I spent a long time watching the central bank balance sheet like it was the only liquidity number that mattered. Reserves go up, risk assets go up, that was the whole model. It works often enough that you stop questioning it. Then you hit a stretch where the balance sheet is flat or even shrinking and markets keep grinding higher, and the tidy story falls apart. What I was missing is that base money is not the same thing as money that actually reaches a business or a household. That part comes from commercial banks deciding to lend.
The credit impulse is the tool that fills that gap. It is not a new idea, credit strategists have used a version of it for years, but it is underused by the average trader building a macro dashboard. The short version is that it measures the change in the flow of new credit, not the level of outstanding credit. That second derivative framing is the whole point, and it is also the part people get wrong the first time they try to build it.
Why the flow matters more than the level
Outstanding credit is a big, slow number. Total loans in an economy do not move much quarter to quarter, so staring at the level tells you almost nothing about the margin, and the margin is where growth and asset prices get set. What you care about is the change in the flow of new borrowing. If new lending is accelerating, there is fresh money hitting the real economy this period that was not there last period. If new lending is decelerating, the marginal buyer of goods, houses, and equipment is quietly stepping back, even while the total loan book still looks healthy.
Mechanically, credit impulse is the change in new credit creation over a period, scaled by the size of the economy. The common construction is the year over year change in the flow of credit, divided by nominal GDP, so you get a percentage that is comparable across time and across countries. Because it is a change in a flow, it is effectively a second derivative of the credit stock, and second derivatives turn before the thing they are derived from. That is the source of the lead.
Constructing it from loan growth data
You do not need anything exotic. Total commercial bank loans, or total private sector credit, is the base series. Most central banks and statistical agencies publish it monthly or quarterly. Here is the workflow I use, and it is deliberately boring.
- Start with the level of outstanding private credit or total bank loans. Get it in nominal local currency, seasonally adjusted if the source offers it.
- Take the flow, which is the change in that level over a period. A quarterly change works, a rolling four quarter change smooths the noise if the raw series is jumpy.
- Take the change in that flow versus the same period a year earlier. This is the acceleration step, and it is the step people skip.
- Divide by nominal GDP for the same period so the number is a percent of output and comparable over time.
- Plot it as a line that crosses zero. Above zero, credit creation is accelerating. Below zero, it is decelerating, even if loans are still growing in absolute terms.
That last point is the one worth sitting with. The credit impulse can be deeply negative while total credit is still expanding. All that means is credit is expanding more slowly than it was. Markets tend to trade the acceleration, not the level, which is exactly why the gauge is useful and also why it feels counterintuitive at first.
Why it often leads the balance sheet read
The central bank balance sheet is a policy instrument. It moves when a committee decides it should move, and those decisions arrive with lags, meetings, and forward guidance. Commercial bank credit is a market outcome. It responds to loan demand, credit standards, collateral values, and bank risk appetite in more or less real time. So the credit impulse frequently turns before the balance sheet does, because lending conditions on the ground shift before the policy response catches up, and sometimes in the opposite direction.
There is a caveat I would not skip. The lead is not fixed and it is not guaranteed. Credit impulse has historically led activity and risk appetite by somewhere in the range of a few quarters, but that lag wanders. In economies where non bank credit, corporate bond issuance, and shadow lending are large, a bank loan series will miss a chunk of the real credit creation and the signal degrades. China's impulse gets a lot of attention precisely because bank lending there is a bigger share of total credit, so the gauge captures more of the picture. In a bond heavy economy you want to widen the base series or treat the read with more humility.
Slotting it into a liquidity composite
I would not trade the credit impulse on its own, and I do not think anyone should. It is one input. The way it earns its place is inside a composite that also carries the balance sheet flow, reserve balances, a fiscal flow proxy, and maybe a cross border dollar liquidity measure. Each input covers a blind spot in the others. The balance sheet tells you what the central bank is doing. The credit impulse tells you what banks are actually doing with it. When both point the same way, the signal is loud. When they diverge, that divergence is the interesting part, and it usually means one of them is about to catch up to the other.
A few rules of thumb I have landed on. Normalize every component to a z score before you combine them, otherwise the biggest, noisiest series dominates the composite by accident. Weight by how much of the economy each input actually touches rather than equal weighting out of laziness. And track the components separately even after you build the composite, because the aggregate can look calm while two inputs are quietly canceling each other out, and that cancellation is a signal in itself. When I fold a liquidity read into the market scorecards on Blockcircle, the credit impulse sits next to the balance sheet line rather than replacing it, for exactly that reason.
The failure mode to watch for is treating the gauge as a timing tool. It is not. The lead is real but loose, the data is revised, and a single quarter can be noise. It works as a regime read, a way of knowing whether the tide is coming in or going out, and it works best when it disagrees with the headline balance sheet number and forces you to ask which one is early.