The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Vol. 71, No.3 May/June 1989 3 T h e 1988 D rought: Its Im pact on District A gricu ltu re 14 Eighth D istrict Banks: in the Black Back 23 T h e Eighth District Business Economy in 1988: Still Expanding, But M ore Slow ly 33 C om parin g Futures and Survey Forecasts o f N ear-T erm T rea su ry Bill Rates 43 Bank Runs and Private Rem edies THE FEDERAL RESERVE BANK of ST. m i IS 1 Federal R eserve B ank of St. Lou is R e v ie w May/June 1989 In This Issue . . . The U.S. agricultural economy endured one of the worst droughts in decades during 1988. In the first article in this Review, “The 1988 Drought: Its Impact on District Agriculture,” Jeffrey D. Karrenbrock examines how the drought affected agriculture in both the U.S. and the Eighth Federal Reserve District. Crop producers experienced sharp drops in yields, while livestock pro ducers faced higher feed costs. Karrenbrock points out that, despite lower net farm income, both farmers and agricultural lending institutions were able to improve their financial positions on average during 1988. Factors helping to limit the financial damage to some farmers included higher grain prices, strong agricultural exports and continued government sup port. On net, the effect of the drought was to slow but not stop the agricultural recovery that started in 1984. * * * In the second article in this Review, "Eighth District Banks: Back in the Black," Lynn M. Barry reports that, for banks in both the nation and the Eighth District, 1988 was a year of recovery from the lackluster earnings reported in 1987. Aggregate bank profit ratios in the United States and the Eighth District improved last year as many banks began to rebound from the negative earnings caused by increased loan loss provisions tied to foreign loans. Barry also reports that profits improved in 1988 across vir tually every asset-size category. Propelled by stronger earnings and im proved asset quality, bank performance at the largest District banks im proved significantly in 1988. Further gains were made in 1988 by the smaller District banks, which posted higher earnings as loan loss provi sions and loan charge-offs declined. Barry expects continued improvement in the coming quarters. Asset quality problems, which have plagued some District banks, appear to be under control; thus, future loan problems should have a less severe effect on earnings. * * * In this issue’s third article, "The Eighth District Business Economy in 1988: Still Expanding, But More Slowly,” Thomas B. Mandelbaum reports that the region’s economy continued to expand in 1988, its sixth successive year of growth. Moreover, District income and employment reached record highs, while the regional unemployment rate declined to its lowest level of the decade. Unlike the previous five years, however, in which regional employment growth approximated the national pace, the District’s job growth was substantially slower than the national average last year. The author discusses the factors that caused this sluggishness and describes other significant developments in the Eighth District's business economy during MAY/JUNE 1989 2 1988. In addition, suggests that a further slowing of the regional economy is likely in 1989. * * * Evidence indicates that Treasury bill futures rates are better predictors of future Treasury bill rates than the forward rates implicit in observed spot rates. Moreover, evidence also shows that survey forecasts often are more accurate than the implicit forward rates. In the fourth article in this Review, “Comparing Futures and Survey Forecasts of Near-Term Treasury Bill Rates,” R. W. Hafer and Scott E. Hein attempt to answer the question, "Does the Treasury bill futures rate provide a better forecast of future short-term interest rates than do survey forecasts?” The authors use survey forecasts of the three-month Treasury bill rate gathered from the Bond and Money Market Letter. This survey polls about 40 to 50 financial market analysts asking for point forecasts for a variety of interest rates three and six months hence. These predictions are com pared with the futures market forecasts, taken from futures contracts traded on the International Monetary Market of the Chicago Mercantile Exchange, of interest rates three months ahead and six months ahead. Based on forecasts from March 1977 through October 1987, Hafer and Hein find that, in general, the futures market forecasts are as good or bet ter than the survey forecasts. They also test the proposition that the futures market efficiently utilizes all publicly available information by testing whether information in the survey forecast could improve upon the futures market forecast. Based on these tests, there is little evidence to suggest that the survey forecast or its revision improves upon the futures rate prediction. Thus, in contrast to previous research, the evidence in this article indicates that the futures rate provides a useful measure of the market’s expectation of future interest rates. * * * Why do we regulate the activities of banks? The reason for much of the current banking regulation in the United States rests on the notion that the banking system is vulnerable to runs that would disrupt the operation of the banking system and other forms of economic activity. Regulation, so the theory goes, is necessary to prevent such runs. In the final article in this Review, "Bank Runs and Private Remedies,” Gerald P. Dwyer, Jr. and R. Alton Gilbert examine the history of banking in the United States prior to the formation of the Federal Reserve to determine whether the banking system, in fact, was vulnerable to such runs. While they find some episodes of runs on the banking system, they also find that there were many years with no evidence of runs at all; moreover, some periods without runs included recession years. The authors find only limited evidence that is consistent with the view that the runs had adverse effects on economic activity. The reasons for the limited effects of the runs can be found in the private remedies developed by banks. Through their clearinghouses, banks created clearinghouse loan certificates, which had an impact on the opera tion of the banking system much like increases in the monetary base. In periods when banks could not meet the demand for currency by depositors through the creation of clearinghouse loan certificates, they acted jointly to restrict currency payments to depositors. Restricting cur rency payments was also a form of private remedy for runs, since it enabled banks to limit the declines in their assets and deposit liabilities. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis 3 Jeffrey D. Karrenbrock Jeffrey D. Karrenbrock is an econom ist at the Federal Reserve Bank o f St. Louis. David H. Kelly provided research assistance. The 1988 Drought: Its Impact on District Agriculture 1VT _L INETEEN eighty-eight will be remembered as the year o f the drought. Crop producers ex perienced sharp drops in yields, while livestock producers faced higher feed costs. The drought slowed, but did not stop the agricultural recov ery that started in 1984. Despite the drought, farmers and agricultural lending institutions im proved their financial positions during 1988. This article examines these and many other effects o f the drought on the agricultural econo my. The article first provides a brief overview of how the U.S. agricultural economy per formed during 1988. The agricultural economy of states in the Eighth Federal Reserve District is then compared to U.S. agricultural per formance.1 U.S. AGRICULTURE AND THE DROUGHT By mid-August, the drought had affected all of the United States, except for the naturally dry Southwest and the East coast, with substantial effects on agricultural production and distribu tion. Low rainfall combined with high tempera 1The Eighth Federal Reserve District includes all of Arkansas and parts of Illinois, Indiana, Kentucky, Mississippi, Missouri and Tennessee. The majority of this report, however, focuses only on the entire states of Arkansas, Kentucky, Missouri and Tennessee. tures caused corn and soybean yields across the United States to fall by 29 percent and 21 per cent, respectively. The decreased supplies sent commodity prices upward throughout the sum mer which helped to limit the drought's impact on some farmers. In addition to reduced sup plies, additional problems arose in moving grain products from elevators to processors and ex port markets. Low water levels on major water ways slowed, and sometimes completely stopped, barge movement of grain. Farm Finances After four years o f increases, net farm in come is currently forecast to have shrunk to $40 billion in 1988.2 Although this figure is down 14 percent from 1987, it is still three times larger than net farm income in 1983. The income statement of the farm sector since 1981 is shown in table 1. The 1988 forecast figures indicate that, while total farm receipts rose more than 9 percent in 1988, farm expenses climbed about 7 percent. Feed, fertilizer and machinery led the list of items increasing in cost. These increasing expenditures plus falling government payments and dwindling grain in ventories resulted in lower net farm income. 2U.S. Department of Agriculture, A gricultural O utlook (April 1989), p. 54, table 32. MAY/JUNE 1989 4 Table 1 Farm Sector Income Statement (billions of dollars) 1981 Farm receipts Government payments Total farm income Total expenses Net farm income2 Net cash income 1982 1983 1984 1985 1986 1987 19881 $144.1 1.9 166.4 139.4 26.9 32.8 $147.1 3.5 163.5 140.0 23.5 37.8 $141.1 9.3 153.1 140.4 12.7 36.9 $146.8 8.4 174.9 142.7 32.2 38.7 $149.1 7.7 166.1 134.0 32.3 46.6 $140.2 11.8 159.8 122.3 37.4 51.4 $143.7 16.8 169.8 123.5 46.3 57.1 $157.0 14.0 172.0 132.0 40.0 58.0 'Values for 1988 are forecasts. 2Total net farm income includes the value of inventory changes. Net farm income totals may not add due to rounding. Data are not adjusted for inflation. SOURCE: A gricultural Outlook (April 1989), p. 54, table 32. While net farm income was expected to fall in 1988, net cash income from farming, another indicator o f farm finances, was expected to rise slightly (see table 1). The difference between net farm income and net cash income from farming is that net farm income measures in come largely generated from a given calendar year’s production, regardless o f whether the commodities are sold, fed or placed in inventory during the year. Net cash income from farming measures the total income that farmers elect to receive from their operation in a given calendar year, regardless of the amount o f production or the year the marketed output was produced. It approximates the income stream available to farmers for purchasing assets such as machinery or land, retiring loans and covering all other expenditures. Since production was low in 1988, net farm income was also lower. But, since some farmers were able to sell stored grain at high prices, net cash income from farm ing was up slightly in 1988. When the number of farms is taken into con sideration, the financial picture changes very lit tle for 1988. Real net farm income per farm is expected to have dropped about 16 percent from 1987 to 1988, while real net cash income from farming per farm is expected to have fallen less than 1 percent.3 Real U.S. net farm income and real cash income from farming per 3The term “ real” here refers to the fact that the data has been adjusted to take into account the impact of inflation. 4See Duncan (February 1989). http://fraser.stlouisfed.org/ Federal Reserve Bank RESERVE BANK OF ST. LOUIS FEDERAL of St. Louis farm since 1950 are shown in figure 1. During the past 30 years, real net farm income per farm has been trending upward, while the real earnings of farmers have been constant to de clining. With few er and few er farms, each re maining farm gets a larger share of the relative ly constant total farm earnings.4 Farm Balance Sheet Despite declining net farm income, the balance sheet of the agricultural sector was ex pected to improve in 1988, chiefly because of rising land values. Farmland values were ex pected to increase approximately 4 percent in 1988.5 While real estate values w ere improving, farmers continued to reduce their real estate debt, paying o ff nearly $4 billion in 1988. Nonreal-estate debt increased about $1.1 billion, al lowing total farm liabilities to fall for the fifth straight year to about $139 billion. Overall, the farm sector’s debt-to-asset and debt-to-equity ratios improved for the third straight year (see figure 2). Agricultural Trade The summer drought had only a limited im pact on agricultural exports. The carry-over of agricultural commodity stocks was large enough to handle increased export demand, despite de creased current year supplies. In 1988, net agricultural exports nearly doubled as exports 5U.S. Department of Agriculture (June 1988), p. 3. 5 m Figure 1 Real U.S. Net and Cash Farm Income per Farm Thousands of 1982 dollars Thousands of 1982 dollars SOURCES: Farm Incom e Data: A H is to rica l P erspective, 1986, p. 16 and A g ric u ltu ra l O utlook (April 1989), p. 54, table 32. Figure 2 U.S. Agricultural Balance Sheet Ratios Ratio Ratio SOURCE: Agricultural Outlook, (January-February 1989), p .62, table 33. MAY/.II INF 1QRQ 6 reached their highest levels since fiscal year 1983. Simultaneously, agricultural imports reached a record high of $21 billion. Agricultur al exports increased 27 percent in dollar value, while imports increased less than 2 percent. The improved agricultural trade surplus was partially a result o f the falling value of the dollar and continued government subsidization o f exports. One example of a U.S. government export subsidy program is the Export Enhance ment Program. This program essentially gives exporters a subsidy for every unit of grain sold so they can compete with other world export ers, mainly the European Community nations, who also subsidize their exports.6 Agricultural Lenders Despite lower real net farm income, agricul tural banks and the Farm Credit System contin ued to improve their financial positions in 1988. The number o f agricultural bank failures in the United States dropped from 53 in 1987 to 24 in 1988. Similarly, agricultural banks reporting negative earnings fell from 488 in 1987 to 261 in 1988. In addition, loans delinquent 30 days or more at agricultural banks dropped to 3.77 per cent of all agriculture loans. This compares with a delinquency rate o f 5.55 percent through the same period last year. Furthermore, agricultural banks' return on assets increased 0.26 of a per centage point to 0.92 percent, while return on equity jumped more than 2 percentage points to 9.69 percent. The Farm Credit System (FCS) also improved its financial position while undergoing a reorga nization in 1988. In the reorganization, the Federal Land Banks (FLB) and the Federal Inter mediate Credit Banks (FICB) of each district merged to form the Farm Credit Bank. The Farm Credit Bank and its affiliates provide farm ers with long-term loans for land purchases as well as short-term loans for operating expenses. The FCS’s Banks for Cooperatives also under went a reorganization in which 11 of the 13 Banks for Cooperatives merged to form the CoBank. The CoBank provides loans to agricultural cooperatives. The Farm Credit Bank in conjunc 6See Coughlin and Carraro (November/December 1988). TThe Farm Credit System is a nationwide system of federal ly chartered agricultural lending institutions cooperatively owned by their borrowers. http://fraser.stlouisfed.org/ Federal Reserve F R A I of St. F R V F R A N K f>F <5T lO I I I S F F n Bank R F S Louis tion with the CoBank make up the Farm Credit System.7 The Farm Credit System’s performance im proved in 1988 when compared to 1987. The FCS reported a combined net income of $704 million for 1988, compared with a net loss in 1987 of $17 million. A major factor in the im proved 1988 results was a substantial negative provision for loan losses of $680 million for the year 1988, more than three times the negative provision of $196 million for 1987. In other words, the FCS decreased the amount o f money it had set aside to cover loans that were at a high risk of defaulting. Although gross loans de clined, the rate of decline was considerably less than in the three preceding years. While things appear to be improving for the FCS, problems still remain; in 1988, for example, the Federal Land Bank in Jackson, Mississippi, was placed in receivership.8 The Farmers Home Administration (FmHA) continues to struggle, but is improving in some areas. The FmHA serves as a lender of last re sort for farmers who cannot secure loans else where. In 1988, delinquencies o f insured individ ual farm ownership loans increased by 2 per cent. New agricultural loan volume fell 30.6 per cent in 1988 when compared to 1987. The FmHA’s current-year operating loss on farmer program loans o f $8.3 billion was substantially less than last year’s loss o f $15.7 billion. The large operating loss in 1987 was partially due to an increase in the FmHA’s allowance for loan losses. Consum er Prices Despite the drought’s effect on commodity prices, the Consumer Price Index (CPI) for all food in 1988 rose near the 1987 rate, about 4 percent. However, food prices did increase more rapidly in the last two quarters of the year than in the first two, with food prices in creasing at a 5.2 percent annual rate during the fourth quarter. Because commodity costs are a small part of the retail price o f food, ranging from about 10 percent to 30 percent, only small upward adjustments in retail prices are needed to reflect farm price increases.9 8Federal Farm Credit Banks Funding Corporation (March 1, 1989). 9U.S. Department of Agriculture (July 1987), p. 12. 7 Figure 3 Government Payments / Net Farm Income Percent 80 Percent 8 0 ------- 70 70 60 60 50 50 40 40 30 30 20 20 10 10 err J im —r r 1950 55 60 65 70 75 80 85 1989F SOURCES: Econom ic Indicators o f the Farm Sector and A g ricu ltu ra l O utlook (January-February 1989), p. 62, table 32. Among food items, fresh fruit and poultry registered price increases of approximately 8.3 percent and 7.2 percent, respectively.1 Other 0 items with price increases of more than 5 percent included beef, fish, fresh vegetables and cereal and bakery products. While poultry price in creases were, in part, due to drought-induced pro duction losses, increased consumer demand also helped push retail prices higher. The United States Department of Agriculture estimates that the drought added only 0.5 percent to the food CPI in 1988.1 The only major food item whose price 1 declined was pork; its retail price fell about 3 per cent last year. GOVERNMENT SUPPORT Direct government payments provided more than 20 percent of U.S. net farm income for the 10Based on comparison of the annual averages of each pro duct’s 1987 and 1988 CPI. "U .S . Department of Agriculture, A gricultural Outlook (January/February 1989), p. 35. 12The Payment-In-Kind program compensated farmers for taking land out of production by paying them with sixth consecutive year in 1988. Government payments as a percent of net farm income have been abnormally high since the record level of 73 percent in 1983 when the Payment-In-Kind program was enacted.1 Historical levels of 2 government payments as a percentage of net farm income are shown in figure 3. Although direct government payments to farmers in 1988 declined more than 16 percent from 1987 lev els, net farm income fell almost 14 percent. Thus government payments made up 35 per cent of net farm income last year. In 1989, di rect government payments to farmers are pre dicted to fall to $11 billion, or about 24 percent o f net farm income.1 3 Commodity program outlays fell in 1988 and will continue to fall in 1989 for two main rea sons. First, loan rates and target prices for most government-owned grain. If a farmer took ground normally planted in corn out of production, he was compensated with government-owned corn. 13U.S. Department of Agriculture, A gricultural Outlook (April 1989), p. 54. MAY/JUNE 1989 8 major commodities fell in 1988 and are sched uled to decline again in 1989. Second, higher grain prices resulting from the drought have decreased the amount of deficiency payments to farmers. Deficiency payments are the target price minus the loan rate, or the target price minus the cash price, whichever is smaller. All major commodity cash prices were above the loan rate this year. Thus, declining deficiency payments have resulted from lower target prices and higher cash prices. In contrast, one agricultural program with ris ing expenditures is the Conservation Reserve Program (CRP). The CRP takes land out of agri cultural production for 10 years or more in ex change for annual payments to the land owner. The CRP differs from other commodity pro grams that are run generally on an annual basis in that it is a multi-year agreement. In 1988, an additional 8.5 million acres were enrolled in the CRP; the total enrolled acreage now runs more than 24 million acres. Estimated total 1988 CRP payments for rent and cover crop establishment w ere $1.5 billion.1 An additional 3.5 million acres 4 are scheduled to be taken out of production in 1989. In 1989, few er acres will be enrolled, and therefore less money will have to be spent establishing cover crops for erosion control. Farmers also got an income boost to counter act the adverse effects of the drought from payments approved by Congress under the Disaster Assistance Act. Budgeted expenditures for the program are $3.9 billion.1 These funds 5 are to be paid out in 1988 and 1989. Corn farm ers are expected to be the largest recipient of aid, getting about $1.7 billion. Payment rates differed depending on the extent of crop dam age. For production losses between 35 percent and 75 percent, the payment rate was for 65 percent of the normal amount of the crop grown on the farm. For losses more than 75 percent, the payment rate was 90 percent of normal production. Disaster payments to crop producers w ere limited to $100,000 per per son.1 Any person with revenues more than $2 6 million was not eligible for assistance. Drought-stricken livestock producers also received disaster assistance. The Secretary of 14Calculated as (total acreage taken out ot production x average weighted rental rates for land in CRP) + (estimated cost sharing for cover crop establishment x new acreage enrolled in 1988). Numbers used in this estima tion were obtained from the Agricultural Stabilization and Conservation Service. http://fraser.stlouisfed.org/ FEDERAL St. Louis Federal Reserve Bank ofRESERVE BANK OF ST. LOUIS Agriculture had several options by which to provide assistance. The two options used most extensively included selling Commodity Credit Corporation-owned feed grain at 75 percent of the county loan rate and partially reimbursing livestock producers for purchased feed and transportation expenses. Low-interest disaster loans were also available from the FmHA. EIGHTH DISTRICT AGRICULTURE AND THE DROUGHT The impact o f the drought on District agricul ture varied from state to state. All states re ported growing season rainfall amounts that ranged from eight to 11 inches below normal (see table 2). While the drought reduced output and net farm income, it did not cripple District agriculture. District waterway activity reflected the severi ty of the summer drought. In mid-June, water depth at the mouth o f the Ohio River at Cairo, Illinois, was 17 feet below normal. Channel widths on parts of the river system narrowed from 500 feet to 200 feet. At Memphis in early August, the Mississippi River flow was 46 per cent below normal for that time of year. Despite low water levels, total grain shipments on the Illinois and Mississippi waterways in 1988 w ere actually larger than total 1987 ship ments. Grain shipments, however, did fall below average from June through November. Monthly grain shipments in 1988 are compared with 1981-87 average monthly grain shipments in figure 4. July saw the sharpest drop in move ment of grain from average, with shipments falling 20 percent. Barge rates skyrocketed in the last week of June as navigation problems became wide spread. For example, rates from Peoria to New Orleans averaged $17.44 per ton in that week in contrast to $6.37 per ton the prior week. Rates, however, declined through August, then started climbing again as prospects for Soviet corn buy ing increased in September. In August, barge 15U.S. Department of Agriculture, A gricultural Outlook (September 1988), p. 28. 