> 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-vi-risk-and-the-real-economy/chapter_27_synthesis.md). # Chapter 27: Claims, Prices, and Holders — Finance and the Real Economy *Part VI: Risk and the Real Economy — Financial Economics: Claims, Prices, and Holders* *** ## Opening Episode: Four Kinds of Efficiency James Tobin delivered the Fred Hirsch Memorial Lecture in New York in May 1984, three years after the Nobel and four decades into a career spent insisting that the financial system was a part of the economy rather than a mirror held up to it. The lecture was published that summer in the *Lloyds Bank Review* under a title that sounds like a survey and is not: "On the efficiency of the financial system." Tobin's opening move was lexical. The profession had one word, *efficient*, doing four jobs, and almost never said which. He separated them. **Information-arbitrage efficiency** is the property that no trader can systematically profit from publicly available information — the weak and semi-strong forms of Chapter 7 §7.1. Tobin granted that markets approximate it, and observed that this is the cheapest of the four to satisfy and the least valuable to have. **Fundamental-valuation efficiency** is the stronger property that a claim's price equals the rationally discounted value of the payments it will actually make. Robert Shiller's volatility tests — prices move far more than the dividend streams they discount — were two years old when Tobin spoke, and Tobin was already doubtful; Chapter 7 §7.4 is where the doubt has since been litigated. **Full-insurance efficiency** is the Arrow-Debreu standard: a complete set of claims contingent on every state of the world, so that any risk anyone bears is a risk someone chose. Chapter 3 §3.4.7 says why this fails — markets are incomplete, and most states carry no traded claim — and Chapter 8 says how far the derivative markets go toward repairing it. Not far enough to matter for most households, was Tobin's verdict, and it still is. **Functional efficiency** is the one he cared about, and it is not a property of prices at all. It asks whether the system performs its economic functions — pooling saving, allocating capital to its best uses, spreading risk, clearing payments — at a reasonable cost in real resources. It is the only one of the four that can be measured in people and buildings and salaries rather than in return regressions. Then the sentence the lecture is remembered for: "I confess to an uneasy Physiocratic suspicion, perhaps unbecoming in an academic, that we are throwing more and more of our resources, including the cream of our youth, into financial activities remote from the production of goods and services, into activities that generate high private rewards disproportionate to their social productivity." The financial sector had been growing considerably faster than the economy it intermediated. Trading volume was growing faster still. Tobin's worry was that the growth was buying more of the first kind of efficiency, which was already close to exhausted, and none of the fourth. He could not settle it. That is the part worth dwelling on. Tobin could see the finance and insurance sector's value added rising in the national accounts, because the Bureau of Economic Analysis published it. What he could not see — because nobody had built it — was who held which claims, at what elasticity, under what constraints, and what happened to prices when those constraints moved. The Financial Accounts existed; the demand system did not. He posed a question about the match between issuers and holders at a moment when neither side of the match was measurable at the level the question required. ![Figure 27.2: The size of finance](https://846781005-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F3EupdX99vVBoNySDtmxb%2Fuploads%2Fgit-blob-e6ed81534deb764bd89918bc78652d30a6b437c7%2Ffig_27_02_the_size_of_finance.png?alt=media) **Figure 27.2: The size of finance.** The series Tobin could see, carried forward forty years. Finance and insurance value added runs 2.8 percent of GDP in 1950 and 7.9 in 2025 — the sector grew roughly three times faster than the economy it intermediates — and at the lecture it stood at 5.4, already double its postwar level. That doubling is what the Physiocratic suspicion was about, and the four decades since have added as much again. The figure is drawn in two colours because it is two series and not one: the Bureau's industry accounts moved from the Standard Industrial Classification to NAICS in 1997, and in that single overlapping year the same sector measures 7.5 on the old basis and 6.7 on the new. Nothing reconciles those; the industries were rebuilt rather than renamed, and a spliced line would invent a fall that did not happen. Two cautions on reading the level. Value added measures what the sector was *paid*, not what it produced for anyone else, so the series is the numerator of Tobin's question and not its answer. And a rising share is consistent with the sector doing more, doing the same for more, or doing something else entirely — which is why the rest of this chapter is about holders and constraints rather than about size. *Source: Bureau of Economic Analysis: historical SIC-basis value added by industry (1947-1997) and the GDPbyIndustry API on a NAICS basis (1997-present). Author's calculations.* Forty years later a good deal of it is. The rest of this chapter is what this book's machinery has to say to Tobin's question, and to the larger one behind it: what an ecology of constrained holders means for what asset prices *are*, and for what they can and cannot tell us about the economy that issues them. *** ## 27.1 The Question, Answered Chapter 1 opened with three questions treated as one: who holds which claims, how are those claims priced, and what do holders' constraints do to prices. Twenty-five chapters later the answer can be assembled. It comes in four parts, each of which has a canonical chapter, and none of which is a cross-reference standing in for an argument; Figure 27.3 is the four of them drawn as the circuit they make. The standard being applied is older than Tobin's lecture: Arrow's account of insurance as a device for moving risk to those best placed to bear it is the compact statement of the functional test, and Chapter 3's state prices — the price today of a dollar in one state — are that account written as a price system. **Prices are** $$E\[mx]$$**, and the** $$m$$ **belongs to somebody.** Chapter 3 established the pricing equation as an implication of no arbitrage, which is why it survives everything the book subsequently does to it: $$p = E\[mx]$$ holds in economies with constrained investors, segmented markets, and vertical demand curves exactly as it holds in a frictionless one. What does not survive is the standard identification of $$m$$ with the marginal utility of a representative household consuming the national accounts' consumption series. Chapter 3 §3.7 stated the objection; Chapter 5 §5.7 measured it, showing that equity risk is borne by a wealthy minority whose consumption is roughly twice as volatile as the aggregate and considerably more correlated with returns, which shrinks the equity premium puzzle by a large factor without closing it; Chapter 19 §19.5 replaced the household altogether for the claims households do not hold, putting the marginal value of intermediary equity capital into the discount factor and finding that it prices credit, sovereigns, options, commodities, and foreign exchange at a common price of risk; and Chapter 20 §20.6 stopped choosing. Whose $$m$$? Everyone's, weighted by wealth and by constraint — and the weights are measurable. That last clause is the book's methodological claim in one line. It converts the identity of the marginal investor from a modeling convenience into an estimate. **Constraints move prices, and the mechanism is balance-sheet mechanics rather than sentiment.