> For the complete documentation index, see [llms.txt](https://laurence-wilse-samson.gitbook.io/textbooks/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://laurence-wilse-samson.gitbook.io/textbooks/financial-economics-claims-prices-holders/part-v-firms-as-issuers-of-claims/chapter_25_firm_financing.md).

# Chapter 25: How Firms Actually Finance Themselves

*Part V: Firms as Issuers of Claims — Financial Economics: Claims, Prices, and Holders*

***

## Opening Episode: One Company's Financing Biography

In the last weeks of 2008 a car company with no profitable product and a few weeks of cash closed a financing round of roughly forty million dollars, structured as convertible debt and bought almost entirely by people who already owned its equity. Tesla Motors had no bank willing to lend to it, no rating, no bond investors, and no public market. It had a small number of existing holders who could be persuaded that the alternative — writing the whole position to zero — was worse. That is what a claim looks like when the only available holder is an insider.

Fifteen years later the same company was rated investment grade, held cash exceeding its debt, and could have placed a benchmark bond with insurance companies at a spread of a hundred-odd basis points over Treasuries. Nothing in between was a single financing decision. It was a sequence of about a dozen of them, and each one sold a *different claim* to a *different holder*, in an order that no theory of the optimal debt ratio predicts and that this chapter is built to explain.

The public record, assembled from the company's own filings and contemporaneous reporting, runs approximately as follows. Amounts are rounded and should be read as magnitudes.

In May 2009 Daimler bought a stake of about ten percent for roughly fifty million dollars — strategic equity, sold to a holder buying an option on a technology rather than a stream of dividends. In January 2010 the US Department of Energy closed a loan facility of about $465 million under a program built to finance advanced-vehicle manufacturing: a senior secured claim held by a government agency that was willing to lend against plant and equipment because its objective function contained something other than the expected present value of repayment. In June 2010 the company sold about thirteen million shares at seventeen dollars in an initial public offering raising roughly $226 million — public equity, sold to the institutional and retail holders of Chapters 14 and 17, the first American automaker to list since the 1950s. In May 2013 it repaid the government loan early, using the proceeds of a combined equity and convertible-note offering that raised about a billion dollars.

Then the convertibles. Between 2014 and 2019 the company issued several billion dollars of convertible senior notes at coupons between roughly a quarter of one percent and two and a half percent — a claim that pays almost nothing in cash because the holder is being compensated in optionality rather than in coupon, and which is bought not by credit investors but by convertible-arbitrage funds who strip out the equity option and hedge the rest. In August 2017 came the first *straight* bond: about $1.8 billion of senior unsecured notes due 2025, at a coupon a little above five percent, rated deep in the single-B category (B− at Standard & Poor's). That door — the high-yield market — had been shut to the company for a decade and opened only when it had enough revenue for a ratings committee to have something to rate.

Then equity again, and at scale: about $2.7 billion of combined equity and converts in May 2019, and in 2020, through three at-the-market programs, roughly twelve billion dollars of common stock sold directly into a rising market — the last of the three announced in December, into the index-inclusion demand shock that is Chapter 17's opening episode, which is the same transaction seen from the buyers' side. By 2021 the 2025 notes had been retired early. Standard & Poor's raised the rating to BBB in late 2022 and Moody's to Baa3 in early 2023, and the company crossed into investment grade — into the eligible universe of the insurers and index funds of Chapter 10 §10.9.

Read that sequence as a table of claims and holders rather than as a corporate history, and the thesis of this chapter is already visible.

The company never chose a leverage ratio. At each date it faced a *menu of doors*, most of them closed, and it walked through the one that was open. In 2008 the only open door was an insider convertible. In 2010 it was a government agency with a policy mandate and, three months later, a retail-and-institutional equity market in a receptive mood. In 2014 it was convertible arbitrageurs. In 2017 it was high-yield funds. In 2023 it was everyone. What changed was not the firm's optimum. What changed was which holders were permitted, and willing, to hold its paper.

Chapter 23 sets out the theories of what a firm's capital structure *should* be — the trade-off between tax shields and distress costs (§23.2), the pecking order (§23.3), the market-timing account (§23.4). Every one of them is a theory of the issuer's problem. This chapter asks the question the standard sequence never asks: *who is on the other side*, and what does the answer do to the set of financings the firm can actually execute? The financing menu is a sequence of holder matches. It is not the solution to a minimization problem.

***

## 25.1 The Menu, Mapped

Start with the fact that dominates the aggregate and gets the least attention in the textbooks.

**Internal funds come first, and they are most of the total.** In the Financial Accounts of the United States, the nonfinancial corporate sector's gross internal funds — undistributed profits plus the consumption of fixed capital — cover the large majority of its capital expenditure in a typical year, and in some years cover all of it, with the sector a net repurchaser of its own equity. External finance is the margin, not the base. Any account of corporate finance that opens with the choice between debt and equity has already skipped the source that funds most investment, which is retained earnings. Myers (1984) built the pecking order on exactly this observation, and the observation has survived far better than the theory erected on it.

![Figure 25.1: Sources of funds](https://846781005-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F3EupdX99vVBoNySDtmxb%2Fuploads%2Fgit-blob-3fc1fe633ede21ae4221ad464518718cf090849e%2Ffig_25_01_sources_of_funds.png?alt=media)

**Figure 25.1: Sources of funds.** Table 25.1's doors as shares of the US nonfinancial corporate sector's own capital expenditure, five-year averages since 1950, from the Financial Accounts' table F.103. Internal funds — undistributed profits plus the consumption of fixed capital — average 100 percent of capital expenditure over the whole period and cover at least four fifths of it in 97 percent of years. Every external door together is the strip above the blue band, and net equity issuance is the strip below zero: negative in every five-year period since the early 1980s, because the sector retires more equity through repurchases and cash mergers than it issues. Three cautions about what the bands are. They are not a decomposition of the liability side — trade payables, taxes payable and miscellaneous liabilities are left out, because they are not doors a firm chooses. They need not sum to 100, and generally exceed it, because the excess over capital expenditure is the sector's net accumulation of financial assets, which is a use of funds rather than a source. And they are an aggregate, so they describe the sector rather than any firm in it: the small firm of §25.3 has no access to four of the five bands, which is the whole point of Table 25.1's last column. *Source: Financial Accounts of the United States (Z.1), table F.103, through the FRED mirror; flows are seasonally adjusted annual rates, so a year is the mean of its four quarters. Author's calculations.*

The five external doors then differ along four dimensions at once: who the holder is, what control the claim conveys, what it costs, and — the dimension this chapter adds — *who is allowed through*.

**Bank loans.** A bilateral loan from a commercial bank, floating-rate, secured or covenant-heavy, usually short (three to five years, often revolving). The holder is the depository sector of Chapter 19 §19.3, which funds itself with deposits and short wholesale borrowing and therefore wants claims that are senior, secured, short-duration, and floating. The control content is the covenant package: a fixed cash-flow right plus a set of decision rights that transfer to the lender on breach, which is row two of Chapter 21's Table 21.3. The bank monitors, and pricing is relationship-specific rather than posted. Open to almost any firm with collateral or a credit history — this is the widest door in the system and the only one most firms will ever use.

**Syndicated and leveraged loans.** The same instrument scaled up and distributed: an arranging bank underwrites a facility of hundreds of millions or billions and sells participations to a syndicate. Above a certain size and below investment grade the syndicate is no longer banks at all. The *institutional* tranche of a leveraged loan is bought overwhelmingly by collateralized loan obligations, and by loan mutual funds and separate accounts alongside them; industry estimates put CLOs at something on the order of two thirds of the outstanding US institutional leveraged-loan market. The control content thins as the syndicate widens — a hundred anonymous holders cannot renegotiate the way one relationship bank can, which is why covenant packages in this market loosened toward the "covenant-lite" structures that now dominate it. Open to firms large enough to justify an arranger's fee and a rating, which in practice means several hundred million dollars of debt at a minimum.

**Public bonds.** A registered, rated, indenture-governed security sold to institutional buyers and traded thereafter. The holders are the insurers and pension funds of Chapter 16 §16.2, the bond funds of Chapter 17, and — on Chapter 2 Table 2.5's corporate-bond row, a block as large as the fund sector — foreign investors; and Chapter 10 §10.9 established what governs their demand: risk-based capital schedules and index eligibility, not a view about expected returns. Control content is minimal — an indenture is looser than a loan agreement, there is no monitoring, and the holder's remedy is a claim in bankruptcy rather than a seat at a renegotiation. The price is a spread, and the door is guarded by two fixed costs: a rating, and the disclosure apparatus of a public issuer.

