The study explores the nexus between economic growth and international financial integration, mediated by country income levels and trade liberalization, across 108 countries. To check the endogeneity of the lagging variable in the model, the two-step GMM robust panel data estimator is used, while the robustness of the S-GMM estimates is examined by the PMG estimator. The findings of this study reveal that (i) the effects of international financial integration (IFI) on economic growth are mediated by the level of trade liberalization within country income groups, indicating that countries with a higher level of trade liberalization relative to their income group peers show a positive impact of financial integration on economic growth; (ii) the effects of IFI on economic growth are influenced by the income level of countries within their respective income groups, whereas countries with higher income levels relative to their peers in the same income group demonstrate a positive impact of financial integration on economic growth. The study's primary contribution is to fill a gap in the literature by confirming the beneficial effects of trade liberalization and by identifying the country's income level as a mediator in the relationship between international financial integration and economic growth.
The emergence of artificial intelligence (AI) marks a transformative shift in central banking, presenting new opportunities for economic forecasting, financial supervision, and operational efficiency. Traditionally, central banks have depended on structured frameworks and statistical models, but AI technologies-particularly machine learning and generative models-are redefining these core functions. AI models enhance central banks' ability to predict economic trends, detect financial irregularities, and streamline administrative tasks, supporting informed decision-making in complex, data-driven environments. Despite these advantages, AI adoption introduces significant challenges, including data security concerns, systemic risk, and algorithmic bias. Central banks must navigate these risks through robust data governance, ethical AI frameworks, and strategic human capital investments. This article examines AI's applications, risks, and strategic requirements in central banking, illustrating how early adopters like the European Central Bank and BIS Innovation Hub leverage AI for forecasting and regulatory compliance. By fostering international collaboration and transparency, central banks can responsibly harness AI's potential to strengthen financial stability and maintain public trust. This balanced approach underscores AI's role in enabling central banks to adapt to an evolving global financial landscape while safeguarding ethical standards and regulatory integrity.
The present study uses the threshold SVAR models to analyse the periods before and after inflation targeting, covering monthly data between 1992 and 2024. It examines whether the Fed reaction function based on the expanded Taylor equation with the US economic and policy uncertainty index is asymmetric with respect to the inflation and output gap. The results indicate that when inflation is below the 2% threshold, the Fed reacts to economic instability rather than inflation and the output gap. The empirical findings show that once inflation exceeds this level during the inflation targeting period, the Fed reacts to inflation in the long term, output gap in the short term, and economic instability in both the short and long term. Also, when inflation is greater than 2%, the Fed reacts more to expected inflation than to current inflation. Moreover, in post-inflation-targeting period, when inflation rises above this level, the Fed reacts to positive shocks in current inflation with a three-month lag, whereas its reaction to expected inflation is instantaneous. According to the output gap threshold value, it was determined that the Fed's reactions were symmetrical before inflation targeting and asymmetrical afterwards. Empirical findings show that the Fed reacts to economic instability before inflation targeting, to both output gap and economic instability in the short term after inflation targeting, and to inflation in the long run. The results show that the asymmetric effects of the Taylor rule are due to the asymmetric preferences of the Fed.
This study evaluates the institutional readiness of central banks in the Western Balkans to adopt and integrate Artificial Intelligence (AI) into their operations. While AI is rapidly transforming central banking functions globally, enhancing forecasting, regulatory oversight, and operational efficiency, its adoption by central banks of the Western Balkans remains underexplored. Using a structured framework adapted from J & ouml;hnk et al. (2021), this research assesses five key readiness dimensions: strategic alignment, resources, knowledge, organizational culture, and data infrastructure; and supplements this with additional questions to gain deeper insights into the perceived practices, expectations, benefits and risks of AI by central bankers. Insights were drawn from senior professionals across six central banks in the region. The findings reveal generally low yet uneven levels of readiness for AI adoption, with particular deficiencies in data infrastructure and the strategic integration of AI. The benefits of using AI in these central banks are widely recognized, including improvements in economic forecasting and task automation. However, key barriers persist, including institutional inertia, fragmented data systems, and a lack of internal expertise. The study emphasizes the importance of aligning AI strategies with core mandates, investing in capacity-building, and fostering regional cooperation as ways to ensure that the central banks in the region can effectively fulfil their mandated functions.