** Chapter 16 §16.5 is the canonical statement: when a leverage ratio, a capital charge, a haircut, a redemption queue, or a mandate contracts a holder's capacity to hold a claim, the price falls by whatever it takes to move the claim to someone whose capacity has not contracted, and the claim's promised cash flows need not change at all. The 2007-2008 haircut schedule of Table 16.8 is the cleanest available exhibit: a rise from three percent to fifty percent on AAA asset-backed paper is a ninety-four percent reduction in the quantity the levered sector can hold, imposed inside eighteen months, with no revision to any cash flow. Chapter 15 §15.5 supplies the complement — why an arbitrageur *with* capital may still decline a trade that can widen before it converges — and Chapter 11 §11.6 shows the same capacity at trade frequency, priced as a spread and a depth. **Issuers answer the holders.** Part V is titled "Firms as Issuers of Claims" because the arrow runs both ways. Chapter 23 §23.7 gives three mechanisms with evidence attached: gap-filling, in which corporations shorten their debt maturity when the Treasury lengthens and lengthen when it shortens, with the effect concentrated among exactly the large unconstrained issuers who can act on it; safe-asset supply, in which a convenience yield on money-like claims (Chapter 9 §9.4) is an invitation to manufacture near-money that the private sector accepts; and credit-supply effects on leverage, in which the reaching-for-yield of capital-charged insurers (Chapter 10 §10.9) compresses spreads and firms lever into the compression. Chapter 25 §25.6 states the general form: the financing menu is the holders' menu, and each door on it exists because some specific balance sheet, under some specific constraint, wants the claim behind it. **And the measurement program exists.** This is the part Tobin did not have. Chapter 2 §2.3's master holdings table is the left-hand side of a demand system — who holds how much of what — and the Financial Accounts, 13F filings, statutory insurance schedules, and the Distributional Financial Accounts supply it quarterly and for free. Chapter 20 is the estimation that consumes it. The result is not a settled parameter set; Chapter 20 §20.5 is candid about which findings are established (demand curves for individual claims slope down; holder identity and mandate affect prices) and which are contested (the exclusion restriction on the investment-universe instrument; the exact magnitude of the aggregate multiplier). But a contested estimate is a different intellectual object from an untestable assumption, and moving the marginal-investor question from the second category to the first is what Part IV was for. ![Figure 27.3: The book in one picture](https://846781005-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F3EupdX99vVBoNySDtmxb%2Fuploads%2Fgit-blob-b6c368bf2241dd94e0389fde2ed6f50b85eecb97%2Ffig_27_03_the_book_in_one_picture.png?alt=media) **Figure 27.3: The book in one picture.** The six Parts as a circuit rather than a sequence. The grey arrows are the reading order: what a claim is, what a price must satisfy, the claims themselves market by market, who holds them, who issues them, and what the whole thing does to output. The two red arrows are what this book adds to that sequence and are the reason it is one volume rather than two. Part IV runs back into Part II because holders' constraints belong inside the discount factor rather than beside it (§§5.7, 19.5, 20.6), and Part V runs back into Part III because the claims that get manufactured are the ones some specific balance sheet is short of (§§23.7, 25.6). Chapter 1's question sits at the centre because every arc is an answer to part of it, and because the answer — everyone's, weighted by wealth and by constraint — is a measurement claim rather than a preference claim. *Source: Author's construction.* Put the four together and the book's answer to its own question is this. **Asset prices are the equilibrium of a market in which the buyers are identifiable institutions with balance sheets, mandates, and regulators, and the sellers are firms and governments who design claims to fit those balance sheets.** Cash flows and risk enter that equilibrium as characteristics, and they matter; they are not the only thing that does. *** ## 27.2 Does Finance Cause Growth? The oldest version of Tobin's question is whether an economy with a bigger financial system grows faster, and the modern literature on it starts with a cross-country regression. Robert King and Ross Levine's design, in one sentence: take about eighty countries over 1960 to 1989, measure financial depth at the *start* of the period — liquid liabilities as a share of GDP, the share of credit allocated by banks rather than the central bank, the share going to private firms rather than to government — and ask whether the initial value predicts subsequent growth. It does, strongly, and it predicts capital accumulation and the productivity residual separately, which is a partial answer to the objection that finance merely finances a boom already under way. Using the initial value rules out the crudest reverse causation. It does not rule out a third factor — property rights, legal origin, human capital, a competent state — that raises both financial depth and growth, and King and Levine said so. Raghuram Rajan and Luigi Zingales closed most of that gap with a design that is now standard. Industries differ in how much external finance they intrinsically require, for technological reasons: pharmaceutical research consumes cash for years before revenue, a restaurant does not. Measure each industry's external dependence in the United States, where financial frictions are plausibly smallest, and treat it as a technological characteristic. Then test the interaction: if financial development matters, industries with high external dependence should grow disproportionately faster in countries with deeper financial systems. Country effects absorb everything about the country, industry effects everything about the industry, and the identification rests entirely on the interaction. They find it, and the point estimate is economically large. The design is a difference-in-differences, and its remaining weak point is the assumption that external dependence is technological rather than itself a response to the financial system a firm grew up in. Then the literature reversed on itself. A body of work associated with Jean-Louis Arcand, Enrico Berkes and Ugo Panizza, and with parallel results from the Bank for International Settlements, re-estimated the relationship with a quadratic term and found it non-monotonic: positive at low levels of private credit to GDP, flat, and then negative above a threshold that the various estimates place somewhere near one hundred percent of GDP. The "too much finance" result should be held at the confidence its authors hold it. The threshold is estimated with wide bands, the countries above it are a small and unusual group, and the sample is dominated by the advanced-economy credit boom that ended in 2008, so the finding partly restates the fact that the largest financial systems had the largest crisis. What survives all of that is the negative claim, which is enough: the mapping from financial size to growth is not monotone, and the policy inference "more depth is better" does not follow from King and Levine at the depths rich countries now have. ![Figure 27.1: Too much finance](https://846781005-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F3EupdX99vVBoNySDtmxb%2Fuploads%2Fgit-blob-ec0b6121cf46c143a9ab3f046466d1a20696a294%2Ffig_27_01_too_much_finance.png?alt=media) **Figure 27.1: Too much finance.