The Hidden Policy Crisis Behind America’s Startup Decline

Research across Cato, Brookings, Mercatus, Kauffman, the SBA, and the Congressional Research Service increasingly points to the same conclusion

Cato Institute policy analyst, Solveig Singleton, published a paper this week titled Addressing Civil Investigative Demand Overreach at the Consumer Financial Protection Bureau, and I fear that approximately zero people in the venture community read it.

That’s not remotely a criticism of their work. It is what motivated me to push you all to engage more.

Singleton’s paper is one of the most consequential startup-policy documents published this year, and the fact that it lives at cato.org under the “Banking and Finance” issue tag instead of being summarized in The Information or TechCrunch is exactly why founders and investors keep losing fights they didn’t know they were in.

I write about public policy and startup ecosystems regularly, because the lesson I learned watching cities, regions, and entire categories of capital formation rise and collapse is that policy is the operating system on which everything else runs. Founders treat it as background noise, if at all; investors treat it as something their LP relations team handles, if at all; economic development professionals treat it as a press release because something was done. None of those framings survive contact with the actual mechanism. I covered this in Strengthening Entrepreneurship Infrastructure: A Policy Guide, again in Founders Can’t Scale What Government Won’t, and again in Startup Economic Policy That Matters for Entrepreneurs, because the message has to be repeated to land, and even then, it mostly doesn’t because, I think, this sector of our economy doesn’t know what to do (and doesn’t fund the lobbyists involved in working with policymakers).

Singleton’s CFPB paper is one document in a much larger body of policy research from Cato, the Mercatus Center, the Brookings Institution, and the Ewing Marion Kauffman Foundation that explores what decides whether your startup survives, whether your fund returns capital, and whether your ecosystem produces anything worth investing in. Almost none of it is being read by the people whose returns depend on it.

What the CFPB Paper Says, my read, Briefly

The Consumer Financial Protection Bureau (CFPB) is the federal agency, created in 2010 by Title X of the Dodd-Frank Wall Street Reform and Consumer Protection Act, that enforces federal consumer financial law across banks, credit unions, payday lenders, mortgage servicers, debt collectors, credit reporting agencies, and a growing list of nonbank financial firms (meaning firms that handle consumer money without being licensed depository banks; think Stripe, Chime, Affirm, Klarna). A Civil Investigative Demand (which I’ll explain why I’m bringing up), or CID, is an administrative subpoena; that phrase deserves a translation because a subpoena is normally a court-ordered command to produce evidence; an administrative subpoena is the same type of command, except issued by the agency itself, on its own authority, without a judge ever looking at it. The recipient must produce documents, tangible items, written reports, or oral testimony, and refusing means a federal court fight you usually lose.

Singleton documents that the CFPB’s CIDs are routinely overbroad, poorly justified, and economically devastating to recipients. The Bureau routinely cites catch-all categories like “any other federal consumer financial law,” demands records spanning years, and frequently fails to identify what specific conduct is under investigation. According to testimony from Rebecca Kuehn of Hudson Cook before the House Financial Services Subcommittee on Financial Institutions in March 2025, the Bureau has used CIDs to investigate entire categories of businesses because they operate in industries it disfavors, requiring those businesses “to prove [they have] not violated a myriad of consumer financial laws.” Compliance costs run from tens of thousands of dollars on the low end to over a million on the high end. Attorney Joel Incorvaia, documented in a public comment to the CFPB that his firm was never told what conduct was under investigation and incurred $110,000 in employee time alone.

We’re here to translate that into venture math.

A Seed-stage fintech with $1.5M in the bank and an 18-month runway receives a CID demanding seven years of customer records, transaction logs, marketing materials, internal Slack messages, and oral testimony from the founders. Outside counsel quotes $200,000 minimum to respond while there is no specific violation alleged. The company cannot raise a bridge round because no rational investor signs a term sheet into an open federal investigation. The company dies, having harmed exactly nobody, and the cap table writes off the position. For a VC, that is the difference between a regulated fintech being a viable thesis and being an entire category to avoid. It explains why so many promising consumer finance categories sit underfunded relative to their underlying market opportunity.

