
George Selgin on the Fed 12/06/2010
What this covers
George Selgin, an economist at the University of Georgia, sits down with Russ Roberts to argue that the Federal Reserve's two main justifications—ending bank panics and stabilizing the business cycle—rest on faulty historical ground. Rather than reviewing the Fed's performance against an idealized standard, Selgin insists on comparing realistic central-bank outcomes to realistic alternatives, chiefly the decentralized free banking that preceded it. He draws on recent research to show that deposit insurance and government-backed lending facilities, not the Fed itself, halted bank panics after 1933, and that revisionist data on output volatility undermine the claim that post-war business cycles were gentler than pre-Fed ones. The Great Moderation of the 1980s and 1990s, he argues, likely resulted from good luck and structural change rather than superior monetary management.
The conversation ranges across the informational and political obstacles built into central banking. Selgin explains why Bagehot rightly distinguished illiquidity from insolvency—and why the Fed blurs that distinction when it rescues insolvent firms, fueling moral hazard that escalates each subsequent crisis. He critiques the Fed's reliance on sterilized lending and sketches how open-market asset purchases could instead provide liquidity without picking winners. He walks through the interwar Gold Exchange standard's fragility, defends the classical gold standard against its worst historical examples, and contends that unit banking restrictions made American banks structurally vulnerable—not a case for central control but an argument against artificial constraints that prevented diversification. The clearest break with convention comes in Selgin's case for decentralized, competitive free banking as a means to distribute decision-making errors across multiple institutions, a position that surfaces as the most contentious claim the episode entertains.
Selgin argues that the conventional case for the Federal Reserve—that it ended bank panics and tamed the business cycle—is unsupported by the best available evidence, and that decentralized free banking would better handle the informational and incentive problems that make central banking inherently unstable.
- Deposit insurance and the RFC, not the Fed's lender-of-last-resort function, actually halted bank panics after 1933
- Revisionist data show post-war business cycles were not less severe than pre-Fed cycles, and the Great Moderation was likely good luck or structural change rather than improved monetary policy
- Both informational deficiencies and public-choice incentives mean the Fed cannot perform as textbooks idealize, so decentralizing money supply decisions would reduce systemic error
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
Unit banking persisted because banking was historically treated as a government privilege requiring a specific legislative act, so established bankers lobbied to block competitors—especially New York banks—and this combined with a public attitude that banks are dangerous and monopolistic so there should be as few as possible.
“banking was thus a government privilege that was very jealously guarded and every banker who succeeded would line up and oppose the entry of other banks particularly in their town”
A classical lender of last resort, per Bagehot, lends freely at high rates of interest but only to solvent firms during a crisis, in order to prevent spillovers from insolvent institutions from bringing down otherwise-solvent ones; this rests on a crucial microeconomic distinction between illiquidity and insolvency.
“a lender of last resort according to Badgett if you're stuck with a central bank is a central bank that lends freely at high rates of interest but only to solvent firms during times of crisis”
The proper comparison is not between gold-standard supply shocks and an idealized perfectly-managed fiat money, but between realistic gold-standard instability and realistic central-bank performance; historically, even the most extreme gold supply shocks (1840s gold rush, 16th-century price revolution) produced annualized inflation rates below the Fed's own implicit 2% target, so people simply don't know the history of prices.
“the relevant comparison should be not between how much the price level might change in a gold standard arrangement... and how much it would change if you had an ideal central bank... It should be between what the gold standard is likely to generate... and what central banks seem likely to generate”
The interwar Gold Exchange standard—not the classical gold standard—caused the deflationary disasters: it was a jury-rigged system depending on central banks treating claims against the Bank of England as cash and practicing forbearance, designed to let Britain return to its pre-war parity without the massive deflation that restoration normally required; when the Bank of France stopped cooperating, the 'house of cards' collapsed.
“when the Bank of France stopped cooperating it all came tumbling down it was a house of cards”
Bagehot himself argued the English banking system was deeply structurally flawed precisely because it had a central bank capable of mismanaging money and triggering gold drains; he pointed to the relatively free Scottish system as ideal and made clear that a decentralized monetary system needs no lender of last resort, calling his own recommendation a second-best solution.
“he also very explicitly said that the English system was deeply structurally flawed and it was structurally flawed because it had a central bank”
US restrictions on branch banking (unit banking) made for a system of many weak, undiversified banks; because banks couldn't buy and sell loans or hold many marketable securities, they were heavily dependent on local firms and farms and easily succumbed to local economic shocks, making runs more likely.
