1、 cGlobal Research21 February 2020EquitiesChinese Independent Power Producers UBS Evidence Lab inside: upgrading our view on IPP A shares in a Goldilocks power marketMajor A-share IPPs could benefit the most in a Goldilocks power marketWe have an anti-consensus upgrade on major A-share IPPs Huaneng-A

2、, GD Power (Neutral to Buy) and Datang-A (Sell to Neutral). We think the market is wary about a significant decline in power demand growth, but we maintain that power demand growth could be sustained at 4.7%/4.1% in 2020/2021E, based on the UBS Evidence Lab Data Science interactive model and the pos

3、sibility of supply-side reform. We have identified a potential Goldilocks situation where: 1) the coal price falls to Rmb530/t;utilisation rates remain reasonable; and 3) the average tariff faces moderate downside. In light of the COVID-19 outbreak, we expect the overall impact on the power sector t

4、o be relatively limited and the performance of major IPP A shares to remain defensive against the A-share market, similar to the situation during the SARS outbreak in 2003. We believe an average 2020 year-on-year earnings rebound of 70% and a recovery of average ROE from 5% in 2019E to 8% in 2020E a

5、re not priced in as IPP A shares are trading at 0.8x 2020E P/BV, implying a long-term ROE of 7%, below our 8% estimate.Power demand growth can be sustained at around 4-5%Using year-on-year construction new starts and steel production growth, we forecast 4.7%/4.1% power demand growth in 2020/2021, ba

6、sed on the UBS Evidence Lab Data Science interactive model. Starting from 2022, we factor in power demand from new industries and forecast 4.4% YoY growth through to 2025.Supply-side reform implies upside potential for power producersIn December 2019, local media reported that SASAC might integrate

7、resources and eliminate outdated plants in trial regions, leading to a one-third to one-fourth capacity cuts in these regions by 2021. Our scenario analysis suggests earnings of A-share IPPs that we cover would increase 26-45%/0-5% if national/regional coal-fired power plant capacity is cut one-thir

8、d by 2021, with the benefits spread equally across all fuel types.Upgrade Huaneng-A, GD Power to Buy and Datang-A to NeutralWe raise our earnings estimates of A-share IPPs 43%/33% on average for 2020/2021 to factor in moderating coal prices and easing competitive pressure on the average coal-fired t

9、ariff with reasonable utilisation rates. Our top pick is Huaneng-A as we think it possesses better asset quality with a long-term ROE of 10%, higher than our sector average estimate of 8%.Figure 1: Valuation summaryElectric UtilitiesChinaAlex Liu Analyst S1460515080001 HYPERLINK mailto:alex-c.liu al

10、ex-c.liu+86-213-866 8857Freda ZhuAnalyst S1460518110002 HYPERLINK mailto:freda.zhu freda.zhu+86-213-866 8879ReutersRatingPrice targetShare ImpliedPEP/BVROEDividend yieldcodeNewOldNewOldpriceUpside2020E2021E2020E2021E2020E2021E2020E2021EHuaneng-A600011.SSBuyNeutralRmb7.00Rmb6.20Rmb4.9940%8.4x9.6x0.8x

11、0.8x10.3%8.6%8.8%7.6%GD Power600795.SSBuyNeutralRmb2.80Rmb2.46Rmb2.1630%9.2x8.0 x0.8x0.8x8.6%9.6%7.2%8.3%Datang-A601991.SSNeutralSellRmb2.30Rmb2.20Rmb2.233%19.0 x19.4x0.9x0.9x4.7%4.4%2.7%2.6%Source: Above data as at 20 February 2020. Source: Reuters, Wind, company data, UBS-S estimates HYPERLINK / T

12、his report has been prepared by UBS Securities Co. Limited. ANALYST CERTIFICATION AND REQUIRED DISCLOSURES BEGIN ON PAGE 44. UBS does and seeks to do business with companies covered in its research reports. As a result, investors should be aware that the firm may have a conflict of interest that cou

13、ld affect the objectivity of this report. Investors should consider this report as only a single factor in making their investment decision.China power sectorUBS Research THESIS MAP MOST FAVOUREDLEAST FAVOUREDHuaneng-ADatang-APIVOTAL QUESTIONSQ: Will an economic slowdown severely drag down power dem

14、and growth and sector profitability?Unlikely. Based on UBS Evidence Labs power demand model, we estimate power demand growth of 4.7/4.1% in 2020/2021. We believe A-share IPPs could be in a Goldilocks situation in the near term easing competitive pressure on tariffs, reasonable utilisation rates, and

15、 moderating coal prices. HYPERLINK l _bookmark1 moreQ: Will the coal supply/demand balance ease in 2020E?No. Ongoing high fixed-asset investment (FAI) in the coal sector has boosted Chinas coal production from the start of 2019. We are concerned that FAI into the sector could continue to rise rapidl