This Matters GREATLY Because Real Startups Are Disappearing

Before we go further into the broader policy landscape, we need to settle a question that will probably surprise most of you.

Is startup formation actually in decline?

Because if you’ve been watching the news for the past four years, you’ve seen the opposite story and I think we’d all agree that it seems like entrepreneurship is booming (and as such, we must be launching a lot more startups, right?!). The Census Bureau’s Business Formation Statistics (BFS, which tracks Employer Identification Number filings with the IRS as a proxy for “people starting a business”) show that monthly business applications nearly doubled from the pre-pandemic average and have stayed elevated ever since. The Small Business and Entrepreneurship Council, the U.S. Chamber of Commerce, and every administration of the last six years have used that surge as proof that American entrepreneurship is booming.

*whew*

And yet, it isn’t, and that’s where we should be alarmed.

The surge in business applications is mostly LLCs filed by gig workers, real estate vehicles, single-member consultancies, and Etsy sellers.

They are not what venture capitalists, economic developers, or economists mean by a “startup.”

The relevant measure is what the Census Bureau calls a new “employer firm,” which is a business that actually hires somebody and runs payroll. That data lives in a different Census product, Business Dynamics Statistics (BDS), and it tells the opposite story. The Bipartisan Policy Center documents that “prior to the COVID-19 pandemic, the rate of new (employer) business creation in the United States had fallen steadily for four decades. From 2010 to 2018, annual new business creation was about 14 percent lower than from 1978 to 2007.” The Congressional Research Service, in its 2024 brief Is U.S. Entrepreneurship Declining?, confirms that “the U.S. establishment entry rate, with some fluctuations, has largely declined since the late 1970s.”

That alone should worry you; it flies in the face of our perception. It gets worse when you look at high-growth firms specifically, which is what VCs and ecosystem builders actually care about. The peer-reviewed Firm Growth and Stagnation in the United States: Key Trends and New Data Opportunities, in the Strategic Management Journal, used a brand-new public Census dataset called BDS-HG (Business Dynamics Statistics – High Growth) tracking every American employer business from 1978 to 2021. Their headline finding: “The US economy has seen two major shifts in firm growth dynamics: a steady decline in the share of high-growth firms, especially among startups, and a notable rise in firms that exhibit no growth.” Read that twice because I’ve exposed extensively the fact that cities throughout the world are struggling with what we call the “missing middle” of venture funding… but this reinforces that such funding seems inaccessible because the startups aren’t actually reaching the point that it’s warranted. They certainly think they are… but is the lack of sufficient Series A funding a result of it lacking or a consequence of circumstances. I argue, circumstances, and these regulatory issues are a MAJOR contributing factor.

The share of fast-growing companies in the American economy has been falling, the share of zombie companies (firms that exist but don’t grow) has been rising, and the effect is concentrated among young and small firms.

The U.S. Small Business Administration’s Office of Advocacy reports something equally disturbing in its analysis of high-propensity applications versus actual employer firm formations. Even among the EIN applications that the Census Bureau classifies as “high-propensity” (meaning they look most likely to become real businesses with employees), the conversion rate has been dropping. More people are filing paperwork to start companies; fewer of them are crossing the threshold into hiring anyone.

This is what makes the surge of entrepreneurship story dangerous. Politicians and economic developers point at the EIN filing data and declare victory; meanwhile the actual production of companies that hire people, scale, raise capital, and exit has been deteriorating for forty years and the deterioration in high-growth firms specifically has been documented across at least three independent peer-reviewed datasets. The Kauffman Foundation, the Brookings Institution, and academic economists publishing in Strategic Management Journal and Economic Affairs are all looking at the same picture and reaching the same conclusion. This is not a fringe finding.

If you are a venture capitalist whose returns depend on the existence of high-growth firms; if you are an economic developer whose ecosystem strategy depends on companies actually crossing the threshold from idea to payroll; if you are a founder competing against an environment where new entrants are rarer than they used to be, then this should change how you read every other paragraph of this article.

