Full Text
ECONOMIC GROWTH AND VITALITY THROUGH AUTOMATIC LOAN GUARANTEES AND EXPEDITED LENDING FOR SMALL ENTERPRISES
Banking exists to move savings into new production through entrepreneurs, yet decades of bank mergers have replaced small local banks, which lend on personal knowledge, with large faceless institutions, which lend on standardized data alone, so capital flows overwhelmingly to large corporations while small enterprises remain rationed. Slowness compounds the rationing: half of small enterprises hold under fifteen days of cash, ordinary approvals take weeks, and many viable companies never apply. Our paper asks how governments can restore fast credit access, and thereby economic growth and vitality, for small enterprises; our paper answers with two complementary instruments, grounded in comparative analysis of guarantee schemes and delivery models. The first instrument is an automatic loan guarantee: any company which has operated for five years may borrow up to the total taxes which the company paid during the previous five years, with the state guaranteeing a partial share, so worst-case public exposure stays bounded by revenue already collected, while the verified payment record screens out unstable borrowers. Automatic eligibility matters: enrollment research shows participation multiplying once application burdens disappear, with the largest gains among discouraged applicants. The second instrument is a required minimum share of bank lending directed to small enterprises, following Community Reinvestment Act and priority sector lending precedents, so banks again serve as infrastructure for distributing capital. Anticipated findings include faster approvals, renewed business entry, rising machinery and technology investment, employment growth, and bounded fiscal cost, matching field-experimental evidence of returns far above interest rates. Open parameters include eligibility age, entitlement ratio, coverage percentage, approval-time targets, and minimum lending share; evaluation compares eligible companies with just-below-threshold peers on lending, defaults, entry, employment, and productivity. Delivery requires no new agency: the tax authority issues digital entitlement certificates, existing banks decide within days, disburse, and retain part of every risk, and, consistent with Fiscal Secularity, eligibility follows a public formula, banks never see tax records, and no official decides who deserves capital.
INTRODUCTION
The purpose of banking is old and simple: banks collect the savings of many people and lend the collected savings to people who can turn borrowed money into new production. The academic foundations agree on the point at every level. Banks and similar intermediaries exist because financial markets fundamentally channel savings toward real investment, with banks acting as delegated monitors on behalf of savers (Diamond 1984). In the classic account of economic development, entrepreneurs need credit to hire resources and put innovations into practice; without access to capital even the most visionary entrepreneur cannot build anything, which gives banks a central role in development, not just as lenders but as gatekeepers who decide which new ventures get funded and which do not (Schumpeter 1934). The empirical record confirms the stakes: across 80 countries over the 1960 to 1989 period, measures of financial development are strongly associated with real per capita growth, physical capital accumulation, and efficiency improvements, and the predetermined component of financial development robustly predicts future growth (King and Levine 1993).
Our paper begins from a diagnosis: the channel connecting savings to small productive enterprises has narrowed, and the narrowing has two dimensions. The first dimension is allocation. Decades of consolidation have replaced small, locally rooted lenders with large hierarchical institutions, and the evidence reviewed below shows that large institutions structurally lend to large firms. The second dimension is time. Lending to small enterprises has become slow and burdensome exactly where speed decides survival: the median small business holds enough cash to withstand roughly 27 days without inflows (JPMorgan Chase Institute 2016), and 50 percent of small businesses operate with fewer than 15 cash buffer days (JPMorgan Chase Institute metropolitan liquidity research), while ordinary loan approval commonly takes weeks and full funding often takes one to three months (Small Business Administration program documentation and industry reporting). We call the mismatch the two-clocks problem: for half of small enterprises, the approval clock runs slower than the survival clock.
Our paper asks one question: how can governments restore fast and reliable credit access for small enterprises, and thereby restore economic growth and vitality? Our paper answers with a policy design built on two principles, automatic eligibility and expedited delivery, embodied in two instruments. The first instrument is an automatic loan guarantee in which eligibility and the borrowing ceiling are computed, without any application or official judgment, from a company's verified tax history. The second instrument is a required minimum share of bank lending directed to small enterprises, with documented regulatory precedents. Our paper is a position paper in the sense used by the Advanced Design Conference: our paper states anticipated findings, names every open parameter, and defines the evaluation procedure by which the design can be tested.
Our paper claims four contributions. First, the entitlement formula: to our knowledge, no existing scheme converts cumulative taxes paid into an automatic borrowing entitlement, although every component of the mechanism is independently validated. Second, a transfer of the automatic-enrollment insight from savings policy to credit policy, with a testable prediction about participation. Third, the two-clocks framing, which converts lending speed from a convenience feature into a solvency requirement. Fourth, a fiscal design in which worst-case public exposure is bounded, by construction, by revenue already collected from the same company.
The remainder of our paper proceeds as follows. The background section reviews the evidence on consolidation, rationing, and declining dynamism. The two design sections present the automatic loan guarantee and expedited delivery through existing banks together with the complementary lending-share instrument. The following sections state anticipated findings, open parameters, and the evaluation design; answer the six strongest objections; state the governance principle, Fiscal Secularity, on which the design rests; state limitations; and conclude.
BACKGROUND AND PRIOR ART
The Loss of Soft Information and the Personal Lender: Small-enterprise lending depends on knowledge which does not fit in a spreadsheet; the owner's character, work habits, and local standing. The theoretical foundation is the distinction between soft and hard information: decentralized organizations with small, single-manager units perform best when information about projects is soft and cannot be credibly transmitted, while large hierarchies perform better when information can be costlessly hardened and passed along inside the firm (Stein 2002). The prediction fits the facts: when large banks acquire small banks, a pronounced decline in lending to small businesses has been documented, exactly what the theory predicts where small-business lending relies heavily on soft information (Stein 2002). Region-level evidence agrees: small firms are less likely to borrow from banks after mergers make those banks more hierarchical (Sapienza 2002), and small businesses located in regions dominated by large banks were more likely to face credit constraints than firms located close to small decentralized banks (Berger, Miller, Petersen, Rajan, and Stein 2005). Relationships are the delivery technology for soft information: the primary benefit of building close ties with an institutional creditor is that the availability of financing increases, with smaller effects on price; relationships are valuable and operate through quantities rather than prices (Petersen and Rajan 1994).
Price cannot clear the market because rationing is an equilibrium outcome: even in equilibrium, credit rationing exists in loan markets and; among applicants who appear identical, some receive credit and others do not, even though the rejected applicants would pay higher interest rates (Stiglitz and Weiss 1981). Consequently, access, not price, is the binding constraint, and any effective remedy must change access.
What Consolidation Did: The structural shift is large. From the 1930s until the 1980s, the United States banking system consisted of more than 18,000 primarily small depository institutions; the number of commercial and savings banks then declined by nearly 30 percent between the end of 1985 and the end of 1995 alone (Peek and Rosengren 1998), and the consolidation wave continued afterward. Institutions which merge lose local identity and refocus their franchises toward capital-market services for large corporate clientele (Udell 1998). The allocation result is measurable: a large-sample study of 178,000 observations from the United States banking sector finds that large banks tend to lend to large firms and small banks tend to lend to small firms, so that in highly concentrated systems consisting only of a few large banks, sufficient small-firm funding cannot be expected. The asymmetry after the 2008 crisis is stark: loans originated in amounts under 1 million dollars remained more than 17 percent below the pre-crisis high in 2015, while loans above 1 million dollars surpassed the pre-crisis level by more than 35 percent (Small Business Administration Office of Advocacy 2021). Deposits and lending decoupled: in one large metropolitan market, deposits increased by 100 percent while small business lending decreased by nearly 40 percent (National Community Reinvestment Coalition). The decline continues: between 2019 and 2023, bank small-business lending by real dollar amount declined by 18 percent nationally despite increasing demand (Federal Reserve Bank of St. Louis 2025). Even conditional on obtaining credit, small firms receive systematically harsher terms: 40 percent of credit lines to small firms matured or were callable within one quarter, against 15 percent short-maturity lines for large firms, and fewer than 5 percent of credit lines to small firms were unsecured, compared with 70 percent for the largest firms (National Bureau of Economic Research Digest summary).