16、y in 2020, further pushing up coal production, while demand may rise only slightly. We thus believe China thermal coal prices could correct to Rmb530/t in 2020 (we previously estimated Rmb580/t), driven by a potential increase in coal supply, and could hover around the same level for two to three ye

17、ars.A-share IPPs: tariff faces downside risk under floating scheme 10/17/2019Q: Will supply-side reform stimulate earnings growth for the power sector?Likely. We are waiting for concrete action plans from the government. Our scenario analysis shows that if national/regional coal-fired power plant ca

18、pacity is cut one-third by 2021 and the benefits are spread equally across all fuel types, power producers earnings would increase 26-45%/0-5%. HYPERLINK l _bookmark21 moreUBS VIEWWe hold a positive view on A-share IPPs. We think they will benefit from a decline in coal prices in a Goldilocks power

19、market. We see less competitive pressure on the average tariff and limited change in utilisation rates.EvidenceLeveraging UBS Evidence Labs model, which is based on year-on-year construction new starts and steel production growth, we estimate power demand growth could be sustained at 4.7/4.1% in 202

20、0/21E. This implies a Goldilocks situation for the power industry, in our view. The UBS basic materials team estimates Chinas thermal coal prices could correct to Rmb530/t in 2020.WHATS PRICED IN?The coal-fired sectors A shares were down 19% in 2019, underperforming CSI 300 by 55ppt. We think invest

21、ors have not factored in further declining coal prices and sustained power demand growth. The sector is trading at 0.8x 2020E P/BV, implying that the market might be factoring in long-term ROE of 7%, below our estimate of 8%.A-share IPPs share price performance vs CSI 300 in 2003Note: 2 January 2003

22、 prices rebased to 100. Source: WindChina power sectorUBS ResearchPIVOTAL QUESTIONS HYPERLINK l _bookmark0 return Q: Will an economic slowdown severely drag down power demand growth and sector profitability?UBS VIEWUnlikely. Based on UBS Evidence Labs power demand model, we estimate power demand gro

23、wth of 4.7/4.1% in 2020/2021. We believe A-share IPPs could be in a Goldilocks situation in the near termeasing competitive pressure on tariffs, reasonable utilisation rates, and moderating coal prices.EVIDENCEUBS Evidence Lab has adopted a bottom-up approach to build a proprietary data science powe

24、r demand model using regional and quarterly data on year- on-year construction new starts and steel production growth. Based on this model, we forecast 4.7%/4.1% YoY power demand growth in 2020/2021.WHATS PRICED IN?A shares in coal-fired sector were down 19% in 2019, underperforming CSI 300 by 55ppt

25、. We think investors have not fully priced in a further decline in coal prices and a Goldilocks power industry that could benefit A-share coal- fired IPPs in the near term.Power demand growth sustainable at 4-5%Based on UBS Evidence Labs proprietary power demand model, we estimate power demand growt

26、h in 2020/2021 could be sustained at around 4.7%/4.1%. We use UBS Evidence Labs year-on-year steel production growth and new construction starts data to derive reasonable estimates for power demand growth. We think this is an innovative way of estimating power demand growth in China. Forecasts from

27、UBSs property and basic materials teams (-3%/-1% YoY growth in new construction starts and 2.5%/0.5% YoY steel production growth in 2020/2021) also provide visibility on power demand growth.We believe the market becoming increasingly curious about the incremental power demand growth from new industr

28、ies. Therefore, from 2022, we expect three key new industries: 5G base stations, electric vehicles and internet data centres to contribute to power demand growth. Our analysis suggests they would contribute only 0.3% additional power demand growth in 2022-25. As such, we estimate 4.4% YoY total powe

29、r demand growth over this period HYPERLINK l _bookmark2 (Figure 2).(%)16%14%12%10%8%6%4%2%2010201120122013201420152016201720182019E2020E2021E2022E2023E2024E2025E0%Base power demand growthEV power demand growth5G BTS power demand growthData center power demand growthSource: CEC, China Electronics Sta

30、ndardization Institute. Greenpeace, North China Electric Power University, UBS Evidence Lab, UBS-S estimatesSector profitability improving: coal-fired IPPs A shares could recoverWe believe coal-fired IPPs could be the near-term beneficiaries of a 4.7% YoY power demand growth rate in 2020, based on:

31、1) UBS estimates for coal prices to fall and normalise at around Rmb530/t; 2) our view that any decrease in the average tariff would be subdued, due to easing competitive pressure; andreasonable utilisation rates, due to sustained moderate power demand growth. We hold a positive view on A-share coal

