Bernanke on the 1907 crisis

Marcus Nunes sent me a very interesting talk by Ben Bernanke, which discusses the 1907 banking crisis:

But even as the banks stabilized, concerns intensified about the financial health of a number of so-called trust companies–financial institutions that were less heavily regulated than national or state banks and which were not members of the Clearinghouse. As the runs on the trust companies worsened, the companies needed cash to meet the demand for withdrawals. In the absence of a central bank, New York’s leading financiers, led by J.P. Morgan, considered providing liquidity. However, Morgan and his colleagues decided that they did not have sufficient information to judge the solvency of the affected institutions, so they declined to lend. Overwhelmed by a run, the Knickerbocker Trust Company failed on October 22, undermining public confidence in the remaining trust companies.

To satisfy their depositors’ demands for cash, the trust companies began to sell or liquidate assets, including loans made to finance stock purchases. The selloff of shares and other assets, in what today we would call a fire sale, precipitated a sharp decline in the stock market and widespread disruptions in other financial markets. Increasingly concerned, Morgan and other financiers (including the future governor of the Federal Reserve Bank of New York, Benjamin Strong) led a coordinated response that included the provision of liquidity through the Clearinghouse and the imposition of temporary limits on depositor withdrawals, including withdrawals by correspondent banks in the interior of the country. These efforts eventually calmed the panic. By then, however, the U.S. financial system had been severely disrupted, and the economy contracted through the middle of 1908.

The recent crisis echoed many aspects of the 1907 panic.

Bernanke’s right that the crises were similar.  He also discusses the advances in policy that have occurred since 1907, such as FDIC, which prevented a run on deposits, and the Federal Reserve, which pumped in massive quantities of base money.  This might partly explain why RGDP only declined slightly over 4% from 2007:4 to 2009:2, as compared to 12% between 1907:2 and 1908:1 (using Gordon and Balke).  But the recovery from the 1907 crisis was also very swift (which that shows financial crises are not inevitably followed by slow recoveries.)  By 1909:4 RGDP was up by nearly 19% from its 1908 low point.

Why was the recovery from the 1907 crisis so swift?  The proximate cause is obvious, monetary policy was much more expansionary in 1908-09 than in 2009-13.  NGDP soared by nearly 28% between 1908:1 and 1909:4.  But why was that?

I believe the gold standard helps explain the rapid recovery in 1908-09.  The international gold standard was a very crude sort of “level targeting” regime.  It didn’t necessarily work when there were international shocks to the real value of gold, due to changes in the supply or demand for gold.  Price levels tended to move in a sort of random walk.  You could have fairly long periods of deflation, as in the 1870s and the 1890s and the early 1930s.  But when the international gold market was relatively stable, a sharp domestic shock that greatly depressed prices and output was likely to be quickly reversed, as prices and output recovered to their original trend line.  In contrast, the Fed made a decision in 2009 to avoid going back to the original trend line, and instead to start a new and lower trend line.  This explains the slow growth in NGDP, and also the slow recovery in RGDP.

I still think that fiat money is better than a gold standard, but it’s also clear that fiat money policymakers could learn something by studying the cyclical properties of the international gold standard.

[Marcus Nunes has a post with some graphs comparing the two episodes.]


Tags:

 
 
 

17 Responses to “Bernanke on the 1907 crisis”

  1. Gravatar of Becky Hargrove Becky Hargrove
    9. November 2013 at 08:02

    Nick Rowe and the gold standard random walk:
    http://worthwhile.typepad.com/worthwhile_canadian_initi/2013/11/naive-vs-rational-expectations-is-a-partly-false-dichotomy.html

  2. Gravatar of jknarr jknarr
    9. November 2013 at 08:19

    People focus on 1907, but is there any comparable discussion of the 1903 “panic”? 1907 just seems to be a continuation of 1903′s vol event, seemingly with higher leverage.

    http://research.stlouisfed.org/fred2/graph/?g=ohV

    (Words of praise for gold, too! All these monetary regimes “work” on paper — to paraphrase an earlier comment, it’s the nature of the cheating in a regime that marks its notable characteristics in hindsight, and is most interesting to study. Which begs the question – what and who are the cheats and cheaters today?)

  3. Gravatar of Ashok Rao Ashok Rao
    9. November 2013 at 09:45

    Quick question. I understand the importance of looking at nominal – rather than real – indicators to gauge the stance of monetary policy (that is, money is neutral in the long run).