16The $100,000 limit per person generally meant a $100,000 limit per farm. The ASCS reviewed each application and determined how much aid each applicant could receive. 9 Table 2 Growing Season Rainfall for Selected Areas (inches) 1988 rainfall April-Septem ber N orm al1 rainfall A pril-S eptem ber Departure from norm al N. Little Rock, AR2 28.45 37.76 -9 .3 1 Paducah, KY 15.21 23.79 - 8 .5 8 St. Louis, MO 11.77 21.65 -9 .8 8 Memphis, TN 14.62 25.80 -1 1 .1 8 ’ Normal is defined as a 30-year average. C um ulative rainfall, January through September. SOURCE: Agricultural Statistical Service of the individual states. Figure 4 Grain Shipments^ Millions of tons Millions of tons SOURCE: Mississippi River Barge Traffic, U.S. Army Corps of Engineers, Rock Island District. Q Grain shipments on the Illinois Waterway and Mississippi River, (Locks 11-22). MAY/JUNE 1989 10 Table 3 District Crop Yields: 1988 vs. 1985-87 Average Yield United States Corn Cotton Rice Sorghum Soybeans Tobacco Wheat M issouri Percent D ifference Corn Cotton Sorghum Soybeans Wheat -3 2 .7 % - 9 .4 - 2 .4 -2 2 .0 27.1 Arkansas Percent D ifference Cotton Rice Sorghum Soybeans Wheat 4.7% 2.9 - 1 .0 7.9 39.5 Percent D ifference - 28.8% -0 .6 -0 .5 - 6 .3 -2 0 .5 - 0 .3 - 7 .2 K entucky Percent Difference Corn Soybeans Tobacco Wheat - 26.5% -1 2 .6 - 1 .0 34.0 r Tennessee Corn Cotton Soybeans Tobacco Wheat Percent D ifference -1 6 .7 % -1 5 .0 -1 .3 0.0 41.5 SOURCE: Agricultural Statistical Service of the four states. rates w ere only 6 percent above the January-toMay average rate. The decline in rates in Au gust was due to decreased demand for exports and increased grain holdings by producers in anticipation of higher grain prices.1 7 Crop Production The most obvious effect o f a drought is its ef fect on crop yields. Crop performance in the District was varied. U.S. and state average crop yields are shown in table 3. Corn yields were most affected by the drought. Major producing states in the District suffered large yield losses that ranged from 17 percent in Tennessee to 33 percent in Missou ri.1 Sorghum yields w ere also down slightly. 8 1 7 U .S . Department of Agriculture (January 1989), pp. 25-27. http://fraser.stlouisfed.org/ FEDERAL St. Louis Federal Reserve Bank ofRESERVE BANK OF ST. LOUIS Soybean and cotton yields w ere mixed across the District. For example, soybean yields rose in Arkansas and fell in Kentucky, Tennessee and Missouri, while cotton yields rose in Arkansas and fell in Missouri and Tennessee. Tobacco yields were essentially unchanged in Kentucky and Tennessee. Wheat and rice crop performance were less affected by the drought. Since winter wheat crops require most o f their moisture in the spring, the summer drought did little damage to the crop. In fact, most District states posted sizable gains in wheat yields. Rice production was not damaged by the drought since much of the crop's water comes from wells and not natural rainfall. Nonetheless, a more normal 18Arkansas is not a major corn-producing state. 11 rainfall pattern in the southern states did help rice and other crops throughout the summer. On a state basis, Arkansas fared the best overall with yield increases in all crops except sorghum. Tennessee and Kentucky, while ex periencing decreased yields, faced losses that w ere generally less-than-average U.S. yield loss es. Missouri experienced large yield losses in both of its most important cash crops, soybeans and corn. Livestock Production Red meat production in the District increased by about 3 percent in 1988.1 Kentucky led the 9 District with a 9.4 percent increase in red meat production. Missouri also increased red meat production, while Arkansas and Tennessee de creased production. U.S. broiler production increased more than 4 percent in 1988 to about 16.1 billion pounds, after increasing nearly 9 percent in 1987. A r kansas, the nation’s largest broiler producer, in creased production about 3.5 percent in 1988. District Farm Incom e District net farm income increased by 26 per cent in 1987, after falling the two previous years. District 1988 data is available with a oneyear lag, but with 1988 U.S. net farm income expected to drop 14 percent, District farmers can expect similar results.2 Similar to the na 0 tion, total farm cash receipts in the District for the first three quarters of 1988 were well ahead of cash receipts for the same period a year ago. All District states were reporting increased crop receipts and livestock receipts. While farm receipts were up, so were expen ditures for District farmers. Especially hard-hit w ere hog producers. Profit margins were squeezed from both sides as increased inven tories pushed hog prices lower and the drought pushed input prices higher. Cattle producers, while also facing higher input costs, enjoyed market prices that were generally higher than 1987 prices. 19Red meat production includes total beef, veal, pork, lamb and mutton slaughtered in federally inspected and other plants, but excludes animals slaughtered on farms. 20Carraro (1988) notes that District net farm income closely follows U.S. net farm income. Broiler producer net returns went as high as 20 cents per pound during July and averaged nearly 5 cents for the year. Higher broiler prices were likely a result of heat stress on pro duction and increased retail sales efforts by fast food restaurants and grocers. District Agricultural Lenders District agricultural bankers improved their financial position again in 1988, outperforming, on average, U.S. agricultural banks as a whole. U.S., District and state data pertaining to agri cultural bank performance are shown in table 4. In 1988, District banks had both higher re turns on assets and equity than did the U.S. agricultural banks on average.2 The District’s 1 agricultural loan net losses as a percent of all agricultural loans was below the national aver age, while the District’s 30-day-or-more delin quent agriculture loans as a percent o f total ag ricultural loans was higher than the U.S. aver age. The District’s non-performing agricultural loans fell for the third straight year to 5.06 per cent of all agricultural loans. The number of ag ricultural banks with negative earnings fell in both the nation and the District. With respect to the individual states, Tennes see agricultural banks had the highest return on assets and Indiana the lowest. Missouri had the highest return on equity, while Illinois had the lowest return on equity. All District states im proved their agricultural losses as a percent of total agriculture loans during 1988. Further more, non-performing agricultural loans as a percent o f total agricultural loans fell in all states except Tennessee. Mississippi agricultural banks saw a substan tial improvement over 1987. Returns on both assets and equity went from negative to positive values in 1988. Return on assets increased 1.4 percentage points and return on equity jumped 17.8 percentage points. Both of the District’s Farm Credit Banks im proved their financial positions during 1988.2 2 22The two Farm Credit Banks in the District are the St. Louis branch, covering the states of Arkansas, Illinois and Missouri, and the Louisville branch, covering Indiana, Ken tucky, Ohio and Tennessee. 2 Based on fourth-quarter FDIC Reports of Condition and In 1 come for Insured Banks. MAY/JUNE 1989 12 Table 4 U.S. and District Agricultural Banking Data U.S. D istrict 1988 Banks with negative earnings Return on assets Return on equity Ag. loan losses/Total ag. loans Ag. nonpf. loans/Total ag. loans' 1987 1988 1987 261 0.92 9.69 0.59 3.77 488 0.66 7.21 1.88 5.55 11 1.04 10.9 0.4 5.06 29 0.76 8.1 2.49 6.94 Arkansas2 1988 1987 Banks with negative earnings Return on assets Return on equity Ag. loan losses/Total ag. loans Ag. nonpf. loans/Total ag. loans' 5 1.05 10.34 0.31 2.75 5 0.96 9.17 1.13 3.08 M ississippi2 1988 1987 Banks with negative earnings Return on assets Return on equity Ag. loan losses/Total ag. loans Ag. nonpf. loans/Total ag. loans' 0 0.98 11.54 0.34 4.14 2 -0 .4 2 -6 .2 6 3.31 9.9 Illin o is2 1988 1987 1 0.94 10.14 0.29 5.73 11 0.7 7.71 3.03 8.36 M issouri2 1988 1987 4 1.17 12.32 0.45 5.96 8 0.86 9.13 2.26 7.4 Indiana2 1988 1987 1 0.9 10.34 1.01 7.41 1 0.51 5.94 2.49 10.93 K entucky2 1988 1987 0 1 10.27 0.5 4.92 1 0.91 9.51 2.2 5.54 Tennessee2 1988 1987 0 1.21 11.25 1.36 4.82 1 0.91 9.02 8.51 3.08 'Nonperforming loans are defined as those loans that are 30 days or more delinquent. 2State data only includes banks within the Eighth District. SOURCE: Fourth-Quarter FDIC Reports of Condition and Income for Insured Commercial Banks. The St. Louis branch had a net income of $99.3 million in 1988, up from $10.4 million last year. The Louisville branch generated a $3.6 million net income, which included an extraordinary $92 million loss on the restructuring of highcost debt. This is the first year since 1983 that the Louisville branch has had positive net in come; in 1987, the branch lost $25.1 million. SUMMARY The summer drought of 1988 has left its mark on the agricultural economy. Real net farm income is lower, consumer prices are slightly higher and drought conditions remain in some areas. Despite lower yields and higher in http://fraser.stlouisfed.org/ FEDERAL RESERVE Federal Reserve Bank of St. Louis BANK OF ST. LOUIS put costs, the average U.S. and District farmer improved his balance sheet in 1988. While most of the results of the drought w ere negative, the drought has had one positive effect on the farm economy. The combination o f lower production and continued strong consumption has left grain stocks at their lowest level in years. These low grain stocks will provide price support for commodities in 1989. REFERENCES Carrara, Kenneth C. “ The 1987 Agricultural Recovery: A District Perspective,” this Review (March/April 1988), pp. 36-37. Coughlin, Cletus C., and Kenneth C. Carrara. “ The Dubious Success of Export Subsidies for Wheat,” this Review (November/December 1988), pp. 38-47. 13 Duncan, Marvin R. “ U.S. Agriculture: Hard Realities and New Opportunities,” Federal Reserve Bank of Kansas City Economic Review (February 1989), p. 6. Federal Deposit Insurance Corporation. “ FDIC Reports of Condition and Income for Insured Commercial Banks,” various issues. ______ Agricultural Resources Situation and Outlook Report, June 1988, p. 3. ______ Economic Indicators o f the Farm Sector, National Financial Summary, 1987, p. 7 & p. 14. _____ . Farm Income Data: A H istorical Perspective, 1986, p. 16. Federal Farm Credit Banks Funding Corporation. “ News Release,” New York, March 1, 1989. _____ . Food Cost Review, 1986, July 1987, p. 12. U.S. Department of Agriculture, Economic Research Service. Agricultural Outlook, various issues. ______ The Drought of 1988, January 1989. MAY/JUNE 1989 14 Lynn M. Barry Lynn M. Barry is an econom ist at the Federal Reserve Bank of St. Louis. Thomas A. Pollmann provided research assistance. Eighth District Banks: Back in the Black OB COMMERCIAL banks in both the nation and the Eighth Federal Reserve District, 1988 was a year of recovery.1 Aggregate bank profit ratios in the United States and the Eighth District improved as many of the nation’s larger banks began to recoup from losses associated with foreign loans. Further gains were made by smaller District banks, which posted higher ear nings as loan loss provisions and loan chargeoffs declined. Asset quality also improved at small banks as nonperforming loans and actual loan losses decreased. $191.6 million from 1987. The U.S. banking in dustry earned $25.1 billion in 1988, up sharply from $3.2 billion in 1987. Sixty-eight banks, 5.3 percent of all District banks, reported negative earnings in 1988, down from 86 in 1987. Na tionally, 13.7 percent o f commercial banks reported net losses for the year compared with 18.2 percent in 1987. Much of the improvement in both District and U.S. bank earnings can be traced to lower loan loss provisions, which had a positive effect on earnings. This article compares the performance of Eighth District commercial banks with their na tional counterparts across several asset-size categories.2 An analysis of bank earnings, asset quality and capital adequacy provides useful in formation on the financial condition, regulation compliance and operating soundness of the District’s banking industry. Return on Assets and Equity EARNINGS Eighth District banks reported year-end earn ings of $1.1 billion in 1988, an increase of 1The Eighth Federal Reserve District consists of the follow ing: Arkansas, entire state; Illinois, southern 44 counties; Indiana, southern 24 counties; Kentucky, western 64 coun ties; Mississippi, northern 39 counties; Missouri, eastern and southern 71 counties and the City of St. Louis; Ten nessee, western 21 counties. 2For more specific bank performance statistics on each Eighth District state, see the Federal Reserve Bank of St. Louis’ June 1989 issue of Pieces of Eight. BANK OF ST. LOUIS FEDERAL RESERVE In analyzing bank earnings, there are two standard measures of bank performance: the return on average assets (ROA) and the return on equity (ROE) ratios.3 The ROA ratio, calcu lated by dividing a bank’s net income by its average annual assets, shows how well a bank’s management is using the company’s assets. The ROE ratio, obtained by dividing a bank’s net in come by its equity capital, indicates to share holders how much the institution is earning on their investment.4 3A major concern with ROA, ROE and other performance measures is that they are calculated using the book values of assets, liabilities and equity not the current market value. 4Equity capital includes common and perpetual preferred stock, surplus, undivided profits and capital reserves. 15 Table 1 Return on Average Assets and Return on Equity Return on Average Assets (ROA) 1987 1988 D istrict All banks < $ 2 5 million in assets $25-$50 million $50-$100 million $100-$300 million $300 million-$1 billion $1-$10 billion > $10 billion U.S. D istrict 0.93% 0.80 0.96 1.01 0.97 1.01 0.82 — 0.83% 0.31 0.62 0.78 0.81 0.66 0.79 0.96 0.80% 0.68 0.90 0.95 0.94 1.07 0.51 — 1986 U.S. 0.11% 0.15 0.46 0.66 0.76 0.61 0.52 -0 .6 5 1985 D istrict U.S. D istrict U.S. 0.87% 0.68 0.85 0.92 0.87 0.66 0.98 — 0.62% 0.02 0.44 0.60 0.70 0.59 0.75 0.57 0.84% 0.71 0.80 0.95 0.97 0.54 0.87 — 0.68% 0.27 0.67 0.74 0.84 0.76 0.85 0.50 Return on E quity (ROE) 1988 1987 1985 1986 D istrict All banks < $25 million in assets $25-$50 million $50-$100 million $100-$300 million $300 million-$1 billion $1 -$10 billion > $10 billion U.S. D istrict U.S. D istrict U.S. D istrict U.S. 11.72% 8.24 10.65 11.46 11.89 12.95 12.50 — 13.02% 3.15 6.98 9.16 10.30 9.52 12.39 19.11 10.28% 7.16 10.13 10.90 11.71 13.67 7.96 — 1.81% 1.55 5.27 7.88 9.80 8.76 8.29 -1 4 .8 8 11.26% 7.37 9.77 10.93 11.09 8.81 14.59 — 9.60% 0.20 5.11 7.46 9.29 8.45 11.72 10.71 10.85% 7.68 9.25 11.41 12.42 7.04 13.47 — 10.65% 2.75 7.70 9.11 11.22 10.34 13.54 10.00 SOURCE: FDIC Reports of Condition and Income for Insured Commercial Banks, 1985-1988. As table 1 reports, the 1988 average ROA and ROE for Eighth District banks was 0.93 percent and 11.72 percent, respectively. Nationally, banks reported an average ROA o f 0.83 percent and an average ROE of 13.02 percent. For each o f the years presented, District ROA averages outperformed national averages. In 1988, ROAs for both the District and the nation improved significantly over 1987, when they w ere de pressed by poor earnings associated with sus pect foreign loans at the nation’s largest banks. Table 1 also shows ROAs and ROEs for seven asset-size categories of commercial banks. Across virtually every category, both Districtwide and nationwide, 1988 was a year of improvement. Of particular note are the strong earnings at banks with assets between $1 billion and $10 billion. District banks in this asset range reported average ROAs of 0.82 percent in 1988, up from 0.51 percent in 1987. Nationally, these banks reported a jump in ROA from 0.52 percent to 0.79 percent. ROA for banks with assets more than $10 bilion (none of which are in the Eighth District) was the highest of the size groupings at 0.96 percent, a substantial im provement from -0.65 in 1987. Another bright note in 1988 was the con tinued earnings improvement at smaller banks. For the periods reported in table 1, 1988 was the year in which District banks with assets less than $100 million earned their highest ROAs and ROEs. Higher earnings for these banks were the direct result o f lower loan loss provi sions and a decline in loan charge-offs. Margin Analysis The financial success of a bank depends on its management’s ability to generate sufficient revenue while controlling costs. Tw o important measures of management’s success are net in terest and net noninterest margins. Net interest margin is the difference between what a bank earned on loans and investments and what it paid its depositors, divided by MAY/JUNE 1989 16 Table 2 Net Interest Margin 1988 1987 1986 1985 D istrict All banks < $25 million in assets $25-$50 million $50-$100 million $100-$300 million $300 million-$1 billion $1-$10 billion > $10 billion U.S. D istrict U.S. D istrict U.S. D istrict U.S. 4.26% 4.33 4.29 4.21 4.26 4.49 4.17 4.27% 4.55 4.56 4.56 4.59 4.56 4.45 3.85 4.27% 4.45 4.35 4.33 4.39 4.55 3.97 4.09% 4.61 4.59 4.59 4.59 4.56 4.36 3.39 4.40% 4.68 4.56 4.56 4.44 4.46 4.14 4.18% 4.73 4.75 4.77 4.68 4.65 4.25 3.60 4.31% 4.58 4.21 4.16 4.54 4.61 4.07 4.21% 4.77 4.60 4.52 4.84 4.75 4.41 3.49 — — — — NOTE: Interest income has been adjusted upward for the taxable equivalence on tax-exempt state and local securities. SOURCE: FDIC Reports of Condition and Income for Insured Commercial Banks, 1985-1988. average earning assets.5 This ratio indicates how well interest-earning assets are being employed relative to interest-bearing liabilities.6 Higher net interest margins were one of the driving forces behind stronger earnings at the larger banks in both the District and the United States in 1988.7 As table 2 shows, District banks with assets between $1 and $10 billion reported an average net interest margin o f 4.17 percent, a 20 basis-point increase from 1987. Nationally, these banks reported average net interest margins o f 4.45 percent, up from 4.36 percent in 1987. The largest banks in the nation, those with assets more than $10 billion, recorded an average net interest margin o f 3.85 percent, up 46 basis points from 1987 averages. At banks with assets less than $1 billion, net interest margins declined both Districtwide and nation wide in 1988. Banks across the nation, however, outperformed banks in the Eighth District for each of the asset-size categories reported in table 2. For District banks, interest income rose from 9.35 percent o f average earning assets in 1987 5Earning assets include: loans (net of unearned income) in domestic and foreign offices; lease financing receivables; obligations of U.S. government, states and political sub divisions and other securities; assets held in trading ac counts; interest-bearing balances due from depository in stitutions; federal funds sold and securities purchased under agreements to resell. 6On the asset side, this includes both interest income and fees related to interest-earning assets. Examples include: interest on loans; points on loans; income on tax-exempt http://fraser.stlouisfed.org/ Federal Reserve Bank ofRESERVE BANK OF ST. LOUIS FEDERAL St. Louis to 9.63 percent in 1988. As figure 1 shows, in terest income as a percent of earning assets was, on average, lower at District banks than at U.S. banks for each year except 1986. Nation ally, interest income as a percent o f earning assets rose from 9.62 percent in 1987 to 10.31 percent in 1988. In contrast, interest-related ex penses, while rising from 5.08 percent of earn ing assets in 1987 to 5.37 percent in the District in 1988, were lower than the 1988 national average o f 6.04 percent. The net noninterest margin is an indicator of the efficiency o f a bank’s operations and its pricing and marketing decisions. The net noninterest margin is the difference between noninterest income (other) and noninterest ex pense (overhead) as a percent of average assets. Since noninterest expense generally exceeds noninterest income, the calculation yields a negative number; it is common practice, however, to report the net noninterest margin as a positive number. Thus, smaller net noninterest margins indicate better bank perfor mance, holding all other things constant. municipal loans and bonds and income from holdings of U.S. government securities. On the liability side, interest expense includes: the amount paid on all categories of interest-bearing deposits; federal funds purchased and capital notes. 7Bank management should be concerned not only with the level of the net interest margin, but also with its variability over time. With volatile interest rates, the stability of the net interest margin indicates that the interest sensitivity of assets and liabilities is matched. 