Financial Intermediation and Provincial Economic Activity in a Dollarised Economy: Panel VAR Evidence from Ecuador

Abstract

In dollarised economies, the absence of autonomous monetary policy shifts the burden of macroeconomic adjustment onto the banking system, where deposits and credit constitute the principal channel through which liquidity is conveyed to commercial activity. The literature has documented this relationship using aggregate national data, yet its behaviour at the monthly provincial scale remains underexplored for Latin America, particularly in fully dollarised economies and over recent periods marked by severe shocks. This article addresses that gap for Ecuador using a monthly panel of its 24 provinces over 2019–2025, estimated as a Panel VAR by two-step GMM, with monthly sales declared to the Internal Revenue Service used as a high-frequency indicator of provincial economic activity. The pandemic is incorporated as an exogenous control. The theoretical framework combines the supply-leading hypothesis, the credit-channel literature on transmission lags arising from financial frictions, and financial intermediation theory on liquidity and asset transformation. The system exhibits a predominantly supply-leading dynamic: deposits and credit retain predictive capacity over provincial sales, with no robust evidence of reverse feedback. Transmission speed is heterogeneous across channels. Deposits affect sales with a one-period lag, whereas credit requires an additional period—a pattern consistent with the differential role of each channel in banks’ asset-transformation function. The provincial-scale evidence for a dollarised economy shows that the macroeconomic relevance of financial intermediation depends on the heterogeneous transmission speeds of its components, with implications for territorial policy.

1. Introduction

The relationship between financial intermediation and economic activity remains an open question in applied financial economics. No consensus has been reached on the direction of causality, since findings vary with the institutional context, with the choice of financial development proxies, and with the econometric strategy adopted (Rousseau & Wachtel, 2000; Shan et al., 2001). This uncertainty has direct implications for the design of credit and financial deepening policies, the calibration of which depends on the direction and speed at which intermediation reaches the real economy. Three areas, however, remain insufficiently explored. Latin America, and in particular the region’s dollarised economies, has received limited attention in studies employing multivariate dynamic designs. The subnational dimension continues to be marginal, even though territorial differences in banking sector depth and productive structure can be as pronounced as those observed across countries. The post-pandemic phase, finally, has been scarcely examined, given that most recent studies close their sample period before or at the onset of 2020 (Asaleye et al., 2018; Khatri Chettri, 2022; Alam & Alam, 2024). The interaction between liquidity, credit and economic activity during the most severe shock of the past decade thus remains unexamined.

Ecuador offers a pertinent case for addressing these three limitations jointly. Following the loss of confidence in the sucre, the country adopted the US dollar as legal tender in 2000, which left it without an autonomous monetary policy (Tuasa, 2023). Under that regime, inflation stabilised in the single-digit range (Toscanini et al., 2020). Cyclical dynamics, however, were substantially modified, and the economy became more exposed to demand shocks (Orellana et al., 2023). Within that institutional setting, banking intermediation takes on greater macroeconomic prominence in the channelling of resources and in the propagation of adverse shocks. That centrality became evident in 2020, when lockdown measures reduced total deposits by 3.4 per cent and credit to the private sector by 0.60 per cent, with heterogeneous effects across credit and deposit categories (Camino-Mogro, 2022). Comparable patterns were documented across Latin American economies, where the pandemic shock contracted local credit with an intensity that varied according to the policy interventions implemented (Norden et al., 2021), and assessments extending into 2023 document the unprecedented challenges that the pandemic posed to the operational efficiency of Ecuadorian banks and cooperatives (Eraso Cisneros et al., 2025). Liquidity, financing, and commercial activity thus underwent simultaneous disruptions in the absence of conventional countercyclical adjustment mechanisms.

Beyond the Ecuadorian case, the international empirical literature on this nexus reports findings that are sensitive to the institutional context. Cointegration analyses uncover significant long-run relationships in Jordan, with bidirectional causality between deposits and growth and unidirectional causality running from private sector credit to growth (Al-Rahamneh et al., 2026). A vector error-correction framework combined with Granger causality tests yields support for both the supply-leading and demand-following hypotheses in Nigeria, with the prevailing direction varying by indicator (Asaleye et al., 2018). Within ARDL bounds testing frameworks, financial deepening indicators show positive effects on output in Nepal (Khatri Chettri, 2022) and prove statistically insignificant in Oman (Alam & Alam, 2024). This body of work shares two features that limit its usefulness for the problem posed here. Estimation is typically conducted at the national level using annual data, which constrains the temporal resolution required to identify transmission lags from deposits and credit to the real economy, while Latin America, and particularly its dollarised economies, remains underrepresented.

In response to that twofold gap, this study analyses the dynamic relationship between deposits, credit, and sales across Ecuador’s 24 provinces over the period 2019–2025 using a Panel VAR model. It examines whether financial intermediation statistically precedes the evolution of provincial sales, whether deposits and credit are transmitted to commercial activity at the same speed, and how the pandemic shock interacts with these dynamics. The central hypothesis posits a predominantly supply-leading structure, whereby financial intermediation precedes economic activity rather than the reverse. As a complementary proposition, transmission is expected to differ across the two financial channels: deposits should transmit more immediately to sales, whereas credit would operate with an additional lag arising from the loan evaluation, disbursement, and maturation processes.

The design adopted here sets this article apart from previous studies in two complementary respects. The comparative literature has estimated the link between intermediation and activity at the national level using annual data, whereas this work employs a monthly panel of the 24 provinces, which increases the temporal and territorial resolution of the analysis. In contrast to the prevailing use of cointegration frameworks applied to low-dimensional systems, the study adopts a multivariate Panel VAR that treats deposits, credit, and sales as a single system and makes it possible to identify channel-specific transmission speeds. The article thus provides subnational evidence for a dollarised Latin American economy and covers a recent period marked by an extraordinary shock. Section 2 presents the literature review. Section 3 describes the data and the empirical strategy. Section 4 reports the results. Section 5 develops the discussion. Section 6 sets out the conclusions and policy implications.

2. Literature Review

2.1. Theoretical Foundations and Empirical Patterns of the Finance-Activity Nexus

The theoretical basis of this study rests on three complementary approaches that explain how financial intermediation may affect regional economic activity. First, the supply-leading hypothesis, associated with the seminal contributions of Schumpeter (1934), McKinnon (1973), and Shaw (1973), argues that financial development precedes and supports real activity by mobilising savings, improving resource allocation, reducing information asymmetries, and facilitating investment. From this perspective, banks do not merely respond to economic expansion; they create financial conditions that enable firms and households to expand production, consumption, and commercial activity. Second, the credit channel literature, developed in particular by Bernanke and Gertler (1995), Bernanke et al. (1999), and Kashyap and Stein (2000), emphasises that the effect of bank lending on the real sector is not necessarily contemporaneous, because credit allocation is affected by information frictions, screening costs, collateral requirements, risk assessment, contractual rigidities, loan approval, and disbursement processes. This framework explains why changes in credit may precede economic activity but materialise with a lag, particularly when lending is directed towards working capital, productive investment, or commercial expansion.