The American startup engine is running at meaningfully lower output than it did in the late 1970s, the deterioration accelerated through the 2010s, and the post-COVID application surge has obscured that fact rather than reversed it.

What Is Causing It: The Pattern Across Ideologies

The case for regulatory accumulation as a meaningful causal contributor is strong, and the strength comes for this argument comes from ideological diversity converging on the same answer.

The Mercatus Center at George Mason University, a free-market academic center, maintains RegData, a database that quantifies regulatory restrictions in the U.S. Code of Federal Regulations across nearly five decades of industry-specific data. Using it, Dustin Chambers, Patrick McLaughlin, and Tyler Richards, there, found that “regulatory growth disproportionally burdens small businesses relative to large business; a 10 percent growth in regulatory restrictions in an industry is associated with a reduction in the total number of small firms within that industry but has no statistically significant association with the number of large firms in the industry.”

Translated: when an industry gets more regulated, small firms disappear and large firms don’t.

A separate Mercatus analysis estimates that regulatory accumulation since 1980 cost the U.S. economy approximately $4 trillion in 2012 GDP alone, if you’d held regulation at 1980 levels.

Other work confirms the same pattern with different methods. Liya Palagashvili’s 2024 study in Economic Affairs, What Is the Relationship Between Industry-Specific Regulation and Technology Startups?, cites Bailey and Thomas (2017) finding that “a 10 percent increase in the intensity of regulation as measured by the RegData index leads to a statistically significant 0.47 percent decrease in overall firm births.” Calcagno and Sobel (2013) found that regulations function as a fixed cost (meaning the cost of compliance is roughly the same whether you’re a five-person startup or a 5,000-person corporation), which mathematically produces larger average firm sizes and crushes the smallest firms first. Crain and Crain (2014) measured the per-employee compliance cost across industries and found it consistently greatest for small businesses and lowest for large ones.

Want to play politics? Let’s lean center-left. The Brookings Institution (and I’m not calling it left, this is; what I want to help establish is this is an issue with broad consensus regardless of political views), has independently documented the same trend. Ian Hathaway and Robert Litan, both Brookings economists, published What’s Driving the Decline in the Firm Formation Rate? A Partial Explanation, establishing that new firm formation has fallen across all 50 states, every major metropolitan area, and every industry, including tech (granted, it’s an old study). The Ewing Marion Kauffman Foundation, which is nonpartisan and operates with the explicit mission of advancing entrepreneurship, has published years of research showing that the number of U.S. companies less than a year old declined as a share of all businesses by nearly 44 percent between 1978 and 2012 (also old, but bear with me to make the point). John Haltiwanger, Ron Jarmin, and Javier Miranda established in Who Creates Jobs? Small vs. Large vs. Young that new and young firms generate disproportionately large contributions to net new jobs, with young high-tech businesses adding jobs at a rate twice that of all firms.

The point: When Cato, Mercatus, Brookings, Kauffman, the Bipartisan Policy Center, the Congressional Research Service, the SBA’s own Office of Advocacy, and academic economists publishing in peer-reviewed journals converge on the same finding using different data and different priors, that finding is not ideology; it’s signal. Not that it is happening nor that it was happening decades ago but the CAUSE underlying. Startup formation has been falling for decades, regulatory accumulation is the meaningful causal contributor, and the burden falls disproportionately on the firms that produce most net new jobs in the economy.