17 Figure 1 Interest Income and Interest Expense as a Percent of Average Earning Assets Percent Percent Source: FDIC Reports of Condition and Income for Insured Commercial Banks, 1985-1988. To supplement income generated from interest-earning assets, banks have attempted to generate more fee-related income. For example, service charges on deposit accounts, leasing in come, trust activities income, credit card fees, mortgage servicing fees and safe deposit box rentals. Noninterest expense includes all the expense items involved in overall bank operations, such as employee salaries and benefits as well as ex penses o f premises and fixed assets. Noninterest expense also covers such items as directors’ fees, insurance premiums, legal fees, advertising costs and litigation charges. For the periods presented in table 3, District banks have lagged national averages in terms of generating noninterest sources of revenue. Noninterest expense, on the other hand, has continually been lower at District banks than for banks across the nation. In 1988, non interest income continued to average around 1 percent of average assets at District banks. Noninterest expense also remained virtually flat at about 3 percent of average assets. Noninter est expenses generally have been declining, par ticularly at District banks with assets between $300 million and $1 billion. In recent years, banks have undertaken numerous consolidation and cost-control measures to reduce fixed overhead costs. For many banks, cost reduc tions, including staff cuts, could have been a main contributor to profits in 1988. Loan and Lease Loss Provision Declining loan and lease loss provision levels helped boost earnings both in the District and the nation last year. In 1987, many large banks allocated huge sums to their loan and lease loss provision account to allow for their deteriora ting foreign loan portfolio. This was a precau MAY/JUNE 1989 18 Table 3 Noninterest Income and Noninterest Expense as a Percentage of Average Assets N oninterest Income 1988 1987 1986 1985 D istrict All banks < $ 2 5 million in assets $25-$50 million $50-$100 million $100-$300 million $300 million-$1 billion $1-$10 billion > $10 billion U.S. D istrict U.S. D istrict U.S. D istrict U.S. 0.98% 0.58 0.55 0.55 0.74 1.23 1.51 — 1.44% 0.89 0.75 0.78 0.87 1.08 1.49 1.84 1.00% 0.57 0.53 0.53 0.77 1.39 1.52 — 1.39% 0.95 0.70 0.74 0.88 1.10 1.44 1.77 1.01% 0.55 0.52 0.52 0.73 1.25 1.69 — 1.27% 0.85 0.70 0.74 0.88 1.11 1.39 1.53 0.94% 0.55 0.52 0.53 0.73 1.14 1.63 — 1.18% 0.79 0.72 0.74 0.86 1.08 1.43 1.28 N oninterest Expense 1988 1987 1986 1985 D istrict All banks < $ 2 5 million in assets $25-$50 million $50-$100 million $100-$300 million $300 million-$1 billion $1-$10 billion > $ 1 0 billion U.S. D istrict U.S. D istrict U.S. D istrict U.S. 2.97% 3.07 2.71 2.56 2.76 3.32 3.27 — 3.27% 3.77 3.29 3.18 3.23 3.33 3.42 3.13 2.98% 3.08 2.69 2.57 2.81 3.37 3.27 — 3.26% 3.83 3.28 3.19 3.23 3.36 3.41 3.10 2.98% 3.09 2.65 2.59 2.74 3.46 3.30 — 3.20% 3.77 3.28 3.21 3.24 3.45 3.35 2.95 2.97% 3.04 2.62 2.57 2.76 3.63 3.28 — 3.13% 3.69 3.24 3.18 3.24 3.42 3.44 2.69 SOURCE: FDIC Reports of Condition and Income for Insured Commercial Banks, 1985-1988. tionary measure to absorb expected future losses. Having taken this action in 1987, many banks saw little need to increase provision levels in 1988. Loan and lease loss provision totaled $450.6 million at District banks in 1988, down $246.6 million from 1987 levels. As reported in table 4, Eighth District banks decreased their provision for loan and lease losses to 0.37 percent of average assets, a sharp drop from 0.60 percent in 1987. This decrease can be traced primarily to the largest District banks. For those banks, provision for loan and lease losses fell from 0.97 percent o f average assets in 1987 to 0.46 per cent in 1988. Nationally, banks decreased their loan and lease loss provision by $20.2 billion and, at year-end 1988, the account stood at $17.2 billion. As a percent o f average assets, loan and lease loss provision was 0.51 percent in 1988, a substantial decline from 1.24 percent in 1987. As with the District, the largest banks were primarily responsible for the decrease as their ratio fell from 2.02 percent in 1987 to 0.42 per cent in 1988. http://fraser.stlouisfed.org/ FEDERAL RESERVE Federal Reserve Bank of St. Louis BANK OF ST. LOUIS ASSET QUALITY As it has for some time, asset quality con tinues to be a primary factor influencing the banking industry’s earnings pattern. With loan losses rising over the past few years at many commercial banks, investors and regulators alike are focusing on asset quality in assessing the health o f the banking industry. Asset quality typically is measured by two in dicators. The first measure, the nonperforming loan rate, indicates both the current level of problem loans as well as the potential for future loan losses. The second indicator, the ratio of net loan losses to total loans, shows the percen tage o f loans actually written o ff the bank’s books. N onperform ing Loans and Leases The level o f nonperforming assets includes all loans and lease financing receivables that are 90 days or more past due, are in nonaccrual status or are restructured because of a deterioration in the financial position of the obligor. In the District, nonperforming assets decreased $246.5 19 Table 4 Loan and Lease Loss Provision as a Percentage of Average Assets 1988 1987 1986 1985 D istrict All banks < $ 2 5 million in assets $25-$50 million $50-$100 million $100-$300 million $300 million-$1 billion $1-$10 billion > $10 billion U.S. D istrict U.S. D istrict U.S. D istrict U.S. 0.37% 0.30 0.32 0.30 0.36 0.36 0.46 0.51% 0.60 0.53 0.47 0.47 0.59 0.63 0.42 0.60% 0.49 0.43 0.41 0.45 0.42 0.97 1.24% 0.82 0.72 0.60 0.56 0.70 0.90 2.02 0.59% 0.68 0.67 0.62 0.64 0.68 0.46 0.78% 1.15 0.97 0.85 0.75 0.85 0.67 0.80 0.59% 0.80 0.76 0.64 0.53 0.61 0.43 0.67% 1.07 0.88 0.81 0.63 0.63 0.57 0.70 — — — — SOURCE: FDIC Reports of Condition and Income for Insured Commercial Banks, 1985-1988. Table 5 Nonperforming Loans and Leases as a Percentage of Total Loans 1988 1987 1986 1985 D istrict All banks < $25 million in assets $25-$50 million $50-$100 million $100-$300 million $300 million-$1 billion $1-$10 billion > $10 billion U.S. D istrict U.S. D istrict U.S. D istrict U.S. 1.62% 1.80 1.74 1.67 1.70 1.25 1.65 — 2.96% 2.65 2.45 2.16 1.89 2.38 1.91 4.48 2.11% 2.13 2.14 2.05 1.95 1.47 2.44 — 3.49% 3.16 2.76 2.45 2.20 2.29 2.41 5.26 2.16% 2.66 2.61 2.45 2.04 2.33 1.81 -- 2.77% 3.76 3.19 2.93 2.53 2.51 2.06 3.37 2.49% 3.26 3.05 2.67 2.11 2.65 2.19 — 2.83% 3.73 3.31 3.06 2.57 2.45 2.24 3.34 SOURCE: FDIC Reports of Condition and Income for Insured Commercial Banks, 1985-1988. million from 1987 to 1988. As reported in table 5, Eighth District banks’ nonperforming loans and leases as a share of total loans fell from 2.11 percent in 1987 to 1.62 percent in 1988. Banks across the nation experienced a similar decline as the nonperforming loan rate dropped from 3.49 percent to 2.96 percent. Across all asset-size categories, District banks reported a decrease in nonperforming loans and leases in 1988. District banks with assets less than $25 million saw nonperforming loans and leases fall from 2.13 percent of total loans in 1987 to 1.80 percent in 1988. The largest District banks saw their nonperforming loan rate drop from 2.44 percent to 1.65 percent during the same one-year period. Nationally, this pattern also held true as most asset-size cat egories reported a decline in the nonperforming loan rate. The only exception was at banks with assets between $300 million and $1 billion where nonperforming loans and leases rose to 2.38 percent of total loans, up from 2.29 per cent in 1987. Figure 2 shows the distribution of nonperfor ming loans by loan type for Eighth District banks. At year-end 1988, nonperforming agricultural loans as a percent of total nonper forming loans was 5.45 percent, down from 6.84 percent in 1987. The percentage o f nonper forming commercial loans fell from 45.91 per cent of the total to 41.14 percent. Consumer nonperforming loans, which accounted for 6.88 percent o f the total in 1987, rose to 8.61 per cent in 1988. Nonperforming real estate loans had a fairly substantial increase in 1988, rising from 35.81 percent in 1987 to 42.45 percent. MAY/JUNE 1989 20 Figure 2 District Distribution of Nonperforming Loans by Loan Type Percent Percent | Agriculture □ Consumer □ Real Estate □ Commercial 50 50 40 40 30 30 20 20 10 10 B A 1985 1986 1987 1988 Source: FDIC Reports of Condition and Income for Insured Commercial Banks, 1985-1988. Note: Percentages may sum to greater than 100 because agricultural loans are included in other categories as well. Loan and Lease Losses The most direct measure o f a bank’s loan pro blems is the percentage of loans and leases charged-off during the year. Net loan and lease losses (adjusted for recoveries) amounted to $510.8 million at District banks in 1988, an in crease of $46.7 million from 1987. Nationally, banks charged-off $17.6 billion in 1988, $1.5 billion more than in 1987. As table 6 shows, the average charge-off rate at banks in the Eighth District rose slightly in 1988, from 0.70 percent of total loans in 1987 to 0.72 percent. Nation ally, the average charge-off rate rose from 0.89 percent o f total loans in 1987 to 0.93 percent in 1988. Across virtually every asset-size category, charge-off rates at District banks were lower than at their national counterparts. The only ex ception was at the largest District banks where net loan losses and leases to total loans jumped http://fraser.stlouisfed.org/ FEDERAL RESERVE Federal Reserve Bank of St. Louis BANK OF ST. LOUIS sharply in 1988, from 0.68 percent in 1987 to 1.18 percent. Table 7 shows the distribution of loan losses by type of loan. For both the nation and the District, commercial loan losses contributed the greatest percentage to overall loan loss. The percent of District commercial loan charge-offs fell in 1988, from more than 50 percent o f total loan losses in 1987 to approximately 44 percent. Farm-related charge-offs declined further in 1988 and now account for slightly more than 2 percent o f total District loan losses. The percen tage of District consumer loan charge-offs also declined in 1988, falling from 23.24 percent in 1987 to 17.88 percent of total loan losses. Only one category, loans held in foreign offices, in creased in 1988. Loan losses for this category increased to 17.51 percent of overall loan loss, up substantially from 1.79 percent in 1987. 21 Table 6 Net Loan and Lease Losses as a Percentage of Total Loans 1988 1987 1986 1985 D istrict D istrict U.S. D istrict U.S. D istrict U.S. 0.72% 0.60 0.51 0.46 0.50 0.42 1.18 All banks < $ 2 5 million in assets $25-$50 million $50-$100 million $100-$300 million $300 million-$1 billion $ 1-$ 10 billion >$10 billion U.S. 0.93% 1.13 0.88 0.71 0.66 0.79 0.94 1.06 0.70% 0.95 0.73 0.70 0.67 0.71 0.68 0.89% 1.50 1.17 0.96 0.78 0.88 0.86 0.88 0.88% 1.32 1.16 1.07 0.99 0.92 0.57 0.94% 2.02 1.61 1.35 1.03 0.99 0.73 0.89 0.89% 1.52 1.38 1.09 0.72 0.78 0.59 0.81% 1.71 1.38 1.22 0.84 0.74 0.64 0.77 — — — — SOURCE: FDIC Reports of Condition and Income for Insured Commercial Banks, 1985-1988. CAPITAL ADEQUACY Table 7 Distribution of Loan Losses 1988 1987 1986 1985 District Agriculture Commercial Consumer Real estate Foreign1 2.24% 44.42 17.88 16.63 17.51 8.26% 51.51 23.24 19.10 1.79 16.10% 62.22 18.56 17.05 0.16 19.44% 65.60 14.03 18.18 0.37 United States Agriculture Commercial Consumer Real estate Foreign’ 0.77% 36.47 26.39 13.42 19.70 3.35% 45.18 28.66 15.20 6.30 7.66% 55.73 26.89 11.76 1.13 10.35% 61.07 23.13 8.59 2.55 1Loans held in foreign offices, Edge and Agreement subsidiaries and International Banking Facilities (IBFs). NOTE: Percentages may sum to greater than 100 because agricultural loans are included in other categories as well. SOURCE: FDIC Reports of Condition and Income for In sured Commercial Banks, 1985-1988. 8The components of primary capital as defined in the FDIC Consolidated Report of Condition and Income are: com mon stock; perpetual preferred stock; surplus; undivided profits; contingency and other capital reserve; qualifying mandatory convertible instruments; allowance for loan and lease losses and minority interests in consolidated sub sidiaries, less intangible assets excluding purchased mor tgage servicing rights. (For the purposes of this paper, on ly the goodwill portion of intangible assets was deducted.) Secondary capital is limited to 50 percent of primary capital and includes subordinated notes and debentures, limited-life preferred stock and that portion of mandatory convertible securities not included in primary capital. Each Bank regulators have a strong interest in en suring that banks maintain adequate financial capital. Bank capital is intended to absorb losses, cushion against risk, provide for asset ex pansion and protect uninsured depositors. Given its importance, the regulatory agencies have set minimum standards of 5.5 percent primary capi tal to assets and 6 percent total capital to assets.8 These standards have been revised re cently and, on December 16, 1988, the Federal Reserve Board approved new risk-based capital guidelines intended to encourage banks to make safer investments.9 The improved performance of District banks had a favorable effect on their capital levels. As table 8 indicates, improvement in bank primary capital ratios is apparent throughout most assetsize ranges. Average primary capital ratios for banks both in the District and nationwide are well above the current minimum standards established by the regulatory agencies. Nation ally, an average primary capital ratio of 7.92 bank’s secondary capital is added to its primary capital to obtain the total capital level for regulatory purposes. 9The guidelines establish a systematic framework whereby regulatory capital requirements are more sensitive to dif ferences in risk profiles among banking organizations. In addition, off-balance sheet activity is evaluated for risk ex posure. The guidelines provide for a phase-in period through the end of 1992 at which time the standards become fully effective. Starting December 31, 1990, the level of capital that banks are required to hold will in crease to 7.25 percent of qualifying total capital to weighted risk assets and, finally, to 8 percent in 1992. MAY/JUNE 1989 22 Table 8 Primary Capital Ratios 1987 1988 1986 1985 D istrict All banks < $ 2 5 million in assets $25-$50 million $50-$100 million $100-$300 million $300 million-$1 billion $1-$10 billion > $10 billion U.S. D istrict U.S. D istrict U.S. D istrict U.S. 8.73% 10.44 9.70 9.49 8.86 8.55 7.66 — 7.92% 10.82 9.68 9.24 8.62 7.90 7.45 7.58 8.73% 10.14 9.53 9.37 8.71 8.50 7.89 — 7.80% 10.59 9.49 9.07 8.53 7.87 7.50 7.26 8.47% 9.97 9.27 9.08 8.50 8.30 7.52 — 7.56% 10.36 9.30 8.82 8.26 7.81 7.31 6.87 8.38% 9.88 9.21 8.91 8.37 8.35 7.18 — 7.44% 10.58 9.33 8.78 8.17 8.07 7.12 6.51 SOURCE: FDIC Reports of Condition and Income for Insured Commercial Banks, 1985-1988. percent was reported, up slightly from 1987. While District aggregate primary capital ratios remained the same in 1988, some asset-size groups showed notable improvement. In par ticular, the smallest District banks reported an average primary capital ratio of 10.44 percent in 1988, up from 10.14 percent in 1987. In con trast, the largest District banks reported a decline in their average primary capital ratio, falling from 7.89 percent in 1987 to 7.66 per cent in 1988. As of December 1988, six banks or 0.5 per cent of all District banks fell short of the minimum regulatory primary capital standards. This number was down from 15 banks in 1987. Nationally, 465 banks had deficient primary capital ratios at year-end 1988, compared with 474 in 1987. CONCLUSION 1988 marked a year of recovery from the overall poor earnings reported by banks across the nation in 1987. Bank performance in the Eighth Federal Reserve District improved in BANK OF ST. LOUIS FEDERAL RESERVE 1988, propelled by lower loan loss provisions. Aggregate bank profit ratios improved as many of the District’s largest banks began to rebound from the negative earnings associated with in creased loan loss provisions tied to foreign loans. Profits recouped across virtually every as set size category o f Eighth District commercial banks. The smaller District banks employed higher earnings as both loan losses and loan loss provisions levels declined. As with most o f the banking industry, better asset quality helped to improve earnings at District banks last year. Finally, a majority of Eighth District banks improved their primary capital ratios in 1988 and are positioned well above the minimum standards set by bank regulators. The banking industry in the Eighth District has returned to profitability, and, barring any shocks, should continue to improve in the com ing quarters. With a continued positive eco nomic environment, loan problems that have plagued District banks should abate and as the level of nonperforming loans declines, future loan problems should be less severe. 23 Thomas B. Mandelbaum Thomas B. Mandelbaum is an econom ist at the Federal Reserve Bank o f St. Louis. Thomas A. Pollmann provided research assistance. The Eighth District Business Economy in 1988: Still Expanding; But More Slowly ll HE BUSINESS economy o f the Eighth Federal Reserve District continued to expand in 1988, its sixth successive year o f growth. District income and employment reached record highs, while the regional unemployment rate declined to its lowest level of the decade. Unlike the previous two years, in which regional eco nomic growth approximated the national pace, however, the District economy grew substantial ly slower than the rest of the nation last year.1 This article discusses the factors that caused this sluggishness and describes other significant developments in the Eighth District’s business economy during 1988. In addition, it provides a perspective on future economic conditions in District states. PERSONAL INCOME AND CONSUMER SPENDING As figure 1 shows, District personal income growth during the current expansion has ex 1Data from Arkansas, Kentucky, Missouri and Tennessee are used to represent the Eighth District. 2Growth rates compare data for the entire year with the average of previous years. The substantially slower growth of District income does not necessarily imply that District output also grew slower than the national average. For ex ceeded the national rate only in 1984. After ap proaching the nation’s 3.2 percent gain in 1987, District real income growth slowed to 2.6 per cent in 1988, while U.S. real income rose 3 percent.2 The region’s relatively weak income growth in 1988 stemmed largely from its sluggish growth in real earnings, which make up about twothirds of total income: real earnings rose 2.3 percent regionally compared with 2.7 percent nationally. The other sources of personal in come—transfer payments and dividends, interest and rent—also grew slower regionally than na tionally in 1988. The drought caused real personal farm in come to fall sharply in Kentucky and Missouri last year, but had little direct effect on the ex pansion o f total personal income. Personal in come earned from farms has accounted for less than 2 percent of the region’s or nation’s total in recent years. Excluding it from income did not substantially change regional or national 1988 growth rates. ample, while the annual growth rate of District income be tween 1982 and 1986 was 0.4 percentage points less than the national rate, the growth rate of total output was just 0.1 percentage points lower. See Mandelbaum (1988/89). MAY/JUNE 1989 24 Figure 1 Annual Percent Change in Real Personal Income Percent Percent United States [ ] Arkansas [ ] Missouri Eighth District | Kentucky □ Tennessee -1 -1 1983 1984 1985 Though District personal income growth trailed the national average in 1988, it was typi cal of states in the nation’s interior. In 1988 and throughout the recovery, the economies o f most interior states have grown more slowly than those of states on either coast.3 Between III/1987 and III/1988, for instance, the District’s growth of real nonfarm income matched that of non 3This comparison excludes Alaska and Hawaii. See Coughlin and Mandelbaum (1988) for an overview of regional growth of per capita income in the 1980s. http://fraser.stlouisfed.org/ Federal Reserve Bank ofRESERVE BANK OF ST. LOUIS FEDERAL St. Louis 1986 1987 1988 coastal states, but was a full percentage point slower than the 4 percent rate posted by coastal states. Much of the strong coastal growth last year stemmed from the sharp expansion of earnings in service-producing industries, construction and, in many states, durables manufacturing. Earlier in the decade, the rapid expansion of 25 Figure 2 Annual Percent Change in Payroll Employment P erce nt P erce nt -1 high-tech industries, often related to defense projects, fueled coastal growth as well. Mean while, the economic expansion o f some interior states was hampered by the decline in com modity prices, particularly energy prices. Despite some strengthening during the year, re latively low energy prices continued to depress growth in energy-producing states in 1988. In 1986 and 1987, the District’s income growth was able to approach the national average largely because o f Tennessee (see figure 1). Unlike most interior states, the Tennessee economy expanded much faster than the na tional average. In 1988, when Tennessee’s in come growth fell back to near the national average, the District’s income growth fell fur ther below the national average, as did most of the nation’s interior. District retail sales have followed national trends during the first five years o f the recov ery, but expanded more slowly in 1988. Be tween 1982 and 1987, sales rose at 3.6 percent and 3.8 percent annual rates in the District and nation, after adjusting for inflation. In 1988, District real retail sales growth was a weak 1.6 percent compared with a national increase of 2.6 percent. LABOR MARKETS Employment data, presented in figure 2, tell essentially the same story as the income and sales data told: the District’s expansion contin ued in 1988 but at a slower pace compared with either the previous few years or the na tional average. Nonagricultural payroll employ ment rose 2.3 percent last year in the District, while growing a robust 3.6 percent nationally. The District’s weaker growth contrasts with the similarity of regional and national growth dur ing the previous years o f the the recovery: be tween 1982 and 1987, payroll employment grew at a 2.9 percent annual rate in the District and at a 2.7 percent rate nationally. The District’s 1988 unemployment rate was the lowest this decade. It averaged 6.5 percent, MAY/JUNE 1989 26 Figure 3 Percent Change in Employment □ District □ United States Total Payroll Employment Wholesale/Retail Trade Manufacturing Services Government Transportation and Utilities Finance, Insurance and Real Estate Construction Mining J_____L -5 -4 -3 -2 -1 0 1 2 3 4 5 6 P e rc e n t C h a n g e , 1987-88 down from 7.2 percent in the previous year and 10.8 percent in 1983. Although