Third, the financial intermediation theory of Gurley and Shaw (1960) postulates that intermediaries collect liquid liabilities from surplus units and transform them into claims on less liquid assets held by deficit units. Later contributions by Diamond and Dybvig (1983), Diamond (1984), and Levine (1997) extended this framework and formalised the functions of liquidity provision, maturity transformation, delegated monitoring, and resource allocation. Under this perspective, deposits represent liquid liabilities collected from households and firms and may capture transactional balances, precautionary savings, and short-term liquidity conditions within a territory. Credit, by contrast, reflects the transformation of those liquid liabilities into less liquid assets subject to maturity, default risk, and monitoring costs. This distinction implies that deposits and credit should not be treated as equivalent expressions of financial intermediation. Deposits may be linked to more immediate liquidity circulation, whereas credit may require additional time before its effects appear in observable economic activity. Taken together, these three perspectives support the central contribution of this article, namely the identification of temporally heterogeneous financial transmission between deposits, credit, and provincial sales in a dollarised economy.

Building on these theoretical foundations, the supply-leading hypothesis posits that financial development precedes the expansion of economic activity by mobilising savings, allocating resources, and reducing information frictions. The demand-following hypothesis holds that growth in real activity increases the demand for financial services and thereby deepens the system. A third position recognises a bidirectional relationship between the two dimensions. Shan et al. (2001) showed that none of these hypotheses can be regarded as universal, since the direction of the linkage depends on the institutional context, the variables selected, and the specification adopted.

A substantial body of empirical evidence is consistent with the supply-leading hypothesis. Rousseau and Wachtel (1998) documented, for industrialised economies, a positive association between financial depth and economic performance. Rousseau and Wachtel (2000), using a Panel VAR for 47 countries, found that the activity of financial intermediaries and stock market liquidity precede per capita output growth. Sahoo (2014) identified an effect of private credit on real GDP in India, Shah et al. (2023) reported similar results for Nepal, and Awad et al. (2025) concluded that banking intermediation favours economic activity in Palestine, although they noted limitations in the efficient channelling of deposits towards productive uses. More recent evidence for Ecuador has examined the differentiated role of bank loans and deposits in shaping economic growth. Using quarterly data for 2000–2022 and an ARDL bounds-testing approach, Naula et al. (2025) report a positive and significant long-run relationship between bank deposits and GDP, whereas bank credit displays no significant long-run effect. The authors attribute this asymmetry to high non-performing loans and weak institutional quality.

Other studies support the demand-following hypothesis. Mushtaq (2016) found that for Pakistan economic growth precedes bank credit, with no solid evidence in the reverse direction. Awad and Al Karaki (2019) reached a similar conclusion for Palestine and attributed this pattern to more conservative lending practices vis-à-vis higher-risk productive sectors. The literature has also documented bidirectional relationships. Durafe and Jha (2018) reported mutual feedback between credit and growth in India, a result consistent with Chakraborty (2008), who had documented bidirectionality between bank credit and industrial production despite finding that, at the aggregate level, growth precedes financial development. Zhou and Tewari Dev (2020) identified a positive bidirectional relationship between shadow banking, credit, and growth in a panel of developed and emerging economies. In sum, the direction and intensity of the linkage depend on the level of financial development, the productive structure, and the way in which both dimensions are operationalised.

2.2. Methodological Strategies and Conditioning Factors

The differences observed in the literature are attributable not solely to the economic context but also to the empirical strategy employed. The transition from time-series VAR models to panel designs broadened the scope of the analysis by incorporating structural differences across units. Hsueh et al. (2013), through a bootstrap panel Granger causality test for Asian economies, showed that causal patterns differ across countries, confirming that the methodology conditions the identification of the linkage. The Toda-Yamamoto procedure (Toda & Yamamoto, 1995), which allows causality testing when the order of integration is uncertain, has yielded mixed results: Wolde-Rufael (2009) found mutual feedback between credit and growth for Kenya, while Sulaiman et al. (2015), for Nigeria, detected no causality in either direction, a result they attributed to the limited capacity of the banking system to intermediate resources towards the productive sector.

These approaches, however, differ in their suitability for the type of research question addressed in this study. The Toda-Yamamoto procedure is valuable when the objective is to test Granger causality in individual time-series settings under uncertainty about the order of integration, since it reduces pre-test biases associated with unit root and cointegration decisions. Its main limitation for the present context is that it does not naturally exploit the cross-sectional structure of a provincial panel. Similarly, bootstrap panel Granger causality tests are useful for detecting heterogeneous predictive relationships across units, but they remain mainly testing procedures. They provide evidence on the direction of predictability, yet they do not offer a complete dynamic system capable of jointly estimating feedback effects, impulse responses, variance decompositions, and the differentiated timing of transmission across financial variables.

For this reason, a Panel VAR estimated by GMM is more consistent with the empirical objective of this article. This approach treats deposits, credit, and sales as jointly endogenous variables, incorporates the dynamic persistence of each series, controls for unobserved time-invariant provincial heterogeneity and uses internal instruments to mitigate the endogeneity generated by lagged dependent variables and simultaneous feedback. These features are particularly relevant in post-crisis settings, where shocks may intensify persistence, reverse causality, and adjustment delays between financial intermediation and economic activity. Recent subnational evidence also supports the use of dynamic panel methods. Flores Segovia and Torre Cepeda (2024), using Mexican state-level data and GMM estimations, showed that bank credit to non-financial firms contributes to regional economic growth, reinforcing the relevance of territorial designs for analysing the finance-growth nexus.

Panel VAR models estimated by GMM have advanced the treatment of endogeneity inherent in dynamic panels. Obed et al. (2024) identified, for Middle Eastern economies, a shift in the causal pattern before and after the Arab Spring. Tinoco-Zermeño (2023) found bidirectional causality between financial development and GDP in a panel of developing economies. Cheng and Hou (2020) showed that the effects on growth differ according to the type of financial service and the time horizon. Alodayni (2016) demonstrated that oil shocks affect output through non-performing loans and credit restriction in GCC countries, and Avdjiev and Zeng (2014) found that the effect of credit varies according to the prevailing economic growth regime.

