Productivity Paradox v Economist Robert Solow the father of the modern theory of economic growth famously said in 1987 that you can see the computer e

Transcription

1 2015 年 5 月 24 日 Issue No: 15/21 美国经济分析 生产率悖论 研究报告 近几年实测生产率增速显著放缓, 我们将潜在趋势的暂行假设下调至 1½%, 与 年的疲软水平相似, 并且远低于 2¼% 的长期均值 增速放缓的直接原因是信息技术的实际贡献大幅回落 然而生产率增速疲软是真实存在的吗? 我们对此心存疑虑 利润率已上升至历史高点, 通胀率总体上低于预期, 股票价格整体大幅上升, 科技股表现领先大盘 这些现象都不像是 IT 造成的生产率放缓 这些观测的一个可能的解释是美国经济的结构性变化可能导致实际 GDP 增速在统计上被低估 有几个领域或许令人担忧, 但软件和数字内容快速增长 经质量调整的价格和实际产出较大部分其他领域难测量得多 似乎尤为重要 具体而言, 我们认为有理由相信, 众所周知的与质量变化和新产品推出有关的通胀统计数据的偏高在软件和数字内容领域尤为严重 对这一影响进行量化分析较为困难, 但有理由认为该影响可能在很大程度上抵消了实测生产率的放缓 我们的分析有三点实际意义 第一, 应审慎对待对生活水平提高速度显著落后于以往的坚定看法 第二, 考虑到围绕 GDP 的不确定性, 更好的做法是关注其他指标 尤其是就业率 来衡量复苏的总体进展和剩余的闲置产能规模 第三, 真实通胀甚至可能低于实测通胀, 这加强了继续实施货币宽松政策的理由 Jan Hatzius (212) jan.hatzius@gs.com Alec Phillips (202) alec.phillips@gs.com Jari Stehn (212) jari.stehn@gs.com Kris Dawsey (212) kris.dawsey@gs.com David Mericle (212) david.mericle@gs.com Chris Mischaikow (212) chris.mischaikow@gs.com Karen Reichgott (212) karen.reichgott@gs.com 投资者不应视本报告为作出投资决策的唯一因素 有关分析师的申明和其他重要信息, 见信息披露附录, 或参阅 全球投资研究

2 Productivity Paradox v Economist Robert Solow the father of the modern theory of economic growth famously said in 1987 that you can see the computer everywhere but in the productivity statistics. At the time, labor productivity was growing at around 1½%, well below the 2½%-3% pace seen until the early 1970s, and the measured contribution of information technology to GDP growth struck most observers as surprisingly small. The Solow paradox was resolved over the following decade, or so it seemed. Exhibit 1 plots the trend in labor productivity growth measured as real GDP per hour worked in the nonfarm business sector against the contribution from the production and use of IT hardware, software, and communications equipment. 1 The chart shows that the reacceleration of productivity growth in the 1990s followed hard on the heels of a big increase in the IT contribution. Exhibit 1: Information Technology and the Productivity Cycle Percentage points 1.8 Percent change, annual rate IT Contribution to Productivity* (left) Labor Productivity* (right) * Nonfarm business sector, 5-year average. Source: Bryne, Oliner, and Sichel (2013). Department of Labor. Goldman Sachs Global Investment Research. But now the paradox is back. The IT contribution has fallen back to levels last seen in the early 1980s, and measured productivity growth has averaged 1½% over the past decade (and just ½% over the past three years). For analysts who had held onto the view that at least part of the weakness was a cyclical and temporary consequence of the deep slump in output and capital accumulation, such as ourselves, the recent numbers have been a clear disappointment. On the back of this disappointment, we have reduced our working estimate for long-term productivity growth from 2% to 1½%, which takes us back to the average pace of the period. Exhibit 2 shows the components of our estimate in a standard growth accounting framework: (1) the contribution from growth in capital services, (2) the contribution from changes in labor composition, and (3) growth in total factor productivity (i.e. the residual component). 2 Based on the outlook for capital spending, labor force growth, educational attainment, and other demographic changes, we still expect contributions from capital services and labor quality that are very similar to our prior estimates. Over the coming ten years, we expect capital services to 1 The IT contribution series is derived in David Byrne, Stephen Oliner, and Daniel Sichel, Is the Information Technology Revolution Over? Federal Reserve Board FEDS Working Paper , March We thank them for providing us with a year-by-year version of their series. 2 For details see Kris Dawsey, Productivity: Down But Not Out, US Economics Analyst, January 4, 全球投资研究 2