**Private credit.** A loan made directly by a fund — a closed-end limited partnership, a business development company, or an evergreen vehicle — to a middle-market borrower, held to maturity rather than syndicated. The fund side is Chapter 18 §18.5's; the intermediation channel is *International Finance*, Chapter 9's. The economically interesting feature for this chapter is that the money behind it is committed capital from insurers and pension plans, which cannot run, and the claim is bilateral, which means the control content is a bank's rather than a bondholder's. Private credit is a bank loan whose holder is not a bank.

**Equity, public and private.** The residual claim, carrying residual control. In private form it is the venture and growth capital of Chapter 18 §18.3, sold in staged rounds of convertible preferred stock whose cash-flow and control rights are negotiated separately (Chapter 21 §21.4). In public form it is the initial and seasoned offerings of Chapter 12, sold to a dispersed holder base that will not monitor and will price the issue at a discount for the adverse selection Myers and Majluf identified. Equity is the most expensive door and the least conditional: it imposes no repayment obligation and no covenant, which is precisely why it is priced as though the issuer knows something.

**Table 25.1: The financing menu, by door**

| Door                                | Typical issuer               | Holder (developed in)                                                        | Control content                                    | Cost, roughly                                         | Who is allowed through                                   |
| ----------------------------------- | ---------------------------- | ---------------------------------------------------------------------------- | -------------------------------------------------- | ----------------------------------------------------- | -------------------------------------------------------- |
| Internal funds                      | All firms, all ages          | The firm's own shareholders, implicitly                                      | None transferred                                   | Opportunity cost of capital; no issuance cost         | Anyone with earnings                                     |
| Bank loan (bilateral)               | Small and mid-size, any age  | Commercial banks (Ch 19 §19.3)                                               | Covenants; monitoring; renegotiable                | Floating, prime or SOFR plus a spread; low fixed cost | Firms with collateral or a credit history                |
| Syndicated / leveraged loan         | Large; leveraged buyouts     | Arranging banks, then CLOs, loan funds (Ch 19 §§19.3-19.4; IF Ch 13)         | Covenants, weakening as the syndicate widens       | Floating, plus arrangement fees                       | Firms above roughly a few hundred million of debt, rated |
| Public bond                         | Large, mature, disclosed     | Insurers, pensions, bond funds, and foreign holders (Chs 10 §10.9, 16, 17)   | Indenture only; no monitoring                      | Fixed coupon plus a spread; high fixed cost           | Rated public filers                                      |
| Private credit                      | Middle market, sponsor-owned | Private credit funds, funded by insurers and pensions (Ch 18 §18.5; IF Ch 9) | Bilateral covenants; bank-like                     | Floating, wider than a syndicated loan                | Firms with a sponsor or predictable cash flow            |
| Private equity (VC, growth, buyout) | Young, or newly private      | Venture and buyout funds and their LPs (Ch 18)                               | Board seats, vetoes, state-contingent control      | Dilution; the highest required return on the menu     | Firms with a growth story or a stable cash flow to lever |
| Public equity                       | Large, disclosed             | Households, funds, foreign investors (Chs 2, 12, 14, 17)                     | Residual control, dispersed and mostly unexercised | Underwriting spread plus the announcement discount    | Firms that clear the listing threshold                   |

*Source: Author's construction. Holder assignments follow Chapter 2's Table 2.5 and the Part IV chapters cited; control content follows Chapter 21's Table 21.3.*

Read the last column down. It is the column the standard treatment does not have, and it is the one that determines what a given firm's menu actually contains. For most firms in most economies, five of the seven rows are closed, and the financing decision is a choice between retained earnings, a bank, and the owner's own money.

One row is missing from the table because it is not a door the firm chooses. Every claim above it is written in the shadow of what happens when the promises are not kept, and the terms on which insolvency is resolved — the automatic stay, the priority ranking, whether management stays in place, how long a plan takes — are part of the price of each rung rather than a separate topic. Chapter 24 §24.1 treats bankruptcy as the last transfer of control in the covenant sequence, and Chapter 10 §10.1 prices the recovery rate that comes out of it. What belongs here is only the direction of the effect: a regime that resolves quickly and predictably makes the secured and covenanted doors cheaper and wider, and a regime that does not pushes borrowers back toward collateral, toward relationships, and toward the owner's own guarantee. Section 25.4 is where that variation becomes a cross-country fact.

***

## 25.2 The Life Cycle

The rows of Table 25.1 open in a rough order as a firm ages, and the order is regular enough to teach as an arc — provided it is taught as a tendency and not a law. Figure 25.2 draws the arc against firm size, with each rung's holder beside it.

**Founder capital and friends and family.** The first claim a firm issues is usually a personal one. Robb and Robinson's work on the Kauffman panel of US startups corrected a widespread misimpression here: new firms rely far more on *formal external debt* — bank loans and, above all, bank credit personally guaranteed by the owner — than the founder-and-friends story suggests. The founder's home equity and personal credit are financing channels for the firm, which is why household balance sheets (Chapter 14) and small-business credit are connected at the root.

**Angels and venture capital.** For the small subset of firms with a plausible tail outcome, the next door is the staged convertible preferred of Chapter 18 §18.3. The claim is designed for the informational problem: capital arrives in rounds conditioned on milestones, so the holder retains an option to abandon, and control migrates to the investor in bad states and back to the founder in good ones.

**Venture debt.** In one sentence, because it is a real rung and a small one: specialist lenders extend term debt alongside a venture round, secured by the firm's assets and by the presence of an equity sponsor who will probably fund the next round, and priced with warrants that give the lender a slice of the upside it cannot get from a coupon.

**Bank relationship lending.** For everyone else — which is to say almost every firm — the first external door is a bank, and what opens it is not collateral alone but *information produced over time*. This is the chapter's most important piece of economics on the small-firm side, and it deserves its own treatment below.

**Syndicated and institutional loans.** As a firm's borrowing outgrows what one bank will hold, the loan is arranged rather than made, and the ultimate holders become the CLOs and loan funds of Table 25.1's third row. The firm has crossed from a relationship to a market without changing instrument.

**High-yield bonds.** With a rating and a public filing history, the firm can sell a fixed-rate security to institutions directly, at a spread that compensates for default risk and illiquidity. The opening episode's 2017 issue is the canonical instance.

**Investment-grade bonds, and then commercial paper.** Above the BBB−/Baa3 line the buyer base changes character: the marginal holder becomes a capital-constrained insurer rather than a high-yield fund, and the spread falls by more than the change in default probability warrants, for the reasons Chapter 10 §10.9 sets out. At the top of the ladder sits the shortest claim of all, commercial paper — unsecured discount notes of up to 270 days, sold to money market funds; the market's mechanics and its 2008 and 2020 episodes are *International Finance*, Chapter 4's.

![Figure 25.2: The financing ladder](https://846781005-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F3EupdX99vVBoNySDtmxb%2Fuploads%2Fgit-blob-a2f844bd6f9836e3818521fd05ad156d78e4a70a%2Ffig_25_02_financing_ladder.png?alt=media)

**Figure 25.2: The financing ladder.** Each door of Table 25.1 drawn over the span of firm size on which it is actually open, with its holder named beside it. The bars carry Table 25.1's last column — who is allowed through — which is the column the standard treatment does not have. Read the left edge rather than the right: the shaded band is where almost every firm in the economy sits, and inside it the menu is three rungs long. The rungs are the chapter's qualitative statements placed on a size axis, not measured thresholds; the axis is logarithmic, so each tick is a tenfold change in the size of the firm, and the apparent width of a rung is a ratio rather than a count of firms. *Source: Author's construction from Table 25.1 and §25.2.*

### The exceptions, which are the point

The arc is a tendency. Three classes of exception establish that.

**Bootstrapped giants.** Firms have reached very large scale without ever selling external equity or issuing a bond, funding growth entirely from retained earnings and supplier credit. Some of the largest privately held companies in the United States and Europe have done so for generations. The arc describes a firm that needs outside capital, and a firm with high margins and low capital intensity may never need it.

**Firms born public, or nearly.** A special purpose acquisition company inverts the sequence entirely: the shell lists first and acquires the operating business afterward, so a company with modest revenue can arrive in the public equity market without an initial public offering at all. Chapter 12 §12.7 works through the structure and the dilution arithmetic that make it contested. A direct listing does something related for a firm that stayed private long enough to need liquidity rather than capital.

**Firms that never leave the bank.** Family-controlled firms with a strong preference for control will decline the doors that dilute it, and remain on bank finance at a scale that would easily support a bond. Section 25.4 shows that whole economies do this.

