Financing costs, credit and capital formation in Albania: integrated reconstruction after three tests of the i-I model
Description
This paper analyzes the relationship between financing costs, credit and gross fixed capital formation (GFCF) in Albania in the quarterly period 2006Q4-2025Q4. The analysis uses the logarithm of gross fixed capital formation as a dependent variable, the logarithm of gross domestic product (GDP) and the consumer price index (CPI) as macroeconomic controls, spreads as indicators of the relative financing cost, and credit-to-GDP ratios as measures of financial intensity. The autoregressive distributed lag (ARDL)/unrestricted error-correction model (UECM) approach is used because the variables have mixed order of integration and the sample is limited. The results support the cost of capital channel: spreads are negatively related to investment and retain consistent interpretation in alternative specifications. The credit channel is interpreted with caution: credit-to-GDP ratios look more convincing as indicators of leverage, the denominator effect or balance sheet stress than as a pure positive funding channel. The evidence for the financial accelerator is partial because direct series on firm balance sheets, collateral and bank portfolio quality are missing. The main contribution is the reconstruction of a conditional macro-financial argument for the Albanian case, where bank-based financial structure and euroization make monetary transmission more complex than in the standard interest rate model. A sectoral extension based on INSTAT quarterly construction gross value added real growth is added as a production-side robustness check, with careful distinction between sectoral co-movement and structural causality.
Files
Steps to reproduce
1. Download all supplementary files listed with this submission. 2. Open the workbook HSSC_iI_construction_GVA_extension_tests_2026_07_28.xlsx. 3. Use core_iI_dataset_with_INSTAT_construction_GVA_extension.csv as the main quarterly dataset. 4. Use instat_construction_gva_real_yoy_quarterly.csv as the sectoral construction gross value added extension. 5. Reproduce the descriptive statistics, unit-root tests, UECM/ARDL specifications, HAC-robust coefficients, long-run effects, diagnostics, and Granger-causality checks using the corresponding CSV files: unit_root_tests_adf_pp_kpss.csv; uecm_hac_coefficients.csv; uecm_diagnostics_and_model_selection.csv; uecm_long_run_coefficients_delta_method.csv; granger_tests_sectoral_extension.csv; Table_4_sectoral_extension_model_diagnostics.csv; Table_5_sectoral_extension_long_run_effects.csv. 6. Recreate the reported sectoral extension figure using Figure_6_sectoral_construction_extension_results.png as the visual reference and the table CSV files as the numerical source. 7. Compare the reproduced outputs with the tables and figures reported in the manuscript. The analysis can be reproduced in any standard econometric environment capable of estimating ARDL/UECM models, unit-root tests, HAC-robust standard errors, long-run coefficient transformations, and Granger-causality tests. The submitted workbook and CSV files contain the data and reported outputs used in the manuscript.
Institutions
- University of TiranaTirana, Tirana