3 contribute about 0.9 percentage point per year to productivity growth, and labor composition to add another 0.1pp. Exhibit 2: Back to a Sluggish Regime for Measured Productivity Growth Percentage points, annual rate Percentage points, annual rate Contribution to Labor Productivity Growth, Nonfarm Business Sector 3.0 Total Factor Productivity Capital Services 2.5 Labor Quality GS Projection 0.0 Source: Labor Department. Goldman Sachs Global Investment Research. However, our prior estimate for growth in total factor productivity (TFP) now looks too high. As the strong-productivity period before 2004 fades further into the background and new data for 2013 and 2014 have become available, our estimate of the TFP trend has fallen back to the ½% trend seen from 1974 to Adding up the components, this implies a reduction in our working assumption for measured trend productivity growth to 1½%. Not Really Feeling It But is the weakness for real? We find the notion of a major productivity slowdown difficult to square with several aspects of the economic and financial environment of the past decade. As shown in Exhibit 3, the productivity slowdown of the 1970s featured declining profit margins, rising inflation, and declining equity valuations. In the past decade, by contrast, profit margins have hit all-time highs, inflation has remained very low, and equity valuations have surged. And to cap it all, tech stocks have outperformed the overall market and are trading at their highest relative levels ever, except for the late-1990s technology bubble. Admittedly, these observations are not conclusive. The impact of genuinely weaker productivity growth on profits, inflation, and equity valuations could have been overwhelmed by the weakness in aggregate demand after 2008, which held down wage and price inflation, kept monetary policy easy, and thus may have pushed up asset prices on net. And the outperformance of tech stocks could be due to a new bubble. But at a minimum it seems safe to say that neither the overall economy nor the technology sector is behaving as if it had been hit by a major IT-centered productivity slowdown. It s Hard to Measure a Digital Economy One possible way to reconcile the weakness of measured productivity growth with the friendly profits, inflation, and stock market picture is that structural changes in the US economy might have resulted in a growing statistical understatement of real GDP growth. 全球投资研究 3

4 Exhibit 3: It Doesn t Feel Like an IT-Centered Productivity Slowdown Percent 14 Corporate Profits/GDP Percent Percent change, year ago Percent change, year ago Core PCE Inflation Avg Avg Avg Ratio 2.5 Equity Market Capitalization/GDP Ratio Ratio NASDAQ Composite/S&P Ratio Source: Federal Reserve Board. NASDAQ. Department of Commerce. Department of Labor. Goldman Sachs Global Investment Research. 0.0 There are several possible reasons for such an understatement, but they generally fall into one of two categories, namely an understatement of nominal GDP because some transactions go unrecorded or an overstatement of the GDP deflator because of biases in price measurement. In today s discussion, we focus on the price issue, which is generally thought to be the more important of the two. 3 The importance and difficulty of correctly measuring prices is best explained if we focus on consumption, although the logic applies equally to other parts of the GDP deflator. A correctly measured consumer price index shows the dollar cost of attaining a given level of utility the total benefit that consumers obtain from their basket of purchases at different points in time. When both the consumption basket and the characteristics of the typical product are stable over time, inflation is easily calculated as a weighted average of all raw price changes. But when new goods are introduced, existing ones change in quality, or consumers substitute between different products and outlets, potential biases creep in. 3 However, there are also some important ways in which nominal GDP might be understated. For example, the shift toward online retailing just like any rapid structural change in the market shares of different retailers might have led to some understatement of nominal consumption and GDP because the Census Bureau only updates the weights for different outlets in the retail sales survey every few years. The shift from licensed taxicabs to Uber and comparable firms might have similar implications. 全球投资研究 4

