The Nasdaq bubble was brief, but still irrational
Almost nobody will deny that the stock market developed a bubble in the last years of the 20th century. Consider famous episodes such as the market's inability to figure out that 3Com was the majority owner of Palm, the maker of handheld computers, the dizzying heights reached by the Nasdaq Composite index as it exploded in late 1999, more than doubling in value in the year up to its late-winter 2000 peak, and the huge bath taken by investors in the Nasdaq from February 2000 to September 2002 as the index lost three-quarters of its value. All this happened, even though the long-awaited recession proved shallower than anyone had forecast and trend productivity growth proved faster than even the most zealous boosters of the “new economy” had dared project. It is next to impossible to interpret these events within the context of a rational-expectations model, in which stock prices provide the best possible forecasts of future values. Only those who want their colleagues to doubt their own rationality even try.
The interesting thing about the late-1990s stock market bubble, however, is how short it was. Conventional histories start with the excitement produced by the initial public offering of Netscape, the internet company, in 1995. But the stock market was not then in a bubble. Those who invested in the Nasdaq in March 1995 and then shadowed the index have earned real returns averaging 9.3 per cent per year; those who invested in the Nasdaq in September 1995 the month after Netscape's IPO have earned 7.3 per cent annually since. Compare this with the average 8.8 per cent real return on the Nasdaq since the start of the 1970s, and it is clear there is no evidence that the Netscape-mad market of 1995 was significantly overvalued.
By the end of 1996, Alan Greenspan, chairman of the US Federal Reserve, was worrying about the stock market. “How do we know,” he asked an audience at the American Enterprise Institute, the Washington think-tank, “when irrational exuberance has unduly escalated asset values?” If you had invested in the Nasdaq while Mr Greenspan was writing his speech, you would have realised a real return from then until now of 8.1 per cent per year. The answer to the question Mr Greenspan implied in December 1996 “Are asset prices unduly escalated by irrational exuberance?” is no.
When, then, did the late-1990s bubble begin? That is a surprisingly hard question to answer. Since the start of 2000, the Nasdaq has been hit by two particular pieces of bad news: the September 11 2001 terror attack on the US, and the fact that it has turned out to be much harder than anyone expected to transform technical excellence in information and communications technology into durable profits. The beneficiaries of high-tech innovation since 2000 have not been the workers, entrepreneurs and financiers of Silicon Valley, but users of their products and ideas, such as Wal-Mart, the US retailer, its shareholders and its customers. Given that these two pieces of fundamental bad news were unknown (and unknowable) in the late 1990s, it seems reasonable to set up three yardsticks to assess the beginning of the bubble. Because the high-tech stocks in the tech-heavy Nasdaq are particularly risky, a rational market would be expected to price the Nasdaq to produce returns higher than the stock market's long-run historic average of 6.5 per cent per year. But we also suspect that actual returns have fallen behind rationally expected returns. If these two factors cancel each other out, and the Nasdaq was in a bubble, then returns from the start of that bubble period until now will have been less than the 6.5 per cent per year we expect from stocks. As a second, more stringent, yardstick, a bubble can be defined as a period during which real returns do not match the 3 per cent per year expected from investment in bonds. The third and most stringent yardstick would define a bubble as a period in which returns are negative.
According to the second and third yardsticks the bubble was remarkably short. Measured from then until now, the realised Nasdaq real return drops below 3 per cent per year in October 1998, and in November 1998 it drops below zero. According to these measures, it was overvalued for less than a year and a half before its peak in March 2000. According to the first yardstick, the bubble was a year and a half longer. Cumulative real returns on the Nasdaq have lagged behind the 6.5 per cent per year we expect from a diversified portfolio of stocks since April 1997. Using this yardstick, the bubble lasted for less than three years before the Nasdaq peaked. Fischer Black, the economist, offered another definition of a “large” bubble: a time when assets are twice their fundamental values or more. On this definition, the Nasdaq bubble became a “large” one only in September 1999, less than half a year before its peak.