### Relationship lending and its information economics

Why should a bank be able to lend to a firm that a bond market will not touch? Not because the bank is less risk-averse. Because the bank has produced information that no one else has, and the production technology is *duration of contact*.

Petersen and Rajan (1994) supplied the canonical evidence, using a national survey of US small businesses. Firms with longer and more concentrated banking relationships — more years with the same lender, fewer lenders overall — obtain *more* credit. The effect on the *quantity* of credit available is large and robust; the effect on the *price* is small, and in some specifications absent. That asymmetry is the finding. A relationship does not principally buy a discount. It buys access.

The reason is that the binding constraint on a small opaque borrower is not the interest rate. It is that at any interest rate an outside lender will offer, the pool of borrowers who accept is adversely selected. A lender who has watched the firm's deposit account for six years, seen its receivables, and renegotiated one covenant breach knows which side of that pool this borrower is on. **Transaction lending** — a credit-score-based small-business loan, a leveraged loan sold to a syndicate, a bond — prices publicly verifiable information. **Relationship lending** prices private information the lender manufactured itself.

Both halves of that arrangement have a cost, and the second is the reason this is not simply a happy story. Information the inside bank alone possesses is a source of monopoly power. When the firm comes back for the next loan, the outside market cannot distinguish a good borrower shopping for a better rate from a bad borrower whose bank has declined to renew — so the outside quote is priced for the worse pool, the firm's outside option is poor, and the inside bank can charge above the competitive rate. This is the **informational holdup** of Sharpe's and Rajan's analyses, and it predicts what borrowers do about it: firms that can afford to maintain more than one banking relationship do so, paying in duplicated monitoring for the option of a credible outside quote. Problem 2 computes the trade-off and finds a threshold — below a certain cost of making the firm's quality verifiable to outsiders, the relationship dominates; above it, the borrower prefers the pooled market even though the pooled market funds projects that should not be funded.

***

## 25.3 The Small-Firm / Large-Firm Divide

Every door in Table 25.1 has a fixed cost, and a fixed cost is a wall.

### The financing wall at the public-market door

A public bond issue requires a rating from at least one and usually two agencies, a registration statement, audited financials to a public-company standard, ongoing disclosure, and an underwriting syndicate. Those costs are largely invariant to the size of the issue. Chapter 12 §§12.1 and 12.7 make the same point for the public equity door — the compliance cost of being listed is a fixed cost, it bites hardest on small firms, and the listing gap Doidge, Karolyi and Stulz document is concentrated exactly where the fixed cost is largest relative to the firm.

Box 25.1 works that arithmetic on plausible five-year parameters, and Problem 1 sets it as an exercise. The wall is arithmetic.

> **Box 25.1 — The financing wall, in numbers**
>
> Take a five-year financing with plausible parameters: a syndicated loan at a floating rate 275 basis points over a benchmark, with an arrangement fee of one and a half percent and about four hundred thousand dollars of legal and agency cost; against a public bond at a fixed coupon of 6.25 percent, an underwriting spread of seven eighths of one percent, and two million dollars of fixed issuance cost.
>
> At a principal of fifty million dollars the all-in annual cost of the loan is about 7.67 percent and of the bond about 8.49 percent — the loan wins by more than eighty basis points. At five hundred million the loan costs about 7.40 percent and the bond about 6.66 percent — the bond wins by nearly seventy-five. The crossover is at roughly ninety-three million dollars of principal, and it moves with nothing about the firm's risk.
>
> Two million dollars of fixed cost is four hundred basis points of upfront charge on fifty million and forty basis points on five hundred million.

There is a second wall behind the first, and it is a holder constraint rather than a cost. A bond too small to enter the major corporate indices — the common threshold is a few hundred million dollars of par outstanding — is invisible to the index-tracking and index-benchmarked buyers who constitute much of the market's demand. A firm can pay for a rating and still find that the natural holders are not permitted to buy in size.

### What small firms hold instead: banks and trade credit

On the far side of the wall, the liability structure looks entirely different. Small firms borrow from banks, from their owners, and from their *suppliers*.

**Trade credit** — the accounts payable arising when a supplier ships goods before being paid — is the quietly enormous item. For small firms it is comparable in magnitude to bank lending as a source of short-term finance, a fact Petersen and Rajan documented directly, and accounts payable run to a substantial share of total assets across the US corporate sector. That a supplier extends credit a bank will not is not a puzzle once the information problem is stated properly. The supplier observes the buyer's order flow, knows the resale value of its own goods better than any bank does, and can repossess or cut off shipments more cheaply than a lender can foreclose. It is a lender with a collateral advantage and a monitoring advantage in one product line, and the trade-off is that its credit is short and, when the discount is forgone, extremely expensive: Problem 3 shows that standard terms of two percent off for payment within ten days, net thirty, carry an implicit annualized rate above thirty-seven percent simple and above forty-four percent compounded. A firm financing itself by stretching payables is paying a rate no bank would charge, which is a reliable sign that no bank would lend.

### When bank credit contracts

If small firms are bank-dependent, then a shock to bank balance sheets is a shock to small firms — and, because those firms cannot substitute, to their investment and employment. The evidence on this is qualitatively strong and is developed in §25.5, where the identification is set out. The pattern established across the Great Recession literature is consistent: firms whose pre-crisis lenders were more exposed to the 2008 funding shock obtained less credit afterward, and the shortfall passed through to real decisions, with the effect concentrated in small and mid-size borrowers and largely absent for large firms that could switch to the bond market. Chodorow-Reich's estimates attribute a substantial share — on the order of a third to a half — of the employment decline at small and mid-size firms in his sample to the withdrawal of credit rather than to demand. The asymmetry across firm size is the finding that matters for this chapter: the same aggregate shock is a financing shock for firms on one side of the wall and a nuisance for firms on the other.

### Private credit as the new middle door

The gap between "too large for one bank" and "too small for the bond market" is where the fastest-growing claim of the last decade has landed. Private credit funds now lend directly into that middle, at a scale the International Monetary Fund's *Global Financial Stability Report* has put in the trillions globally when committed-but-undrawn capital is included. The fund economics, the fee structure, and the measurement problems that afflict any non-traded claim are Chapter 18 §18.5's. The intermediation question — whether this is bank disintermediation, regulatory arbitrage, or a genuinely better match between a long-lived illiquid loan and a holder that cannot be forced to sell it — belongs to *International Finance*, Chapter 9.

![Figure 25.3: Loans, bonds, and private credit](https://846781005-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F3EupdX99vVBoNySDtmxb%2Fuploads%2Fgit-blob-8ad823931a472d751f3fb908e2a68dd3e82d2d45%2Ffig_25_03_loans_bonds_and_private_credit.png?alt=media)

**Figure 25.3: Loans, bonds, and private credit.** The three middle doors, in dollars outstanding. Two of them the Financial Accounts publish, and their divergence is the disintermediation this section is about: corporate and municipal bonds outstanding go from 2.2 trillion dollars in 2000 to 8.0 trillion, while bank loans to nonfinancial corporates go from 0.9 to 1.2 — up by a third in a quarter century, against nearly fourfold for the bond market. A middle door had to open somewhere. The third line is that door, and it comes with more caveats than any other series in this book, all of them stated on the figure. It is built from the Commission's own roster of filers holding a `814-` file number — the number assigned on electing regulation as a business development company — joined to the SEC's XBRL frames endpoint. It begins in 2022 because investment companies came under inline XBRL only then, and nothing earlier is spliced onto it. The frames endpoint matches about seven roster filers in ten, so the level is what the matched filers report. And business development companies are the *listed and reporting* slice of private credit, not the market: the drawdown funds and evergreen vehicles that hold most of it do not file this way at all, which is why the paragraph above reaches for the IMF's estimate rather than a number. What the line shows is not the size of private credit. It is the shape and the direction of the only part of it a reader can count for themselves. *Source: Financial Accounts of the United States (Z.1) via the FRED mirror; the SEC's business development company roster joined to the XBRL frames endpoint. Author's calculations.*

What belongs here is the borrower's view, and it is simple. A middle-market firm that in 1995 would have been rationed now has a door, and the door is expensive: private credit spreads run wide of syndicated loan spreads for comparable leverage. The spread is not the whole price. A bilateral lender that holds the loan to maturity takes a concentrated position and prices for the illiquidity, and it protects that position with maintenance covenants, information rights, and a seat at every material decision — so what the borrower gives up on this rung, relative to the covenant-lite syndicated market above it, is control as much as basis points. That is the liquidity-control trade-off in its corporate form: the claims that are easiest for a holder to sell are the ones that convey the least authority over the borrower, and a firm moving up the ladder is buying dispersion in its holder base and selling monitoring. Whether that widening is the price of a genuinely new service or the rent earned by the only lender in the room is Discussion Question 1.

