Since the end of July this year, photovoltaic modules have experienced price increases due to the three major factors of “partial polysilicon production interruption”, “bulk raw material prices”, and “some leading companies competing for industry pricing power”, ending the module since the end of 2016 The trend of “falling and steadily alternating” prices.

Since entering September, we have observed the following facts:

(1) Xinjiang Silicon Material Factory has begun to resume production gradually, and one of the four affected lines has been restored.

(2) Leading silicon wafer companies in the industry have slowed down their purchases of silicon materials since September this year, which has led to the fact that silicon wafer factories with low inventory of other raw materials are afraid to purchase silicon materials at high levels.

(3) Some silicon material traders have already “profited” their stocks of silicon materials held at low prices in the previous period at a high price in the range of 93-95 yuan/kg.

(4) The price of photovoltaic glass is still continuously hitting new highs. The spot purchase price of 3.2mm glass from small factories has exceeded 32-33 yuan/square meter, and even 34-35 yuan/square meter has been sold.

(5) The domestic “competitive electricity price” project has a strong willingness to land. Major photovoltaic module manufacturers have a full schedule in the fourth quarter, and even some second-tier manufacturers are already full (but third-tier manufacturers are affected by various factors and have difficulty in operating ).

At present, the photovoltaic industry generally believes that due to various factors such as demand-driven and commodity costs, module prices will remain stable in the fourth quarter; even if silicon material prices subsequently decline due to the gradual recovery of supply from Xinjiang silicon material plants, module prices are the most It only showed a state of “slight decline”.

So, how do we look at subsequent component price changes? This article puts forward some thinking directions for everyone to criticize and correct.

1. Is the price of photovoltaic modules a “cause” or “effect”?

For most commodities, price is a dependent variable and function of supply and demand. The same is true for photovoltaic modules in the era of “benchmark electricity prices”.

However, since the photovoltaic industry entered 2019, the logic of the global photovoltaic market is gradually changing: (1) The domestic “benchmark price” projects have gradually turned into “competitive power prices” or “parity projects”, (2) more and more overseas markets In the centralized power station project, bidding based on the core variable of electricity price was launched and the PPA agreement was signed.

As a result, the demand attributes of photovoltaic modules are undergoing profound changes, from a rigid demand curve (steep) with “price has a small effect on demand” to an elastic demand curve (with a greater effect on demand) (flat). The “reflexive” effect of price on demand is constantly increasing.

It is based on this that the module prices of the photovoltaic industry have plummeted since June 2018, which has promoted a round of rapid growth in overseas demand since the fourth quarter of 2018. But it is precisely based on this that the rebound in module prices since late July this year has hindered the logic of “price reduction to stimulate demand” in the photovoltaic industry.

Why is the demand in the second and fourth quarters better?

The current general quotation of module prices rose to 1.7 yuan/W, which was about 0.25 yuan/W higher than the quotation in mid-July. At the same time, SOLARZOOM new energy think tank learned through investigations that the demand in the fourth quarter was better (this is also the common understanding of the industry). So, since rising component prices will affect the logic of “price reduction to stimulate demand”, why is the demand in the fourth quarter so good?

We believe that there are at least the following two reasons:

(1) Domestic electricity bidding projects in the fourth quarter of 2020 will be implemented intensively. According to SOLARZOOM New Energy Think Tank’s article at the end of June, “20H2 or 21H1? “Comments on the Landing of Competitive Electricity Price Projects”, given that the module prices in the first half of this year are at a low level due to the global epidemic, if there is no increase in module prices since late July, the room for the decline in module prices in the first and second quarters of next year will inevitably not make up for the electricity prices in each quarter. The 1 cent reduction will have a negative impact on the project’s IRR, so at that time, we believed that it was more reasonable for the 2020 bidding projects to “rush to install” in the fourth quarter of this year. Now, although module prices are rising due to the influence of the supply side, there will still be a large number of domestic bidding electricity project owners who will choose to complete the grid connection in the fourth quarter of this year.