The mechanism is physical as well as organizational. Branch closings lead to a persistent decline in local small business lending; annual originations fall by roughly 453 thousand dollars off a 4.7 million dollar baseline and remain depressed for up to six years, with effects dissipating within six miles; and the decline translates into a 2 percentage point reduction in local employment growth, driven primarily by tighter constraints on the size of entering firms (Nguyen 2019). Decisively, the lending contraction persists even after a new branch opens, suggesting the loss is the information destroyed when a local relationship is severed, not reduced competition (Nguyen 2019). European evidence reaches the same conclusion: branch closures significantly reduce employment, sales and working capital of local firms, firm exit probability rises by around 0.6 percentage points over three years if 10 percent of local branches close, and local firms cannot offset the decline with funding from other sources (SUERF policy analysis). In fairness, parts of the merger literature found offsetting responses i.e. static reductions in small business lending after mergers were mostly offset by the reactions of other banks in the same market (Berger, Saunders, Scalise, and Udell 1998),but the offset presupposes surviving small banks able to absorb abandoned borrowers, and the offsetting population is precisely what consolidation has been eliminating. Our design therefore does not attempt to resurrect the past cure; our design engineers a substitute channel which works with the banking structure that actually exists.
What Is at Stake for Growth and Vitality: Small and medium enterprises are the majority of the productive economy: they represent around 90 percent of all businesses and account for more than half of global employment (World Bank), and in the United States, small businesses account for 99 percent of establishments, 48 percent of workers, over 41 percent of net job creation, and 45 percent of gross domestic product (JPMorgan Chase Institute 2016). The credit gap is measured in trillions: more than 200 million formal and informal micro, small and medium enterprises are estimated to be unserved or underserved (International Finance Corporation).
The job-creation evidence carries an important refinement: once firm age is controlled for, there is no systematic relationship between firm size and growth; startups and young businesses drive job creation, and young firms exhibit an up-or-out dynamic (Haltiwanger, Jarmin, and Miranda 2013). Vitality, in our paper, therefore means business dynamism: entry, young-firm growth, and productivity-enhancing reallocation. Dynamism is in documented decline: the economy has become less dynamic, with declining rates of gross job and worker flows and declining rates of entrepreneurship and young-firm activity, pervasive across industries, regions, and firm size classes (Decker, Haltiwanger, Jarmin, and Miranda 2016), and the share of employment at young firms has dropped from 20 percent to 10 percent over several decades (Decker, Haltiwanger, Jarmin, and Miranda 2016). The connection to growth is direct: a large fraction of aggregate productivity growth is accounted for by the movement of resources from less-productive to more-productive businesses, so reduced dynamism implies a slowdown in productivity-enhancing reallocation (Decker, Haltiwanger, Jarmin, and Miranda 2016). Credit conditions are among the suspected causes of diminished dynamism (Davis and Haltiwanger 2015), and the machinery of our design targets exactly the entry-and-young-firm margin.
Finally, the payoff to relaxing the constraint is unusually high at the small end. It is consistently the smallest firms that are most constrained, and it is again the small firms that benefit most when obstacles weaken (Beck, Demirgüç-Kunt, and Maksimovic 2005); the growth of most small firms is constrained by internal finance (Carpenter and Petersen 2002); and randomized evidence shows why the constraint is costly: providing cash and equipment grants to small firms and measuring the profit increase from the exogenous capital shock, the average real return to capital was around five percent per month, substantially higher than market interest rates (De Mel, McKenzie, and Woodruff 2008). When small firms do invest, the purchases match the growth mechanism: technology and information technology top the small-firm investment list at 60 percent, followed by plant, machinery and vehicles and staff training at 52 percent each (Bank of England 2024).
The Established Instrument and the Gap Our Design Fills: Credit guarantee schemes are the standard remedy, with a global evidence base. The evaluation literature finds that credit guarantees provide financial additionality, increasing the availability of credit and reducing its cost (OECD and European Commission 2017); evaluations of the European Union guarantee facility found significantly positive effects on employment and turnover, with micro and young enterprises benefiting most (Asdrubali and Signore 2015); firm-level evidence from China finds guarantees raising the probability of obtaining loans, loan amounts, research spending, fixed asset investment, and total factor productivity; and loan guarantees are documented as a relatively low-cost way to increase lending to small enterprises (Bachas et al. 2021). Design principles are codified: guarantees should be partial and provide capital relief to lenders (World Bank and FIRST Initiative 2015), because full coverage invites lowered lending standards (Yale Program on Financial Stability 2020) and the coverage ratio is the instrument which limits moral hazard for borrowers and lenders alike; United States regulation already requires that the lender retain an economic interest in, and the ultimate risk of loss on, the unguaranteed portion (Code of Federal Regulations, Title 13, Part 120).
Two documented weaknesses motivate our redesign. First, discretion: applications assessed case by case invite selection bias, with intermediaries off-loading their riskier portfolio tranches onto the public guarantee (Asdrubali and Signore 2018), and political incentives to supply guarantees generously while concealing long-term fiscal costs behind uncertain expected losses (Honohan 2010). Second, burden and slowness, which suppress participation among exactly the firms the schemes target: around half of small enterprises are permanent non-borrowers (Bank of England 2024), discouraged borrowers experience outcomes as adverse as those formally denied (Bouslama 2025), and for every three discouraged firms, one would have been approved (Cole and Sokolyk 2016, as reported in Bank for International Settlements Working Paper No. 1041).
Why Automatic and Expedited Loans Matter: The behavioral evidence on automatic eligibility is unambiguous in a neighboring domain. When automatic enrollment replaced voluntary sign-up in a large employer's retirement plan, participation rose from 37 percent to 86 percent, and the switch dramatically changed behavior although no economic feature of the plan changed (Madrian and Shea 2001); industry-scale data show participation tripling from 28 percent to 91 percent (Vanguard 2020); and, critically for our setting, automatic designs equalize participation across groups, with the largest effects among those least likely to participate under affirmative-election regimes (Madrian and Shea 2001). The administrative-burden literature generalizes the lesson to public programs: non-take-up is commonly misread as intentional choice when research shows the assumption does not reflect reality (United States Office of Management and Budget 2022); reducing learning and compliance costs substantially increases access (Health Affairs 2020); the scale of burden-driven loss is large, with an estimated 140 billion dollars in authorized benefits left unclaimed each year in the United States (Institute for Responsive Government); burdens fall hardest on the least resourced (Deshpande and Li 2019); and delivery through the tax system is the documented low-burden channel, as the relatively high take-up and durability of tax-system-delivered benefits attests (Russell Sage Foundation Journal 2023). Our design transfers the insight to credit: existing guarantee schemes are voluntary enrollment; our instrument is automatic enrollment for eligibility. The prediction; participation multiplying, concentrated among discouraged firms, is stated below as an anticipated finding.
Speed has equally hard evidence behind feasibility. Technology-based lenders process applications about 20 percent faster than traditional lenders (Fuster, Plosser, Schnabl, and Vickery 2019); at the frontier, one technology-based lender serving over 20 million small businesses; more than three-quarters of whom had never received a bank business loan, assesses creditworthiness in minutes with zero human interaction using transaction histories (Bank for International Settlements Working Paper No. 1041), and machine-learning credit scoring better predicts default than traditional indicators (Gambacorta, Huang, Qiu, and Wang 2019). Public programs have demonstrated the same at national scale: one scheme launched within 11 days delivered most of 47 billion pounds within two months, aiming to pay borrowers within 24 to 48 hours of application (House of Commons Committee of Public Accounts 2021); a parallel program offered effectively instant support to firms in good standing, because once the stated conditions were met, no further risk assessment was needed (Review of Political Economy analysis of the KfW pandemic response); and in ordinary times, delegated-authority lenders approve guaranteed loans in five to ten business days without agency review, with an express channel carrying a 36-hour turnaround (Small Business Administration program documentation and industry reporting). Against the survival arithmetic above, expedited delivery is not a preference; expedited delivery is the condition under which credit arrives while the borrower is still alive.