***

## 25.4 Cross-Country Variation, Held to Firm Financing

Firms in different countries finance themselves differently, and the difference is a fact about firms' liabilities, not a verdict on financial systems.

The canonical contrast is Germany against the United States. A German *Mittelstand* manufacturer of several hundred employees is very likely to be financed by a small number of long-standing bank relationships — historically the *Hausbank* arrangement, in which one bank holds the primary relationship, provides most of the credit, and expects to be consulted — and is very unlikely to have issued a public bond or to be listed. An American firm of the same size and age is more likely to have several banking relationships, a higher probability of having tapped a public market at some point, and a shorter average relationship length. Continental European and Japanese firms have historically carried more bank debt and less market debt than American firms of comparable size and industry; American firms reach the bond market earlier and more often.

Two qualifications keep this from becoming a caricature. First, Rajan and Zingales's comparison of the G-7 established that aggregate *leverage* is far more similar across these countries than the bank-versus-market contrast suggests, and that the firm-level correlates of leverage — size, tangibility, profitability, growth opportunities — carry over with the same signs. What differs is the composition of the debt, not principally its quantity. Second, the gap has narrowed since 2000. European non-financial corporate bond issuance grew substantially after the introduction of the euro and again after 2008, when bank deleveraging pushed borrowers toward markets and the European Central Bank's corporate purchase program from 2016 added a large price-insensitive holder to the market. Convergence has been real, partial, and largely one-directional.

The right reading of all this is institutional in the sense of Chapter 21 §21.5. Contract enforceability, creditor rights in bankruptcy, disclosure standards, and the historical structure of the banking sector determine which claims can be written and held cheaply, and therefore which doors exist. Chapter 21 promised that this book would cash in North's argument here; this is the cash.

One amendment to that reading, because "institutions" too easily becomes a synonym for slow-moving legal origin. Some of what determines which claims can be written is not inherited at all but legislated, and legislated recently. Interest-rate ceilings are the oldest instance — usury restrictions predate every other financial regulation, they still bind in much of the world and in parts of US consumer and small-business lending, and their effect is not to lower the price of credit but to remove the borrowers whose risk cannot be priced under the cap from the market entirely, which is the same rationing §25.3 describes arriving by statute instead of by fixed cost. Debt contracts are also renegotiated politically after the fact: moratoria, forced restructurings, mortgage relief acts, and emergency stays are recurrent, they cluster in exactly the states of the world in which creditors most need the contract to hold, and a lender that anticipates them prices accordingly and lends less against pledged assets ex ante. Bolton and Rosenthal's analysis of political intervention in debt contracts is the systematic treatment. The point for this chapter is that the last column of Table 25.1 has a legislature in it, and that a firm's menu can be widened or narrowed by a vote.

*Boundary.* Whether a bank-based or market-based system allocates capital better, how financial structure relates to growth, and the comparative architecture of financial systems are questions about systems rather than about firms. They belong to the *International Finance* and *Macroeconomics* volumes in this series, and Chapter 27 returns to the finance-and-growth literature at the level this book needs. This section stops at the firm's liability side.

***

## 25.5 What Identification Has Established — and What It Cannot

Everything above is a claim about *supply*: that the credit available to a firm depends on the state of its lenders, and not only on the state of the firm. That claim is hard to establish, because lenders and borrowers are matched. A bank whose lending falls may have lost funding, or may lend to industries that just stopped wanting to invest. The observed correlation between bank health and borrower outcomes is consistent with both, and the policy implications are opposite.

### The workhorse design

Khwaja and Mian (2008) solved this with a research design that has since organized an entire literature. Their setting was Pakistan's unanticipated nuclear tests in 1998, which triggered a freeze on foreign-currency deposits. Banks differed in how much of their funding came from those deposits, so the shock hit lenders in different sizes for a reason unrelated to their borrowers. Their data had the feature that makes the design work: many firms borrowed from *more than one bank at the same time*.

That permits the comparison to be made **within a firm**. Write the change in lending from lender $$b$$ to firm $$i$$ over the window as

$$
\Delta L\_{i,b} = \mathrm{FE}\_i + \beta\mathrm{Shock}\_b + \varepsilon \_{i,b},
$$

where $$\mathrm{FE}\_i$$ is a firm fixed effect. The fixed effect absorbs everything about the borrower — its investment opportunities, its demand for credit, its industry, its own solvency — because those things are common across the firm's lenders in the same window. What is left in $$\beta$$ is the differential response of lending across *different lenders to the same borrower*, identified off variation in the lenders' funding shocks alone. Khwaja and Mian estimate an elasticity of roughly 0.6: a bank suffering a one percent larger liquidity shock cut credit to a given firm by about six tenths of a percent more than that firm's other banks did.

The design's second result is the one this chapter's §25.3 needs. Having established that credit supply moved, they ask whether firms undid it by borrowing elsewhere. Large firms substantially did. Small firms did not, and their total borrowing fell close to one for one with the supply shock. The transmission from lender balance sheets to real activity runs through bank-dependent firms, and it runs through them because substitution is unavailable — which is the financing wall of §25.3 observed in a natural experiment.

The design has since been applied to the euro-area sovereign crisis, to the 2008 shock through pre-crisis syndicate composition (Chodorow-Reich's approach), to central-bank lending programs, and to bank mergers. The consistent finding across settings is the one to carry: **credit supply is a real shock for firms that cannot substitute across doors.**

### ★ What the firm fixed effect absorbs, and what it does not

The within-firm estimator is often described as though the fixed effect solves the problem outright. It does not, and the residual issues are where the frontier of this literature sits.

The fixed effect absorbs *level* differences in a firm's credit demand. It does not absorb an *interaction*: if a firm's demand for credit from its stressed bank differs from its demand for credit from its healthy bank — because, say, the stressed bank is the one that also provides its cash management and the firm has begun to move that business — then $$\mathrm{Shock}\_b$$ is correlated with a demand component the fixed effect leaves in the error. Matching is the source of this: firms and banks chose each other, and the same characteristics that determined the match may determine the response.

The estimator also identifies a *relative* effect, not an aggregate one. It tells you how much less a stressed lender lent than a healthy lender to the same firm. It cannot tell you what happened to the total, because the firm fixed effect has differenced out precisely the aggregate the policy question asks about. Recovering the aggregate requires assumptions about general equilibrium — whether healthy banks expanded to fill the gap, and whether the credit they extended would otherwise have gone elsewhere. And the sample is a selected one: only firms with multiple lenders contribute, and multi-bank firms are systematically larger and less opaque than the single-bank firms whose dependence is the object of interest.

### What remains unidentified

Two questions this part of the book cares about are not settled by any of this, and the honest statement is that they are not close.

**The optimal financing structure of a firm is not identified.** The credit-supply literature establishes that *shifting* a firm's financing changes its behavior. It does not establish what the firm's financing should have been. Chapter 23 §23.6 reports the corresponding fact from the capital-structure side: leverage ratios are highly persistent, poorly explained by the observable determinants the theories nominate, and stable across firms in ways that neither the trade-off theory nor the pecking order predicts. A literature that can measure the effect of a change with great precision and cannot characterize the optimum is in an uncomfortable position, and this one is in it.

**The welfare cost of the financing wall is not identified either.** We can measure that small firms cannot substitute, and that their investment and employment respond to their lenders' balance sheets. That is not the same as measuring how much output is lost because a productive firm was rationed. Doing so requires knowing the counterfactual return on the projects that were not funded, and the firms that were rationed do not report them. The quasi-experimental designs identify the effect of credit on the firms that got it; the welfare question is about the firms that did not.

One clause connects this to the book's other half. The credit-supply literature is the *liability-side mirror* of Chapter 20's demand system: there, holders' constrained demand curves for claims are estimated and prices derived from them; here, lenders' constrained supply of one particular claim is shocked and quantities derived from it. Same economics, opposite side of the balance sheet, and — a fair summary of the state of both literatures — much better identified in the quantity dimension than in the price dimension.

***

## 25.6 Who Holds the Claim Behind Each Door, and What Their Constraints Do to Its Price

Return to Table 25.1 and read it in the other direction — not as a firm's set of options, but as a list of what particular holders want to own.

A commercial bank funds itself with deposits and short wholesale borrowing, faces a capital schedule, and is examined on the credit quality of its loan book. It therefore wants a claim that is senior, secured, floating-rate, short, and renegotiable. The bilateral bank loan of row two is not a natural object. It is a claim engineered to fit that balance sheet, which is why it has covenants (the lender wants control before, not after, insolvency) and floats (the lender's funding cost floats).