** The result re-estimated rather than quoted: 1,842 non-overlapping five-year country cells since 1960, private credit to GDP against real GDP per capita growth, both averaged within the cell, from the World Bank's own data. The fitted quadratic is concave and turns at 94 percent of GDP, which is where the published threshold estimates cluster. Two things the figure is built to show rather than hide. The cells above the threshold are drawn in red, and there are only 185 of them out of 1,842 — the countries the result is about are a small and unusual group, and the downward arm of the curve is fitted on a tenth of the sample. And because the headline sample overlaps the advanced-economy credit boom that ended in 2008, the same fit is run on cells that end before 2005; it turns at 101 rather than 94. That the two agree is worth stating, because it is the caveat the paragraph above raises and this particular cut does not sustain it: the non-monotonicity is not an artifact of the boom. What the figure cannot do is identify anything. It is a conditional mean through a cloud with no controls, no instrument, and no attempt at causality, and the negative claim in the text — that the mapping from financial size to growth is not monotone — is all it can carry. *Source: World Bank open data, private credit by deposit money banks and other financial institutions to GDP (FS.AST.PRVT.GD.ZS) and real GDP per capita growth (NY.GDP.PCAP.KD.ZG). Author's calculations.* **Table 27.1: The finance-and-growth evidence, and what each design establishes** | Finding | Design | Status | | ------------------------------------------------------------------------------------ | --------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------- | | Initial financial depth predicts subsequent growth (King-Levine) | Cross-country, initial values, \~80 countries, 1960-89 | **Established as a correlation** robust to lagging; vulnerable to an omitted institutional third factor | | Externally dependent industries grow faster where finance is deeper (Rajan-Zingales) | Industry × country interaction, both main effects absorbed | **The identification improvement**; rests on external dependence being technological | | The relationship turns negative above roughly 100% private credit to GDP | Quadratic in cross-country panels | **Contested**; wide bands, few countries above the threshold, sample overlaps the pre-2008 boom | | The unit cost of financial intermediation has not fallen in a century (Philippon) | Ratio of financial income to intermediated assets, US, \~1880-present | **Established as a measurement**; its interpretation as inefficiency is disputed | *Source: Author's assessment of the literature discussed in this section.* Now connect it back to the lecture. The growth evidence is evidence about **functional efficiency in the claim-issuing and rationing functions** — the subject of Chapters 21 through 25, and specifically of Chapter 25 §25.3's financing wall, where firms without a rating, a shelf registration, or a banking relationship simply do not get funded. It is not evidence about the trading functions of Chapters 11 through 13 — market making, the spread, the turnover of claims already issued. Credit to GDP is a stock of claims outstanding; it says nothing about turnover, and the two have moved very differently. Thomas Philippon's measurement is the sharpest statement of the gap: the unit cost of financial intermediation in the United States — the sector's income divided by the quantity of assets it intermediates — has hovered near two percent for well over a century, through the arrival of the telephone, the computer, and the electronic exchange. Whatever the information technology bought, it did not buy a cheaper conversion of household saving into corporate investment. Tobin's suspicion was that the resources were going into information-arbitrage efficiency, which is rival, zero-sum in large part, and already nearly exhausted. Philippon's series is consistent with the suspicion without proving it. What this book's ecology adds is a distinction the growth literature does not draw. **A financial system grows when holders' demand for claims grows, and that is not the same event as firms' investment opportunities improving.** Chapter 2 §2.4's fifty years of reallocation is mostly a story about the liability side of households — retirement systems, defaults, the shift from defined benefit to defined contribution — pushing a rising stock of saving into intermediaries that must hold something. Chapter 20 §20.3's reading is that when those holders' demand curves are steep, the arriving money mostly moves prices rather than quantities, and Chapter 23 §23.7's is that the claims which do get manufactured are the ones the arriving holders are short of, not the ones the highest-return projects need. A credit-to-GDP ratio that rises for that reason is a measure of holder demand wearing the clothes of a measure of capital allocation. Any regression that treats it as the second when it is partly the first will attribute to finance's productive function an expansion that belongs to its portfolio function — which is one reason the same variable can predict growth in a poor country and predict a crisis in a rich one. > **Box 27.1 — How the national accounts measure a bank** > > Every claim that finance is some percentage of GDP rests on a measurement convention, and the convention is unusual enough to state. Most of a bank's output is not billed to anyone. > > A firm that sells software has revenue that is a price times a quantity. A bank makes some of its money that way — fees for advice, for custody, for arranging an issue — but the larger part historically came from the margin between the rate it pays depositors and the rate it charges borrowers. Nobody is invoiced for that margin, so it cannot be measured as a sale. The national accounts impute it, under the heading **financial intermediation services indirectly measured**, or FISIM. > > The method is a reference rate. The accounts choose a rate meant to represent the pure cost of funds with no service content, and the imputed output is the difference between what borrowers pay and that rate, plus the difference between that rate and what depositors receive, applied to the relevant balances. The imputation is then allocated to the users of the services, and it enters value added. > > Three consequences follow, and the chapter needs all three. A widening interest margin