📈 economics

Where Is Measured Financial Development Common? Components, Frequency, and Long-Run Income in 142 Economies

This paper analyzes financial development across 142 economies from 1993 to 2023, finding that while the IMF Financial Development Index shows high fit with income, this performance is largely attributable to high-persistence common factors rather than a distinct financial driver, with institutional access and market depth emerging as more informative components than efficiency.

Original authors: Thomas Panagiotou, Constantinos Katrakilidis

Published 2026-09-04

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Thomas Panagiotou, Constantinos Katrakilidis

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Economies are like living organisms that need a circulatory system to grow. In the world of finance, this system is made up of banks, stock markets, and the rules that govern them. For decades, economists have tried to measure how well this system works and whether a stronger financial network automatically leads to a richer country. The hope has been that if a nation builds deeper banks and wider access to credit, its people will inevitably become wealthier. To track this, researchers use a composite score called the Financial Development Index, which combines many different measurements into a single number. This score looks at how big the financial system is, how many people can reach it, and how efficiently it operates. The big question has always been whether this score truly captures the unique economic magic of a specific country or if it is simply reflecting a global trend that lifts everyone at the same time.

A team of researchers from the Aristotle University of Thessaloniki decided to test the reliability of this score by looking at data from 142 different economies over a thirty-year period, from 1993 to 2023. They wanted to know if the high correlation often seen between financial development and national income was a sign of a powerful economic force or just a statistical illusion caused by the way the data moves over time. They found that when financial numbers and income numbers both drift upward steadily over decades, they naturally look like they are connected, even if they are not influencing each other directly. To check this, the team created a simulation where they generated 500 sets of completely independent economic histories. They programmed these imaginary economies to have the same steady, slow upward drift as the real ones, but with no shared causes or common forces linking them. When they ran their standard tests on these fake, unrelated histories, the results were startling: the simulated economies showed an even stronger statistical link between finance and income than the real world did. This suggests that the high scores often seen in real data might just be a side effect of time and persistence, rather than proof of a deep economic connection.

The researchers then peeled back the layers of the financial index to see if some parts were more meaningful than others. They broke the big score down into six smaller pieces: the depth, access, and efficiency of both financial institutions (like banks) and financial markets (like stock exchanges). They discovered that the parts of the index measuring the sheer size of the system and how many people can reach it were the ones that moved most closely together across all countries. These "size and reach" numbers rose and fell in sync globally. However, the parts of the index that measured efficiency—how cheaply and quickly money moved through the system—did not move in lockstep. The efficiency numbers were much more unique to each country and less influenced by global trends. This tells us that while the expansion of financial systems is a global phenomenon, the actual quality of how those systems work remains a local story.

The study also looked at how these connections changed depending on the time frame. When the researchers examined the data year by year, the strong global link almost vanished. The countries did not seem to be changing their financial systems in unison from one year to the next. The strong connection only appeared when looking at the long, smooth paths of the data over many years. This means that while the overall size of financial systems tends to grow together worldwide, the specific changes happening within a single year are driven by local conditions and decisions, not by a single global wave.

Finally, the team tested whether having a better financial system actually led to higher income in the long run. When they used the most careful statistical methods to account for these global trends, the link between financial development and income became very weak and uncertain. The data did not provide a clear answer on whether building a bigger financial system guarantees a richer population. However, they did find a clearer pattern regarding financial openness, which refers to how easily a country allows money to flow in and out across its borders. In countries that already had strong institutions and a decent level of wealth, opening up to foreign capital was associated with higher income. In countries without that foundation, opening up did not show the same benefit. This suggests that the ability to use foreign money effectively depends heavily on a country's existing capacity to manage it. The study concludes that while financial development indices are useful for describing the scale of a system, their high correlation with income is often a statistical artifact of slow-moving trends, and the true economic benefits depend heavily on the specific local conditions and the quality of a country's institutions.

Technical Summary: Where Is Measured Financial Development Common? Components, Frequency, and Long-Run Income in 142 Economies

Problem Statement
Cross-country estimates of the relationship between financial development (FD) and long-run income remain unsettled, with results varying significantly by sample, period, and estimator. While existing literature attributes this instability to reverse causality, nonlinearity, and parameter heterogeneity, this paper posits a more fundamental issue: measurement instability. Specifically, the authors question whether high-level explanatory shares (common-proxy fits) observed in the IMF Financial Development Index (FDI) reflect genuine shared economic forces or are artifacts of persistent trends and mechanical index construction. The paper investigates whether the index cleanly separates country-specific financial change from common global expansion, noting that theory assigns income relevance to screening and productive allocation rather than mere financial size or participation.