This paper examines the effectiveness of monetary policy transmission to bank liquidity in countries operating under constrained monetary regimes, drawing on the case of Bosnia and Herzegovina's currency board arrangement. Using monthly bank-level data (2006 - 2024), we apply a two-step empirical framework combining a Panel Vector Autoregression (PVAR) and a Panel Vector Error Correction Model (PVECM) to analyse the short and long-run effects of reserve requirements and the remuneration of excess reserves on bank liquidity. Our findings indicate that liquidity dynamics are largely driven by banks' internal portfolio decisions, while monetary instruments can influence liquidity in the short term, but their overall impact is modest and considerably outweighed by internal bank-level factors such as lending intensity and foreign asset exposure. A stable long-run relationship is confirmed between liquidity, policy tools, and balance sheet fundamentals. Although the study considers bank size and ownership structure, the estimated effects represent system-wide averages. Nonetheless, the observed patterns are broadly consistent with the hypothesis that large and foreign-owned banks may exhibit lower sensitivity to domestic monetary impulses. The results suggest that in a currency board system, the transmission of monetary impulses is constrained and highly dependent on structural banking characteristics.
The question that an asset could serve as a good store of value has been brought up a number of times, especially due to major market shocks. Despite the long-held belief that gold serves as a safe haven, a growing body of contemporary literature perpetually assesses this notion. Whether cryptocurrencies, namely Bitcoin as the most well-known of them, as an asset, can serve as a safe haven or hedge gains traction in research. This study investigates the relationship of gold and Bitcoin to S&P 500 stock index and the possibility that these might serve as a safe haven in respect to stock indexes. The results show that, when analysed the safe haven properties of either gold or Bitcoin, there is evidence that Bitcoin has negative and statistically significant relationship with the stock market making it good proponent for safe haven. On the other hand, gold has non-positive correlation with the stock market, which adds to the basic readings that it is good hedge asset.
This paper presents two offline, on-premise NLP proof-of-concept assistants built on a shared architecture for internal knowledge access in the Central Bank of Bosnia and Herzegovina: (i) a semantic document search tool for internal Word/PDF repositories and (ii) an HR chatbot that applies retrieval-augmented generation (RAG) over indexed HR policies and procedures. Rather than proposing a novel NLP method, the paper contributes by documenting a reusable offline architecture for institutional AI assistants in a security-constrained central banking environment and by providing pilot evidence on how established semantic retrieval and RAG techniques can be adapted to strict requirements of confidentiality, data sovereignty, and governance. The semantic search assistant combines exact phrase matching with embedding-based retrieval and hybrid re-ranking, while the HR chatbot generates source-grounded answers using locally hosted language models under explicit governance constraints, including B/H/S-only output, strict fallback behaviour, and transparent display of retrieved passages. Pilot results indicate that hybrid retrieval offers the most reliable performance across representative internal queries, while the HR chatbot demonstrates the feasibility of document-grounded employee support under offline institutional constraints. The findings provide preliminary evidence that offline NLP assistants can improve access to internal institutional knowledge while remaining compatible with the security and operational risk requirements typical of central banking environments.
This paper tests whether Islamic equity indices exhibit systematically different sensitivities to high-frequency monetary policy surprises than their matched conventional benchmarks, and whether any Islamic-conventional wedge is universal or provider-specific. We construct daily return differentials (Islamic minus conventional) in USD total returns, expressed in basis points, for three global index families: FTSE All-World vs FTSE IdealRatings All-World Islamic; MSCI World vs MSCI World Islamic; and S&P Global BMI vs S&P Global BMI Shariah. We regress the differential on orthogonalised FOMC monetary policy surprises and two ECB surprises: a Target component measured around the press release (OIS-1M) and a Forward-Guidance component measured around the press conference (OIS-2Y). Shocks are mapped to close-to-close returns over an event window ([t-1, t, t+1]), with inference based on Newey-West (HAC) standard errors; event-only and ECB "Monetary Event" (ME) robustness checks are reported. Results show significant, opposite-signed announcement-day effects for FTSE and S&P: Islamic indices relatively outperform on Target surprises and underperform on Guidance surprises (larger magnitudes for S&P), while FOMC surprises are not significant. MSCI differentials are weak and inconclusive. Collapsing ECB shocks into ME largely nets out same-day effects, with a modest (t+1) Guidance effect in event-only estimates. The findings are consistent with distinct funding-cost (Target) versus policy-path/term-premium (Guidance) channels and with provider-level composition differences.
A model based on evolving splines is proposed to address seasonal patterns in daily cash demand observed in banknotes issued by the national central banks of Spain and Germany. Statistical indexes are applied to measure and compare weekly, monthly and yearly seasonal variations in these two series. Changes in these seasonal patterns are identified. The main finding is that seasonal variations are less relevant in the German series. By contrast, in the Spanish case, the magnitude of seasonal variations is increasing. Therefore, high levels of dissimilarity are observed between both cases for each one of the three seasonal patterns. However, the lower values of complementarity indexes suggest that the shapes of these seasonal patterns are not so different.