INSTRUMENT ONE: THE AUTOMATIC LOAN GUARANTEE
The Lending Rule: Any company which has operated and paid taxes for the trailing five years holds an automatic entitlement to borrow, from any participating bank, up to an amount equal to the total taxes the company paid during those five years, multiplied by a tunable entitlement ratio. The state guarantees a fixed partial share of each entitlement loan. There is no application for eligibility, no business-plan review by any official, and no committee. The tax record, data the government already holds, is simultaneously the eligibility test, the credit ceiling, and the primary risk signal. Eligibility is continuous and rolling: each year, the window advances, and the ceiling recomputes.
Why Five Years and Why Lending on Taxes Paid: Three properties converge on our formula. First, survival screening. Roughly 22 percent of new businesses fail within the first year and nearly half close within five years; by the sixth year only around half survive (United States Bureau of Labor Statistics, Business Employment Dynamics). A five-year threshold therefore places the entire early mortality zone outside the guarantee, and conditional survival rates rise with age, so the eligible population carries structurally lower hazard than the applicant pool a bank ordinarily faces. Second, verified truth. Taxes paid cannot be inflated by optimism: overstating them requires actually paying real money to the state, making the signal costly to fake in the only direction that matters, while understating them is already unlawful. The predictive power of exactly such data is commercially demonstrated: underwriting practice reports that a history of verified tax payments over a long period is a stronger indicator of financial health than personal credit data for thin-file small businesses, and that integrating comprehensive tax data into underwriting may reduce default rates by upwards of 50 percent while increasing approved volume by more than 20 percent (commercial underwriting practice reports). Current practice already uses the record for verification; tax-return verification is a required underwriting step in the largest national guarantee program (Small Business Administration Standard Operating Procedure 50 10), but only as a check on an application. Our design promotes the record from verification document to the entitlement itself; the promotion is the novelty. Third, production-proportionality. The ceiling scales with demonstrated real activity: a growing firm earns more borrowing power and a shrinking firm earns less, automatically. The objections section shows why the property matters.
If richer signals are ever needed, the feasibility bar is low: even simple digital-footprint variables match the information content of credit bureau scores, and their discriminatory power extends to previously unscorable customers (Berg, Burg, Gombović, and Puri 2020). Five years of state-verified payments is a stronger signal than most inputs modern underwriting already trusts; the design is technologically conservative.
Bounded Fiscal Exposure: Because the ceiling equals revenue already collected from the same company, no company can become a net lifetime drain on the treasury: the worst case per company is the return, in guarantee payouts, of a fraction of money the same company previously contributed, and only the guaranteed share of the defaulted fraction is ever paid. Aggregate exposure is transparent by construction; the sum of trailing five-year tax receipts of eligible companies, times the entitlement ratio, times the coverage share, i.e. a number publishable in real time. The design answers the documented fiscal-opacity critique of guarantee schemes structurally rather than by promise. Guarantees remain contingent liabilities with high leverage: schemes can responsibly guarantee loan volumes several times larger than their capital funds (Asian Development Bank 2021).
Moral Hazard Controls: Coverage is partial, never total, per the codified principles above; the lender's retained share preserves the incentive to underwrite and to collect. The borrower's incentive is preserved because default consumes an earned entitlement and carries ordinary insolvency consequences. Fraudulent misrepresentation, including structured fragmentation of one enterprise into several artificial small ones, is classified as fraud and detected algorithmically through shared beneficial ownership and circular transaction flows, which is investigative work on data rather than a new inspection bureaucracy. The governing sentence of the wider research program applies: a company's money is inviolable; lying to the system about who you are is not.
The Entitlement Certificate and Data Minimization: The tax authority issues a digitally signed certificate stating exactly three facts: company identity, current entitlement ceiling, validity period. The bank sees the certificate, never the underlying tax records; the state learns that a loan was requested, never the business plan. Sealed data, transparent design. The certificate also attacks discouragement directly, because a company no longer initiates a plea into the unknown: the company walks into any participating bank already knowing the ceiling printed on a government document. Given the evidence that non-application reflects burden rather than choice, visibility of the entitlement is itself an intervention.
INSTRUMENT TWO AND DELIVERY: EXPEDITED LENDING THROUGH EXISTING BANKS
Division of Labor: The government does three things only: define the standard product, compute entitlements from records already in state hands, and operate the guarantee fund with a central reporting portal, for which national precedent exists; a single scheme portal used by all accredited lenders to report guaranteed facilities and claims, allowing the state to track exposure continuously (British Business Bank scheme documentation). Existing banks do everything else: receive the application, know the customer, price, decide, disburse, service, and collect. The model is proven at national scale: the leading promotional bank does not interact with customers directly but supports banks, with favourable funding and by assuming risks, under the on-lending principle, using the branch networks of savings banks, cooperative banks and private banks as the distribution channel, and relying on the local banks' expertise in assessing creditworthiness and risk and on their balance-sheet capacity (KfW institutional documentation). Subsidiarity is the governing idea: the state addresses market weaknesses without disrupting or crowding out private enterprise (KfW institutional documentation).
Delegated Authority, Accreditation, and Approval-Time Targets: Participating banks earn delegated approval authority; processing, closing, servicing, and liquidating guaranteed loans without prior agency review (Code of Federal Regulations, Title 13, Part 120), under earned, renewable accreditation with periodic portfolio examination, mirroring the documented preferred-lender model in which full delegation is granted to the best lenders, renewed at least every two years, with portfolios periodically examined (Small Business Administration program documentation). Accreditation quality control matters: pandemic-era analysis found new lenders experienced significantly higher default rates than the main banking sector (Economics Letters study, 2024). Approval-time targets are set as explicit service standards; for illustration, five business days standard and forty-eight hours for small amounts; feasibility being established by the precedents above. Supervision applies to lenders, never to individual borrowers.
Why Banks Decide and Not Officials: The strongest reason is empirical. In the universe of corporate lending in one emerging market over 1996 to 2002, politically connected firms borrowed 45 percent more and defaulted 50 percent more often, and the preferential treatment occurred exclusively in government banks, with private banks providing no political favors, at an estimated economy-wide cost of 0.3 to 1.9 percent of gross domestic product per year (Khwaja and Mian 2005). Cross-country evidence shows government-owned banks increasing lending in election years relative to private banks (Dinç 2005), and plant-level evidence shows politicians using bank lending to shift employment toward politically attractive regions just before competitive elections (Carvalho 2014). Our design removes both discretion points simultaneously: no official chooses the eligible (the formula does), and no official directs the flow (banks risking their own retained share do). The state's legitimate strengths remain in the architecture through pluralism, not discretion: public and promotional banks participate as accredited lenders alongside private and cooperative banks, preserving the documented countercyclical capacity of public lenders, national development and public retail banks significantly increased lending during crises while private banks contracted, with development bank lending rising 36 percent during 2007 to 2009 against 10 percent for private credit (de Luna-Martínez and Vicente 2012), inside a plural system where no single allocator exists.
The Complementary Minimum Lending Share: The second instrument requires a minimum share of bank lending to reach small enterprises. Precedent one: the Community Reinvestment Act is estimated, using regression-discontinuity and fixed-effects methods on 2004 to 2016 census-tract panels, to increase the number of small business loans by 3 to 7 percent and dollar volume by 6 to 10 percent in treated areas, and institutions maintain higher lending even after agreements expire (United States evaluation literature). Precedent two: priority sector lending frameworks mandate that a portion of bank loans be directed toward designated sectors, with targets historically set at one-third and later 40 percent of adjusted net bank credit and shortfalls triggering contributions to development funds (Reserve Bank of India Master Directions). Our design imports the most useful refinement: tradable compliance certificates, which let banks with surplus achievement sell to banks with shortfalls through a market mechanism, softening the rigidity of quotas (Reserve Bank of India Master Directions). Honest engagement: directed credit has documented costs, including increased transaction costs and nonperforming assets, which is precisely why the quota is the complementary instrument and the production-proportional guarantee is the primary one.