What Else Cato Has Been Publishing (And Why You Should Care)

Singleton’s paper sits within a research program that touches most variables founders and VCs need to track. By topic:

Capital formation and securities law. Federal securities law decides who is legally allowed to invest in your startup. Under Rule 501 of Regulation D, the SEC’s “accredited investor” definition limits most private securities offerings (meaning private company stock sales not registered with the SEC, including basically every venture deal) to individuals earning $200,000 annually or with $1 million in net worth excluding their primary residence. If you’re below those thresholds, the federal government has decided you are not “sophisticated” enough to invest. Cato’s The Case for Micro-Offerings documents that this wealth test excludes more than 87 percent of Americans from a private capital market that raised $2.7 trillion in exempt offerings in 2019 alone, compared to $1.2 trillion in registered (public) offerings. Their earlier policy analysis, Your Money’s No Good Here, makes the case (which frankly, is obvious) that the wealth threshold is a poor proxy for sophistication and locks out exactly the people most likely to invest locally in founders they know. Hell, it’s a terrible proxy, and I’ve constantly ranted that the reason we lack risk tolerant startup investors is not because the capital doesn’t exist but because the investors themselves are qualified merely by being rich enough or not. I don’t want startup investors because they have money; we need those that know startups!

Cato’s recent commentary on the INVEST Act of 2025 addresses both the accredited investor framework and Regulation Crowdfunding, commonly called Reg CF. Reg CF is the SEC rule, created by the JOBS Act, that lets any company raise up to $5 million per year by selling stock to ordinary investors (not just wealthy ones) through SEC-registered online platforms such as Republic, Wefunder, and StartEngine. The cap, the disclosure requirements, and the platform rules all sit in this body of regulation that the SEC quietly adjusts and that almost no founder reads. There is also Regulation A+, the JOBS Act’s other path, which lets a company raise up to $75 million from the general public in what amounts to a mini-IPO without the full registration burden. For any founder raising under Rule 506(b) or 506(c) (the two most common private-placement exemptions for accredited-only and accredited-plus-marketing rounds), this is the rulemaking environment that determines what is actually possible.

And by the way, I can count on one hand how many founders have any idea what we’re talking about when we say Reg CF, 506(b), or accredited investor, let alone what it means to them.

Financial regulation generally. Cato’s Fifteen Years of Dodd-Frank: A Legacy of Missed Targets and Regulatory Overreach addresses what produced both the CFPB and the Volcker Rule. The Volcker Rule is the Dodd-Frank provision that bans banks from using their own balance sheets to make venture capital, private equity, or proprietary trading investments. It’s the reason commercial banks largely can’t run their own VC arms. The Center for Monetary and Financial Alternatives, where Singleton sits, produces ongoing work on bank capital requirements, the Financial Stability Oversight Council (a regulatory body created by Dodd-Frank that designates firms as “systemically important” and subjects them to enhanced federal supervision), and the Orderly Liquidation Authority (the framework for winding down failed large financial institutions). Cato’s Regulation quarterly journal covers administrative law topics relevant to any operator in a regulated sector.

Tech and platform policy. Content Creators, Entrepreneurial Users, and the Impact of Tech Policy catches my attention Jennifer Huddleston, examines how proposed changes to Section 230 (the federal law that shields online platforms from legal liability for what their users post, and that effectively makes the modern internet possible), antitrust enforcement, and the Children’s Online Privacy Protection Act, known as COPPA, affect creator-economy entrepreneurship. Panagiotis Mallios documented that social media users increasingly view starting a business as feasible specifically because the platforms enable it; constrict the platforms, and you constrict the entrepreneurship. Cato’s AI and Privacy Rules Meant for Big Tech Could Hurt Small Businesses Most makes the case that rules sold as anti-incumbent measures predictably benefit incumbents, because compliance is a fixed cost. This is the same dynamic economists call “regulatory capture” (the process by which the firms regulators are supposed to constrain end up shaping the regulations in ways that suppress new entrants).

Mercatus research on lobbying and business dynamism finds that “regulations and lobbying reduce the entry and growth of small businesses relative to large businesses, and combined with rising entry costs, this has led to a less dynamic and adaptable economy.”

Occupational licensing and zoning. Most startup ecosystem conversations fixate on tech and fintech while ignoring the categories that account for the majority of new business formation. Occupational licensing is the state-level legal requirement that you obtain a government-issued license to perform certain types of work, including tutoring and telehealth practice, where we’re seeing startups try. Cato’s coverage of occupational licensing reform addresses the regimes that determine whether someone can legally start a business at all. Their Facilitating Personal Improvement and Entrepreneurship at Home-Based Businesses treats residential zoning as a startup killer, noting that homes are low-cost incubators where founders test ideas before they can afford commercial space. For state and municipal economic development professionals, these are foundational regulatory environments your ecosystem either supports or doesn’t.