    But I’ve always wondered about statements like “But it’s not the right statistic, as it reflects both supply and demand shocks. Ideally you want a measure of changes in aggregate demand or total spending, i.e. NGDP.”

    Why wouldn’t AS affect NGDP? If demand is sufficiently inelastic, an contraction in productivity will increase NGDP. That would not be a reflection of changes in demand. Similarly, a simultaneously positive supply and demand shock (shale gas + depreciation) would leave NGDP unchanged but, again, says nothing about the state of demand.

    What am I missing?

  4. Gravatar of Bill Woolsey Bill Woolsey
    9. November 2013 at 11:00

    Rao:

    What are you holding constant? The quantity of money? Then it turns out you are saying that the demand for money is “real” income elastic.

    Or are you looking at where the long run and short run aggregate supply curves intersect?

    If what is held constant _is_ nominal GDP, then the aggregate demand curve is unit elastic.

  5. Gravatar of ssumner ssumner
    9. November 2013 at 11:09

    jknarr, Good point. It’s easy to cheat with inflation targeting, it’s almost impossible to cheat with NGDPLT. That’s one of my key arguments.

    Ashok, The slope of the AD curve depends on the policy regime, under NGDPLT it would be unit elastic.

  6. Gravatar of Steve Steve
    9. November 2013 at 11:20

    “However, Morgan and his colleagues decided that they did not have sufficient information to judge the solvency of the affected institutions, so they declined to lend. ”

    Did the DOJ spend 1908-1916 suing J.P. Morgan for manipulating the interbank lending market?

  7. Gravatar of Ashok Rao Ashok Rao
    9. November 2013 at 12:32

    “Ashok, The slope of the AD curve depends on the policy regime, under NGDPLT it would be unit elastic.”

    Right, but whatever we have now is certainly not a nominal level target – in which case nominal income also includes supply-side factors.

  8. Gravatar of Rajat Rajat
    9. November 2013 at 14:15

    Simple question – how was monetary policy made very expansionary in 1908/09 under a gold standard? I understand that a gold standard makes it harder for real wages to adjust and so a negative demand shock is more likely to result in rising UnN than under a fiat currency. But then what causes the rapid turnaround if the country stays on gold?

  9. Gravatar of ssumner ssumner
    9. November 2013 at 16:38

    Ashok, Yes, but it’s still far better than the price level. I don’t think supply shocks have a very significant impact on US NGDP.

    Rajat, That’s a tough question. Perhaps expectations that prices and AD would return to the trend line caused current nominal spending to rise.

  10. Gravatar of Jake Jake
    9. November 2013 at 19:40

    I’m still not sure why full reserve banking is not preferable to fractional reserve. No bank runs, no FDIC, and the monetary authority is just as capable of conducting monetary stabilization as in the current system.

    What’s the downside?

  11. Gravatar of Geoff Geoff
    10. November 2013 at 10:54

    “Why was the recovery from the 1907 crisis so swift? The proximate cause is obvious, monetary policy was much more expansionary in 1908-09 than in 2009-13. NGDP soared by nearly 28% between 1908:1 and 1909:4. But why was that?”

    “I believe the gold standard helps explain the rapid recovery in 1908-09. The international gold standard was a very crude sort of “level targeting” regime.”

    A wiser person would believe that a “level targeting regime” is a very crude sort of gold standard.

  12. Gravatar of Geoff Geoff
    10. November 2013 at 11:00

    “It didn’t necessarily work when there were international shocks to the real value of gold, due to changes in the supply or demand for gold.”

    What would cause such changes in the demand for gold?

    “But when the international gold market was relatively stable, a sharp domestic shock that greatly depressed prices and output was likely to be quickly reversed, as prices and output recovered to their original trend line. In contrast, the Fed made a decision in 2009 to avoid going back to the original trend line, and instead to start a new and lower trend line. This explains the slow growth in NGDP, and also the slow recovery in RGDP.”

    The Fed made a decision in 2009 to continue avoiding any adherence to a gold standard, and instead inflated far more than what a gold standard would have allowed. This explains the slow recovery in RGDP.

    See what I did there?

  13. Gravatar of Geoff Geoff
    10. November 2013 at 11:18

    Sumner, you really should learn to distinguish ideas that follow from historical data, such as identification of past events, and ideas that are established a priori and utilized to interpret historical data, such as economic theories.