ANTICIPATED FINDINGS, OPEN PARAMETERS, AND EVALUATION DESIGN
Anticipated Findings: First, approval times fall to days, per the delegated-authority and instant-approval precedents. Second, participation among eligible small enterprises multiplies relative to application-based schemes, with the largest gains among previously discouraged firms, per the enrollment evidence. Third, machinery, equipment, and technology investment rises, per the guarantee evidence showing increases in fixed asset investment, research spending, and total factor productivity through those channels. Fourth, employment grows, per the documented positive relationship between guaranteed lending and employment growth, estimated at 3 to 3.5 jobs per million dollars of loans (Brown and Earle 2017). Fifth, business entry recovers at the margin, per the dynamism mechanism. Sixth, fiscal cost stays bounded and transparent, by construction.
Open Parameters: Eligibility age (anchor: five years; an existing national program already uses five years as an eligibility threshold for its entrepreneur loan); entitlement ratio (anchor: 100 percent of trailing five-year taxes, tunable downward); coverage share (within the documented 60 to 85 percent band, per the 75 to 85 percent guarantee structure of the largest national program); approval-time targets; maximum term and pricing; minimum lending share and certificate tradability.
Evaluation Procedure: Identification uses the eligibility threshold: regression discontinuity comparing companies just above five years of operation with just-below-threshold peers, supplemented by difference-in-differences under staggered regional rollout. Outcome metrics: lending volumes, approval times, default rates, participation and application rates, business entry, employment, investment composition, and productivity. Two guardrail metrics are monitored explicitly: the zombie share of eligible borrowers, using the subsidized-interest-rate detection method of the zombie literature, and aggregate reallocation, addressing the productivity concern below. The framework matches the standard evaluation frame in which economic additionality is assessed on employment, turnover, sales and default probability causally influenced by the guarantee (OECD and European Commission 2017).
OBJECTIONS AND ANSWERS
The Zombie Objection: The cautionary tale is real: banks engaged in sham loan restructurings that kept credit flowing to otherwise insolvent borrowers, suppressing competition, congesting markets, and discouraging entry and investment by healthy firms, with zombie-dominated industries showing depressed job creation and productivity, and zombie shares exceeding 25 percent for every year after 1994 (Caballero, Hoshi, and Kashyap 2008). But the anatomy of the zombie mechanism was discretionary evergreening i.e. banks quietly rolling over dead loans to hide their own losses. Our instrument inverts every relevant axis: the ceiling is backward-looking and recomputed from verified production, so a shrinking firm's entitlement shrinks automatically; no sympathetic banker can renegotiate the formula; partial coverage keeps the bank's own money at risk in every rollover decision; and the guarantee attaches to new lending, not to the concealment of old losses. The design is, precisely, an anti-zombie guarantee, and the zombie share is a monitored evaluation metric.
The Political Allocation Objection: Answered structurally above: eligibility is a formula, allocation is a private decision with retained risk, and the empirical record showing political favoritism operating exclusively through government banks and not through private banks (Khwaja and Mian 2005) is the justification for the division of labor.
The Young Firms Objection: Young firms, not small firms as such, create the jobs and the design excludes firms younger than five years. Correct, and deliberate. The five-year screen is what bounds fiscal exposure and removes the early mortality zone. The excluded earliest stage is served by separate instruments, including existing start-up loan channels; and the eligible population; young survivors of the up-or-out filter, is exactly the population whose conditional growth is highest, a targeting choice consistent with evidence that guaranteed-loan employment effects are stronger for younger firms (Brown and Earle 2017). The design trades startup coverage for boundedness, and states the trade openly.
The Productivity Objection: The strongest counter-finding: a crisis-era guarantee program had positive employment effects but dampened worker reallocation toward more productive firms, translating into a reduction in aggregate productivity (Barrot, Martin, Sauvagnat, and Vallée 2024), and the broader evaluation record shows mixed economic additionality (OECD and European Commission 2017). Our reply: selection differs. Crisis rescue selects on distress; our formula selects on demonstrated production, so the subsidy flows in proportion to output rather than in proportion to trouble, and the guarantee evidence where investment channels operate shows robust total factor productivity improvements through research and fixed-asset investment. Aggregate reallocation is a monitored metric, so the concern is testable rather than rhetorical.
The Fraud Objection: Speed plus automation caused fraud in the pandemic schemes, and the documented failure teaches the design: the fast scheme did not require lenders to check the information on the application form or perform credit checks, and provided a 100 percent guarantee precisely because of the absence of checks, and fraud and loss concerns followed (National Audit Office 2020; House of Commons Committee of Public Accounts 2021). The failed component was self-declared data under total coverage. Our design replaces self-declaration with state-verified payment history, a number which cannot be inflated without paying real money to the state, and replaces total coverage with partial coverage. Speed is retained; the fraud vector is removed. The same schemes also demonstrated the upside: up to 500,000 businesses and between half a million and 2.9 million jobs may have been saved (House of Commons Library 2021).
The Fiscal Cost Objection: The instrument is a contingent liability, not spending; exposure is bounded per company by that company's own past contributions; the aggregate ceiling is publishable in real time; and guarantees are documented as a relatively low-cost way to increase small-enterprise lending, with taxpayer cost per job created estimated at 21,000 to 25,000 dollars in the largest evaluated program (Brown and Earle 2017). The comparison baseline is not zero cost but the ongoing cost of rationing: forgone returns documented at 55 to 63 percent per year in field-experimental settings (De Mel, McKenzie, and Woodruff 2008), forgone entry, and forgone dynamism.
GOVERNANCE PRINCIPLE: FISCAL SECULARITY
The design rests on one governance principle, named Fiscal Secularity within the broader research program to which our paper belongs: the constitutional and technical separation of money, taxation, and governance, so that the financial machinery of the state cannot be weaponized against the citizen. In the present instrument the principle takes four concrete forms. Eligibility follows a public formula which no official can override, removing the most weaponizable form of discretion. The entitlement certificate carries only the derived ceiling, never the underlying records: banks never see tax data, and the state never sees business plans; sealed data, transparent design. Access and denial are symmetric and rule-bound, so no company can be starved of the entitlement for holding the wrong opinions. And the enforcement boundary is judicial, reserved for theft and fraud. A credit system that citizens do not fear is not only a matter of rights; the enrollment and burden evidence shows fear and friction measurably suppress participation, so the governance principle and the growth objective point in the same direction.
LIMITATIONS
Our paper is a design proposal with anticipated findings, not an evaluation of a deployed system. Five limitations deserve statement. First, survival statistics used for the screening argument track establishments rather than firms, a measurement caveat inherited from the underlying data. Second, parameter calibration, entitlement ratio, coverage share, pricing, is left open by intention and must be set by pilot evidence, not by argument. Third, general-equilibrium effects, including possible crowding of non-eligible borrowers and interactions with monetary conditions, are not modeled and belong to the evaluation stage. Fourth, transferability varies with state capacity: the design presumes a functioning tax administration and a supervised banking sector, and the on-lending literature warns that the flagship national model may not be easily replicable where the institutional ecosystem of local banks is absent, raising misallocation risks (Roosevelt Institute 2024). Fifth, the instrument deliberately excludes firms younger than five years and nonproduction financial borrowers; complementary instruments carry those margins.
CONCLUSION
The channel which once moved savings into small productive enterprises has narrowed on two dimensions at once, allocation and time, while the enterprises in question remain the majority of employment and the seedbed of dynamism. Our paper proposes to reopen the channel with the thinnest possible machinery: an automatic entitlement computed from taxes already paid, delivered as a certificate through banks which already exist, decided in days, partially covered, fiscally bounded by construction, complemented by a minimum lending share with tradable compliance, and governed by a formula no official can bend. Every component is independently validated in the documented record; the contribution is the bundle. The anticipated result is stated in the title: economic growth, through capital reaching the highest-return producers, and vitality, through renewed entry of the young firms which create the jobs. The evaluation design makes every promise falsifiable.
