Why Some Firms Gain More from Being Legible
The subject in this chapter is the uneven value of a firm's payment trail. The question is why the association between receiving electronic payments and credit access is larger for small, young, less documented, and less productive firms than for firms that already generate substantial conventional information. The answer supports a practical judgment: a payment trail is most valuable when it fills a specific information gap, not when it simply adds more data to an already well-documented business.
That judgment matters because a larger estimated benefit can be misread. It does not necessarily mean that a small firm is safer, that a young firm suddenly becomes more profitable, or that electronic payments cure the risks attached to weak operations. It may instead mean that the lender had more difficulty distinguishing the firm's existing performance from its risks. Payment records can change how visible a business is before they change anything about the business itself.
The gradient of missing conventional information
The World Bank study estimates the association between receiving electronic payments and the probability that a firm is fully credit constrained separately across several firm characteristics. The pattern is consistent: the estimated association is larger where the firm is less likely to have a long, detailed, independently documented record.
Size provides the clearest starting point. The estimated marginal effect associated with receiving electronic payments is −0.040 for small firms, −0.030 for medium firms, and −0.018 for large firms. These figures are percentage-point changes in the modeled probability of full credit constraint, not changes in a firm's revenue or a lender's return. The difference between −0.040 and −0.018 says that the relationship is more pronounced for small firms in the study's specification. It does not say that every small firm would experience a four-point improvement, or that size alone explains the outcome.
Why might scale matter? A larger firm often leaves more evidence through repeated transactions, established relationships, formal reporting, and a longer operating history. Scale is not proof of quality, but it can make a business easier to compare with itself over time and with other firms. A smaller firm may have fewer periods of data and fewer independent ways for a lender to verify its claims. In that setting, a record of incoming payments can supply information that would otherwise be expensive or impossible to obtain.
Age shows a similar gradient. The reported marginal effect is −0.044 for start-ups, −0.038 for firms aged 2–10 years, and −0.029 for firms older than 10 years. A young firm has had less time to establish a borrowing history, build audited records, or demonstrate how it behaves across different business conditions. An electronic record of customer payments cannot recreate a decade of history. It can make the limited history that does exist more observable.
This distinction is important for interpreting the word “young.” The evidence does not establish that digital payments remove the uncertainty inherent in a new business model. A start-up may have a clear stream of receipts and still face demand risk, execution risk, or a short runway. The payment trail can help answer one question—whether revenue is arriving through a recorded channel—while leaving other questions open. Better evidence is not the same as complete evidence.
The innovation and productivity results add another layer. The estimated effect is −0.037 for firms reporting no innovation and −0.016 for firms innovating in process, product, and research and development. For labor productivity, the effect is −0.046 in the lowest quartile and −0.019 in the highest. The paper interprets these differences as consistent with lower-capacity or less transparent firms having more to gain from an additional source of information.
Here, too, the safest interpretation concerns documentation rather than a technology-driven transformation of performance. A firm that has not introduced a new product or process may have fewer formal indicators through which a lender can understand its organization and prospects. A firm with low measured productivity may be harder to assess using standard performance comparisons. An incoming payment record can reveal sales activity that those other indicators do not capture. It does not prove that the firm is a hidden high performer, and it does not make low productivity irrelevant to repayment.
Audited accounts sharpen the same point. The estimated marginal effect is −0.038 for firms without audited financial statements and −0.026 for firms with audited statements. An audit does not guarantee that a firm will receive credit, but it can give a lender an established way to examine reported financial information. Where that channel is absent, payment records may carry more marginal information. Where it is present, electronic receipts may still help, but they add less relative to what the lender already knows.
These results form a gradient rather than a binary division. Firms are not simply transparent or opaque. A business may have audited accounts but little electronic revenue, or many electronic receipts but no long operating history. It may be large in headcount but concentrated in one unstable customer relationship. The useful question is not “Does this firm use digital payments?” It is “Which material part of this firm's activity remains difficult to verify, and does the payment trail speak to that part?”
What heterogeneity can and cannot tell us
Heterogeneous estimates are useful because they test whether a proposed mechanism predicts where effects should be larger. If electronic payments reduce credit constraints by improving information, the association should be stronger for firms with less reliable conventional information. The observed patterns across size, age, innovation, productivity, and audit status fit that prediction. They are more informative than a single average coefficient because they connect the proposed channel to differences among firms.