civilian employ ment (upon which the jobless rate is based) rose only 1.8 percent in 1988, the unemployment rate fell as the labor force grew even more slowly, rising by only 1 percent. The relatively sluggish District job growth raises two questions. First, where in the nation have these new jobs been created? As was true of personal income, a disproportionate share of the nation’s new jobs in 1988 were created in coastal states. California, Florida, New York and Virginia gained the most jobs, for example, generating approximately one quarter of the na tion’s new jobs last year. Illinois, Ohio, Penn sylvania and Texas also posted large job gains. The second question raised by the pattern of job growth is why the District’s job growth was slower this year, after several years of near national growth. As was true of income growth, Tennesee’s faster-than-national employment ex pansion allowed the District workforce to grow at nearly the national pace in most previous years of the recovery (see figure 2). In 1988, http://fraser.stlouisfed.org/ FEDERAL RESERVE Federal Reserve Bank of St. Louis BANK OF ST. LOUIS however, employment in Tennessee grew at well below the national rate and District em ployment growth followed suit. Another way to understand the factors that account for the relatively slow District job growth in 1988 is to consider the performance o f the District economy’s major sectors. As figure 3 shows, employment in all major sectors of the District economy grew more slowly in 1988 than the national average. In the figure, the eight divisions o f payroll employment are ordered in descending size, ranging from the wholesale and retail trades sector, which employed almost a quarter o f the 1988 District workforce, to mining, which employed less than 1 percent. The largest four sectors—trades, manufacturing, services and government— account for more than four-fifths o f all total payroll employment. GOODS-PRODUCING SECTORS Although the 1.8 percent increase in District manufacturing employment last year trailed the nation’s 2.5 percent increase, it represents a 27 slight acceleration from the 1.2 percent annual rate o f increase over the previous five years. In 1988, manufacturing growth was stimulated by a surge in exports. The value of U.S. manufac tured exports rose 26.3 percent in 1988. Pro ducers o f nonelectrical machinery, an industry that includes computers and most capital goods, accounted for much of the rise in exports. Employment in the District’s nonelectrical machinery industry rose 3 percent in 1988 com pared with the nation’s 6.7 percent rise. Among the District’s other large manufactur ing industries, the fabricated metals and elec trical equipment sectors also experienced moderate growth last year. The number of jobs in the apparel and textiles mill products de clined, however, because of a fashion shift away from denim products and rising textile inven tories. Transportation equipment employment also declined. The District transportation equip ment sector is dominated by defense-related aerospace production—in which employment rose—and motor vehicles production—in which periodic layoffs, strikes and a Missouri auto assembly plant closure led to job losses. After expanding sharply in the first two years of the recovery, District construction activity had leveled o ff in subsequent years until 1988, when building weakened substantially. The real value of District building contracts fell 8.3 per cent last year. District building activity has fol lowed national trends during the current recovery, although last year’s drop in building contracts was somewhat more severe than the nation’s 5.3 percent decline. The weaker District performance stemmed from the residential sector; District residential contracts dropped by 11.8 percent in 1988, al most twice the national decline. Building permit data also show that residential activity declined last year: District housing permits fell 6.1 per cent in 1988 while dropping 4.7 percent nation ally. Both single-family and multi-family residen tial building weakened last year. Throughout the recovery, District residential construction has expanded more slowly than the national average (reflecting the region’s slower population growth), while nonresidential building has been stronger than at the national level. In 1988, the real value of nonresidential contracts dropped 4.3 percent regionally and 6.6 percent nationally. While mining employment rose 1.6 percent na tionally, it dropped 4.9 percent in the District. This differential is due largely to differences in the composition of mining. Nationally, oil and gas extraction accounts for most mining jobs while coal mining dominates the District mining workforce. The nation’s increase in mining jobs last year was due entirely to increases in oil and gas extraction; employment was flat in the remaining mining sectors. Coal production has remained strong in re cent years, as much of the the nation’s expan ding industrial activity has been fueled by coal generated electricity. Mine production in Ken tucky, which accounts for most of the District’s mining output, reached its highest point of the decade in 1987, then fell slightly, by 1.9 per cent, in 1988. Rapidly increasing productivity allowed a more severe 6.3 percent drop in Ken tucky coal mine employment in 1988. SERVICE-PRODUCING SECTORS The wholesale and retail trades sector is the District’s largest in terms o f employment. Since its growth is related to a region’s income growth, the slower 1988 growth of District trades is not surprising. Trades employment rose 2.9 percent regionally and 4 percent na tionally in 1988. During the first five years of the recovery, both the District and national trades sectors grew at annual rates of a little less than 4 percent. Much of the slower District employment growth is attributable to the slower expansion of the services sector. Approximately half o f the services jobs are in business and health services with the remainder in legal, personal and miscellaneous services. At both the regional and national levels, services has generated more new jobs—in 1988 and throughout the recovery—than any other sector. The increase of District services jobs of 3.3 percent in 1988, however, fell far short of the national average of 5.2 percent. The District’s 1988 growth rate also represented a deceleration from its 4.9 per cent annual pace over the previous five years. To some extent, the slower growth of the District services sector is related to the relative ly sluggish growth o f District manufacturing, for which the services sector provides business and legal services. Government employment grew at a 1.4 per cent annual rate during the first five years of the recovery in both the District and the nation. In 1988, growth was slightly stronger as govern MAY/JUNE 1989 28 Table 1 1988 Growth of Selected Economic Indicators U.S. Real personal income Payroll employment Manufacturing Construction Mining Wholesale and retail trade Services Government Finance, insurance and real estate Transportation and public utilities Real value of building contracts1 D istrict Arkansas 3.0% 3.6 2.5 5.8 1.6 4.0 5.2 2.2 2.6% 2.3 1.8 3.0 -4 .9 2.9 3.3 2.0 2.8% 3.0 3.7 0.0 3.7 2.9 4.4 2.2 K entucky M issouri Tennessee 2.3% 3.3 4.4 8.6 - 6 .3 5.0 4.0 0.9 2.2% 1.4 0.4 0.6 -1 .7 2.3 2.0 1.1 3.2% 2.5 0.8 3.2 - 4 .7 2.2 3.9 3.7 2.0 0.8 0.5 1.2 0.7 0.8 3.7 - 5 .3 1.6 - 8 .3 1.7 -1 7 .4 0.4 - 7 .2 0.6 -6 .1 3.7 - 8 .6 'Excludes nonbuilding construction. SOURCE: F. W. Dodge Construction Potentials. ment employment rose by 2 percent in the Dis trict and 2.2 percent nationally. In addition to directly providing jobs, the federal government influences the District economy through its spending in District states. Federal expenditures include grants-in-aid, di rect payments to individuals and procurement contracts as well as salaries and wages. In fiscal year 1988, federal government spending in District states totaled $55.4 billion, or 2.6 per cent more than in the previous year. Expen ditures increased in each of the District states, ranging from 2.3 percent in Arkansas to 3.1 percent in Kentucky. Of the District total, $7.9 billion was received in the form of defense pro curement contracts, down 4 percent from 1987. The decline was largely due to an 8.2 percent drop in Missouri, which received $5.5 billion in defense contracts in fiscal year 1988. Defense contracts have declined in recent years follow ing rapid growth during the first half of the decade. After identical 4.2 percent annual growth rates between 1982 and 1987, employment in both the District and national finance, insurance and real estate sectors grew more slowly in 1988, rising by just 0.8 percent regionally and 2 percent nationally. Consolidations among banks and thrifts and a slowdown in the sales and construction of homes contributed to the slowdown. http://fraser.stlouisfed.org/ FEDERAL RESERVE Federal Reserve Bank of St. Louis BANK OF ST. LOUIS Employment in the District’s transportation and public utilities firms rose just 1.6 percent in 1988, after growing at a slightly stronger 2.3 percent rate during the previous five years. In 1988, as well as throughout the recovery, much of the District’s weakness was concentrated in Kentucky and Missouri. INTERSTATE COMPARISONS Although discussions o f the District’s economy provide an overview o f broad regional trends, they obscure substantial differences among in dividual states. Wide variations in annual in come and employment growth, for example, can be seen in figures 1 and 2 as well as in table 1. This section highlights these and other differ ences among the state economies. Arkansas Arkansas enjoyed moderate economic growth during 1988, despite a sharp decline in con struction activity. As figure 1 shows, the state’s 2.8 percent rise in real personal income was just slightly below the national average. The earnings component o f personal income rose marginally faster in Arkansas than in the na tion, in part because of the strong growth from farms. Arkansas’ income growth was impeded, however, by the relatively slow expansion of transfer payments and dividends, interest and rent. 29 Figure 4 Unemployment Rates: United States and District States Percent Percent in 1988. Arkansas’ rate has declined relatively slowly during the recovery, however. The 7.7 per cent jobless rate for 1988 is just 2.1 percentage points below its 1982 rate, compared with a 4.2 percentage-point drop nationally. Payroll employment rose at a moderate 3 per cent in 1988, the same rate as during the first five years of the recovery. Manufacturing, one of the state’s chief strengths in the period, continued strong in 1988, with manufacturing jobs growing by 3.7 percent. Employment in factories making fabricated metals and transportation equipment rose rapidly. Although the state lacks a major vehicle-assembly plant, parts suppliers expanded in response to the growing needs of car and truck makers throughout the region. Employment growth in food processing- the state's largest manufacturing industry-slowed in 1988 to 1.9 per cent from a 5.1 percent pace during the previous five years. Kentucky As figure 4 shows, unemployment rates fell in Arkansas, as they did in each of the District states Kentucky’s economic growth during 1988 was mixed. While employment growth was fairly Construction activity, weak throughout the recovery, declined sharply in 1988 (see table 1). After rising in 1983 and 1984, the real value of building contracts declined each subsequent year and dropped 17.4 percent in 1988, the most severe decline among the District states. Both residential and nonresidential building declined sharply last year. MAY/JUNE 1989 30 strong, income growth was rather weak and construction activity declined. Real personal in come rose 2.3 percent in 1988. The drought was partially responsible for this sluggish in crease. Kentucky’s 2.6 percent rise in nonfarm personal income matched the District average. Payroll employment rose 3.3 percent last year, the second consecutive year o f moderate growth. Despite these job gains in recent years, the state’s unemployment has remained high. In 1988, the unemployment rate averaged 8 per cent, the highest rate among District states. The jobless rates in the Louisville and Lexington areas are considerably lower, but rates general ly are higher in rural areas, particularly where coal mining has been dominant.4 Much of Kentucky’s recent employment strength stems from the state’s largest sectors: wholesale and retail trades, services and manufacturing. Employment in services and in wholesale and retail trades rose by 4 percent and 5 percent in 1988, with much of the growth in the Louisville area. Kentucky’s manufacturing job growth of 4.4 percent last year was its highest since 1984. Much of this growth can be traced to the ex pansion o f motor vehicle production. The Toyota assembly plant in the LexingtonGeorgetown area hired thousands o f new workers and the light truck plant in Louisville expanded production. Many parts suppliers for these and other assembly plants in the region either expanded or began operations in Ken tucky last year. The state has benefited from the shift to “just-in-time” inventory strategies during the 1980s which require parts suppliers to be near assembly plants. Fabricated metals plants in Kentucky also sharply increased their workforces in 1988, largely because o f increas ed orders from motor vehicle parts suppliers. After growing by nearly 5 percent a year be tween 1982 and 1987, the real value of con struction contracts fell 7.2 percent in 1988, with both residential and nonresidential building con tracts falling. Residential building contracts in the Louisville and Lexington areas remained strong, however, falling only slightly. th ro u g h o u t the nation, unemployment rates tend to be higher in nonmetropolitan areas than in metropolitan areas. In the fourth quarter of 1988, for example, the average U.S. unemployment rate outside of metropolitan areas (not seasonally adjusted) was 5.8 percent compared http://fraser.stlouisfed.org/ Federal Reserve Bank of RESERVE BANK OF ST. LOUIS FEDERAL St. Louis Missouri Missouri’s economic expansion has slowed to a sluggish pace in recent years. In the recovery’s first four years, personal income and employ ment rose moderately, expanding at near the the national rate (see figures 1 and 2). In 1987 and 1988, however, growth trailed the national and District averages. Personal income rose 2.2 percent last year, after adjusting for inflation. The drought severely affected the northern part of the state, but had only a minor impact on overall personal income growth. Rather, it was the slow growth of nonfarm earnings, reflecting the sluggish job expansion in the state, that was largely responsible for Missouri’s relatively slow personal income growth. Missouri’s payroll employment rose just 1.4 percent last year, less than half the national rate. Nevertheless, since the state’s labor force was flat, the unemployment rate declined to 5.6 percent in 1988 from 6.3 percent a year earlier. As table 1 shows, employment in every major sector rose more slowly than in the rest of the nation in 1988. Services and trades—the major sources of Missouri’s job growth between 1982 and 1987—grew at their slowest rate o f the recovery last year. On a positive note, manufac turing employment rose 0.4 percent in 1988, its first increase since 1984. Employment rose slightly in many industrial sectors, but fell in food processing, textiles and apparel, and trans portation equipment industries. The latter, which employs almost one of every six of Missouri’s manufacturing workers, experienced job gains in aircraft manufacturing but had larger losses among producers of motor vehicles and parts. As in the other District states, the real value of construction contracts fell in 1988 in Missouri. The decline in the residential sector was particularly severe, with most o f the weakness in the multi-family housing market. Last year's construction decline contrasts with strong growth during the first five years of the recovery. with 4.9 percent in metropolitan areas. See U.S. Depart ment of Labor (1989), pp.79-80. For a discussion of the slower economic growth in nonmetropolitan areas during the 1980s, see Carraro and Mandelbaum (1989). 31 Tennessee Although moderate, Tennessee’s economic growth in 1988 was considerably slower than in 1986 and 1987 (see figures 1 and 2). Personal income increased by 3.2 percent in 1988, a drop from approximately 5 percent in each o f the previous two years. Payroll employment rose 2.5 percent in 1988, after increasing 4.1 percent in 1987. Tennessee’s 1988 income growth, how ever, exceeded the national average, while its job growth was sufficient to allow the unem ployment rate to fall to 5.8 percent in 1988 from 6.6 percent in 1987. Growth in most sectors of the Tennessee economy slowed last year. The 1988 employ ment increases in Tennessee’s services and trades sectors w ere the smallest in several years. Manufacturing employment growth rose 0.8 percent last year. Employment shrank in many nondurables sectors, such as chemicals and textiles and apparel, while most of the sec tors producing durables rose moderately. Em ployment in the Memphis area rose only slightly in 1988 after several years of strong gains. The impending construction of a $1.2 billion cereal plant should help boost Memphis area job growth in 1989, however. The real value of construction contracts awarded in Tennessee fell 8.6 percent last year after growing 8 percent in 1987. The 1987 strength stemmed from the sharp expansion of nonresidential building. In 1988, nonresidential contracts rose only marginally, while residential contracts plunged 16.4 percent after adjusting for price changes. OUTLOOK FOR 1989 Projections from academic and governmental institutions in each District state suggest that the states’ economies will continue to grow in 1989, but at a slower rate than in 1988. This slowing reflects the strong ties between the states’ economies and the national economy, whose growth is also expected to slow. Many observers of the national economy feel that it cannot continue to expand at the pace of the last few years, given the low level of unemploy ment and high rates o f capacity utilization.5 Table 2 Economic Projections for District States 1988 Unemployment Rate United States Arkansas Kentucky Missouri Tennessee 1989 5.5% 7.7 8.0 5.6 5.8 5.5% 8.1 7.2 6.1 6.3 Percent change’ 1988 Payroll employment United States Arkansas Kentucky Missouri Tennessee 3.6% 3.0 3.3 1.4 2.5 1989 2.2% 1.5 1.3 1.1 1.1 Manufacturing employment United States 2.5 Arkansas 3.7 Kentucky 4.4 Missouri 0.4 Tennessee 0.8 0.6 2.5 1.5 N.A. 0.4 Personal income (current dollars) United States 7.3 Arkansas 7.0 Kentucky 6.5 Missouri 6.4 Tennessee 7.5 7.9 4.6 7.0 8.3 7.1 ’ Percent changes compare entire year with previous year. SOURCES: United States: DRI/McGraw-Hill, Review of the U.S. Economy, January 1989; Arkansas: University of Arkansas at Little Rock, Arkansas Economic Outlook, January 1989; Kentucky: Kentucky Finance and Ad ministration Cabinet based on DRI/Mcgraw-Hill projec tions, January, 1989; Missouri: College of Business and Public Administration, University of Missouri-Columbia, M issouri Econom ic Indicators; 2nd Quarter, 1988; Tennessee: Center for Business and Economic Research, University of Tennessee, Knoxville, An Economic Report to the Governor o f the State of Tennessee On the S tate’s Economic Outlook, January 1989. Table 2 presents actual data for 1988 and pro jections for 1989 for several economic in- 5The Blue Chip consensus forecast (from February 1989) of 51 private economists, for example, indicates that real GNP growth will slow to 2.7 percent in 1989 (full year-overyear comparison) from its 3.8 percent increase in 1988. MAY/JUNE 1989 32 dicators. Projections for the national economy made by DRI/McGraw-Hill also are provided. Although different methodologies were used to generate the various projections, they are all consistent in their forecast of an employment slowdown in 1989. U.S. payroll employment growth is expected to slow to 2.2 percent in 1989 from its 3.6 percent rise last year. Employ ment in each of the states is expected to grow even more slowly than in nation. Reflecting the slow job growth, unemployment rates are ex pected to rise in Arkansas, Missouri and Ten nessee. Lower rates are anticipated in Kentucky, however. To the extent that the projections are correct, manufacturing will provide few er new jobs in 1989 than in 1988. Nationally, manufacturing employment growth is expected to slow to a 0.6 percent rise from the 2.5 percent increase in 1988. Each of the District states for which data are available show a similar pattern of decelera tion. In Arkansas, manufacturing growth is ex pected to slow, in part, because anticipated higher interest rates may slow orders for durables goods, particularly those related to residential investment. In Tennessee, a continua tion of the weakness in the textiles, apparel, lumber and wood products sectors are expected to retard manufacturing growth in 1989. In contrast to employment, national personal income, measured in current dollars, is ex pected to grow faster in 1989 than in 1988. There are several reasons for this. First, ex pected higher inflation in 1989 raises the year’s estimated nominal income figure. Second, DRI expects interest rates will be higher in 1989 which will raise interest income. Finally, BANK OF ST. LOUIS FEDERAL RESERVE transfer payments, particularly for Medicare and Medicaid, are expected to grow rapidly. Personal income is expected to grow more slow ly in 1989 than in 1988 in Arkansas and Ten nessee, while accelerating in Kentucky and Missouri. CONCLUSION The Eighth District’s business economy during 1988 was strong and growing, albeit at a slower rate than in the previous two years. The unemployment rate fell to its lowest level of the decade, and personal income and employment in most sectors continued to expand. Regional growth was weaker than at the national level, however, as job growth in each of the major sectors of the economy trailed the national average and District construction activity de clined more severely. Although projections sug gest that economic growth may slow this year, it is expected that 1989 will be the seventh suc cessive year of growth for both the Eighth District and the nation. REFERENCES Carraro, Kenneth C., and Thomas B. Mandelbaum. “ Rural Economic Performance Slows in the 1980s,” Pieces of Eight - An Economic Perspective on the Eighth D istrict, Federal Reserve Bank of St. Louis, (March 1989), pp. 5-8. Coughlin, Cletus C., and Thomas B. Mandelbaum. “ Why Have State Per Capita Incomes Diverged Recently?” this Review (September/October 1988), pp. 24-36. Mandelbaum, Thomas B. “ Gross State Product Series Pro vides New Perspectives on Regional Economies,” Business: An Eighth D istrict Perspective, Federal Reserve Bank of St. Louis (Winter 1988/89). U.S. Department of Labor. Employment and Earnings, (January 1989). 