“The lack of branch banking alone made for a system of very weak banks... banks were heavily dependent on the success of the firm's in the local economy... when you have local shocks they relatively easily succumbed to those shots and runs”
Maintaining a fiat money requires at minimum a public monetary authority to guarantee its scarcity (since paper is cheap to print) and to prevent private replicas; only a commodity-based standard like gold needs no central authority to guarantee the unit's scarcity and value.
“in a minimal sense having a fiat money means having a public monetary authority if only in order to safeguard the paper stuff from being our augmented”
Because of informational deficiencies, the Fed frequently mistakes the stance of money—thinking it has tight money when money is loose (or vice versa) and then adjusting in the wrong direction—creating an unstable feedback loop that makes conditions worse, analogous to tangled lab hoses where you adjust the wrong tap.
“what the Fed has done many many times it thinks it has tight money and so it loosens but actually it's got loose money and it's doing making things worse at other times it takes there's loose money so it tightens when in fact there's tight money already”
During the crisis the Fed was too busy making sterilized loans to insolvent institutions to attend to the illiquidity of the general market; it should instead have engaged in aggressive open-market operations without sterilization, and (following Buiter) considered purchasing a broader range of assets at their actual worth rather than lending to particular insolvent firms.
“it was too busy making sterilized loans to insolvent institutions to to attend to the illiquidity of the general market”
Until the bank holiday of 1933, banking panics under the Fed were not less frequent or less severe than they had been before the Fed's establishment; if anything the Fed's early years saw an increase in panics, including the most severe ever, with about five panics in the 1930s.
“the few studies that try consistently to account for panics find that until the bank holiday of 1933 panics were not less frequent or less severe than they had been before the establishment of the Fed”
Each bailout raises the ante through moral hazard, leading inevitably to an even bigger insolvency failure: the Fed's rescue of insolvent Franklin National in the 1970s was followed by the insolvent Continental Illinois in 1984, escalating to the recent crisis's massive interventions.
“every time it does that it raises the ante because of moral hazard and there will inevitably be a even bigger failure from insolvency when this process happened so along comes continental Illinois 1984 also insolvent”
The historical record shows that bank runs in the absence of insurance overwhelmingly targeted banks that were in danger of failing because of bad loans, not banks failing simply because of panic or 'animal spirits'; runs typically hit institutions on the verge of insolvency or heavily tied to such institutions.
“the historical record overwhelmingly shows that bank runs in the absence of insurance have tended to be runs on banks that were in danger of failing and not because of the runs but because of bad loans”
There are two distinct kinds of spillover: too-big-to-fail interconnection (where one firm's insolvency makes others insolvent) and Bagehot-style panic (where one failure triggers a liquidity shortage for otherwise-solvent firms); only reserve primary was actually insolvent among money market funds, so the rest faced a liquidity—not solvency—crisis, which is the case for open-market liquidity support but not for keeping Lehman alive.
“there are two questions here one is the too-big-to-fail question... the other kind of spillover effect is the kind Badgett was arguing central banks should prevent and that is where panic generated by one firms failure... leads to a run on or shortage of liquidity for other otherwise solvent institutions”
Buiter's proposal shows the Fed could conduct open-market purchases even of questionable assets via a Dutch auction that lets the market determine the securities' value, thereby providing liquidity while avoiding moral hazard and protecting taxpayers from buying assets at artificially inflated prices.
“if you would design the open market operation with a Dutch auction or something like that you can pretty much allow the the option to determine the value of the secure of the securities and do so in a way that will avoid the moral hazard problem”
Mandatory 'living wills'—plans by large financial firms specifying procedures to expedite bankruptcy in the event of insolvency—are a sensible reform that addresses the legitimate worry that ordinary bankruptcy is too slow to avoid spillovers, and the US should adopt them.
“the English proposal the legislation that requires larger financial firms... to create so-called living wills these are specific plans... amount to procedures for expediting bankruptcy by specifying exactly what should happen in the event that they are insolvent”
A more realistic model of the Fed than the technocratic 'bathtub taps' view is a public-choice model in which the Fed, as a political institution, responds to economic actors (the banks) with strong incentives to influence its decisions—not through conspiracy but through general political forces that reward keeping Wall Street and the administration happy—implying the Fed mostly helps banks at the expense of everyone else.
