In short
- System stability peaked in 2017 and fell most sharply in 2020.
- In this sample, the state banks pulled system stability down the most.
- Use one time window consistently, because the window changes the result.
Summary of a peer-reviewed paper
How stable is a banking system, and who moves the needle?
A simple accounting measure can track how stable each bank is and how much each one adds to risk across the system. We tested it on seventeen years of Turkish banking data.
The Z-score asks a plain question: how far would a bank's profits have to fall before its equity is wiped out? It measures that distance in units of the bank's usual earnings volatility, so a higher score means a safer bank. It needs only accounting data, which makes it cheap to track and easy to compare.
Turkey makes an interesting test case. After a severe banking crisis in 2000 and 2001, reform made its banks considerably more resilient. Since then they have faced new capital and liquidity rules, a currency crisis and a pandemic. My co-author Murat Donduran and I used that history to see what the Z-score can tell a supervisor or a bank.
What we did
We studied Turkey's deposit-taking banks over 71 quarters, from March 2003 to September 2020. We built eight versions of the Z-score: one based on total assets and one on risk-weighted assets, each measured over four different time windows. We then used a "leave-one-out" method, recalculating the system's score with each bank removed, to see how much each bank contributes to risk across the sector.
What we found
The choice of time window changes the answer, so comparisons only make sense on the same window. The two versions of the score moved together until 2016, then drifted apart as new risk-weight rules, the LCR and a new accounting standard came in.
Stability across the system peaked in 2017, then slipped in 2018, the year the lira fell 23% against the dollar in a single day, and continued to deteriorate in 2019 and 2020. That 2020 fall was the largest in the decade, most likely driven by COVID-19.
The leave-one-out results were the most striking. The state banks had the largest effect on the system's score, and taking them out made the system look more stable. Smaller banks generally added less to systemic risk.
Why it matters
A leave-one-out Z-score is a simple way to see which banks move risk across a system, and a sudden fall can serve as an early warning. It has clear limits. It relies on accounting data and does not capture contagion, currency or concentration risk, so it should sit alongside other tools rather than replace them.
The paper
Bank stability and systemic risk measurement: application of Z-score variations to the Turkish banking sector