33 R. W. Hafer and Scott E. Hein R. W. Hafer is a research officer at the Federal Reserve Bank of St. Louis. Scott E. Hein is the First National Bank at Lub bock Distinguished Scholar, Departm ent of Finance, Texas Tech University. Kevin L. Kliesen provided research assistance. Com paring Futures and Survey Forecasts of NearTerm Treasury Bill Rates p JL REVIOUS research indicates that Treasury bill futures rates are better predictors of the future Treasury bill rate than forward rates. In a recent paper, MacDonald and Hein (1989) analyze 44 separate contracts delivered during the period 1977-87 for forecast horizons rang ing from two days ahead to 91 days ahead. Their evidence shows that the Treasury bill futures rate generally delivers a smaller forecast error of the three-month Treasury bill rate than the forward rate implicit in the spot market, and that the forward rate adds little informa tion about future Treasury bill rates that is not already incorporated into the futures rate. There also is evidence from other studies that survey forecasts of future Treasury bill rates contain information that improve upon forward rate forecasts. Studies by Friedman (1979) and Throop (1981), for example, reveal that survey forecasts often are more accurate than the forecasts from implicit forward rates. Given the results of this research, a natural question to ask is "Does the Treasury bill fu tures rate provide a better forecast o f future short-term interest rates than do survey fore casts?” In addition, since theories of financial market efficiency suggest that financial asset prices should include all available information, a related question is “Could one improve upon the Treasury bill futures forecasts using the infor mation contained in the survey projections?” Addressing these questions, the object o f this paper, is interesting for several reasons. One is that forecasts o f future interest rates are a crucial factor in forming investment strategies or purchasing plans. Incorrect interest rate forecasts can have large effects on investors’ wealth. Moreover, to the extent that interest rate risk is directly related to the level of in terest rates, accurately predicting the future level of rates is an important avenue to reduc ing interest rate risk exposure.1 In a related vein, policymakers often consider the effect on interest rates as an important factor in predic ting the outcome of policy changes. Knowing that the futures market provides an accurate gauge of the market’s expectation for future rates provides a practical benchmark prediction TQn this, see Belongia and Santoni (1987). MAY/JUNE 1989 34 to which policymakers can compare their forecasts.2 This article compares futures market and survey forecasts o f short-term Treasury bill rates in two ways. First, considering general ac curacy, we compare forecasting results of the two predictions over the 10-year period, 1977-87. General forecast accuracy is compared along with the extent of bias in the two reported forecasts.3 Second, we investigate whether information in the survey forecast could reduce the forecast error of the Treasury bill futures market prediction. This relates to the efficiency of the Treasury bill futures market, an issue that previously has been ad dressed by comparing futures and forward rates in terms o f the arbitrage opportunities that differentials in these two rates indicate.4 THE DATA This study uses two quarterly interest rate forecasts: one from a widely circulated survey of market participants; the other from the Treasury bill futures market. Survey Forecasts The survey forecasts are published in the Bond and Money Market Letter.5 This survey has been taken quarterly since 1969. On each survey date, approximately 40 to 50 financial market 2As Poole (1978) notes, “ Unless policymakers have solid evidence that their own forecasts are more accurate than market forecasts, they cannot afford to ignore the T-bill futures market.” (p. 18) 3Belongia (1987) also compares the relative accuracy of futures and survey forecasts of Treasury bill rates, using the semiannual survey published by the Wall Street Journal. 4For examples of such studies, see Hegde and Branch (1985) or MacDonald, et al (1988) and the references cited therein. 5We would like to thank the publishers of the Letter for allowing us to use their survey results in this study. For previous analyses of this survey data, often referred to as the Goldsmith-Nagan survey, see Prell (1973), Friedman (1980), Throop (1981) and Dua (1988). 6The survey actually asks for forecasts of 11 different in terest rates, ranging from the federal funds rate to conven tional mortgage rates. H'he newsletter in which the survey results are published also provides the interest rates on the day the question naires are mailed and the latest close before publication, a period of about two weeks. 8One such survey is conducted by the American Statistical Association-National Bureau of Economic Research (ASANBER). This quarterly survey also asks participants to FEDERAL RESERVE BANK OF ST. LOUIS analysts representing a variety o f financial insti tutions are asked for their point forecast o f a number o f different interest rates, three months and six months hence.6 In this study, we focus on the survey forecasts o f the three-month Treasury bill rate. The respondents’ forecasts are compiled, and the mean value is published in the Letter. Since the approximate date of the survey response is easily identified, these forecasts can be easily matched with futures market rates for similar dates.7 This feature makes the survey more attractive than other ex isting surveys for empirical comparison with in terest rate forecasts from the futures market.8 Futures Market Rates Trading in Treasury bill futures contracts takes place on the International Monetary Mar ket (IMM) of the Chicago Mercantile Exchange between the hours o f 8 a.m. and 2 p.m.9 The futures contracts traded call for delivery o f $1 million of Treasury bills maturing 90 days from the delivery day of the futures contract. The in strument and maturity of the deliverable instru ment match well with the survey forecasts of the Treasury bill rate. These contracts call for delivery four times a year: March, June, September and December.1 0 The futures market forecasts were gathered so that the futures market rate was taken on the same approximate date that the survey forecast the Treasury bill rate one quarter and two quarters ahead. Unfortunately, the questionnaire does not ask respondents for a forecast of the rate on any certain date in the future. It is unclear, therefore, whether the resulting forecast is a quarterly average, the peak rate for the quarter or the rate expected to hold at quarter’s end. Another interest rate survey already referred to is the se miannual Wall Street Journal poll of financial market analysts. This survey asks participants for their forecast of the three-month Treasury bill rate six months hence. Because this survey has been conducted only since December 1981, the limited number of forecasts restricts its usefulness for the type of empirical analysis used in this study. 9The discussion of the futures contract is based on informa tion available in the 1983 Yearbook of the IMM and the 1987 Yearbook, volume 2, of the Chicago Mercantile Exchange. 10The volume of futures contracts traded on the IMM grew substantially from their introduction in January 1976, when the total volume for all delivery months was 3,576 con tracts, through August 1982, when the number of con tracts traded reached 738,394. Since 1982, however, the number of contracts traded has decreased: in December 1987, the total number of contracts was 131,575. The decline in the Treasury bill contracts also coincides with the introduction of a Eurodollar futures contract. This new contract may be viewed as a substitute for the Treasury bill contract. 35 Figure 1 T-Bill, Futures and Survey Forecasts Forecast Horizon: Three Months Percent Percent forecast was made. It is the approximate date, because the exact date when each survey respon dent made his or her forecast cannot be deter mined. For example, the questionnaire asking ‘‘At what level do you see the following rates on September 30, 1987, and December 31, 1987?” was mailed to survey participants on June 16, 1987. The results of this survey subsequently were published on July 2, 1987. directly compared with the three-month and sixmonth-ahead Treasury bill rate survey forecasts published on July 2, 1987. For example, the July 1987 survey forecase of the September 30, 1987, Treasury bill rate was 5.81 percent. The futures market forecast was slightly higher, 6.15 percent. The actual rate turned out to be 6.64 percent. To make the analysis in this study tractable, we have chosen the midpoint of this two-week interval between the mailing date and publica tion date as the representative forecase date. Continuing with the example, two Treasury bill futures contracts were gathered from the Wall Street Journal for June 24, 1987: those for the September and December 1987 delivery dates.1 1 These futures market predictions are then A PRELIMINARY LOOK AT THE FORECASTS " I t also should be noted that a slight disparity between the date of the two forecasts is expected to prevail. The survey participants presumably are projecting rates for the last business day of each quarter. Alternatively, the futures To illustrate the overall relationship between the different series over the full sample period, we plotted the actual three-month Treasury bill rate and the different forecasts for the full sam ple period, from March 1977 through October 1987. These are shown in figures 1 and 2. market is concerned with rates on the delivery day of the futures contract, usually the third Thursday of the final month in each quarter. The maximum disparity, however, is only six business days. MAY/JUNE 1989 36 Figure 2 T-Bill, Futures and Survey Forecasts Forecast Horizon: Six Months Percent Three-Month-Ahead Forecasts Figure 1 presents the two different threemonth-ahead forecasts along with the actual three-month Treasury bill rate. The general pat tern shown is similar for both forecasts. In fact, both appear to have a closer relationship to each other than they do to the actual Treasury bill rates. For example, both forecasts overpredicted the actual rate in 1980.1 The forecast 2 error (actual minus predicted) for June 1980 from the futures market was -630 basis points; for the survey it was -642 basis points. Another relatively large forecasting error occur red when the actual rate fell sharply in late 1982. For September 1982, the futures market 12The Special Credit Control program was administered dur ing this period. For a description of the program and a discussion of monetary policy during this period, see Gilbert and Trebing (1981). 13These subperiods represent those during which monetary policy was thought to be influenced by the behavior of the http://fraser.stlouisfed.org/ FEDERAL RESERVE Federal Reserve Bank of St. Louis BANK OF ST. LOUIS Percent forecast error is -571 basis points compared with the survey forecast error of -487 basis points. Since 1984, although the differences have become smaller, the forecast errors from the futures market and the survey have tended to systematically overpredict rates. To provide some statistical basis for assessing the accuracy of these two forecasts, table 1 presents summary measures o f the relative ac curacy of the two three-month Treasury bill forecasts over the full period and two subper iods.1 Both the mean absolute error (MAE) and 3 the root mean squared error (RMSE) are reported for the forecasts. As a benchmark, we also report the results based on a simple no- monetary aggregates (1980-82) and the behavior of in terest rates (1983-87). Gilbert (1985) and Thornton (1988) suggest that the behavior of policy under borrowed reserve targeting was quite similar to that under a federal funds rate targeting procedure. 37 Table 1 Summary Forecast Statistics, Three-Month Treasury Bill Rate March 1977O ctober 1987 Forecast MAE RMSE March 1980Decem ber 1982 MAE RMSE March 1983O ctober 1987 MAE RMSE Three-m onth forecasts Futures 1.18% 1.90% 2.79% 3.32% 0.53% 0.86% Survey 1.25 1.97 2.92 3.44 0.62 0.93 Naive 1.20 1.91 3.10 2.53 0.54 0.79 Six-m onth forecasts Futures 1.52 2.28 2.94 3.64 1.03 1.48 Survey 1.60 2.23 3.10 3.63 1.03 1.30 Naive 1.68 2.31 3.12 3.63 0.83 1.01 NOTE: MAE is the mean absolute error; RMSE is the root mean squared error. change forecast model, where the no-change model’s forecast is the spot three-month Treasury bill rate observed on the same day that the futures rate forecast also is gathered. The overall forecast accuracy of the threemonth-ahead futures and survey predictions are quite close. For the full period, the MAE is 1.18 percent for the survey and 1.25 percent for the futures rate, both about the same as the no change forecast (1.20 percent). The RMSEs also are quite similar across forecasts. The subperiod results reflect the difficulty in forecasting the Treasury bill rate during the early 1980s: the MAEs for the different forecasts are, on average, five times greater during the 1980-82 period than the 1983-87 period. Still, the forecast statistics in dicate that the relative accuracy of the forecasts is similar.1 4 during 1980 in figure 2 contrasted with figure 1. The prediction error for December 1980 from the futures rate was -704 basis points and, for the survey forecast, -744 basis points. For the three-month-ahead forecasts, the respec tive errors were positive and smaller: 239 basis points for the futures market forecast and 409 basis points for the survey forecast. Note also the magnitude of the post-1984 overprediction in figure 2 relative to figure 1. The summary statistics in table 1 reveal that the accuracy of the six-month-ahead futures and survey forecasts is comparable for the full period and the subperiods. Generally, there is little difference between the MAEs and RMSEs for the two forecast series. Bias Tests Figure 2 is a plot of the six-month-ahead forecasts together with the actual Treasury bill rate. The size and pattern of the two six-monthahead forecast errors contrasts sharply with the three-month-ahead forecasts. Note, for example, the relative magnitude of the forecast errors Observers generally argue that rational indi viduals do not make the same forecasting mis take over and over again, because forecasts that consistently over- or underpredict the actual series presumably reduce the investor's wealth relative to forecasts that are unbiased. Consis tent with the notion o f wealth-maximization and rationality, forecasts therefore should be unbiased. 14This observation is corroborated by a statistical test of the futures and survey forecasts’ mean square errors (MSE). This test, suggested by Ashley, Granger and Schmalensee (1980), revealed that one could not reject the hypothesis that the futures market and survey forecasts’ MSEs are equal. Six-Month-Ahead Forecasts MAY/JUNE 1989 38 To test forecasts for bias, researchers usually estimate a regression of the form forecast error (ut) has a unit root. Moreover, it should be the case that E(u,) = 0. (1) r, = a + Pt_s E + u, , rt To implement this test procedure, we first test for unit roots in the actual and forecast in terest rate series. Again, if it is shown that the actual interest rate series has a unit root, then so should the forecast series under the assump tion of rational expectations.1 To test for unit 9 roots, the Dickey-Fuller (1979) test procedure is used wherein the change in each series is re gressed on a constant and one lagged value of the serie’s level. Specifically, a regression o f the form where r, is the actual rate of interest at time t, ,_ ,r, is the expectation of the rate for time t held at time t-s, and u, is a random error term.1 The null hypothesis, that expectations 5 are unbiased, implies the testable hypothesis that the estimated values of the coefficient a is zero and the coefficient ft is unity. Moreover, the error term (ut) should not display characteristics of autocorrelation.1 6 A problem in estimating equation 1 arises if the actual and forecast series are characterized by unit root processes.1 In such a case, 7 estimating equation 1 will produce downwardbiased coefficient estimates, an increased pro bability of rejecting the null hypothesis and, therefore, an incorrect finding of bias when it doesn’t exist.1 8 As an alternative to estimating equation 1 directly, one can test for bias by imposing the null hypothesis conditions and determine whether the data reject them. Imposing the null restrictions yields the relationship (2) r, - ,_,r® = ut . If the actual interest rate series and the forecasts are characterized by unit root pro cesses and the forecasts are unbiased, then the data also should reject the hypothesis that the 15Webb (1987) has argued that such tests may lead one to reject the null hypothesis when it is true. He argues that rejection of unbiasedness may reflect several factors, all of which are known to the econometrician ex post but not to the forecaster ex ante. He argues that forecasts that fail bias tests may in fact have originally been formulated op timally. This criticism is most forceful for examining forecasts of series that are revised many times following the original forecast date. Such a problem does not exist, however, with the interest rate series used here. 16This restriction, as Friedman (1980) notes, strictly applies only to the one-step-ahead forecasts. 17lf the fundamental moving-average representation of some series X has an autoregressive representation, then it can be written in the form [1-a(L)] X, = e, , where L is the lag operator (i.e., LX, = Xt_, and a(L) = Xa,L‘. The polynomial in the lag operator a(L) can be written as a(L) = (l-E^LJBfL). If there exists a root B i that is equal to unity, then the series X is characterized by a unit root. It is useful to note that a random walk is a par ticular type of unit-root process. 18We would like to thank Jerry Dwyer for pointing this out. This issue is discussed at length in Dwyer, et al (1989) from which the following draws. http://fraser.stlouisfed.org/ Federal Reserve Bank of RESERVE BANK OF ST. LOUIS FEDERAL St. Louis (3) AX, = a0 + AXt_, + e, is estimated, where A is the difference operator (i.e., AXt = X, - Xt_,). If the t-ratio associated with the lagged variable is less than the relevant critical value, then we can reject the existence of a unit root. The results of this test for the Treasury bill rate and its forecasts are reported in the upper half of table 2. In every instance, we find that the estimated t-ratio on the lagged level of the selected variable is greater than the 5 percent critical value, about -3.50.2 This evidence in 0 dicates that we cannot reject the notion that each series has a unit root. Given this finding, the forecast errors are ex amined to determine whether they do not con tain unit roots, as hypothesized under the con19ln other words, the process generating the expectations should be the same as the one generating the actual series. 20The critical value is taken from Fuller (1976), table 8.5.2. We should note that Schmidt (1988), extending the work of Nankervis and Savin (1985), argues that these critical values are incorrect in the presence of significant drift in the variable. Given the estimated constant terms found in the upper panel of table 2, the critical value to test for unit roots according to Schmidt is about -1.86 at the 5 percent level and about -2.60 at the 1 percent level. Using these critical values, our estimates suggest that, while unit roots are rejected at the 5 percent level, they are not at the 1 percent level. If we take the results using the 5 percent level, then it is possible to estimate equation 1 directly. Doing so gives the following results: the calculated F-statistic and related marginal significance level testing the joint hypothesis that a = 0 and p = 1 in equation 1 is 2.51 (0.09) for the threemonth futures forecast; 3.26 (0.05) for the six-month futures forecast; 1.66 (0.20) for the three-month survey forecast; and 1.80 (0.18) for the six-month survey forecast. Except for the six-month futures forecast, these results in dicate that unbiasedness cannot be rejected. 