Banking exists to move savings into new production through entrepreneurs, yet decades of bank mergers have replaced small local banks, which lend on personal knowledge, with large faceless institutions, which lend on standardized data alone, so capital flows overwhelmingly to large corporations while small enterprises remain rationed. Slowness compounds the rationing: half of small enterprises hold under fifteen days of cash, ordinary approvals take weeks, and many viable companies never apply. Our paper asks how governments can restore fast credit access, and thereby economic growth and vitality, for small enterprises; our paper answers with two complementary instruments, grounded in comparative analysis of guarantee schemes and delivery models. The first instrument is an automatic loan guarantee: any company which has operated for five years may borrow up to the total taxes which the company paid during the previous five years, with the state guaranteeing a partial share, so worst-case public exposure stays bounded by revenue already collected, while the verified payment record screens out unstable borrowers. Automatic eligibility matters: enrollment research shows participation multiplying once application burdens disappear, with the largest gains among discouraged applicants. The second instrument is a required minimum share of bank lending directed to small enterprises, following Community Reinvestment Act and priority sector lending precedents, so banks again serve as infrastructure for distributing capital. Anticipated findings include faster approvals, renewed business entry, rising machinery and technology investment, employment growth, and bounded fiscal cost, matching field-experimental evidence of returns far above interest rates. Open parameters include eligibility age, entitlement ratio, coverage percentage, approval-time targets, and minimum lending share; evaluation compares eligible companies with just-below-threshold peers on lending, defaults, entry, employment, and productivity. Delivery requires no new agency: the tax authority issues digital entitlement certificates, existing banks decide within days, disburse, and retain part of every risk, and, consistent with Fiscal Secularity, eligibility follows a public formula, banks never see tax records, and no official decides who deserves capital.
INTRODUCTION
The purpose of banking is old and simple: banks collect the savings of many people and lend the collected savings to people who can turn borrowed money into new production. The academic foundations agree on the point at every level. Banks and similar intermediaries exist because financial markets fundamentally channel savings toward real investment, with banks acting as delegated monitors on behalf of savers (Diamond 1984). In the classic account of economic development, entrepreneurs need credit to hire resources and put innovations into practice; without access to capital even the most visionary entrepreneur cannot build anything, which gives banks a central role in development, not just as lenders but as gatekeepers who decide which new ventures get funded and which do not (Schumpeter 1934). The empirical record confirms the stakes: across 80 countries over the 1960 to 1989 period, measures of financial development are strongly associated with real per capita growth, physical capital accumulation, and efficiency improvements, and the predetermined component of financial development robustly predicts future growth (King and Levine 1993).
Our paper begins from a diagnosis: the channel connecting savings to small productive enterprises has narrowed, and the narrowing has two dimensions. The first dimension is allocation. Decades of consolidation have replaced small, locally rooted lenders with large hierarchical institutions, and the evidence reviewed below shows that large institutions structurally lend to large firms. The second dimension is time. Lending to small enterprises has become slow and burdensome exactly where speed decides survival: the median small business holds enough cash to withstand roughly 27 days without inflows (JPMorgan Chase Institute 2016), and 50 percent of small businesses operate with fewer than 15 cash buffer days (JPMorgan Chase Institute metropolitan liquidity research), while ordinary loan approval commonly takes weeks and full funding often takes one to three months (Small Business Administration program documentation and industry reporting). We call the mismatch the two-clocks problem: for half of small enterprises, the approval clock runs slower than the survival clock.
Our paper asks one question: how can governments restore fast and reliable credit access for small enterprises, and thereby restore economic growth and vitality? Our paper answers with a policy design built on two principles, automatic eligibility and expedited delivery, embodied in two instruments. The first instrument is an automatic loan guarantee in which eligibility and the borrowing ceiling are computed, without any application or official judgment, from a company's verified tax history. The second instrument is a required minimum share of bank lending directed to small enterprises, with documented regulatory precedents. Our paper is a position paper in the sense used by the Advanced Design Conference: our paper states anticipated findings, names every open parameter, and defines the evaluation procedure by which the design can be tested.
Our paper claims four contributions. First, the entitlement formula: to our knowledge, no existing scheme converts cumulative taxes paid into an automatic borrowing entitlement, although every component of the mechanism is independently validated. Second, a transfer of the automatic-enrollment insight from savings policy to credit policy, with a testable prediction about participation. Third, the two-clocks framing, which converts lending speed from a convenience feature into a solvency requirement. Fourth, a fiscal design in which worst-case public exposure is bounded, by construction, by revenue already collected from the same company.
The remainder of our paper proceeds as follows. The background section reviews the evidence on consolidation, rationing, and declining dynamism. The two design sections present the automatic loan guarantee and expedited delivery through existing banks together with the complementary lending-share instrument. The following sections state anticipated findings, open parameters, and the evaluation design; answer the six strongest objections; state the governance principle, Fiscal Secularity, on which the design rests; state limitations; and conclude.
BACKGROUND AND PRIOR ART
The Loss of Soft Information and the Personal Lender: Small-enterprise lending depends on knowledge which does not fit in a spreadsheet; the owner's character, work habits, and local standing. The theoretical foundation is the distinction between soft and hard information: decentralized organizations with small, single-manager units perform best when information about projects is soft and cannot be credibly transmitted, while large hierarchies perform better when information can be costlessly hardened and passed along inside the firm (Stein 2002). The prediction fits the facts: when large banks acquire small banks, a pronounced decline in lending to small businesses has been documented, exactly what the theory predicts where small-business lending relies heavily on soft information (Stein 2002). Region-level evidence agrees: small firms are less likely to borrow from banks after mergers make those banks more hierarchical (Sapienza 2002), and small businesses located in regions dominated by large banks were more likely to face credit constraints than firms located close to small decentralized banks (Berger, Miller, Petersen, Rajan, and Stein 2005). Relationships are the delivery technology for soft information: the primary benefit of building close ties with an institutional creditor is that the availability of financing increases, with smaller effects on price; relationships are valuable and operate through quantities rather than prices (Petersen and Rajan 1994).
Price cannot clear the market because rationing is an equilibrium outcome: even in equilibrium, credit rationing exists in loan markets and; among applicants who appear identical, some receive credit and others do not, even though the rejected applicants would pay higher interest rates (Stiglitz and Weiss 1981). Consequently, access, not price, is the binding constraint, and any effective remedy must change access.
What Consolidation Did: The structural shift is large. From the 1930s until the 1980s, the United States banking system consisted of more than 18,000 primarily small depository institutions; the number of commercial and savings banks then declined by nearly 30 percent between the end of 1985 and the end of 1995 alone (Peek and Rosengren 1998), and the consolidation wave continued afterward. Institutions which merge lose local identity and refocus their franchises toward capital-market services for large corporate clientele (Udell 1998). The allocation result is measurable: a large-sample study of 178,000 observations from the United States banking sector finds that large banks tend to lend to large firms and small banks tend to lend to small firms, so that in highly concentrated systems consisting only of a few large banks, sufficient small-firm funding cannot be expected. The asymmetry after the 2008 crisis is stark: loans originated in amounts under 1 million dollars remained more than 17 percent below the pre-crisis high in 2015, while loans above 1 million dollars surpassed the pre-crisis level by more than 35 percent (Small Business Administration Office of Advocacy 2021). Deposits and lending decoupled: in one large metropolitan market, deposits increased by 100 percent while small business lending decreased by nearly 40 percent (National Community Reinvestment Coalition). The decline continues: between 2019 and 2023, bank small-business lending by real dollar amount declined by 18 percent nationally despite increasing demand (Federal Reserve Bank of St. Louis 2025). Even conditional on obtaining credit, small firms receive systematically harsher terms: 40 percent of credit lines to small firms matured or were callable within one quarter, against 15 percent short-maturity lines for large firms, and fewer than 5 percent of credit lines to small firms were unsecured, compared with 70 percent for the largest firms (National Bureau of Economic Research Digest summary).