Two factors recurrently appear as conditioning elements of the linkage. Institutional quality moderates the effect of financial intermediation, though not in a simple linear manner. Nabi and Suliman (2009) showed that the causal channel running from banking development to economic growth intensifies as the institutional environment improves, while Abuzayed and Al-Fayoumi (2016) found that, in contexts characterised by weak institutions, banking concentration can substitute for a weak legal framework and sustain a positive effect on growth. Pradhan et al. (2023) confirmed, for a broad panel of lower-income countries, that both institutional quality and financial development independently drive long-run growth, with the institutional effect dominating. Taken together, these studies suggest that the direction and magnitude of the finance-growth nexus are shaped by the quality of the regulatory and governance environment rather than by financial depth alone. Crises also alter the relationship. Li and Zhang (2022) showed that, following the 2008 crisis, the linkage between banking and growth in US states shifted from a unidirectional pattern to mutual feedback, and Jeke et al. (2025) documented that crises can weaken or reverse the effect of credit in emerging markets. This evidence is pertinent to the analysis of the pandemic period, where the shock to real activity and the response of the financial system may have modified the transmission channels in an analogous way.

2.3. Research Gaps

The foregoing review identifies the gaps that motivate the present study. Geographically, the evidence is concentrated in studies of Asia, the Middle East, and Africa, as well as in broad international panels. Tinoco-Zermeño (2023) incorporated Latin American countries within a larger sample, but without disaggregating regional dynamics or distinguishing specific monetary contexts. Jungo et al. (2022) examined this relationship for Latin America and Sub-Saharan Africa using a PVAR framework; however, their analysis assumes economies with autonomous monetary systems, in which conventional transmission channels operate without structural constraints. Recent evidence for Ecuador has advanced the debate by examining bank loans, deposits, and economic growth at the national level (Naula et al., 2025), as well as financial inclusion and provincial economic activity from a spatial perspective (Hidalgo Correa et al., 2025). However, comparable evidence remains limited for fully dollarised economies where bank intermediation is analysed through deposits, credit, and high-frequency territorial activity indicators within a dynamic Panel VAR framework.

Regarding the analytical scale, the dominant literature relies on national data or cross-country comparisons. Li and Zhang (2022) examined US states, and Jula and Jula (2013) applied panel data models to Romanian subnational units; however, the use of Panel VAR frameworks at the subnational level in developing economies remains infrequent. Recent subnational evidence, such as Flores Segovia and Torre Cepeda (2024) for Mexican states, confirms the relevance of territorial approaches and dynamic GMM estimations for analysing the finance-growth nexus. Nevertheless, the dynamic interaction between deposits, credit, and provincial sales in a dollarised Latin American economy remains insufficiently explored. In temporal terms, a significant share of studies end their sample before 2020. Giyasova et al. (2026) and Al-Rahamneh et al. (2026) provide more recent coverage—extending to 2024 and 2021, respectively—yet both operate at the national level in Middle Eastern and Central Asian contexts, without a subnational focus on Latin America.

Table 1. Summary of studies on financial intermediation and economic activity.

Authors	Country/Region	Method	Key Variables	Main Finding
Rousseau and Wachtel (2000)	47 countries	Panel VAR	Financial intermediation, stock market, GDP p.c.	Financial intermediation leads to per capita output growth
Shan et al. (2001)	9 OECD/Asia	Multivariate VAR, Granger	Financial development, GDP, trade, investment	Bidirectional or absent causality depending on context
Chakraborty (2008)	India	Granger, cointegration	Bank credit, market cap., GDP	Growth causes intermediation; bidirectional credit-industry
Wolde-Rufael (2009)	Kenya	Toda-Yamamoto VAR	M2, M3, bank credit, GDP	Bidirectional in 3 of 4 financial indicators
Hsueh et al. (2013)	Asia (panel)	Bootstrap panel Granger	Financial development, GDP	Heterogeneous patterns across countries
Sahoo (2014)	India	ARDL, Granger	Private credit, market cap., GDP	Bank credit causes growth; market has no effect
Sulaiman et al. (2015)	Nigeria	Toda-Yamamoto	Financial intermediation, GDP	No causal relationship in either direction
Alodayni (2016)	GCC countries	System GMM, Panel VAR	Oil, NPLs, credit, GDP	NPLs restrict credit and depress output
Mushtaq (2016)	Pakistan	Johansen, Granger	Deposits, bank credit, GDP	Growth drives credit; not the reverse
Durafe and Jha (2018)	India	Granger, correlation	Bank capital, credit, GDP	Bidirectional credit-growth; procyclical behaviour
Cheng and Hou (2020)	8 OECD countries	Cointegration, Granger	Intermediation, non-intermediation, GDP	Heterogeneous effects by type of financial service
Zhou and Tewari Dev (2020)	28 economies	Panel GLS, Granger	Shadow banking, credit, GDP	Positive bidirectional shadow banking-growth relationship
Tinoco-Zermeño (2023)	23 developing	Panel VAR-GMM	Credit, liquidity, energy, CO2, GDP	Bidirectional financial development-GDP
Shah et al. (2023)	Nepal	VECM, Granger	Private credit, bank assets, M2, GDP p.c.	Intermediation leads long-run growth
Obed et al. (2024)	12 Middle East	GMM Panel VAR, IRF	FII, FMI, M3, GDP p.c.	Structural break: causal pattern shifts after political crisis
Awad et al. (2025)	Palestine	ARDL, simulation	Loans, deposits, GDP	Credit positive in the long run; deposits not channelled
Al-Rahamneh et al. (2026)	Jordan	VECM, Granger	Private credit, deposits, M2, GDP	Bidirectional deposits; credit causes growth
Giyasova et al. (2026)	Turkey	VAR, Toda-Yamamoto	Exports, GDP, inflation, credit	Real activity and inflation cause domestic credit
Flores Segovia and Torre Cepeda (2024)	Mexico, states	Dynamic Panel GMM	Bank credit, regional GDP p.c.	Bank credit to non-financial firms contributes to regional economic growth
Naula et al. (2025)	Ecuador	ARDL, cointegration	Bank loans, deposits, GDP	Banking variables are linked to long-run economic growth
Hidalgo Correa et al. (2025)	Ecuador, provinces	SAR, SARAR	Multidimensional financial inclusion index, provincial gross value added	Financial inclusion exerts a positive effect on provincial GVA

Note. The table summarises the geographical context, method, key variables, and main findings of the reviewed studies. The evidence is concentrated at national and international scales, with scarce presence of dollarised economies, subnational analyses, and post-2020 periods.

synthesises the reviewed studies, highlighting the geographical concentration of the evidence, the predominance of national scales, and the limited presence of post-2020 samples.

