The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
____________ Review____________ Vol. 68, No. 10 December 1986 5 Coping with Bank Failures: Some Lessons from the United States and the United Kingdom 15 W hy Has M anufacturing Em ploym ent Declined? The Review is published 10 times p e r year by the Research and Public In form ation D epartm ent o f the Federal Reserve Rank o f St. Louis. Single-copy subscriptions are available to the public fr e e o f charge. Mail requests f o r subscriptions, back issues, o r address changes to: Research and Public In form ation Departm ent, Federal Reserve Rank o f St. Louis, P.O. R o t 442, St. Louis, M issouri 63166. The views expressed are those o f the individual authors and do not necessarily reflect official positions o f the Federal Reserve Rank o f St. Louis o r the Federal Reserve System. A rticles herein may be reprin ted provided the source is credited. Please provide the Rank's Research and Public In form ation D epartm ent with a copy o f reprin ted material. Federal Reserve Bank of St. Louis Review December 1986 In This Issue . . . Although U.S. bank failures are occurring at an unusually rapid rate, they have neither eroded public confidence in the banking system nor induced bank runs by depositors. The stability o f the U.S. banking system in the face o f increased bank failures reflects the success o f policies designed to prevent banking panics. In the first article in this Review, "Coping w ith Bank Failures: Some Lessons from the United States and the United Kingdom ,” R. Alton Gilbert and Geoffrey E. W ood examine the history o f banking panics and the evolution o f government policies designed to eliminate them in the United Kingdom and the United States. The last U.K. banking panic occurred in 1866. In response to that panic, the Bank o f England accepted the responsibility o f acting as lender o f last resort — by increasing bank reserves w hen the public withdraws large shares o f their deposits in the form o f cash — and bank runs ceased to be a problem. In contrast, the United States experienced panics throughout the 19th and early 20th centuries. In response to these panics, the Federal Reserve System was established in 1917 and federal deposit insurance was established in 193:5. Since 1933, although individual banks have failed, no banking panics have occurred in the United States. * * * Manufacturing em ployment in the U.S. econom y has declined since 1979, furthering the view that the United States is losing out in an international competition for manufacturing jobs. In the second article in this Review, "W hy Has Manufacturing Employment Declined?” John A. Tatom examines the factors that determine manufacturing employment. He argues that the decline in manu facturing em ployment has occurred for two reasons. Part o f the decline repre sents a transitory cyclical phenom enon. However, the decline is also due to the relatively rapid growth o f productivity in manufacturing that has taken place throughout the post-World War II period. The only recent period in which manufacturing em ploym ent grew relatively rapidly since W orld W ar II was in the early ’60s, when manufacturing wages declined sharply relative to wages paid in the rest o f the economy. Tatom explains that the relatively rapid growth in manufacturing productivity has been an important source o f the rising standard o f living in the United States and that it has been associated with a declining relative price o f these goods. Consumers, however, have not chosen to realize all o f the gain in their standard o f living through greater consumption o f manufactured goods. Instead, they have dem anded more o f other goods and services as well. Thus, the proportion o f labor resources em ployed in manufacturing has declined fairly steadily especially since 1969. International com petition has played only a small role in overall developments, the author savs. In the 1980s, he argues, manufacturing output and em ploym ent have strengthened relative to the experience abroad. 3 FEDERAL RESERVE BANK OF ST. LOUIS DECEMBER 1986 Coping with Bank Failures: Some Lessons from the United States and the United Kingdom /{. Alton Gilbert and Geoffrey E. Wood T JL HE number o f U.S. bank failures has risen dra matically in the past few years. Banks failed at the rate o f about six per year from 1950 through 1981. In 1984, however, 79 banks failed, and 120 banks failed in 1985.' Yet this sharp rise in the rate o f bank failure has not produced the kind o f public “panic” that accom pa nied bank failures during much o f U.S. history. system w ill not be disrupted by the failure o f one bank or even by the failure o f several banks. The government policies that create this public confidence in the sta bility o f the banking system reflect the histoiy o f each nation. This paper contrasts the experience with banking panics in the United Kingdom to that in the United States. In a banking panic, the failure o f one bank leads people to fear for the safety o f their funds at other banks. Subsequent attempts to w ithdraw their de posits from other banks put these banks in jeopardy as well. Recent experience suggests that bank failures no longer cause banking panics. There are now wellestablished and frequently tested principles for pre venting a bank failure from turning into a panic. The last banking panic in the United Kingdom oc curred in 1866. At that time the Bank o f England acted to prevent the disruption o f the banking system when banks failed and the public in England came to believe that the Bank o f England had accepted that responsi bility and w ould be successful in canying it out. To fully appreciate the importance o f these princi ples in preventing panics, it is necessary to review some episodes o f history during which panics oc curred. History illustrates the inherent vulnerability of the banking industry to panics, when there are no policies in place to prevent them; it also illustrates the adverse effects o f panics on the operation o f banking systems and econom ic activity. To prevent banking panics, it is necessaiy to con vince the public that the operation o f the banking R. Alton Gilbert is an assistant vice president at the Federal Reserve Bank of St. Louis and Geoffrey E. Wood is a professor of economics at City University, London. Sandra Graham provided research assis tance. 'A bank is declared to have failed when it is closed by the govern ment authorities. The government authorities close a bank when its net worth falls close to or below zero. The United States established a central bank in 1914, but the Federal Reserve System failed to prevent bank ing panics in the early 1930s. Thus, the public in the United States did not have the experience of observing a central bank successfully dealing with banking pan ics. The last banking panic in the United States (1933) occurred in the same year when the federal govern ment established deposit insurance. I bis observation indicates that federal deposit insurance has been an important feature o f the policies in the United States for preventing banking panics. WHAT ARE BANKING PANICS AND WHY ARE THEY DANGEROUS? Tw o features o f the operation o f commercial banks make the banking system vulnerable to disruptions when depositors lose confidence in their banks. First, 5 FEDERAL RESERVE BANK OF ST. LOUIS a large part o f the liabilities o f banks is payable to depositors on demand. Second, tbe cash reseives of banks are a small fraction o f their deposit liabilities. Thus, if large numbers o f depositors suddenly wanted to convert their deposits to currency, the banking system w ould not immediately have enough cash on hand to honor their demands. W hen a banking panic occurs, people attempt to be among the first to convert their deposits to currency because they remember that during previous banking panics, only those w ho dem anded currency early enough were able to get it.2 Microeconomic Effects o f Banking Panics Deposits and reseives o f the banking system decline one-for-one as depositors w ithdraw currency. If total reserves w ere just equal to required reserves before the withdrawals o f currency, reseives w ould be de ficient after the withdrawals. Each bank responds to its reserve shortage by selling assets, producing a decline in demand deposits that exceeds the initial conversion o f dem and deposits to currency. The vulnerability o f the banking system to panics is illustrated in tables 1 and 2 by the use o f balance sheets. Table 1 presents the hypothetical balance sheet o f an individual bank that is required bv some regulatory authority to keep a cash reserve o f at least 10 percent o f total deposits. Because o f concern about the viability o f the bank, customers w ithdraw $10 m illion in the form o f currency, reducing the bank’s cash reserves to zero. To raise cash reseives, the bank sells $9 m illion o f its interest-earning assets.3 W hen the bank sells its assets to increase its cash reserves, however, it draws cash from other banks, causing their reserves and deposits to decline. These banks must n ow sell some o f their assets to eliminate their reseive deficiencies. Thus, the initial withdrawal o f currency by depositors produces a chain reaction o f reductions in deposits payable on demand. The effects on the banking system o f the currency withdrawals are illustrated in table 2. Prior to the banking panic, the banking system has assets o f $1.1 2Several recent studies develop theoretical models of the behavior of banks and their depositors to investigate the conditions that are likely to cause a banking panic. See Batchelor (1986), Bryant (1980), Diamond and Dybvig (1983), Gorton (1985a), Ho and Saunders (1980), and Jacklin and Bhattacharya (1986). 3lf many banks sell assets at the same time, the prices of bank assets may fall. In that case, the bank would have to sell additional assets and charge losses against net worth. 