Reading Into Jobless Claims

在目前用于经济预测的各种数据模型和复杂工具中，有一项简单的数据在预测经济衰退的起止时间上历来十分准确，那就是首次申请失业救济人数，该数据反映的是每周有多少美国人新申请了失业救济，它每周四由劳工部(Labor Department)发布。虽然自本轮经济衰退开始以来，每周首次申请失业救济人数一直在大幅上涨，不过这一数字最近已显现了一些趋稳态势。上周，经季节因素调整的首次申请失业救济人数骤减53,000人，降至61万人，但一些人提醒说这可能是受到了复活节假日因素的影响。四周平均首次申请失业救济人数上周下降了8,500人，降至65.1万人。这是该数据三个月来首次下降。预测人士之所以喜欢追踪研究首次申请失业救济人数，既因为它能及时反映劳动力市场状况，也因为该数据一贯是判断经济活动下滑势头何时接近尾声的良好指标。美国西北大学(Northwestern University)的经济学教授戈登(Robert J. Gordon)任职于美国一个负责确定经济衰退起止时间的委员会，他发现首次申请失业救济人数是一个很有价值的数据。戈登研究了上世纪60年代末至今的这一数据后发现，官方正式宣布的经济衰退截止时间出现在四周平均首次申请失业救济人数见顶后约一个月的几率相当高。截至目前，四周平均首次申请失业救济人数的峰值出现在截止于4月4日的那一周，为65.95万人。如果这一峰值今后不被打破，那么从以往情况看，有可能意味着本轮经济衰退将在3月底至5月初之间结束。这种判断要比当前所有的普遍预期都乐观许多。 (《华尔街日报》对经济学家的最新调查显示，受访者平均预计本轮经济衰退将于9月份结束。)戈登说，离隧道尽头可能只有几周了。当然，要想让人相信戈登的乐观预期确实很不容易。首先，开始于2007年12月的本轮衰退是二战以后持续时间最长的一次经济衰退，且以许多指标来衡量也是最严重的一次，所以首次申请失业救济人数这一指标的预测力是否还能像以往一样强大仍有待观察。其次，该指标的波动性很大，如果美国目前为8.5%的失业率像许多人预计的那样攀升至两位数，那么每周首次申请失业救济人数也很有可能继续增加。高频经济公司(High Frequency Economics)的经济学家谢弗德森(Ian Shepherdson)说，如果每周首次申请失业救济人数明显呈现下行趋势，则很有可能表明整体经济来到了转折点，但这种情况几个月内不会出现。Kelly Evans（“经济实时报”提供有关经济美国联邦储备委员会政策和经济事件的独家新闻分析和评论，《华尔街日报》记者和编辑均参与供稿，由Sudeep Reddy和Phil Izzo主笔。）相关阅读美国就业状况2010下半年以前难好转 2009-04-10


Amid all the data, models, and sophisticated tools that today's economic forecasters have at hand, there's one simple series with a pretty impressive track record for predicting when recessions are coming to an end: jobless claims, a weekly tally of how many new claims Americans are filing for unemployment benefits, released every Thursday by the Labor Department.Jobless claims have risen sharply during this recession, but shown some stabilization lately. Last week, newly filed claims plunged by a seasonally adjusted 53,000 to 610,000 - which some cautioned was a quirk of the timing of the Easter holiday - and the four-week average of new claims fell by 8,500 to 651,000, the series' first decline in three months.Forecasters love tracking jobless claims because they're a timely read on the labor market, but also because they have historically been a great way to determine when declines in economic activity are nearing an end.Robert J. Gordon, an economics professor at Northwestern University who sits on the committee tasked with dating recessions, is one who finds enormous value in this series. Going back to the late 1960s, he has found that the four-week average of new claims peaks about a month before the declared end of recessions with remarkable accuracy.As of right now, the four-week average claims series peaked at a level of 659,500 in the week ended April 4. If that number holds, based on the series' past performance it would mean the recession ended somewhere between late March and early May - a far more optimistic read on the economy than any consensus forecast (the latest WSJ survey of economists shows on average they expect the recession to end in September). 'The end of the tunnel may only be weeks away,' says Mr. Gordon.Of course, that's a pretty big 'if.' First of all, the current recession - which began in December 2007 - has been longer and by many measures more severe than any other postwar recession, so it remains to be seen whether jobless claims will have the same predictive power they've shown in the past. Secondly, the weekly series is volatile and could well keep rising as the nation's unemployment rate, now 8.5%, heads towards the double-digits many expect.'A clear downturn in claims would be a strong signal of a turn in the broad economy, but we think that is still a few months off,' said Ian Shepherdson, an economist with High Frequency Economics, in a note following the release.Kelly Evans