Therefore, the gap addressed in this study is not only geographical or methodological, but also conceptual. Existing research has extensively examined whether finance precedes economic activity, follows it, or interacts with it bidirectionally. However, less attention has been paid to whether different components of bank intermediation transmit to real activity with different temporal profiles. By distinguishing deposits from credit and examining their dynamic relationship with provincial sales in a dollarised economy, this article contributes to the literature by identifying temporally heterogeneous financial transmission at the subnational level. To operationalise this design, the empirical strategy combines a balanced provincial panel with a multivariate Panel VAR.

3. Data, Variables, and Empirical Strategy

3.1. Data and Variables

The analysis is based on a balanced panel of Ecuador’s 24 provinces at a monthly frequency, covering January 2019 through July 2025. The panel comprises 79 monthly observations per province, for a total of 1896 observations. Monthly sales declared by province were obtained from the Internal Revenue Service (SRI), while deposit and credit data came from the Superintendency of Banks. Declared sales are employed as a high-frequency indicator of provincial economic activity. This choice is consistent with the practice of the Central Bank of Ecuador, which uses the SRI’s VAT Form (F104) as an input for the construction of the Monthly Index of Short-Term Economic Activity (IMAEc) and as a component of its coincident business cycle indicator. The estimated correlations support this choice: 0.87 with the IMAEc, 0.79 with adequate employment, 0.63 with tax revenue, and 0.61 with energy consumption, with the maximum association concentrated at lag zero in all cases.

The endogenous variables of the model are deposits, credit, and sales. Deposits correspond to the stock of deposits held by the public in supervised financial institutions, assigned according to the province of the receiving branch. Credit captures total lending by the financial system, including private banks, public banks, and finance companies, classified by credit type, portfolio status, and geographical origin. Sales represent the monthly value declared by taxpayers in each province. The three series were deflated using a 2018 base year, seasonally adjusted through the X-11 procedure applied independently by province, and transformed into natural logarithms. This sequence ensures that the estimated dynamics reflect real variations, free from seasonal patterns and nominal effects.

The vector of endogenous variables is defined as $Y_{i,t}=(de{p}_{i,t},cre{d}_{i,t},{sales}_{i,t})$, where i = 1, …, 24 identifies the provinces and $t$ the monthly periods. An exogenous variable, ${covid}_t$, is included, taking the value 1 between March 2020 and December 2021 and 0 otherwise, in order to control for the shock associated with the pandemic period.

Table 2. Variable definitions, data sources, transformations and prior empirical support.

Variable	Definition	Source	Transformation	Prior Empirical Use/Theoretical Support
$de{p}_{i,t}$	Stock of public deposits in financial institutions, province i, month t	Superintendency of Banks	Deflation, seasonal adjustment, natural logarithm	Banking intermediation and liquidity mobilisation: Mushtaq (2016); Awad et al. (2025); Al-Rahamneh et al. (2026); Diamond and Dybvig (1983).
$cre{d}_{i,t}$	Total credit granted by the financial system, province i, month t	Superintendency of Banks	Deflation, seasonal adjustment, natural logarithm	Bank credit and real-sector transmission: Rousseau and Wachtel (2000); Sahoo (2014); Shah et al. (2023); Bernanke and Gertler (1995).
${sales}_{i,t}$	Monthly declared sales, province i, month t	Internal Revenue Service	Deflation, seasonal adjustment, natural logarithm	Proxy for economic activity, comparable to output, GDP or regional activity indicators: Chakraborty (2008); Jula and Jula (2013); Li and Zhang (2022).
${covid}_t$	Dummy variable: 1 if March 2020 ≤ t ≤ December 2021; 0 otherwise	Authors’ construction	Exogenous variable	Crisis-related disruption in financial transmission: Li and Zhang (2022); Obed et al. (2024); Jeke et al. (2025).

Notes. Monetary variables are expressed in real dollars. $de{p}_{i,t}$, $cre{d}_{i,t}$, and $sale{s}_{i,t}$ correspond to the transformed series used in the model estimation. The econometric specification is estimated on the first differences in their logarithms, so the captured dynamics should be interpreted in terms of relative changes rather than levels. The final column reports prior empirical or theoretical support for the inclusion of each variable in the finance-activity framework.

details the operational definition, source, transformation, and prior empirical or theoretical support for each variable included in the model.

Figure 1 presents the monthly variation in the three series by province. The asymmetry in volatility is marked: sales exhibit a standard deviation of 14.5%, compared with 2.6% for credit and 5.1% for deposits. Ninety-seven per cent of monthly credit variations fall within the ±5% range, confirming its character as a slow-adjusting variable relative to commercial activity. Provincial heterogeneity is considerable: Guayas and Pichincha display sales volatility of 4.7% and 6.8%, whereas smaller provinces such as Zamora Chinchipe, Galápagos, and Pastaza reach 48.2%, 20.9%, and 15.1%. This disparity justifies the panel estimation, which exploits temporal and cross-sectional variation jointly.

3.2. Panel VAR Specification and Estimation

The relationship between deposits, credit, and sales was modelled through a Panel VAR, an approach that extends the VAR framework to the panel context and allows all variables in the system to be treated as endogenous, capturing their dynamic interdependencies both over time and across provinces (Abrigo & Love, 2016; Love & Zicchino, 2006). The general specification of the model is:

$$\Delta Y_{i,t}={\Gamma }_1\Delta Y_{i,t-1}+{\Gamma }_2\Delta Y_{i,t-2}+B\hspace{0.17em}covi{d}_t+u_{i,t}$$

(1)

where $\Delta Y_{i,t}$ denotes the vector of first differences in the endogenous variables, ${\Gamma }_1$ and ${\Gamma }_2$ are the coefficient matrices associated with the two lags of the system, $B$ is the coefficient vector of the exogenous variable, and $u_{i,t}$ is the idiosyncratic error term. Estimation in first differences eliminates time-invariant provincial fixed effects and, since the variables were previously expressed in logarithms, the captured dynamics are interpreted in terms of relative changes.

Estimation was carried out by a two-step Generalised Method of Moments (GMM), following the logic of dynamic panel models developed by Arellano and Bond (1991) and adopted in the Panel VAR literature (Abrigo & Love, 2016). The first-difference transformation introduces a mechanical correlation between the lagged regressors and the transformed error, which invalidates ordinary least squares estimation and requires the use of internal instruments constructed from lagged levels. The instrument matrix was collapsed and restricted to lags 2 and 3 in order to limit instrument proliferation. The resulting number of instruments is 33, a figure that slightly exceeds the number of groups (24); therefore, the validity of the instrumental scheme is formally assessed through the Hansen test, whose null hypothesis states that the moment conditions are jointly valid.