6 DECEMBER 1986 Table 1 Balance Sheet of an Individual Bank (millions of dollars) Before the banking panic: Assets Cash Interest-earning assets Liabilities $ 10 100 Deposits payable on demand Time deposits Net worth $50 50 10 After withdrawal of $10 million in currency: Assets Cash Interest-earning assets Liabilities $ 9 91 Deposits payable on demand Time deposits Net worth $40 50 10 trillion and deposits payable on dem and o f $500 bil lion. As the banking panic begins, bank customers withdraw $10 billion in currency from their deposit accounts payable on demand, (liven the 10 percent reserve requirement, total deposits must decline until the remaining cash reseives o f $!)0 billion are 10 per cent o f total deposits. This shrinkage in the assets o f the banking system may reduce the confidence o f the public in the bank ing system even more, inducing additional w ith drawals o f deposits in the form o f currency. The addi tional loss o f reserves w ou ld force even larger reductions in bank deposits, interest-earning assets, net worth o f banks and number of banks. Macroeconomic Consequences o f Banking Panics A banking panic causes a sharp reduction in the m oney supply (currency held by the public plus bank deposits payable on demand). Sharp and unexpected reductions in the m oney supply usually cause reduc tions in econom ic activity and, consequently, an in crease in unemployment and business failures. The panic will end when the public becomes convinced that banks are safe and that it can withdraw currency from deposit accounts whenever it wishes. At that time, the public will again deposit part o f its currency with banks. FEDERAL RESERVE BANK OF ST. LOUIS DECEMBER 1986 evolving traditions, w ere implicit rather than spelled out in the Bank’s charter or other legislation. For instance, by the 1800s, the government expected the Bank to buy any part o f its new debt issues not pur chased by others. Table 2 Balance Sheet of the Banking System (billions of dollars) Before the banking panic: Assets Cash Interest-earning assets Liabilities $ 100 1,000 Deposits payable on demand Time deposits Net worth $500 500 100 After withdrawal of $10 billion in currency:' Assets Cash Interest-earning assets Liabilities $ 90 900 Deposits payable on demand Time deposits Net worth $400 500 90 'In the banking panic, sales of interest-earning assets cause the prices of those assets to fall. In this illustration, banks reduce their net worth by $10 billion, recognizing that loss on the sale of assets that had a value of $100 billion before the panic. HOW TO PREVENT BANKING PANICS — THE BRITISH EXPERIENCE How can the failure o f one bank be prevented from spilling over into the w hole banking system with such catastrophic consequences? Only by removing the fear that all banks are in danger o f failing. Can this be done in practice? It can, and the wav to do it was discovered before the theoiy behind the m ethod was developed. The Bank o f England The histoiy o f the British approach to preventing banking panics involves the histoiy o f the Bank o f England. The British government chartered the Bank in 1694 as a means o f raising funds to fight a war with France. Those subscribing to the stock o f the Bank made loans to the British government. In return, the Bank was given some exclusive l ights to function as a commercial bank.4 Although the Bank was privately owned, there was always a close relationship between it and the govern ment. Some aspects o f the relationship, based on “Clapham (1944), Fetter (1965) and Santoni (1984). The Bank o f England maintained a large inventory o f gold upon which it could draw in a panic to meet the public’s demand for gold. Legislation in 1844 gave the Bank a m onopoly on issuing bank notes and made the notes o f the Bank legal tender. That legislation set a limit on the amount o f the Bank’s notes that could be outstanding, though it specified that the limit could be exceeded in an emergency. The limit on the notes o f the Bank could be lifted at the discretion o f the govern ment. Thus, the Bank o f England could expand the monetary base (currency plus reserves) in an em er gency, since its notes were used as currency and were held as part o f bank reserves. One aspect o f the policies that evolved over time was the Bank’s response to a banking panic. The evolution o f that policy is described in this section by discussing fir st, what happened during two banking panics that occurred in England during the 1800s and second, why no panics have occurred in the British banking system since 1866. The Panic o f 1825 In December 1825, a banking panic occurred in London after the failure o f a bank (Sir Peter Pole and Company). As people fled from deposits at other banks to gold, gold reserves w ere drained from the Bank o f England. To convince people that their bank deposits w ere safe, the Bank lent gold freely from its holdings.’’ The panic was allayed when it became clear to the public that there was nothing about which to panic — that there was indeed sufficient gold to meet the public’s increased demand for gold. As a result, the failure o f one bank did not turn into a general financial crisis. Unfortunately, however, banking panics contin ued to occur in England after 1825 because the Bank o f England had not made a public comm itm ent to act as 5The actions of the Bank in the panic of 1825 are described vividly by Jeremiah Harman, director of the Bank, in Bagehot (1978), p. 73: We lent it [gold] . . . by every possible means and in modes we had never adopted before; we took in stock on security, we purchased exchequer bills, we made advances on exchequer bills, we not only discounted outright, but we made advances on the deposit of bills of exchange to an immense amount, in short, by every possible means consistent with the safety of the Bank and we were not on some occasions over nice. Seeing the dreadful state in which the public were, we rendered every assistance in our power. 7 FEDERAL RESERVE BANK OF ST. LOUIS the “lender o f last resort” in all financial crises. A lender o f last resort acts to increase the monetary base if many people want to w ithdraw cash (gold and notes o f the Bank o f England) from their banks. The significance o f a lender o f last resort in a bank ing panic can be illustrated by referring to the balance sheets in table 2. If depositors w ithdraw cash, the lender o f last resort acts to increase the reserves o f the banking system, thus preventing the contraction o f the m oney supply that could be caused by a banking panic. Until the English public became convinced that the Bank o f England w ould act to increase the m one tary base (cash in the hands o f the public plus bank reserves) in financial crises, many o f them tended to withdraw cash from banks when there w ere problems in the financial system. The Lesson o f 1866 The last major banking panic in England occurred in 1866. Since then, although events have occurred that could have triggered banking crises (in 1873,1890, 1907, 1914, 1931 and 19731, they did not do so.1’ What changed after 1866? The panic o f 1866 began with the failure o f a major English bank. Overend, Gurney and Company was a large bank, founded early in the 19th century from the amalgamation o f two banks that had been important and active in the 18th century. Hit by a variety of setbacks, Overend’s was com pelled to seek assistance from the Bank o f England on May 10, 1866. The Bank refused to provide assistance and that afternoon Over end, Gurney and Company was declared insolvent. The next day, there w ere runs on all banks. People scrambled for cash because no bank was trusted.7The Bank o f England, which was being drained o f notes, briefly made things worse by hesitating over w hether to make its usual purchases o f newly issued govern ment debt. By the evening o f Friday, May 11, however, the Bank gave assurance that it w ould provide support to the banking system, and, though demands for small bills continued for a week, the panic was essentially broken in one day* The important consequence of this episode was that the Bank had im plicitly accepted 6A succinct survey of the history of these episodes can be found in Schwartz (1986). 7“ Panic, true panic, came with unexpected violence that day.” (Clapham, 1944), p. 263. “ Terror and anxiety took possession of men's minds for that and the whole of the succeeding day." (Bank ers Magazine, 1866). “ No one knew who was sound and who was unsound.” (The Economist, 1866). 8A detailed description of the failure of Overend, Gurney, and Com pany and the events surrounding that failure can be found in Batchelor (1986). 8 DECEMBER 1986 the responsibility o f acting as lender o f last resort and the public understood that the Bank had accepted that responsibility. For a discussion o f the historical developm ent o f the concept o f a lender o f last resort, see the insert on the opposite page. HOW TO PREVENT BANKING PANICS — THE U.S. EXPERIENCE The U.S. econom y suffered the effects o f banking panics long after the British discovered h ow to prevent them. The United States established the Federal Reserve as the central bank in 1914. There w ere eight major banking panics before then and additional financial crises that had more lim ited regional im pact.” The formation o f the Federal Reserve, however, did not end the problem o f banking panics; the last panic occurred in 1933. The period since the last banking panic coincides w ith the period o f federal deposit insurance. Banking Until the 1860s (Civil War Period)"' Bank Structure and Regulation — After the Revo lutionary War, the new U.S. government confined its monetary role to the minting o f gold and silver coin. State governments assumed responsibility for charter ing and supervising commercial banks. State banks issued bank notes, w hich circulated as currency, and had deposit liabilities against which customers could write checks. Both the bank notes and dem and de posits w ere payable on dem and in the form o f the coins m inted by the federal government. The first banking panic occurred in 1814 during the War o f 1812 with the British. In response to fears about the outcom e o f the war, many people attem pted to redeem bank notes and convert their bank deposits into coin. The banking system responded by suspend ing coin payments, which kept the contractions o f the m oney supply and bank assets from being as large as they w ould have been (see the insert on page 10). In all o f the major U.S. banking panics through 1907, the banking system suspended cash payments to deposi tors and holders o f bank notes. The Panic of 1837 — The panic described above was unusual in that it was triggered by anxieties about the war. Other banking panics in this period occurred 9The nine major banking panics occurred in 1814,1837,1857,1861, 1873,1884,1893,1907 and 1931-33. 