registers as more financial production, whether or not any additional service was performed. The choice of reference rate matters, and it is a choice: different plausible rates produce visibly different levels of measured financial output. And because the sector's output is inferred from a spread rather than observed as a transaction, the sector's measured share of GDP is not directly comparable to the measured share of an industry that sells things. > > One further boundary is worth naming, because it moves the number. In the North American Industry Classification System finance and insurance is sector 52, and real estate, rental and leasing is 53. Statements about "the size of finance" that include or exclude real estate are describing different objects, and the two are not close. > > This is the construction behind Table 27.1's Philippon row and the reason his unit-cost measure — the cost of intermediation per dollar intermediated — is a more informative object than the sector's share of value added. *** ## 27.3 The Financial Accelerator, in One Statement The mechanism that connects all of this to the business cycle can be written on a line. Because lenders cannot observe or enforce everything, a borrower who funds a project externally pays more than the opportunity cost of internal funds, and the wedge — the **external finance premium** — falls as the borrower's net worth rises, because net worth is what the borrower can pledge: $$ \mathrm{EFP}\_t = \Phi(\text{borrower net worth}\_t), \qquad \Phi' < 0 . $$ That single sign is the accelerator. A shock that lowers net worth raises the premium, which lowers investment, which lowers cash flow and asset values, which lowers net worth again. Bernanke and Gertler built the agency-cost version; Kiyotaki and Moore built the collateral version, in which the constraint is a borrowing limit tied to the liquidation value of an asset that the borrowing itself helps to price, and showed that small temporary shocks can generate large persistent output movements through it. Both versions turn on the same four steps, none of which needs the model: net worth falls; the collateral a borrower can pledge is worth less, so the constraint it secures tightens; credit and investment contract; and the pledged assets are marked down again, closing the loop. That is how a shock to a balance sheet becomes a shock to output. **The machinery belongs to the companion volume.** *Institutionalist Macroeconomics* develops the accelerator formally in its Chapters 12 and 19, with the general-equilibrium closure, the calibration, and the quantitative business-cycle results; this book does not reproduce them, and a reader who wants the model rather than the mechanism should go there. What this book contributes to that story is the other leg. The accelerator as usually taught runs through the *borrower's* balance sheet — the firm, the household, the entrepreneur. Chapters 16 and 19 establish that it also runs through the *lender's*, and that this is where it binds in the episodes we actually observe. Chapter 16 §16.5's margin spiral is an accelerator in which the constrained agent is the levered holder rather than the borrower, and the collateral being repriced is the holder's inventory. Chapter 19's March 2020 episode is the demonstration: Treasury yields *rose* during the largest flight to safety since 2008, because dealer balance sheets were full and the premium being repriced was the price of intermediation itself, not the price of anybody's project. The 2022 gilt episode of Chapter 16 §16.2 is the same shape with pension funds in the dealer's seat, where a rise in yields triggered collateral calls on liability-driven overlays, forced sales of the very gilts whose price was falling, and required a central bank to stand between the sellers and the market. In both cases nothing happened to the net worth of any borrower. The supply of balance sheet contracted, and the premium the accelerator is built around is a price that constrained holders charge. *** ## 27.4 Asset Prices as Transmission If financial conditions matter for output, the transmission runs through prices, and this book has been describing the pipes. Three channels, each read through its owning chapters. **The cost-of-capital channel.** A change in the policy rate does not reach investment directly; it reaches it by repricing a chain of claims. It moves the short end of Chapter 9 §9.2's curve and, through expectations and the term premium demanded for holding duration, the long end that actually matters for long-lived projects. It reprices Chapter 13's mortgage market, where the convexity hedging of §13.5 — mortgage holders sell duration as rates rise and buy it as they fall — amplifies the rate move before any household refinances. It changes the hurdle rate in Chapter 22 §22.1's capital budgeting arithmetic, where a firm's weighted average cost of capital is assembled from exactly the asset-market prices Parts II and III derive — and where Chapter 22's evidence that firms use sticky, internally administered hurdle rates is a reason to expect the transmission to be lumpy rather than smooth. And it opens or closes the issuance windows of Chapter 23 §23.4, which is the margin firms respond to fastest: the maturity, the size, and the timing of an offering move long before the capital expenditure does. **The wealth channel.** A rise in asset prices raises household net worth and, at some marginal propensity, consumption. Chapter 14 §14.6 supplies the qualification that makes this a claims-and-holders statement rather than an aggregate one: the top decile of the wealth distribution holds on the order of eighty-five percent of directly held corporate equity and the bottom half about one percent. A ten percent equity rally distributes almost entirely to households whose consumption is least responsive to wealth, while the housing wealth of Chapter 14 §14.4 is distributed very differently and is the collateral against which the middle of the distribution actually borrows. The size of the wealth channel is therefore a fact about the distribution of holdings, not a coefficient. **The collateral channel.** The margin spiral of Chapter 16 §16.5 again — falling prices raise haircuts, higher haircuts force sales, forced sales lower prices — now read as macroeconomics. A price decline is a reduction in the borrowing capacity of everyone who had pledged the asset, and the reduction is mechanical, immediate, and independent of anyone's forecast. This is the channel through which an asset-market event becomes a funding event within days, and it is why the 2008 volume — the ecology's fullest exhibit — is a story about haircuts and repo as much as about mortgages. **Inelastic markets revisited.