爱思英语编者按:经典回归模型是建立在平稳数据变量的基础之上的,对于非平稳变量,不能使用经典回归模型,否则会出现虚假回归等诸多问题。

Soft science no more
不再是软科学
 
This year's Nobel prize shows how far number crunchers have come in economics
今年的诺贝尔奖表明,与数字打交道的人已经在经济学中走得很远了。
 
ONCE, a beautiful turn of phrase would take you a long way in economics. From Adam Smith to John Maynard Keynes, economists were content to put their theories and ideas into (mostly) English prose, and leave it at that. Their big, breezy thoughts made great, but imprecise reading. Contradictions were glossed over. So prolix was Keynes, for example, that he is thought to have said everything at least once.
 
以前,一段漂亮的措辞会带着你在经济学中徜徉徘徊。从亚当·斯密到约翰·梅纳德·凯恩斯,经济学家满足于把他们的理论和思想变成散文(主要是英文),然后就置之不问。他们那些重大且玄妙思想造就了伟大但不精确的阅读。矛盾之处被掩盖了起来。例如,凯恩斯的冗长,据说他把所有事情至少都说了一遍。
 
This will no longer do. Since 1945 or so, practitioners of what was once called “political economy” have become more demanding. They sought to test their grand thoughts against the hard facts of the real world. Incomes, interest rates, and prices of all sorts could be measured. Did they behave as theory supposed? National accounts, the detailed measures of GDP, were just being created, at Keynes's behest. But economic data were, and still are, messy; analysing them even more so. It took decades to develop the tools to detect and measure economic relationships with much certainty.
 
这种情况将不会再有了。自1945年前后开始,曾被称为“政治经济学”的实践者变得更加苛刻。他们力求通过现实世界的硬事实来验证他们的宏大理论。收入、利率以及各种各样的价格都可能得到衡量。它们如理论所设想的那样表现了吗?国民核算——GDP的详尽衡量——正是在在那个时候,应凯恩斯的要求问世了。但是,经济数据过去是、现在仍是一团乱麻;对它们的分析更是如此。直到几十年才开发出了探寻并衡量经济关系的工具。
 
This year's Nobel prize has gone to two economists who epitomise the rise of statistical techniques: Robert Engle, an American economist at New York University, and Clive Granger, a Briton at the University of California at San Diego. They have crafted some of the most sophisticated tools to analyse economic data. Their contributions, developed during the 1980s, deal especially with “time-series” data: share prices, household consumption, inflation—anything, in fact, that changes over time, and thus poses difficulties for older forms of statistical analysis.
 
今年的诺贝尔奖花落两位代表统计技术崛起的经济学家:纽约大学经济学家、美国人罗伯特·恩格尔 (Robert Engle)和加州大学圣迭戈分校、英国人克莱夫·格兰杰(Clive Granger)。他们精心打造出了一批高精尖的工具,以分析经济数据。他们在上世纪80年代做出的贡献专门对付“时间序列”(time-series)数据:股票价格、家庭消费、通货膨胀——实际上,凡是随时间而变化、因而给老式统计分析带来困难的东西。
 
Poets and plumbers
诗人和工匠
 
The Nobel committee seems to be highlighting the wide range of ideas and skills that comprise modern economics: a healthy equilibrium, you might say, between poets and plumbers. Last year one of the two winners was a psychologist whose findings contradict many of the assumptions of economic theory. This year the plumbers were back. Only three years have passed since James Heckman and Daniel McFadden were also honoured for sharpening econometrics, the statistical methods with which economic data are analysed.
 
诺贝尔评委会似乎是为了突出构成当代经济学的广泛的思想和技能:或许是诗人与匠人间的一种健康的均衡。去年,两位获奖者中的一人曾是一位其发现与许多经济理论的假设想矛盾的心理学家。今年,工匠回来了。这距离詹姆斯·哈克曼和丹尼尔·麦克法登因为优化计量经济学——经济数据以之得到分析的统计方式——也受到奖励才刚刚过去三年。
 
Mr Engle and Mr Granger have crafted techniques that demand even greater virtuosity at maths, but which are nevertheless crucial in separating wheat from chaff. Mr Engle's work has helped build the foundations for measuring and avoiding myriad types of risks in the modern economy. He has studied the volatility—the severity of swings—of time series ranging from inflation to the prices of securities. Anyone who watches the stockmarkets knows that they undergo periods of wild adolescent swings as well as times of geriatric languor. Until Mr Engle came along, people interested in such things—financial types, mostly, but also regulators—used crude measures of historical volatility, looking back over a year, say, to see what the average of the swings was. They would then use this as a gauge of likely future volatility.
 