“wouldn't a more fruitful model of the Fed be a public choice model that says I would never think that they're gonna run the taps in the drain... I would normally expect them to be a political institution to respond to economic actors that... would have a big incentive to influence their decisions which would be the banks”
Rule-based reforms (a Friedman steady-money-growth rule, or a Taylor rule) do not solve the underlying informational challenge, because we lack reliable real-time measures of the price level or high-powered money, and high-powered money is no longer regarded as the lever the Fed actually uses.
“the problem with all these those they don't really solved the informational challenge you talked about earlier which is you know these are sort of textbook in the sense that I don't really have reliable measures... of what the real price levels doing or what the what high-powered money is”
The abandonment of the gold standard during World War I was driven by belligerent countries wanting inflationary finance to spend more than they could without printing money; the US imposed gold embargoes and inflated, but far less than England and incomparably less than Germany by 1923.
“the belligerent countries... wanted to resort to inflationary finance to cover the the expenses of the war they wanted to spend more than they could spend without printing money and to do that they had to abandon the gold standard”
The Great Moderation (roughly 1983–2007) was real—nobody denies the reduction in output volatility—but the most recent data suggest it has ended, and a large body of research has steadily chipped away at the early presumption that improved monetary policy caused it, attributing it instead to good luck (fewer external shocks) or structural changes (dynamism, financial innovation, demography, inventory management) while testing and rejecting the improved-monetary-policy hypothesis.
“that research has steadily chipped away at the early presumption that improved monetary policy was behind it”
Revisionist research (chiefly Christina Romer and others) has drastically changed our understanding so that the post-war business cycle has not been less severe overall than the pre-Fed cycle: contractions have not been less frequent, shorter, or less severe, and output has not been substantially less volatile.
“the new research suggests that the post-war business cycle has not been less severe overall than the pre fed business cycle that is contractions have not been less frequent they haven't been shorter they haven't been less severe and output has not been substantially less volatile”
The big run that started at the end of February 1933 was mainly a run on gold driven by fear of devaluation; once the executive-order suspension of gold withdrawals removed the ability to obtain gold, the motive for running on banks was taken away.
“the big run that started at the end of February 1933 and continued to the time of the national holiday was mainly a run on gold because people feared devaluation of course once they couldn't get gold anymore that took that motive away for running on the bank”
A classical gold standard could have been restored in the 1920s in only two ways: returning to pre-war parities (requiring massive deflation) or adopting new parities reflecting the much-higher post-war price levels; the sensible 'let bygones be bygones' option of new parities was avoided because Britain wanted to restore its pre-war parity, producing the fragile Gold Exchange standard instead.
“if you chose to go back to the pre-war definitions of the paper currency... then big deflation was necessary... alternatively... England and other countries could have let bygones be bygones they could have reestablished a classical gold standard but with new parities reflecting the reality”
The only way to avoid major monetary errors tied to informational deficiencies is to decentralize—and thereby de-centralize—the money supply via a competitive free-banking system, because just as we prevent any one firm from monopolizing other markets so its mistakes don't sink the whole market, multiple money-issuing institutions making separate decisions would have their errors largely offset, conditional on a banking system not dependent on routine government bailouts.
“the only way to avoid major errors of monetary change connected with informational deficiencies... is to decentralize the decision-making behind adjustments in the money supply which means these centralizing the money supply”
Even if interlocking systemic risk were entirely real, accepting that 'we can't let the first domino fall' creates a permanent welfare program for Wall Street that lets institutions enlarge themselves with borrowed money, take riskier bets, and pay large salaries—appearing successful until collapse, then being rescued and repeating—a vicious cycle that only worsens absent draconian regulation that would turn finance into a public utility.
“once you say that we can't let that happen you basically create a welfare program for Wall Street that allows them to perpetually enlarge themselves with borrowed money and correlate the size of their institutions make riskier and riskier bets”
Too-big-to-fail is internally incoherent: if large spillovers are the worry, the proper response is ample liquidity provision (Bagehot-style) to still-solvent firms, never bailing out a particular firm; moreover too-big-to-fail has always been a lie, since after-the-fact analysis (e.g., George Kaufman on Continental Illinois showing only two banks would have lost more than half their capital) shows the predicted catastrophes did not materialize.
