Financial and market risks of bitcoin adoption as legal tender: evidence from El Salvador

The paper examines policy-level implications of adopting Bitcoin as legal tender in El Salvador (legalized June–September 2021). It addresses the lack of comprehensive regulatory and legal frameworks (AML/CTF, taxation), Bitcoin’s fixed supply and price volatility, and their implications for monetary policy and financial stability. Using a pass-through perspective, it evaluates how Bitcoin price shocks affect remittances, the money multiplier, the USD index, and gold prices. Research questions: (1) Is adopting Bitcoin as legal tender credible for economic stability? (2) What is the impact on international transfers (remittances)? (3) What is the relationship between Bitcoin and gold in stimulating monetary policy transmission? The study adopts an SVAR with long-run restrictions to measure Bitcoin price pass-through into these macro-financial variables.

Classical and modern theories of money (Jevons, Wicksell, Kiyotaki & Wright) emphasize acceptability and social convention for a medium of exchange. Chartalist views highlight taxation as conferring value to money and promoting acceptability. Prior crypto literature spans risks, arbitrage, asset pricing, network effects, production costs, and determinants of prices. Empirical work documents correlations and hedging roles of cryptocurrencies with traditional assets (USD, gold). The Salvadoran case tests whether accepting a digital asset for tax payment suffices for its acceptance as currency and extends literature by evaluating policy channels (money multiplier), international transfers, and Bitcoin–gold relationships using SVAR pass-through measures.

Data: Monthly series from September 2010 to February 2022. Variables include Bitcoin price in USD (investing.com), remittances to El Salvador in USD (Central Bank of El Salvador), money multiplier (IMF Monetary and Financial Statistics), USD index DXY (investing.com), and LBMA gold price (investing.com). All series are transformed to stationarity for VAR estimation. Two policy dummies capture El Salvador’s Bitcoin policy: bdummy (June 9, 2021 announcement) and btdummy (September 7, 2021 implementation); both are significant at 1%. Model: A multivariate VAR/SVAR with lag p=1. Reduced-form VAR is estimated and then structurally identified. Granger causality tests reject the null that remittances, money multiplier, USD index, and gold do not Granger-cause Bitcoin (p=0.001), with caveats about interpretation and omitted variables. Identification: Primary identification follows Blanchard and Quah (1988) long-run restrictions; specifically φ12(1)=0 to separate temporary shocks (demand) and permanent components (supply), using Cholesky factorization on long-run impact matrices. Impulse response and pass-through: Structural impulse responses (IRFs) are computed, and Bitcoin price pass-through Y_j is defined as the ratio of the cumulative impulse response of variable i (remittances, money multiplier, USD index, gold) to the cumulative response of Bitcoin after a Bitcoin price shock. Pass-through is reported at multiple horizons (1, 3, 6, 12, 19 months) with 95% bootstrap confidence intervals. Robustness checks apply alternative SVAR identification (A-model with contemporaneous restrictions) and re-estimate pass-through. Estimation details: Implemented in R using the SVAR package; data-driven identification methods considered per Lanne et al. (2017) and related literature. Lag selection p=1 is retained. Figures present IRFs, historical decomposition (HD), and forecast error variance decomposition (FEVD).

Baseline SVAR (Blanchard-Quah identification):

• Remittances (Bt_rem): Initially decrease by about 10.9% at 1 month; effects moderate and show a small sustained long-run positive response (~6.1% from 12 months onward). Reported points: −10.94% (1m), 9.15% (3m), 5.89% (6m), 6.06% (12m and 19m). Authors interpret long-run as moderately positive for remittances.
• Money multiplier (Bt_momu): Persistent negative response: −8.81% (1m), −8.23% (3m), −7.53% (6m), stabilizing around −7.6% (12–19m).
• USD index (Bt_usd): Large immediate negative response: −24.5% (1m), then around −5.8% to −6.0% from 3–19 months.
• Gold prices (Bt_gold): Modest negative response: −4.42% (1m), around −5.9% to −6.02% from 3–19 months. Dynamic effects: Demand disturbances generate hump-shaped effects on Bitcoin and remittances that fade by ~7 months; supply disturbances produce mirror patterns with remittances turning positive in the long run. FEVD/HD: Gold prices, USD index, and remittances contribute substantially to Bitcoin’s variance; Bitcoin behaves as an asset “between gold and the dollar.” Robustness (A-model identification):
• Remittances: Strong positive short-run pass-through 43.58% (1m), then small but positive 2.56% (3m), ~1.9–1.95% (6–19m).
• Money multiplier: Negative and persistent around −6.7% (1m) to −6.05% (6m), stabilizing ~−6.09% (12–19m).
• USD index: More pronounced negatives: −27.54% (1m), roughly −10.6% to −11.23% (3–19m).
• Gold prices: Larger negatives than baseline: −8.72% (1m) deepening to ~−23.7% (12–19m). Policy signals: Bitcoin price shocks tend to depress the money multiplier and USD index and reduce gold prices; remittances exhibit sensitivity with sign depending on identification, but authors emphasize a modest long-run positive response in the baseline.

Findings suggest that adopting Bitcoin as legal tender introduces channels affecting monetary policy transmission in El Salvador. Bitcoin price shocks reduce the money multiplier, implying potential constraints on traditional credit creation if deposits shift toward crypto holdings. The USD index and gold respond negatively to Bitcoin shocks in these models, indicating intertwined dynamics between crypto and traditional safe assets/currencies. FEVD results show gold and USD index significantly influence Bitcoin dynamics; Bitcoin appears positioned between gold and the dollar in portfolio terms. While remittances may face short-run reductions amid volatility, baseline results point to modest long-run gains, suggesting potential for international business and improved transfer channels if volatility and frictions are managed. However, risks remain: crypto’s volatility and speculative nature can threaten financial stability, with potential bubble dynamics and spillovers to indebted households and institutions, especially under limited regulatory oversight. These results inform debates on whether accepting a digital asset for tax payments ensures its acceptance as money and how coexistence with state money affects policy efficacy.

Adopting Bitcoin as legal tender in El Salvador is unlikely to deliver broad economic stability benefits due to volatility and limited transactional use. Empirically, Bitcoin price shocks lower the money multiplier and affect key financial benchmarks (USD index, gold), with mixed effects on remittances—short-run declines but a modest long-run positive response in the baseline. Policymakers should weigh innovation benefits against monetary control and financial stability risks, considering regulations, design features (access rules, use beyond intraday, interest-bearing options), and robust oversight. The experience underscores that central banks can maintain the role of state money through credible policies, limiting crypto’s role as a medium of exchange. Future research should test generalizability to other countries and examine behavioral drivers of remittance decisions under crypto volatility.

• External validity: Results are specific to El Salvador’s dollarized economy and policy context; outcomes may differ in countries with other monetary regimes.
• Identification sensitivity: The sign and magnitude of remittance pass-through vary across identification strategies (long-run restrictions vs. A-model), indicating model dependence.
• Measurement scope: Remittances are measured in USD; Bitcoin-related transfers or informal channels may be undercaptured, implying indirect effects.
• Data frequency and period: Monthly data from 2010–2022 may smooth short-lived dynamics; early-period Bitcoin market immaturity and later regime changes could influence estimates.
• Causality caveats: Granger causality reflects predictive power, not structural causation; omitted variables and contemporaneous correlations may bias inferences.
• Stationarity and specification: Transformations to achieve stationarity and lag selection (p=1) may affect dynamics; alternative specifications could yield different responses.