5 The best-known study of these biases is the 1996 report by the so-called Boskin Commission. 4 It estimated that the CPI as of 1996 overstated the true inflation rate by 1.1 percentage points per year. About half of this distortion was due to substitution bias, i.e. a failure to consider shifts in the consumption basket away from products or retail outlets that have risen in price and toward ones that have fallen in price. The other half was due to a failure to capture the implications of quality improvements and new products for the cost of living. Since 1996, methodological changes in the CPI have reduced the substitution bias by about 0.3 percentage points per year. But there are reasons to believe that both the quality change bias and the new goods bias have increased significantly because of structural changes in the economy, especially the growing importance of software and online content. 5 And unlike the substitution bias, these biases fully feed through to the deflator and thus to measured real GDP Quality Change Bias One reason why the IT revolution of the 1990s showed up so clearly in the productivity numbers is that the statisticians devised ways of translating the increases in computer performance faster processors, more memory, better graphics, and the like into rapid quality-adjusted price declines. Exhibit 4 shows that the measured price of computer hardware has plunged by 9% since 1995, with most of the decline occurring in the first half of that period. 7 Since real output is equal to nominal output divided by the price level, this meant a sharp increase in the measured contribution of computer hardware to real GDP growth. 4 See Michael Boskin, Ellen Dulberger, Robert Gordon, Zvi Griliches, and Dale Jorgenson, Toward a More Accurate Measure of the Cost of Living, Final Report to the Senate Finance Committee, December 4, 1996, as well as the retrospective by member Robert Gordon, The Boskin Commission Report: A Retrospective One Decade Later, NBER Working Paper No , June Potential price measurement problems are not confined to the technology sector. The healthcare industry is another area where quality is changing quickly and measuring prices is difficult. If true healthcare prices are growing more slowly than measured prices, this will also result in an understatement of real GDP growth, and potentially a growing understatement given the increasing GDP share of healthcare. 6 The substitution bias does not fully feed through to real GDP because the PCE and GDP deflators are based on CPI and PPI inputs but are then chain-weighted, i.e., calculated in a way that allows for substitution between different products. 7 More recently, the measured price decline for semiconductors and computers has slowed sharply, and there is now a debate whether an upward bias has recently crept into these price indices. This would reinforce the notion that the official statistics currently understate the IT contribution to productivity growth. See David Byrne, Stephen Oliner, and Daniel Sichel, How Fast Are Semiconductor Prices Falling? NBER Working Paper No , April 全球投资研究 5

6 Exhibit 4: Measured Prices Have Fallen Much Faster in Hardware than in Software Index (1995=100) Index (1995=100) GDP Price Index, Private Investment in: Software Computers & Peripherals Source: Department of Commerce. 0 But Exhibit 4 also illustrates that the statisticians have not found a way to capture the improvements in software and digital content in a similar manner, with measured software prices only edging down slightly over the past two decades. This is not surprising because the performance of software and digital content is inherently a more amorphous and subjective concept. How much better are the inventory management systems that retail companies contract out or develop for their own account compared with those of twenty years ago? How much better is Grand Theft Auto V than Grand Theft Auto IV? And how much more value do we now derive from our internet connection compared with a decade ago? 8 It is very difficult for a statistician to know, and when we do not know our default assumption tends to be that there is little change. This measurement asymmetry is important because the IT sector s center of gravity has shifted away from hardware to software and digital content. Exhibit 5 shows that these products now make up more than half of the output and market capitalization of the sector, a shift that has come at the expense of computer hardware, communication equipment, and semiconductors. In fact, the chart probably understates the full extent of the shift because the output of Apple by far the biggest technology company is included in hardware even though the company sells a combination of hardware and software. 8 Currently, the price index for internet access is not even adjusted for the speed of the connection. More broadly, there are now a lot more sites worth visiting than ten or twenty years ago, so the quality of the experience has improved immeasurably. But it is hard to see how this latter issue will ever be captured in a price index. 全球投资研究 6