10This section is based largely on Hammond (1963). DECEMBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS The Lender of Last Resort and Walter Bagehot The name and the work o f Walter Bagehot recur continually in the discussion o f banking and bank failures.' His main proposal called for the Bank o f England to announce that it was willing to act as lender o f last resort and that it would do so unhesi tatingly whenever necessary. By lender o f last re sort, he meant that the Bank would, in times o f panic, "lend freely, at high interest rates.” It w ould lend freely, so that banks could satisfy the demands o f their customers for cash and thus allay panic. It w ould do so at penally high interest rates to ensure that the Bank was truly the lender o f last resort; banks would com e to it only when the w hole bank ing system was short o f cash. The policy o f setting a high lending rate was designed both to prevent excessive monetary expansion in normal times and to guarantee that banks repaid their borrowings when interest rates dropped after the panic, so that the money stock was not permanently boosted by crisis borrowing. 'Bagehot was, among other things, a journalist. He became editor of The Economist, and wrote voluminously on many subjects (including The British Constitution). But he is most widely remem bered and discussed for his book, Lombard Street (first published in 1873) (Bagehot, 1978), and for his campaign in The Economist to have adopted his recommendations for the conduct of the Bank of England. when bank failures caused the public to lose con fidence in the value o f their bank notes and deposits. The panic o f 1837 shows the nature of panics in this period before the U.S. Civil War. The U.S. econom y experienced an econom ic boom from 1834 through 1836, supported by large invest ments in the United States by Europeans. Many of these large investments w ere in railroads and pur chases o f public land by those moving to the western frontier. The boom stopped in 1837. Gold flowed from the United States to Europe as European investors d e m anded payment o f their loans and liquidated their U.S. investments. This outflow o f gold reduced the cash reserves o f banks, which, along with failures by business firms, caused some banks to fail. The Dry Dock Bank, a major bank in N ew York City, closed on Bagehot em phasized that the Bank should not only behave in this way, but also should announce in advance that it w ould do so whenever necessaiy. He saw this “precom m itm ent,” w hich the Bank had never made before the episode o f 1866, as vital. A credible precom m itm ent w ould give assurance that sound banks w ould not be allowed to fail as a result o f the failure o f some other bank. Once this assurance was given, panic w ould be less likely. Indeed, the Bank might not actually have to act as lender o f last resort at all; m erely standing ready to do so might be sufficient to provide stability.2 Although now traditional, Bagehot s recommendation was not accepted without demur. Thomson Hankey, a director of the Bank, was particularly critical of the proposal. After the Overend and Gurney affair, Hankey denied that the Bank had an unequivo cal duty to lend freely in panics. He was concerned with what has become known as “ moral hazard." If bankers know that the central bank will lend freely in a panic, he argued, they will take more chances: hold lower reserves, make riskier loans or pay higher dividends. Hankey is plainly correct. The issue, however, is which is the lesser evil, slightly riskier banks or the prospect of a collapse in the money stock. Hankey's criticism of Bagehot's principles for running a central bank did not represent the official views of the Bank. Officially the Bank neither accepted nor rejected Bagehot’s principles but came to act in a manner consistent with these principles. May 8, 1837. All other banks in N ew York City experi enced runs by depositors the next day. The N ew York City banks suspended coin payments on May 10, and Philadelphia banks follow ed suit on May 11. Within the next 10 days, banks in all the leading cities sus pended coin payments. N ew York City banks resumed coin payments to holders o f bank notes and deposits on May 10, 1838, exactly oneyear after the suspension. Banks in the rest o f the nation resumed coin payments between August 1838 and early 1839. This episode illustrates the vulnerability o f the banking system to disruption. Without a central bank, the supply o f cash in the econom y was determined by the coins minted by the federal government and inter national movements o f precious metals. The bank runs following the failure o f the Dry Dock Bank 9 FEDERAL RESERVE BANK OF ST. LOUIS DECEMBER 1986 Suspending Cash Payments Commercial banks in the major urban areas in the United States suspended payments o f cash to depositors and note holders during each o f the major banking panics from 1814 through 1907. Dur ing suspensions, banks in an area w ould agree to act together in refusing to pay out cash. Until the 1860s, the form o f cash dem anded by depositors during panics was metal coins. In the early 1860s, the cash demand during panics included currency. During these general suspensions o f cash pay ments, banks remained open for business and per mitted their customers to use deposits to make payments to others. Checks w ere cleared and de posit liabilities were transferred among banks. N on redeem ed bank notes continued to circulate as currency. When banking panics w ere over, banks showed that the public’s confidence in banks could be undermined quickly when an important bank went under. At this time, however, the U.S. banking system had no institution comparable to the Bank o f England, which had a reputation for financial strength and an inventory o f cash that could cut short a panic. Conse quently, some banking panics in the United States, like the panic o f 1837, w ere follow ed by long periods o f suspended cash payments. Money and Banking from the 1860s to the Formation o f the Federal Reserve System in 1914" The National Banking System — Reforms were begun in the 1860s to achieve two purposes: to estab lish a national currency, with all currency accepted at par value in exchange throughout the nation, and to make the banking system less vulnerable to panics. As a first step, the federal government began chartering national banks whose notes w ere to be the prim aiy national currency. National banks w ere required to hold both collateral with the Treasury Department against their notes as w ell as cash reserves that w ere a percentage o f their deposit liabilities and notes. The collateral and reserve requirements w ere im posed to "This section is based largely on Cagan (1963), Robertson (1964), pp. 302-30, Scroggs (1924), and Sprague (1910). 10FRASER Digitized for resumed the payment on dem and o f coin for bank notes and deposits. The significance o f the suspension o f bank pay ments in the form o f coin or currency can be illus trated by referring to table 2, the balance sheet o f the banking system. As soon as bankers realized that the public was attempting to convert its m oney holdings to coin or currency, the bankers as a group refused to meet this demand for cash. In this case, their cash reserves w ould stay at $100 billion, and with that cash available to meet reserve require ments, the banking system w ould not have to sell assets and reduce its liabilities. By suspending cash payments, the banks w ould prevent the contraction of the assets and liabilities o f the banking system. restrain the growth o f bank liabilities and to prom ote greater public confidence in the banking system. The basic flaw in the design o f the new system was the absence o f a central bank w ith the p ow er to in crease the monetary base should the public lose con fidence in the value o f bank deposits. In this period, there w ere major banking panics in 1873, 1884, 1893 and 1907. Banks acted cooperatively during these pan ics to increase their reserves by creating clearing house loan certificates (see the insert on the opposite page). The creation o f clearing house loan certificates, however, did not permit banks to meet the public demand for currency. During each o f these panics they also suspended payments o f coin and currency to depositors. The Panic of 1907 — The nature o f banking panics in this period can be illustrated by the panic o f 1907, which occurred in October and Novem ber o f that year. This panic is interesting for several reasons. Its effects on the nonfinancial sectors o f the econom y w ere rela tively severe, and its events provide a good illustration o f h ow the loss o f public confidence in one bank can lead to loss o f confidence in the banking system. Finally, political reaction to this panic led to the for mation o f the Federal Reserve System. For several years prior to 1907, gold flow ed from Europe to the United States because Europeans in vested heavily in the U.S. econom y. In the fall o f 1906, FEDERAL RESERVE BANK OF ST. LOUIS DECEMBER 1986 Increasing the Monetary Base by Creating Clearinghouse Loan Certificates The banking system attempted to cope with the banking panics through emergency actions other than the suspension o f cash payments. One ap proach involved the issuance o f clearinghouse loan certificates. Clearinghouses were cooperative insti tutions established by banks to econom ize on the resources used in clearing checks and notes among themselves. Checks and notes drawn on other