A life insurer holds long-dated liabilities and is charged capital by rating category, nearly flat within investment grade and stepping up sharply at the boundary (Chapter 10 §10.9). It therefore wants long, fixed-rate, investment-grade paper with no need for monitoring and no covenant to administer. The public bond of row four is that claim. Its minimum size, its rating requirement, and its thin control content are all consequences of who buys it.

A collateralized loan obligation issues rated tranches against a pool and must pass coverage tests. It therefore wants floating-rate, broadly syndicated, ratable loans in a granular pool, and it is a price-insensitive buyer of exactly that when its own liabilities can be sold. The leveraged loan of row three exists in its present form — covenant-lite, floating, minimum tranche sizes, ratings on individual credits — because the CLO is the marginal holder, which is Chapter 24 §24.1's explanation of the cov-lite migration read from the issuer's side. Chapter 19 §19.4 and *International Finance*, Chapter 13 develop that vehicle's own constraints.

A private credit fund holds committed capital that cannot be redeemed. It is therefore the one holder in the ecology that can promise a borrower funding certainty across a downturn, and it charges for that. A venture fund holds a portfolio whose return comes from a tail, so it wants a claim with unlimited upside and a liquidation preference, negotiated round by round.

The generalization is the one Part V has been building toward since Chapter 21 §21.4. **Each door on the menu exists because a specific holder with a specific constraint wants the claim behind it.** The firm does not choose from an abstract menu of financing forms. It chooses from the set of claims that some balance sheet in Part IV is currently able to hold.

Which gives the crossover in the opening episode its meaning. When a firm's claims migrate from relationship holders to market holders — from a bank that monitors to a CLO that does not, from a venture board seat to a dispersed public float — two things change at once, and neither is on any pro forma. Its *governance* changes: the party holding residual control is now a population that cannot easily exercise it, which is Chapter 24 §24.4's subject and Chapter 17 §17.7's common-ownership problem in another guise. And its *crisis exposure* changes: a relationship lender with private information and a long horizon renegotiates; a market holder facing a redemption or a capital charge sells. The companion volume on the 2008 crisis is a long demonstration of what happens when a firm's funding sits with holders who can be forced to stop holding, and Chapter 19 §19.6 is this book's statement of the same mechanism.

In Chapter 1 §1.2's terms, the price a firm pays for credit reports a constraint binding far more often than it reports news: Khwaja and Mian's borrowers learned nothing about their own prospects when their banks' foreign-currency deposits froze, and the small ones, who had no second door, paid all of it.

### Part V, read backward

That is the close of this part, and it is best stated in reverse.

Chapter 21 established that **the firm is a claim factory**: a boundary drawn by transaction costs and control rights, inside which cash flows are assembled and against which securities can be issued. Chapter 22 **calibrated** the factory, asking what its assets are worth and what it should build, and found that the answer runs through a discount rate the asset market sets. Chapter 23 asks how the resulting cash flows are **divided into claims** and paid out, and finds theories that explain less than they should. Chapter 24 asks who **decides** — the covenants, boards, and control rights that determine what happens when the contracts run out.

And this chapter asks what firms actually do, and answers: they finance themselves out of retained earnings, and beyond that they are **rationed by holder access**. The firm's supply of claims meets Part IV's demand for them, and the meeting point is not a price alone. It is a set of doors, most of them closed to most firms, each of them open only because some constrained balance sheet on the other side wants what is behind it.

Part VI takes the two sides together. Chapter 26 asks how firms and financial institutions manage the risks the preceding twenty-five chapters have priced — and finds, in §26.6, that the measure they manage by is itself a price-setting mechanism. Chapter 27 §§27.1 and 27.6 then return to the question Chapter 1 opened with, reading this section's proposition as the supply half of a match: what an ecology of constrained holders means for what asset prices are, and for what they can and cannot tell us about the economy.

***

## Elsewhere in the Series

* **Commercial paper**: the market's mechanics, its dealer and money-fund structure, and the 2008 and 2020 episodes — *International Finance*, Chapter 4. This chapter keeps one line: it is the top rung of the ladder.
* **Private credit as an intermediation channel**: bank disintermediation, regulatory arbitrage, funding structure, and resilience — *International Finance*, Chapter 9. Section 25.3 keeps the borrower's view; Chapter 18 §18.5 keeps the fund economics.
* **Corporate treasury operations and the practice of executing a financing** — *International Finance*, Chapter 10, which owns the treasury-desk view of banking relationships, cash management, and the mechanics of raising capital. This chapter owns the economics of which door opens.
* **Leveraged loan and CLO market structure** — *International Finance*, Chapter 13, for the instrument's trading, clearing, and investor base.
* **Aggregate credit, the credit cycle, and the financial accelerator** — the *Macroeconomics* volume. This chapter stops at the firm.
* **What happens when a firm's funding sits with holders who can be forced to stop holding** — the companion volume on the 2008 crisis.
* **This book**: the master holdings table every row of Table 25.1 draws on — Chapter 2 §2.3. The insurers' capital schedule that defines the investment-grade door, and the pricing of every instrument on the menu — Chapter 10, especially §10.9. Listing costs, the listing decline, and SPACs as an alternative route to the public equity door — Chapter 12 §§12.1, 12.7. Venture capital, private credit's fund side, and staged convertible preferred — Chapter 18 §§18.3, 18.5. Bank balance sheets, dealer capacity, and the constrained-intermediary core — Chapter 19 §§19.3, 19.6. The demand system this chapter mirrors — Chapter 20. Securities as bundles of cash-flow and control rights, and institutions as the determinant of which contracts can be written — Chapter 21 §§21.4, 21.5. The investment this financing pays for — Chapter 22. Capital structure theory, payout, and gap-filling — Chapter 23 §§23.2-23.5, 23.7. Covenants, boards, and control after the crossover — Chapter 24 §§24.1, 24.3-24.4. Risk management and the closing synthesis — Chapters 26 and 27.

***

## Summary

1. **Internal funds are the dominant source of corporate finance in aggregate.** The Financial Accounts show the nonfinancial corporate sector's undistributed profits plus depreciation covering the large majority of its capital expenditure in a typical year. External finance is the margin, and a course that opens with debt versus equity has skipped the base.
2. **The external menu has five doors, and they differ in who the holder is as much as in what they cost** (Table 25.1): bank loans, syndicated and leveraged loans, public bonds, private credit, and equity in private and public form. Each carries a distinct control content, following Chapter 21's Table 21.3.
3. **The doors open in a rough life-cycle order** — founder and formal external debt, angel and venture rounds, venture debt, bank relationships, syndicated loans, high-yield bonds, investment-grade bonds, commercial paper — and the order is a tendency, not a law. Bootstrapped firms, SPAC-listed firms, and family firms that never leave their banks are the standing exceptions.
4. **Relationship lending buys access, not price.** Petersen and Rajan found that longer and more concentrated banking relationships raise the *quantity* of credit available to small firms substantially and lower its price only slightly. The lender's asset is private information it manufactured by watching.
5. **The same private information is a source of holdup.** An outside lender cannot distinguish a good borrower shopping around from a bad borrower whose bank declined to renew, so the inside bank can price above the competitive rate. Firms respond by maintaining multiple relationships, and Problem 2 finds the threshold at which a borrower prefers the pooled market despite its inefficiency.
6. **The financing wall is arithmetic** (Box 25.1). Two million dollars of fixed issuance cost is four hundred basis points on a fifty-million-dollar issue and forty on a five-hundred-million-dollar one, which puts the loan-versus-bond crossover near ninety-three million dollars of principal with nothing about the firm's risk entering the comparison. Index-eligibility minimums put a second wall behind the first.
7. **Trade credit rivals bank lending for small firms and is extremely expensive at the margin.** Terms of two percent off for ten days, net thirty, imply an annualized rate above thirty-seven percent simple and above forty-four percent compounded. A firm stretching payables is paying a rate no bank would charge.
8. **Credit supply is a real shock for firms that cannot substitute.** The Great Recession evidence, and Khwaja and Mian's original design, agree: shocks to lenders pass through to borrowers' investment and employment, concentrated in small and mid-size firms and largely absent for firms that can reach the bond market.
9. **The within-firm estimator is the workhorse and has known limits.** Comparing lending from differently shocked lenders to the same borrower absorbs credit demand, but not demand that interacts with the lender's identity; it identifies a relative effect rather than an aggregate; and it is estimated on the selected sample of multi-lender firms.
10. **Cross-country variation in firm financing is real, and is about composition rather than quantity.** German and Japanese firms have historically carried more bank debt and less market debt than comparable American firms, while Rajan and Zingales found aggregate leverage far more similar across the G-7 than the contrast suggests. The gap has narrowed since 2000.
11. **What is not identified: the optimal structure, and the welfare cost of the wall.** The literature measures the effect of changing a firm's financing with precision and cannot say what the financing should have been, nor what output was lost to the firms that were rationed and left no record of the projects they did not undertake.
12. **The financing menu is the holders' menu.** Each door exists because a specific Part IV balance sheet wants the claim behind it — banks want senior secured floating short paper, insurers want long investment-grade paper, CLOs want ratable floating loans, private credit funds hold committed capital that cannot run, venture funds want a claim on a tail. When a firm's claims cross from relationship holders to market holders, its governance and its crisis exposure change together.

