行业资讯

加入亿拓客·流量大师 撬动财富之门!!!

解读高盛研报:何种因素制约了中国的企业利润?又该去何处寻找增长机遇?

wang 2026-11-13 行业资讯
解读高盛研报:何种因素制约了中国的企业利润?又该去何处寻找增长机遇?

文章分析

反常识观点与高信息熵内容

Goldman Sachs Global Strategy Paper No. 80
《What's ailing Chinese profits, and where to find growth?》

作者:Kinger Lau / Timothy Moe / Si Fu / Kevin Wang

发布:2026年11月11日 · 分析日期:2026-11-12

一、总体结论

这篇高盛全球策略长文(系列第80篇)试图回答一个问题:25年新增5万亿美元GDP之后,为什么中国股票的回报还是没能跟上宏观。作者的核心答案是:市场熟知的"低通胀、产能过剩、内卷"只是表象,更深处是政策、社会与经济因素共同把中国企业的资本成本(COC/WACC)压到了全球低位,由此改变了企业的最优资本开支与利润最大化行为——融资成本下降降低投资门槛,让"资本开支仍强劲、ROE/ROIC/产能利用率却持续走低"的组合得以持续,即所谓"利润无增长式增长"(profitless growth)。

基于此,高盛把中国公司的"正常利润"锚定为约12%的稳态ROE + 未来5年10%的趋势EPS增速,对应指数10-15%的重估空间;并用一个把6000+公司、47个GICS三级行业聚成五个成长桶的框架给出配置路线图:核心配置AI与"出海",看多稳定/现金回报(国企+分红),在"降级成长"与"周期成长"中战术性寻找盈利拐点与自由现金流,回避"压抑成长"直到房价2028/2029年企稳。

全文最反常识的两点:

① 把"资本成本过低"当作利润侵蚀的隐藏根源
低融资成本不是解药,而是压低投资门槛、延长过度投资的机制。

② "避开增长陷阱"比"拥抱成长"更重要
剔除"压抑成长"(Depressed Growth)后,五个成长桶过去4年的平均已实现EPS增速从12%跃升到35%。

二、反常识观点

01 中国利润问题的深层根源是"资本成本太低",而非低通胀/产能过剩/内卷本身——低COC是利润侵蚀的隐藏机制

原文支撑

"Low inflation, abundant capacity, and intense domestic competition are well-discussed and understood drivers for subdued profit growth and profitability. However, in our view, they may stem from and reflect deeper economic, social, and political reasons that have pressured the cost of capital for Chinese corporates, and influenced their capital budgeting decisions and profit-maximization behaviors."

"China has probably one of the lowest COCs among key markets globally at present, predominately helped by the declining benchmark rates and corporate funding cost domestically, particularly at the time when global bond yields are rising and credit spreads are beginning to widen."

02 资本成本下降不是利好,而是"低回报下继续扩张"的机制:投资门槛降低→资本开支保持强劲→ROE、ROIC、产能利用率同步下行

原文支撑

"A falling COC also implies that Chinese companies' hurdle rate for investment projects is decreasing, consistent with the still-robust capex growth and intensity but the downtrend in ROE, ROIC, and capacity utilization in recent years."(Exhibit 8:2006-2026年 WACC、ROIC、ROE与资本开支强度四线)

03 政府补贴与隐性支持压低资本成本→资源错配:中国工业企业拿到的补贴约为海外竞争者的数倍

原文支撑

"Government subsidies, financial incentives, and other implicit/indirect government support measures could effectively lower the cost of capital for companies (e.g. SOEs) that are seen as strategically, economically, or socially important... some of which might lead to resource mis-allocation and other non-profit-driven behaviors.";OECD研究显示中国工业企业"补贴/年营收"最高接近2.5%(含政府补助、所得税减免、低于市场利率的借款),而OECD亚太与欧洲仅约0.3%

04 监管常态化会把民企利润"公用事业化":企业的理性应对是放弃利润最大化、转向份额扩张——这本身是"内卷"的制度成因

原文支撑

"In a hypothetical situation where POEs' normal profit is regulated as a public utility or subject to a theoretical cap, companies' optimal decision could shift from profit maximization to volume/market share expansion, potentially driving over-investment, homogeneous competition, and other 'involuted' behaviors, all else being equal."

05 供给侧刺激偏好是一种"文化基因":用制造业产出、有形资产、贸易量衡量经济实力,导致政策系统性偏向供给端,挤占消费与福利的财政资源

原文支撑

"Culturally, the strength of the Chinese economy has often been measured by its manufacturing output, scale of tangible assets, and trade volume, possibly explaining the entrenched macro and industrial policy bias towards the supply side."

"likely at the expense of fiscal budgets and resources for consumption and other social welfare programs."