members o f a clearinghouse were presented to the clearinghouse for collection, rather than being sent to each bank for collection. Banks deposited cash with their clearinghouses and, in turn, received cer tificates that were accepted by the other banks for covering net debit positions in clearinghouse settle ments. Originally, the clearinghouse certificates were merely receipts for cash held by clearing houses. In each o f the banking panics from 1857 through 1907, the banking system acted cooperatively to expand the m onetaiy base by increasing the amount o f clearinghouse certificates outstanding. The certificates issued during the panics w ere not simply receipts for cash held by the clearinghouses but w ere loans to banks with their assets pledged as collateral. The standard practice was for a clearing house association to accept some o f the assets o f a bank as collateral and issue certificates payable by the clearinghouse association. A bank that received such certificates w ould pay interest on them until they w ere redeem ed after the panic was over. These European investors began liquidating their U.S. invest ments, resulting in large gold outflows from the United States. This disinvestment was associated with sharp drops in U.S. stock prices in March and August 1907. The U.S. econom y went into a recession after May 1907; the rate o f decline in real econom ic activity was relatively low until the banking panic in the fall o f that year, but relatively rapid after the panic. The Panic of 1907 began with a depositor run on the Mercantile National Bank in N ew York City, which had suffered large losses. The N ew York clearinghouse transactions effectively converted the clearing houses into fractional reserve banking institutions by creating certificates that exceeded their holdings o f cash. Initially, clearinghouse certificates circulated only among banks that w ere members o f the clear inghouse. During panics, clearinghouse members agreed to accept these certificates in payment from other member banks, rather than insisting on pay ment in cash. The circulation o f these certificates among banks in place o f cash payments allowed the participating banks to use their cash to meet the cash demands o f their depositors. In some panics, however, clearinghouse associations issued certifi cates in small denominations that banks offered to their depositors as substitutes for cash. These small-denomination certificates then circulated as currency. This use o f clearinghouse certificates was not legal, but the government banking authorities did not challenge their use during panics.1 'Gorton (1985a,b) and Timberlake (1984). The Aldrich-Vreeland Act of 1908 established a procedure for the emergency issuance of national bank notes that was copied after the procedures that banks had used for issuing clearinghouse loan certificates during panics. The panic of 1914 tested the effectiveness of this innova tion just before the Federal Reserve System began operations. By issuing notes that were available for such an emergency, banks did not have to suspend cash payments. See Cagan (1963), pp. 26-28, and Sprague (1915). came to the aid o f the Mercantile National Bank in October 1907, after the bank was put under new man agement. This action, however, was insufficient to stem the panic. Depositor runs began at other institu tions, reflecting a general loss o f public confidence in the stability o f the banking system. Within a few days, all depository institutions in N ew York City faced depositor runs. Banks in N ew York City suspended cash payments in Novem ber 1907, and the suspension o f payments spread quickly to other cities. The panic and suspension o f payments ended in early 1908, but only after a sharp decline in econom ic activity and a 11 DECEMBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS rise in bankruptcies in the nonfinancial sectors o f the economy. Banking Crises o f the 1930s and the Beginnings o f Federal Deposit Insurance'2 After the long series o f banking panics, culminating in the Panic o f 1907, Congress finally responded by passing legislation in 1913 to establish the Federal Reserve System. There w ere no banking panics from 1914, w hen the Federal Reserve System began its oper ations, until the early 1930s. Then, however, a series o f banking crises resulted in the closing o f all banks in the nation in March 1933. The Federal Reserve did not respond to these bank ing crises as the Bank o f England had nearly 70 years before. U.S. commercial banks came under liquidity pressures because o f cash withdrawals by depositors and outflows o f gold from the United States. Yet, ex cept for a few months in 1932, the Federal Reserve did not increase the monetary base in response to these cash withdrawals from banks. Moreover, commercial banks did not act cooperatively to suspend cash pay ments to depositors as they had in earlier banking crises.13 Consequently, the deposit liabilities o f the banking system declined sharply. Congress took various approaches to dealing with the general collapse o f the banking system in the 1930s. The most significant legislation was the estab lishment o f federal deposit insurance. There have been no general banking panics in the United States since 1933. HAS DEPOSIT INSURANCE PREVENTED BANKING PANICS IN THE UNITED STATES? The Pros and Cons o f Federal Deposit Insurance Recent experience indicates that large numbers o f bank failures do not induce nationwide banking pan ics. A controversial issue, however, is w hether federal deposit insurance could be elim inated without under mining public confidence in the banking system. ,2This section is based largely on Friedman and Schwartz (1963). ,3Friedman and Schwartz (1963, pp. 311 -12 ) argue that banks did not suspend cash payments because they thought the need to do so had been eliminated by the establishment of the Federal Reserve. 12FRASER Digitized for The British solved the problem o f banking panics m ore than 100 years before they adopted a program of deposit insurance administered by the governm ent.14 Some argue that it is time to eliminate federal deposit insurance in the United States.15In their view, banking panics are best prevented by a credible len der o f last resort, and they argue that the Federal Reserve has learned h ow to function as such. Federal deposit insurance gives depository institutions the incentive to assume greater risk than if deposit insurance w ere eliminated or offered by private firms. An opposing view is that federal deposit insurance is essential for preventing banking panics. Since fed eral deposit insurance has been in effect for over 50 years, depositors rely on it, rather than on their assess ment o f the financial condition o f their banks. In this view, banks w ou ld be vulnerable to runs by depositors as they had been prior to 1933 if federal deposit insur ance w ere cancelled. The Importance o f Credible Insurance: The S&L Experience Developments in Ohio and Maryland in 1985 pro vide some evidence on the importance o f federal d e posit insurance in preventing banking panics in the United States. The deposits o f 80 Ohio savings and loan associations (S&.Ls) had been insured by the Ohio Depository Guarantee Fund (ODGF), a private deposit insurance fund founded by the S&.Ls themselves to insure their deposits. On March 4, 1985, the largest S&.L insured by the ODGF incurred losses because o f the failure o f a government securities dealer with which the S&L had large investments. These losses exceeded the capital o f the S&.L and the entire reserves o f the ODGF. When these events w ere publicized, depositors at other ODGF-insured S&.Ls began to withdraw their deposits. Their confidence in the safety of their funds was destroyed when the reserves o f the ODGF w ere w iped out.1" Eleven days later, the governor ordered all o f the S&.Ls insured by the ODGF closed. One o f the conditions for reopening was that 14The British program of deposit insurance was introduced under the Banking Act of 1979. With few exceptions, all depository institutions are covered and must contribute to an insurance fund. Coverage is 75 percent of each account, with a maximum compensation of £10,000 for each depositor. This program was introduced in re sponse to the secondary banking crisis of the early 1970s. 15Ely (1985), England (1985), O’Driscoll and Short (1984), Short and O’Driscoll (1983), and Wells and Scruggs (1986). 16Federal Resen/e Bank of Cleveland (1985). FEDERAL RESERVE BANK OF ST. LOUIS they obtain federal insurance for their deposits.17 The loss o f confidence in the ODGF-insured institu tions did not lead to a general loss o f confidence in depositoiy institutions. There were no runs on feder ally insured banks or S&.Ls in Ohio. Similar events transpired in M aiyland in May 1985. A private fund insured the deposits o f 102 Maryland S&Ls. Losses at the largest S&L insured by the private fund triggered runs by depositors on the privately insured S&Ls throughout the state. Once again, there were no runs on federally insured institutions. The Maryland state government required the privately in sured S&.Ls to obtain federal deposit insurance. In reaction to these developments in Ohio and Maryland, several other states have required their privately in sured thrift institutions to obtain federal deposit in surance coverage. DECEMBER 1986 bank with the responsibility o f acting as the lender of last resort if a banking panic occurred. The Federal Reserve failed in that responsibility in the early 1930s, which resulted in a nationwide banking panic in the United States in 1933. There have been no banking panics in the United States since the federal govern ment established deposit insurance in the 1930s. Runs by depositors on privately insured savings and loan associations in Ohio and Maryland during 1985 pro vide some evidence that federal deposit insurance is an essential feature o f the policies in preventing bank ing panics in the United States. REFERENCES Bagehot, Walter. Lombard Street, reprinted in Norman St. JohnStevas, ed., The Collected Works of Walter Bagehot (Hazell Wat son and Viney Ltd., 1978). Bankers Magazine, June 1866, pp. 45-46. SUMMARY AND IMPLICATIONS FOR THE UNITED STATES The rate o f bank failure in the United States is currently high relative to failure rates in most years since W orld War II. There have been many episodes in U.S. histoiy when increased bank failures led to bank ing panics that disrupted the operation of the nation’s banking system. To prevent banking panics, it is essential that the public maintain confidence in the safety o f their d e posits even though som e banks are failing. In the United Kingdom, public confidence in the stability of the banking system was established through the com mitment o f the Bank of England to act as the lender of last resort in financial crises. This policy was estab lished in the banking panic o f 1866, and the: U.K. banking system has not experienced a banking panic since. The basic feature o f that policy involves a com mitment to increase the monetary base (currency held by the public plus bank reserves) when bank runs occur. Policies in the United States reflect a different histor ical development. After various banking panics, the Federal Reserve was established in 1914 as the central ,7The Federal Reserve attempted to stop the depositor runs by lending cash to the privately insured S&Ls. Federal Reserve em ployees from throughout the System were put on special assign ment to accept the assets of these S&Ls as collateral for the cash loans. This response, however, did not stop the depositor runs. This indicates that, in a nation in which depositors have come to rely on deposit insurance to maintain their confidence in the safety of their funds, the central bank may not be able to maintain that confidence by lending cash to depository institutions when the protection of deposit insurance is suddenly eliminated. Batchelor, Roy A. “The Avoidance of Catastrophe: Two Nineteenthcentury Banking Crises,” in Forrest Capie and Geoffrey E. Wood, eds., Financial Crises and the World Banking System (St. Martin’s Press, 1986), pp. 41-73. Bryant, John. “ A Model of Reserves, Bank Runs, and Deposit Insurance,” Journal of Banking and Finance (December 1980), pp. 335—44. Cagan, Phillip. “The First Fifty Years of the National Banking Sys tem — An Historical Appraisal,” in Deane Carson, ed., Banking and Monetary Studies (Richard D. Irwin, 1963), pp. 15—42. Clapham, Sir John. The Bank of England: A History (Cambridge University Press, 1944). Diamond, Douglas W., and Philip H. Dybvig. “ Bank Runs, Deposit Insurance, and Liquidity,” Journal of Political Economy (June 1983), pp. 401-19. The Economist, June 1866, pp. 15-16. Ely, Bert. “Yes — Private Sector Depositor Protection is a Viable Alternative to Federal Deposit Insurance!” in Federal Reserve Bank of Chicago, Proceedings of a Conference on Bank Structure and Competition, held in Chicago, May 1-3, 1985, pp. 338-53. England, Catherine. “A Proposal for Introducing Private Deposit Insurance," in Federal Reserve Bank of Chicago, Proceedings of a Conference on Bank Structure and Competition, held in Chicago, May 1-3, 1985, pp. 316-37. Federal Reserve Bank of Cleveland. 1985 Annual Report. Fetter, Frank Whitson. Development of British Monetary Orthodoxy (Harvard University Press, 1965). Friedman, Milton, and Anna J. Schwartz. A Monetary History of the United States, 1867-1960 (Princeton University Press, 1963). Gorton, Gary. “ Bank Suspension of Convertibility,” Journal of Mon etary Economics (March 1985a). ________ _ “ Clearinghouses and the Origin of Central Banking in the United States,” The Journal of Economic History (June 1985b). Hammond, Bray. “ Banking before the Civil War,” in Deane Carson, ed., Banking and Monetary Studies (Richard D. Inwin, 1963), pp. 1 14 . Ho, Thomas, and Anthony Saunders. “A Catastrophe Model of Bank Failure,” The Journal of Finance (December 1980), pp. 1189— 207. 13 FEDERAL RESERVE BANK OF ST. LOUIS DECEMBER 1986 Jacklin, Charles J., and Supipto Bhattacharya. “ Distinguishing Pan ics and Information-Based Bank Runs: Welfare and Policy Implica tions,” First Boston Working Paper Series 86-16 (February 1986). Short, Eugenie D., and Gerald P. O’Driscoll, Jr. “ Deregulation and Deposit Insurance,” Federal Reserve Bank of Dallas Economic Review (September 1983), pp. 11-22. O’Driscoll, Gerald P., Jr., and Eugenie D. Short. “ Safety-Net Mech anisms: The Case of International Lending,” Cato Journal (Spring/ Summer 1984), pp. 185-204. Sprague, O. M. W. History of Crises under the National Banking System, National Monetary Commission, Sen. Doc. No. 538, 61 Cong. 2 Sess. (U.S. Government Printing Office, 1910). Robertson, Ross M. History of the American Economy, 2nd ed. (Harcourt, Brace and World, 1964). Santoni, G. J. “A Private Central Bank: Some Olde English Les sons,” this Review (April 1984), pp. 12-22. Schwartz, Anna J. “ Real and Pseudo-financial Crises” in Forrest Capie and Geoffrey E. Wood, eds., Financial Crises and the World Banking System (St. Martin’s Press, 1986), pp. 11-31. Scroggs, William O. A Century of Banking Progress (Doubleday, Page and Company, 1924). 14FRASER Digitized for _________ “The Crisis of 1914 in the United States,” American Economic Review (September 1915), pp. 499-533. Timberlake, Richard H., Jr. “The Central Banking Role of Clearing house Associations,” Journal of Money, Credit and Banking (Feb ruary 1984), pp. 1-15. Wells, Donald R., and L. S. Scruggs. “ Historical Insights into the Deregulation of Money and Banking,” Cato Journal (Winter 1986), pp. 899-910. FEDERAL RESERVE BANK OF ST. LOUIS DECEMBER 1986 Why Has Manufacturing Employment Declined? u John A. Tatom NITED STATES manufacturing em ploym ent grew little in 1986. Currently at about 19 million workers, it is below the 21 million em ployed at its peak in 1979. This disappointing performance often is at tributed to the declining international com petitive ness o f U.S. manufacturing. Such arguments, however, are tenuous at best: U.S. manufacturing output ex panded more rapidly during the period o f dollar ap preciation fro in 1980-84 than it had over the previous four years when the dollar’s value was falling. More importantly, the growth o f manufacturing abroad has been anemic during this decade. A variety o f output, cost and productivity measures reveal that the competitiveness o f U.S. manufacturing has actually im proved.1 Concern over the recent performance o f manufac turing employment, however, is not so easily rebutted. Indeed, viewed alongside the strength o f U.S. manu facturing output growth, there seems to be a “JekyllH yde” quality to the U.S. manufacturing sector perfor mance.- A longer-run perspective on manufacturing employment and an understanding o f econom ic forces contributing to it, however, reveals that the recent decline is not unusual and simply reflects the John A. Tatom is an assistant vice president at the Federal Reserve Bank of St. Louis. Michael L. Durbin provided research assistance. 1See Tatom (1986). Clark (1986) has pointed to the unusual strength of manufacturing output in recent years. 2See Clark (1986). strength o f U.S. manufacturing productivity growth in the 1980s. DOMESTIC MANUFACTURING EMPLOYMENT: CYCLICAL W ITH NO PERSISTENT TREND Chart 1 shows manufacturing em ployment and out put (1982 prices) since 1948. As one can see by examin ing the shaded periods o f business recession, both manufacturing em ployment and manufacturing out put are strongly cyclical. What is equally evident is that manufacturing em ployment has shown little ten dency to grow over the prior three decades, except for its sharp rise from 1960 to 1967. Indeed, at its peak in 1979, there were fewer than one m illion more workers in the manufacturing sector than in mid-1969, and only about four million more workers than in 1956 and early 1957. Thus, temporarily negative growth in man ufacturing em ployment is neither unprecedented, nor should it be assessed relative to a presumption that manufacturing em ployment has exhibited any signifi cant growth since 1948. The cyclical explanation, however, does not fully account for the decline in em ployment from 1979 to 1986. At manufacturing em ploym ent’s peak in 1979, unemployment equaled 5.8 percent o f the civilian labor force. If the nation’s output increased enough to reduce the current unem ploym ent rate (7.0 percent) back to 5.8 percent, about 1.4 m illion jobs w ould result, given today’s labor force. Up to one-half o f these jobs w ould likely be in manufacturing. Even w ith these 15 FEDERAL RESERVE BANK OF ST. LOUIS DECEMBER 1986 C hart 1 U.S. M anufacturing O u tpu t and Employment Latest d a t a p lo t t e d : 3 r d q u a r te r 1986 additional jobs, however, manufacturing em ployment w ould remain low er than in 1979.:‘ WHAT DETERMINES MANUFACTURING EMPLOYMENT? Economic theoiy points to several factors that in fluence manufacturing employment. At the simplest level, firms choose their desired em ploym ent o f labor based on a comparison o f the expected cost and the expected revenue obtained from hiring additional workers. The latter depends on both the change in output associated with em ploying more (or less) work 3The appendix to this article further discusses the importance of cyclical movements in the decline of manufacturing employment since 1979. Digitized for 16 FRASER ers and the expected output price. Another w ay o f expressing this choice is to compare the relative price o f labor, the wage relative to the price o f the output produced, and the productivity o f additional workers.4 A rise in the manufacturing wage or a fall in the price o f manufactured goods raises the cost o f labor relative to its productivity, reducing the incentive to em ploy labor. Similarly, a rise in the productivity o f workers for a given level o f em ployment increases the incentive to em ploy workers, given the relative cost o f labor. “The relevant productivity measure is the marginal product of labor; normally, however, output per worker, or average productivity, is the most commonly used measure. As long as the ratio of the marginal to average product of labor does not