The modern American Economy is an impressive machine. Over this century, it has generated enormous advances in technology, absorbed millions of immigrants and women into the labor force, and seen massive investments in capital goods and a sharp rise in educational attainment, all of which have translated into huge increases in living standards. Yet, there are periods - more than 20 since 1900 - where the economy contracts and production drops. For households and firms, the declines in employment and income that result cause considerable distress, both financial and psychological.

Why and how does this happen? What causes a healthy economy with a highly educated work force and a large store of modern equipment to falter? Why don't prices and wages adjust to reach full employment? Are these economic downturns inevitable?

You may not be heartened to learn that 70 years after the Great Depression, and more than 60 years after the publication of the book that inaugurated the field of macroeconomics, John Maynard Keynes's The General Theory of Employment, Interest, and Money, economists still aren't sure. Although one of the most serious issues that a modern economy faces, it may also be the least well understood. In 1997, the American Economic Association saw reason to have a session at its annual meeting entitled, "Is There a Core of Practical Macroeconomics That We Should All Believe?” The answer was a weak "Yes,” but the causes of recession were notably absent from the list.

WHAT HAPPENED LAST TIME: 1990-91

It has been nine years since the start of the most recent U.S. recession. It began, according to the National Bureau of Economic Research (the people in charge of dating the peaks and troughs), in July 1990, about a month before the United States invaded Kuwait. It was relatively brief, ending eight months later in March 1991, although a sluggish early recovery made the downturn seem to last much longer and kept unemployment rising - up to 7.8 percent in June 1992 - even as the economy started to come back.

By the time production had climbed past its earlier peak, the cumulative cost to the nation in lost output was about three-quarters of a month of GDP. Personal income had declined in real terms (that is, in relation to prices), as is typical. So had spending, particularly on big-ticket, discretionary purchases that can be deferred until income begins rising and confidence in the economy returns. According to former Federal Reserve Bank of Boston economist Stephen McNees, household spending on residential housing and consumer durables, such as cars, appliances, and furniture, declined 10 percent and 7.1 percent, respectively, from peak to trough. Business spending on capital goods such as computers and factories dropped 7.3 percent.

Thus, manufacturing and construction firms bore the brunt of the downturn, shedding one million jobs between them. The production of cars and light trucks was hammered hard, McNees found, dropping 28 percent. Services sector employment, which tends to be less sensitive to the business cycle, rose by 0.6 percent.

As is typical, the pain of unemployment or a significant loss in income was not spread randomly across the labor market. Manufacturing and construction workers experienced higher rates of unemployment than services workers, especially managers and professionals. Less-educated workers, particularly those without high school diplomas, also saw their unemployment rates rise by more than rates for college graduates. Unemployment rates for black male teenagers, always extraordinarily high (around 32 percent in 1988-89), reached 40 percent in September 1991.

But the 1990-91 recession was also atypical as compared to past recessions in certain ways. Employment declines for less-skilled workers were still greater than for the more skilled, but the difference was narrower as this recession hit more educated and white-collar workers hard as compared to past downturns. Services jobs, although continuing to grow slightly, also fared badly compared to the past; in the previous five recessions, services employment had risen an average of almost 3 percent.

And, as cycles go, this one was milder than average. The downturn was relatively short, about eight months long as compared to an average of 11 months for recessions after World War II. The cumulative loss of GDP was also less than average. And the unemployment rate saw the smallest increase of any recession since the war. (New England, however, was hit hard. See the sidebar.)

Compare this to the Great Depression of 1929, the most calamitous economic event in the United States this century. Within less than three years, the nation lost production equal to almost half of the entire output loss from all other recessions this century, according to calculations by U.C. Berkeley economist Christina Romer. Measured unemployment rates soared to 25 percent, with the true rate probably far worse, as many people just gave up looking for work. As MIT Professor Peter Temin observed, "Policies that avoid similar catastrophes may be more important than policies that fine-tune the economy.”

PROXIMATE CAUSES

To avoid catastrophes, it helps to know what causes them. In a paper prepared for a 1998 conference, "Beyond Shocks What Causes Business Cycles?,” sponsored by the Federal Reserve Bank of Boston, Professor Temin attempted to determine the dominant cause of each of the recessions since 1890. In particular, he looked for a break - a shock or change - to an important economic relationship that might have triggered the subsequent decline in production and employment. Such a break might come from a fall-off in demand (a drop in consumption or investment spending in response to a war or other event that erodes confidence) or events on the supply side (an increase in the cost of an important input such as oil). It might, as his MIT colleague Rudiger Dornbusch has suggested, originate with the Federal Reserve and contractionary monetary policy, which by choking off liquidity and raising short-term interest rates slows spending on housing, capital goods, and other interest-sensitive demand. Or it might be imported from abroad, as fallout from economic or financial distress that begins in a foreign country.