06 房地产下行不是"已出清"的旧周期,而是仍在运转的"通缩机制":需求损失持续压制利率、通胀、产能利用率与利润;且GS预计房价要到2028/2029年才企稳

原文支撑

"the cumulative loss of aggregate demand and the resulting negative output gaps are still weighing on interest rates, inflation, capacity utilization, and profits, especially when both the stock and flow of system-wide supply stays abundant."

"We would remain cautious and stay Underweight housing-associated equities until property prices stabilize, possibly in 2028/2029 on GSe."

07 AI普及对中国在中期是"通缩性"的(劳动力替代+成本/价格下行),不是单纯的增长引擎

原文支撑

"Technological innovation is in most cases productivity-enhancing, but could be dis-inflationary via the channels of labor displacement and falling product costs/prices, particularly in China where manufacturing and transportation infrastructure is well-built and frictional cost is low."(Exhibit 17:AI对中国总就业的净冲击在2020年代末见底,其后回补并在2030年代初转正)

08 "避开增长陷阱"比"拥抱成长"更重要:剔除"压抑成长"后,五个成长桶过去4年平均已实现EPS增速从12%提高到35%

原文支撑

"The 5 buckets have averaged a realized EPS growth of 12% over the past 4 years; but stripping out Depressed Growth, the average growth rate would have improved to 35%. This means that avoiding out-of-favor, ex-growth stocks has been equally, if not more, important than simply embracing growth opportunities to stay exposed to the Growth factor."

09 "结构性成长"没那么结构性:5年前被互联网/消费主导,如今被硬件科技与HALO资产取代——中国公司(尤其民企)的护城河可能不如想象的持久

原文支撑

"5 years ago, our Structural Growth bucket was dominated by Internet and Consumer-related stocks but now consists mainly of Hardware Tech industries and HALO assets. Not only are the shifts reflecting the rise of AI but are also indicating that competitive moats of Chinese companies, POEs in particular, could be less durable than would otherwise be the case given potential technology disruptions and regulatory uncertainty."

10 成长股在PEG上比衰退股还便宜:Structural Growth 0.6x(-2.0个标准差),低于Depressed(1.0x)和Stable(1.3x)——"大量悲观已被价格消化"

原文支撑

"Investor concerns about the durability and ROI of global AI capex are valid, but with Structural Growth trading at meaningful discounts to Cyclical and Depressed Growth in PEG terms, this leads us to believe that a great deal of pessimism is already well-discounted in equity prices."

"we'd take the weakness to build strategic positions on select Structural/AI Growth stocks, notably our Chinese AI Export Leaders, Domestic Champions, and the more diversified Select AI Portfolio."

11 过去两年中国股票中风险调整后回报最好的是"稳定增长"(国企+高分红),而非成长:滚动2年夏普Stable 1.3 vs Structural 0.7

原文支撑

"Stable Growth has continued to generate superior Sharpe ratios among key themes in the equity universe since the once-every-decade Nine Measures document... was unveiled in April 2022, and has remained a critical component to our barbell strategy in a difficult beta environment."(Exhibit 36末值:Stable 1.3、Structural 0.7、Cyclical 0.5、Depressed 0.2、De-rated 0.0;该组2026E共识股息率6%;原文并称美国(上行)与中国(下行)利率反向、美中利差维持高位时该组更占优,对应其近期将中国银行上调至Overweight)

12 "政府干预经济"与"企业ROE"在全球样本中没有稳定关系(自由度检验"inconclusive");中国的稳态趋势ROE或在13%——干预并不必然压低回报

原文支撑

"we note an inconclusive pattern between jurisdictions with low rankings (i.e. high State influence in the economy) and the corresponding corporate ROEs in the listed universe. This could be attributable to high barriers of entry, tight industry regulation, protectionism, and monopolistic/oligopolistic abnormal profit in jurisdictions which embrace State or crony capitalism, in our view."

"Assuming China's IEF (relative) ranking doesn't change materially from now, the trend ROE for China could stay at around 13% in the foreseeable future."

三、高信息熵内容

01 一组"连年失望"的硬数据:上市公司ROE连续6年下降(伴随杠杆上升、资产周转停滞);指数EPS连续9年低于年初一致预期;过去20年净发行(增发-回购)年均占市值0.8%,累计稀释约15%的EPS

原文支撑

"listed companies' ROE has been falling for 6-consecutive years alongside rising leverage and stagnant asset turnover ratios, realized index EPS has underperformed the start-of-year consensus expectations for 9 years in a row, and total equity issuance (follow-on) has outpaced stock repurchases for the most part in the past 2 decades, implying around 15% of cumulative EPS dilution for shareholders."