39 Table 2 Unit Root Tests on Actual Treasury Bill and Forecasts Sample: March 1977-October 1987 Estim ated co e ffic ie n t1 Constant Lagged level Treasury bill 2.17 (2.34) -0 .2 3 9 (-2 .4 0 ) Futures (3-month) 2.40 (2.40) - 0.263 (-2 .4 8 ) Survey (3-month) 2.84 (2.70) -0.32 1 (-2 .7 9 ) Futures (6-month) 2.21 (2.26) -0 .2 4 0 (-2 .3 4 ) Survey (6-month) 2.07 (2.26) -0 .2 3 4 (-2 .3 2 ) -0 .1 1 2 (-0 .3 7 ) -1 .2 0 3 (-7 .6 6 ) 0.201 (0.64) -1 .2 1 4 (-7 .7 6 ) -0 .1 2 2 (-0 .3 5 ) -0 .6 9 8 (-4 .5 5 ) 0.128 (0.37) -0 .7 0 2 (-4 .5 9 ) A ctual series Forecast errors Futures (3-month) Survey (3-month) Futures (6-month) Survey (6-month) always less than the critical value. These results indicate that the imposed restrictions associated with unbiased forecasts are not rejected. The different forecast error series also are ex amined to decide whether their mean values differ from zero. In every instance, the hypothesis that the mean forecast error is not statistically different from zero could not be re jected. In fact, the largest t-statistic calculated is far below unity. Thus, the evidence is largely consistent with the notion that the futures market and survey forecast errors are unbias ed.2 2 MARKET EFFICIENCY TESTS The evidence to this point tells us little about the efficiency of the Treasury bill futures mar ket. The hypothesis of market efficiency asserts that financial markets use all available informa tion in pricing securities. If this is true, there should be no more accurate forecast of future security prices than that in today’s price. To investigate the efficiency o f the futures market forecasts, a test proposed by Throop (1981) is used to determine whether knowledge o f the survey forecast o f Treasury bill rates could reduce the forecast error made by the futures market. The answer to this question can be found by estimating the regression (4) rt - t_,r* = c5(,_s t—t_sr^) + e „ t , rs 'Values of t-ratios are reported in parentheses. The 5 percent critical value taken from Fuller (1976) is about -3 .5 0 . dition of unbiasedness.2 Regressing the change 1 in the respective forecast error on a constant and a lagged level o f the forecast error pro duces the results reported in the lower half of table 2. For both the three-month and six-month forecasts, the futures market and survey fore casts o f the Treasury bill rate satisfy the condi tion of unbiasedness: the calculated t-ratio is 21As Dwyer, et al (1989) stale, “ A unit root in the forecast errors would indicate that the distribution of the forecast errors has a random walk component which has no counterpart in the innovations in the events being forecast.” (p. 15) 22The bias of the no-change forecasts also was tested. Like the results based on the futures market and survey forecasts, the reported t-ratios allow us to reject the hypothesis of a unit root in the forecast errors of the no change models. Moreover, the mean forecast error is not statistically different from zero. where rt is the three-month Treasury bill rate at date t, ,_,r^ is the futures market rate at t-s for delivery at t, t_,rf is the survey forecast taken at t-s for rates prevailing at t and eS is a )t random error term.2 The hypothesis of market 3 efficiency requires that the estimated value of the coefficient < is zero, indicating that the in 5 formation in the survey forecast already is in corporated in the futures market’s projection. To see this, rewrite equation 4 as rt = c5,_,r® + (1 —c5),_s Under the market efficiency require r^ ment that d = 0, the survey forecast drops 23Throop (1981) used this approach to test the efficiency of Treasury bill forward rate projections and found evidence of inefficiencies in the forward market. Kamara and Lawrence (1986) and MacDonald and Hein (1989) use this approach and find that Treasury bill futures rates are more accurate forecasts when compared with the forward rates. Other examples employing a similar type of analysis are Fama (1984a,b) and French (1986). MAY/JUNE 1989 40 Table 3 Efficiency Test Regressions Sample: March 1977-October 1987 Estimated Equations: A) r, ~ B)r, - ,.,rf = d2 ( Estim ated coe fficie n ts Equation 6 R2 1 DW Three-m onth forecasts A -0 .0 2 0.08 (0.16) 0.44 (0.95) B 2.41 0.02 2.50 Six-m onth forecasts A 0.05 0.71 (1.54) B 0.44 (1.36) 1.40 0.04 1.58 NOTE: Absolute value of t-statistics in parentheses. month-ahead Treasury bill futures market forecasts are reported as equation A in table 3.2 The evidence indicates that the hypothesis 5 of a semi-strong form of market efficiency can not be rejected at the 5 percent level of significance. Using the information differential between the survey forecast and the futures rate, the estimated value o f 6 is only 0.08 (t = 0.16) for the three-month forecast horizon. For the six-month horizon, the estimated value of 6 is 0.71 (t = 1.54). In both instances, we cannot reject the hypothesis of efficiency as applied to the futures market forecast. A weak-form market efficiency test also was considered by replacing the survey forecast with the current spot market rate. The result is reported as equation B in table 3. When compared with the no-change forecast, efficiency again cannot be rejected for the futures rate: the results indicate that, for the three-month and six-month forecasts, the estimated value of 6 is never significantly different from zero. Rewriting equation 4 as above also indicates that it imposes the restriction that the sum of the weights on the two forecasts sum to unity. W e have re-estimated the equation without this restraint and found that we still could not reject efficiency of the futures rate forecasts when compared with either the survey or no-change forecasts. from the equation and one is left with r, = t-srf + e.,r If the estimated value of 6 is different from zero, however, knowledge of the differential between the survey forecast and the futures rate would significantly reduce the forecast er ror in the futures rate.2 This would be incon 4 sistent with the notion that market participants efficiently utilize all available information. In the terminology of Fama (1970), our test is a “ semistrong” form test of market efficiency, since all the information in the survey projections would not have been publicly available when the futures market was sampled. Estimates of equation 4 to test the efficiency of both the three-month-ahead and the six- 24This same procedure can be used to test if there is infor mation in the futures rate that is not present in the survey forecast. In this case, the left-hand side of equation 4 is the forecast error from the survey prediction. The results from this test (not reported) indicate that the survey forecasts are efficient with respect to the futures market forecasts. http://fraser.stlouisfed.org/ Federal Reserve Bank of RESERVE BANK OF ST. LOUIS FEDERAL St. Louis The R o le o f the R evision in the Survey Forecast Since the survey participants are asked for their three- and six-month-ahead forecasts every three months, they essentially are providing two forecasts of the same event, taken at two dif ferent points in time. For example, survey par ticipants are asked in December of the previous year and then again in March to forecast the June Treasury bill rate. One piece o f new infor mation that survey respondents have in making their March forecasts is the revision o f the December forecast itself. Nordhaus (1987) has suggested that, for forecasts to be efficient, the information in the revision also should be incor- 25The results reported are those excluding a constant term in the regression. Including a constant term does not alter the conclusions reached. Also, W hite’s (1980) test failed to reject the null hypothesis of homoskedasticity in the residuals. 41 porated in the current forecast. Knowledge of the revision should not allow a reduction in the forecast error under the hypothesis of efficiency. A similar argument can be applied to the futures rate forecasts. In particular, knowledge of the revision in the survey forecast of future Treasury bill rates should not help reduce the futures market’s forecast error if the latter is formed efficiently. The survey’s revision is part of today’s information set and should already be incorporated into the market’s projection.2 To 6 test whether knowledge of the survey’s revision could help reduce the forecast error in the futures market, equation 4 is modified to in clude the survey revision itself: (5) rt- t_,r* = a0 + y , (t_ , r f - ,_,rf) + y( ,-.rf - .-2 + e, • 2 rf) The term ( t. ,r s - t_2 f ) reflects the revision in t r the survey’s forecast o f next quarter’s Treasury bill rate. Efficiency requires not only that the futures rate contains all the information in the survey forecast, but also that it reflects the survey forecast revision. If the futures rate forecast is efficient, estimated values o f both y, and y 2 in equation 5 should not be different from zero. The results from estimating equation 5 (with absolute value of t-statistics in parentheses) are:2 7 (6) r, - t_ ,rf = -0.066 + 0.104( t_,rf— t_,rf) (0.20) (0.18) -0.312(t_ , r f - t_2 rf) ( 1 . 86 ) R2 = 0.034 DW = 1.92 The intercept of the equation is not statistical ly different from zero, indicating no bias in these projections. W e also find that the esti mated slope coefficients (y, and y 2) are not 26The reader again is reminded that this is a semi-strong form efficiency since the information in the survey revision would not have been released to the public at the time that we sampled the futures rates. 27W hite’s (1980) test indicated that we could not reject the null of homoskedastic residuals. 2eWe should note, however, that the y2 slope coefficient is significant at about the 7 percent level. Based on this level of significance, the result of estimating equation 5 is con sistent with the notion that the futures rate forecasts may significantly different from zero using a conven tional 5 percent level of significance. This out come is consistent with the efficient markets hypothesis that there is little information in the survey forecast or its revision that is not already incorporated into the futures rate forecast.2 8 CONCLUSION In this study, we compared futures market and survey forecasts of the three-month Trea sury bill rate both three and six months ahead. Our test results generally support the percep tion that the forecasts are unbiased predictors of future rates. Moreover, futures market fore casts of near-term interest rates usually are as accurate as those produced by professional fore casters. Compared with a popular survey of professionals used in this study, we find little difference in the relative forecasting accuracy o f the two. Our results also indicate that no in formation in the survey forecast or its revision could reliably improve upon the futures rate prediction. This conclusion about market efficiency con trasts sharply with that found for the forward market. Previous evidence has shown that the Treasury bill forward rate does not incorporate all o f the information contained in the same survey considered here. Such a conclusion, along with the evidence presented in this paper, is consistent with the belief that there is a timevarying premium in the forward rate that ap parently is absent in the Treasury bill futures rate. The results presented here should not be in terpreted as proof that the Treasury bill futures market rate is always the most accurate interest rate forecast. The evidence does suggest, how ever, that for investment decisions and mone tary policy discussions, the futures rate provides a useful measure o f the market’s expectation of future interest rates. Consequently, it is a valuable benchmark to which other forecasts can be compared. not be the optimal projection of the Treasury bill rate. Given the results in equation 6, the optimal forecast ( ,_,r") would take the form ,-,rf = ,rf— 0.312( ,_,rf— ,_*rf) • This result implies an overreaction on the part of the futures market to a revision. That is, if the survey revises its interest rate forecast upward, the optimal forecast would scale down the forecast from the futures market. MAY/JUNE 1989 42 REFERENCES Ashley, Richard, Clive W. J. Granger, and Richard Schmalensee. “Advertising and Aggregate Consumption: An Analysis of Causality,” Econometrica (July 1980), pp. 1149-67. Belongia, Michael T. “ Predicting Interest Rates: A Compari son of Professional and Market-Based Forecasts,” this Review (March 1987), pp. 9-15. Belongia, Michael T., and Gary J. Santoni. “ Interest Rate Risk, Market Value, and Hedging Financial Portfolios,” Journal of Financial Research (Spring 1987), pp. 47-55. Chicago Mercantile Exchange. 1987 Yearbook (1988). Dickey, David A., and Wayne A. Fuller. “ Distribution of the Estimators for Autoregressive Time Series with a Unit Root,” Journal o f the Am erican S tatistical Association (June 1979), pp. 427-31. Dua, Pami. “ Multiperiod Forecasts of Interest Rates,” Jour nal of Business and Economic Statistics (July 1988), pp. 381-84. Gilbert, R. Alton, and Michael E. Trebing. “ The FOMC in 1980: A Year of Reserve Targeting,” this Review (August/September 1981), pp. 2-22. Hegde, S., and B. Branch. “ An Empirical Analysis of Ar bitrage Opportunities in the Treasury Bills Futures Market,” Journal o f Futures Markets (Fall 1985), pp. 407-24. International Monetary Market. 1983 Yearbook (Chicago Mer cantile Exchange: 1984). Kamara, A., and C. Lawrence. “ The Information Content of the Treasury Bill Futures Market Under Changing Monetary Regimes,” Working Paper No. 51 (Center for the Study of Banking and Financial Markets, University of Washington, 1986). MacDonald, S. Scott, and Scott E. Hein. “ Futures Rates and Forward Rates as Predictors of Near-Term Treasury Bill Rates,” Journal of Futures Markets (June, 1989) pp. 249-62. MacDonald, S. Scott, Richard L. Peterson, and Timothy W. Koch. “ Using Futures to Improve Treasury Bill Portfolio Performance,” Journal o f Futures M arkets (April 1988), pp. 167-84. Nankervis, J. C., and N. E. Savin. “ Testing the Autoregres sive Parameter with the t Statistic,” Journal o f Economet rics (February 1985), pp. 143-61. Dwyer, Gerald P., Jr., Arlington W. Williams, Raymond C. Battalio, and Timothy I. Mason. “ Tests of Rational Expecta tions in a Stark Setting,” Federal Reserve Bank of St. Louis Research Paper 89-001 (April 1989). Nordhaus, William D. “ Forecasting Efficiency: Concepts and Applications,” Review o f Economics and S tatistics (November 1987), pp. 667-74. Fama, Eugene F. “ Forward and Spot Exchange Rates,” Journal of Monetary Economics (November 1984a), pp. 319-38. Poole, William. “ Using T-bill Futures to Gauge Interest-Rate Expectations,” Federal Reserve Bank of San Francisco Economic Review (Spring 1978), pp. 7-19. _______ . “ The Information in the Term Structure,” Journal o f Financial Economics (December 1984b), pp. 509-28. Prell, Michael J. “ How Well do the Experts Forecast Interest Rates?” Federal Reserve Bank of Kansas City M onthly Review (September-October 1973), pp. 3-13. ________ “ Efficient Capital Markets: A Review of Theory and Empirical Work,” Journal of Finance (May 1970), pp. 383-417. French, Kenneth R. “ Detecting Spot Price Forecasts in Futures Prices,” Journal of Business (April 1986), pp. 539-54. Friedman, Benjamin M. “ Survey Evidence on the ‘Ratio nality’ of Interest Rate Expectations,” Journal o f M onetary Economics (October 1980), pp. 453-65. ________ “ Interest Rate Expectations Versus Forward Rates: Evidence from an Expectations Survey,” Journal of Finance (September 1979), pp. 965-73. Fuller, Wayne A. Introduction to S tatistical Time Series (John Wiley & Sons, Inc., 1976). Gilbert, R. Alton. “ Operating Procedures for Conducting Monetary Policy,” this Review (February 1985), pp. 13-21. FEDERAL RESERVE BANK OF ST. LOUIS Schmidt, Peter. “ Dickey-Fuller Tests with Drift,” unpublished manuscript, Michigan State University (June 1988). Thornton, Daniel L. “ The Borrowed-Reserves Operating Pro cedure: Theory and Evidence,” this Review (January/ February 1988), pp. 30-54. Throop, Adrian W. “ Interest Rate Forecasts and Market Effic iency,” Federal Reserve Bank of San Francisco Economic Review (Spring 1981), pp. 29-43. Webb, Roy H. “ The Irrelevance of Tests for Bias in Series of Macroeconomic Forecasts,” Federal Reserve Bank of Richmond Economic Review (November/December 1987), pp. 3-9. White, Halbert. "A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity,” Econom etrica (May 1980), pp. 817-38. 43 Gerald P. Dwyer > Jr. and R. Alton Gilbert Gerald P. Dwyer, Jr., a professor o f econom ics at Ciemson University, is a visiting scholar and R. Alton G ilbert is an assistant vice president a t the Federal Reserve Bank o f St. Louis. Erik A. Hess and Kevin L. Kliesen provided research assistance. Bank Runs and Private Remedies iURRENT banking regulatinn in the United States is based in part on the notion that both the banking system and the economy must be protected from the adverse effects of bank runs. An example often cited as typical is the string of bank runs from 1930 to 1933, which conventional wisdom holds responsible for thousands of bank failures and the Banking Holiday o f 1933 when all banks closed. The runs on savings associations in Ohio and Mary land in 1985 are more recent examples. This conventional view is reflected in a recent comment on the “Panic of 1907” in the Wall Street Journal (1989): Long lines o f depositors outside the closed doors o f their banks signaled yet another financial crisis, an all-too fam iliar event around the turn o f the century. Research in the last few years on bank runs indicates that the conventional view is mistaken. Runs on the banking system were not common place events, and their impact on depositors and the economy easily can be overstated. Prior to the formation o f the Federal Reserve System in 1914, banks responded to runs in ways that 'Salant (1983) provides a general analysis of the breakdown of such arrangements as bank redemption of its lessened their impact. These private remedies did not solve the problem of runs, but they did mitigate the effects of the runs on the banks and the economy. In this article, we explain the private remedies for runs and provide some evidence on the frequency and severity of runs on the banking system. BANK RUNS: THE THEORY Before examining the history of bank runs, it is useful to consider why banks are vulnerable to runs. This examination establishes a frame work for determining the kinds of observations that would be consistent with their occurrence. Runs on Individual Ranks In a run, depositors attempt to withdraw cur rency from a bank because they think the bank will not continue to honor its commitment to pay on demand a dollar of currency for a dollar of deposits.1 One aspect of the contract banks make with their customers is central to under standing why depositors would run on their bank. Banks make contractual promises that they cannot always honor: exchange of gold or liabilities at par. The mapping from speculative attacks into bank runs is discussed by Flood and Garber (1982). MAY/JUNE 1989 44 currency at par value for bank liabilities.2 When banks issued notes as a form of currency, the promise was a contractual agreement to deliver specie (gold or silver) in exchange for the bank’s notes at par value. Banks currently promise to deliver U.S. currency to depositors on demand at par value. Because banks hold reserves that are only a fraction o f their liabilities payable on demand, they cannot honor this promise if all of their depositors try to convert deposits into currency at the same time. Fractional-reserve banking by itself is not suf ficient to make it impossible for banks to honor their promises to deliver currency in exchange for deposits on demand. Banks always could honor a promise to pay currency at a variable exchange rate o f currency for deposits. If all depositors want to exchange their deposits for currency at the same time, banks do not have sufficient currency (or other reserves that can be transformed into currency on a dollar-fordollar basis instantaneously) to meet that de mand for currency at a price of one dollar of currency for one dollar o f deposits.3 The effects o f a run by depositors on one bank can be illustrated by an example. Table 1 shows the balance sheet of a hypothetical na tional bank in New York City in the national banking period (1863 to 1914). Its liabilities in clude deposits and national bank notes backed by securities deposited with the Treasury. In the event of the bank’s failure, the notes were guaranteed by the U.S. Treasury, whether or not the deposited bonds were sufficient backing for the notes. Apparently as a result o f this guarantee, runs on banks in the national bank ing era were runs by depositors, not by note holders.5 In the normal course o f affairs, the inability of all depositors to exchange their deposits for currency is irrelevant. As some depositors with draw currency from a bank, others deposit it. The low probability of every depositor closing his or her account at the same time is the reason a bank usually can operate with frac tional reserves and pay currency on a dollarfor-dollar basis. During this period, national banks in New York City were required to maintain reserves of specie and legal tender equal to 25 percent or more of deposits, with the required ratio of reserves to deposits lower for national banks in other cities. Banks generally held excess re serves as a buffer stock to meet deposit with drawals, but we use a reserve ratio o f 25 per cent to keep the numerical example simple. The second part of table 1 shows the initial loss of reserves upon withdrawal o f $2 million of deposits, while the last part indicates the reac tion of the bank to the decrease in deposits. An individual bank can replenish its reserves by selling assets; in the example, the bank returns its reserve ratio to 25 percent by selling $1.5 million in assets. At least part of the reserves are from other banks, thereby transmitting the reserve loss to other banks. A low probability is not the same as a zero probability though. Information or rumors which suggest a capital loss by a bank may in duce its depositors to attempt to convert their deposits to currency.4 The mere expectation that other depositors will attempt the same con version also can cause a run on a bank. A run on a single bank is unlikely, however, to have substantial effects on the economy. The primary effect of a single bank closing is that the bank winds up its affairs and no longer operates. In a run on a single bank, the specie and legal tender withdrawn from the bank are likely to be largely deposited in other banks. As a result, a run on a single bank is not likely to drain reserves from the banking system or increase currency held by the public. If the currency withdrawn is deposited in other banks, the net effect on the bank’s balance sheet is that shown in table 1, and the deposit and reserve loss at this bank is matched by a similar increase in deposits and reserves at other banks. 2Whether this promise is a result of market forces or government regulation is an open question. Davis (1910) summarizes the laws in the United States in the 19th cen tury, and Schweikart (1987) provides the historical devel opment of these laws in the South in the 19th century. 3Promises that cannot be kept in all states of the world are hardly unique to banking. For instance, firms often cannot make payments on debt if there is a large decrease in the demand for their products. The common legal word for failure to honor contractual commitments is “ default.” BANK OF ST. LOUIS FEDERAL RESERVE While default generally is not the expected outcome of a contract, it does happen. 4Among others, Diamond and Dybvig (1983) and Gorton (1985a) present models of runs. 5ln banking panics prior to the national banking era, customers of banks attempted to redeem their bank notes for specie. For details on the backing for notes in the na tional banking era, see Friedman and Schwartz (1963), pp. 20-23, 781-82. 