The two-lag specification was adopted as the main model after comparing models with p = 1, …, 5 using the MMSC criteria of Andrews and Lu (2001). The criteria did not converge on a single optimal order—the MMSC-BIC and MMSC-HQIC favour one lag, while the MMSC-AIC selects two—but the two-lag specification captures the more gradual transmission of credit to sales, a central aspect of the study’s second hypothesis. The one-lag version was retained as a robustness exercise.

Preliminary tests included the verification of stationarity in first differences through the Levin et al. (2002) test, complemented by the CIPS test of Pesaran (2007) to control for possible cross-sectional dependence among provinces, and the assessment of predictability through the panel Granger causality test of Dumitrescu and Hurlin (2012). Regarding the adequacy of the GMM scheme, the diagnostics were complemented with auxiliary Arellano and Bond (1991)-type equations to test for second-order serial autocorrelation in the residuals.

3.3. Dynamic Analysis and Robustness

The system dynamics were examined through generalised impulse response functions (GIRFs), which measure the response path of each variable to a one-standard-deviation shock in another variable of the system over an eight-month horizon. Unlike orthogonalised impulse response functions, which depend on the ordering of the variables and may yield different results according to the assumed causal hierarchy, GIRFs are invariant to the ordering (Love & Zicchino, 2006). Confidence intervals were constructed at the 90% level by means of non-parametric bootstrap resampling with 500 replications, resampling entire provinces with replacement and preserving the internal temporal structure of each unit.

The interpretation of the dynamic responses was complemented with the forecast error variance decomposition (FEVD), following the approach of Cadena-Silva et al. (2025), which quantifies the proportion of each variable’s variability, at different horizons, that is attributable to its own shocks and the proportion explained by shocks in the remaining variables of the system. This decomposition was computed up to an eight-month horizon. Both tools require the system to be stable, a condition verified when all eigenvalues of the companion matrix have modulus less than 1, ensuring that the effects of shocks are transitory.

Robustness was assessed through an alternative specification with p = 1 and 24 instruments, equating the number of instruments to the number of provincial groups. This version imposes a more parsimonious structure and strictly satisfies the condition on the number of instruments. The comparison between the two specifications makes it possible to examine whether the linkage between financial intermediation and sales holds under a more restricted temporal structure and whether the identification of the credit channel depends on the lag length of the specification.

4. Results

4.1. Preliminary Evidence and Model Adequacy

The unit root tests of Levin et al. (2002), reported in

Table 3. Unit root tests.

Variable	Z-Statistic	p-Value	CIPS Statistic	CIPS Lags	CIPS p-Value	Conclusion
Credit	−32.238	<0.00	−6.8707	2	≤0.01	Stationary
Deposits	−34.611	<0.01	−0.9083	2	≤0.01	Stationary
Sales	−48.571	<0.00	−7.6203	2	≤0.01	Stationary

Note. Variables are expressed in log differences.

, decisively reject the null hypothesis of non-stationarity for all three variables in first differences. The CIPS test of Pesaran (2007) confirms this result in the presence of possible cross-sectional dependence. Both tests support the estimation of the system on stationary transformed variables.

The results of the Dumitrescu and Hurlin (2012) panel Granger causality test, presented in

Table 4. Dumitrescu & Hurlin panel Granger causality test results.

Causal Relationship	Z-Statistic	p-Value	Decision
Credit → Deposits	2.575	0.010	Null hypothesis rejected
Sales → Deposits	−0.367	0.714	Null hypothesis not rejected
Deposits → Credit	1.021	0.307	Null hypothesis not rejected
Sales → Credit	1.333	0.183	Null hypothesis not rejected
Deposits → Sales	7.082	0.000	Null hypothesis rejected
Credit → Sales	8.665	0.000	Null hypothesis rejected

Note. The null hypothesis of the Dumitrescu & Hurlin test posits homogeneous Granger non-causality across all panel units. A p-value below 0.05 allows the null to be rejected, suggesting predictive capacity of the explanatory variable over the dependent variable in at least a subset of provinces. Values reported as 0.000 correspond to p-values below 0.001.

, reveal a directional pattern consistent with the central hypothesis of the study. Deposits and credit exhibit significant predictive capacity over the evolution of provincial sales, whereas sales show no statistical predictive capacity over the financial variables. The only additional significant linkage runs from credit to deposits, suggesting an internal interaction within the territorial financial system. This result is consistent with the supply-leading hypothesis documented by Rousseau and Wachtel (2000) and is in line with the evidence reported by Sahoo (2014), Shah et al. (2023), and Awad et al. (2025), all of whom document temporal precedence from credit to output. The absence of causality running from sales to the financial variables marks a departure from studies where demand-following or mutual feedback patterns prevail, such as Mushtaq (2016), Shan et al. (2001), and Zhou and Tewari Dev (2020).

Table 5. Lag order selection using Andrews & Lu MMSC criteria.

Lags	MMSC-BIC	MMSC-AIC	MMSC-HQIC
1	−192.795	−38.183	−100.831
2	−176.896	−39.176	−95.020
3	−140.990	−20.088	−69.149

Note. The Andrews and Lu MMSC criteria allow comparison of GMM-estimated specifications. Lower values indicate better relative model performance under each criterion. The MMSC-BIC and MMSC-HQIC favour a one-lag specification, while the MMSC-AIC selects two lags. The main analysis adopts p = 2 on theoretical and empirical grounds and uses p = 1 as a robustness check.

presents the comparison of specifications using the MMSC criteria of Andrews and Lu (2001). The criteria do not converge on a single optimal order: the MMSC-BIC and MMSC-HQIC favour one lag, while the MMSC-AIC selects two. The two-lag specification is adopted as the main model because it captures the more gradual transmission of credit to sales, a central aspect of the study’s second hypothesis. The one-lag version is employed as a robustness check.

The econometric diagnostics of the main model are summarised in

Table 6. Econometric diagnostics of the main Panel VAR model (p = 2).

Indicator	Result
Estimation method	Two-step GMM
Transformation	First differences
Number of groups	24
Number of observations	1824
Number of instruments	33
Collapsed instruments	Yes
Restricted instrumental lags	Yes
Exogenous variable	COVID
Hansen statistic (p-value)	0.288
Maximum eigenvalue modulus	0.981
Stability condition	Satisfied
Auxiliary AR(1)—deposits equation (p-value)	0.0147
Auxiliary AR(2)—deposits equation (p-value)	0.0915
Auxiliary AR(1)—credit equation (p-value)	0.0037
Auxiliary AR(2)—credit equation (p-value)	0.0991
Auxiliary AR(1)—sales equation (p-value)	0.0264
Auxiliary AR(2)—sales equation (p-value)	0.0752

Note. The main model was estimated by a two-step Panel VAR-GMM with first-difference transformation and collapsed instruments. The null hypothesis of the Hansen test posits the joint validity of the moment conditions. System stability is verified when all eigenvalues of the companion matrix lie within the unit circle. AR(1) and AR(2) statistics are obtained from auxiliary Arellano-Bond-type difference-GMM regressions estimated equation by equation, since the Panel VAR estimator does not provide a direct system-level serial correlation test. The AR(1) rejection is expected under the first-difference transformation. The AR(2) tests do not reject the null of no second-order serial correlation at the 5% level, although the p-values are close to the 10% threshold. Therefore, these diagnostics are interpreted as supportive but not conclusive evidence regarding the validity of the moment conditions.