32、-fired IPPs in the near term as we expect them to benefit from falling coal prices.Coal-fired power sectorWe upgrade Huaneng-A, GD Power from Neutral to Buy and Datang-A from Sell to Neutral. We raise our earnings estimates an average 43%/33% for 2020/2021 to factor in moderating coal prices, easing

33、 competitive pressure on the average coal-fired tariff and reasonable utilisation rates. HYPERLINK l _bookmark3 Figure 3 illustrates our ROE estimates for A-share power producer stocks in our coverage. We expect the average ROE of coal-fired IPP A shares recover to 8% in 2020 from 5% in 2019E. We no

34、w estimate their long-term ROE to remain at around 8%, higher than our previous 5% estimate. The three coal-fired IPP A shares are trading at around 0.8x 12-month forward P/BV. We think this valuation is attractive, as we forecast an average 8% long-term ROE for the sector. We prefer Huaneng-A as it

35、 we think it possesses better asset quality and estimate a long-term ROE of 10%.25%20%15%10%5%0%20152016201720182019E2020E2021E2022EHuaneng-AGD PowerDatang-A Source: Company data, UBS-S estimatesChanges to our EPS estimates and price targetsThe key changes in our assumptions for A-share IPPs are as

36、follows:The UBS basic materials team expects a QHD 5500kc coal price of Rmb530/t in 2020 and Rmb540/t in 2021. This is below its previous forecast of Rmb580/t for both 2020 and 2021. The decline in coal prices could serve as a catalyst for an increase in earnings.We raise our assumption for the perc

37、entage of market power sales from around 80% to 85% for 2021 onwards. We expect the market power sales to contain two parts: power sales under a floating tariff scheme, and direct power sales. We expect 10% of power sales to be sold under a floating tariff and the remaining 75% to be sold directly t

38、o customers from 2021 onwards.We raise our 2020/2021 utilisation rate assumptions 0.3% to reflect sustained power demand growth.Figure 4: Changes in our key assumptionsHuaneng-AGD PowerDatang-A2020E2021E2020E2021E2020E2021ECoal price (Rmb/t)New530540530540530540Old580580580580580580Utilisation rateN

39、ew48.1%48.1%52.7%52.7%54.6%54.6%Old47.8%47.8%52.4%52.4%54.3%56.3% of competitive market salesNew80%85%80%85%80%85%Old80%90%80%80%80%80%Weighted average tariff (Rmb/kWh, ex-VAT)New0.3480.3480.2870.2840.3220.318Old0.3480.3450.2870.2840.3220.322Source: UBS-S estimates（RMB/t)9008007006005004003002008 20

40、09 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019E 2020E 2021ESource: Wind, UBS-S estimatesFigure 6: Earnings sensitivity analysis of coal-fired IPPsHuaneng-A2020E2021EGD Power2020E2021EDatang-A2020E2021E+1% average coal price-4.8%-5.0%-4.8%-3.6%-5.1%-4.9%+1ppt thermal utilisation rate5.9%6.4%1.6

41、%1.4%8.9%8.8%+25bps interest rate-3.1%-3.4%-7.2%-5.9%-4.8%-4.6%+1% coal-fired tariff7.2%8.2%7.4%6.1%9.5%10.1%Source: Company data, UBS-S estimatesOur price targets now imply around 1.1x 12-month forward P/BV for the sector. We think this is justified, as we forecast long-term ROE for the sector of a

42、round 8%.Figure 7: Changes to our ratings, price targets, and EPS estimatesReutersRatingPrice targetCurrentPrice2019E EPS2020E EPS2021E EPScodeNewOldNewOldpriceUpsideNewOldChgNewOldChgNewOldChgHuaneng A600011.SSBuyNeutralRmb7.00Rmb6.20Rmb4.9940%0.380.4-5%0.60.3760%0.520.3547%GD Power600795.SSBuyNeut

43、ralRMB2.80Rmb2.46Rmb2.1630%0.190.1714%0.240.1834%0.270.1944%Datang - A601991.SSNeutralSellRmb2.30Rmb2.20Rmb2.233%0.050.13-60%0.120.0935%0.110.119%Note: *EPS data in Rmb. Above data at 20 February 2020. Source: Company data, Wind, UBS-S estimatesPotential sector driversWith reasonable power demand gr

44、owth, we expect the following drivers to materialise and benefit A-share coal-fired IPPs.Stable utilisation rates at a reasonable levelFigure 8: Utilisation rate of A-share coal-fired IPPs70%65%60%55%50%45%40%35%2007200820092010201120122013201420152016201720182019E2020E2021E2022E2023E2024E2025E30%So

45、urce: CEC, UBS-S estimatesThe A-share coal-fired power industry has been in severe oversupply since 2015, with the average utilisation rate falling around 50% from about 60% in the previous decade. We expect power demand to grow 4.7%/4.1% in 2020/2021. Our estimates assume oversupply will likely per