** Now put Chapter 20 §20.3's estimate on top of the three channels and the boundary between "financial" and "real" shocks stops being well defined. If the aggregate price multiplier is anywhere near five — dollars of market value created per dollar of flow, the reciprocal of an aggregate demand elasticity near $$0.2$$ — then a portfolio reallocation carrying no information about any firm's cash flows is a first-order determinant of the level of the market. Take the arithmetic literally. One hundred billion dollars of flow into equities creates something like five hundred billion dollars of market value; at a marginal propensity to consume out of financial wealth in the three-to-five-cent range, that is fifteen to twenty-five billion dollars of consumption, or about a tenth of a percent of US consumption spending, from an event with no news in it. Gabaix and Koijen's own statement is stronger: they argue that a large share of the observed variation in aggregate equity prices is flow-driven, which would make the composition and timing of institutional demand a macroeconomic variable in its own right. Chapter 20 §20.5 is where the contest over that claim is recorded — the instrument, the magnitude, and Haddad, Huebner and Loualiche's argument that a shrinking active sector trades more aggressively and partly offsets the mechanical effect of rising passive ownership. Take the middle of the range rather than the headline and the conclusion still holds in the direction that matters here: **flows are a macro force, and a shock that originates in who is holding rather than in what is being produced propagates through the same three channels as any other.** *** ## 27.5 What Asset Prices Can and Cannot Tell Us The standard reason to read prices for macroeconomic information is Chapter 7's: a price aggregates the beliefs of everyone trading, weighted by their willingness to back them, and therefore embeds information no survey collects. Chapter 7 §7.2's Grossman-Stiglitz argument even explains why the aggregation is good but never perfect — if the price revealed everything, no one would pay to learn anything, and there would be nothing to reveal. Everything since Chapter 7 has been a qualification. Chapter 1 §1.2 set the instrument out on a single oil-price move: four readings of any price change, only one of which is news. Every chapter since has run it once, at its holder section. Here it is run at full depth, through the breakeven, the basis, and the VIX. The four readings, and the section that owns each: * **News.** The cash flows or the risks changed, and the price moved because the discounted value did. This is Chapter 7 §7.3's event-study case, and in a narrow window around an identified announcement it is usually the right reading. * **A constraint binding.** A holder's capacity contracted and the claim moved to a less willing owner (Chapter 16 §16.5). Nothing about the economy changed; something about the balance sheet holding the claim did. * **A flow.** Money arrived or left, for reasons of payroll dates, index rules, or mandate, and met a steep demand curve (Chapter 20 §20.3). * **A model becoming convention.** The price reflects what the standard model says the price should be, because everyone is using the standard model — Chapter 7 §7.6's performativity, where a model everyone uses makes prices conform to it; Chapter 8 §8.5's post-1987 smile, where crash insurance has been priced above Black-Scholes ever since; and Box 26.1's treatment of value at risk as a convention that makes institutions sell in unison because they measure risk in unison. The four are not always separable, and pretending otherwise is the characteristic error of reading markets for macroeconomic content. The practical version: when a policymaker reads a price, the price is a holder artifact as much as a forecast, and the good practitioners already know it. The ten-year breakeven inflation rate is the leading example. Chapter 9 §9.3 decomposes it into expected inflation, an inflation risk premium, and a TIPS liquidity premium, and notes that the last term widens exactly when liquidity is scarce — which is why the breakeven collapsed toward zero in the autumn of 2008 and no forecaster read it as a forecast of a decade of price stability. The cross-currency and Treasury bases of Chapter 19 are the same object in a purer form: a basis is a price gap arbitrage has left open: not a forecast of anything, but the price of a balance sheet, and central banks read it precisely as a gauge of intermediary capacity. The VIX is a third case. It is a risk-neutral expectation, so by Chapter 3 §3.4.5's construction it is expected volatility multiplied by a variance risk premium, and Chapter 8 §8.6 identifies the premium as compensation demanded by the dealers who are structurally short options because everyone else wants to be long them. A rising VIX is a joint statement about expected volatility and about the willingness of a small set of intermediaries to bear it, and the two components move for different reasons. None of this makes prices uninformative. It makes them informative about something more specific than "the economy": they are informative about the economy *and* about the ecology holding the claims, jointly, and separating the two requires knowing who the holders are. Which is, again, a measurement problem rather than a philosophical one. > **Box 27.2 — May Day, 1 May 1975** > > One date makes the chapter's distinction between issuing and trading concrete, because after it the cost of trading fell by orders of magnitude and the cost of intermediation did not. > > Until 1975 the New York Stock Exchange enforced a schedule of minimum commissions, fixed by rule and unnegotiable. A member firm could not compete on price for a customer's order, so competition took other forms: research, entertainment, and the allocation of new issues. Following an amendment to the Exchange Act and a Commission rule, fixed commissions ended on 1 May 1975. The industry called it May Day and expected carnage; a substantial part of the retail brokerage industry was in fact reorganised out of existence over the following decade, and the discount brokerage of Chapter 15's Box 15.1 was born within months. > > What happened to the price of a trade since is not in dispute. Institutional commissions fell from tens of cents a share toward fractions of a cent; retail commissions fell from tens of dollars a trade to zero; the bid-ask spread on a liquid stock fell from an eighth of a dollar to a fraction of a cent, helped by decimalisation in 2001 and by the market structure of Chapter 11. On any measure of the cost of *transacting*, the last fifty years are one of the great efficiency gains in the history of services. > > Set that against the two facts §27.2 records. Philippon's unit cost of intermediation — what the financial sector charges per dollar of intermediation provided — is roughly flat across more than a century, and the sector's share of GDP has approximately doubled since 1975. A market whose transaction costs collapsed and whose intermediation costs did not is telling us that the two activities are different, and that the second is the one Tobin was suspicious of. > > That is §27.2's distinction with a date attached, and §27.5's caution about what a price can tell us with a case attached. Cheaper trading is not the same thing as better allocation, and the fifty years since May Day are the largest available natural experiment on the difference. *** ## 27.6 Claims, Prices, Holders Return to the lecture. Tobin asked whether a financial system that had grown enormously was performing its functions better, and could not answer because the functions are performed by a match between issuers and holders that nobody could observe. The book's answer is that **the quality of that match is what functional efficiency is**, and that it is now largely measurable. Stated in the book's own terms: the system is functionally efficient to the extent that the claims