收入/盈利与名义GDP的增速也在10年间同步降档:2006-16年CAGR为盈利15%/收入15%/名义GDP 13%,2017-26E降为7%/8%/6%;总市值自2002年的6000亿美元升至如今的约20万亿美元(仅次于美国),上市公司数量从1,600家增至7,000家(A+H+ADR)

02 2020-2026全球指数回报分解:中国累计约-22%(PE贡献为负),对比SPX +137%、EM ex-CN +75%、日本+63%、EU +53%、APJ +47%;中国指数PE从周期峰值的17x降至9.7x

"After reaching their historical peak in early 2021, Chinese equities have fallen 45% since then (a 63% peak-to-trough drawdown), trailing DM and EM (ex-China) equity blocs by 110pp and 80pp respectively. Correspondingly, index PE has de-rated from 17x at the cycle peak 6 years ago to 9.7x at the moment."

(Exhibit 6:2020-2026年 SPX/Japan/EU/China/APJ/EM ex-CN的PE与EPS分解)

03 地产链贡献的坍缩幅度:GDP占比29%→16%、盈利占比15%→1%、收入占比12%→5%(2020 vs 2025);2021年峰顶住房名义价值超60万亿美元、"可能是当时全球最大的资产类别"

"the housing sector contributed to about 30% of GDP directly and through its adjacent value chains, and boasted over US$60tn in notional value, perhaps the world's largest asset class back then... 12% of equity market revenues and 15% of earnings were exposed to the extended housing value chain in 2020."

"Aggregate net profit for Depressed Growth has plunged from Rmb1.1tn in 2020 to Rmb150bn last year."(较峰值降幅超70%)

04 过剩产能的量化全景(中国产能/全球需求,2026-28E)

太阳能组件
167%
锂电池
122%
空调
102%
工程机械
95%
电动车
88%
钢铁
73%
功率半导体(IGBT)
49%

工业产能利用率总水平73%(2021年:77),其中汽车66%为全行业最低、医药69%、采矿70%(原文支撑:Exhibit 14、Exhibit 16)

05 中国AI采用的三个量化锚:中国大模型产出约占全球AI token消费的70%;电商已占零售额30%以上;多数中国消费者对AI持正面或接受态度

"e-commerce already representing more than 30% of retail sales, a majority of Chinese consumers taking a favorable or receptive view on AI, and Chinese LLMs producing around 70% of the world's AI token consumption."(中国玩家token市占率从2025年4月的17%升至2026年8月的72%)

06 "正常利润"的四种独立估算殊途同归:约12%的稳态ROE

(a)政府在经济中的角色(按经济自由度分组):13%

(b)全球国企-民企盈利能力差距外推(非中国上市国企317家,其中EM 223家、DM 94家):13%

(c)成本加成经验模型(MSCI China 576股、含244家国企;国企10年净利率10.7% vs 民企11.4%):10-11%

(d)社会主义经济下的资本回报(皮凯蒂 r=g 逻辑:中国未来5年潜在名义GDP增速6%、债务成本3%,要求股权回报至少11%)

"we conclude that Chinese corporates (SOEs and POEs combined) could generate about 12% ROE in our envisaged steady state."

07 三个可以分头验证的估值上行口径:按PB-ROE历史关系→约10%重估空间;若派息率与回购延续升势→估值+15%;按DDM,理论化PE为10.7x→仍有8%空间

"First, the gaps between our estimated normal ROE and the ex-post market COE would imply about 10% of potential re-rating upside for MSCI China... Second, in a scenario where the aggregate payout ratios and share buybacks extend their rising trends... we see index valuations appreciating about 15% from current levels... we estimate that the theoretical, intrinsic-value-derived P/E for MSCI China would be at around 10.7x, implying 8% potential fair value upside."

敏感性:ERP从6.5%降到6.0%,重估空间从+8%跳至+24%

08 中国资本成本(WACC)的完整底稿:MSCI China 5.4%、CSI300 3.9%、All China 4.6%;同口径下中国MSCI WACC从2020年的7.5%降至5.4%,美国从4.9%升至6.0%

MSCI China无风险利率=80%美国主权收益率+20%中国主权收益率,ERP 4.3%,Ke=8.3%;CSI300以中国政府债为无风险利率、ERP 3.8%、Ke=5.5%;中国整体Ke=6.3%(MSCI China与CSI300按1:2.5市值加权)

All China资本结构D/E 0.63x、债务权重38.7%、权益权重61.3%;MSCI China WACC 5.4%、CSI300 WACC 3.9%、All China WACC 4.6%(附录2;Exhibit 7)

09 五个成长桶的完整量化画像(6000+公司、47个GICS三级行业)