The mechanism is physical as well as organizational. Branch closings lead to a persistent decline in local small business lending; annual originations fall by roughly 453 thousand dollars off a 4.7 million dollar baseline and remain depressed for up to six years, with effects dissipating within six miles; and the decline translates into a 2 percentage point reduction in local employment growth, driven primarily by tighter constraints on the size of entering firms (Nguyen 2019). Decisively, the lending contraction persists even after a new branch opens, suggesting the loss is the information destroyed when a local relationship is severed, not reduced competition (Nguyen 2019). European evidence reaches the same conclusion: branch closures significantly reduce employment, sales and working capital of local firms, firm exit probability rises by around 0.6 percentage points over three years if 10 percent of local branches close, and local firms cannot offset the decline with funding from other sources (SUERF policy analysis). In fairness, parts of the merger literature found offsetting responses i.e. static reductions in small business lending after mergers were mostly offset by the reactions of other banks in the same market (Berger, Saunders, Scalise, and Udell 1998),but the offset presupposes surviving small banks able to absorb abandoned borrowers, and the offsetting population is precisely what consolidation has been eliminating. Our design therefore does not attempt to resurrect the past cure; our design engineers a substitute channel which works with the banking structure that actually exists.
What Is at Stake for Growth and Vitality: Small and medium enterprises are the majority of the productive economy: they represent around 90 percent of all businesses and account for more than half of global employment (World Bank), and in the United States, small businesses account for 99 percent of establishments, 48 percent of workers, over 41 percent of net job creation, and 45 percent of gross domestic product (JPMorgan Chase Institute 2016). The credit gap is measured in trillions: more than 200 million formal and informal micro, small and medium enterprises are estimated to be unserved or underserved (International Finance Corporation).
The job-creation evidence carries an important refinement: once firm age is controlled for, there is no systematic relationship between firm size and growth; startups and young businesses drive job creation, and young firms exhibit an up-or-out dynamic (Haltiwanger, Jarmin, and Miranda 2013). Vitality, in our paper, therefore means business dynamism: entry, young-firm growth, and productivity-enhancing reallocation. Dynamism is in documented decline: the economy has become less dynamic, with declining rates of gross job and worker flows and declining rates of entrepreneurship and young-firm activity, pervasive across industries, regions, and firm size classes (Decker, Haltiwanger, Jarmin, and Miranda 2016), and the share of employment at young firms has dropped from 20 percent to 10 percent over several decades (Decker, Haltiwanger, Jarmin, and Miranda 2016). The connection to growth is direct: a large fraction of aggregate productivity growth is accounted for by the movement of resources from less-productive to more-productive businesses, so reduced dynamism implies a slowdown in productivity-enhancing reallocation (Decker, Haltiwanger, Jarmin, and Miranda 2016). Credit conditions are among the suspected causes of diminished dynamism (Davis and Haltiwanger 2015), and the machinery of our design targets exactly the entry-and-young-firm margin.
Finally, the payoff to relaxing the constraint is unusually high at the small end. It is consistently the smallest firms that are most constrained, and it is again the small firms that benefit most when obstacles weaken (Beck, Demirgüç-Kunt, and Maksimovic 2005); the growth of most small firms is constrained by internal finance (Carpenter and Petersen 2002); and randomized evidence shows why the constraint is costly: providing cash and equipment grants to small firms and measuring the profit increase from the exogenous capital shock, the average real return to capital was around five percent per month, substantially higher than market interest rates (De Mel, McKenzie, and Woodruff 2008). When small firms do invest, the purchases match the growth mechanism: technology and information technology top the small-firm investment list at 60 percent, followed by plant, machinery and vehicles and staff training at 52 percent each (Bank of England 2024).
The Established Instrument and the Gap Our Design Fills: Credit guarantee schemes are the standard remedy, with a global evidence base. The evaluation literature finds that credit guarantees provide financial additionality, increasing the availability of credit and reducing its cost (OECD and European Commission 2017); evaluations of the European Union guarantee facility found significantly positive effects on employment and turnover, with micro and young enterprises benefiting most (Asdrubali and Signore 2015); firm-level evidence from China finds guarantees raising the probability of obtaining loans, loan amounts, research spending, fixed asset investment, and total factor productivity; and loan guarantees are documented as a relatively low-cost way to increase lending to small enterprises (Bachas et al. 2021). Design principles are codified: guarantees should be partial and provide capital relief to lenders (World Bank and FIRST Initiative 2015), because full coverage invites lowered lending standards (Yale Program on Financial Stability 2020) and the coverage ratio is the instrument which limits moral hazard for borrowers and lenders alike; United States regulation already requires that the lender retain an economic interest in, and the ultimate risk of loss on, the unguaranteed portion (Code of Federal Regulations, Title 13, Part 120).
Two documented weaknesses motivate our redesign. First, discretion: applications assessed case by case invite selection bias, with intermediaries off-loading their riskier portfolio tranches onto the public guarantee (Asdrubali and Signore 2018), and political incentives to supply guarantees generously while concealing long-term fiscal costs behind uncertain expected losses (Honohan 2010). Second, burden and slowness, which suppress participation among exactly the firms the schemes target: around half of small enterprises are permanent non-borrowers (Bank of England 2024), discouraged borrowers experience outcomes as adverse as those formally denied (Bouslama 2025), and for every three discouraged firms, one would have been approved (Cole and Sokolyk 2016, as reported in Bank for International Settlements Working Paper No. 1041).
Why Automatic and Expedited Loans Matter: The behavioral evidence on automatic eligibility is unambiguous in a neighboring domain. When automatic enrollment replaced voluntary sign-up in a large employer's retirement plan, participation rose from 37 percent to 86 percent, and the switch dramatically changed behavior although no economic feature of the plan changed (Madrian and Shea 2001); industry-scale data show participation tripling from 28 percent to 91 percent (Vanguard 2020); and, critically for our setting, automatic designs equalize participation across groups, with the largest effects among those least likely to participate under affirmative-election regimes (Madrian and Shea 2001). The administrative-burden literature generalizes the lesson to public programs: non-take-up is commonly misread as intentional choice when research shows the assumption does not reflect reality (United States Office of Management and Budget 2022); reducing learning and compliance costs substantially increases access (Health Affairs 2020); the scale of burden-driven loss is large, with an estimated 140 billion dollars in authorized benefits left unclaimed each year in the United States (Institute for Responsive Government); burdens fall hardest on the least resourced (Deshpande and Li 2019); and delivery through the tax system is the documented low-burden channel, as the relatively high take-up and durability of tax-system-delivered benefits attests (Russell Sage Foundation Journal 2023). Our design transfers the insight to credit: existing guarantee schemes are voluntary enrollment; our instrument is automatic enrollment for eligibility. The prediction; participation multiplying, concentrated among discouraged firms, is stated below as an anticipated finding.
Speed has equally hard evidence behind feasibility. Technology-based lenders process applications about 20 percent faster than traditional lenders (Fuster, Plosser, Schnabl, and Vickery 2019); at the frontier, one technology-based lender serving over 20 million small businesses; more than three-quarters of whom had never received a bank business loan, assesses creditworthiness in minutes with zero human interaction using transaction histories (Bank for International Settlements Working Paper No. 1041), and machine-learning credit scoring better predicts default than traditional indicators (Gambacorta, Huang, Qiu, and Wang 2019). Public programs have demonstrated the same at national scale: one scheme launched within 11 days delivered most of 47 billion pounds within two months, aiming to pay borrowers within 24 to 48 hours of application (House of Commons Committee of Public Accounts 2021); a parallel program offered effectively instant support to firms in good standing, because once the stated conditions were met, no further risk assessment was needed (Review of Political Economy analysis of the KfW pandemic response); and in ordinary times, delegated-authority lenders approve guaranteed loans in five to ten business days without agency review, with an express channel carrying a 36-hour turnaround (Small Business Administration program documentation and industry reporting). Against the survival arithmetic above, expedited delivery is not a preference; expedited delivery is the condition under which credit arrives while the borrower is still alive.