firms and governments can issue are the claims that holders, given their liabilities and their constraints, want to hold — and to the extent that the holders whose demand is steepest are not the ones setting the price of capital for the projects that matter. Part V describes the supply side of that match, claim by claim and door by door. Part IV describes the demand side, balance sheet by balance sheet. Chapter 25 §25.6 is where the two are laid on top of one another, and the result is not reassuring: the menu is real, most doors are closed to most firms, and which door opens depends on which balance sheet happens to want the claim behind it. That is a description of a matching technology, and matching technologies can be evaluated. Tobin could only gesture at the sector's value-added share; we can ask which claims cleared, at what elasticity, held by whom, and what happened to the firms for whom no claim cleared at all. Three things remain genuinely open, and a closing chapter that pretended otherwise would be worse than useless to a student deciding what to work on. **The demand system's identification is unsettled.** Chapter 20 §20.5 marks the line: that demand curves slope down and that holder identity matters are established; that the investment-universe instrument — a manager's mandated asset menu, used as a demand shifter — satisfies its exclusion restriction, and that the aggregate multiplier is five rather than two or eight, are not. Since nearly every macro-financial implication in §27.4 is proportional to that multiplier, the difference between two and eight is the difference between an interesting cross-sectional finding and a reordering of macroeconomics. **The equity premium's resolutions have not been jointly tested.** Chapter 5 §5.5 sets out habit, long-run risk, and disasters — a consumption benchmark that risk aversion tracks, small persistent shocks to expected growth, and a catastrophe absent from the sample — and Chapter 5 §5.7 sets out limited participation and the intermediary alternative, which move the marginal holder away from the average consumer. Each family can be calibrated to the premium. None has been shown to survive the full battery — the premium, the riskless rate, the volatility of both, the predictability evidence of Chapter 7 §7.4, and the option-market evidence of Chapter 8 §8.5 — simultaneously and out of sample. **The welfare cost of the financing wall is not identified.** Chapter 25 §25.5 states it plainly. We can measure that small firms cannot substitute across the menu and that their investment responds to their lenders' balance sheets. We cannot measure the output lost because a productive firm was rationed, because that requires the return on the projects that were never funded, and the firms that were rationed do not report them. This is the single largest gap between what the finance-and-growth literature claims to measure and what it does measure, and it sits precisely on Tobin's question. So end on the map. Chapter 2 §2.3's master table is the thing to watch, because it is the object every argument in this book ultimately rests on: sectors down the side, claim classes across the top, and in the cells the quantities that fix the weights in Chapter 20's answer to Chapter 3's question. It is not a static object. Fifty years ago its largest holder column was households, holding equities directly; today it is intermediaries, holding on households' behalf under mandates the households never read. The passive complex, the insurance sector, the foreign official holders of Treasuries, the private funds whose positions the table sees worst — each of those rows has moved within a generation, and each movement changed a price before it changed anyone's forecast. When the holders change, everything downstream of this book changes with them: which claims can be issued, which firms can issue them, what the resulting prices mean, and what a policymaker reading those prices is actually reading. A claim is a promise about a payment. A price is what somebody paid for it. A holder is who ended up with it, and why. The first is the province of contract, the second of theory, and the third — the one this book has insisted on throughout — is the province of measurement, and it is the one that decides the other two. *** ## Elsewhere in the Series * **The financial accelerator, formally** — *Institutionalist Macroeconomics*, Chapters 12 and 19, which own the model, its general-equilibrium closure, and the quantitative business-cycle results. Section 27.3 keeps the one-line statement, the loop written out in words, and the supply-side extension. * **Wealth distribution, aggregate demand, and what a flow-driven asset market does to both** — *Institutionalist Macroeconomics*, Chapter 25. Section 27.4's wealth channel stops at the household balance sheet. * **Global transmission** — *International Finance*, for how the channels of §27.4 cross borders: the dollar's international role, global banking and the cross-currency basis, and capital-flow transmission to small open economies. This chapter's transmission is domestic throughout. * **The crisis as the ecology's fullest exhibit** — the 2008 crisis volume, which narrates in full what §§27.3 and 27.4 state as mechanisms: the collateral channel, the run on repo, and the fire sales that priced claims by who could no longer hold them. * **The intellectual history of these questions** — the *History of Economic Thought* volume: the efficient-markets debate, the finance-and-growth tradition from Schumpeter and Gurley-Shaw forward, and the reception of Tobin's own program. This chapter uses Tobin's taxonomy as a working tool and does not narrate its history. * **This book**: the organizing question — Chapter 1. The master holdings table — Chapter 2 §2.3. The pricing equation and "whose $$m$$?" — Chapter 3 §§3.5, 3.7. Efficiency and performativity — Chapter 7. Constrained capital as the canonical mechanism — Chapter 16 §16.5. Intermediary asset pricing — Chapter 19 §19.5. The demand system and the multiplier — Chapter 20 §§20.3, 20.5-20.6. Issuance as a response to holder demand — Chapter 23 §23.7. The financing menu and the wall — Chapter 25 §§25.3, 25.5-25.6. Risk constraints as a pricing mechanism — Chapter 26 §26.6; value at risk as convention — Box 26.1. *** ## Summary 1. **Tobin's 1984 taxonomy still organizes the question.** Information-arbitrage, fundamental-valuation, full-insurance, and functional efficiency are four different properties, and a financial system can improve on the first while doing nothing for the fourth. Tobin suspected exactly that and could not test it, because the match between issuers and holders was not measurable in 1984. 2. **The book's answer to its own question has four parts.** Prices are $$E\[mx]$$ with an $$m$$ that belongs to identifiable constrained holders (Chapters 3, 5, 19, 20); balance-sheet constraints move prices with no change in cash flows (Chapter 16 §16.5); issuers manufacture the claims holders are short of (Chapters 23 §23.7, 25 §25.6); and the weights that decide whose $$m$$ prices what are data rather than assumptions (Chapters 2 §2.3, 20). 3. **Whose** $$m$$**? Everyone's, weighted by wealth and by constraint — and the weights are measurable.** That single line is what separates this book's spine from the representative-agent canon, and it is a claim about measurement rather than about preferences. 