. The Hansen statistic does not reject the null hypothesis of joint instrument validity (p = 0.288), and the stability condition is satisfied, with the maximum eigenvalue modulus below unity. The auxiliary Arellano and Bond (1991)-type tests confirm the expected significance of AR(1) and do not reject the absence of second-order autocorrelation at the 5% level, although the AR(2) p-values lie close to the conventional threshold (0.091 for deposits, 0.099 for credit, and 0.075 for sales), which warrants caution in the interpretation.

4.2. Estimated Coefficients and Transmission Across Financial Channels

Table 7. Main Panel VAR model estimates (p = 2).

Variable	Deposits	Credit	Sales
lag1 Deposits	0.6537 ***	0.0273	0.3111 **
lag1 Credit	0.2277 ***	0.8405 ***	−0.2038
lag1 Sales	0.0665 ***	−0.0052	0.5799 ***
lag2 Deposits	0.0739	0.1166	−0.1914
lag2 Credit	−0.0008	−0.0048	0.2063 ***
lag2 Sales	0.013	−0.0122	0.1214 **
COVID	0.0446 ***	−0.0259 *	−0.1061 ***
Constant	0.1829 *	0.3370 ***	0.0245

Note. Estimated coefficients of the main two-lag Panel VAR model, estimated by two-step GMM with first-difference transformation. All endogenous variables—Deposits, Credit, Sales—are expressed in natural logarithms. *** p < 0.001, ** p < 0.01, * p < 0.05. The COVID variable corresponds to the exogenous control associated with the pandemic period.

reports the coefficients of the main model with two lags. The relationship between financial intermediation and provincial sales is dynamic and heterogeneous across financial channels. In the deposits equation, autoregressive persistence is high (0.654 at the first lag), one-period lagged credit exerts a significant positive effect, and lagged sales display a positive coefficient of smaller magnitude. This latter result, absent in the bivariate Granger test reported in Table 4, reflects a conditional feedback that emerges when the simultaneous dynamics of all variables in the system are controlled for.

In the credit equation, own inertia dominates. The first-lag coefficient reaches 0.841, and no robust effects from deposits or sales are observed. Provincial credit dynamics depend, to a large extent, on their own past values—a feature compatible with lending processes subject to prior portfolio decisions, risk assessment, and bank supply conditions. This evidence echoes the findings of Mushtaq (2016) and Awad et al. (2025), who show that credit does not always respond immediately to variations in economic activity.

The central result of the study is concentrated in the sales equation. One-period lagged deposits exert a positive and statistically significant effect, while credit operates through its second lag, as the first lag is not significant. The temporal difference between the two channels supports the study’s second hypothesis and suggests that deposits function as a faster-transmitting liquidity component, whereas credit requires an additional period to translate into observable commercial activity. This pattern is consistent with the evidence of Rousseau and Wachtel (2000), Sahoo (2014), and Shah et al. (2023), although it adds a nuance that Cheng and Hou (2020) had already anticipated: the effects of the financial system depend on the composition of the service analysed and the time horizon considered.

The estimated coefficients in the sales equation can be read as short-run elasticities, given that all variables enter the model in first differences in natural logarithms. A one per cent increase in deposits in the previous month is associated with an increase of approximately 0.31 per cent in provincial sales in the current month, while a one per cent increase in credit two months earlier is associated with an increase of approximately 0.21 per cent in current sales. The deposit elasticity is therefore larger in magnitude than the credit elasticity, although the latter materialises one period later. This pattern is consistent with the second hypothesis on the differential timing of transmission across financial channels. Both elasticities are economically meaningful at the provincial scale, where commercial activity is closely tied to available liquidity and to the financing capacity of the productive sector. In a dollarised economy that operates without the buffer of an autonomous monetary policy, magnitudes of this order justify treating the calibration of credit and liquidity policy as a key instrument for regional economic activity.

The coefficients of the pandemic exogenous variable display signs consistent with expectations: positive in the deposits equation, negative in credit, with the largest contraction estimated for sales. The increase in deposits is consistent with a greater preference for liquidity and precautionary saving, the decline in credit reflects a more cautious supply in a high-uncertainty environment, and the contraction in sales indicates that the most intense adjustment fell on commercial activity. This evidence is consistent with the literature documenting disruptions in transmission channels during crisis periods (Jeke et al., 2025; Li & Zhang, 2022; Obed et al., 2024).

4.3. Dynamic Responses to Financial Shocks

Figure 2 reports the generalised impulse-response functions of the baseline Panel VAR-GMM model. The figure is organised as a response-by-shock matrix: shocks are displayed in columns and response variables in rows. The solid line represents the point response, while the shaded areas denote 90% bootstrap confidence bands. Since the magnitude of the responses differs across variables, each panel is displayed with its own vertical scale. Responses are expressed in log points to a one-standard-deviation shock and should therefore be interpreted as dynamic responses to innovations, rather than as one-to-one elasticities.

The main result is the response of sales to a credit shock. In the baseline two-lag specification, the point estimates show a positive response throughout the eight-month horizon. The initial response is 0.123 log points and gradually declines to 0.027 log points by the eighth month. This pattern suggests that credit may operate as a dynamic channel for provincial economic activity. However, the bootstrap confidence bands indicate that the estimated responses should be interpreted with caution, since statistical precision varies across horizons. The evidence therefore points to an economically meaningful credit-sales transmission in terms of magnitude, but not to a uniformly significant response across all horizons.

The response of sales to a deposit shock is weaker and less direct. It is initially negative and then remains close to zero, suggesting that higher deposits do not translate automatically into immediate commercial expansion. This result is consistent with a more indirect liquidity mechanism, in which deposits may support activity only if they are subsequently channelled into credit or transactional spending. By contrast, sales display a strong own response, indicating short-run persistence in provincial commercial activity.

