What High Profits Tell Us—and What They Do Not

The evidence from Mauritania’s banking system now permits a direct answer to the series’ opening question. High bank profits do not necessarily mean strong financial intermediation because profitability and lending are different balance-sheet outcomes, and because credit is concentrated in a few institutions whose lending movements carry disproportionate weight. This final synthesis connects the EADG puzzle, the granular mechanism, the lending and profitability regressions, and the policy choices they support—while keeping the case’s limits visible.

The answer is a transmission problem, not a profit problem

The central finding is not that profit is unimportant or suspicious. A profitable bank may be building capital, provisioning against losses, preserving liquidity, or being compensated for operating and credit risk. Those are legitimate financial functions. The finding is that none of them, by itself, establishes that the banking system is expanding access to credit or deepening intermediation. Earnings describe what the bank captures; intermediation describes what funding and balance-sheet capacity make available to the wider economy.

The EADG diagnostic puts that distinction in the right order. It identifies a situation in which the efficiency or profitability dimension runs ahead of access and depth. That gap does not prove that market power caused every profit, or that high margins are pure rents. It says that the income side and the service side do not move together closely enough to be treated as one outcome. In Mauritania, the paper uses that mismatch as a starting point for asking which institutions and which shocks actually determine the flow of credit.

The answer depends on concentration. The paper’s comparison of system-wide lending with the five largest banks and the remaining institutions shows that aggregate credit is overwhelmingly driven by the largest lenders. The country’s final analytical panel contains 15 banks observed monthly from June 2020 through June 2024, but the number of banks is less important than the uneven distribution of their balance sheets. When a few banks hold a large share of system assets or loans, an unexpected movement at one of them is not diluted in the way it would be in a system of many equally sized institutions.

Granularity is the link between that structure and the aggregate outcome. The researchers remove selected common macroeconomic conditions, observed bank characteristics, and persistent bank differences. They then weight the remaining bank-level residuals by lagged bank size. This produces a measure that preserves exposure: a residual at a dominant bank counts more because more of the system is attached to that bank’s balance sheet. The equal-weight alternative removes precisely that exposure difference.

The lending results show why this matters. The macro-only aggregate regression has an R-squared of 0.110. Adding any of the eight size-weighted lending-shock measures produces positive coefficients from 0.634 to 0.923, all reported at p < 0.01, and raises R-squared to between 0.872 and 0.981. The paper describes the relationship as close to one-for-one and summarizes the contribution as roughly 80 percent of short-run financial-intermediation fluctuations. The reported association is unusually stable across the alternative first-stage specifications.

The equal-weight placebo makes the interpretation more specific. When the same broad residual idea gives every bank equal influence, the coefficients are about −0.003 with standard errors about 0.007. R-squared rises only from 0.110 to at most 0.115, and adjusted R-squared falls slightly. The result is not simply that unexplained bank movement predicts credit. It is that unexplained movement attached to a skewed bank-size distribution carries the aggregate relationship in this case.

Profitability behaves differently. The eight size-weighted profitability-shock coefficients range from 0.031 to 0.064, with standard errors around 0.088–0.091.