What to Watch When Liquidity Policy Changes
The central question is now clear in its proper form: why can less required reserves be followed by later settlement of large payments in Colombia’s CUD? The answer is not that a reserve reduction mechanically slows every transfer. It is that the reduction changes the feasible set of intraday choices. This final synthesis connects the policy change to liquidity allocation, participant position, payment recycling, infrastructure, and the distribution of value across the day. It also turns the CUD evidence into a disciplined way to judge whether a change has improved settlement resilience or merely moved pressure into the upper tail.
The answer is about allocation, not a shortage of money
A lower reserve requirement first changes a regulatory constraint. It can release funds that credit institutions previously had to hold against deposits. But “released” does not mean “sent through CUD immediately.” The institution decides whether to leave the funds in its settlement account, use them for a payment, obtain or repay repo funding, place them in another short-term use, or preserve them for an obligation whose timing is not yet certain. The policy changes the value of those alternatives.
That is the first condition behind the CUD result: released liquidity has competing uses. If the opportunity cost of leaving funds at the central bank rises relative to the return or strategic value of deploying them elsewhere, an institution may redeploy part of the additional balance before sending some outgoing payments. The institution can be better funded overall while its payment schedule becomes later at particular percentiles. A balance-sheet improvement and an earlier settlement timetable are related outcomes, not identical ones.
The second condition is that obligations are not interchangeable. A bank can settle smaller or more immediate transfers early and preserve liquidity for a high-value payment tied to a securities transaction, a money-market operation, or an expected funding event. That produces the pattern observed in the CUD data: lower percentiles can move earlier while upper percentiles move later. The finding is not a contradiction; it is what a redistribution of payment value looks like when the day is treated as a sequence of obligations rather than one average transaction.
The third condition is interdependence. A payment sent by one participant becomes usable liquidity for another. A participant that expects an incoming payment may rationally wait rather than acquire expensive replacement funds. If the expected payment is delayed, the recipient’s own outgoing payment can be delayed, passing the timing pressure through the network. A locally sensible decision can therefore increase the amount of value left for the late window. Settlement liquidity is partly a stock of balances, but it is also the timing and reliability of flows between those balances.
Repo conditions make the intraday sequence visible. The CUD study associates a higher ratio of daily repos to available reserves with more pressure in the lower and middle percentiles, followed by easier settlement later as central-bank operations meet demand for repo funding. A high ratio is not simply evidence of abundant liquidity; it can indicate that institutions are relying more heavily on a funding source relative to the reserves they already hold. Early in the day, the funding may not yet be available where it is needed. Later, the same source can relieve the constraint.
The interest-rate differential tells a related story. When the interbank rate rises relative to the policy rate, the opportunity cost of holding reserves and the pressure in funding markets can increase. In the paper’s estimates, that condition is associated with delays in earlier percentiles, while it matters less for later percentiles after participants have largely met their liquidity needs. The effect is therefore not a universal “high rates mean late payments” rule. It is evidence that the price of liquidity interacts with the stage of the operating day.
The late-day fee does not overturn this mechanism. CUD’s second pricing window is more expensive for a large transfer than the first window, yet the study finds greater concentration of high-value settlement in that later window in the latest period. The underlying market transaction, the need to complete a funding adjustment, and dependence on incoming liquidity can outweigh the local fee. A participant may accept a higher settlement charge because moving earlier would require more expensive or less reliable liquidity.