.366). This counter-intuitive finding provides strong support for the “working capital” channel: in capital-scarce environments, access to credit relaxes supply-chain bottlenecks and lowers the marginal cost of production, effectively shifting the short-run aggregate supply curve outward. Simultaneously, the investment response is powerful: a 1% point credit impulse raises GFCF growth by 3.0% points. Taken together, this implies that for the Global South, credit is primarily a production input rather than a consumption smoother.

In contrast, for advanced economies, the link between credit and the real economy is tenuous. The inflation coefficient is effectively zero, and while the investment coefficient is positive (1.590), it is statistically insignificant and half the magnitude of the developing world. This suggests that in the developed world, the marginal unit of credit is increasingly absorbed by non-productive activities, potentially financing asset transfers or household leverage, consistent with the “too much finance” hypothesis.

Table 5 presents the subgroup analysis per each continent of the world, which allow us to map the “geography of credit efficiency.” These results offer a detailed explanation for the differential growth trajectories of the Global North and South.

Countries in the Sub-Saharan African continent displays a remarkable sensitivity to credit shocks, with Credit impulse (ci) coefficient of −23.325 for GFCF growth. This signals a high negative marginal product of capital; when credit is available, it is immediately deployed into demand-pull effect. However, it is theMiddle East and North African region countries exhibit the most sustainable capital deepening, as evidenced by the highly significant positive coefficients on the GFCF growth of 14.933. In our theoretical framework, this corresponds to a low leakage parameter (Λ), where institutional quality ensures credit finances long-duration assets.

Conversely, the results for Eastern Europe states and Eastern and Central Europe and North American nations reveal the fragility of their credit regimes. While Eastern Europe shows a weak investment response, it is accompanied by a significant disinflationary effect (−0.953), typical of supply-side catch-up growth. North America, however, lacks a significant estimation for the GFCF level, suggesting that its credit booms are often “shallow,” financing consumption or short-lived cycles rather than permanent shifts in the LRAS curve.

Figure 3 provides a decade-level comparison of the LRAS-Leaning Barometer across major world regions, revealing two important patterns. Panel 3a shows that developing and advanced economies broadly follow the same global “credit-regime cycle”: the 2000s were the only decade with strongly positive LLB values,indicating supply-enhancing credit, whereas the 2010s and early 2020s are dominated by negative values, consistent with demand-driven or speculative credit dynamics. However, the amplitude differs markedly: developing economies exhibit deeper negative cycles in both the late 1990 and 2010s, reflecting vulnerabilities to misallocation, commodity boom–bust cycles, and weaker financial supervision.

Panel 3b highlights that the distinct trajectory of Eastern and Central Europe relative to the rest of the world. These economies converted credit into productive capital more effectively during the 2000s, consistent with post-transition financial deepening, EU convergence reforms, and improved banking regulation. Yet the sharp decline of the LLB in the 2010s indicates a reversion toward more demand-driven credit, likely associated with deleveraging after the global financial crisis, rising real-estate cycles, and slower structural reform momentum. Together, these patterns underscore the importance of institutional quality and credit allocation mechanisms in shaping whether credit expansions translate into sustainable supply-side gains or merely fuel temporary demand imbalances.

The results for Eastern and Central Europe are of particular interest regarding the transition experience. The region exhibits a statistically significant disinflationary response to credit (−0.953) alongside insignificant real investment drop. This suggests that the financial deepening process in transition economies has successfully matured from the instability of the 1990s into a functional mechanism for relaxing supply-side constraints. This contrasts sharply with North America, where similar credit impulses fail to generate permanent capital deepening, pointing to persistent structural rigidities in the latter's financial architecture.

Importantly, the consistency of results across income groups and continental subsamples suggests that the estimated credit responses are not driven by a small number of extreme observations but reflect systematic differences in credit allocation across institutional environments.

Finally, we synthesize these findings into the LLB, presented in Table 6. The temporal evolution of the LLB provides a quantitative history of global financialization.

The 1990s, characterized by the Asian Financial Crisis and Latin American volatility, show negative LLB values for developing economies (−2.03), confirming that credit during this era was largely destabilizing. The 2000s, however, stand out as a “Golden Age” of productive finance. The positive LLB values across both Advanced (2.80) and Developing (2.68) economies indicate a period where global liquidity was successfully funneled into capacity expansion, the “Great Moderation” was underpinned by a supply-side credit expansion.

Crucially, our data identifies a structural break following 2010. In the post-GFC decade, the LLB collapses to −4.09 for developing economies and −2.65 for advanced economies. This deterioration persists and deepens into the 2020s. This signals a fundamental regime change that the marginal unit of credit is no longer generating productive capacity but is instead increasingly leaking into inflation or asset prices. This “Great Decoupling” of finance from the real economy represents a critical policy warning like without structural reform to the allocation of credit, future financial deepening is likely to be stagflationary rather than growth-enhancing.