change, movements in the average product of labor will reflect the same proportional move ments in the marginal product of labor. FEDERAL RESERVE BANK OF ST. LOUIS The manufacturing sector is only one part o f the econom y. Producers o f m anufactured products, therefore, must com pete with producers in other sec tors, such as agriculture, services, construction, m in ing, transportation, utilities and government, for sales and for resources, including workers. Thus, manufac turing wages and prices must be competitive in order to attract workers and sales. A simple statement o f this relationship can be derived from the identical em ploy ment decisions made by firms throughout the econ omy. In particular, if wages equal some fraction ((3mfor manufacturing, or (3 for the w hole econom y) o f the revenue per worker in manufacturing and in the w hole economy, then: (1) W,„/W = (0„,/0) (P„,/P) (TT../TT), where W,„ and W are wages in manufacturing and in the w hole economy, respectively, P,„ and P are the prices o f output in the two sectors, and tt,,, and i t are the output per worker, or productivity, in the respec tive sectors. Because productivity is measured as the ratios o f output to the number o f workers in each sector, equation 1 can be rearranged to the following: 12! L„/L = ip„,/(3l (X„/X) (P,„/P) iWm/Wr', where L,„ is the em ployment in manufacturing and L is total civilian employment, and X„, and X represent their respective output levels. According to equation 2, the share of manufacturing em ploym ent (L„/L) de pends positively on the share o f manufacturing in the nation’s total output (X,,,/X) and the price o f manufac turing output relative to prices generally (P„,/P), and is inversely related to wages in manufacturing relative to wages generally (W,„/W). Relative wages, o f course, depend on relative skill differences, nonpecuniaiy dif ferences o f jobs in manufacturing com pared with the remainder o f the economy, and barriers to labor m ove ment across sectors o f the econom y. Differences in the relative degree of unionization or in regulation can affect the latter factor. Manufacturing output’s share in total output de pends on the dem and for manufacturing output com pared with other goods. This demand is influenced by permanent or transitoiy movements in real income and by the relative price o f manufactured product. The share o f manufacturing product in total output can also be influenced by international trade. Low er prices for im ported manufactured products could re duce both the share o f domestic manufacturing pro DECEMBER 1986 duction and its relative price. Similarly, a rise in the relative price o f manufactured goods due to a rise in foreign demand can increase domestic manufacturing production (for export) relative to the econom y’s total output. Manufacturing output's share is o f interest not just because o f its influence on employment; more im por tantly, it indicates the direct role o f manufacturing in generating real incom e in the economy. In addition, comparisons o f the em ployment and output shares o f the manufacturing sector indicate the relative perfor mance o f productivity, or output per worker. The next section examines the em ployment and output shares in the manufacturing sector. Then the implications of productivity growth for prices and output are dis cussed. The discussion links two o f the three factors influencing the em ployment share, according to equa tion 2. The third factor, relative wages, is discussed subsequently. THE SHARE OF MANUFACTURING EMPLOYMENT AND OUTPUT Chart 2 shows the share o f manufacturing em ploy ment and output as percentages o f civilian em ploy ment and real gross national product (GNP) respec tively. The share o f m anufacturing output has fluctuated cyclically, but shows no trend. Em ploy ment in manufacturing has been declining as a share o f total em ployment for a long time. The principal factor accounting for this decline has been relatively more rapid growth in labor productivity in manufac turing than in the remainder o f the economy. Chart 3 shows the ratio o f labor productivity in manufacturing to that for the business sector as a whole. Labor productivity is measured by output per worker. From 1948 to 1960, there was little difference in the growth rates o f productivity in manufacturing and elsewhere, so the relative productivity level shown in the chart changed little. Note that in chart 2, the share o f labor em ployment in manufacturing also changed little over this period. Since then, productiv ity has grown faster in the manufacturing sector, so that between 1960 and 1985, labor productivity in manufacturing increased almost 50 percent more in the manufacturing sector than in the business sector. As chart 2 shows, this rise in productivity was associ ated with a decline in the share o f labor employment rather than a rise in the share o f manufacturing output. 17 DECEMBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS C hart 2 Share of Employment and O u tpu t in M an u factu rin g 1947 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83 1985 NOTE: P e r c e n ta g e s o f c iv ilia n e m p l o y m e n t a n d r e a l G N P , re sp e ctiv e ly . Productivity Growth, Prices and Output Why have productivity gains in manufacturing re sulted in a relative decline in em ployment rather than a rise in the share o f output? A simple perspective on this question is to examine the effect o f productivity growth in a supply-demand framework. In figure 1, the initial supply curve and demand curves are la beled S and D, respectively. Given other factors that influence supply or dem and decisions, the curves indicate that as the price o f manufactured product rises, the quantity supplied rises and the quantity dem anded falls. At the initial equilibrium price, P,„ producers desire to produce and sell exactly the quantity of product that buyers wish to purchase. A gain in output per worker, or productivity, raises the quantity that producers could profitably produce, given factor and product prices. Such a gain shifts the Digitized for 18 FRASER supply curve to the right, as shown in the shift from S to S' in the figure. The shift in the supply results in an excess supply.’ Buyers are unwilling to purchase more, given the price, P„, and the other factors in fluencing demand. Thus, the product price falls as producers com pete to enlarge their sales. At a new equilibrium price, P, in the figure, buyers purchase more and sellers are selling exactly the output they Productivity growth in manufacturing also has a significant effect on real GNP since this sector accounts for more than 20 percent of real GNP. For example, a 10 percent increase in output per worker would tend to increase real GNP by (0.2) (0.1) or 2 percent, other things the same. This change in real GNP would raise the demand for all normal goods and sen/ices. This shift is omitted in the figure. The initial excess supply created by a productivity improvement in manufacturing is reduced somewhat by this shift, as is the associ ated decline in price. DECEMBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS C hart 3 Relative Productivity in M anufacturing profitably choose to sell along the new supply cuive S'. Thus, productivity growth increases output only to the extent that buyers are willing to increase their purchases; this willingness is influenced by the re sponsiveness o f dem and to a decline in the price of the product. The effect o f productivity growth on the size o f the output increase in an industry is determ ined by pur chasers o f the product, not by the producers. If d e mand is quite responsive to price, then price falls relatively less and the quantity purchased rises rela tively more. Economists refer to this responsiveness as the “own price elasticity o f dem and” ; it measures the percentage change in quantity dem anded induced by a given percentage change in price. If the elasticity equals one, a given percentage-point decline in price induces an equal percentage rise in the quantity d e manded. If the elasticity exceeds one, the product is said to have elastic demand; a given percentage d e cline in price induces a larger percentage rise in quan tity demanded. If the own price elasticity o f demand is less than one, demand is said to be inelastic, indica ting a low er degree o f responsiveness o f dem and to price changes. An important implication o f the magnitude o f the dem and response to a price change is the effect o f a supply shift on total spending on the product. When supply shifts from S to S' in the figure, the product o f price times quantity, or total spending on the product, can change. If dem and is elastic, the percentage rise in 19 FEDERAL RESERVE BANK OF ST. LOUIS quantity dem anded will exceed the percentage d e cline in price that caused it; as a result, total spending (P, X,) w ill rise (P, X, exceeds P„ XJ. If dem and is unit elastic, total spending w ill not change. If dem and is inelastic, the price w ill fall relatively more than quan tity dem anded rises and total spending falls. DECEMBER 1986 Figure 1 The Supply and Dem and for M anufacturing Output Implications f o r the Manufacturing Sector The estimated dem and for manufacturing output shown in the appendix has a price elasticity that is less than one, or inelastic. Thus, according to equation 2, faster productivity growth in manufacturing has resulted in a declining share o f em ployment because relative price reductions have more than offset the price-induced gains in output Relatively faster productivity growth in manufactur ing also has reduced the share o f nominal incom e generated in manufacturing products. In effect, the gain in the nation’s incom e and output occasioned by productivity growth in manufacturing has been real ized in increased output elsewhere. To the extent that consumers o f manufactured and other products are unwilling to buy the increased manufacturing output, resources that are saved by productivity improvement are moved into other activities to produce goods or services. The rise in the price o f nonmanufactured product relative to prices o f manufactured goods reflects this shift. Moreover, the share o f incom e spent on the manufactured product declines, or the share of income spent on other products rises.7 The relative price o f the manufactured product is shown in chart 4; it is the ratio o f the implicit price deflator for manufacturing output to that for business sector output, where the price indexes are set to 1 in 1982. The share o f nominal GNP originating in domes- 6The price elasticity is not the only factor that influences the share of spending on manufacturing output. The “ income elasticity," the sensitivity of demand to real income changes, is also an important determinant of the share of such output and spending in a growing economy. As real income expands, the demand for all goods and services normally rises, given unchanged prices. But if the income elasticity of demand for manufactured product is less than one, then the share of manufacturing output in total output would fall, given unchanged product prices. This elasticity, with respect to permanent income, is estimated to be less than one in the appendix. Transitory or cyclical changes in income have much larger effects. T h e agricultural sector is a more well-known area in which produc tivity gains have given rise to sharp increases in the nation’s real income, despite a declining share of income being spent on the product and relatively large flows of resources out of the sector. 