    You have a habit of believing that believing that your a priori theories are empirical, and hence have a monopoly on explaining historical data.

    Two theories:

    1. The Fed brings about more inflation than a free market in money system would allow, and so economic recoveries are greater than what they otherwise would have been.

    2. The Fed brings about more inflation than a free market in money system would allow, and so economic recoveries are smaller than what they otherwise would have been.

    There is no way to establish which of these two theories is correct by recourse to historical events alone. Why? Because history agrees with both of these mutually incompatible theories.

    Economic science is purely logical. We utilize empirical concepts to give contextual interpretations of history, but the intellectual tools are grounded a priori. Geometricians utilize empirical concepts such as door stops and inclined surfaces to give a contextual interpretation of the Pythagorean theorem. But the intellectual tools are grounded a priori. Economics is the same way.

    In another context, those who argued that the economic recovery post-2009 would be sluggish on the basis that the Fed acted to reverse the deflation, are just as correct about the outcome as you were when you argued that the economic recovery post-2009 would be sluggish on the basis of insufficient reversal of deflation.

    The way economic theories are exposed as flawed is not on the basis of history. Only historical arguments can be refuted by history. Economic arguments can only be refuted by discursive reasoning, because they are of a categorically different logical nature.

    The economic question of money can only be settled by discursive reasoning on the nature of money, and its place in human activity.

  14. Gravatar of Geoff Geoff
    10. November 2013 at 11:25

    The Panic of 1907 was caused by prior inflation.

    From wiki.Mises.org:

    In the forty years 1890 to 1930, the population of the country doubled, the value of farm property increased three and a half times, pig iron production four and a half times, exports five times, coal production five times, and freight traffic five and a half times, but commercial bank deposits increased over seventeen and a half times. Thus, while the gold stock had increased proportionally with the increase of industrial production, the expansion in bank credit had far outstripped both and had thus been at the expense of a thinning gold reserve. The monetary gold stock available to support and redeem this tremendous amount of bank liabilities that was being created, which had been 25.3 per cent of total note and deposit liabilities of banks in 1865, and 23.9 per cent in 1880, steadily dropped under the pressure of the public upon the banking interest for more and more credit, standing in 1900 at 20.4 per cent, in 1910 at 14.2 per cent and in 1930 at 10.4 per cent. Such had been the diminution of reserves that by the decade 1920-1930, banking was being conducted “on a shoestring.” In 1900 the ratio of vault cash to deposit liabilities of commercial banks had dropped to 14.8 per cent, and in 1910 to 12.7 per cent—vault cash in those years being respectively $706,302,000 and $1,366,164,000.[3]

    After attempts to create a central bank failed, the Secretary of Treasury Leslie Shaw attempted to continue and expand the experiments of his predecessor Lyman Gage in making the U.S. Treasury function like a central bank. In particular, Shaw made open-market purchases in recessions and violated the Independent Treasury statutes confining Treasury funds to its own vaults, by depositing Treasury funds in favored large national banks. In his last annual report of 1906, Secretary Shaw urged that he be given total power to regulate all the nation’s banks.

    The Panic of 1907 struck in October, the result of an inflation stimulated by Secretary of the Treasury Leslie Shaw in the previous two years.[2]

    ———————

    References

    1. Rothbard, Murray N. History of Money and Banking in the United States.
    2. Rothbard, Murray N. The Case Against The Fed, 1994, p.106, 108.
    3. Elgin Groseclose. Money and Man, 1934, p.209-210.

  15. Gravatar of ssumner ssumner
    11. November 2013 at 06:20

    Jake, Perhaps a compromise is 100% reserves (cash or T-bills) for FDIC insured deposits, and no legal reserve req. for other deposits.

  16. Gravatar of Brian McCarthy Brian McCarthy
    11. November 2013 at 12:32

    You say that “fiat money is better than a gold standard,” but this assertion would seem to assume that it is possible to find, appoint, and get confirmed an FOMC that will follow optiomal policies, as you would define them.

    So while an ideally managed fiat currency would be superior to a gold standard, might not the gold standard be preferable in reality.

    You certainly seem to suggest that it would have performed better in the past 5 years.

  17. Gravatar of ssumner ssumner
    12. November 2013 at 07:35

    Brian, No, because a gold standard also depends on good policymakers. Compare ideal fiat to ideal gold standard, or actual fiat to actual gold standard.

Leave a Reply