4. **The finance-and-growth evidence is real, identified, and narrower than it is usually read to be.** King and Levine established the correlation from initial financial depth; Rajan and Zingales identified it off the industry-by-country interaction in external dependence; the "too much finance" literature found the relationship turning negative above roughly one hundred percent private credit to GDP, on estimates whose bands are wide and whose sample overlaps the pre-2008 boom. 5. **The growth evidence is about issuing and rationing, not about trading.** It speaks to Chapters 21-25 — whether a firm with a project can get funded — and not to Chapters 11-13's turnover. Philippon's finding that the unit cost of intermediation has not fallen in more than a century is the sharpest statement of the gap between the two. 6. **Financial deepening and improving investment opportunities are different events.** A credit-to-GDP ratio rises when holders' demand for claims rises, which Chapter 2 §2.4's fifty-year reallocation shows is largely a story about retirement systems. Against Chapter 20's steep demand curves, arriving money moves prices more than quantities, and the claims manufactured are the ones holders are short of rather than the ones the best projects need. 7. **The accelerator is one sign.** The external finance premium falls in borrower net worth, so a shock to net worth amplifies itself (Bernanke-Gertler, Kiyotaki-Moore). The machinery is the Macro volume's; this book's contribution is the supply side — in March 2020 and in the 2022 gilt episode the constrained agent was the *holder*, and the premium being repriced was the price of intermediation itself. 8. **Asset prices transmit through three channels, each of which is a claims-and-holders statement.** Cost of capital, through Chapter 9's curve, Chapter 13's mortgage convexity, Chapter 22's hurdle rates, and Chapter 23's issuance windows; wealth, through concentrated holdings (Chapter 14 §14.6), which makes the channel's size a distributional fact; and collateral, through Chapter 16 §16.5's margin spiral, in which falling prices raise haircuts and forced sales lower prices again. 9. **If flows move prices several-for-one, flows are a macro force.** A multiplier near five makes an informationless reallocation a first-order determinant of the market's level and blurs the line between financial and real shocks. The magnitude is contested (Chapter 20 §20.5); the sign and order of magnitude are not. 10. **A price move has at least four readings — news, a constraint binding, a flow, and a model becoming convention** — and separating them requires knowing who holds the claim. The breakeven inflation rate, the Treasury and cross-currency bases, and the VIX are each read by policymakers as forecasts and are each, in part, a holder artifact. 11. **Functional efficiency is the quality of the match between issuers and holders, and it is now measurable in ways Tobin could not manage.** Part V is the supply side of that match, Part IV the demand side, and Chapter 25 §25.6 is where they are laid on top of each other. What remains open: the demand system's identification, whether any resolution of the equity premium survives the full battery of tests, and the welfare cost of the financing wall. *** ## Key Terms * **Functional efficiency**: Tobin's fourth sense — whether the financial system performs its economic functions (pooling saving, allocating capital, spreading risk, clearing payments) at a reasonable cost in real resources. A property of the system, not of prices * **Information-arbitrage efficiency**: Tobin's first sense — the absence of systematic profit from public information; the weak and semi-strong forms of Chapter 7 §7.1 * **Fundamental-valuation efficiency**: Tobin's second sense — price equal to the rationally discounted value of the payments a claim will actually make * **Full-insurance efficiency**: Tobin's third sense — the availability of claims contingent on every state, so that every risk borne is a risk chosen * **Financial deepening**: Growth in the stock of financial claims relative to output, usually measured as private credit or liquid liabilities to GDP; a measure of holder demand as much as of capital allocation * **External finance premium**: The wedge between the cost of funds raised externally and the opportunity cost of internal funds, arising from monitoring and enforcement costs and decreasing in borrower net worth * **Financial accelerator**: The feedback by which a shock to net worth raises the external finance premium, lowers investment, and lowers net worth again, amplifying and propagating the original shock * **Cost-of-capital channel**: Transmission from asset prices to real activity through the discount rate applied to investment projects and through the terms on which claims can be issued * **Wealth channel**: Transmission from asset prices to consumption through household net worth; its magnitude depends on how the repriced assets are distributed across households * **Collateral channel**: Transmission from asset prices to activity through the borrowing capacity of everyone who has pledged the repriced asset; mechanical, immediate, and independent of forecasts *** ## Readings ### Required * Tobin, J. (1984). "On the Efficiency of the Financial System." *Lloyds Bank Review* 153: 1-15. *Fifteen pages that set this chapter's agenda: the four-way distinction among senses of efficiency, and the question of whether a rapidly growing financial sector was improving the one that matters. Read it before the chapter, and again after.* * King, R. and R. Levine (1993). "Finance and Growth: Schumpeter Might Be Right." *Quarterly Journal of Economics* 108(3): 717-737. *The cross-country result that founded the modern literature. Read §§II-III for the design and for the authors' own statement of what initial values do and do not rule out.* ### Recommended * Arrow, K. J. (1971). "Insurance, Risk and Resource Allocation." In *Essays in the Theory of Risk-Bearing*. North-Holland. *The functional test stated well before Tobin's lecture and in fewer pages: what an insurance market accomplishes is a reallocation of risk-bearing to those able to carry it, and the welfare gain is real even though nothing is produced. The pre-history of §27.1's answer, and the verbal form of Chapter 3's state prices.* * Rajan, R. and L. Zingales (1998). "Financial Dependence and Growth." *American Economic Review* 88(3): 559-586. *The identification improvement: external dependence measured as a technological characteristic, interacted with financial development, with country and industry effects absorbed. The paper to imitate when a cross-country correlation needs a design.* * Bernanke, B. and M. Gertler (1989). "Agency Costs, Net Worth, and Business Fluctuations." *American Economic Review* 79(1): 14-31. *The agency-cost version of §27.3's one line, in general equilibrium.* * Kiyotaki, N. and J. Moore (1997). "Credit Cycles." *Journal of Political Economy* 105(2): 211-248. *The collateral version, in which the borrowing limit depends on the price of an asset that the borrowing itself helps determine. The clearest statement of why small shocks persist.