Temin examined evidence from the historical literature and concluded that there was no single underlying source for all of this century's recessions. About two-thirds had their origins in the domestic economy, the other one-third in external events, he decided. Roughly half could be attributed to a monetary event; half began with a shift in the supply of or demand for goods and services. Temin also looked for systematic differences in the causes of big and small recessions, and found none.

But determining causation is a tricky undertaking, as Temin notes in his paper. Your underlying theory about how the economy works will matter; those (very few) who doubt on theoretical grounds that monetary policy is capable of affecting anything in the economy except prices would never ascribe any recession to actions of the Federal Reserve. It also requires separating out the "cause” from a chain of connected events, a subtle task and one that is open to interpretation and subject to disagreement. For example, all recessions since World War II have been preceded by rising inflation that has, in turn, prompted tighter monetary policy. While many would argue that such policy was appropriate and should, in fact, be given credit for dampening business cycles over that period, Dornbusch and others argue that the Fed "caused” these recessions.

Assigning a single cause - even to any particular recession - may be impossible, for downturns often result from a combination of events. In 1990-91, for instance, you can look for causes as far back as March 1988. With the economy arguably at or near capacity, the Federal Reserve raised the federal funds rate and short-term interest rates soon followed. Around the same time, problems in the real estate industry meant that many bankers and bank regulators became cautious in approving loans, and the resulting "credit crunch” may have squelched some investment demand. Once Iraq invaded Kuwait in summer 1990, consumer confidence fell and the price of oil rose, resulting in declines in spending and personal income. And Robert Hall of Stanford has suggested that, with economic conditions hard to discern, the Fed reacted cautiously in lowering the federal funds rate later that summer. All these factors may have contributed to the subsequent recession; perhaps, none alone would have been enough.

What does seem important is whether the economy is vulnerable - weakened enough so that it is unable to handle a disturbance that might have only mild consequences when the economy was more robust. In 1990-91, for instance, the economy was already slowing and, thus, may have been particularly sensitive to the spending drop that occurred around the time of the invasion of Kuwait. Perhaps the recent currency crisis that began in Asia and spread to Russia and Latin America could have triggered a recession in the United States if the domestic economy had been more vulnerable.

And it may be a mistake to look for recessions' causes only in economic factors. "Virtually all recessions have occurred around the time of some highly distinctive, not purely economic event such as a war, a massive change in the price of imported oil, a major strike, or wage, price, and credit controls,” observes McNees. "Recessions almost always come as a surprise,” he reminds us, "though they seem easy to 'explain' after the fact.”

WHY DON'T PRICES AND WAGES FALL?

So now we have a vulnerable economy and a proximate cause - some disturbance, perhaps a war, and a drop in consumer confidence and spending, as in 1990-91. But if wages and prices adjusted quickly, the economy might not experience such destructive declines in GDP and employment. More specifically, if real wages declined when demand fell, then firms might not have to cut production and lay off workers, and a recession might be averted.

But many economists believe that prices and wages, while not completely rigid, are somewhat "sticky” and slow to adjust. Industrial economists have long observed that manufactured goods seem to have relatively inflexible prices as compared to agricultural products. Several have tracked the price changes of individual products, such as newspapers and magazines, and found that prices change infrequently, even when aggregate prices are rising. And Robert Gordon has looked at the overall level of prices and found evidence of stickiness at various times in U.S. history, including the years before World War I, as well as later in this century.

Why do prices and wages fall only slowly, especially when the alternative is layoffs and shutdowns? This is one of the big unanswered questions in macroeconomics. While a variety of hypotheses have been offered, good data have been tough to come by. Many of the conjectures - such as that firms put off lowering prices because price is a signal for quality to customers, or because it is difficult and costly to reprice contracts, print new signs, and the like - require measures not available with the precision necessary for statistical testing. And the usual labor market data on compensation and employment can't say much about why wages are slow to adjust.

So two economists went out and asked business people to see if that would help resolve the question.

Yale economist Truman Bewley spoke to more than 300 managers in firms in the northeastern United States and asked them about wages. Traditional thinking had focused on why workers refuse to accept lower wages and risk unemployment, with much of the speculation emphasizing the role of unions.

But Bewley found that resistance to pay cuts came primarily from employers, not from workers or unions. And the main reason that employers gave was their belief that cutting pay hurts morale and increases labor turnover. According to Bewley, employers would like their employees to identify with the objectives of the firm and to cooperate with coworkers and supervisors in that spirit. They want workers to operate autonomously, show initiative, use imagination, and be willing to take on extra tasks. Workers who are scared or disheartened do not do these things. So long as employers believe that pay cuts or reduced hours are harmful to workers'morale, Bewley explains, they are not a useful alternative to layoffs. Layoffs also affect morale, but the damage is likely to be brief. "Those laid off may suffer terribly, but once they are gone, they cannot disrupt the workplace.” The damage is also limited to a relative few. Even in the most severe recessions, 90 percent of workers still have jobs, with little incentive to support a wage cut.