***

## Key Terms

* **Internal funds**: Undistributed profits plus the consumption of fixed capital — the firm's own retained cash flow, and the source of most corporate investment in aggregate
* **Relationship lending**: Credit extended on the basis of private information the lender produced over a continuing relationship, rather than on publicly verifiable data; raises credit availability more than it lowers price, and generates informational holdup as a by-product
* **Transaction lending**: Credit extended on publicly verifiable information — a credit score, an audited statement, a rating — and therefore contestable by any lender with access to the same data
* **Syndicated loan**: A large loan arranged by one or more banks and distributed to a syndicate of participants; above a certain size the ultimate holders are institutions rather than banks
* **Leveraged loan**: A syndicated loan to a below-investment-grade borrower, floating-rate and senior secured, whose institutional tranche is bought principally by collateralized loan obligations and loan funds
* **CLO (collateralized loan obligation)**: A vehicle holding a pool of leveraged loans and issuing rated tranches against it; the marginal holder of the leveraged-loan market, and the reason that market's terms take the form they do
* **Private credit**: Direct lending by a fund — closed-end partnership, business development company, or evergreen vehicle — to a middle-market borrower, held rather than syndicated, funded by committed capital that cannot be redeemed
* **Trade credit**: Financing extended by a supplier that ships before being paid; comparable in scale to bank lending for small firms, and carrying a very high implicit rate when an early-payment discount is forgone
* **Financing wall**: The discontinuity in a firm's menu created by the fixed costs of public-market access — rating, disclosure, underwriting, index-eligibility minimums — which are invariant to issue size and therefore exclude small firms from the doors that price most efficiently at scale
* **Credit supply shock**: A change in the credit a lender is willing to extend that originates in the lender's own balance sheet rather than in the borrower's prospects; identified in practice by comparing differently shocked lenders to the same borrower
* **Bank-dependent firm**: A borrower with no access to the syndicated, bond, or private-credit doors, whose total borrowing therefore moves close to one for one with its lenders' capacity

***

## Readings

### Required

* Myers, S. C. (1984). "The Capital Structure Puzzle." *Journal of Finance* 39(3): 574-592. *The presidential address that states the pecking order and, more usefully for this chapter, states the aggregate fact the ordering is built on: firms finance investment out of retained earnings first, debt next, and equity last and rarely. Read it for the framing of the puzzle rather than for the resolution, which Chapter 23 shows has not held up as well as the framing.*
* Petersen, M. A. and R. G. Rajan (1994). "The Benefits of Lending Relationships: Evidence from Small Business Data." *Journal of Finance* 49(1): 3-37. *The paper behind §25.2. Note carefully which coefficient is large: relationship length and concentration raise the availability of credit substantially and lower its price only a little, and that asymmetry is the whole economics of the arrangement.*

### Recommended

* Degryse, H. and S. Ongena (2005). "Distance, Lending Relationships, and Competition." *Journal of Finance* 60(1): 231-266. *The European continuation of Petersen-Rajan, and the one that puts geography into the relationship. Loan rates fall with the distance between the borrower and competing lenders and rise with the distance to the lender that actually makes the loan, which is informational holdup measured in kilometers.*
* Khwaja, A. I. and A. Mian (2008). "Tracing the Impact of Bank Liquidity Shocks: Evidence from an Emerging Market." *American Economic Review* 98(4): 1413-1442. *The design that organized the credit-supply literature. Read §25.5 alongside it and watch what the firm fixed effect does — and read the second half, on which firms could and could not substitute, because that is the part this chapter needs most.*
* Rajan, R. G. and L. Zingales (1995). "What Do We Know about Capital Structure? Some Evidence from International Data." *Journal of Finance* 50(5): 1421-1460. *The corrective to any simple bank-based-versus-market-based story. Leverage across the G-7 turns out to be more similar than the institutional contrast suggests, and the firm-level correlates carry over; what differs is the composition of the debt.*
* Bolton, P. and H. Rosenthal (2002). "Political Intervention in Debt Contracts." *Journal of Political Economy* 110(5): 1103-1134. *The reading behind §25.4's amendment. Ex-post intervention — moratoria, stays, forced restructurings — can be efficient in aggregate states where enforcement would destroy value, and is anticipated by lenders, so the political process is inside the credit contract rather than outside it.*
* Glaeser, E. L. and J. Scheinkman (1998). "Neither a Borrower Nor a Lender Be: An Economic Analysis of Interest Restrictions and Usury Laws." *Journal of Law and Economics* 41(1): 1-36. *Interest ceilings as social insurance rather than as simple price control, with the historical range of the restriction. The oldest version of §25.3's question about which door opens, and the link to the household credit of Chapter 14.*
* Chodorow-Reich, G. (2014). "The Employment Effects of Credit Market Disruptions: Firm-Level Evidence from the 2008-9 Financial Crisis." *Quarterly Journal of Economics* 129(1): 1-59. *Khwaja-Mian's identification transplanted to the United States, using pre-crisis syndicate composition as the source of lender variation. The estimate of how much of the small-firm employment decline is attributable to credit rather than demand is the number §25.3 cites.*
* Robb, A. M. and D. T. Robinson (2014). "The Capital Structure Decisions of New Firms." *Review of Financial Studies* 27(1): 153-179. *The correction to the founder-and-friends story of startup finance, using the Kauffman panel. New firms use far more formal bank debt — much of it personally guaranteed — than the folk account allows, which is why household and small-business credit are the same subject.*
* Edwards, F. R. and E. R. Morrison (2005). "Derivatives and the Bankruptcy Code: Why the Special Treatment?" *Yale Journal on Regulation* 22(1): 91-122. *Where §25.1's menu ends. Bankruptcy's safe harbors exempt one class of counterparty from the stay that binds everyone else, so the resolution regime is not neutral across the claims on the firm — which is the sharpest available illustration that insolvency law is part of the price of each rung. Read with Chapter 10 §10.1 and Chapter 26.*

***

## Discussion Questions

1. **Does private credit widen the middle door, or only re-price it?** A middle-market firm that was rationed in 1995 can now borrow from a direct-lending fund, at a spread wider than a comparable syndicated loan. Build the case that this is genuine widening: name the friction the fund overcomes that a bank or a syndicate cannot, and say what the borrower is buying with the extra spread. Then build the case that it is re-pricing: the fund is the only lender in the room, its capital is committed and must be deployed, and the spread is rent rather than a service charge. What evidence would separate the two — and would you expect to be able to observe it, given Chapter 18 §18.2's account of how a non-traded claim is marked?
2. **What does the opening episode say about the pecking order?** Myers's ordering is retained earnings, then debt, then equity. The company in the opening episode issued equity repeatedly while unprofitable, then convertibles, then a high-yield bond, then more equity at the top of a rally, and reached investment grade holding almost no debt. State which parts of that sequence the pecking order predicts, which parts it gets backward, and whether the market-timing account of Chapter 23 §23.4 does better. Then consider a defense of the pecking order: that its ordering is conditional on the *availability* of each source, and this firm's availability set was changing throughout. Is that a rescue or an evacuation of the theory's content?
3. **Is the financing wall a market failure?** Section 25.3 shows that a fifty-million-dollar borrower pays more, per dollar, for market access than a five-hundred-million-dollar borrower, because the costs are fixed. Argue first that this is efficient: the fixed costs buy disclosure and verification that real resources produce, and a small firm's alternative — a monitoring bank — is the cheaper technology for its size. Then argue that it is not: name at least two of the fixed costs that are regulatory or conventional rather than technological, and say what would happen to the crossover point in Problem 1 if they were removed. Which argument does the growth of private credit support?
4. **Relationships and competition.** A policy maker proposes to increase competition in small-business banking, on the grounds that concentrated local markets let banks overcharge. Section 25.2's holdup analysis says that competition erodes the inside bank's rent — but a bank that cannot recover its monitoring investment through future rents may not make the investment. State the prediction each view makes for the *quantity* of credit available to young opaque firms in more competitive banking markets, and design the test. What does the growth of credit-score-based transaction lending do to both predictions?
5. **What would settle the welfare question?** Section 25.5 says the welfare cost of the financing wall is not identified because the rationed firms leave no record of the projects they did not undertake. Propose a research design that would get closer. You may use any of: a lottery or threshold in a public credit-guarantee program, the entry of a new lender into a local market, a bank branch closure, or a change in a disclosure threshold. For your design, state precisely what population the estimate applies to, and be explicit about the gap between that population and "all rationed firms."

