The Capacity Squeeze Before Growth
Public procurement is the subject here, but the focus is inside the winning firm: what happens when a government order arrives before the firm has enough capacity to serve every customer? The answer matters for judging the financing value of a contract. A procurement award can raise total sales and improve future borrowing while still reducing private-sector sales in the short run, so the reader needs to follow the sequence rather than treat “growth” as one immediate outcome.
Total sales can rise while private sales fall
The Spanish evidence shows a pattern that is easy to miss in a single revenue statistic. After a firm obtains a procurement contract, its total sales increase by around 9 percentage points in the study's cumulative growth measure. That sounds like a straightforward expansion. But sales to private-sector customers fall immediately after the procurement shock. Only over subsequent years does that private business recover and move above its earlier level. The same contract therefore produces two different readings depending on the horizon: at impact, it reallocates the firm's activity; over the medium run, it is associated with a larger business.
The distinction is not just accounting. A firm cannot necessarily add a complete second operation at the moment it wins a public project. It may need to hire, buy materials, reserve production time, or install equipment. Those resources are limited even when the order is profitable. If the new contract has to be delivered on a schedule, the firm may direct workers and equipment away from private customers. A fall in private sales can then indicate that the firm is meeting a new obligation under a binding capacity constraint, not that public demand has destroyed the private business permanently.
The data are consistent with that interpretation because the initial private-sales loss varies with the firm's apparent financial position. The study uses fixed assets, leverage, and age as conventional proxies for financial constraint. The crowding-out effect has a negative monotonic relationship with assets and age: firms with less of each experience a larger immediate reduction in private sales. The leverage relationship is less smooth, but the strongest crowding out appears among firms above the 95th percentile of leverage. These are not perfect measures of an invisible borrowing limit. They are observable indicators that make the pattern more informative than a pooled average.
A firm with little fixed capital may have high-return opportunities in both markets but no way to finance all of them at once. A heavily leveraged firm may be able to win a contract and still lack the spare collateral needed to expand production for private customers. An older firm may have more established relationships and assets, so the same procurement award creates less disruption. The point is not that age or size causes the response. It is that the effect is larger where the evidence suggests the constraint is tighter, which is what a financing mechanism predicts.
The sales path also helps distinguish a temporary allocation problem from a permanent loss of private demand. If government work simply replaced private customers forever, private sales would not be expected to rebound above their pre-contract level. The observed later increase is consistent with the firm using procurement profits to build capacity. That interpretation remains an interpretation: the data do not observe every internal decision about labor, equipment, or retained earnings. But the timing is difficult to summarize as “public contracts crowd out private business” without adding the second half of the path.
Persistence raises a related question. Perhaps the apparent medium-run effect is only a long contract that keeps generating the same activity, or a sequence of repeat awards rather than a lasting change in the firm. The authors use separate evidence to narrow that concern. In European public-contract data, about 75% of contracts last one year or less, 89% last two years or less, and 93% last three years or less, although those figures may be an upper bound for typical contracts because very large contracts are over-represented. The procurement-credit relationship is also similar among firms that receive only one contract during the sample period. These checks do not prove that every firm's later growth comes from the first award, but they make a purely mechanical duration explanation less sufficient.
The recovery depends on how finance and capacity interact
The model explains the two-stage sales pattern by making the firm's resources explicit. Firms sell to private customers and, when selected, to the government. They use capital and labor to produce for both markets. A financially unconstrained firm can increase capital and labor for the public order while leaving its private operation broadly unchanged. The procurement shock adds a new source of demand, and the firm's capacity is flexible enough to respond in both directions.
A constrained firm faces a different choice. Its borrowing limit depends on some combination of net worth, tangible capital, private-sector earnings, and public-sector earnings. When the firm wins a procurement project, it must allocate the scarce financing capacity between two demands. Even if public earnings are more pledgeable than private earnings, that advantage may cover only part of the additional credit needed. The firm can shift capital and labor toward the government project and away from private production. The immediate result is crowding out.
This is why a contract can be valuable and disruptive at the same time. The public project may carry a higher financing value than a private sale because its expected earnings support more borrowing. Yet the firm still has to perform the project with real workers and equipment. The contract's financial value arrives through a chain: expected earnings support initial borrowing; production generates revenue and profit; retained profit increases net worth; net worth and new assets relax later borrowing constraints. The first link can improve the firm's future, while the middle link creates a temporary operating squeeze.
Imagine a contractor that already has a profitable private order but cannot borrow enough to add a second production line. A government award may be attractive because its revenue is more pledgeable, but the award does not create a second line instantly. The firm may use the same supervisors, machines, or working-capital limit for both jobs. It can fulfill the public order by postponing a private delivery, accepting fewer private orders, or moving labor toward the government work. The measured fall in private sales is then the visible cost of reallocating capacity while the firm waits for the financing and earnings effects to arrive.