Princeton economist Alan Blinder undertook a comparable exercise, quizzing managers from more than 200 companies on their pricing practices and motivations. He found that most companies changed their prices annually, although some changed them even less frequently and 10 percent changed them more than 50 times a year. Most did not respond right away to changes in market conditions; firms reported that it took an average of three months after a cost or demand shift before they made a change in their product's price. But companies also showed great variability 20 percent reported changing prices immediately, while 13 percent said they delayed as much as six months. Services firms adjusted prices slowly, trade firms did so more rapidly, with manufacturers in between. And firms seemed to take about as much time before lowering prices as before raising them.

About two-thirds of the companies reported that, in their view, they had an implicit agreement with customers not to raise prices when markets were tight - which perhaps makes sense considering that 70 percent of sales were made to other businesses, and 85 percent to repeat customers. Blinder estimates that more than one-quarter of private nonfarm GDP is actually sold under explicit written contracts that fix prices for some period of time.

But when asked why prices are slow to adjust, the answers that companies gave were more helpful for understanding why firms are slow to increase prices than for why they are slow to lower them. Less than half the firms reported that the costs of repricing were significant; and repricing costs tended to be largest for price increases. The most commonly cited reason was that managers hesitate to raise prices out of fear that competitors will not follow suit, but this doesn't reveal much about why firms are slow to reduce prices when demand drops. The concern that frequent price adjustments would "antagonize” customers because they would think it "unfair” also doesn't explain firms' reluctance to lower prices. And the fact that many firms say they respond to market conditions by varying delivery lags, sales effort, and product quality rather than by changing prices is intriguing, but puzzling. Neither the reasons that firms would choose to make adjustments in this manner nor the implications for the macroeconomy are clear. So why don't wages and prices fall quickly in response to a shortfall in demand? As Blinder himself put it, "Theorists have a long way to go.”

STICKY PRICES VERSUS REAL BUSINESS CYCLES

Thus, the matter remains fundamentally unsettled. Some economists dispute the importance of sticky wages and prices. They point out that what looks like wage stickiness might in fact be the outcome of long-term employment relationships; workers and firms agree to maintain a stable "average” wage rather than compensation that rises and falls over the cycle with output and productivity. Others note that a slow wage adjustment process implies that when demand drops and the economy enters a downturn, we should observe real wages that are "too high.” But empirical efforts are inconclusive on this point, either showing no relationship between real wages and the cycle or showing that real wages are, in fact, slightly lower early in a recession. Some are just plain dubious about whether we can tell if wages or prices are sticky. "As compared to what?” they ask. Just because prices don't change for a while is not in itself proof they are rigid. What would flexible prices look like? Are the pricing practices that Blinder cites evidence of flexibility? Or inflexibility?

Even those who believe that prices and wages adjust slowly take issue with some of the formal models that incorporate this process in an ad hoc way or impose price rigidity on the data rather than testing it. Moreover, in simulation exercises, many of the formal models can't generate theoretical cycles that resemble the ones that we observe in the real world.

The main intellectual competition for the past decade or so, known as "real business cycle theory,” has problems of its own. Its proponents maintain that recessions are not the result of price stickiness or other problems in the way markets adjust. They argue, instead, that downturns occur when productivity or other shocks to supply cause workers to reallocate their desired hours of work over time. Suppose, for example, a change in technology (or an increase in oil prices, or bad weather) lowers wages. Workers might decide to reduce their current hours of work because the return to work is low, and then increase work hours later on, when wages have returned to normal levels. Thus, both employment and production would fluctuate over the business cycle in response to these productivity shocks. And unemployment would consist of people who are voluntarily choosing leisure They may be willing to work, but only if they are offered the wages they receive in nonrecession years.

But whether technology shocks - negative ones - are large enough to produce recessions with the frequency and magnitude we observe is open to question. Moreover, according to evidence from labor markets, people do not seem to respond to wage changes by substantially reallocating leisure over time, as the theory suggests. And many find it hard to believe that most unemployed people are voluntarily choosing leisure over work. If this were so, say the doubters, the 1930s should be known not as the Great Depression but as the Great Vacation.

So economists in both camps - and those who cross party lines - keep working, adding refinements, and testing them against the data. Some are looking at labor markets and studying consumption and investment decisions to learn more about why we observe sticky prices. Perhaps price stickiness is desirable or serves some purpose, as yet not clearly identified. Some have focused on identifying patterns in job creation and job destruction by industry, firm size, and other variables. Perhaps these patterns can provide clues as to how changes in technology or demand in a particular sector of the economy might eventually reduce demand economywide and trigger a national recession. Perhaps the explanation has been hidden in aggregate data and will be uncovered in plant or industry data. There is still much we do not know.

