Data Note
An average lending rate can fall while the rate charged to every borrower category rises. A two-group decomposition identifies how changing loan volumes can reverse the interpretation of euro-area financing conditions.
A lower average lending rate does not necessarily mean that a comparable borrower can obtain cheaper credit. The average reflects both the rates charged and the business being written. If new lending shifts toward lower-rate categories, the published average can decline even while rates rise within every category.
The ECB's general description of MFI interest-rate statistics distinguishes rates on new agreements from rates applied to outstanding loan stocks. It also describes instrument categories, loan-size and interest-fixation breakdowns, and weighted euro-area aggregation. These definitions matter because the set of contracts behind an average can change from one observation to the next.
Consider two hypothetical categories of new loans, labeled A and B. The labels identify products, not individual borrowers or actual euro-area industries. Initially, A carries a rate of 4 percent and B a rate of 6 percent. Each accounts for half the amount of new business, giving an average rate of 5 percent.
In the next period, both category rates rise by half a percentage point, but A accounts for 90 percent of new lending:
| Category | Initial rate | Initial volume share | Next rate | Next volume share |
|---|---|---|---|---|
| A | 4.0% | 50% | 4.5% | 90% |
| B | 6.0% | 50% | 6.5% | 10% |
The next average is 0.9 × 4.5 + 0.1 × 6.5 = 4.7 percent. It falls by 0.3 percentage point, although both constituent rates increase. At the original composition, the next-period average would instead be 5.5 percent.
The difference between 5.5 and 4.7 is a composition effect of −0.8 percentage point under this decomposition. It offsets a within-category rate increase of 0.5 percentage point, leaving the observed aggregate decrease of 0.3. This is a transparent illustration, not a decomposition of an actual ECB release.
For an average , changing both the weights and the rates changes the result. An exact two-period decomposition can assign the rate effect the initial weights and the composition effect the final rates. Other ordering conventions allocate the interaction differently. The convention must be reported if contributions are compared.
Holding broad categories fixed does not necessarily hold borrower quality fixed. A loan-size band can still contain different maturities, collateral, industries and borrower risks. A credible estimate of the price offered to an otherwise comparable firm therefore needs more detailed controls than a comparison of two national averages.
Selection can also occur outside the observed rate. If some applicants receive no loan, their prospective borrowing costs do not enter an average of completed agreements. A decline in the average rate could accompany tighter access for the applicants no longer represented. The numerical example does not establish that such exclusion occurred; it identifies evidence that a credit-conditions assessment must examine separately.
The ECB includes new negotiations of existing contracts within its definition of new business, while automatic extensions without renegotiation are treated differently. Accordingly, a new-business measure is not simply the rate on newly constructed productive assets. An outstanding-stock measure answers another question: the terms applied across loans still on bank balance sheets.
A financing assessment should report which measure it uses, inspect the associated volumes and compare relevant product categories. For an existing borrower, contractual repricing dates remain decisive. For a prospective borrower, new agreements with comparable terms are more relevant. An average rate is informative once its changing composition is visible; it is not an offer that every firm could have accepted.