Table of Contents
1. Introduction & Overview
This research investigates the equilibrium dynamics between demand and supply for foreign currency within the non-cash segment of the Ukrainian Interbank Foreign Exchange Market (UIEM). The study is motivated by the persistent challenges faced by emerging economies like Ukraine in managing exchange rate volatility and capital flows. The authors posit that the observed trade-offs in the forex market are a direct outcome of the existing foreign exchange arrangement, administrative measures enacted by the National Bank of Ukraine (NBU), and a set of fundamental economic variables critical to Ukraine's context.
The core objective is to construct and analyze an equilibrium model to uncover problematic aspects of market functioning, providing insights for more effective monetary policy.
2. Research Methodology & Model Framework
The study employs a Factor-Augmented Vector Autoregression (FAVAR) modeling approach to build the equilibrium model. Empirical data from the UIEM is utilized, segmented into distinct periods as proposed by the authors to account for structural breaks or regime changes.
2.1. FAVAR Modeling Approach
The FAVAR framework extends the traditional VAR model by incorporating a large set of informational variables summarized by a few estimated factors. This is particularly useful for capturing the influence of many potential fundamental variables without facing the "curse of dimensionality." The model can be represented in state-space form, where the factors are treated as latent variables.
2.2. Data Segmentation & Periods
A crucial step involved splitting the time-series data into specific periods. This segmentation likely corresponds to different phases of NBU policy (e.g., periods of strict administrative controls versus more liberalized phases) or significant economic events, allowing the model to capture non-linearities and structural shifts in the equilibrium relationship.
3. Model Specification & Technical Details
3.1. Log-Linearized Specification
The paper presents a log-linearized specification of the equilibrium model. Log-linearization is a common technique to transform non-linear economic relationships into a linear form suitable for estimation, often around a steady state. For an equilibrium condition $S(P, Z) = D(P, X)$, where $S$ is supply, $D$ is demand, $P$ is price (exchange rate), and $Z$ and $X$ are vectors of supply and demand shifters, the log-linearized version might take a form like:
$\hat{s}_t = \alpha_s \hat{p}_t + \beta_s' \hat{z}_t$
$\hat{d}_t = -\alpha_d \hat{p}_t + \beta_d' \hat{x}_t$
Equilibrium implies $\hat{s}_t = \hat{d}_t$, solving for the equilibrium log-price $\hat{p}_t^*$.
3.2. Cointegration Analysis
The efficiency of testing for cointegration among the fundamental variables' time series is reported. Cointegration tests (e.g., Johansen test) are essential to determine if there exists a long-run equilibrium relationship among non-stationary variables. The results are presented as critical statistics values, indicating whether a stable long-run relationship between demand, supply, and their determinants exists.
4. Empirical Results & Analysis
4.1. GAP Analysis of Equilibrium Deviations
The authors propose and implement a GAP analysis tool. This involves calculating the deviation of the actual exchange rate or market state from the model's implied equilibrium path ($GAP_t = Y_t - Y_t^*$). Analyzing these gaps helps identify periods of market overvaluation or undervaluation and assess the persistence of disequilibrium.
4.2. Disconnection Properties in the Model
A significant finding discussed is the "disconnection properties" within the model. This likely refers to instances where the traditional link between fundamental variables (e.g., interest rate differentials, trade balance) and the exchange rate breaks down or becomes weak, possibly due to dominant administrative interventions or market segmentation.
References
19
Figures
3
Tables
5
5. Policy Implications & Regulatory Analysis
The study provides a detailed analysis of the NBU's regulatory style. It critically examines the impact of administrative controls versus market-based mechanisms. A key argument is that heavy-handed interventions, while potentially stabilizing in the short term, can create distortions, shortages, and increased volatility, as evidenced by the "disconnection" findings.
6. Key Findings & Conclusions
The research concludes that the increased share of cash held outside the banking system (de-dollarization in the form of physical cash hoarding) has significantly undermined price stability in Ukraine. The paper's central policy recommendation is that NBU interventions would be more effective if a flexible exchange rate regime is coupled with a credible and flexible inflation-targeting framework. This combination could help anchor expectations and reduce the need for disruptive administrative measures.
7. Original Analysis: Core Insight & Critical Evaluation
Core Insight: This paper delivers a crucial, albeit painful, diagnosis: Ukraine's forex market dysfunction is a self-inflicted wound. The NBU's historical reliance on blunt administrative controls, while politically expedient, has systematically eroded the very market mechanisms needed for a stable equilibrium. The identified "disconnection properties" aren't a statistical anomaly; they are the scar tissue of repeated policy interventions, severing the link between economic fundamentals and price signals. This aligns with the broader literature on emerging market forex regimes, such as the work of Calvo and Reinhart (2002) on "fear of floating," where the desire for stability paradoxically breeds fragility.
Logical Flow: The authors' logic is robust. They start from the observable dilemma (volatility vs. shortage), build a sophisticated FAVAR model to quantify the equilibrium, and use its breakdowns (the gaps and disconnections) as forensic evidence to pinpoint policy failure. The use of GAP analysis is particularly astute—it transforms abstract model output into a tangible dashboard for policy error measurement.
Strengths & Flaws: The major strength is the application of a high-dimensional FAVAR model to a messy, intervention-driven market. This is a significant technical contribution, moving beyond simple OLS or standard VARs that would fail in this environment. However, the paper's flaw is its vagueness on the "fundamental variables." For a model-centric paper, the opacity of the factor composition is a critical weakness. It echoes the "black box" criticism sometimes leveled at machine learning in finance—great predictive power, limited explanatory insight. Furthermore, while citing the BIS or IMF on inflation targeting would strengthen the argument, the external referencing is light.