That timing matters for both the firm and the buyer. A firm that is paid late or earns less than expected can remain constrained even after winning. A firm that delivers successfully but distributes all of the profit rather than retaining it may accumulate less net worth for the next period. The model abstracts from many contract-level details, so it does not tell us which of these outcomes occurs for a particular supplier. It does establish why the route from award to growth has intermediate states, and why evaluating only the award or only the eventual sales level can miss the risk borne during delivery.
The contrast with an unconstrained firm is equally important. If the firm already has sufficient collateral and cash, the public project can add production for the government without forcing a comparable reduction in private output. The contract still raises demand and may increase profit, but it does not have to solve a binding financing problem. In the constrained case, the award changes the allocation of a scarce resource; in the unconstrained case, it is closer to an additional market. This is why the average sales response cannot by itself reveal the mechanism.
A temporary decline in private sales is also not automatically evidence that public procurement is harmful. The public project may have a higher social value, or the firm may be building productive capacity that benefits both markets later. Nor is later private-sector growth proof that the initial displacement was benign. The buyer may have obtained its output by drawing labor and capital away from other customers, and the model warns that collateral value can pull resources toward public production even when private-sector returns are high. The relevant judgment depends on the full path and on which output the government is trying to secure.
The model also clarifies why the same procurement award can have different effects at different wealth levels. In the empirically relevant calibration, firms with substantial existing wealth are more likely to experience a tightening of their financial constraint after taking on the public project: they must split collateral between public and private operations, and the higher pledgeability of public earnings does not fully offset the extra demand. Firms with zero or nearly zero wealth can be an exception. For them, the financing value of public earnings may dominate the reallocation pressure, allowing procurement to produce a positive private-sector spillover.
That result is not a claim that the poorest firm always benefits most, nor that a public contract should be assigned by wealth alone. A firm with no wealth may be unable to finance the application, hire the necessary workers, or absorb a delay in payment. The model's point is narrower and more useful: the effect of procurement depends on which constraint binds. If asset collateral is the main limit, an additional demand can intensify the squeeze. If pledgeable public earnings are the missing link, the same demand can relax the constraint. The two forces can coexist within the same supplier population.
The dynamic choice extends beyond one award. A firm can spend current resources on saving, which builds net worth and relaxes future asset-based borrowing limits, or on preparing applications that increase the chance of winning a future procurement project. A current supplier often saves more because it earns higher profits and has a higher probability of obtaining procurement again. Experience can therefore make access persistent, while the ability to deliver larger projects makes the value of access depend on productivity and wealth. The procurement system is not only allocating today's order; it is influencing tomorrow's preparation and capacity decisions.
The calibrated model reproduces the direction and broad timing of the sales response. It predicts private-sector crowding out on impact and crowding in from the following horizon, as profits raise net worth and reduce the shadow cost of the borrowing constraint. But it does not reproduce every magnitude. The model predicts about 8 log points of private-sales crowding out at impact, compared with about 13 log points in the data, and its crowding-in phase is slower. Four years after the shock, the accumulated increase in private sales is about 2 log points in both the model and the data. This is a useful pattern of partial fit: the model captures the transition mechanism while leaving room for real firms to use balance-sheet channels the model omits.
The model's treatment of public earnings also carries a risk that matters for interpretation. If government revenue is more pledgeable, a constrained firm has an incentive to direct scarce capital toward public production. That may be financially rational for the firm but not socially optimal if the allocation is driven by collateral value rather than by the highest productive return. Within the firm, capital can move toward the sector whose earnings support more borrowing. Across firms, the government may select suppliers partly for their financing capacity. The contract helps the firm grow out of a constraint, but the path to growth can involve temporary private losses and misallocation.
The evidence therefore supports a precise conclusion about the capacity squeeze. Procurement is associated with higher total sales, an immediate reduction in private-sector sales, and later private-sector expansion. The initial loss is stronger among firms that appear more financially constrained, which supports the interpretation that scarce collateral and capacity mediate the response. The model explains how retained procurement profits can reverse the initial displacement, but its fit is directional and incomplete, not a guarantee about any individual supplier.
This time path changes how access should be evaluated. A procurement policy judged only by the first quarter may appear to harm a winning firm's private customers. A policy judged only by later sales may overlook the risk that the firm had to displace valuable business to deliver the public order. Both observations belong to the same lifecycle. The remaining question is upstream: which firms can reach this contract-to-finance path in the first place, and how does access policy distribute its benefits and costs?