INSTRUMENT ONE: THE AUTOMATIC LOAN GUARANTEE
The Lending Rule: Any company which has operated and paid taxes for the trailing five years holds an automatic entitlement to borrow, from any participating bank, up to an amount equal to the total taxes the company paid during those five years, multiplied by a tunable entitlement ratio. The state guarantees a fixed partial share of each entitlement loan. There is no application for eligibility, no business-plan review by any official, and no committee. The tax record, data the government already holds, is simultaneously the eligibility test, the credit ceiling, and the primary risk signal. Eligibility is continuous and rolling: each year, the window advances, and the ceiling recomputes.
Why Five Years and Why Lending on Taxes Paid: Three properties converge on our formula. First, survival screening. Roughly 22 percent of new businesses fail within the first year and nearly half close within five years; by the sixth year only around half survive (United States Bureau of Labor Statistics, Business Employment Dynamics). A five-year threshold therefore places the entire early mortality zone outside the guarantee, and conditional survival rates rise with age, so the eligible population carries structurally lower hazard than the applicant pool a bank ordinarily faces. Second, verified truth. Taxes paid cannot be inflated by optimism: overstating them requires actually paying real money to the state, making the signal costly to fake in the only direction that matters, while understating them is already unlawful. The predictive power of exactly such data is commercially demonstrated: underwriting practice reports that a history of verified tax payments over a long period is a stronger indicator of financial health than personal credit data for thin-file small businesses, and that integrating comprehensive tax data into underwriting may reduce default rates by upwards of 50 percent while increasing approved volume by more than 20 percent (commercial underwriting practice reports). Current practice already uses the record for verification; tax-return verification is a required underwriting step in the largest national guarantee program (Small Business Administration Standard Operating Procedure 50 10), but only as a check on an application. Our design promotes the record from verification document to the entitlement itself; the promotion is the novelty. Third, production-proportionality. The ceiling scales with demonstrated real activity: a growing firm earns more borrowing power and a shrinking firm earns less, automatically. The objections section shows why the property matters.
If richer signals are ever needed, the feasibility bar is low: even simple digital-footprint variables match the information content of credit bureau scores, and their discriminatory power extends to previously unscorable customers (Berg, Burg, Gombović, and Puri 2020). Five years of state-verified payments is a stronger signal than most inputs modern underwriting already trusts; the design is technologically conservative.
Bounded Fiscal Exposure: Because the ceiling equals revenue already collected from the same company, no company can become a net lifetime drain on the treasury: the worst case per company is the return, in guarantee payouts, of a fraction of money the same company previously contributed, and only the guaranteed share of the defaulted fraction is ever paid. Aggregate exposure is transparent by construction; the sum of trailing five-year tax receipts of eligible companies, times the entitlement ratio, times the coverage share, i.e. a number publishable in real time. The design answers the documented fiscal-opacity critique of guarantee schemes structurally rather than by promise. Guarantees remain contingent liabilities with high leverage: schemes can responsibly guarantee loan volumes several times larger than their capital funds (Asian Development Bank 2021).
Moral Hazard Controls: Coverage is partial, never total, per the codified principles above; the lender's retained share preserves the incentive to underwrite and to collect. The borrower's incentive is preserved because default consumes an earned entitlement and carries ordinary insolvency consequences. Fraudulent misrepresentation, including structured fragmentation of one enterprise into several artificial small ones, is classified as fraud and detected algorithmically through shared beneficial ownership and circular transaction flows, which is investigative work on data rather than a new inspection bureaucracy. The governing sentence of the wider research program applies: a company's money is inviolable; lying to the system about who you are is not.
The Entitlement Certificate and Data Minimization: The tax authority issues a digitally signed certificate stating exactly three facts: company identity, current entitlement ceiling, validity period. The bank sees the certificate, never the underlying tax records; the state learns that a loan was requested, never the business plan. Sealed data, transparent design. The certificate also attacks discouragement directly, because a company no longer initiates a plea into the unknown: the company walks into any participating bank already knowing the ceiling printed on a government document. Given the evidence that non-application reflects burden rather than choice, visibility of the entitlement is itself an intervention.
INSTRUMENT TWO AND DELIVERY: EXPEDITED LENDING THROUGH EXISTING BANKS
Division of Labor: The government does three things only: define the standard product, compute entitlements from records already in state hands, and operate the guarantee fund with a central reporting portal, for which national precedent exists; a single scheme portal used by all accredited lenders to report guaranteed facilities and claims, allowing the state to track exposure continuously (British Business Bank scheme documentation). Existing banks do everything else: receive the application, know the customer, price, decide, disburse, service, and collect. The model is proven at national scale: the leading promotional bank does not interact with customers directly but supports banks, with favourable funding and by assuming risks, under the on-lending principle, using the branch networks of savings banks, cooperative banks and private banks as the distribution channel, and relying on the local banks' expertise in assessing creditworthiness and risk and on their balance-sheet capacity (KfW institutional documentation). Subsidiarity is the governing idea: the state addresses market weaknesses without disrupting or crowding out private enterprise (KfW institutional documentation).
Delegated Authority, Accreditation, and Approval-Time Targets: Participating banks earn delegated approval authority; processing, closing, servicing, and liquidating guaranteed loans without prior agency review (Code of Federal Regulations, Title 13, Part 120), under earned, renewable accreditation with periodic portfolio examination, mirroring the documented preferred-lender model in which full delegation is granted to the best lenders, renewed at least every two years, with portfolios periodically examined (Small Business Administration program documentation). Accreditation quality control matters: pandemic-era analysis found new lenders experienced significantly higher default rates than the main banking sector (Economics Letters study, 2024). Approval-time targets are set as explicit service standards; for illustration, five business days standard and forty-eight hours for small amounts; feasibility being established by the precedents above. Supervision applies to lenders, never to individual borrowers.
Why Banks Decide and Not Officials: The strongest reason is empirical. In the universe of corporate lending in one emerging market over 1996 to 2002, politically connected firms borrowed 45 percent more and defaulted 50 percent more often, and the preferential treatment occurred exclusively in government banks, with private banks providing no political favors, at an estimated economy-wide cost of 0.3 to 1.9 percent of gross domestic product per year (Khwaja and Mian 2005). Cross-country evidence shows government-owned banks increasing lending in election years relative to private banks (Dinç 2005), and plant-level evidence shows politicians using bank lending to shift employment toward politically attractive regions just before competitive elections (Carvalho 2014). Our design removes both discretion points simultaneously: no official chooses the eligible (the formula does), and no official directs the flow (banks risking their own retained share do). The state's legitimate strengths remain in the architecture through pluralism, not discretion: public and promotional banks participate as accredited lenders alongside private and cooperative banks, preserving the documented countercyclical capacity of public lenders, national development and public retail banks significantly increased lending during crises while private banks contracted, with development bank lending rising 36 percent during 2007 to 2009 against 10 percent for private credit (de Luna-Martínez and Vicente 2012), inside a plural system where no single allocator exists.
The Complementary Minimum Lending Share: The second instrument requires a minimum share of bank lending to reach small enterprises. Precedent one: the Community Reinvestment Act is estimated, using regression-discontinuity and fixed-effects methods on 2004 to 2016 census-tract panels, to increase the number of small business loans by 3 to 7 percent and dollar volume by 6 to 10 percent in treated areas, and institutions maintain higher lending even after agreements expire (United States evaluation literature). Precedent two: priority sector lending frameworks mandate that a portion of bank loans be directed toward designated sectors, with targets historically set at one-third and later 40 percent of adjusted net bank credit and shortfalls triggering contributions to development funds (Reserve Bank of India Master Directions). Our design imports the most useful refinement: tradable compliance certificates, which let banks with surplus achievement sell to banks with shortfalls through a market mechanism, softening the rigidity of quotas (Reserve Bank of India Master Directions). Honest engagement: directed credit has documented costs, including increased transaction costs and nonperforming assets, which is precisely why the quota is the complementary instrument and the production-proportional guarantee is the primary one.