7 Exhibit 5: Software and Digital Content Have Taken Over Percent of total technology spending 70 Percent of total technology equity market capitalization Software and Comm. Services Spending, NIPA (left) Software and Services Equity Market Capitalization (right) Source: Department of Commerce. Goldman Sachs Global Investment Research If part of the gap between the price declines in hardware vs. software is a statistical mirage, then an increase in the relative weight of software and digital content will result in a spurious slowdown in real GDP growth. How large might the bias be? It is difficult to say, but we would note that the software and digital content sectors account for almost 4% of nominal GDP and measured prices are approximately flat. If we assume, purely as an illustration of the potential magnitude of any biases in this area, that true prices were falling at the 20-year average pace of the computer hardware industry of about 5% per year, this would imply a statistical understatement of the growth contribution from software and digital content of about 0.2pp per year. 2. New Products Bias The new products bias is even harder to address. How does one calculate the impact of a product that did not exist last year on the overall amount of consumer surplus and thus the correctly measured cost of living? A practical answer is difficult to give, and some amount of new products bias in the price statistics is probably unavoidable. But the extent of this bias may well have increased substantially in recent years, partly because of the increased availability of free or nearly-free digital products. We can illustrate the basic issue via a simple diagram such as Exhibit 6. The left-hand side shows the demand curve for a traditional product. Immediately after its introduction, the price of the traditional product is p 1. At this price and the resulting sales, the buyers realize a consumer surplus i.e., a difference between the price they pay and the value they derive from the purchase that is equal to area A 1. Since a conventional price index can only incorporate new products after they have been introduced, this consumer surplus is unmeasured and introduces an upward bias into the price index and a downward bias into real GDP. However, this is a relatively small issue because the first recorded price p 1 is high and the initial sales volume correspondingly low. In subsequent periods, prices fall from p 1 to p 2, sales increase accordingly, and consumer surplus increases by A 2. This price decline is recorded in the price and GDP statistics. 全球投资研究 7

8 Exhibit 6: Free Digital Products Create a Large Amount of Unmeasured Consumer Surplus Price (P) Price (P) A 1 P 1 A 2 P 2 A 1 * P 1 * Traditional Product Quantity (Q) "Free" Digital Product Quantity (Q) Source: Goldman Sachs Global Investment Research. Now turn to the right-hand side of Exhibit 6, which illustrates the demand curve for free digital content. 9 Immediately after its introduction, the price of the digital product drops to p 1*, a number that is equal to or close to zero. At this price, demand for the product is very high and the resulting consumer surplus is equal to area A 1*. The difference compared with the left-hand side chart is that the unmeasured surplus and the resulting bias relative to a correctly measured costof-living index are now much bigger. How important is this issue for GDP measurement? Hard evidence is again hard to come by, but Erik Brynjolfsson and Joo Hee Oh estimated in 2012 that the growth in time spent on free internet sites between 2007 and 2011 created incremental consumer surplus worth ¾% of GDP each year. 10 Although the uncertainty around these types of estimates is obviously very large, as the authors acknowledge, we do think it is plausible that increased new products bias could be holding down the measured real GDP growth rate by a meaningful amount. Take It Easy Our discussion of the potential biases in the GDP statistics resulting from the growth of software and digital content has only scratched the surface, and a numerical estimate for the size of these biases would require a much more comprehensive analysis. But when we look broadly at 1) the sharp measured productivity slowdown, 2) the much friendlier message sent by the broad economic and market environment, and 3) the potential ways in which the official statistics might be missing real output growth, we walk away persuaded by the notion that productivity mismeasurement could be a significant issue. Pending a more comprehensive analysis, we see three practical implications from our analysis today. First, we would be skeptical of confident pronouncements that the standard of living is growing much more slowly than in the past. Second, given the uncertainty around GDP it is better to focus on other indicators especially employment to gauge the cumulative progress of 9 The story does not change materially if we allow for the fact that these free products are typically financed by exposing consumers to advertising, partly because advertising is treated as an intermediate input and thus does not show up directly in GDP. 10 See Erik Brynjolfsson and Joo Hee Oh (2012), The Attention Economy: Measuring the Value of Free Digital Services on the Internet, Thirty-Third International Conference on Information Systems, Orlando, Their approach is based on the earlier study by Austan Goolsbee and Peter Klenow, Valuing Consumer Products by the Time Spent Using Them: An Application to the Internet, American Economic Review Papers and Proceedings, Vol. 96, No. 2, May 2006, pp 全球投资研究 8