46、sist for the next few years and gradually ease in the long term (see HYPERLINK l _bookmark4 Figure 8 for utilisation rates of coal-fired power plants in the industry). Our analysis suggests the utilisation rate of an average A-share coal-fired IPP could remain at the 52% level in the near term. For

47、Huaneng-A, GD Power, and Datang-A, we factor in an additional 0.3% to our previous forecasts for 2020/2021 to reflect industry trends HYPERLINK l _bookmark5 (Figure 9).Figure 9: Utilisation rates of the three A-share IPPs we cover70%60%50%40%30%20%10%0%20122013201420152016201720182019E2020E2021EHuan

48、eng-AGD PowerDatang-A Source: Company data, UBS-S estimatesEasing competitive pressure on tariffsWe expect competitive pressure on the tariff for A-share coal-fired IPPs to ease, because: 1) the percentage of power sold through market power sales reached about 51% in H119; and 2) we forecast power d

49、emand growth to remain above 4% in 2020-21.In HYPERLINK l _bookmark7 Figure 11, we list our average tariff and percentage of market power sales assumptions. We assume the magnitude of tariff cuts from market sales in 2020 and 2021 is maintained at the 2019 level. This is way lower than that in 2015

50、to 2016 when power demand growth was weak and competition was irrational as market sales were introduced HYPERLINK l _bookmark6 (Figure 10).Figure 10: Average tariff cut for market power sales vs power demand growth(Rmb/kKWh, ex-VAT)0.068%7%0.056%0.045%0.034%3%0.022%0.011%-20152016201720182019E2020E

51、0%2021EA-share coal-fired IPPs (LHS)Power demand growth (RHS)Source: Company data, UBS-S estimatesFigure 11: Average tariff assumptions for A-share IPPs under our coverageRmb/kWh, ex-VAT2019E2020E2021EAverage tariff of Huaneng-A, GD Power and Datang-A0.3240.3190.317average % of market power sales51%

52、80%85%Source: UBS-S estimatesModerating coal pricesThe UBS basic materials team expects China thermal coal prices to correct to Rmb530/t in 2020, below its previous Rmb580/t forecast, and hover around the same level for two to three years. This is because: 1) ongoing high FAI into the coal sector mi

53、ght continue to increase coal supply in 2020; and 2) coal demand growth could slow in 2020, pressured by hydropower generation, lower steel production, a slowing manufacturing capex cycle, and high coal inventories. Therefore we see further downward pressure on coal prices in the near term HYPERLINK

54、 l _bookmark8 (Figure 12).(Rmb/t)9008007006005004003002008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019E 2020E 2021ESource: Wind, UBS-S estimatesUBS Evidence Labpower demand modelIn the following sections, we discuss how the model works and back-test it.How closely does it track actual pow

55、er demand growth?We back-test UBS Evidence Labs power demand model with actual power demand and the model yields 61% R-squared HYPERLINK l _bookmark9 (Figure 13), indicating the 61% power demand growth variation could be explained by the model. A mean absolute error calculates the average of absolut

56、e differences between modelled power demand growth and actual power demand growth, and could be treated as a proxy for standard deviation. The UBS Evidence Lab model has a 0.01-0.02 mean absolute error.Figure 13: Power demand modelling resultsR-squaredMean absolute errorRegion (average prediction er

57、ror; YoY)In-sampleIn-sampleOut-of-sampleCentral72%0.030.04Northern74%0.030.02Southern62%0.020.03North Eastern49%0.030.02North Western80%0.020.03Eastern55%0.030.03National (Method*)61%0.010.02*Sum the predictions from six regions. Source: UBS Evidence Lab, WindUsing UBSs 2019 steel and new constructi

58、on starts estimates, the model derives 2019 full-year power demand growth of 4.8%, compared with 4.7% for 9M19. We are not back-testing the model from Q109 to Q412 because aluminium production data started in 2013 January. HYPERLINK l _bookmark10 Figure 14 illustrates actual power demand, and UBS Ev

59、idence Labs modelled power demand. It shows the model closely tracks actual power demand with a level of error that we view as acceptable.(bn KWh)2,1001,9001,7001,5001,3001,100900700Mar-09Mar-10Mar-11Mar-12Mar-13Mar-14Mar-15Mar-16Mar-17Mar-18Mar-19500Model Power Demand (bn KWh)Actual Power Demand (b

60、n KWh)Source: UBS Evidence Lab, WindHow does the UBS Evidence Lab power demand model work?UBS Evidence Lab adopts a bottom-up approach to build a proprietary data science power demand model using quarterly and regional data.Figure 15: Breakdown of power demand in 2018Tertiary industry 15%Residential