***

## Problems

**Problem 1 — Two doors, two firm sizes.** A firm is raising capital for five years, with the principal repaid in full at the end. Two doors are available.

*Syndicated term loan:* floating at a benchmark of 4.25 percent plus a spread of 275 basis points, an arrangement fee of 1.5 percent of principal paid at closing, four hundred thousand dollars of one-time legal and agency cost, and an ongoing agency fee of fifty thousand dollars a year. Assume the benchmark stays flat.

*Public bond:* a fixed coupon of 6.25 percent, an underwriting spread of 0.875 percent of principal, two million dollars of one-time issuance cost (legal, accounting, printing, and initial ratings), and five hundred thousand dollars a year of ongoing rating surveillance and incremental reporting cost.

(a) For a principal of fifty million dollars, compute the net proceeds and the annual cash outflow under each door, and then the all-in cost of each financing as the internal rate of return on its actual cash-flow stream. Which door is cheaper, and by how much?

(b) Repeat for a principal of five hundred million dollars. Which door is cheaper now?

(c) Express the one-time costs of each door in basis points of principal at both sizes. Explain in two sentences why the ranking reverses.

(d) Find, to the nearest million, the principal at which the two doors cost the same. State one reason a firm just below that principal might still choose the bond, and one reason a firm well above it might still choose the loan.

(e) Now suppose the firm's bonds would be too small to enter the major corporate indices at a principal below three hundred million dollars. Describe what this does to the comparison in (d), and identify which holder from Table 25.1 has just left the market.

**Problem 2 — The value and the price of a relationship.** A firm has a project requiring an outlay of 100 at date 1, paying off at date 2. With probability 0.8 the firm is of good type and the project pays 125; with probability 0.2 it is of bad type and the project pays 60. All parties are risk-neutral and the riskless rate is zero. Lenders are competitive and break even in expectation.

(a) Suppose type is public information. What face value does a good firm pay, and what does it keep? Is the bad firm's project funded? Compute total expected surplus.

(b) Now suppose no lender can observe type, and any lender that offers terms must lend to every applicant. Compute the break-even face value, the good firm's payoff, the cross-subsidy the good firm pays, and total expected surplus. Verify that the fall in surplus from (a) equals the expected loss on funding the bad project.

(c) Suppose instead that one bank has built a relationship and observes the firm's type. It lends only to good firms. If outside lenders infer that any firm leaving this bank is of bad type, what is the highest face value the inside bank can charge a good firm, and what rent does it capture?

(d) Suppose a good firm can make its type verifiable to outside lenders at a cost $$c$$, borne by the firm. Show that the inside bank charges a face value of $$100 + c$$. For $$c = 1$$, $$c = 8$$ and $$c = 20$$, compute what the good firm keeps, and compare each with its payoff under (b).

(e) Find the threshold value of $$c$$ at which the good firm is indifferent between the relationship and the pooled market of (b). Then state the uncomfortable implication: for $$c$$ above the threshold, which arrangement does the *firm* prefer, which does *society* prefer, and why do the two disagree? Relate $$c$$ to firm size and opacity, and hence to §25.3.

**Problem 3 — What a supplier charges.** A supplier offers terms of "2/10 net 30": a two percent discount for payment within ten days, with the full invoice due in thirty.

(a) Compute the periodic cost of forgoing the discount, as a percentage of the amount actually financed. Be explicit about what the denominator is and why.

(b) Compute the number of such periods in a 365-day year, and hence the simple annualized rate and the effective annual rate.

(c) Repeat for terms of "1/10 net 30" and "2/10 net 60". Which change matters more — halving the discount, or doubling the credit period? Explain the arithmetic in one sentence.

(d) A firm's bank offers a revolving credit line at 9 percent. The firm is stretching payables past day 30 anyway. Compute what it saves per hundred dollars of purchases, per year, by drawing on the line and taking the discount instead. Why might it not do so?

(e) Section 25.3 says a firm financing itself through payables is signaling that no bank would lend to it. State the alternative interpretation — that the supplier has an information or collateral advantage the bank lacks — and identify one observable that would distinguish the two.

**Problem 4 — Designing an identification.** A researcher wants to know whether a contraction in bank credit reduces employment at small firms. She has a panel of firms with their lenders, annual employment, and a measure of each bank's exposure to a funding shock in a given year.

(a) She first regresses firm employment growth on the exposure of the firm's lender, with industry and year controls. State the two distinct reasons this coefficient need not be the causal effect of credit supply.

(b) She restricts the sample to firms borrowing from two or more banks and estimates $$\Delta L\_{i,b} = \mathrm{FE}\_i + \beta\mathrm{Shock}\_b + \varepsilon \_{i,b}$$. State precisely what the firm fixed effect absorbs and what identifying assumption is still required for $$\beta$$ to be a credit-supply elasticity.

(c) Explain why $$\beta$$ from (b), even if perfectly identified, does not answer the question in the preamble. What further step is required, and what new assumption does that step introduce?

(d) Her multi-lender sample turns out to have a median employment of 400 while the full population's median is 25. Discuss what this does to external validity, and say whether the bias in the estimated pass-through to *total* borrowing is likely to be upward or downward, with a reason.

(e) Propose one supplementary design, using a different source of variation, that would address the weakness you identified in (d). State its own principal threat to validity.

**Problem 5 — Reading a liability structure.** Three firms report the following liabilities, in millions.

| Item                        | Firm A | Firm B | Firm C |
| --------------------------- | ------ | ------ | ------ |
| Accounts payable            | 40     | 120    | 900    |
| Bank line drawn             | 25     | 60     | 0      |
| Term loan / syndicated loan | 0      | 250    | 0      |
| Public bonds                | 0      | 0      | 4,000  |
| Commercial paper            | 0      | 0      | 600    |
| Owner-guaranteed debt       | 15     | 0      | 0      |
| Common equity (book)        | 30     | 400    | 6,500  |

(a) For each firm, compute total liabilities, the ratio of debt to debt-plus-book-equity, and the share of interest-bearing debt supplied by banks.

(b) Place each firm on the life-cycle arc of §25.2 and name the doors in Table 25.1 that are plausibly open to it and those that are not. Justify each closure with a specific fixed cost or holder constraint.

(c) Firm A's payables are larger than its bank line. Using Problem 3's arithmetic, state what this implies about its effective cost of short-term finance if it is forgoing early-payment discounts, and what it implies about its bank's willingness to lend.

(d) A shock cuts the lending capacity of the banking sector by 20 percent. Rank the three firms by expected reduction in total borrowing, and state the mechanism for each. Which firm's response is closest to zero, and why?

(e) Firm B is acquired by a buyout fund, which refinances the term loan into a covenant-lite institutional facility. State what changes in the *control* content of the claim, who the new holders are, and what that does to the firm's prospects in a downturn.

***

## Selected Solutions

*Solutions to Problems 1 and 3 follow. Solutions to the remainder are in the instructor materials.*

**Problem 1.**

(a) At a principal of fifty million: the loan's net proceeds are $$50{,}000{,}000 \times (1-0.015) - 400{,}000 = 48{,}850{,}000$$, and its annual outflow is $$50{,}000{,}000 \times 0.07 + 50{,}000 = 3{,}550{,}000$$, with the principal repaid at year five. The internal rate of return on that stream — proceeds in at date 0, the annual payment out at dates 1 through 5, plus principal at date 5 — is **7.67 percent**. The bond's net proceeds are $$50{,}000{,}000 \times (1-0.00875) - 2{,}000{,}000 = 47{,}562{,}500$$, its annual outflow is $$50{,}000{,}000 \times 0.0625 + 500{,}000 = 3{,}625{,}000$$, and its all-in cost is **8.49 percent**. The loan is cheaper by about 82 basis points, despite carrying a *higher* coupon than the bond.