This study investigates the relationship between monetary policy and financial stability by examining non-performing loans (NPLs) as a key indicator of credit risk in nine Southeast European countries over the period 2007-2022. The empirical framework is built on an unbalanced panel dataset and employs fixed effects (FE) estimation, feasible generalized least squares (FGLS), and panel-corrected standard errors (PCSE) to address potential econometric challenges and enhance the robustness of results. The findings reveal that lending interest rates are a significant determinant of NPLs, indicating that tighter monetary conditions increase repayment burdens and amplify credit risk. By contrast, the official exchange rate does not show a consistent effect on loan performance, a result that may reflect relative stability in currency markets and the presence of regulatory safeguards in the region. An unexpected but noteworthy outcome is the negative association between unemployment and NPLs, which could be attributed to more cautious borrowing behaviour in periods of economic uncertainty and the role of government support mechanisms. These results highlight the complex transmission channels through which monetary policy interacts with financial stability, while also underscoring the importance of country-specific conditions in shaping these dynamics. The study contributes to the literature by providing evidence from a region marked by transition, crisis episodes, and external vulnerabilities, offering insights into how monetary policy can be calibrated to safeguard the resilience of the banking sector.
The study investigates the determinants of bank non-performing loans (NPL) in European and African countries, focusing on 32 European and African countries from 2010 to 2021. The results based on the two-stage least squares regression methodology show that the number of commercial bank branch, bank liquid reserves to bank assets ratio, inflation rate, exchange rate, real interest rate and the lending rate are significant determinants of bank NPL in the full sample. Size of domestic private credit, bank capital to asset ratio, bank liquid reserve to bank asset ratio, unemployment rate, inflation rate, exchange rate, real interest rate and lending rate are significant determinants of bank NPL in European countries. Bank capital to asset ratio, bank liquid reserve to bank asset ratio and inflation rate in Africa are significant determinants of bank NPL in African countries. The implication of the results is that the determinants of bank NPL in European countries are not necessarily the drivers of bank NPL in African countries.
This paper explores the ongoing debate on the role of monetary policy in preventing and addressing financial crises, known as the "lean vs. clean" dilemma. The central question is whether central banks should act preventively to avoid financial bubbles and imbalances (the lean approach), or whether it is more effective to respond only after they burst, focusing on mitigating the consequences (the clean approach). Through a review of theoretical literature and historical experiences, the paper highlights the advantages and limitations of both approaches. Special emphasis is placed on post-crisis reforms and the role of macroprudential policy as a complementary instrument to monetary policy. The paper shows that neither approach offers a universally applicable solution, but places a slight emphasis on the lean approach and suggests that the new framework for monetary policy must include a combination of preventive measures, effective responses after a crisis outbreak, international coordination of central banks, as well as improvements in forecasting models and early warning systems.
The objective of this study is to examine the effect of corruption on non-performing loans in 17 Central and Eastern European countries from 2004 to 2021. The study investigates the influence of corruption on the incidence of non-performing loans while controlling for macroeconomic, bank-specific, and governance-related factors. The analysis is based on panel data, and the fixed effects method is applied. The results indicate that several variables, including corruption, unemployment, loan-to-deposit ratio, voice and accountability index, and credit growth significantly affect non-performing loans. The findings contribute to the existing literature by examining the role of external factors in the persistence of non-performing loans, revisiting this issue in the specific context of the Central and Eastern European region.
The traditional trade-off between banks' safety and income should be amended with a green factor creating the green trilemma. Banks must find the balance between the three mentioned goals. Firstly, we build up a formal model of the green trilemma and point out the need for incentives to support green lending. The introduction of green differentiated capital requirements can be a solution. However, there is little empirical experience about the application of this policy tool. Secondly, we assess the Green Preferential Capital Requirement Program (GPCRP) of the Central Bank of Hungary, which is a pioneer green supporting factor program. We measure the cost efficiency of this program. The cost is prudential, meaning that the benefit of prudential release is distributed between bank owners and green borrowers. The program's unit cost is much below the current EU ETS prices. Our results underline the effectiveness of the program: without material increase in prudential risk, the GPCRP contributes to avoiding significant amounts of carbon emission.
Many central banks adopted inflation targeting under pressure from the IMF. Adoption of inflation targeting happened on pretty favourable macroeconomic terms whose distinctive features were the absence of supply shocks, low budget deficit and foreign currency access. It was a 'period conducive to price stability' with inflation on a downward trajectory in many countries, especially developed ones, even before the introduction of inflation targeting. That could have contributed to efficiency of inflation targeting considering other monetary strategies. The most widely used model in designinig monetary policy under inflation targeting is a macroeconomic model of a small open economy from the group New Keynesian model. The results of the econometric analysis in this paper show that inflation targeting is an inefficient monetary strategy in the face of negative supply shocks (financial crises, pandemic, rising energy prices, tariffs), as it leads to rising interest rates, falling GDP, and rising unemployment. The results of the econometric analysis in this paper show that inflation targeting is an inefficient monetary strategy in the face of negative supply shocks (financial crisis, pandemic, rising energy prices, tariffs, etc.), which leads to rising interest rates, falling GDP, rising unemployment, and ultimately to an "inflationary pandemic".