Feedback from the real sector to the financial block is present but more moderate. A shock to sales generates positive responses in credit and deposits after the initial periods, but these effects are smaller than the response of sales to credit shocks and are estimated with wider confidence bands. Overall, the impulse-response evidence does not suggest a symmetric feedback structure. Rather, it points to a dynamic system in which the financial block—particularly credit—retains greater precedence over provincial sales.

Appendix A reports the corresponding orthogonalised impulse-response functions and the one-lag robustness specification, all including 90% bootstrap confidence bands to assess the precision of the estimated responses.

Figure 3 presents the forecast error variance decomposition over the eight-month horizon. For deposits, own shocks explain all forecast error variance in the first period, but their contribution declines progressively to 57.2% by horizon 8. At the same horizon, credit shocks explain 34.2% of deposit forecast error variance, while sales shocks account for 8.6%. This pattern indicates that deposit dynamics are initially dominated by their own innovations, but become increasingly connected to credit conditions as the forecast horizon expands.

Credit exhibits the greatest autonomy within the system. Own credit shocks explain 94.7% of credit forecast error variance in the first period and still account for 91.4% by horizon 8. The contribution of deposit shocks increases moderately from 5.3% to 8.4%, while sales shocks explain less than 1% of credit variability throughout the horizon. This suggests that credit is highly persistent and largely self-driven, with limited contemporaneous dependence on innovations originating in sales.

For sales, own shocks remain the dominant source of forecast error variance, explaining 74.7% in the first period and 75.8% at horizon 8. However, credit shocks account for a sizeable and increasing share of sales variability, rising from 20.5% in the first period to 21.5% by horizon 8. In contrast, deposit shocks explain a relatively small and declining fraction, from 4.8% to 2.7%. Read jointly with the estimated coefficients and the impulse-response functions in Figure 2, the variance decomposition reinforces the interpretation that credit is the main financial transmission channel in the system. This evidence should be read as complementary to the impulse-response analysis, since FEVD quantifies the relative contribution of each innovation to forecast uncertainty rather than providing a standalone causal estimate.

4.4. Robustness

To assess the sensitivity of the results to instrument construction, additional specifications were estimated using more restrictive instrument sets. In particular, two-lag models were re-estimated by further limiting the GMM-style instruments while maintaining collapsed matrices and the same first-difference transformation. However, the two-lag specifications in which the number of instruments was forced below the number of provincial units led to Hansen test p-values below the conventional 5% threshold, indicating rejection of the overidentifying restrictions. For this reason, these specifications were not retained. Instead, a stricter one-lag Panel VAR specification is reported, in which the number of instruments does not exceed the number of cross-sectional units. This model provides a conservative robustness check focused on instrument parsimony, while the two-lag specification is retained as the baseline because it better captures the delayed credit-sales transmission mechanism and satisfies the Hansen test under collapsed and restricted instruments.

Table 8. Robustness Panel VAR model estimates (p = 1).

Variable	Deposits	Credit	Sales
lag1 Deposits	0.7029 ***	0.0714 *	0.4139 ***
lag1 Credit	0.2447 **	0.8158 ***	−0.0038
lag1 Sales	0.0460 ***	0.0255	0.4868 ***
COVID	0.0473 ***	−0.0292 ***	−0.1357 ***
Constant	0.461	1.2675 ***	−2.5523 ***

Note. Estimated coefficients of the one-lag robustness Panel VAR model, estimated by two-step GMM with first-difference transformation. All endogenous variables—Deposits, Credit, Sales—are expressed in natural logarithms. *** p < 0.001, ** p < 0.01, * p < 0.05. The COVID variable corresponds to the exogenous control associated with the pandemic period.

reports the coefficients of the alternative specification with a single lag and 24 instruments. Credit continues to exert a positive and significant effect on deposits, and deposits retain a positive effect on sales. The overall direction of the system does not change. The Hansen statistic remains within a range compatible with instrument validity (p = 0.126), and the system preserves the stability condition (maximum eigenvalue of 0.937). The main difference appears in the sales equation: the effect of credit ceases to be significant when the system is restricted to a single lag. This result does not invalidate the central conclusion but confirms that the transmission of credit to sales is not instantaneous and requires a broader time horizon to become visible, as suggested by the dynamic trajectories in Figure 2. The deposit channel, by contrast, remains significant under both specifications, reinforcing its role as a more immediate transmission component.

The impulse-response evidence reinforces this interpretation. The GIRFs from the one-lag specification are qualitatively consistent with the baseline model: sales continue to respond positively to credit shocks throughout the eight-month horizon, although the response is more gradual and less pronounced than in the two-lag specification. Deposits and sales also exhibit positive and persistent own responses, while feedback from sales to the financial variables remains moderate. The main sensitivity across specifications concerns the response of sales to deposit shocks, which changes more noticeably with the lag structure. Overall, the robustness exercise indicates that the credit-sales transmission is the most stable cross-variable result, whereas the deposit-sales mechanism should be interpreted with greater caution.

As a complementary check, Appendix B reports both generalised and orthogonalised impulse-response functions for the baseline and robustness specifications, all including 90% bootstrap confidence bands. These figures confirm that the main dynamic trajectories are not driven by the specific presentation of the baseline GIRF.

As an additional sensitivity exercise, the baseline two-lag Panel VAR-GMM was re-estimated excluding the COVID-19 dummy. This exercise evaluates whether the dynamic relationships among deposits, credit, and sales are mechanically driven by the inclusion of the pandemic-period control. The model without COVID uses 30 instruments and preserves the same first-difference transformation, collapsed instrument matrix, and restricted GMM-style lag structure. The Hansen test does not reject the overidentifying restrictions at the 5% level, although the p-value decreases to 0.064, indicating that the specification without the pandemic control is less satisfactory in terms of instrument validity than the baseline model. The results of this sensitivity specification are reported in

Table 9. Robustness Panel VAR model estimates without COVID control (p = 2).

Variable	Deposits	Credit	Sales
lag1 Deposits	0.7043 ***	0.0482	0.3502 ***
lag1 Credit	0.1245 *	0.8869 ***	−0.1059
lag1 Sales	0.0617 **	−0.0015	0.5692 ***
lag2 Deposits	0.1204 *	0.0995 *	−0.1552
lag2 Credit	0.0127	−0.0539	0.0187
lag2 Sales	−0.0001	−0.0152	0.1621
Constant	0.1756	0.3205	0.0209

Note. Estimated coefficients of the two-lag Panel VAR model excluding the COVID control, estimated by two-step GMM with first-difference transformation and collapsed instruments. The model includes 1824 observations, 24 provincial groups, and 30 instruments. All endogenous variables—Deposits, Credit, Sales—are expressed in natural logarithms. The Hansen test does not reject the overidentifying restrictions at the 5% level, although the p-value is close to the conventional threshold, χ²(9) = 16.12, p = 0.064. *** p < 0.001, ** p < 0.01, * p < 0.05.