“There is no better place for low-cost product experimentation than an entrepreneur’s home. Homes are low-cost incubators to test business ideas before larger investments are made,” wrote Chris Edwards. “Startups are risky and have high failure rates, so entrepreneurs need early, low-cost feedback from consumers. Food entrepreneurs, for example, want to test recipes with consumers but may not be able to initially afford commercial kitchen space. Home production can give entrepreneurs the confidence, skills, and capital they need to later open a brick-and-mortar location.”

Tax and capital gains. Chris also researched and published, We’re Going to Need Startups, arguing that capital gains tax rates directly affect the flow of risk capital into early-stage companies.

Cato’s broader tax policy track frequently engages the Qualified Small Business Stock exemption of the Internal Revenue Code (a tax provision that lets early shareholders of a qualifying small company exclude up to $10 million of capital gains from federal tax when they sell), carried interest treatment (the rule that lets VC and PE fund managers pay capital gains rates rather than ordinary income rates on their fund profits), R&D expensing, and the Section 174 fight. Section 174 is the IRS provision that, between 2022 and partial relief in 2025, forced software companies to amortize their research and development spending over five years instead of deducting it immediately. The practical effect was that startups owed federal income tax on money they hadn’t actually earned yet, and software hiring collapsed for two years as a direct result. Most founders affected by it didn’t know it was happening until their accountants told them.

Hopefully you’re seeing why this all matters so much and why I’m so passionate that our startup sector needs more attention on it. There’s more…

Constitutional and administrative law. Brent Skorup’s Will ‘Administrative Subpoenas’ Survive? provides the frame for Singleton’s CFPB paper, which is why I’ll end my assessment with it. Loper Bright Enterprises v. Raimondo, decided by the U.S. Supreme Court in 2024, overturned the so-called Chevron doctrine. Chevron, named after a 1984 Supreme Court case involving Chevron USA, was the rule that required federal courts to defer to a federal agency’s own interpretation of any ambiguous statute the agency administered. In practice, that gave agencies like the SEC, the FTC, the EPA, and the CFPB enormous discretion to interpret their own authority. Loper Bright killed that with the practical effect being that founders facing agency action now have meaningfully better odds in federal court than they did two years ago, because judges are no longer required to take the agency’s word for what the statute means.

This is a representative slice of how policy supersedes the conventional startup advice and largely decides success or failure. This is the Go/No-Go work.

The fact that almost none of this material is cited in the venture capital trade press, in board decks, or in letters to limited partners (the investors who fund VC firms) is aninformation failure that benefits incumbents and disadvantages new entrants. Policy variables that determine whether a startup can form, raise capital, hire, scale, and exit. Which, if you read the regulatory capture literature, is exactly what the dynamic predicts.

The Economics, Not the Politics

The economic argument is independent of political affiliation, and that is the part founders and investors need to internalize: Regulations impose compliance costs.

Compliance costs are largely fixed (meaning the cost is roughly the same regardless of company size). Fixed costs fall most heavily, per dollar of revenue, on the smallest firms. The smallest firms are every startup. Therefore, increases in regulatory complexity systematically disadvantage startups relative to incumbents, regardless of the merit of any individual regulation. This holds whether the rule was written by a Democratic administration, a Republican administration, or a bipartisan congressional consensus.

Incumbents have rational incentives to support regulatory expansion, even regulation ostensibly targeting them, because the marginal compliance cost is trivial for them and prohibitive for their would-be competitors.

When founders cheer for AI regulation that will primarily harm new AI entrants; when fintech CEOs publicly support consumer protection rules they can afford and their seed-stage competitors cannot; when established VCs back accredited investor rules that lock retail capital out of the asset class they themselves operate in; the resulting market structure changes incentives.