45 Table 1 Balance Sheet of a National Bank in New York City with a Large Withdrawal of Deposits (millions of dollars)________________ INITIAL BALANCE SHEET Liabilities Assets Reserves (specie and legal tender) Interest-earning assets Total assets Deposits $ 2.5 Notes Net worth 11.0 Total liabilities $13.5 $10.0 1.0 2.5 $13.5 IMMEDIATELY AFTER WITHDRAWAL OF $2 MILLION BY DEPOSITORS Liabilities Assets Reserves Interest-earning assets Total assets Deposits Notes Net worth 11.0 $11.5 $ 8.0 1.0 2.5 Total liabilities $ 0.5 $11.5 AFTER RESTORATION OF RESERVE RATIO TO 25 PERCENT Liabilities Assets Reserves $ 2.0 Interest-earning assets Total assets 9.5 $11.5 Runs on the Banking System Runs on a single bank can develop into runs on the banking system.6 An important, if seem ingly obvious, aspect of banking is that the like lihood of a bank’s default on its deposit agree ment is not known with certainty by depositors. Instead, depositors estimate this likelihood as best they can with available information. One type of information that can be useful in estimating the value o f a bank's assets is infor mation on the value of assets at other banks. News about the failure of one bank can cause depositors at other banks to raise their estimate o f the probability that their bank will default. Contagious bank runs can be defined as runs which spread from one bank or group o f banks to other banks. A term sometimes used for a period of a run on the banking system is a "banking panic,” a Deposits Notes Net worth $ 8.0 1.0 2.5 Total liabilities $11.5 term that has a connotation of unreasoning fear or hysteria. Contagious runs, however, can be based on the optimal use of all information by all agents. As a simple example, suppose that two banks are identical in all respects known by depositors, and one o f the two fails because of loan losses. Because o f the first failure, de positors will increase their estimate o f the pro bability that the second bank will fail. If this estimate increases sufficiently, depositors will run on the second bank, even though no other information has appeared. This use o f informa tion is quite consistent with rational behavior. Depositors use the information available, and one part of that information is the condition of other banks. Simultaneous runs on many banks need not be contagious runs though. For example, an ex ogenous event can increase simultaneously de- 6Gorton (1985a) and Waldo (1985) provide models of aspects of the process which we discuss in this section. MAY/JUNE 1989 46 Table 2 Balance Sheet of the Banking System with a Large Withdrawal of Deposits (millions of dollars)___________________________ Liabilities Assets Reserves Interest-earning assets Total assets $ 250 1,100 $1,350 Deposits Notes Net worth $1,000 100 250 Total liabilities $1,350 After withdrawal of $200 million by depositors: Liabilities Assets Reserves Interest-earning assets Total assets $ 50 500 $ 550 positors’ estimated probability that many banks will fail to redeem at par. Myers (1931) suggests that bank runs in 1914 resulted from the public’s expectation that the war would result in a restriction of convertibility o f notes and deposits into specie.7 Whether a contagious or a simultaneous run, a run on the banking system is associated with a drain of reserves from the banking system. The effect of this withdrawal of reserves is shown in table 2, which illustrates the effect of a $200 million increase in the demand for cur rency. For each bank individually, the initial im pact is a withdrawal o f reserves. Banks no longer have a reserve ratio of 25 percent, and, as a result, they attempt to increase their holdings of reserves by selling assets. The sale of assets by one bank drains reserves from other banks though, and these banks then sell some of their assets to acquire reserves. Unlike the previous example, the $200 million of reserves is gone from the banking system. As 7See Myers (1931), p. 421. Empirically distinguishing be tween contagious runs and simultaneous runs is a tricky issue, which requires distinguishing between bank runs due to information that affects banks’ assets and those due to information about some banks’ assets. One way of BANK OF ST. LOUIS FEDERAL RESERVE Deposits Notes Net worth $ 200 100 250 Total liabilities $ 550 table 2 shows, the result of this process is a contraction o f deposits and assets that is a multiple of the initial decrease in reserves. If banks sell relatively large amounts of their assets quickly in a run, they can drive down the market value of their assets and drive up market interest rates. Table 2 could be modified to reflect this effect, with an additional decline in the value of bank assets and their net worth. If the declines in net worth are large enough, the response of the banks to the run indicated in table 2 will cause some banks to fail. Thus, an additional effect o f a bank run might be a rise in the rate o f bank failure. Observations Consistent with the Occurrence o f Runs The definition of a run is based on depositors' estimated probability of non-par redemption by banks. While it is possible to use an economic model to estimate this probability, we use a lessdemanding basis to examine data for evidence doing this is to define contagious bank runs as those that would not have occurred without runs on earlier banks. There is at least one successful attempt at providing detailed evidence of a contagious run: W icker’s (1980) analysis of the runs in November and December 1930. 47 of runs: we examine the data for consequences of runs.8 A leading example o f an event consistent with a run on the banking system is a joint restric tion of convertibility by banks. Without an of ficial central bank, banks can limit the effects of a run by jointly agreeing to restrict currency payments to depositors.9 The effects o f such a restriction can be illustrated by referring to table 2. Suppose that, after depositors withdraw $50 million in currency, the banks agree to stop making currency payments. In this illustration, deposits decline by only $200 million, to $800 million. The demand for more currency by depositors will not cause a further decline in deposits because some or all of that demand is refused by the banks. Hence, one observation that provides clear evidence of a run on a banking system is a restriction o f currency payments by banks in the system. An individual bank resorts to a restriction o f currency payments if it cannot meet its commitment to pay currency to deposi tors on demand. Banks will resort to this action jointly if they face a common problem o f cur rency withdrawals. If the restriction of payments results in signifi cant restrictions on depositors’ ability to trans form deposits into currency, a market for trans forming currency into deposits may develop. If there is such a market, there will be a premium for currency in terms of deposits.1 0 A bank run need not result in restriction though. The following developments also would be consistent with the occurrence of a run on a banking system, although they are not inevi table effects of runs and they can occur in the 8Gorton (1988) does estimate a particular model for runs and finds them generally consistent with our analysis. He also defines runs on the banking system, or in his terms “ banking panics,” as periods when convertibility was restricted in New York City, clearing house loan cer tificates were authorized by the New York Clearing House or both (1988, pp. 222-23). We prefer not to identify periods with runs based on a single criteria. If we were to pick a single criteria, it would be restriction of payments by banks. With any penalties on nonpar payments, banks will not do this unless they at least believe that they can not continue payments at par indefinitely. For the use of a multiple set of criteria along our lines, see Bordo (1986). 9The names “ restriction of cash payments" or “ restriction of convertibility of deposits into currency” are suggested by Friedman and Schwartz (1963, p. 110, fn. 32) rather than the traditional name of “ suspension of currency payments.” Following this suggestion avoids confusion of “ suspension of currency payments” with “ suspension of operations” and is more consistent with the fact that absence of a run. Perhaps most importantly for our purposes, these indicators of runs can be lessened by a restriction o f payments to deposi tors. They are: 1. a decline in the ratio of reserves to deposits. 2. a rise in the ratio of currency to deposits. 3. for a given monetary base, a decline in the money supply (because the decline in de posits is a multiple of the decline in bank reserves). RESTRICTION OF CONVERTIRILITY The view that the banking system is vulner able to runs may be based primarily on the ex perience of the early 1930s, but the most rele vant period to examine for evidence o f runs is before the operation o f the Federal Reserve System. Prior to late 1914, the United States had no official central bank.1 W e focus on the 1 banking system beginning with the 1850s. While events in earlier years also are o f interest, 1853 marks the beginning of a weekly data set on reserves and deposits in banks in New York Ci ty which is very useful. In addition, by the 1850s, New York City was the most important financial center in the United States. Many banks in other parts of the country held cor respondent balances in New York City banks, and pressures affecting banks in the rest o f the country affected New York City banks through these balances.1 2 Restrictions on Paym ents As table 3 indicates, banks in New York City restricted payments on five occasions banks commonly did not completely stop converting deposits into currency. Currency payments were non-price rationed, not suspended. Evidence for the post-Civil-War period that payments generally were restricted, not suspended, is presented by Sprague (1910), pp. 63-65, 121-24, 171-78, 286-90, and Andrew (1908), pp. 501-02. A more general and precise, but also quite pedantic, name for restrictions would be “ restriction of convertibility at par of bank liabilities with promised par redemption on de mand.” ,0As we show below, banks remained open for deposits. Hence, a discount on currency could not persist. "F riedm an and Schwartz (1963) and, in more detail, Timberlake (1978) discuss the central banking activities by the Treasury in the national banking period. As argued forcefully by Dewald (1972), the New York Clearing House acted as a central bank at times. 12See Myers (1931) and Sprague (1910). MAY/JUNE 1989 48 Table 3 Dates of General Restriction of Payments in New York City, 1857 to 1933 Year Beginning date Ending date 1857 1861 1873 1893 1907 1933 October 13 December 28 September 24 August 3 October 26 March 3 December 11 April 1862 October 22 September 2 December 28 March 15 Sources: see data appendix available on request. between the 1850s and 1914.1 In the episodes 3 from 1857 to 1907, banks across much of the country restricted currency payments, but the restrictions were not universal.1 The last such 4 restriction was the banking holiday o f March 1933. In the banking holiday of 1933, the fed eral government closed all banks in the country and gradually reopened those that regulators judged to be in satisfactory financial condition. In the earlier restrictions, in contrast, banks re mained open and processed transfers o f depos its for their customers. Other than for the restriction of payments in 1907, it is difficult to obtain precise estimates of how widespread or binding these restrictions were. Shortly after the panic of 1907, A. Piatt Andrew surveyed banks in 147 cities in the United States with populations greater than 25,000. Andrew (1908) found that, of the 145 cities for which he had responses, 53 had no restriction of payments or emergency response. Of the remaining 92, the only restriction of payments in 20 cities was a request by the banks that larger depositors mark their checks as "payable only through the clearing house.” In the remaining 72 cities, limits on withdrawals w ere often discretionary. Even in the 36 cities where there was joint agreement between the banks in the city to limit withdrawals, there was substantial variations across them. For ex 13A data appendix, available on request from the authors, gives the sources of these dates and the other data in this paper. 14For a discussion of 1873 and 1893, see Sprague (1910), pp. 63-74, 168-69. Andrew (1908) presents the results of a survey for 1907. BANK OF ST. LOUIS FEDERAL RESERVE ample, in Atlanta, depositors could withdraw up to $50 per day and $100 per week from their banks. At the same time, depositors in two of these 36 cities, South Bend, Indiana, and Youngs town, Ohio, could withdraw nothing from their checking accounts. The Relative P rice o f Currency and Deposits During the periods of restrictions o f currency payments in the national banking era, markets developed in New York City for the exchange of currency for certified checks. Holders o f cer tified checks marked "payable through the clearing house” could obtain currency in this market if they were willing to accept less than the face amount of the certified checks. Figure 1 shows the premiums on currency quoted in these markets in the three periods of restric tions in New York City in the national banking era. These markets operated for about four months in this period. The maximum premiums on currency are about 4 percent to 5 percent, but for most of the days in which these mar kets operated, the premiums are much smaller. Nonetheless, the important issue is whether the premiums are nonzero, which they are. Clearinghouses and Restriction During these restrictions o f payments, banks remained open for much o f their regular busi ness and processed checks for their customers as they usually did. In some parts of the coun try, banks in a local area processed checks bilaterally, but in other areas, banks used clear inghouses to process checks. From 1857 to 1914, these clearinghouses developed an emer gency currency used during restrictions for clearing checks. C lea rin ghou ses f o r ba nks — In the second half of the nineteenth century, banks in many cities established clearinghouses to decrease the resources used in clearing checks and exchang ing gold and currency with other banks.1 5 Rather than sending checks received to the of fices of each bank for collection, members o f a clearinghouse sent checks drawn on other mem ber banks to the clearinghouse. Those with net ^D escriptions of clearinghouses are provided by Cannon (1910), Myers (1931), pp. 94-97, and Redlich (1968), ch. XVII. 49 Figure 1 Currency Premiums during Restrictions of Currency Payments September and October 1873 Percent Percent September and October 1893 Percent Percent MAY/JUNE 1989 50 November and December 1907 Percent outflows of deposits at the clearinghouse paid those with net inflows in gold and currency or, more conveniently, with clearinghouse certifi cates. These certificates were receipts for banks’ deposits of gold and legal tender at the clearing house. C lea rin g h ou se loa n c e rtific a te s — In some periods, clearinghouses issued additional certifi cates called “clearinghouse loan certificates” that could be used to clear checks. These certificates w ere a commonly used expedient in runs from 1860 until the creation of the Federal Reserve.1 6 The precursor of these loan certificates was an extraordinary issue o f clearinghouse cer tificates in the run on banks in 1857. Fears about the solvency of banks resulted in a drain o f specie and ultimately a run on the banks in New York City in 1857.1 At this time, banks 7 issued notes that were used as currency, and the banks redeemed them in gold or silver on demand. If a bank failed though, holders of the 16This section owes much to the analyses in Timberlake (1984) and Gorton (1985b). 17This account is based on Gibbons (1859), ch. XIX; Myers http://fraser.stlouisfed.org/ Federal ReserveFEDERAL RESERVE BANK OF ST. LOUIS Bank of St. Louis Percent notes could wind up with less than the prom ised amount of specie. In 1857, holders o f bank notes were concerned about the likelihood that banks in various parts of the country would be able to continue converting their notes into specie at par. As a result o f the continuing redemption of their notes, these banks con verted their correspondent balances in New York City banks into specie for redeeming their own notes. Thus, specie balances in New York City banks dwindled and this drain o f reserves culminated in a run on banks in New York City. On October 13, banks in New York City re stricted specie payments, with restriction in many other parts of the United States following. In part, the effect o f this specie drain on banks in New York City was alleviated by a joint agreement of the banks in the New York Clearing House on November 7. New York state banks that w ere not redeeming their notes agreed to pay 6 percent interest on them, and the clearinghouse agreed that the notes of the (1931), pp. 97-99, 141-44; Calomiris and Schweikart (1988), pp. 31-56. 51 banks could be used as backing for clearing house certificates. Until they were gradually retired, these certificates w ere used for clearing checks just as if they were clearinghouse certi ficates backed by deposits of specie. Clearinghouse loan certificates w ere first issued in 1860. In anticipation o f war, Southern ers converted their deposit balances in Northern banks into specie and, just as in 1857, banks in New York City were confronted with a drain of their specie reserves.1 After the election of 8 Abraham Lincoln in November, the banks in the New York Clearing House responded to the drain by jointly agreeing to allow bonds issued by the federal government and the state o f New York to be used as backing for certificates, call ed "clearinghouse loan certificates,” which could be used for clearing checks. The procedure adopted in 1860 was basically the same as in every later instance when such certificates were issued. A loan committee was established which examined collateral and is sued certificates based on the collateral. Upon using a loan certificate, a bank was required to pay interest, at a rate fixed by the clearing house, to any bank that held its loan certifi cates.1 The members o f the clearinghouse, 9 however, w ere jointly liable for any loss atten dant on holding a loan certificate. In addition, the clearinghouse agreements specified a date at which loan certificates would no longer be ac ceptable for settling balances at the clear inghouse. Several features o f the practices of clear inghouses indicate that, in issuing loan cer tificates, members o f a clearinghouse were pool ing their resources to deal with a common pro blem of withdrawals. Clearinghouse members pledged to absorb any losses on loan certificates as a group, with losses allocated according to each bank’s capital. Losses w ere not likely, however, because the borrowing banks pledged assets with the clearinghouse, receiving loan certificates for a fraction of the value of the assets. In some panics, clearinghouse members stopped the weekly publication of their individ 18Swanson (1908) provides a detailed account of this episode. ual balance sheets and published combined balance sheets o f their members, thus withhold ing information on the relative weakness o f in dividual members.2 0 Clearinghouse loan certificates were created several times in the 55 years from 1860 to 1914. Table 4 shows the dates when these cer tificates were issued by the New York Clearing House.2 As a quick comparison of tables 3 and 1 4 shows, clearinghouse loan certificates were issued whenever convertibility o f deposits into currency was restricted. This is no coincidence, because clearinghouse loan certificates w ere an important part of banks’ strategy for staying open after a run on the banking system. Although first issued in 1860 in New York City only, the use o f clearinghouse loan certifi cates became widespread over time (Stevens 1894; Andrew 1908; Cannon 1910). In 1873, the clearinghouses in New York City, Boston, Cincin nati, New Orleans, Philadelphia and St. Louis issued them. In 1884, New York City again was the only clearinghouse to issue loan certificates, but in 1890 it was joined by Boston and Phila delphia. In 1893, clearinghouses in at least 12 cities issued loan certificates, and in 1907, banks in 42 of 145 cities in the United States with more than 25,000 people used such certificates. Loa n c e rtific a te s and re s tric tio n s — Even with access to clearinghouse loan certificates, banks could provide currency in a run only un til they exhausted their inventories of specie and legal tender.2 During restrictions, banks ra 2 tioned currency, meeting the requests by some customers for their customary withdrawals of currency and denying requests by others. Banks that were members of the clearinghouse con tinued to accept checks drawn on other clear inghouse members when deposited by their customers. As a result, depositors could make payments by writing checks drawn on their ac counts or with certified checks issued by their banks. The major limitation was that the checks generally could not be exchanged for specie or currency by the recipient of the check. panics. Sprague (1910), pp. 46, 120; Myers (1931), pp. 408-20. 19The annual rates were 7 percent in 1860 and 1873 and 6 percent in every other instance when they were issued. Comptroller of the Currency (1915, vol. 1), p. 103. 21The New York Clearing House authorized but did not issue loan certificates in December 1895 and August 1896. Gorton (1985b), p. 280, fn. 11. 20Members of the New York City Clearing House agreed to pool reserves in the panic of 1873 but not in the following 22This section draws heavily on Sprague (1910). MAY/JUNE 1989 52 Table 4 Clearinghouse Loan Certificates Issued by the New York City Clearing House: Dates, Duration and Magnitudes M onths until all redeemed Year November 23 September 19 November 6 March 7 September 22 May 15 November 12 June 21 October 26 August 3 Deposits Date firs t issued 1860 1861 1863 1864 1873 1884 1890 1893 1907 1914 Maximum am ount created 3 7 2 3 3 4 2 4 5 4 1/2 1/4 3/4 1/4 3/4 1/4 3/4 2/3 $ 6.9 22.0 9.6 16.4 22.4 21.9 15.2 38.3 88.4 109.2 $ 99.6 99.3 159.5 168.0 174.8 317.2 386.5 398.0 1023.7 Sources: see data appendix available on request. If a check was not deposited at the issuing bank but at another bank in the local clear inghouse, the issuing bank could obtain more loan certificates to settle with the bank that ac cepted the check. If the check was deposited at a bank in another area, the receiving bank could deposit the certified check with a corres pondent in the clearinghouse of the issuing bank. Initially, these certificates were used only as a means of payment by other members o f the clearinghouse, but in later years, they also were used as currency.2 In 1893, clearinghouse loan 3 certificates w ere issued in small denominations by some clearinghouses as a substitute for cur rency. In addition, banks in several cities with no clearinghouse issued notes that were jointly guaranteed by the banks in the cities. In 1907, banks in many parts of the country created loan certificates which temporarily were used as currency. In 53 of the 71 larger cities in which banks jointly created loan certificates, banks issued the certificates to the public as cur rency. These issues of currency, which were extra-legal, were given legal status by the AldrichVreeland Act, which permitted associations of 23Andrew (1908) and Cannon (1910), pp. 107-112, ch. XI, discuss this aspect of clearinghouse loan certificates. Stevens (1894), pp. 145-48, provides some information for 1893 based on contemporary correspondence. Timberlake (1981) discusses the significance of private money in non panic periods. http://fraser.stlouisfed.org/ Federal ReserveFEDERAL RESERVE BANK OF ST. LOUIS Bank of St. Louis national banks to issue temporary currency. Na tional banks used that privilege in 1914. RESERVES, CURRENCY AND MONEY Ratio o f R eserves to Deposits Clearinghouse loan certificates w ere at least a partial remedy for runs on the banking system because, with access to them, banks could op erate with relatively low reserve ratios.2 Figure 4 2 shows the reserve ratios for banks in New York City weekly from 1853 through 1909. The vertical lines in the figure indicate the first week when the extraordinary certificates of 1857 or clearinghouse loan certificates were issued. As one can see, the reserve ratios gener ally drop around the dates when the New York clearinghouse issued loan certificates, reflecting the effects o f bank runs. During several periods when they used loan certificates to cover adverse clearings among themselves, the reserves o f banks in New York City fell below required levels (25 percent o f deposits after 1874) for at least a short period. 