“if big spill overs are what you're worried about then big support for the still liquid still solvent firms is what's called for there's no reason to bailout any firm because of big spillover effects”
What actually halted bank panics after 1933 was not the Fed but the Reconstruction Finance Corporation buying large amounts of bank stock and the FDIC providing deposit insurance, which removed one of the more important proximate causes of panic.
“what really brought panics to a halt for a while was first of all the RFC which was bought tremendous amounts of bank stocks... FDIC and the FF LIC by providing Deposit Insurance eliminate what had become a more one of the more important proximate causes of panic”
The systemic/interconnection argument for bailouts is an assertion without evidence that we are asked to take on faith; like the boy who cried wolf, past assertions of contagion (Continental Illinois, LTCM, Lehman) turned out untrue, with Lehman's spillover effects surprisingly minor as Peter Wallison has shown.
“it is an assertion without evidence and we are expected to take it as a matter of faith and as with the boy who cried wolf we we know from the past assertions of that type that they have been untrue”
The Fed's post-1933 record is one of persistently violating Bagehot's rule by lending to insolvent institutions rather than letting them fail, which not only throws good money after bad but creates a moral hazard problem that can itself become a potent cause of reckless lending.
“the feds record has been one of persistently violating Paget's rule to lend only to solvent institutions and let the insolvent ones fail”
The best policy is not what actually happens; the political 'sausage factory' gives banks far more say than ordinary citizens, and many economists who benefit from the Fed's size and power also favor avoiding messy bankruptcy, so political incentives—not optimality—determine which policy emerges.
“what actually happens is the stew the ferment of the political sausage factory in that world the the banks get a lot more say than you and me”
The Fed acted only as an emergency lender, not a classical lender of last resort; in the early 1930s it lent only to member banks, leaving many banks 'in the cold', and it played an increasingly emergency role starting in the 1970s and escalating to the recent crisis.
“it did do some last resort lending during the crises of the early 1930s in February 1932 for example but only to member banks it did not see itself as responsible for other banks and that left a lot of banks in the cold”
Economists wrongly idealize central banks in textbooks, portraying them as performing the way they could only perform with information that is actually unavailable to them; in reality the Fed determines monetary policy in a 'finger in the wind' manner and is constantly making very severe mistakes, so textbooks and news should describe how central banks actually behave, not how they ideally would.
“what appalls me is the way economists idealized the Fed and other central banks writing in textbooks about them as if they actually performed the way they might ideally perform”
Runs and panics were a much bigger problem in more heavily regulated banking systems, with the US at the top of the list among industrialized nations; in relatively less regulated systems panics were either unknown or infrequent.
“you'll find that runs and panics were a much bigger problem in the more heavily regulated systems and the u.s. is at the top of that list among industrialized nations at the time and when you go to systems that were relatively less regulated you find the panics where either unknown or infrequent”
The paper's limited claim is a burden-of-proof argument: the best available statistical evidence (on output volatility, banking panics, recession length and frequency, lender-of-last-resort conduct, price-level behavior, and inflation predictability) provides no prima facie case that the Fed has improved economic performance, so defenders bear the burden and we should stop being complacent and do systematic research on alternatives.
“the available evidence out there doesn't provide a case an initial case showing that the Fed clearly has made things better”
Money market funds lent to Lehman in search of higher yield precisely because Bear Stearns's creditors had been rescued, which encouraged the expectation that Lehman would be rescued too—an example of how prior bailouts distort behavior.
“they lent the money because Bear Stearns had been rescued in search of higher yield they lent money to Lehman”
Free banking can only work if the system is first weaned off bailout guarantees to create strong private banks capable of taking charge of money creation; it will not work with a crummy banking system dependent on routine government bailouts and the expectation thereof, such as exists in the US today.
“it's not going to work with a crummy banking system dependent on routine government bailouts and the expectation thereof such as exists today in the United States so the first step has to be to wean our system off from guarantees”
The mainstream view until recently is that the Fed improved economic performance by eliminating 19th-century bank runs and, once monetary economics was better understood, smoothing the business cycle into the Great Moderation.
“The mainstream view at least until recently is of course the Fed has improved economic performance as a lender of last resort its eliminated the bank runs and panics of the 19th century”
You can follow Russ Roberts on Twitter at @econtalker for upcoming guests and observations.
“you can follow me on Twitter at AZ econ talker”