Methodology
The study utilizes an annual panel of 142 economies from 1993 to 2023. The empirical strategy relies on "common-proxy regressions" and a high-persistence null benchmark to disentangle measurement artifacts from economic substance.

Common-Proxy Share Measurement: For each country, the authors regress its financial development series on the cross-sectional averages of other variables (leave-one-out design). The arithmetic mean of the resulting country-level $R^2$ values constitutes the "common-proxy share," measuring the statistical association with international benchmarks.
Design Safeguards: To prevent mechanical inflation of the fit, three increasingly strict designs are employed:
- Target Exclusion: Excludes the international average of the target variable itself.
- Parent-Index Exclusion: Removes the Financial Institutions (FI) and Financial Markets (FM) sub-index averages when the target is the composite FD index.
- Symmetric External-Proxy Design: Conditions both FD and a physical comparator (energy use) on an identical set of external variables (income, openness, ecological footprint, CO2), removing FD, energy, FI, and FM from the information sets.
High-Persistence Null Benchmark: To test if high $R^2$ values are merely a function of trending data, the authors simulate 500 independent panels of 142 economies with calibrated drift and autoregressive persistence ( $\rho$ ) ranging from 0.90 to 0.99. If the empirical $R^2$ exceeds the simulated benchmark, it suggests shared economic forces; if not, the fit may be driven by persistence alone.
Component and Frequency Analysis: The FDI is disaggregated into six components (Institutional/Market Depth, Access, Efficiency). The analysis compares these components across four data transformations: levels, country-specific detrended levels, annual differences, and three-year differences.
Long-Run Income Estimation: A heterogeneous Cross-Sectionally Augmented ARDL (CCE-ARDL) model is used. The preferred specification is selected based on residual dependence diagnostics (Pesaran CD, CD*, and dependence exponent) rather than the sign of the finance coefficient.

Key Results

The Persistence Benchmark: The primary finding is that the magnitude of the common-proxy share is not informative on its own. At a persistence level of $\rho = 0.97$ , all 500 simulated independent panels produced a higher mean $R^2$ than the empirical data across all designs. The empirical shares (0.823 under target exclusion, 0.749 under parent-index exclusion, and 0.726 under the symmetric design) fall within the distribution of the placebo simulations. This indicates that high level fit can arise from shared persistence without shared economic innovations.
Component Heterogeneity: While absolute levels are uninformative, comparative differences are robust. Under identical proxy spaces:
- Institutions: Access (0.884) and Depth (0.852) exhibit significantly higher common-proxy fit than Efficiency (0.592).
- Markets: Depth (0.836) exceeds Access (0.701) and Efficiency (0.666).
- Interpretation: Measures of system size and reach are more closely aligned across countries than measures of efficiency.
Frequency Sensitivity: The alignment is concentrated in persistent trajectories. When data is transformed into annual differences, the common-proxy shares drop sharply (e.g., Institutional Depth falls from 0.852 in levels to 0.327 in annual changes). This suggests that year-to-year financial changes are largely country-specific, while the commonality lies in long-run trends.
FD vs. Physical Comparators: Under the symmetric design, FD and energy use per capita do not differ statistically in their common-proxy fit (0.726 vs. 0.750), though equivalence within five percentage points is not established.
Long-Run Income Associations:
- The preferred factor-augmented coefficient for composite FD is positive (0.114) but statistically imprecise ( $p=0.119$ ).
- Disaggregating the index does not yield stable, precise coefficients for any specific component.
- Financial openness shows a conditional gradient: it has a more positive association with income in economies with higher initial income or stronger institutional capacity, consistent with the "absorptive capacity" hypothesis.

Significance and Claims
The paper claims three primary contributions, maintaining a modest tone regarding causal inference:

Benchmarking Necessity: It demonstrates that a large level $R^2$ in persistent international panels requires a suitable high-persistence benchmark before it can be assigned economic meaning. Without such a benchmark, high fit may simply reflect shared drift.
Internal Index Structure: It identifies clear, descriptive differences within the IMF index. Commonality is concentrated in dimensions measuring system scale and reach (Depth and Access) rather than efficiency. This suggests that the "common" movement in financial development indices may reflect global expansion and formalization rather than improvements in the core financial functions of screening and monitoring.
Measurement vs. Estimation: It separates sensitivity to common-factor augmentation from sensitivity to the structural conditioning set. The instability in income estimates appears linked to the lack of a robust, precise coefficient once common factors are adequately absorbed, rather than a failure of the estimation technique itself.

The authors conclude that while broad financial-development indices are useful summaries of system scale and participation, high level fit is not a direct measure of intermediation quality or a uniform policy effect. The evidence suggests that the "common" element in financial development is largely a feature of persistent, slow-moving trajectories shared across economies, rather than a synchronized response to global financial shocks in the short run.

More like this