Companion DDataset for FMVM Calibration and Validation: Kazakhstan, 2015–2024 (Monthly)
Description
This dataset accompanies the study “Multi-Premium Valuation in Frontier Markets: Evidence from Kazakhstan” and provides the full calibration inputs, component estimates, and validation benchmarks for the Frontier Market Valuation Model (FMVM) applied to Kazakhstan. It covers the period 2015Q1–2024Q4, with a companion monthly dataset constructed by linear interpolation for visualization and alignment with valuation anchors. The dataset is organized in a structured Excel file and contains: Core FMVM components: Sovereign risk premium (CRP), Liquidity premium (LP), Behavioral premium (BP), and Institutional quality premium (IQP), along with their sum (FMVM-implied cost of equity). Baseline model scenarios: CAPM (Rf + β×GERP), CAPM+CRP, and FMVM (full specification). Market anchors: Inverse P/E ratio and earnings yield, used for empirical validation of model estimates. Regime flags: Indicators for four structural periods in Kazakhstan’s financial development—(i) 2015 FX float and devaluation transition, (ii) 2017–2019 stabilization and AIFC/AIX launch, (iii) 2020–2021 COVID-19 and oil price shock, and (iv) 2022–2024 unrest, tightening, and disinflation. Cross-country comparison inputs: Benchmark estimates for Azerbaijan, Georgia, and Uzbekistan, harmonized to the same CAPM and FMVM structure for regional positioning. The dataset supports all calibration, estimation, and out-of-sample validation exercises reported in the paper, including regression fits (Table 6), predictive error metrics (Table 6b), and comparative valuation (Table 7). All core analyses are based on quarterly data (2015Q1–2024Q4), while the monthly version ensures smoother presentation of figures (e.g., Figure 2). By providing both raw inputs and processed FMVM outputs, the dataset allows replication of the study’s results, extension to alternative specifications (e.g., LP scaling, IQP multipliers), and further comparative research on frontier equity markets. File format: Excel (.xlsx) Temporal coverage: January 2015 – December 2024 (monthly companion); Q1 2015 – Q4 2024 (quarterly core) Geographic coverage: Kazakhstan (with comparator entries for Azerbaijan, Georgia, Uzbekistan) Intended use: Academic research, policy analysis, and replication of FMVM calibration and validation in credibility-sensitive frontier markets.
Files
Steps to reproduce
Steps to Reproduce 1. Obtain the dataset. Download the Excel file Kazakhstan FMVM Companion Dataset (Monthly, 2015–2024). The file contains quarterly calibration data and a companion monthly series interpolated for visualization. 2. Identify the components. Use the four FMVM components provided— a. Sovereign Risk Premium (CRP), b. Liquidity Premium (LP), c. Behavioral Premium (BP), d. Institutional Quality Premium (IQP)— along with the baseline CAPM (Rf + β×GERP) and CAPM+CRP scenarios. 3. Reconstruct FMVM. Apply the additive specification: kₑ= Rf + (β × GERP) + CRP + LP + BP + IQP The dataset includes both the individual components and the summed FMVM cost of equity. 4. Replicate model validation. a. Run OLS regressions of model-implied cost of equity against valuation anchors (inverse P/E, earnings yield) for in-sample fit (2015Q1–2024Q4). b. Perform out-of-sample validation by splitting 2015Q1–2021Q4 (train) and 2022Q1–2024Q4 (test), freezing coefficients estimated in the training window. c. Compute RMSE and MAE as in Tables 6 and 6b of the paper. 5. Conduct realized-return check. Test the directional relationship between changes in ERP (ΔERP) and subsequent quarterly USD excess returns for 2022–2024 to verify predictive content. 6. Use regime flags. Filter data by regime dummies (2015 float, 2018 AIFC/AIX launch, 2020–21 COVID-19 and oil shock, 2022–24 unrest/disinflation) to replicate Figure 2 and Table 5. 7. Cross-country comparison. Use the benchmark estimates for Azerbaijan, Georgia, and Uzbekistan included in the file to replicate Table 7. CAPM assumptions (Rf=4.0%, GERP=5.0%, β=1.0) are fixed across countries; differences come from CRP, LP, BP, and IQP. 8. Sensitivity analysis. To reproduce Appendix A, vary: a. LP scaling factor γ<sub>LP</sub> ∈ {0.50, 0.75, 1.00, 1.25, 1.50}, b. IQP multiplier m ∈ {10, 12, 15, 18, 20}, and recompute out-of-sample errors to confirm robustness. 9. Visualization. Use the monthly companion dataset for smoother graphical representation of co-movement between FMVM estimates and valuation anchors (e.g., Figure 2). By following these steps, researchers can fully reproduce the calibration, estimation, validation, and comparative analysis reported in the study. The dataset is designed to be transparent, modular, and extensible for future applications to other frontier and emerging markets.