But a pattern that fits a mechanism is not the same as proof of that mechanism. The same firm characteristics may influence payment adoption and credit access through other routes. A larger firm may have better management, more bargaining power, or stronger demand, all of which can affect its financing. A start-up that adopts electronic payments may also be one of the better-organized start-ups. A firm with audited accounts may differ from an unaudited firm in ways that the survey controls do not fully capture. The heterogeneity therefore supports the information-asymmetry interpretation while leaving room for selection and omitted variables.
The World Bank paper tests the interpretation in a further way. It first estimates a firm's predicted probability of being fully credit constrained from standard firm, sector, and country characteristics without the electronic-payment variables. It then examines the residual component—the part not explained by those fundamentals. Receiving electronic payments is associated with lower residual constraint, with a reported effect of −8.097 for the receiving indicator and −10.216 for the percentage of payments received electronically.
This residual analysis asks whether payment use appears to carry information beyond the standard characteristics in the model. That is a useful test of incremental information, but the residual is not a pure measure of information asymmetry. It can also contain credit-supply limits and any other influence left unexplained by the first-stage model. A lower residual among firms using electronic payments is consistent with an added information channel; it does not establish that the channel is the only reason for the difference.
The paper's alternative outcomes reinforce the main pattern without eliminating this qualification. Receiving electronic payments is positively associated with being Unconstrained. In an ordered model, it is associated with a 2.4-percentage-point lower probability of being Fully Credit Constrained, a 1.4-point lower probability of being Partially Credit Constrained, and a 3.8-point higher probability of being Unconstrained. These results suggest that the relationship is not confined to one arbitrary boundary between constrained and unconstrained firms. They still come from the same observational survey framework.
The instrument results also show why the content and setting of a record matter. Bank transfers and other electronic methods, including mobile money and e-money, have significant negative associations with full constraint in the alternative specification, while cards do not. The difference is compatible with different instruments producing different informational traces. It may also reflect where each instrument is used: cards are more common in high-income economies where conventional credit information may already be stronger. The table identifies a contrast; it does not by itself identify the reason for that contrast.
A useful way to apply the evidence is to imagine two firms with similar sales. The first is a young local business with no audited statements and mostly cash-based customers. The second has operated for years, has audited accounts, and already receives most sales through channels visible to formal finance. If both begin receiving electronic payments, the first may gain more incremental visibility because the new record fills a larger gap. The second may still benefit operationally, but the lender may learn less that is genuinely new. The study's heterogeneity results make this comparison plausible; they do not predict a loan decision for either firm.
The distinction between visibility and risk should remain explicit. A payment trail can show that money arrived. It may not show whether the customer will return, whether the margin was sufficient, whether the firm owes more than it can repay, or what happens after a shock. It may also reflect a selection decision: firms with stronger operations may be more likely to accept electronic payments in the first place. Better screening can improve capital allocation only if the information is relevant and interpreted alongside the risks it does not measure.
The diagram captures the chapter's central limit. The payment trail can occupy the gap between activity and evidence, but it does not replace the firm's underlying operations. The line from visibility to a possible credit outcome is the relationship observed in the research, not a guaranteed sequence for an individual borrower.
Applying the opacity test
For a real firm, the evidence suggests four questions. First, what important activity is difficult for an outside lender to verify: sales, timing of receipts, customer concentration, costs, or something else? Second, does the firm receive electronic payments in a way that speaks directly to that gap, or does it merely have a small digital footprint? Third, is the record genuinely additional to audited accounts, established credit histories, and other conventional evidence? Fourth, which risks remain outside the record even if the payment stream is clear?
The answers matter more than the presence of a payment app. A record of incoming sales may be decision-relevant for a business whose revenue has been largely invisible, particularly when conventional credit information is weak. It may be less valuable for a firm whose financial statements, history, and institutional records already tell a lender much the same story. The same technology can therefore have a different marginal contribution across firms.
This is why the World Bank findings should not be translated into individualized financial advice. They do not show that a particular small firm will obtain a loan after adopting electronic payments, nor that a lender should disregard collateral, profitability, governance, or repayment capacity. They show a structured association: the firms for which conventional information is likely to be thinner display the larger relationship between receiving electronic payments and lower measured credit constraint.
The next question is whether this firm-level gradient changes when the surrounding credit system changes. A payment trail may add a great deal where registries and formal records are weak, yet add little where conventional information is already deep. The value of being legible depends not only on the firm being observed, but also on the institutions capable of using what is observed.