That dynamic does not require anyone to be malicious; it requires only that everyone act rationally in their own interest. Which is why ecosystem policy, securities reform, occupational licensing reform, and administrative law reform all matter. The question is whether the policy produces more startups, more capital formation, and more economic mobility, or not.

We talked about recently the Best and Potential States for Startups: A Policy-First Look at Who’s Helping Founders, where the underlying point is that policy trajectory matters more than current ecosystem scoreboards, and that the states actually doing the regulatory work are the ones where capital and founders will eventually concentrate. The same logic applies at the federal level. The agencies, doctrines, and statutes Cato, Mercatus, Brookings, and Kauffman are litigating in public are the ones that will decide which sectors are investable in 2030.

What Founders and Investors Should Actually Do

Reading the papers is the first move but believe me, I know that’s tough, we all publish a lot. So, not all of them; pick three per quarter that touch your sector. If you operate in consumer finance, read Singleton. If you raise capital from individuals, read the micro-offerings paper. If you run a regulated marketplace, read the Section 230 work. If you allocate to early-stage funds, ask your general partners (the people who actually run the venture fund) which policy variables they consider material to their thesis, and if they cannot name three, that is information about the fund.

Engage with the trade associations that file comments on rulemakings. Reg CF reform, accredited investor redefinition, CID procedure, occupational licensing reciprocity, and zoning preemption are all decided through federal and state comment periods that almost no actual founders or VCs participate in. The Securities and Exchange Commission’s Office of Small Business Policy reads the comments it receives; the Federal Trade Commission reads its comments; the CFPB reads its comments. The comment records on these rulemakings are dominated by incumbent law firms representing incumbent clients, which is one reason the rules end up looking the way they do.

Push your local economic development organization or someone like me, to engage with policy work, not just programming. Founders Can’t Scale What Government Won’t. Accredited investor reform, tax incentives that make angel investment rational in secondary markets, Small Business Innovation Research (SBIR) grants that reduce the risk profile of early-stage technical R&D, and securities frameworks that enable regional fund formation are all better-documented and more replicable than the ecosystem culture interventions that try.

If you had never heard of Solveig Singleton until this morning, that’s fine. If you don’t know what Cato, Mercatus, Brookings, the Kauffman Foundation, the Bipartisan Policy Center, the Congressional Research Service, or the SBA’s own Office of Advocacy, have published on the variables that decide whether your startup or your portfolio company survives. The research exists; it has been peer-reviewed; it converges across the ideological spectrum on findings that should change how you think about which sectors are fundable, which ecosystems are emerging, and which fights are worth picking.

Six Things to Do This Quarter

1. Subscribe to three policy sources this week. Cato’s Center for Monetary and Financial Alternatives, Mercatus’s regulatory research feed, and the Kauffman Indicators of Entrepreneurship newsletter; together they cost about ten minutes a week to read and they will surface rulemakings your competitors are already commenting on.
2. Track the federal agency dockets for your sector on regulations.gov. Set email alerts on docket numbers that touch your business; you want to see proposed rules during the comment window when they can still be shaped, not after they finalize when your only option is litigation.
3. Co-sign a coordinated comment letter the next time a rulemaking lands in your sector. A solo comment from one founder carries weight; a coordinated comment from twelve operators in the same category carries materially more, and the bottleneck is almost always that nobody volunteers to write the draft, so write it.
4. Build a direct relationship with two policy researchers (heck, I’m right here and you know I’m easy to connect with) whose work touches your business. Email the authors of the papers you read; most respond, and the exchange improves both their research and your operating context.
5. Watch the Supreme Court and federal circuit cases this term that affect your industry. SCOTUSblog summarizes argued cases for free; the difference between a 6-3 win and a 6-3 loss in an administrative law case is sometimes the difference between your sector being investable in 2030 and being a graveyard.
6. Put money into the research itself. A grant of $5,000 to $25,000 to a researcher or policy center producing work in your sector returns more leverage than the same dollars spent on another conference sponsorship; policy research is structurally underfunded relative to its impact on the outcomes you are trying to produce.