Central banks that intend to implement Central Bank Digital Currency (CBDC) must decide whether to leverage blockchain as their technology or introduce digital currency on their own terms. We argue that the first option is not a good idea, as it implies the application of a very complicated, non-intuitive, and expensive solution, which cannot meet some central banks' expectations. Instead, we propose a special entity - a Digital Currency Bank (DCB) - as a method to implement CBDC. The DCB would use traditional information technology successfully employed by banks for years, instead of the intricate blockchain technology used to implement cryptocurrencies.
This study investigates whether the money supply (MS) is endogenously or exogenously determined in 10 developing Asian countries. The study implemented the Panel FMOLS, DOLS, and ARDL/PMG approaches with quarterly panel data from 1980Q1 to 2020Q4. The results reveal that bank lending and income positively influence the MS, while the inflation rate has a negative impact. These findings support the idea that the MS is endogenously determined while rejecting the view of the Monetarists that the MS is exogenously determined or that there is a "helicopter drop" of money. Central banks should espouse a flexible approach to monetary policy that considers the broader economic environment. Recognizing the endogeneity of the MS can lead to more prudent strategies for achieving sustainable economic outcomes.
The aims of the study include examining the relationship between assets, funding, income diversity, and bank performance using a panel dataset over the period ranging from 2011 to 2023 by applying a two-step system GMM procedure for South Asian banks. The findings reveal that diversity in funding sources and assets leads to decreases in the profitability of banks in South Asia. The findings imply that overdoing it in funding and asset diversity is not good for South Asian banks. However, the diversity in income sources causes the performance of South Asian banks to boost up. Implying that an increase in income sources causes an increase in the profits of South Asian banks. Moreover, the empirical analysis remains consistent for the outcome of well- and under-capitalized banks. The findings are also in line with the economics and finance theories, including portfolio diversification, resource-based theory, and agency hypothesis. The findings suggest that regulators, economists, policymakers, and managers should revise the composition of their banks' assets and funding sources to optimize benefits in South Asian regions.
Since the increases of policy interest rates in the years 2022-2023, a number of central banks are suffering significant losses from the materialisation of interest rate risk. These losses erode the capital buffers and raise questions about the cost-efficiency of monetary policy. This warrants a closer look at the topic of central bank profitability. What drives central bank profits? What is the problem with central bank losses exactly? And what possibilities do central banks have to influence their profits and manage public perception? In this paper we revisit these questions for central banks in general, with a particular focus on the Eurosystem and De Nederlandsche Bank.Although central bank losses can be an accepted consequence of necessary monetary policy (risks), they are regrettable as they constitute public money that could have been otherwise used for public purposes such as education and healthcare. But even low (positive) profits are undesirable. In general, central bank profits contribute to maintaining a strong balance sheet and support financial independence from the government. A central bank should preferably generate sufficient income over time to grow its capital in line with GDP (Gross Domestic Product). Here, we use the concept of "capital" in a broad sense, i.e. shareholder capital and provisions, acting as risk buffer. This risk buffer should develop in line with GDP as that is roughly proportional to the underlying latent risks of the central bank from the economy and the banking sector.Central bank profits are mainly driven by the monetary policy interest rates - which have little room for including "efficiency" considerations. However, central banks should understand the outlook of their profits under different (interest rate) scenarios. This is also important for Eurosystem national central banks and the ECB which are exposed to the financial consequences of the ECB's monetary policy decisions via income and cost sharing arrangements. Some of the balance sheet items allow for profitability considerations to be included in their management. The central bank's own investment portfolio is the most prominent example. With the significant losses of a number of central banks, it may be wise to consider profitability more explicitly in the central bank policies. This paper attempts to offer input on that question.
Procyclicality in the banking sector is one of the important indicators that may encourage the systemic risk in the banking system. This study examines banking behavior of the economy during the COVID-19 pandemic and analyze the amplitude and frequency of the credit cycle of Islamic and conventional banks. This study is primarily focused on the credit property of Islamic and conventional banks from 2014 to 2020 with application of Ordinary Least Square (OLS), Frequency Base Filter Analysis (FBF) and Turning Point Analysis. Our study finds that the size of Islamic bank's amplitude is larger than the size of conventional bank's amplitude. This is characteristic of Islamic banks based on the pattern of financing of the real sector. Meanwhile, conventional banks encourage the creation of bubble capital because it is related to the credit pattern grounded in speculative activities based on the interest system. Therefore, conventional banks need to encourage credit patterns based on capital. Meanwhile, the size of the frequency of Islamic banks has a longer frequency measure than conventional banks, but the number of cycles formed is the same as a perfect cycle.
