A Giffen good is something people consume more of as its price rises, violating supply and demand. The most common example of a Giffen good is potatoes in the Irish potato famine – as the price of potatoes went up, people were poorer, and so consumed more potatoes. This strange outcome happens because the substitution effect (the substitution across goods as relative prices change) is outweighted by the income effect (the adjustments that occur because of the budget constraint).
Giffen goods are little more than curiosities, but they can help you understand one of the perverse implications of the poor design of the NSFR. The NSFR, which the Federal Banking Agencies put out for public comment in June, is designed to encourage banks to maintain a more liquid balance sheet. It establishes incentives that lead banks to shift from less to what it defines as more stable funding, and from less liquid to more liquid assets. A number of flaws were introduced in the design of the NSFR in the final round of revisions, when rushed horse trading between the jurisdictions on the Basel Committee superseded sound design. (For a comprehensive discussion of the NSFR and its many design flaws, see this TCH Research Note from July.) For one, the regulation was hijacked to make reverse repos more costly in an effort to make matched repo books more costly. Specifically, whereas before a bank was required to hold no stable funding to back an overnight reverse Treasury repo, after the change the bank needed to fund the reverse repo with 10 percent stable funding (long-term debt or equity).
The problem is that reverse Treasury repos (overnight loans backed by Treasury securities) are one of the safest and most liquid assets – a regulation designed to make banks want to be more liquid is being used to discourage investment in a highly liquid asset. The change made to the NSFR leads to a greater need for stable funding, which encourages banks to shift toward more liquid assets such as reverse repo; this is the income effect. Although the change also encourages banks to substitute away from reverse repos, it is easy to think of simple and plausible cases where the “income effect” dominates the “substitution effect.”
For instance, consider a bank that is backed by a fixed amount — $1 — of equity and whose size and balance sheet composition is determined exclusively by the regulations it complies with. In addition to equity, the bank is funded by long term debt and Treasury repo. The bank can invest in loans, reverse repos, and cash. The NSFR requires the bank to fund loans with long term debt or equity. Loans earn more than reverse repos which earn more than cash, and equity is more expensive than long-term debt, which is more expensive than repo.
1. Size is determined by a leverage ratio requirement of 10 percent. Equity divided by assets equals 10 percent, so assets equal $10.
2. Long-term debt and repo are determined by a “total loss-absorbing capital” or “TLAC” requirement, which requires the bank to have long-term debt equal to 10 percent of assets, so long-term debt equals $1. Because liabilities have to equal assets minus equity, repo equals $8.
3. Now suppose the NSFR requires loans to be backed entirely by stable funding but requires no stable funding for reverse Treasury repos (the requirement before the final round of revisions). Equity and long-term debt count as stable funding but repo does not. The bank would extend $2 in loans and $8 in reverse repo (assuming that cash earns a substantially lower return than reverse repo, then the bank would hold no cash.)
3’. But suppose instead that the NSFR required 10 percent stable funding for reverse Treasury repos (the current proposal). The bank would make $1.10 in loans, cutting its lending nearly in half, and $8.90 in reverse repo, an 11 percent increase.
In this case, assuming simply that the bank complies with regulations, raising the stable funding requirement on reverse repo, a change that is intended to discourage reverse repo, would result in more reverse repo and less lending.