9 the recovery and the remaining amount of slack. And third, if true inflation is even lower than measured inflation and especially if this gap is bigger than it has been historically the case for keeping monetary policy accommodative strengthens further. At the margin, the analysis therefore reinforces our view that it would be better for Fed officials to delay monetary tightening until Jan Hatzius Kris Dawsey 全球投资研究 9

10 The US Economic and Financial Outlook (% change on previous period, annualized, except where noted) (f) (f) (f) (f) Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 OUTPUT AND SPENDING Real GDP Consumer Expenditure Residential Fixed Investment Business Fixed Investment Structures Equipment Intellectual Property Products Federal Government State and Local Government Net Exports ($bn, '09) Inventory Investment ($bn, '09) Industrial Production, Mfg HOUSING MARKET Housing Starts (units, thous) 1,001 1,097 1,323 1,447 1, ,081 1,128 1,203 1,254 1,307 1,349 1,383 New Home Sales (units, thous) Existing Home Sales (units, thous) 4,920 5,037 5,280 5,346 5,218 4,967 5,000 5,045 5,136 5,208 5,232 5,298 5,381 Case-Shiller Home Prices (%yoy)* INFLATION (% ch, yr/yr) Consumer Price Index (CPI) Core CPI Core PCE** LABOR MARKET Unemployment Rate (%) GOVERNMENT FINANCE Federal Budget (FY, $ bn) FINANCIAL INDICATORS FF Target Range (Bottom-Top, %)^ Year Note^ Euro ( /$)^ Yen ($/ )^ * Weighted avg of metro-level HPIs for 366 metro cities where the weights are dollar values of single-family housing stock reported in the 2000 Census. ** PCE = Personal consumption expenditures. ^ Denotes end of period NOTE: Published figures are in bold. Source: Goldman Sachs Global Investment Research. 全球投资研究 10

11 Economic Releases and Other Events Time Estimate Date (EDT) Indicator GS Consensus Last Report Tue May 26 8:30 Durable Goods Orders (Apr) +% -% +4.7% 8:30 Durable Goods Orders Ex-Transport (Apr) Flat +0.3% +0.3% 8:30 Core Capital Goods Orders (Apr) Flat +0.3% +0.6% 8:30 Core Capital Goods Shipments (Apr) Flat +0.3% +0.9% 9:00 S&P/Case Shiller Home Price Index (Mar) +1.2% +0.9% +0.93% 9:00 FHFA House Price Index (Mar) n.a. +0.7% +0.7% 9:45 Markit PMI Prel (May) n.a. n.a :00 New Home Sales (Apr) +9.0% +5.0% -11.4% 10:00 Consumer Confidence (May) :00 Richmond Fed Survey (May) n.a. 0-3 Thu May 28 8:30 Initial Jobless Claims n.a. 270, ,000 8:30 Continuing Claims n.a. 2,200,000 2,211,000 10:00 Pending Home Sales (Apr) n.a. +0.8% +1.1% 17:00 GS Analyst Index (May) n.a. n.a Fri May 29 8:30 Real GDP Q1 Annualized (Second) -1.1% -0.9% +0.2% 8:30 Personal Consumption (Q1) +% +% +1.9% 9:45 Chicago Purchasing Managers Index (May) :00 UMich Consumer Sentiment Final (May) Source: Goldman Sachs Global Investment Research. 全球投资研究 11