.

The exclusion of COVID modifies the coefficient estimates in the sales equation. In particular, the second lag of credit is no longer statistically significant, while sales remain strongly persistent and deposits continue to exert a positive effect. This suggests that the pandemic period represents a relevant common shock in the monthly provincial system. When it is omitted, part of the extraordinary variation associated with 2020–2021 is absorbed by the autoregressive structure of the model, weakening the direct credit coefficient in the sales equation.

However, the generalised impulse-response functions, reported in Appendix C, remain qualitatively consistent with the baseline specification. The response of sales to a credit shock remains positive throughout the eight-month horizon. The initial response is 0.124 log points in the model without COVID, compared with 0.123 log points in the baseline model, and gradually declines to 0.022 log points by the eighth month. This similarity indicates that the main dynamic credit-sales trajectory is not mechanically generated by the COVID dummy. Nevertheless, given the lower Hansen p-value and the loss of significance in the direct credit coefficient, the model without COVID is interpreted as a sensitivity check rather than as the preferred specification.

This exercise also clarifies the role of the COVID variable in the baseline model. The dummy is not intended to measure the heterogeneous intensity of the pandemic across provinces, but to control for a common disruptive period that simultaneously affected deposits, credit, and sales. Therefore, the pandemic coefficient is interpreted cautiously as a period control, while the central conclusions of the paper are based on the dynamic responses of the financial-real system rather than on a structural interpretation of the COVID dummy itself.

5. Discussion

This article provides the first subnational evidence on the relationship between banking intermediation and economic activity for a dollarised Latin American economy. The comparative literature has examined the Ecuadorian and Salvadoran cases through designs based on country-level time series, such as that of Cáceres (2021) for El Salvador, while the comparative analysis of the three members of the dollarised group, namely Ecuador, El Salvador, and Panama, has remained restricted to aggregate macroeconomic comparisons (Tutiven-Desintonio, 2025). The estimated system is consistent with a supply-leading structure in which the first lag of deposits and the second lag of credit retain predictive capacity over provincial sales without robust feedback in the reverse direction. This direction of causality converges with the evidence that Khatri Chettri (2022) reports for Nepal through an ARDL framework and with the findings documented by Asaleye et al. (2018) for Nigeria through a VECM with Granger causality. What the Ecuadorian result adds is an internal temporal disaggregation of the supply-leading hypothesis, since banks’ asset-transformation function operates at different speeds depending on the balance-sheet component examined. Deposits, as liquid liabilities directly available for transactional use, are transmitted to provincial sales with a one-month lag. Credit, subject to risk assessment, disbursement, and maturation processes, requires an additional period before reaching the real sector. The dynamic reading provided by the impulse-response functions reinforces this interpretation with a notable nuance. Sales display sustained positive responses to a credit shock over an eight-month horizon, whereas the response to a deposit shock is initially weak and only materialises once those liquid balances are transformed into effective lending. Banking intermediation thus operates not as two parallel channels but as a sequential mechanism. Deposits fund credit, while credit, in turn, channels the sustained transmission to provincial sales.

The behaviour of the system during the pandemic period offers evidence consistent with this analytical framework and, at the same time, a test of its operation under stress. The accumulation of deposits, the contraction of credit, and the fall in sales form an asymmetric pattern in which the first stage of intermediation remains intact while the second breaks down. This result aligns with the evidence that Norden et al. (2021) report for Brazil, where the policy interventions implemented during the pandemic modulated the magnitude of the impact on local credit, and with the short-term reading that Camino-Mogro (2022) provides for Ecuador. The extension of the monthly panel over 22 months reveals that this asymmetry was not a transient episode but a persistent disconnect between deposit-taking and lending. In a monetary regime that lacks autonomous policy instruments, the persistence of this desynchronisation reveals a structural vulnerability specific to the dollarised arrangement, since no contemporaneous adjustment of the monetary base can accelerate the credit recovery once the intermediation chain has broken down.

6. Conclusions and Policy Implications

This study examined the dynamic relationship between financial intermediation and provincial economic activity in Ecuador using a Panel VAR model estimated with monthly data for 24 provinces over the period 2019–2025. The evidence obtained is consistent with a predominantly supply-leading structure in which deposits and credit statistically precede the evolution of provincial sales. The most relevant finding of the article does not lie in the direction of the linkage, already documented for other contexts, but rather in the temporal heterogeneity within the financial block itself. Deposits are associated with sales over an immediate horizon, whereas credit operates with an additional lag. That difference reflects the distinct nature of the two channels. Available liquidity is channelled rapidly toward the commercial circuit, while credit financing passes through processes of evaluation, disbursement, and maturation before materialising as observable economic activity. In a dollarised economy, where the absence of conventional monetary instruments constrains the avenues for macroeconomic adjustment, this distinction acquires direct practical relevance.

The behaviour of the system during the pandemic period reinforces this reading. The simultaneous increase in deposits, the contraction in credit, and the decline in sales reveal that adjustments to severe shocks are transmitted with differentiated intensity depending on the financial channel considered. This asymmetry suggests that crisis response strategies should envisage channel-differentiated interventions. Preserving the continuity of credit during adverse episodes is particularly relevant, because if its effect on sales operates with a lag, a prolonged contraction in lending can intensify recessionary phases and delay territorial recovery. At the same time, improving the transformation of deposits into credit at the provincial scale constitutes a complementary policy objective, since territorial financial depth depends not only on the volume of funds captured but also on the institutional capacity to channel them toward activities with an impact on employment and local commercial circulation.

A further implication arises from the order of magnitude of the estimated responses. Provincial sales react to changes in deposits and credit with short-run elasticities of approximately 0.3 and 0.2, respectively, materialising at distinct lags. In a dollarised regime without monetary instruments, these magnitudes provide policymakers with a quantitative reference for calibrating the scale and timing of credit and liquidity interventions, a benchmark that aggregate national models rarely supply at the territorial level.

The results should be interpreted within the limits of the empirical design. Provincial variables reflect the administrative location of operations and not necessarily the final destination of financial flows. The Panel VAR specification imposes an average structure that does not capture the full heterogeneity across provinces. The approach allows the identification of temporal precedence and joint dynamics, yet it does not constitute a structural causal identification strategy in the strict sense. These limitations delimit the inferential scope and, at the same time, open avenues for future research. Promising directions include disaggregating credit by segment to identify which component most intensely drives the transmission toward sales, exploring provincial heterogeneities according to productive structure and banking density, incorporating alternative measures of territorial economic activity, and moving toward more demanding identification designs.