ANTICIPATED FINDINGS, OPEN PARAMETERS, AND EVALUATION DESIGN
Anticipated Findings: First, approval times fall to days, per the delegated-authority and instant-approval precedents. Second, participation among eligible small enterprises multiplies relative to application-based schemes, with the largest gains among previously discouraged firms, per the enrollment evidence. Third, machinery, equipment, and technology investment rises, per the guarantee evidence showing increases in fixed asset investment, research spending, and total factor productivity through those channels. Fourth, employment grows, per the documented positive relationship between guaranteed lending and employment growth, estimated at 3 to 3.5 jobs per million dollars of loans (Brown and Earle 2017). Fifth, business entry recovers at the margin, per the dynamism mechanism. Sixth, fiscal cost stays bounded and transparent, by construction.
Open Parameters: Eligibility age (anchor: five years; an existing national program already uses five years as an eligibility threshold for its entrepreneur loan); entitlement ratio (anchor: 100 percent of trailing five-year taxes, tunable downward); coverage share (within the documented 60 to 85 percent band, per the 75 to 85 percent guarantee structure of the largest national program); approval-time targets; maximum term and pricing; minimum lending share and certificate tradability.
Evaluation Procedure: Identification uses the eligibility threshold: regression discontinuity comparing companies just above five years of operation with just-below-threshold peers, supplemented by difference-in-differences under staggered regional rollout. Outcome metrics: lending volumes, approval times, default rates, participation and application rates, business entry, employment, investment composition, and productivity. Two guardrail metrics are monitored explicitly: the zombie share of eligible borrowers, using the subsidized-interest-rate detection method of the zombie literature, and aggregate reallocation, addressing the productivity concern below. The framework matches the standard evaluation frame in which economic additionality is assessed on employment, turnover, sales and default probability causally influenced by the guarantee (OECD and European Commission 2017).
OBJECTIONS AND ANSWERS
The Zombie Objection: The cautionary tale is real: banks engaged in sham loan restructurings that kept credit flowing to otherwise insolvent borrowers, suppressing competition, congesting markets, and discouraging entry and investment by healthy firms, with zombie-dominated industries showing depressed job creation and productivity, and zombie shares exceeding 25 percent for every year after 1994 (Caballero, Hoshi, and Kashyap 2008). But the anatomy of the zombie mechanism was discretionary evergreening i.e. banks quietly rolling over dead loans to hide their own losses. Our instrument inverts every relevant axis: the ceiling is backward-looking and recomputed from verified production, so a shrinking firm's entitlement shrinks automatically; no sympathetic banker can renegotiate the formula; partial coverage keeps the bank's own money at risk in every rollover decision; and the guarantee attaches to new lending, not to the concealment of old losses. The design is, precisely, an anti-zombie guarantee, and the zombie share is a monitored evaluation metric.
The Political Allocation Objection: Answered structurally above: eligibility is a formula, allocation is a private decision with retained risk, and the empirical record showing political favoritism operating exclusively through government banks and not through private banks (Khwaja and Mian 2005) is the justification for the division of labor.
The Young Firms Objection: Young firms, not small firms as such, create the jobs and the design excludes firms younger than five years. Correct, and deliberate. The five-year screen is what bounds fiscal exposure and removes the early mortality zone. The excluded earliest stage is served by separate instruments, including existing start-up loan channels; and the eligible population; young survivors of the up-or-out filter, is exactly the population whose conditional growth is highest, a targeting choice consistent with evidence that guaranteed-loan employment effects are stronger for younger firms (Brown and Earle 2017). The design trades startup coverage for boundedness, and states the trade openly.
The Productivity Objection: The strongest counter-finding: a crisis-era guarantee program had positive employment effects but dampened worker reallocation toward more productive firms, translating into a reduction in aggregate productivity (Barrot, Martin, Sauvagnat, and Vallée 2024), and the broader evaluation record shows mixed economic additionality (OECD and European Commission 2017). Our reply: selection differs. Crisis rescue selects on distress; our formula selects on demonstrated production, so the subsidy flows in proportion to output rather than in proportion to trouble, and the guarantee evidence where investment channels operate shows robust total factor productivity improvements through research and fixed-asset investment. Aggregate reallocation is a monitored metric, so the concern is testable rather than rhetorical.
The Fraud Objection: Speed plus automation caused fraud in the pandemic schemes, and the documented failure teaches the design: the fast scheme did not require lenders to check the information on the application form or perform credit checks, and provided a 100 percent guarantee precisely because of the absence of checks, and fraud and loss concerns followed (National Audit Office 2020; House of Commons Committee of Public Accounts 2021). The failed component was self-declared data under total coverage. Our design replaces self-declaration with state-verified payment history, a number which cannot be inflated without paying real money to the state, and replaces total coverage with partial coverage. Speed is retained; the fraud vector is removed. The same schemes also demonstrated the upside: up to 500,000 businesses and between half a million and 2.9 million jobs may have been saved (House of Commons Library 2021).
The Fiscal Cost Objection: The instrument is a contingent liability, not spending; exposure is bounded per company by that company's own past contributions; the aggregate ceiling is publishable in real time; and guarantees are documented as a relatively low-cost way to increase small-enterprise lending, with taxpayer cost per job created estimated at 21,000 to 25,000 dollars in the largest evaluated program (Brown and Earle 2017). The comparison baseline is not zero cost but the ongoing cost of rationing: forgone returns documented at 55 to 63 percent per year in field-experimental settings (De Mel, McKenzie, and Woodruff 2008), forgone entry, and forgone dynamism.
GOVERNANCE PRINCIPLE: FISCAL SECULARITY
The design rests on one governance principle, named Fiscal Secularity within the broader research program to which our paper belongs: the constitutional and technical separation of money, taxation, and governance, so that the financial machinery of the state cannot be weaponized against the citizen. In the present instrument the principle takes four concrete forms. Eligibility follows a public formula which no official can override, removing the most weaponizable form of discretion. The entitlement certificate carries only the derived ceiling, never the underlying records: banks never see tax data, and the state never sees business plans; sealed data, transparent design. Access and denial are symmetric and rule-bound, so no company can be starved of the entitlement for holding the wrong opinions. And the enforcement boundary is judicial, reserved for theft and fraud. A credit system that citizens do not fear is not only a matter of rights; the enrollment and burden evidence shows fear and friction measurably suppress participation, so the governance principle and the growth objective point in the same direction.
LIMITATIONS
Our paper is a design proposal with anticipated findings, not an evaluation of a deployed system. Five limitations deserve statement. First, survival statistics used for the screening argument track establishments rather than firms, a measurement caveat inherited from the underlying data. Second, parameter calibration, entitlement ratio, coverage share, pricing, is left open by intention and must be set by pilot evidence, not by argument. Third, general-equilibrium effects, including possible crowding of non-eligible borrowers and interactions with monetary conditions, are not modeled and belong to the evaluation stage. Fourth, transferability varies with state capacity: the design presumes a functioning tax administration and a supervised banking sector, and the on-lending literature warns that the flagship national model may not be easily replicable where the institutional ecosystem of local banks is absent, raising misallocation risks (Roosevelt Institute 2024). Fifth, the instrument deliberately excludes firms younger than five years and nonproduction financial borrowers; complementary instruments carry those margins.
CONCLUSION
The channel which once moved savings into small productive enterprises has narrowed on two dimensions at once, allocation and time, while the enterprises in question remain the majority of employment and the seedbed of dynamism. Our paper proposes to reopen the channel with the thinnest possible machinery: an automatic entitlement computed from taxes already paid, delivered as a certificate through banks which already exist, decided in days, partially covered, fiscally bounded by construction, complemented by a minimum lending share with tradable compliance, and governed by a formula no official can bend. Every component is independently validated in the documented record; the contribution is the bundle. The anticipated result is stated in the title: economic growth, through capital reaching the highest-return producers, and vitality, through renewed entry of the young firms which create the jobs. The evaluation design makes every promise falsifiable.