You can use this analysis to give an optimistic reading of stock market performance: when the great and good Mr Greenspan, for example were puzzled and cautious, the market was still cranking out valuations that look, in retrospect, remarkably close to the fundamental values of the underlying assets. In spite of unprecedentedly rapid technological change with very uncertain long-run consequences and the fact that humans are prone to irrational exuberance, the stock market kept doing its job of feeding the real economy an appropriate shadow value of capital, and did not succumb to irrationality until 1997 or 1998.
Today many people including us are worried about high levels of bond prices and real estate in the US and elsewhere that seem incompatible with likely scenarios for the world economy. This look back at the bubble of the 1990s is somewhat reassuring: financial markets were not as dumb as we feared then, and so are probably not as dumb as we fear now. Like the late 1990s, any bond or real-estate bubble this decade will probably be of short duration.
However, short duration does not necessarily mean small magnitude. The size of the late 1990s stock market bubble at its peak still supports a very pessimistic reading of stock market rationality. Those who followed the advice of the “hyper-bulls” the Kevin Hassetts, James Glassmans, George Gilders and their ilk and bought the Nasdaq at the end of February 2000 expecting it to rise much further, have seen a realised real rate of return of minus 16 per cent per year. In spite of the recovery since late 2002, they are still down by more than half. Brad DeLong is professor of economics at the University of California, Berkeley, and Konstanin Magin is a post-doctoral fellow at the Center of Integrated Nanomechanical Systems 非理性的纳斯达克泡沫
几乎没有人会否认，股票市场在 20 世纪的最后几年形成了泡沫。想想这些著名的事件吧，比如市场没能弄清楚，其实 3Com 拥有手持电脑制造商 Palm 的多数股权； 1999 年末纳斯达克综指飙升，达到了令人眩目的高位，在至 2000 年冬季末市场见顶的一年里市值翻了一倍还多；还有，从 2000 年 2 月到 2002 年 9 月，纳斯达克综指跌幅达四分之三，投资者在纳斯达克市场上损失巨大。所有这些事情都发生了，尽管期待已久的衰退比任何人事先预测的程度都要浅，而生产力趋势增长如此之快，甚至超过了“新经济”最狂热支持者的大胆预测。要想在一个理性预期模型的背景下去解释上述事件几乎是不可能的，因为在这个模型中，股价为未来价值提供了最佳预测。只有那些想要同事质疑自己是否理性的人，才会去尝试这么做。