12 信息披露附录 申明 我们,Jan Hatzius Alec Phillips Jari Stehn Kris Dawsey David Mericle Chris Mischaikow Karen Reichgott, 在此申明, 本报告所表述的所有观点准确反映了我们的个人看法, 没有受到公司业务或客户关系因素的影响 本报告首页所列作者为高盛全球投资研究部分析师, 除非另有说明 高盛信息披露 全球产品 ; 分发机构 高盛全球投资研究部在全球范围内为高盛的客户制作并分发研究产品 高盛分布在其全球各办事处的分析师提供行业和公司的股票研究, 以及宏观经济 货币 商品及投资组合策略的研究 本研究报告在澳大利亚由 Goldman Sachs Australia Pty Ltd(ABN ) 分发 ; 在巴西由 Goldman Sachs do Brasil Corretora de Títulos e Valores Mobiliários S.A. 分发 ; 在加拿大由 Goldman Sachs Canada Inc 或分发 ; 在香港由高盛 ( 亚洲 ) 有限责任公司分发 ; 在印度由高盛 ( 印度 ) 证券私人有限公司分发 ; 在日本由高盛证券株式会社分发 ; 在韩国由高盛 ( 亚洲 ) 有限责任公司首尔分公司分发 ; 在新西兰由 Goldman Sachs New Zealand Limited 分发 ; 在俄罗斯由高盛 OOO 分发 ; 在新加坡由高盛 ( 新加坡 ) 私人公司 ( 公司号 : W) 分发 ; 在美国由分发 高盛国际已批准本研究报告在英国和欧盟分发 欧盟 : 高盛国际 ( 由审慎监管局授权并接受金融市场行为监管局和审慎监管局的监管 ) 已批准本研究报告在英国和欧盟分发 ;Goldman Sachs AG 和 Goldman Sachs International Zweigniederlassung Frankfurt( 由联邦金融监管局监管 ) 可能也会在德国分发 一般性披露 本研究报告仅供我们的客户使用 除了与高盛相关的披露, 本研究报告是基于我们认为可靠的目前已公开的信息, 但我们不保证该信息的准确性和完整性, 客户也不应该依赖该信息是准确和完整的 我们会适时地更新我们的研究, 但各种规定可能会阻止我们这样做 除了一些定期出版的行业报告之外, 绝大多数报告是在分析师认为适当的时候不定期地出版 高盛是一家集投资银行 投资管理和证券经纪业务于一身的全球性综合服务公司 高盛全球投资研究部所研究的大部分公司与我们保持着投资银行业务和其它业务关系 美国证券经纪交易商高盛是 SIPC 的成员 ( 我们的销售人员 交易员和其它专业人员可能会向我们的客户及我们的自营交易部提供与本研究报告中的观点截然相反的口头或书面市场评论或交易策略 我们的资产管理部门 自营交易部和投资业务部可能会做出与本报告的建议或表达的意见不一致的投资决策 本报告中署名的分析师可能已经与包括高盛销售人员和交易员在内的我们的客户讨论, 或在本报告中讨论交易策略, 其中提及可能会对本报告讨论的证券市场价格产生短期影响的推动因素或事件, 该影响在方向上可能与分析师发布的股票目标价格相反 任何此类交易策略都区别于且不影响分析师对于该股的基本评级, 此类评级反映了某只股票相对于报告中描述的研究范围内股票的回报潜力 我们以及我们的关联机构 高级职员 董事和雇员, 不包括股票分析师和信贷分析师, 将不时地对本研究报告所涉及的证券或衍生工具持有多头或空头头寸, 担任上述证券或衍生工具的交易对手, 或买卖上述证券或衍生工具 在高盛组织的会议上的第三方演讲嘉宾 ( 包括高盛其它部门人员 ) 的观点不一定反映全球投资研究部的观点, 也并非高盛的正式观点 在此提到的任何第三方, 包括销售人员 交易员和其它专业人士或其家庭成员, 可能持有本报告提及的且与本报告分析师所表达的观点不一致的产品头寸 在任何要约出售股票或征求购买股票要约的行为为非法的司法管辖区内, 本报告不构成该等出售要约或征求购买要约 本报告不构成个人投资建议, 也没有考虑到个别客户特殊的投资目标 财务状况或需求 客户应考虑本报告中的任何意见或建议是否符合其特定状况, 以及 ( 若有必要 ) 寻求专家的意见, 包括税务意见 本报告中提及的投资价格和价值以及这些投资带来的收入可能会波动 过去的表现并不代表未来的表现, 未来的回报也无法保证, 投资者可能会损失本金 外汇汇率波动有可能对某些投资的价值或价格或来自这一投资的收入产生不良影响 某些交易, 包括牵涉期货 期权和其它衍生工具的交易, 有很大的风险, 因此并不适合所有投资者 投资者可以向高盛销售代表取得或通过 取得当前的期权披露文件 对于包含多重期权买卖的期权策略结构产品, 例如, 期权差价结构产品, 其交易成本可能较高 与交易相关的文件将根据要求提供 所有研究报告均以电子出版物的形式刊登在我们的内部客户网上并向所有客户同步提供 并非所有研究内容都转发给我们的客户或者向第三方整合者提供, 高盛也并不对由第三方整合者转发的我们研究报告承担任何责任 有关某特定证券的研究报告 模型或其它数据, 请联络您的销售代表或登陆 披露信息可以查阅 或向研究合规部索取, 地址是 200 West Street,New York,NY 高盛版权所有 2015 年未经公司事先书面同意, 本材料的任何部分均不得 (i) 以任何方式制作任何形式的拷贝 复印件或复制品, 或 (ii) 再次分发 高华证券信息披露 一般披露 本报告在中国由高华证券分发 高华证券具备证券投资咨询业务资格 本研究报告仅供我们的客户使用 本研究报告是基于我们认为可靠的目前已公开的信息, 但我们不保证该信息的准确性和完整性, 客户也不应该依赖该信息是准确和完整的 我们会适时地更新我们的研究, 但各种规定可能会阻止我们这样做 除了一些定期出版的行业报告之外, 绝大多数报告是在分析师认为适当的时候不定期地出版 高盛高华为高华证券的关联机构, 从事投资银行业务 高华证券 高盛高华及它们的关联机构与本报告中涉及的大部分公司保持着投资银行业务和其它业务关系 我们的销售人员 交易员和其它专业人员可能会向我们的客户及我们的自营交易部提供与本研究报告中的观点截然相反的口头或书面市场评论或交易策略 我们的自营交易部和投资业务部可能会做出与本报告的建议或表达的意见不一致的投资决策 全球投资研究 12

