张军|复旦发展研究院副院长

复旦大学经济学院院长

复旦大学中国经济研究中心主任

The Truth about Chinese Unemployment Rates

SHANGHAI – Since 2002, China’s economy has undergone significant changes, including a shift from acceleration to deceleration of GDP growth. Yet the official urban unemployment rate, jointly issued by the National Bureau of Statistics (NBS) and the Department of Labor and Social Security, has remained remarkably steady, at around 4-4.1%. Since 2010, it has stood at precisely 4.1%. This is surprising, to say the least – and has led some to ask whether the NBS could be fudging the numbers.

中国经济从2002年起经历了显著的变化，其中包括GDP增速由加快转为放缓。然而由国家统计局与人力资源和社会保障部联合发布的城镇登记失业率却牢牢地维持在4%-4.1%左右。尤其是从2010年起，失业率一直稳定在4.1%。这让人不得不讶异，或者可以说，会让人怀疑国家统计局是否篡改了数据。

The NBS is not lying; it simply lacks data. The unemployment rate that the NBS provides reflects how many members of the registered urban population have reported to the government to receive unemployment benefits. But, unlike in developed countries, China’s piecemeal unemployment insurance and underdeveloped reemployment programs weaken the incentive for people to seek assistance. As a result, the NBS unemployment figures are far from accurate.

然而国家统计局并没有篡改数据，只是由于缺乏数据。国家统计局提供的登记失业率反映的只是拥有城镇户籍的失业人口中向政府申请失业补助的那部分。与发达国家不同的是，中国失业保险的碎片化与再就业项目的薄弱消减了人们寻求失业救助的积极性，这也导致了国家统计局的失业数据失真。

China’s government has moved to remedy this, by carrying out urban unemployment surveys. But, despite having been collected a decade ago, those statistics have yet to be released.

为了着手补救，中国政府已经开始进行大城市失业调查以获得更为科学的调查失业率。但尽管调查已历经10年，但仍没有完善的体系使相关统计数据得以披露。

In lieu of convincing official figures, some economists have taken matters into their own hands, using data from the urban household survey (UHS) to estimate the real unemployment rate. Extrapolating from UHS data gathered in six provinces, Jun Han and Junsen Zhang, for example, concluded that, in 2005-2006, Chinese unemployment stood at around 10%. Using UHS data from almost all of China, Feng Shuaizhang, Hu Yingyao, and Robert Moffitt calculated an average urban unemployment rate of 10.9% from 2002 to 2009 – the highest estimate ever produced.

一些经济学家开始自行收集数据，特别是中国城镇住户调查数据（简称UHS）来测算真实失业率以替代官方公布的登记失业率。举个例子，韩俊和张俊森通过六个省份的城镇住户调查数据，推断出2005年至2006年中国的城镇失业率在10%左右。冯帅章、胡颖尧和罗伯特•墨菲特则根据几乎涵盖全国范围的城镇住户调查数据计算出2002年至2009间，中国平均城镇失业率为10.9%——这是目前为止最高的测算值。

But these estimates are just that – estimates. Because UHS data are not freely available, different people obtain results for different years and provinces from the various sources they could access. This has caused considerable frustration for researchers, and has resulted in estimates with ranges so wide as to be statistically insignificant.

但是这些测算也仅仅是测算。由于城镇住户调查（UHS）数据无法轻易获得，不同的人只能从各自的渠道得到不同年份和省份的数据。如此一来给研究人员带来了相当大的困难，并且导致估算范围过大而丧失统计意义上的显著性。

In our own research at Fudan University in Shanghai, my two PhD students, Liheng Xu and Huihui Zhang, and I managed to obtain a reasonably broad supply of official statistics: the 2005-2012 data for four provinces, the 2005-2009 data for three provinces, and monthly data for 2010-2012 for four of these seven provinces. While the sample is technically small, the provinces for which we acquired data represent the coastal, inland, and northeast regions. With the right adjustments and processing, we were able to infer the unemployment rates in different kinds of provinces and municipalities, thereby estimating the real nationwide unemployment rate.

我的两个博士生徐力恒和张慧慧协助我一起在复旦大学开展研究，我们获得了2005年至2012年间4个省份的数据，2005年至2009年间3个省份的数据，此外还有2010年至2012年间7个省份中4个的月度数据。尽管从技术上来说样本并不大，但涵盖了沿海、内陆和东北等地区。经过适当的调整和处理，能够推断出不同省（或直辖市）的失业率，并以此估算全国范围的失业率。

We found that, although China’s urban unemployment rate was probably quite high in 2005, standing at 10.7%, it has most likely dropped over the last decade, reaching 7% in 2012. That puts the annual average for the 2005-2012 period at 8.5%. (These and our other findings correlate with a cross-sectional analysis of the official data, meaning that the data for registered unemployment, subjected to such an analysis, might serve as a proxy for the real unemployment rate.)