* * Caballero, R. J., E. Farhi and M. L. Hammour (2006). "Speculative Growth: Hints from the U.S. Economy." *American Economic Review* 96(4): 1159-1192. *Asset prices and investment in one model, with the mechanism §27.4 asserts made explicit and given a speculative component: a rise in valuations that is not warranted by fundamentals still raises investment while it lasts. The closest thing in the literature to §27.4's transmission argument taken seriously as macroeconomics.* * Philippon, T. (2015). "Has the US Finance Industry Become Less Efficient? On the Theory and Measurement of Financial Intermediation." *American Economic Review* 105(4): 1408-1438. *Constructs a unit cost of intermediation over more than a century and finds it roughly flat. The measurement is the contribution; the interpretation is the argument.* * Merton, R. C. (1982). *Finance Theory*. Unpublished lecture notes, MIT Sloan School of Management, chapters I and XII. *Chapter I is the field's own statement of what it is for, worth setting beside Tobin's four efficiencies; chapter XII, on risk-spreading through life insurance, is the functional answer to §27.2's question worked through one industry rather than one cross-country regression.* *** ## Discussion Questions 1. **Run Tobin's four efficiencies over the book.** Take each of the four senses — information-arbitrage, fundamental valuation, full insurance, functional — and name the Part of this book that bears most directly on it, the chapter that supplies the best evidence, and the verdict that evidence supports. Where two Parts speak to the same sense, say which is the stronger evidence and why. Which of the four does this book have the least to say about, and is that a gap in the book or in the literature? 2. **Is a five-to-one multiplier a macro shock source?** Suppose Chapter 20 §20.3's aggregate price multiplier is five. Write down the chain from a one-hundred-billion-dollar informationless inflow to a change in consumption, naming the channel of §27.4 at each step and the chapter that supplies the elasticity. Then state the two assumptions your chain most depends on, and what evidence would falsify each. Finally: at what value of the multiplier would you stop calling flows a macro force, and why that value? 3. **Which finance grew?** Credit to GDP and equity market turnover are both measures of "financial size" and they have moved very differently over the past forty years. Which of the two does the King-Levine and Rajan-Zingales evidence actually speak to? Construct a hypothetical economy in which one rises sharply and the other does not, say which of Tobin's efficiencies improves in each case, and explain why the "too much finance" threshold results are hard to interpret without making this distinction. 4. **Read three prices.** Pick the ten-year breakeven inflation rate, the cross-currency basis, and the VIX. For each, state what a central banker would conventionally infer from a large move, and then state the holder-side reading of the same move using §27.5's four categories. Design one additional observation, available for free, that would let you distinguish the two readings in each case. 5. **The accelerator's other leg.** Section 27.3 argues that in March 2020 and in the 2022 gilt episode the binding constraint was the *holder's* balance sheet rather than the borrower's. Take one of the two episodes and state what a standard borrower-side accelerator model would have predicted, what actually happened, and which chapter's mechanism accounts for the difference. Does the supply-side version amplify or dampen the borrower-side one, and under what conditions could it do the opposite? *** ## Data Exercise: Tobin's Question, in Three Series Every part runs on free public data. **Part A — The size of finance (BEA, free).** From FRED or directly from the Bureau of Economic Analysis GDP-by-industry tables, download value added for **Finance and insurance** and for **Real estate, rental and leasing** separately, together with total GDP, annually from 1947. 1. Plot finance and insurance value added as a share of GDP over the full period. Report the level in 1950, in 1984 (the year of Tobin's lecture), in 2007, and in the most recent year. State the multiple by which the share has risen since 1950. 2. Add the real estate line. Explain why the two are separated in the accounts, and why including imputed rent on owner-occupied housing would make the combined series answer a different question from the one Tobin asked. 3. Value added is revenue minus purchased inputs; it is not a measure of the service delivered. Using Philippon (2015) §I as a guide, state in one paragraph what would have to be measured in the denominator to convert this series into a unit cost of intermediation, and which of those quantities the Financial Accounts of Chapter 2 could supply. **Part B — Credit to GDP against the thresholds (BIS or FRED, free).** The Bank for International Settlements publishes credit to the private non-financial sector as a share of GDP for about forty-five countries; the US series is also on FRED. 1. Plot the US series from 1960 to the present, and mark the one-hundred-percent line that the "too much finance" estimates cluster near. Report the first year the US crossed it and the peak value. 2. Add four other countries of your choosing, including at least one middle-income economy. Rank them at the most recent date. Then, for each, state whether the King-Levine reading (more depth, more growth) or the threshold reading is the one you would apply, and why. 3. Pair the credit series with a turnover measure — the World Bank's Global Financial Development Database publishes stock market turnover ratios — and plot the two for the United States on the same axes, indexed to 1980. Describe how differently they have moved. Write one paragraph on what this does to any regression that uses a single variable to represent "financial development", referring to §27.2's distinction between the issuing and trading functions. **Part C — The standing exercise: rebuild the map (Z.1, free).** Return to Chapter 2's Data Exercise and run it again against the current Z.1 release. 1. Rebuild Table 2.5's holder shares for Treasuries, corporate equity, and corporate bonds. Compare each cell against the figure you (or the chapter) recorded previously, and list every share that has moved by more than one percentage point. 2. For each such movement, name the chapter of this book that owns the holder in question and state, in one sentence, what its mechanism predicts should have happened to the price of that claim. 3. Write a one-page memorandum in the form: *these holders changed, therefore these prices should have moved, therefore the following real-economy channel of §27.4 should have transmitted.* Then check the third step against the data. This exercise is designed to be repeated annually; it is the smallest complete version of what this book asks a reader to do. **★ Extension (WRDS or licensed data).** Using CRSP and Compustat, construct Rajan and Zingales's external-dependence measure by industry for the most recent decade — capital expenditures less cash flow from operations, divided by capital expenditures, at the industry median. Compare the ranking to theirs for the 1980s. Industries whose dependence has changed rank are the interesting cases: for two of them, say whether the change reflects technology or a change in the financing menu of Chapter 25 §25.1, and what evidence would separate the two.