(b) At five hundred million: the loan's proceeds are 492,100,000 against an annual outflow of 35,050,000, an all-in cost of **7.40 percent**. The bond's proceeds are 493,625,000 against an annual outflow of 31,750,000, an all-in cost of **6.66 percent**. The bond is now cheaper by about 74 basis points. The ranking has reversed with no change in the firm's risk, in the instruments, or in either quoted rate.

(c) One-time costs, in basis points of principal: at fifty million, the loan's 230 (150 of arrangement fee plus 80 of legal and agency) against the bond's 488 (87.5 of underwriting spread plus 400 of fixed cost). At five hundred million, the loan's 158 against the bond's 127. The bond's fixed two million dollars is 400 basis points of a fifty-million issue and 40 basis points of a five-hundred-million one; the loan's costs are mostly proportional and so barely move. The bond's *rate* advantage is constant in percentage terms and its *cost* disadvantage shrinks with scale, so the ranking flips once scale is large enough for the rate to dominate.

(d) Solving for equality of the two internal rates of return gives a principal of about **93 million dollars**. A firm just below that might still issue a bond to obtain fixed-rate funding and a longer maturity than any bank will hold, to establish a public credit history before it needs one, or to avoid a covenant package. A firm well above it might still borrow from banks for the flexibility of a revolver, to preserve a relationship it will need in a downturn, or because it wants to avoid the disclosure a registered issue requires.

(e) An index-eligibility minimum of three hundred million converts a smooth cost comparison into a step. Below that threshold, the bond's coupon would have to rise — perhaps materially — because the index-tracking and index-benchmarked buyers that constitute much of the demand cannot hold the issue in size, so the crossover principal moves *up*, and possibly above the threshold itself, leaving a range in which no bond is worth issuing at all. The holder that has left the market is the bond fund and the index-benchmarked separate account of Chapters 10 §10.9 and 17 — a demand constraint, not a cost.

**Problem 3.**

(a) Paying on day 10 costs 98 for every 100 of invoice; paying on day 30 costs 100. The extra 2 buys the use of **98**, not of 100, for the additional twenty days, so the periodic cost is $$2/98 = 0.020408$$, or **2.0408 percent**. The denominator is the amount actually borrowed, which is the discounted price, not the invoice face.

(b) The extra credit period is $$30 - 10 = 20$$ days, so there are $$365/20 = 18.25$$ such periods in a year. The simple annualized rate is $$0.020408 \times 18.25 = 0.3724$$, or **37.24 percent**. The effective annual rate is $$(1.020408)^{18.25} - 1 = 0.4459$$, or **44.59 percent**.

(c) For 1/10 net 30 the periodic cost is $$1/99 = 1.0101$$ percent over the same 18.25 periods: 18.43 percent simple, 20.13 percent effective. For 2/10 net 60 the periodic cost is unchanged at 2.0408 percent but there are only $$365/50 = 7.3$$ periods: 14.90 percent simple, 15.89 percent effective. **Doubling the credit period matters more** than halving the discount, because the discount enters the periodic cost roughly proportionally while the credit period enters the number of periods as a reciprocal — going from twenty days of extra credit to fifty cuts the count by more than half.

(d) Per hundred dollars of purchases, taking the discount and borrowing 98 on the line for twenty days costs $$98 \times 0.09 \times (20/365) = 0.483$$ in interest against a saving of 2.00 on the invoice, a net gain of about 1.52 per hundred per cycle, or roughly $$1.52 \times 18.25 \approx 27.8$$ per hundred of annual purchase volume. It might not do so because the line is capacity-constrained and reserved for another use, because drawing it would breach a covenant or trip a borrowing-base test, because the line does not exist in the size required, or — most simply — because the firm has no line and is stretching payables precisely because it could not obtain one.

(e) The alternative interpretation is that the supplier lends because it is a *better* lender to this borrower: it observes the order flow, values its own goods in resale better than any bank can, and can withhold future shipments as a sanction that is faster and cheaper than foreclosure. One observable that distinguishes the two: whether the firm is *forgoing early-payment discounts* or paying within the discount window. A firm that pays on day 10 and takes the discount is using trade credit as a transaction convenience at close to zero cost; a firm that consistently pays on day 30 or later is paying above thirty-seven percent, which no firm with an unused bank line would do. Rejected-applicant data from a small-business credit survey would corroborate directly.

***

## Data Exercise: Who Lends to Business, and What Happens When They Stop

Parts A through C run entirely on free data. Part D is the licensed extension.

**Part A — The composition of business liabilities (free, Z.1 via FRED).** The Financial Accounts of the United States report the liabilities of nonfinancial *corporate* business (the L.103 and B.103 tables in recent releases) and of nonfinancial *noncorporate* business (L.104) separately. Work from the released tables rather than from memory of FRED mnemonics, which have changed across vintages; confirm from the table's own notes which line is which before dividing.

1. For nonfinancial corporate business, construct the quarterly shares of total liabilities accounted for by (i) debt securities, (ii) loans (depository institution loans plus other loans and advances), and (iii) trade payables, from the earliest available date to the present. Plot the three shares on one axis and mark 1990, 2008, and 2020.
2. Repeat for nonfinancial noncorporate business. Put the two loan-share series on a single chart. Write one paragraph on what the difference between the two panels shows about §25.3's divide, and be explicit about which firms fall into which sector and why that is an imperfect proxy for firm size.
3. Retrieve the corporate sector's gross internal funds (undistributed profits plus consumption of fixed capital) and its capital expenditure. Plot internal funds as a share of capital expenditure. Report the mean, the minimum, and the number of years in which the ratio exceeded one. Compare your answer with Summary point 1 and state whether the data support it as written.

**Part B — The credit-supply gauge (free, SLOOS via FRED).** The Federal Reserve's Senior Loan Officer Opinion Survey reports, quarterly, the net percentage of domestic banks tightening standards on commercial and industrial loans, separately for large and middle-market firms and for small firms.

4. Plot both series from the survey's start. Mark the four episodes in which the net tightening percentage exceeded fifty.
5. Plot the small-firm series against the large-firm series in a scatter and report the correlation. Then plot the *difference* between them over time. In which episodes does the gap widen, and what does §25.3 predict about which firms bear the adjustment?
6. Overlay the small-firm tightening series on the loan share you built in Part A, question 2. Lag the survey by one, two, and four quarters and report the correlation at each lag. Write two sentences on why a correlation here is not an identification, referring to §25.5's first paragraph.

**Part C — The borrower's side (free, Fed Small Business Credit Survey).** The Federal Reserve Banks publish the *Small Business Credit Survey Report on Employer Firms* annually, free, with detailed tables on application rates, approval rates by lender type, and reasons for not applying.

7. Assemble, across the available years, the share of applicant firms that were fully approved, by lender type (large bank, small bank, credit union, online lender, and — where reported — finance company or CDFI). Plot the series and identify which lender type's approval rate is most cyclical.
8. Extract the reported *reasons for not applying* and separate discouraged borrowers ("did not think they would be approved") from firms with sufficient funding. Plot the discouraged share over time. Explain in one paragraph why the discouraged share is exactly the quantity §25.5 says is not identified, and what the survey does and does not tell you about it.
9. Combine questions 7 and 8 with your Part B chart. Write a one-page account of the transmission from bank balance sheets to small-firm credit, being explicit at each step about whether the evidence you are citing is descriptive or causal.

**Part D ★ — Lender shocks and firm outcomes (WRDS).** With Compustat, DealScan, and (if available) the FDIC Call Reports:

10. Build a firm-lender panel from DealScan's syndicated loan facilities, identifying each borrower's lead arranger and the set of lenders in each syndicate over a window spanning 2006-2010. Merge to Compustat on the standard linking table and document how many borrowers you lose and whether the losses are systematic in size.
11. Construct a lender-level shock — the lead arranger's exposure to the 2007-08 funding disruption, using whatever measure your data support, and state its weakness explicitly. Estimate the within-firm specification $$\Delta L\_{i,b} = \mathrm{FE}\_i + \beta\mathrm{Shock}\_b + \varepsilon \_{i,b}$$ on the multi-lender subsample and report $$\beta$$ with standard errors clustered by lender.
12. Then estimate the firm-level regression of employment growth (or capital expenditure growth) on the firm's *weighted average* lender shock, with and without industry-by-year fixed effects. Report both coefficients. Write a paragraph on the relation between what you estimated in question 11 and what you estimated in question 12: which one requires the general-equilibrium assumption §25.5 names, and what would have to be true for the second to be interpreted as an aggregate effect? Compare your multi-lender subsample's size distribution with the full Compustat distribution and state the population your estimate describes.