I am on the 2026–27 academic job market. My research sits at the intersection of macroeconomics, credit markets, and banking — combining theory with empirical analysis to understand how household and bank balance sheets shape the transmission of monetary and fiscal policy. Below you will find everything a committee needs; please email me with any questions.
Job Market Paper
Who Bears the Burden of Bank Capital Regulation? Buffers, Risk Weights, and the Supply of Credit
Job Market Paper
Abstract
Who bears the burden of bank capital regulation? Using narratively identified capital-requirement shocks and a quarterly panel of U.S. banks (1994–2019), we document two cross-sectional facts: within banks, commercial lending contracts more than real estate lending, reflecting differential risk weights; across banks, institutions with thin voluntary buffers cut lending most. The buffer gradient extends well beyond the two percent of bank-quarters at the regulatory frontier—which a capital constraint alone cannot produce—and so identifies a cost of regulatory capital that is not risk-weighted. A heterogeneous macro-banking model built on that mechanism reproduces the gradient as an untargeted moment, 1.32 against 1.32 in the data. Statutory size tiers locate the exposed banks poorly, explaining 1.8% of the cross-sectional variance in regulatory slack. Risk weights and size tiers redistribute the burden of a tightening but barely change its aggregate magnitude—a gross transition cost excluding the stability benefits of capital.In one paragraph: Who bears the burden of bank capital regulation? Using narratively identified capital-requirement shocks and a quarterly panel of U.S. banks (1994–2019), we document two cross-sectional facts: within banks, commercial lending contracts more than real estate lending, reflecting differential risk weights; across banks, institutions with thin voluntary buffers cut lending most. The buffer gradient extends well beyond the two percent of bank-quarters at the regulatory frontier—which a capital constraint alone cannot produce—and so identifies a cost of regulatory capital that is not risk-weighted. A heterogeneous macro-banking model built on that mechanism reproduces the gradient as an untargeted moment, 1.32 against 1.32 in the data. Statutory size tiers locate the exposed banks poorly, explaining 1.8% of the cross-sectional variance in regulatory slack. Risk weights and size tiers redistribute the burden of a tightening but barely change its aggregate magnitude—a gross transition cost excluding the stability benefits of capital.
Three-Minute Speech
Since the financial crisis, capital requirements have risen a lot. And the regime regulators built is deliberately differentiated — business loans carry higher risk weights than mortgages, and requirements step up at ten billion and fifty billion dollars of assets.
Both of those design choices presume something: that regulators know where the burden of capital regulation actually falls.
Now, there’s a large literature on whether requirements are worth it. It computes an aggregate cost — how much credit or GDP you give up — and weighs it against fewer crises. That literature is good, and I don’t argue with it. But it answers a question banks aren’t asking. When banks object to higher requirements, the honest reply isn’t “the benefits exceed the costs.” It’s that nobody has said clearly who pays them.
That’s the question in my title, and my answer is this. The burden is organised by two variables that appear on every bank’s balance sheet — and in no statute.
The first is the risk-weight composition of the loan portfolio. It determines which sectors absorb the contraction: commercial and industrial lending falls about a third more than real estate. The second is the voluntary capital buffer a bank holds above its requirement. It determines which banks absorb it: thin-buffer banks cut commercial lending about 1.3 times more than well-buffered ones.
And here is the finding I’d want a supervisor to take away. The statute sorts banks by size. The mechanism sorts them by buffer. Those two sorts are almost unrelated. Statutory tier explains under two percent of the cross-sectional variance in regulatory slack, and eighty-six percent of the banks in the thinnest-buffer tercile face the lowest statutory requirement.
So the instrument regulators actually control is aimed at nearly the wrong variable.
I show this in two steps. Empirically, I trace narratively identified capital-requirement shocks through a quarterly panel of about forty-eight hundred US banks using local projections. Structurally, I build a heterogeneous banking model where funding frictions make banks hoard precautionary capital, so the distribution of regulatory slack is an equilibrium object rather than something I assume.
The model reproduces the buffer gradient without targeting it — 1.32 in the model against 1.32 in the data.
And the policy message is a separation. The level of requirements determines the aggregate cost — about 1.4 percent of credit-market surplus per percentage point. The design — risk weights, size tiers — barely moves that total. What design does is decide who pays.
Debates about tailoring are debates about incidence. And current tailoring doesn’t identify the people bearing it.Application Materials
References
| Jonathan Thomas | Professor, School of Economics, University of Edinburgh · Jonathan.Thomas@ed.ac.uk |
| Corrado Di Guilmi | Associate Professor, University of Florence · corrado.diguilmi@unifi.it |
| Mengheng Li | Senior Lecturer, University of Technology Sydney · Mengheng.Li@uts.edu.au |
Teaching
I have taught macroeconomics, econometrics, forecasting, and programming/numerical methods at the University of Edinburgh, UTS, ANU, and UCA — see Teaching for the full list and my open course materials: Programming & Numerical Methods and Deep Learning for Macroeconomics (interactive notebooks used by my students).
Last updated: September 2026. For the most recent versions of all papers, see Research.