20FRASER Digitized for tic manufacturing is also shown in chart 4. The decline in the relative price o f manufacturing output since 1960 has been quite rapid and reflects the relative gain in labor productivity in that sector." Since the propor tion of output has been unchanged (chart 21, the share o f incom e originating in or spent on manufacturing has declined in line with the falling relative price o f manufactured product. Tw o o f the principal factors determining the share o f labor employment devoted to manufacturing in equation 2 are summarized in the nominal spending share in chart 4. The dominant factor o f the two has been the declining relative price o f manufacturing output, which reflects relative productivity gains in the sector. Of course, its share o f output and its rela tive price could both fall if the demand for manufac- 8The sharp decline in the relative price of the manufactured product from 1971 to 1973 and subsequent recovery to its previous path may be due to errors in measurement. Darby (1974) has argued that wage and price controls in this period initially biased down price measures and artificially raised real output measures. If wage and price patterns in 1971-75 were artificially distorted by controls, the share of employment (chart 2) would not have been so flat in 1971— 73, nor would it have subsequently declined so sharply in 1973-75. DECEMBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS C h a rt 4 The R elativ e Price a n d Share of N o m in a l Income in M a n u fa c tu rin g Index Per cent A n n u a l D ata 31.0 Price |SCALE tured goods w ere declining. Chart 2 clearly indicates, however, that this has not been the case; the share o f manufacturing output has been nearly unchanged for the past 40 years.0 RELATIVE WAGES AND EMPLOYMENT IN MANUFACTURING The final factor in equation 2 that influences the share o f em ployment in manufacturing is the relative level o f compensation in manufacturing. W hen wages 9ln agriculture, even the share of output has declined, making it more difficult to see the sector as an important source of expanding real income. rise more (less) in one sector relative to the rest o f the economy, the relative amount o f em ployment gener ally is reduced (increased), given initially unchanged relative price and output levels. One w ay to under stand this makes use o f equation 1. If relative wages in manufacturing rise, it either reflects a relative im provement in the value o f manufacturing productivity for a given level o f em ployment or will be reflected in such an improvement obtained by changing em ploy ment."’ In the latter case, a rise in wages relative to prices forces firms to both substitute other factors of ,0That is, the relative employment demand depends on relative wages. If relative wages change, there is either a movement along, or a shift in, the relative demand for labor in manufacturing. 21 FEDERAL RESERVE BANK OF ST. LOUIS DECEMBER 1986 C h a rt 5 Relative Labor Compensation in M anufacturing production for labor to offset some o f the cost increase and to reduce production, which tends to raise prod uct prices. Both types o f adjustment raise productiv ity, but output declines and product prices rise when the source o f the productivity gain is an increase in relative wages. Relative wage movements have not been the dom i nant force in U.S. manufacturing. Chart 5 shows com pensation in manufacturing relative to compensation in the business sector generally. Over the past 38 years, there has only been one major shift in the relative compensation levels that w ould induce a ma jor change in relative output, price or employment patterns. From 1948 to 1960, compensation was over 20 percent higher in the manufacturing sector. This differential narrowed from 1960 to 1966, resulting in em ployment growth that was quite rapid (charts 1 and 22FRASER Digitized for 2). With the exception o f that period, however, m ove ments in relative wages do not appear to have been large enough to have affected the share o f labor em ployed in manufacturing significantly. INTERNATIONAL TRADE The view that foreign com petition has led to rela tively large losses in manufacturing em ploym ent in the 1980s is w idely held. But there is no evidence above that the share of domestic manufacturing (chart 2) has been depressed by the appreciation o f the dollar or by increased imports." There is also no apparent "Fieleke (1985) has shown that there was no significant negative correlation between employment changes in domestic employment in manufacturing industries and changes in import penetration in these industries over the period 1980 to 1984. FEDERAL RESERVE BANK OF ST. LOUIS evidence that relative wages in manufacturing (chart 5) have been depressed in the early part o f this decade due to trade-induced reductions in the demand for U.S. manufacturing output and employment. More careful attention to the argument w ould further clarify the analysis, however. Domestic manufacturers com pete with foreign pro ducers. The dollar price o f domestic manufactured product, therefore, must be competitive with the dol lar price of the foreign product. The latter price can be expressed as (P*/EI, where P* is the price of the foreign product in its own currency and E is the price o f a dollar in units o f foreign currency. In the analysis in the figure, productivity improvement lowers the price of domestic product; for foreign goods, this requires that the value o f the dollar, E, rise; to the same extent for foreign goods to remain competitive with U.S. products. In other words, productivity improvement in U.S. manufacturing, given foreign prices, tends to raise the value o f the dollar. Many analysts, however, em phasize the causality running in the opposite direction. Falling prices o f foreign goods or a rise in the value o f the dollar depress the domestic prices o f foreign goods. CM' course, a decline in P,„ due to foreign competition alone w ould lead to a reduction in the quantity o f U.S. output supplied and increased purchases along tin; demand curve; the difference between U.S. purchases o f manufactured products and U.S. production (sup ply) w ould be made up by imports o f foreign products. The evidence presented earlier is inconsistent with the trade hypothesis. If this hypothesis w ere correct, the share o f domestic manufacturing output in total real incom e w ould have fallen in the 1980s. Instead, the share has been relatively strong, especially when adjusted for the domestic business cycle.12Also, if the international hypothesis w ere correct, the growth of manufacturing output and em ploym ent abroad w ould have risen. But neither, in fact, occurred.11 DECEMBER 1986 Moreover, the appendix to this article shows that the exchange value o f the dollar has not significantly af fected the dem and for dom estic m anufacturing output. CONCLUSION Manufacturing em ployment in the United States has declined slightly in recent years, but this decline should be assessed against a previous sharply declin ing trend relative to overall em ployment in the econ omy. Part o f the recent decline is associated with a reduction in the relative dem and for the manufactur ing product due to cyclical forces in the U.S. economy. In 1979, w hen m anufacturing em ploym ent was slightly larger, the nation’s unem ploym ent rate for civilian workers was 5.8 percent, compared with re cent levels o f about 7 percent.14 Losses in incom e associated with cyclical increases in unemployment reduce the demand for manufacturing output rela tively more than demand in other sectors o f the econ omy. But the longer-term "problem " is the strength o f productivity improvement in the manufacturing sec tor generally. Faster productivity growth in this sector has contributed significantly to real incom e growth in the nation; it has also contributed to a significant decline in the relative price o f manufactured goods, reflecting their increased availability. While the share o f manufacturing output has been maintained, its shares o f em ployment and total spending have de clined. This long-standing pattern has continued from 1979 to 1985. Thus, there is no need to blame other popular villains for manufacturing em ploym ent’s fail ure to regain its previous peak level. REFERENCES Clark, Lindley H., Jr. “ Productivity’s Cost: Manufacturers Grow Much More Efficient, But Employment Lags,” Wall Street Journal, December 4, 1986. Darby, Michael R. “ Price and Wage Controls: Further Evidence,” in Karl Brunner and Allan Meltzer (eds.), Carnegie-Rochester Confer ence Series, vol. 2 (Amsterdam: North Holland, 1976). l2The share of manufacturing output in real GNP was 21.7 percent in 1985 and the first three quarters of 1986. This was higher than the 1948-80 average of 21.3 percent, despite the fact that measures of transitory income losses due to unemployment or low capacity utilization indicate a significantly lower-than-average share would have been expected. Tatom (1986) indicates that manufacturing sector growth exceeded that predicted by income growth alone by about 1.6 percent per year for the period 1980-85. 