恩格尔与格兰杰精心打造出了在数学方面要求有更大的精确性、但对去芜存菁来说却是至关重要的技术。恩格尔的研究促成了测算基础的建设,并为避免当代经济中花样繁多的风险提供了帮助。他研究了从通胀到股价等时间序列的波幅,即波动的严重程度。凡是观察股市的人都知道,它们经历过疯狂的成长波动阶段,也经历过暮年的软弱无力时期。恩格尔之前,对此类事情感兴趣的人——主要是金融人士,也包括监管机构——曾经使用粗略的历史波幅测算,比如说向后看一年,来预测波动的平均值。之后,他们会把这个平均值用作可能的未来波幅的一个参照物。
 
Mr Engle's approach, ARCH (for autoregressive conditional heteroscedasticity, should you insist on knowing) gave researchers the power to test whether and how volatility in one period is related to volatility in earlier times. There often is a link, as casual observation suggests. After several days of stockmarket upheaval, there may be several days of calm. A 3% rise or fall in shares is often heralded by increasing volatility, much as an earthquake is preceded by tremors. Mr Engle's high-powered maths has made market risk easier to forecast. Thus banks and investors who use “value at risk” techniques to analyse their portfolios owe much to Mr Engle. So does the Basel committee which is drawing up new rules for banks' capital requirements.
 
恩格尔的方法——ARCH(如果非想知道的话,它代表autoregressive conditional heteroscedasticity,即自回归条件异方差性)给了研究者验证一个周期中的波幅是否与先前周期中的波幅相关以及如何相关的力量。如偶尔的观测所示,这里经常存在一种联系。数天的股市上涨后,可能有几天的平静期。股票上涨或下跌3%经常是以日渐加大的波幅为前兆,这同地震以震动为先兆非常相似。恩格尔的高性能数学让市场风险更容易预测。因而,使用“风险估值”(value at risk)技术分析投资组合的银行和投资者亏欠恩格尔许多。正在为银行资本金要求制定新规的巴塞尔委员会也是如此。
 
Mr Granger's research was aimed more at coming to grips with longer-term swings in economic growth, inflation and currencies than with shorter bouts of risk and volatility. Macroeconomic data often share some common features. GDP per head, for example, has tended to grow over time (at least for as long as it has been reliably measured). But the “trend” rate of growth discussed by forecasters—and, yes, by journalists—is never as fixed as they make it seem: it can be influenced by shocks like rising oil prices or wars.
 
格兰杰的研究更多地聚焦于把握经济增长、通胀和货币中的长期波动,而不是短期的风险和波动。宏观经济数据经常具有一些共性。例如,人均GDP往往是随着时间而增长(至少)。但是,预测者——当然还有记者——所讨论的增长“趋势”率,从未像他们使其看上去那样一成不变:它可能受到不断上涨的油价或战争等冲击波的影响。
 
This may obscure deeper relationships hidden in the data, posing tricky problems for statisticians. Using standard statistical tools—derived from things that do not change much over time, such as the distribution of people's heights—can be misleading. An economist using these tools might conclude that a casual relationship existed where none really does, and thus be fooled by a statistical mirage.
 
这可能模糊隐藏在数据中的更深层次的关系,给统计人员造成棘手的难题。使用标准的统计工具——从身高分布等不太随时间变化的事物中推导出来的——可能让人误入歧途。使用这些工具的经济学家可能得出因果关系存在于实际上毫无关系的地方的结论,从而上了统计假象的当。
 
Mr Granger devised a clever solution to this, called co-integration. He made use of the notion of economic equilibrium—the idea that variables tend to move towards particular values, and thus in a predictable direction. He found that when two sets of economic data are compared, for example inflation and exchange rates, they can often be treated with standard techniques. In collaboration with Mr Engle, he worked to create tests for economists to ensure that they were getting reliable results when making such comparisons.
 
对此,格兰杰设计了一种聪明的解决方案——协整 ( co-integration)。他用到了经济平衡的概念——即各种变量往往是向着特定的价值、因而是在一种可以预测的方向中运动的思想。他发现,当两套经济数据,如通胀和汇率,被拿来对比时,它们可能经常被待之以标准技术。在与恩格尔的合作中,格兰杰曾为了给经济学家创造出能够确保他们在做此类对比时得到可信结果的验证而全力以赴。
 
Despite these sophisticated techniques, which economists now apply as a matter of course, the analysis of economic data remains messy. “Driving a Mercedes down a cow-track” is how Thomas Mayer, an American academic economist, once described the application of fancy tools to real-world phenomena that are not easy to model, much less to measure.
 
除了这些经济学家如今想当然地加以应用的高端技术,经济数据的分析仍旧是一团乱麻。“在慢车道上开奔驰车”正是美国经济学家托马斯·梅耶(Thomas Mayer)对于复杂工具之于不容易建立模型、并且更不容易去衡量的真实世界现象之应用曾经做出的比喻。
 
Even so, the place of econometricians at the centre of economics is now confirmed. Indeed, there now seems to be a dearth of the grand theorists of days past. Specialisation—to use an economists' term—is the order of the day. But then a good plumber is in greater demand than any poet.
 
即便如此,计量经济学家在经济学中心的地位,如今已经得到了确认。实际上,现在似乎存在一种昔日理论大家的缺乏。套用经济学家的话说,专业化是当今的秩序。但是,那时,好工匠比任何诗人都走俏。