我们发现，尽管2005年中国的城镇失业率可能高达10.7%，但它在过去的十年间逐步降低，并于2012年降至7%，这使得2005年至2012年平均的城镇失业率维持在8.5%（我们测算的不同省份的数字与官方登记失业率数据在横断面上具有正相关性，这意味着在横截面分析下，登记失业率可以作为真实失业率的代理变量）。

Moreover, while rapid GDP growth contributed to falling unemployment in, say, 2007, unemployment continued to decline even after the global financial crisis of 2008 began to weaken economic performance. Most economists would assume that declining unemployment amid falling GDP growth is related to a decline in labor-force participation. But our calculations, based on the UHS data, show that labor-force participation in China actually increased slightly after 2008, as the proportion of workers exiting the labor market decreased. In other words, China’s employment growth has accelerated in recent years, even as GDP growth has slowed.

此外，尽管高速的GDP增长有助于解释失业率的不断下滑（比如2007年GDP增速达到14.20%），但是在2008年全球经济危机削弱中国经济表现（GDP增速下降为9.60%）后失业率仍然继续下降。大多数经济学家都认为失业率数字的下降可能是劳动参与率降低的结果（因为退出劳动力市场的人口并不统计在失业人口之中）。但我们通过对城镇住户调查数据进行计算发现，由于退出劳动力市场的那部分人口比例下降，中国的劳动参与率（这里劳动力年龄限制在16岁至退休年龄）在2008年后反而略有增长。这表明，尽管2008年以后中国的GDP增长已经降速，但是就业增长率却得到了保持。

This can be explained partly by an ongoing structural shift in the Chinese economy, from a manufacturing-driven growth model to a services-led model that empowers private innovators. And, indeed, as UHS data show, this shift led to continuous job creation in the services sector from 2005 to 2012.

我们可以从中国正在推进的经济结构转型中得到部分答案。中国经济正从制造业驱动型增长模式转向服务业主导的增长模式，这带动了大量私有的创新型企业的兴起。正如城镇住户数据所显示的，这一转型的确导致了2005年至2012年间服务业工作机会的不断增加。

What has not happened is significant destruction of jobs in the state sector and manufacturing industries, especially since 2009. As the UHS data suggest, the average time it takes an unemployed worker to find a new job in the services and non-state sectors is shorter than in the manufacturing and state sectors. If the manufacturing and state sectors do begin to lay off more employees, urban unemployment rates are bound to rise.

然而国有企业与制造业并没有出现大规模释放失业的现象，特别是2009年以后。正如城镇住户调查数据所示，平均而言在制造业和国有部门失业的劳动者其失业持续期要比在服务业和非国企部门失业的劳动者来的长。也就是说一旦制造业企业与国有企业开始大量释放失业，城镇失业率必然上升。

The reason that hasn’t already happened is that the government has, to some extent, been propping up these sectors since the global financial crisis, by implementing massive stimulus packages focused on investment in infrastructure and real-estate development. This has sustained the rapid growth of the secondary sector (especially manufacturing and construction), which has thus been absorbing large numbers of low-skill workers. In fact, upon closer examination of the UHS data, we found that the least-educated workers largely accounted for the decline in overall unemployment.

上述情况之所以没有发生，是由于中国政府在全球金融危机后实施了大规模经济刺激计划，主要的投资也集中于投资基础设施建设和房地产领域。此举维持了第二产业的快速增长（尤其是制造业与建筑行业），也吸收了大量低技术工人。而事实上，透过对城镇住户调查数据的进一步分析，我们发现总体失业率的下降大概可以用低教育程度的劳动者的失业率下降来解释。

This stimulated infrastructure and real-estate construction boom has also led to the expansion of heavy industry, including state-owned steel, cement, chemicals, glass, and other enterprises, causing employment growth to accelerate from 2005 onward. The NBS data show that the employment growth rate in the state sector was negative before 2009, when it turned positive.

基础设施和房地产建设受政策刺激影响，也导致了包括钢铁、水泥、化工、玻璃等在内的国有重工业的扩张，加速了2005年后的就业增长。国家统计局的数据也显示，国有部门的就业增长率在2009年由负转为正。

The fact that the unemployment rate is declining while GDP growth slows suggests that China’s labor productivity is actually worsening – a trend that is likely to lower China’s long-term potential growth rate. Since the effects of the stimulus obviously cannot last, the sectors that were being propped up will soon begin to shed more workers, causing the unemployment rate to rise. Only further government intervention could prevent this outcome; but that might mean delaying structural reforms that are needed to sustain productivity growth. China’s leaders are thus being confronted with a difficult choice: higher near-term unemployment or slower long-term growth.

中国GDP增速放缓而失业率的下滑表明中国的劳动生产率的增长事实上出现了恶化，这一趋势有可能降低中国的长期潜在增长率。鉴于经济刺激的效果无法持续，一度被扶持的部门不得不开始释放劳动力，进而可能导致失业率的上升。只有政府进一步干预才能规避这种结果，但这也意味着维持生产率增长所必须的经济结构改革将被延缓。因此，中国的领导人面临着两难选择：要么是短期内更高的失业率，要么是长期内更缓慢的经济增长。