13See Tatom (1986). The other countries examined were Canada, Belgium, Denmark, Germany, Italy, the Netherlands, Norway, Sweden and the United Kingdom. l4The change in the unemployment rate is an accurate index of cyclical output and income losses when the “ natural rate” of unem ployment, the noncyclical component, is unchanged. The substan tial slowing in the growth of new entrants into the labor force in the 1980s, especially young, inexperienced people, reduced the natural rate significantly. Of course, the latter implies that a return to a 5.8 percent rate would leave the economy with a larger percentage cyclical output loss than that associated with the same unemploy ment rate in 1979. 23 FEDERAL RESERVE BANK OF ST. LOUIS Fieleke, Norman S. “The Foreign Trade Deficit and American In dustry,” New England Economic Review, Federal Reserve Bank of Boston (July/August 1985), pp. 43-52. Tatom, John A. “ Domestic vs. International Explanations of Recent DECEMBER 1986 U.S. Manufacturing Developments,” this Review (April 1986), pp. 5-18. ------------ “ Economic Growth and Unemployment: A Reappraisal of the Conventional View,” this Review (October 1978), pp. 16-22. Appendix Cyclical Changes in Manufacturing Output and Employment Output and em ployment in U.S. manufacturing are strongly cyclical: transitoiy incom e changes associ ated with recessions or booms have a greater impact on demand for manufacturing output and the de mand for labor in this sector than in the rem ainder of the economy. Thus, some part o f the reduction in manufacturing em ployment from 1979; when such em ployment averaged 21.0 m illion workers, to 1986, when it averaged 19.2 million, is due to the cyclical rise in the unemployment rate over the period from 5.8 percent to 7.0 percent. Some simple rules o f thumb allow an assessment o f the current magnitude of cycli cal em ployment losses in manufacturing. The first useful relationship in such an assessment is called Okun's Law, which relates cyclical m ove ments in the unemployment rate to cyclical losses in real GNP. According to recent estimates, each percent age point o f unemployment is associated with a 2'U percent loss in real GNP.' Thus, the rise in unem ploy ment from 1979 to 1986 is associated with a loss of real GNP o f about 2.7 percent, 12'/»I (1.2 percent). This means that if the unemployment rate in 1986 had been 5.8 percent, nominal GNP w ould have been $115 bil lion larger in the first three quarters o f 1986, given prices. To see how this gain in incom e w ould have been distributed between manufacturing and the rest o f the economy, the demand for manufacturing output must be estimated. The demand for such output is a func tion o f the relative price o f the manufactured product 'See Tatom (1878). Digitized for 24FRASER and income; manufacturing output, however, is rela tively more sensitive to transitory fluctuations in real incom e than permanent changes [see Tatom (1986) ]. Using potential real GNP, XP, to measure permanent incom e and real GNP to measure actual real income (permanent plus transitoiy income), X, the estimated demand for annual manufacturing sector output, in growth rate form, for the period 1949-85 is: A in XM, = K2 = 0.86 -0 .5 3 3 Alnl PM/FI, + 2.284 Ain X, - 1.444 A in XP„ (- 3 .7 4 ) (22.59) (- 1 1 .5 6 ) SE = 1.35% DVV = 2.02 where XM is manufacturing sector output, X is real GNP, (PM/P) is the implicit price deflator for manufac turing output deflated by the GNP deflator and XP is potential real GNP. The constant is om itted because it is not significant. When potential and actual real GNP grow at the same rate, the dem and for manufacturing output ex pands at about the same rate, but cyclical fluctuations in real GNP result in much larger variations in the demand o f manufacturing output. The permanent income elasticity o f dem and is the sum o f the actual and potential GNP coefficients, or 0.84; the cyclical income elasticity is much larger, 2.28. The price elas ticity o f demand for manufacturing output is —0.53, or less than one. To test w hether the dem and for dom es tic manufacturing output is negatively related to the exchange value o f the dollar, changes in the logarithm o f the Federal Reserve Board’s trade-weighted ex change rate were added to the equation. None o f the coefficients above w ere significantly altered and the exchange rate coefficient was positive, 0.003 (t = 0.07), FEDERAL RESERVE BANK OF ST. LOUIS although insignificant.- According to these estimates, a 2.7 percent rise in real GNP, given prices and poten tial output, w ould result in a 6.2 percent gain in manufacturing output. Such a gain w ould put the share o f manufacturing output at about 22.5 percent, essentially the same as at the post-World War 11 peak achieved in 1966 and 1973. O f course, a cyclical gain in manufacturing output of this size w ould be associated with a cyclical rise in output per worker, so that the increase in em ploy ment w ould be smaller than that for output. Equation 2 in the text and the demand equation estimate above may be used to find the manufacturing em ployment gain. The product (PM/P) (Xm/X) in equation 2 in the text is the share o f nominal spending (GNP) on manu facturing product. Changes in this spending share result in proportionate changes in manufacturing em ployment relative to total employment. Cyclical variations in the share o f nominal GNP 2When the relative price of imports is used instead of the tradeweighted exchange rate, its coefficient has the “ expected" negative sign, -0 .0 2 , but it is not statistically significant (t = -0 .7 2 ). None of the elasticity estimates is significantly affected in this test either. The relative price of imports is the ratio of the implicit price deflators for imports from the National Income and Product Accounts and for the domestic manufacturing sector. DECEMBER 1986 originating in domestic manufacturing equal (Ain XM — AlnX + Aln(PM/P) ]; according to the dem and equa tion estimate above, holding (PM/P) and XP constant, this sum is 1.284 AlnX. For a 2.7 percent change in real GNP (AlnX = 2.7 percent), the change in the nominal spending share is 3.5 percent. With an unchanged relative compensation level, equation 2 in the text and the demand function here indicate that a movement from a 7 percent to a 5.8 percent unem ploym ent rate w ill result in a difference (Ain LM — Ain L) equal to 3.5 percent; since Ain L is about 1.2 percent, Ain LM is about 4.7 percent.3 Thus, manufacturing em ployment w ould increase from about 19.2 million workers in manufacturing to about 20.1 million, still below the 21 million level observed in 1979.4 3A more direct method of estimation gives about the same conclusion. When Ain LM, where LM is manufacturing employment, is regressed on a constant and the current and past two quarters' growth rates of real GNP, quarterly for the period IV/1948—11/1986, the sum of the coefficients on real GNP growth yield a manufacturing employment elasticity of 1.5, so that a 4 percent gain in manufacturing employment is associated with a 2.7 percent rise in real GNP, about the same as that indicated above. “These calculations presume that relative wages and prices would be unchanged by a cyclical rise in real GNP. There is no indication, either in the charts of these variables in the text, or in correlation analysis, that these variables are cyclical. 25 FEDERAL RESERVE BANK OF ST. LOUIS DECEMBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS REVIEW INDEX 1986 JANUARY AUGUST/SEPTEMBER Michael T. Belongia, “ Estimating Exchange Rate Effects on Exports: A Cautionary Note” Daniel L. Thornton, “The Discount Rate and Market Interest Rates: Theory and Evidence” Keith M. Carlson, “ Recent Revisions of GNP Data” Salam K. Fayyad, “A Microeconomic System-Wide Ap proach to the Estimation of the Demand for Money” FEBRUARY R. W. Hafer, “The FOMC in 1985: Reacting to Declining M1 Velocity” R. Alton Gilbert, “ Requiem for Regulation Q: What It Did and Why It Passed Away” MARCH R. IV. Hafer, “The Response of Stock Prices to Changes in Weekly Money and the Discount Rate” G. J. Santoni, “The Effects of Inflation on Commercial Banks” APRIL John A. Tatom, “ Domestic vs. International Explanations of Recent U.S. Manufacturing Developments” Kenneth C. Carraro and Daniel L Thornton, “The Cost of Checkable Deposits in the United States” OCTOBER Jerry L. Jordan, “The Andersen-Jordan Approach after Nearly 20 Years” Dallas S. Batten and Daniel L Thornton, “The MonetaryFiscal Policy Debate and the Andersen-Jordan Equa tion” Keith M. Carlson, “A Monetarist Model for Economic Stabilization: Review and Update” Leonall C. Andersen and Jerry L. Jordan, “ Monetary and Fiscal Actions: A Test of Their Relative Importance in Economic Stabilization” Leonall C. Andersen and Keith M. Carlson, “A Monetarist Model for Economic Stabilization” NOVEMBER MAY Alex Cukierman, “ Central Bank Behavior and Credibility: Some Recent Theoretical Developments” Zalman F. Shifter, “ Adjusting to High Inflation: The Israeli Experience” G. J. Santoni, “The Employment Act of 1946: Some History Notes” Michael T. Belongia, “The Farm Sector in the 1980s: Sudden Collapse or Steady Downturn?” JUNE/JULY DECEMBER Mack Ott and Paul T. W. M. Veugelers, “ Forward Ex change Rates in Efficient Markets: The Effects of News and Changes in Monetary Policy Regimes” R. Alton Gilbert and Geoffrey E. Wood, “ Coping with Bank Failures: Some Lessons from the United States and the United Kingdom” John A. Tatom, “ How Federal Farm Spending Distorts Measures of Economic Activity” John A. Tatom, “Why Has Manufacturing Employment Declined?” 27