13 本报告中署名的分析师可能已经与包括高华证券销售人员和交易员在内的我们的客户讨论, 或在本报告中讨论交易策略, 其中提及可能会对本报告讨论的证券市场价格产生短期影响的推动因素或事件, 该影响在方向上可能与分析师发布的股票目标价格相反 任何此类交易策略都区别于且不影响分析师对于该股的基本评级, 此类评级反映了某只股票相对于报告中描述的研究范围内股票的回报潜力 高华证券及其关联机构 高级职员 董事和雇员, 不包括股票分析师和信贷分析师, 将不时地对本研究报告所涉及的证券或衍生工具持有多头或空头头寸, 担任上述证券或衍生工具的交易对手, 或买卖上述证券或衍生工具 在高盛组织的会议上的第三方演讲嘉宾 ( 包括高华证券或高盛其它部门人员 ) 的观点不一定反映全球投资研究部的观点, 也并非高华证券或高盛的正式观点 在任何要约出售股票或征求购买股票要约的行为为非法的地区, 本报告不构成该等出售要约或征求购买要约 本报告不构成个人投资建议, 也没有考虑到个别客户特殊的投资目标 财务状况或需求 客户应考虑本报告中的任何意见或建议是否符合其特定状况, 以及 ( 若有必要 ) 寻求专家的意见, 包括税务意见 本报告中提及的投资价格和价值以及这些投资带来的收入可能会波动 过去的表现并不代表未来的表现, 未来的回报也无法保证, 投资者可能会损失本金 某些交易, 包括牵涉期货 期权和其它衍生工具的交易, 有很大的风险, 因此并不适合所有投资者 外汇汇率波动有可能对某些投资的价值或价格或来自这一投资的收入产生不良影响 投资者可以向高华销售代表取得或通过 取得当前的期权披露文件 对于包含多重期权买卖的期权策略结构产品, 例如, 期权差价结构产品, 其交易成本可能较高 与交易相关的文件将根据要求提供 所有研究报告均以电子出版物的形式刊登在高华客户网上并向所有客户同步提供 高华未授权任何第三方整合者转发其研究报告 有关某特定证券的研究报告 模型或其它数据, 请联络您的销售代表 北京高华证券有限责任公司版权所有 2015 年未经北京高华证券有限责任公司事先书面同意, 本材料的任何部分均不得 (i) 以任何方式制作任何形式的拷贝 复印件或复制品, 或 (ii) 再次分发 全球投资研究 13