然而， 20 世纪 90 年代后期股市泡沫的有趣特点是，它如此短暂。 1995 年，网络公司网景 (Netscape) 首次公开发行 (IPO) ，令市场兴奋不已，人们普遍认为这就是纳斯达克历史的开端。但股市当时并没有出现泡沫。那些在 1995 年 3 月投资纳斯达克股市、随后又跟踪纳指的人，获得了 9.3% 的实际平均年回报率，那些在 1995 年 9 月网景首次公开发行后投资纳斯达克股市的人，此后的年回报率为 7.3% 。与之相比，纳斯达克市场自 20 世纪 70 年代开始以来实际平均年回报率为 8.8% ，显然没有证据能表明， 1995 年对网景狂热的市场价值被大大高估。
到 1996 年底，美联储 (Federal Reserve) 主席艾伦?格林斯潘 (Alan Greenspan) 开始担心股市。“我们如何知道 , ”他在华盛顿智囊机构美国企业研究所 (American Enterprise Institute) 对听众表示，“非理性繁荣已过度夸大了资产价值？”如果你在格林斯潘先生写这篇演讲时投资了纳斯达克股市，从那时到现在，你本可以实现 8.1% 的实际年回报率。对于格林斯潘在 1996 年 12 月所暗指的“非理性繁荣是否过度夸大了资产价格？”这个问题，答案是“否”。
那么， 20 世纪 90 年代末的泡沫是什么时候开始的呢？这是个特别难以回答的问题。自 2000 年年初以来，纳斯达克市场一直受到两条特别不利的消息冲击：其一， 2001 年 9 月 11 日，美国遭受了恐怖袭击；其二，事实表明，要把信息通讯的卓越技术转化为持续不断的利润，其难度远远高出任何人的预期。自 2000 年以来，高科技创新的受益者一直不是硅谷的工人、企业家和金融家，而是诸如美国零售商沃尔玛 (Wal-mart) 等使用他们产品与创意的用户、以及这些用户的股东和顾客。
鉴于在 20 世纪 90 年代末无人知晓 ( 也无从知晓 ) 上述两条重要的坏消息，设立三条标准来评定泡沫的开端似乎是合理的。由于在以科技企业为主的纳斯达克市场上，高科技股票风险特别大，预期理性市场在对纳斯达克股市定价时，它的回报率要高出股市 6.5% 的长期历史年平均水平。但我们也怀疑，实际回报率一直没有达到理性预期的回报率。如果这两个因素相互抵消，而纳斯达克市场也的确存在泡沫，那么从泡沫开始直至现在，回报率将会一直低于我们对股票所预期的每年 6.5% 的水平。第二条标准更加苛刻，即泡沫可以定义为实际回报率达不到 3% 的时期。 3% 是债券投资的预期年回报率。第三条标准最为苛刻，它将泡沫定义为回报为负数的时期。
根据第二和第三条标准，泡沫的持续时间非常短暂。根据从当时直至现在的计算， 1998 年 10 月纳斯达克市场实现的年实际回报率低于 3% ， 1998 年 11 月则回报率跌到了负值。根据这些衡量标准，在达到 2000 年 3 月峰值前，市场被高估的时间不足一年半。而根据第一条标准，泡沫时期要多出一年半。纳斯达克的累积实际回报率每年都不到 6.5% ，这个比率是 1997 年 4 月以来我们对多样化股票投资组合的期望回报率。利用这一尺度衡量，在纳斯达克见顶之前，泡沫持续了近 3 年时间。经济学家费雪?布莱克 (Fischer Black) 对“大型”泡沫有另外一个定义：当资产为其基本价值的两倍或更多时即为泡沫时期。根据这一定义，纳斯达克仅在 1999 年 9 月才成为“大型”泡沫，在市场见顶前持续时间还不到半年。
你可以用这一分析对股票市场的表现做一个乐观的解读：比如说，当伟大善良的格林斯潘先生迷惑不解并表示谨慎时，市场仍在推高股票估值，而现在回头看一下，这个估值看起来十分接近基础资产的基本价值。尽管科技变革空前迅速，并带来了很不确定的长期后果，而且事实上人类也容易受非理性繁荣的影响，但股票市场仍继续执行着它的任务：给实体经济提供一个恰当的资本影子价值，并且在 1997 年或 1998 年前都没有屈服于非理性状态。
今天，包括我们在内，很多人都在担心美国和其它地方的债券和房地产价格的高水平，这些价格水平似乎与全球经济的可能情况并不协调。这里对 20 世纪 90 年代泡沫的回顾多少能使人放心：金融市场并不像我们当时担忧的那样无能为力，或许也没有我们现在所担心的这样。和上世纪 90 年代后期一样， 2001 至 2010 这 10 年间出现的任何债券或房地产泡沫可能都将是短期现象。
然而，短期现象并不一定意味着规模不大。上世纪 90 年代后期股市在顶峰时的泡沫规模太大，从而使人们仍然对股市的理性持悲观看法。凯文?哈塞特 (Kevin Hassett) 、詹姆斯?格兰斯曼 (James Glassman) 、乔治?吉尔德 (George Gilder) 等人当时是“超级看涨派”，而那些听从了他们的建议，并在 2000 年 2 月底买入纳斯达克股票，期望股市能继续大幅上涨的人，他们每年实现的实际回报率为负 16% 。尽管自 2002 年末以来股市出现复苏，但他们的股票价值仍跌去了一半以上。
布拉德?德朗是加州大学伯克莱分校经济学教授；康斯坦丁?马金是综合纳米机械系统中心 (Center of Integrated Nanomechanical Systems) 的博士后研究员