Actionable Insights: For the NBU and similar institutions, the message is clear: Stop fighting the market. The path forward isn't more sophisticated control but a credible commitment to a rules-based framework. The paper implicitly argues for a transition similar to Poland's successful shift to Inflation Targeting. The technical recommendation is to institutionalize the GAP analysis as a real-time monitoring tool to guide market-conforming interventions (e.g., smoothing operations) rather than market-defying ones (e.g., hard caps). The future of Ukraine's monetary stability depends less on perfecting the model of a distorted market and more on having the courage to stop distorting it.
8. Technical Appendix
8.1. Mathematical Formulations
The core equilibrium condition can be derived from the log-linearized supply and demand functions:
$\hat{p}_t^* = \frac{\beta_d' \hat{x}_t - \beta_s' \hat{z}_t}{\alpha_s + \alpha_d}$
Where $\hat{p}_t^*$ is the log-deviation of the equilibrium exchange rate. The FAVAR model incorporates dynamic factors $(F_t)$ representing the unobserved fundamental drivers:
$\begin{pmatrix} Y_t \\ F_t \end{pmatrix} = \Phi(L) \begin{pmatrix} Y_{t-1} \\ F_{t-1} \end{pmatrix} + v_t$
where $Y_t$ contains observable market variables (exchange rate, volumes), and $F_t$ is estimated from a large dataset of potential fundamentals.
8.2. Experimental Results & Chart Descriptions
Figure 1 (Hypothetical Reconstruction): Likely depicts the estimated equilibrium exchange rate path ($\hat{p}_t^*$) against the actual observed exchange rate. Periods of significant and persistent positive GAP (actual > equilibrium) would indicate overvaluation, often preceding a correction or requiring NBU supply interventions.
Figure 2: Probably illustrates the estimated dynamic factors $(F_t)$ extracted by the FAVAR model. One factor might correlate with global risk sentiment (like a VIX index for Ukraine), another with domestic monetary policy stance, and a third with terms of trade or current account dynamics.
Figure 3: Could show the results of the GAP analysis over time, highlighting specific episodes (e.g., 2014 crisis, post-2015 stabilization) where deviations from equilibrium were extreme, along with annotations of major NBU policy actions during those periods.
Tables (1-5): Would present descriptive statistics, unit root and cointegration test results (Johansen trace and max eigenvalue statistics), FAVAR model estimation outputs (factor loadings, variance decompositions), and regression results for the GAP analysis on policy variables.
8.3. Analysis Framework: A Conceptual Case Study
Scenario: Analyzing the impact of a sudden stop in capital inflows.
Framework Application:
1. Data Input: Update the dataset with high-frequency indicators: NBU reserves data, non-resident portfolio flow data, CDS spreads, and interbank offer rate spreads.
2. Factor Estimation: The FAVAR model would immediately show a shift in the "capital flow factor" and the "risk perception factor."
3. Equilibrium Shift: The model's implied equilibrium exchange rate ($p_t^*$) would depreciate, reflecting the reduced supply of foreign currency from inflows.
4. GAP Analysis: If the actual exchange rate is pegged or slow to move, a large negative GAP (actual < equilibrium) emerges, signaling mounting devaluation pressure.
5. Policy Insight: The model quantifies the pressure. A small, temporary GAP might be ignored. A large, growing GAP indicates the need for a policy response: either allow the exchange rate to adjust (flexible regime) or prepare to expend significant reserves to defend the peg, with the model estimating the potential scale of intervention needed.
9. Future Applications & Research Directions
1. Real-Time Monitoring System: This FAVAR-GAP framework can be operationalized into a real-time dashboard for central banks, providing early warning signals of market misalignment and stress.
2. Machine Learning Integration: Future work could replace or complement the FAVAR's factor estimation with non-linear dimension reduction techniques from machine learning (e.g., Autoencoders, as used in feature extraction for image data like in the CycleGAN framework, but applied to financial time series) to capture more complex, non-linear relationships among fundamentals.
3. Cross-Country Analysis: Applying the same methodology to a panel of emerging markets (e.g., Georgia, Moldova, Serbia) could identify common patterns of disequilibrium and the effectiveness of different policy responses, contributing to the academic literature on optimal forex regimes in transition economies.
4. Agent-Based Model (ABM) Calibration: The empirical results from this equilibrium model, especially the disconnection properties, could be used to calibrate the parameters of an Agent-Based Model of the UIEM, simulating how different trader behaviors (e.g., herd mentality, heterogeneous expectations) interact with central bank rules.
10. References
- Bernanke, B. S., Boivin, J., & Eliasz, P. (2005). Measuring the effects of monetary policy: a factor-augmented vector autoregressive (FAVAR) approach. The Quarterly Journal of Economics, 120(1), 387-422.
- Calvo, G. A., & Reinhart, C. M. (2002). Fear of floating. The Quarterly Journal of Economics, 117(2), 379-408.
- International Monetary Fund. (2020). Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER). Washington, DC: IMF.
- Johansen, S. (1991). Estimation and hypothesis testing of cointegration vectors in Gaussian vector autoregressive models. Econometrica, 59(6), 1551-1580.
- Kuznyetsova, A., Misiats, N., & Klishchuk, O. (2017). The equilibrium model of demand and supply at the Ukrainian Interbank Foreign Exchange Market: disclosure of problematic aspects. Banks and Bank Systems, 12(4), 31-43.
- National Bank of Ukraine. (Various Years). Monetary Policy Reports. Kyiv: NBU.
- Zhu, J. Y., Park, T., Isola, P., & Efros, A. A. (2017). Unpaired image-to-image translation using cycle-consistent adversarial networks. Proceedings of the IEEE international conference on computer vision (pp. 2223-2232).