When a Contract Becomes Credit
When a firm wins a public procurement contract, the important financial question is not simply whether its sales increase. It is whether the award changes the firm's access to credit before the revenue has become cash or collateral. The evidence examined here helps the reader separate that immediate borrowing response from the later accumulation of assets, and judge how strongly the pattern supports the idea that government demand can function as a source of financing capacity.
The first change is not “more credit” but different credit
The New York Fed study follows Spanish firms that received at least one procurement project and compares their within-firm credit growth in periods with and without procurement. Its main empirical sample contains about 14,000 firms. The researchers estimate local projections: rather than asking only whether a winner is larger than a non-winner, they measure the cumulative change in an outcome at several horizons after a procurement event. Firm fixed effects remove time-invariant differences between companies, while sector-by-year effects absorb common movements affecting an industry in a particular year.
That design matters because procurement firms are not typical firms. They are larger and older on average, and they differ in their credit composition even before the contract. A simple comparison could therefore mistake pre-existing scale or financial strength for an effect of winning. The within-firm approach is not a complete solution to selection, but it asks a more useful question: when the same firm moves from a period without procurement to a period in which it obtains a project, how does its credit path change?
The first result is a rise in total credit of about 5.5 percentage points on impact. The figure is a change in the study's cumulative growth measure, not a claim that every firm's loan balance rises by exactly 5.5%. More revealing is where the change comes from. At the same horizon, non-collateralized credit increases by roughly 6 percentage points, while collateralized credit shows an effect of approximately zero. The immediate credit response is therefore not primarily a new loan secured by machinery or buildings that the firm has already acquired. It is borrowing supported by a different kind of expectation.
Collateralized credit does appear later. At the five-year horizon, its cumulative effect is about 4 percentage points. That timing produces a two-stage picture. First, a contract can make it possible to borrow against expected earnings even though the firm has not yet expanded its asset base. Later, if the project generates profits that remain inside the business, the firm can accumulate net worth or fixed capital. Those assets then support borrowing through the more familiar collateral channel.
The persistence of the total-credit effect also deserves a qualification. A long contract could mechanically keep a firm classified as a procurement supplier, and repeated contracts could make the first award appear more powerful than it is. The authors address these possibilities with separate evidence. In data on European public contracts, about 75% lasted one year or less, 89% lasted two years or less, and 93% lasted three years or less, though those duration figures may overstate typical length because very large contracts are over-represented. The credit response is also similar in a subsample of firms that received only one contract during the sample period. These checks do not eliminate every alternative explanation, but they make a purely mechanical “the same contract simply lasted forever” account less persuasive.
The result is best understood as a change in credit composition over time. A public award does not merely add one more source of sales to an income statement. It can change which part of a firm's future income a lender is willing to recognize, and when. That is the first reason a government customer may behave differently from an ordinary customer.
Why public earnings can unlock borrowing
A firm needs financing for two related reasons. The first is straightforward: delivering a new project may require labor, materials, or equipment before the buyer pays. The contract creates a credit-demand shock because the firm's desired operating scale has increased. The second is less obvious: the award may make the firm's future repayment capacity more credible. If a lender treats the public revenue as sufficiently reliable, the contract becomes a credit-supply shock as well. More borrowing is then possible not only because the firm wants to expand, but because the lender is willing to support that expansion.
The paper's model represents borrowing capacity as a combination of assets and earnings. In plain language:
borrowing capacity = asset collateral + pledgeable private earnings + pledgeable public earnings
The distinction between the two earnings terms is crucial. In the preferred specification, the authors infer that firms can pledge about 24% of annual earnings from private-sector sales and about 54% of annual earnings from procurement, alongside about 54% of the firm's capital stock. These are calibrated parameters, not legal rules stating that a government invoice can always be borrowed against at a 54% rate. They are the values that allow the model to match the observed relationship between leverage and earnings in the Spanish data.
Why might public earnings be more pledgeable? The paper offers a risk-based interpretation: a government contract may reduce uncertainty about near-future revenue because a public buyer is more likely to pay than the average private customer. That explanation is plausible, but the model does not directly measure payment delays, contract enforcement, interest rates, or invoice terms. The evidence supports an asymmetry in how earnings are reflected in borrowing capacity; it does not identify one institutional reason as settled fact.