市值占比 36% / 15% / 15% / 13% / 22%

1Q25-2Q26中位EPS增速 28% / (1%) / 21% / (29%) / 2%

26-28E CAGR 30% / 25% / 8% / 20% / 7%

Exhibit 31(G1 Structural含电子元件、半导体设备等8个行业;G4 Depressed含化工、开发商、建筑、软件等11个行业;G5 Stable含银行、医药、无线电信、食品、纺服零售等9个行业);分桶方法:47个GICS三级行业(自73个行业筛出,标准为总市值>1000亿美元或成分公司≥50家),对收入增速与净利率分别计算水平、动量、波动性六项z分数并按+1/0/-1加权,叠加行业间月回报相似度(2015年以来)调整;数据自1Q2006至2Q2026,观测值按近重远轻加权(10年前观测权重约为当前的1/4)

10 盈利分化与"稀缺溢价"的量化:顶部与底部四分位个股的EPS增速差从2011-15年的平均66pp扩大到2021-25年的90pp;实现正EPS增长的公司占比从10年前的约60%收缩到约40%

"Average realized EPS growth differential between the top- and bottom-quartile stocks stands at 90% during 2021-2025, widening from 66% during 2011-2015. This suggests that growth concentration has increased in the equity market, and the premium for sustained growth may also rise in an environment where growth is more difficult to come by."

"the proportion of companies that have managed to deliver positive EPS growth has contracted from around 60% ten years ago to 40% in recent years.";全市场中位股2025年EPS增速仅4%、2021年以来CAGR 3%(过去十年平均约10%)

11 三大主题驱动的EPS加速读数:AI、出海、以及在较小程度上的反内卷政策,可把年度化指数EPS增速从6%的趋势水平提升至未来5年的10%左右(此后正常化至约6.5%)

Exhibit 19分主题读数:出海10%→6.5%、反内卷8.8%→6.5%、AI 8.5%→6.5%、基准趋势6.4%→5.9%(2026-30E | 2031-35E)

12 AI产业链FCF的环节分化 + 互联网盈利拐点信号

RMB bn口径下,AI应用段 554→831→748→275(2025年为低谷)、2026E-28E回升至383/670/960;半导体段2022-2025年连续为负((77)/(112)/(124)/(38))、2026E起转正并放量(326/486/591);电力段始终为正(104→174→133→167→165→249→377)

互联网板块(占"降级成长"市值40%)的盈利周期可能在3Q26财报季出现拐点。"the earnings cycle for the Internet sector (40% of De-rated Growth in market cap terms) could inflect in the ongoing 3Q26 results season—all pointing to potentially attractive tactical trading opportunities for nimble investors."(Exhibit 39:China AI universe FCF (RMB bn),五段为AI应用/AI模型/基础设施/半导体/电力)

13 配置建议落地为三张可执行名单

GS China Select AI Portfolio(50只,价格截至2026-11-06)
平均AI收入暴露79%、2026-28E平均收入CAGR 40%、EPS CAGR 39%、fPEG 0.7x;含中际旭创(收入CAGR 69%、fPEG 0.1x)、新易盛(51%)、天孚通信(53%)、长鑫存储CXMT(42%)等

GS China Shareholder Returns Portfolio(30家,价格截至2026-11-09)
合计市值约1.7万亿美元、6个月日均成交124亿美元;样本含茅台(5年平均派息率62%)、平安(27E股息率6.8%)、五粮液(6.9%)、JD.com(回购收益率7.4%)

De-rated/Cyclical FCF名单(19只Buy)
平均26E FCF收益率6%、27E/28E FCF增速29%/18%、27E PE 16.6x

原文支撑:Exhibit 35/37/40(GS Ratings: B=Buy、*为Conviction List)

14 监管被量化为"周期变量":POE Regulation Proxy(文本挖掘z分数)显示2018、2021两个紧缩期,2025年"31条"与座谈会后进入宽松期,最新2026年10月初读值仍在宽松区间

"The regulatory tightening intensity seems to follow a cyclical pattern";"POE Regulation Proxy reflects the text-mining results of news among POE-heavy sectors... The latest bar shows the preliminary reading of Oct 2026."(Exhibit 11)

15 国企与民企的盈利能力对比(MSCI China口径):2026E ROE为POE 13.1%、整体11.7%、SOE 10.5%——民企仍领先,但差距在过去几年有所收窄

Exhibit 12标题"The profitability gaps between POEs and SOEs have narrowed somewhat in the past few years";10年净利润率口径为POE 11.4% vs SOE 10.7%

四、一句话总括

读完这篇研报,最应该记住并可能改变认知的是:中国股票的"利润病"被误诊了——真正的病灶不在价格(通胀/内卷),而在"资金的价格太低";相应的纪律也不只是找盈利反转,而是分清五个成长桶,把"避开增长陷阱"提到与"拥抱成长"同等的地位,并用"12%正常ROE + 10%趋势EPS增速"作为估值锚去衡量10-15%的重估空间是否值得为之承担风险。

原始附件:《What's ailing Chinese profits, and where to find growth?》Goldman Sachs Global Strategy Paper No. 80

猜你喜欢