24lt also was possible for the banks to create sufficient loan certificates that interest-earning assets as well as deposits expanded. According to some authors [Cannon (1910), pp. 75-136; Sprague (1910), pp. 45-46, 171], one of the objec tives of clearinghouses in authorizing loan certificates was to expand loans by clearinghouse members. 53 In 1873, 1893 and 1907, the banks in the New York City clearinghouse restricted convertibility shortly after they had begun borrowing clear inghouse loan certificates. The reserve ratio rose sharply after the banks restricted pay ments, and they built up substantial excess re serve positions before resuming payments to depositors. The New York City banks also built up their excess reserves substantially after they created these certificates in 1860 and 1884, and after the creation o f the extraordinary cer tificates of 1857. The decreases in reserve ratios at the time of runs were short-lived. Indeed, the quarterly data in figure 3 for all banks in the United States from 1853 to 1935 do not show these sharp declines in the reserve ratio. They do show, though, the increases in the ratio after banks restricted convertibility. Ratio o f Currency to Deposits W e would expect a rise in the ratio of non bank money held by the nonbank public to bank money in a run on the banking system, at least until banks limited the reserve outflow by restricting payments. The year-end data for 1856 and 1857 show some indication o f an ef fect o f withdrawals in the panic of 1857, which occurred in the fall of that year. The ratio of specie held by the public relative to bank notes and deposits rose from 47 percent in December 1856 to 57 percent in December 1857. Figure 4 shows these data and quarterly data on the currency-to-deposit ratio for the U.S. banking system from 1867 to 1935. This ratio generally increases around the dates when banks in New York City issued clearinghouse loan certificates or restricted currency payments. The most extreme rise in the currency ratio in figure 4 occurs in the early 1930s. Friedman and Schwartz (1963) argue that the rise in the currency ratio was more extreme in the early 1930s than before the operation of the Federal Reserve System because, rather than restricting currency payments, the banks expected the Fed to provide reserves. In the event, the Federal Reserve failed to provide sufficient reserves.2 5 25See Friedman and Schwartz (1963), pp. 167-72, 308-12. 26Friedman and Schwartz (1963), ch. 2, discuss this period in detail. They attribute these movements to runs on banks M o n ey Growth As the example in table 2 illustrates, a bank run results in a decrease in the money stock for a given monetary base. Table 5 shows the quar ters with relatively large decreases in the money stock from 1867 to 1935 and zero or positive growth of the monetary base. Every quarter with a decrease in the money stock at greater than a 2 percent annual rate and non negative growth of the monetary base is includ ed in the table. Of the six periods in table 5, only one — 1877 to 1878 — is not associated with a restriction of convertibility or the creation o f clearinghouse loan certificates in New York City. The de creases in the money supply in 1877 and 1878 occur during the Treasury’s retirement of greenbacks prior to resumption of dollar con vertibility into gold on January 1, 1879.2 All of 6 the dates o f general restriction — 1873, 1893, 1907 and 1933 — are periods in which the money stock fell and the base increased for at least one quarter. The year 1884 has some characteristics of bank runs: banks in New York City created clearinghouse loan certificates, but conversion of deposits into currency was not restricted. As the table indicates, the highest rates o f decrease in the money stock occurred from 1931 to 1933, after the Federal Reserve was established. EFFECTS OF BANK RUNS While the previous section presents evidence that there were several episodes o f runs on the U.S. banking system before the Federal Reserve was formed, it provides little indication of the importance of their effects. This section pro vides some perspective on the impact of those runs. Losses b y Depositors The premiums on currency provide one mea sure of the cost of runs to bank depositors. In terms of currency, depositors suffered a loss on their deposits during these periods. The premiums indicate that, immediately after runs on the banking system, some people were will ing to exchange currency for certified checks at 96 cents or more on the dollar and, within a outside New York City. Friedman and Schwartz (1963), pp. CC-CQ O O " ’ IUIAV/.IIINP 1QRQ 54 Figure 2 Reserve Ratio Weekly Data 0 .4 - 0 .3 - 0.3 i i i i i i i i i i i i— i— r 1853 1860 1870 1880 Note: Vertical lines are the dates on which the New York City clearinghouse began issuing loan certificates. month, the currency premiums w ere less than 2 percent.2 7 Depositors also suffered losses when banks closed. The total losses borne by depositors in closed banks from 1865 through 1933 were at an annual rate of .21 percent of total deposits. Before the Great Depression, the general trend o f these loss rates was downward. The loss rates were .19 percent in 1865-80, .12 percent in 1881-1900, .04 percent in 1901-20, and rose to a peak of .34 percent in 1921-33. These figures are for all years and understate the loss rates in years with runs. Depositors’ losses on total deposits exceed .25 percent in 12 years: 1873, 1875-78, 1884, 1891, 1893 and 27lt is worth noting that these losses by depositors were counter-balanced at least in part by gains by holders of currency. The bid-ask spread would be a measure of the direct real resource cost of nonpar trades. http://fraser.stlouisfed.org/ Federal Reserve n F B A Iof R P ^LouisP St. P R V F F Bank R IN K H P « T I n illC 1930-33. The average loss rate in these 12 years is .78 percent of total deposits. In all but two of these periods, either convertibility of deposits was restricted or clearinghouse loan certificates were issued in New York City. In only one year was convertibility of deposits into currency restricted, loan certificates issued, and the loss rate less than .25 percent: 1907.2 8 In the 1930s, for which data on individual years are available, it is possible to get reason ably accurate estimates o f loss rates borne by the depositors in closed banks. The losses were not borne evenly across the population: an average loss rate per dollar of total deposits of .47 percent of total deposits in 1930 does not 28Unfortunately, the data before 1920 are provided only as averages for periods of several years; we know that the loss rates in 1907 and 1908 were not as high as .25 per cent, but we do not know more about them. 55 Weekly Data 0.5 0.0 1 r~ I I r l— I | — i— n — i— i— i— i— i— I i— i— i— i— i— i— n — r — — — 1880 1890 1900 1909 Note: Vertical lines are the dates on which the New York City clearinghouse began issuing loan certificates. convey the losses borne by individual depositors in individual banks. The average loss rates for depositors in banks that failed are about 28 per cent in 1930 and 15 percent in 1933.2 9 In sum, two things seem to be clear from these data. First, some holders o f bank liabilities did bear significant losses during periods with runs. These losses w ere not necessarily caused by the runs themselves. The runs and the losses both may have been triggered by events outside the banking system. It is possible, though, that 29The loss rates in closed banks for every year from 1921 through 1929 are higher than from 1930 to 1933. This decrease in depositors’ loss rate in banks closed in years with runs is not necessarily surprising because runs can force banks to liquidate with positive net worth or net worth less negative than it might be otherwise. This latter observation is consistent with the FDIC’s observation that the runs increased the losses from what they might have been under different institutional arrangements. Second, before the creation of the Federal Reserve, depositors’ loss rate from failed banks were declining over time. In this regard, it is worth noting that depositors' loss rate in 1907 was not as high as in as many previous periods, even though the panic of 1907 was the appar ent impetus for the creation of the Federal Reserve System. loss rates are less after the 12 “ crisis years” than in other non-crisis years. FDIC (1940), pp. 65, 69. The loss rates for the national banking period are substan tially, but not always, lower than some of the loss rates estimated by King (1983) and Rolnick and Weber (1988) for the earlier free banking period (1838 to 1863). M AV/.II I N F 1QRQ 56 Figure 3 Reserve to Bank Money Ratio Note: Vertical lines are the dates on which the New York City clearinghouse began issuing loan certificates. Figure 4 Nonbank Money to Bank Money Ratio Quarterly Data Note: Vertical lines are the dates on which the New York City clearinghouse began issuing loan certificates. http://fraser.stlouisfed.org/ Federal Reserve BankAof St. .Q F R U F P P n P R I R F Louis RANK O F QT in ill«5 57 Losses b y Bank Shareholders: Bank Failures Table 5 Growth Rates of the Money Supply and the Monetary Base in Periods in Which the Money Supply Declined at More Than a 2 Percent Annual Rate and the Growth Rate of the Monetary Base was Zero or Positive: 1867 through 1935 (annual growth rates of quarterly data, seasonally adjusted)__________ Money supply Period M onetary base 1873: 4 -1 3 .7 % 1877: 2 3 4 - 4 .5 - 3 .8 - 7 .2 0.1 0.1 1.1 1878: 1 2 - 4 .9 - 3 .5 1.1 4.3 1884: 3 - 2 .6 8.0 1893: 2 3 -8 .7 -1 0 .7 2.9 16.9 1907: 1908: 3 4 1 - 8 .0 -1 1 .1 - 5 .5 0.0 41.3 2.1 1929: 1 -3 .1 1.9 1931: 2 3 4 -1 1 .2 -1 4 .7 -3 0 .8 12.5 11.2 13.3 1932: 2 3 -1 3 .6 -5 .9 13.9 5.7 1933: 1 - 3 9 .8 20.7 2.5% Sources: see data appendix available on request. 30See Dowrie (1913), Krueger (1933) and Economopoulos (1988). During restrictions, two things happened. Banks w ere able to stop the drain o f reserves and possibly the sale o f assets at distress prices. In addition, they w ere able to take stock and determine which banks might survive the panic. The importance o f this effect perhaps is most clearly indicated by a comparison of Illinois and Wisconsin banks just before the Civil War. Banks in Illinois did not restrict specie payments and, ultimately, 93 out o f 112 of the banks closed. With similar portfolios o f assets, banks in Wisconsin did restrict specie payments and few er of them, 50 out o f 107 banks, ultimately closed.3 0 Another way o f getting an idea of the costs to banks is to compare failure rates in banking panics before 1933 with the failure rate in 1933. At the onset of the Depression, banks did not issue clearinghouse loan certificates or restrict currency payments. While the Federal Reserve increased the monetary base, the base was not increased sufficiently to prevent re peated runs until the restriction o f payments in the Banking Holiday. As a result, 1933 provides a contrasting indicator of how serious banking panics can be. Figure 5 shows that banking panics can in deed be associated with relatively large num bers of banks failing. Nonetheless, it is notewor thy that, before 1933, the only year with restriction and a large increase in the failure rate is 1893. M a croecon om ic Effects3 1 Figure 6 shows the monthly average call loan rate for 1857 through 1935. Call loans are over night loans with stock as collateral that are callable without notice. Because call loans were a part of their assets that they were not con tractually obligated to continue for longer per iods, banks in New York City reduced their call loans when they wished to convert part of their assets into reserves. In figure 6, vertical lines denote the periods when banks in New York Ci ty restricted convertibility or had large drains (1983), Bordo (1986), Gorton (1988), Kaufman (1988), Tallman (1988) and Grossman (1989). 3 For other discussions of the macroeconomic effects of 1 bank runs, see Friedman and Schwartz (1963), Bernanke MAY/JUNE 1989 58 Figure 5 Bank Suspension and Failure Rate 1864 1870 1880 1890 1900 1910 1920 1933 Note: Vertical lines are the dates on which the New York City clearinghouse began issuing loan certificates. Figure 6 Call Rate Note: Vertical lines are the dates on which the New York City clearinghouse began issuing loan certificates. FEDERAL RESERVE BANK OF ST. LOUIS 59 Table 6 Bank Runs and the Timing and Severity of Recessions 1857 to 1933______________________________________ Recession peak Recession troug h Clearinghouse loan certifica te s issued R estriction o f curre ncy paym ents June 1857 Dec. 1858 i Oct. 1857 Oct. 1860 June 1861 April 1865 Dec. 1867 n.a. June 1869 Dec. 1870 - 2 .8 % Oct. 1873 March 1879 March 1882 May 1885 March 1887 Sept. 1873 n.a. n.a. Sept. 1873 April 1888 July 1890 Nov. 1860 Percentage change in pro du ction -1 9 .3 May 1884 -1 3 .7 May 1891 Nov. 1890 -2 9 .2 Jan. 1893 June 1894 June 1893 Dec. 1895 June 1897 - 9 .2 Aug. 1893 -5 4 .9 -2 1 .9 June 1899 Dec. 1900 - 6 .7 Sept. 1902 Aug. 1904 -2 0 .0 May 1907 June 1908 Jan. 1910 Jan. 1912 Oct. 1907 Oct. 1907 Jan. 1913 Dec. 1914 Aug. 1918 Mar. 1919 - 8 .8 Jan. 1920 July 1921 -7 1 .3 May 1923 July 1924 -5 3 .8 Oct. 1926 Nov. 1927 Aug. 1929 Mar. 1933 -5 0 .8 -2 1 .1 Aug. 1914 -4 5 .8 -1 7 .9 March 1933 -8 5 .6 1 Banks created certificates backed by bank notes. Sources: see data appendix available on request. of reserves to which they responded by issuing clearinghouse loan certificates. While the in creases in the call loan rate associated with restrictions and drains are not unique, some are extraordinary. Evidence that would support the view that bank runs had adverse effects on the economy would be as follows: bank runs occurred just prior to the onset o f recessions, and more severe recessions followed banking panics. Table 6 provides information on the timing and severity of recessions and the timing o f bank runs. The data do not support a simple conclu sion on the macroeconomic effects of bank runs. Other than the episode in 1873, banks created clearinghouse loan certificates and restricted currency payments several months after the beginning o f the recessions. While some of the more severe recessions occurred when banks restricted currency payments, this is consistent with two very different conclu sions: restrictions led to severe recessions, or severe recessions led to restrictions. Table 6 also indicates that several recessions occurred without runs on the banking system. These observations provide information about the stability o f the U.S. banking system without a federal safety net. Several recessions, with declines in real output and losses to businesses, occurred apparently without undermining the confidence of the public in the safety of bank deposits to the point of starting runs on the banking system. MAY/JUNE 1989 60 CONCLUSION The federal safety net for the banking system includes the Federal Reserve as the lender of last resort, federal deposit insurance, and bank supervision and regulation designed to limit the risk assumed by banks. The rationale for this safety net is that, in its absence, the banking system would be vulnerable to the kind of run on the banking system that occurred in the ear ly 1930s. The run in the early 1930s, however, was, perhaps, the most extreme run on the banking system in U.S. history. While several runs on the banking system took place before the formation o f the Federal Reserve System in 1914, banks took actions that limited their effects. By issuing clearinghouse loan certificates that other banks accepted to clear checks, banks operated temporarily with relatively low reserve ratios. In the more severe runs, bankers jointly restricted payments but continued operating. Moreover, even prior to the creation o f the federal safety net in the United States, runs on the banking system were infrequent. The banking system can operate for many years without runs on the banking system, even in recessions. Dewald, William G. “ The National Monetary Commission: A Look Back,” Journal of Money, Credit and Banking (November 1972), pp. 930-56. Diamond, Douglas W., and Philip H. Dybvig. “ Bank Runs, Deposit Insurance, and Liquidity,” Journal o f Political Economy (June 1983), pp. 401-19. Dowrie, George W. The Development o f Banking in Illinois, 1817-1865. Volume 11, No. 4 of the Series, University of Il linois Studies in the Social Sciences (University of Illinois, December 1913). Economopoulos, Andrew J. “ Illinois Free Banking Exper ience,” Journal of Money, Credit, and Banking (May 1988), pp. 249-64. Federal Deposit Insurance Corporation. Annual Report, (Federal Deposit Insurance Corporation, 1940). Flood, Robert P., and Peter M. Garber. “ Bubbles, Runs, and Gold Monetization,” in Paul Wachtel, ed. Crises in the Economic and Financial Structure (Lexington Books, 1982). Friedman, Milton, and Anna J. Schwartz. A M onetary History of the United States, 1867-1960 (Princeton University Press, 1963). Gibbons, J.S. The Banks o f New York, their Dealers, the Clearing House, and the Panic o f 1857. Original publication, 1859. (Reprint edition, Greenwood Press Publishers, 1968). REFERENCES Gorton, Gary. “ Bank Suspension of Convertibility,” Journal of M onetary Economics (March 1985a), pp. 177-93. Andrew, A. Piatt. “ Substitutes for Cash in the Panic of 1907,” Q uarterly Journal o f Economics (August 1908), pp. 497-516. ________ “ Clearinghouses and the Origin of Central Banking in the United States,” Journal o f Economic History (June 1985b), pp. 277-83. Bernanke, Ben. “ Nonmonetary Effects of the Financial Crisis in the Propagation of the Great Depression,” American Economic Review (June 1983), pp. 257-76. ________ “ Banking Panics and Business Cycles,” Oxford Economic Papers (December 1988), pp. 751-81. Bordo, Michael D. “ Financial Crises, Banking Crises, Stock Market Crashes and the Money Supply: Some International Evidence, 1870-1933,” in Forrest Capie and Geoffrey E. Wood, eds., Financial Crises and the World Banking System (St. Martin’s Press, 1986), pp. 190-248. Grossman, Richard S. “ The Macroeconomic Consequences of Bank Failures Under the National Banking System,” U.S. Department of State. Bureau of Economic and Business Af fairs, Planning and Economic Analysis Staff, Working Paper 14, April 1989. Calomiris, Charles W., and Larry Schweikart. “ Was the South Backward?: North-South Differences in Antebellum Banking during Normalcy and Crisis,” unpublished paper, Northwestern University, August 1988. Cannon, James G. Clearing Houses, U.S. National Monetary Commission, Senate Document No. 491, 61 Cong., 2nd Sess. (U.S. Government Printing Office, 1910). Comptroller of the Currency. Report (U.S. Department of the Treasury, 1915). Davis, Andrew M. The O rigin o f the National Banking Sys tem, U.S. National Monetary Commission, Senate Docu ment No. 582, 61 Cong., 2nd Sess. (U.S. Government Prin ting Office, 1910). http://fraser.stlouisfed.org/ FEDERAL RESERVE Federal Reserve Bank of St. Louis BANK OF ST. LOUIS Kaufman, George G. “ The Truth About Bank Runs,” in Catherine England and Thomas F. Huertas, eds. The Financial Services Revolution: Policy Directions for the Future (Kluwer Academic Publishers, 1988), pp. 9-40. King, Robert G. “ On the Economics of Private Money,” Journal of M onetary Econom ics (July 1983), pp. 127-58. Krueger, Leonard. History o f Com m ercial Banking in Wisconsin (University of Wisconsin, 1933). Myers, Margaret G. The New York Money Market, Vol. I, Origins and Development (Columbia University Press, 1931). 61 Redlich, Fritz. The M olding o f Am erican Banking, 2nd ed. (Johnson Reprint Corporation, 1968). Rolnick, Arthur J., and Warren E. Weber. “ Explaining the Demand for Free Bank Notes,” Journal of M onetary Economics (January 1988), pp. 47-71. Salant, Stephen W. “ The Vulnerability of Price Stabilization Schemes to Speculative Attack,” Journal of Political Economy (February 1983), pp. 1-38. Schweikart, Larry. Banking in the American South from the Age of Jackson to Reconstruction (Louisiana State Universi ty Press, 1987). Sprague, O. M. W. History of Crises under the National Banking System, U.S. National Monetary Commission, Senate Document No. 538, 61 Cong., 2 Sess. (GPO, 1910). Stevens, Albert C. “Analysis of the Phenomena of the Panic in the United States in 1893,” Quarterly Journal of Economics (January 1894), pp. 117-48, 252-60. Swanson, William Walker. “ The Crisis of 1860 and the First Issue of Clearing-House Certificates,” Journal of Political Economy (April 1908), pp. 212-26. Tallman, Ellis. “ Some Unanswered Questions about Bank Panics,” Federal Reserve Bank of Atlanta Economic Review (November/December 1988), pp. 2-21. http://fraser.stlouisfed.org/ Federal Reserve BankRoffSt. Louis A N K O F S T I O U IS FFnFRAI F iF R V F R Timberlake, Jr., Richard H. The Origins of Central Banking in the United States (Harvard University Press, 1978). ______ “ The Significance of Unaccounted Currencies,” Journal of Economic H istory (December 1981), pp. 853-66. _____ . “ The Central Banking Role of Clearinghouse Associations,” Journal of Money, Credit and Banking (February 1984), pp. 1-15. Waldo, Douglas G. “ Bank Runs, the Deposit-Currency Ratio and the Interest Rate,” Journal of M onetary Economics (May 1985), pp. 269-77. Wall Street Journal. “ Panic Button Finally Triggers Reform, 1907,” February 16, 1989. Wicker, Elmus. “A Reconsideration of the Causes of the Banking Panic of 1930,” Journal o f Economic History (September 1980), pp. 571-83. Federal Reserve Bank of St. Louis Post Office Box 442 St. Louis, Missouri 63166 The Review is published six times per year by the Research and Public Information Department o f the Federal Reserve Bank 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, or address changes to: Research and Public Information Department, Federal Reserve Bank o f St. Louis, P.O. Box 442, St. Louis, Missouri 63166. The views expressed are those o f the individual authors and do not necessarily reflect official positions o f the Federal Reserve Bank o f St. Louis or the Federal Reserve System. Articles herein may be reprinted provided the source is credited. Please provide the Bank’s Research and Public Information Department with a copy o f reprinted material.