An important question is whether the post-2010 decline in the LLB reflects weaker investment responses to credit or stronger inflationary responses. Because the barometer is defined as the ratio of cumulative investment responses to inflation responses, a decline could in principle arise from either channel. Examining the underlying impulse responses reveals that the post-2010 deterioration is primarily driven by stronger inflation responses to credit shocks rather than a collapse in investment responses. In other words, credit expansion increasingly translates into price pressures while its ability to generate additional capital formation has remained comparatively stable. This pattern is consistent with a gradual shift of credit toward demand-driven or speculative uses rather than productive investment.

Figure 4 summarizes the full cross-regional heterogeneity in credit regimes over the past 3 decades. A common pattern is the “golden decade” of the 2000s, when most regions, especially Western Europe, Asia–Pacific, MENA, and Sub-Saharan Africa, display positive LLB values, suggesting that credit expansions were, on average, associated with stronger capital formation than with inflationary pressure. By contrast, the 2010s mark a clear shift toward inflationary or misallocated credit in several regions that North America, MENA, Latin America, and parts of Sub-Saharan Africa exhibit strongly negative LLBs, consistent with post-crisis deleveraging, commodity booms and busts, and rising real-estate and consumption credit. Eastern Europe sits in between these extremes that its LLB is mildly positive in the 2000s, reflecting successful post-transition financial deepening, but hovers around zero or slightly negative thereafter, indicating a more balanced but fragile mix of productive and demand-driven credit. Overall, the figure highlights that the same global liquidity waves translate into very different macro outcomes depending on regional institutions and the efficiency of credit allocation, a central theme of our analysis.

This paper has revisited one of the oldest questions in macroeconomics, the relationship between money, credit, and physical capital through a novel supply-side lens. By developing a theoretical framework that explicitly models the allocative friction of credit markets, and by introducing the LRAS-Leaning Barometer to measure it empirically, we have proposed a resolution to the “Schumpeter-Minsky paradox.”

We argue that credit is neither inherently a source of growth nor inherently a source of instability; rather, its macroeconomic character is determined entirely by its transmission elasticity specifically, whether liquidity absorption occurs through quantity adjustments (capital formation) or price adjustments (inflation). Our analysis of 266 economies over 3 decades yields three findings that fundamentally challenge the “credit view” consensus.

First, we document that the global inflationary pass-through of credit expansion is remarkably weak. In many developing economies, credit impulses act as disinflationary supply shocks, suggesting that access to finance relaxes working capital constraints and lowers marginal costs more rapidly than it stimulates aggregate demand. Second, we uncover a stark divergence in allocative efficiency.

The “Asian Miracle” and the stability of core European economies appear underpinned by a structural capacity to convert liquidity into gross fixed capital formation. In contrast, Latin America and parts of the MENA region exhibit a high “leakage parameter”, where credit deepening dissipates into price instability rather than capacity expansion. Third, and perhaps most worryingly, our temporal analysis reveals a secular deterioration in the quality of global credit.

While the 2000s were characterized by a positive global LLB, indicating a synergy between finance and the real economy, the post-2010 and post-pandemic eras have seen the LLB turn negative. This signals a structural decoupling where the marginal unit of debt is increasingly fueling asset prices, consumption, or refinancing rather than new productive capacity. This supports the hypothesis of diminishing returns to financialization in the modern era.

These findings have immediate implications for monetary and macro-prudential policy. Central banks operating in high-LLB environments (where credit creates supply) should be more tolerant of rapid credit growth, even if it exceeds standard speed limits, as this credit is disinflationary. Conversely, in low-LLB environments, monetary tightening may be required earlier, as credit is purely demand-pull. Policy must pivot from monitoring the volume of credit (e.g. Credit-to-GDP gaps) to monitoring the composition of credit. The LLB offers a tractable, real-time metric for this purpose. Regulators should consider differentiated capital requirements that penalize “high-leakage” lending (speculative real estate, import-based consumption) while easing constraints on industrial capacity-enhancing lending (manufacturing, infrastructure). For developing economies with lower LLBs, the priority is not simply more access to finance, but institutional reforms to reduce screening costs and improve collateral registries, thereby lowering the friction Λ (leakage) in our theoretical model.

The central implication is that financial development should not be evaluated solely by the quantity of credit but by the institutional mechanisms that govern its allocation toward productive investment.

Future Research This paper opens several avenues for future inquiry. First, granular loan-level data could better identify the specific sectors responsible for the “leakage” identified in our aggregate results. Second, research should examine whether digital finance and Fintech are improving the LLB by reducing information asymmetries, or worsening it by facilitating easier consumption credit. Finally, determining the causal drivers of the post-2010 decline in credit quality, whether due to regulatory arbitrage, secular stagnation, or the rise of intangible assets, remains a critical task for the next generation of growth empirics.

The author acknowledges financial support in the form of a PhD scholarship from Adelaide University.