ARE RECESSIONS INEVITABLE?

Since World War II, the nation's economic expansions have lasted longer and its recessions have become less frequent (although they are not noticeably shorter or less severe than those between 1886 and World War I). This is due partly to better macroeconomic policy. Tighter management of inventories and a growing share of employment in the less cyclically sensitive services industries may have also helped. Still, most economists think it unlikely that the United States has seen its last recession.

Some point out that economic "shocks” will always be with us. Wars, technological upheaval, foreign economic instability, and bad economic policy are probably inevitable, if ultimately unforeseeable. When they do occur, the economy, caught off guard, takes time to adjust.

Others see recessions as originating in the expansions that precede them. Victor Zarnowitz, for example, argues that business profits, investment, and credit conditions tend to interact in a way that turns booms into busts. In an expanding economy, high profits and a rising stock market encourage firms to invest, he notes. As the boom continues, firms may begin to undertake riskier and riskier investments and the danger of overconfidence grows. Eventually, growth begins to slow, and profits and investment decline as well. Business failures must be written off; credit markets may move to safer and more liquid assets, which can create a credit crunch. Investment and consumption spending fall off, as boom turns to bust. Just as overconfidence can lead to asset bubbles, "herd” psychology can amplify declines. As earnings reports disappoint, the stock market may also drop and exacerbate the downturn. Thus, the seeds of recessions may be sown in the expansions that went before.

Nor is there necessarily any way to prevent them. As Paul Samuelson has noted, financial economists have crafted clever new products - options and other derivatives - to improve efficiency in the pricing of specific financial assets, such as an individual stock, but no opportunity exists to make money by trying to correct mispricing in the general level of stock market prices.

So whether recessions are generated by "shocks” or they are inherent in the process of expansion, the next one is probably inevitable. And while it is possible to assess the economy's vulnerability and to implement policies that reduce its exposure to risk, there is no way to predict when the next recession will arrive or what its cause will be. Economists will have to keep looking for the answers.

WHICH INDUSTRIES ARE HIT HARDEST?

When a downturn hits, few industries go completely unscathed, but some suffer greater job loss than others. The hardest hit are those connected to the production of capital goods and to construction where demand is sensitive to interest rates and can be delayed until times get better. Thus, furniture has the most cyclically sensitive employment; beverages the least. Except for temporary help agencies, the softest blows are in services, especially food-related activities and education.

THE 1990-91 RECESSION WALLOPS NEW ENGLAND

The most recent U.S. recession was especially rough in New England, where it began earlier, hit harder, and lasted longer. The downturn began after February 1989, a good 16 months ahead of the nation, as the collapse of the 1980s construction and real estate boom sent companies into bankruptcy and some banks into failure. Over the next three and one-half years, New England lost a staggering 650,000 jobs - one in 10 - and the unemployment rate exceeded 7 percent for 28 straight months. The recession accelerated the exodus of manufacturing jobs from the region that had begun in 1984; construction employment also fell dramatically, dropping about 40 percent. But employment declined in all sectors, with finance, insurance, and real estate falling by nearly 7 percent and even services jobs dropping 1 percent. The economy finally began its slow climb to recovery in April 1992, 11 months after the national upturn. It was not until November 1997 that New England finally surpassed its employment level at the recession's start.

Other parts of the nation were not nearly so hard hit. The Mountain region, for example, barely saw a dip in its job numbers before employment began to rise again, and some of the Pacific states were also relatively unscathed. This was in sharp contrast to the recessions of the early 1980s, when some Midwestern states saw large employment declines and New England suffered much less. Boston Fed Economist Robert Triest believes that regional differences may play a part in creating and propagating national recessions. Problems in New England and California may have been a contributing factor in the 1990-91 U.S. recession, he claims, with the resulting reductions in personal income and consumption eventually helping to bring about declines in demand and income across the country.

HARD HIT

Industry employment has a high correlation with business cycle fluctuations Household furniture

Miscellaneous plastic products

Personnel supply services

Plumbing supplies

Stone, clay, and misc. mineral products Metal coating and engraving

Concrete, gypsum, and plaster product

Cutlery, handtools, and hardware

SOFTER BLOW

Industry employment has a low correlation with business cycle fluctuations

Beverages

Agricultural chemicals

Accounting and auditing

Educational services

Commercial sports

Communications equipment

Membership organizations

Museums, botanical gardens

NOTE: CALCULATED FOR RECESSIONS AFTER 1977

SOURCE: JAY BERMAN AND JANET PFLEEGER, "WHICH INDUSTRIES ARE SENSITIVE TO BUSINESS CYCLES?" MONTHLY LABOR REVIEW, FEBRUARY 1997