The difference also explains why the first credit response can be unsecured. Suppose a firm has little tangible collateral but wins a project with credible future revenue. The lender cannot seize an asset that does not yet exist, but can still assess whether the expected earnings increase repayment capacity. The firm may obtain a working-capital loan without pledging new machinery. If the project succeeds, the resulting profits can finance equipment or strengthen the balance sheet. The next borrowing decision then has more conventional collateral behind it.
Pledgeability is therefore a forward-looking property, not the same thing as cash already received. A bank is asking whether a future stream of earnings can support repayment under realistic conditions. The answer depends on the contract's credibility, the firm's ability to deliver, the volatility of its other business, and the amount of debt already outstanding. A public award can improve that calculation without eliminating execution risk. If the firm misprices the job, faces a cost overrun, or cannot find enough workers, expected revenue may not become realized revenue.
The channel also has a multiplier built into it. Borrowing against expected public earnings can finance the labor and capital needed to deliver the first project. Profits from that project can then increase net worth, and net worth can support additional borrowing against equipment or other assets. In this sense, the contract's financial value is not limited to the margin on one sale. It can change the firm's feasible path in later periods. But the same multiplier works in reverse when access is lost: a firm may have to postpone investment, operate at a smaller scale, or rely more heavily on private customers whose earnings lenders treat as less pledgeable.
The loan-application result is useful because it looks at the bank's decision conditional on a firm seeking credit. The reported increase of approximately 2 percentage points in the probability of receiving a loan in the quarter of winning a project is not merely a larger loan balance among firms that were already approved. It is evidence that procurement may affect the supply of credit at the margin. Even so, it remains an estimate from the Spanish setting, and it cannot reveal whether the lender relied on the government buyer, the firm's improved prospects, or information correlated with winning.
This sequence should not be read as a promise that every award relaxes a binding constraint. In the calibrated application, only 0.4% of procurement awards lead immediately to a relaxation of the awardee's financial constraint. Many firms can already finance the project, while others may remain constrained even after winning. The broader effect is dynamic: procurement changes the expected return to saving, the firm's future net worth, and its chance of borrowing later.
The model makes that dynamic choice explicit. A firm can use current resources to accumulate net worth, or it can spend resources preparing applications that raise the probability of winning a future government project. The value of either choice depends on productivity, wealth, experience, and the way public earnings can be pledged. A procurement contract therefore affects not only current borrowing but also the firm's incentives to save and compete for the next opportunity.
There is evidence supporting the credit-supply interpretation, but it is important to distinguish evidence from interpretation. The immediate increase being concentrated in non-collateralized credit is difficult to explain solely as asset accumulation. The initial crowding out of private-sector sales is stronger among firms that appear more financially constrained, which fits a story in which credit limits production capacity. And the probability of receiving a bank loan conditional on having applied rises by approximately 2 percentage points in the quarter when a firm wins a procurement project. Together, these patterns point toward lenders responding to the award, not merely firms demanding more loans.
Still, the empirical design does not provide one clean, quasi-experimental assignment of contracts. Firms choose whether to participate, governments choose among applicants, and unobserved characteristics—such as managerial quality or direct relationships with public buyers—could affect both winning and credit access. The authors address selection and reverse causality with quarterly data, firm-year fixed effects, close-competition comparisons, credit ratings, matching, and other robustness exercises. Those checks strengthen the interpretation, but they do not turn the Spanish evidence into a universal causal estimate for every public contract.
The model adds another caution. It reproduces the persistence and direction of the credit response, but its predicted credit increase is much larger than the data's: about 24.5 log points in the model compared with 5.5 in the empirical estimate. The authors suggest that real firms may finance investment by reducing cash or increasing other liabilities, channels the model does not represent fully. A model that gets the sequence right but overshoots the magnitude is still useful for identifying a mechanism, but it should not be mistaken for a precise balance-sheet forecast.
The most defensible conclusion is therefore specific. In the Spanish evidence, obtaining procurement is associated with a persistent credit increase that begins with non-collateralized borrowing and later includes collateralized credit. The pattern is consistent with public earnings being more pledgeable than private earnings, and with procurement easing a financial constraint over time. It does not mean that a government contract is a guaranteed loan, that every supplier is financially rescued, or that the credit channel is independent of the firm's need to finance a larger operation.
Credit is the first bridge between a government order and firm growth. It is not the whole crossing. A firm can receive more borrowing capacity and still have too little labor, capital, or collateral to serve every customer at once. The next question is what happens inside the operation when public demand arrives faster than capacity can expand.