With reference to the situation in 2019, a similar conclusion can be drawn: the module prices in the second half of 2019 began to fall after a year of sideways trading after the “531” in 2018. According to the rational analysis of game theory, the IRR of the projects that the bidding price owners choose to connect to the grid in the first half of 2020 is higher than that of the projects that choose to connect to the grid in the second half of 2019 (in fact, it is true). However, the actual results show that there were still more than 10GW grid connection in a single month in December 2019. Among them, there is the consideration of the performance of central enterprises, and some owners cannot accurately predict the trend of component price changes. Standing at the time of the second half of 2020, based on the component prices in the second half of 2020 rising from a low level, more central SOEs will choose to implement in the fourth quarter of 2020 and implement in the first half of 2021. Choose the former (of course, some owners may choose to “abandon the project”).

Returning to our above framework, the domestic “competitive electricity price” project has a certain “rigid demand” characteristic due to the rule and time constraints of “the electricity price drops by 1 point one quarter late in the grid”. Therefore, in terms of the “rush to install” the domestic “competitive electricity price” project in the fourth quarter, “demand is the cause of price” is stronger than “price is the cause of demand”.

(2) In high-end markets such as the United States, Japan, and Europe, due to high land costs and labor costs, the absolute level of photovoltaic BOS and the proportion of system costs are relatively high. Therefore, these high-end markets are far less sensitive to the increase in photovoltaic module prices than mid-range and low-end markets such as China and India. Among them, these concentrated photovoltaic projects in the high-end market are less affected by the current labor constraints of the epidemic, and still benefit from the loose liquidity to a certain extent (the main high-quality energy giants have reduced financing costs). While the industrial and commercial distributed and residential rooftop projects in these markets are slightly affected by the epidemic (for example, the installed capacity of industrial and commercial and residential projects in the United States in the second quarter fell by 12% and 23% respectively from the previous quarter), but their photovoltaic BOS is in the system cost. As a result, its demand is negatively affected by the increase in component prices than centralized projects. Therefore, overall, the demand in overseas high-end markets can be expected in the fourth quarter (especially the fourth quarter in the United States is the peak season, and the seasonality of other countries is not particularly obvious).

However, in the low-end overseas markets, on the one hand, this year has been greatly affected by the epidemic (for example, India has a great impact, with less than 0.8GW of new installed capacity in the second quarter; and Brazil’s second quarter data also fell 45% from the first quarter. About), on the one hand, its project IRR is greatly affected by the increase in component prices, so there should not be too much expectation on the demand of these middle and low-end markets.

Based on the above two reasons, the demand in the fourth quarter of this year is still relatively good and shows certain rigid characteristics. The extent to which “the demand in the fourth quarter is negatively affected by price increases” will not appear to be so large due to the existence of domestic “competitive electricity pricing” projects and the low sensitivity of overseas high-end markets (Note: overseas low-end markets in the second quarter of this year The scale is already poor, and even if it is poor, its incremental impact on demand is limited).

3. The impact of supply and demand in various links of manufacturing on profit

Among the four links in the photovoltaic manufacturing industry, the tighter supply and demand relationship in 2020 will be silicon materials and solar cells, and the looser supply and demand relationship will be silicon wafers and modules.

Among them, the profit rate of the component link is currently at an extremely low level, so it is impossible to release too much profit space. In the case of good interim demand in the fourth quarter, the cell segment is unlikely to release profit margins. In the silicon material segment, the current profit margin has been at the best stage in the past for a long time. We also analyzed the impact of the future price drop and profit release in the “Analysis of the Impact of Xinjiang Silicon Plant’s Production on the Photovoltaic Industry” last week. process.

Then, the only uncertainty is the silicon chip link.

The wafer segment is the most violent expansion of the manufacturing segment in 2020. According to the plans of various manufacturers, by the end of 2020, if only the monocrystalline silicon wafer production capacity of several leading companies such as Longi, Zhonghuan, Jinko, JA Solar, and Shanghai Machine CNC are considered, it will reach 170GW; and all monocrystalline wafer companies will reach 170GW by the end of the fourth quarter The estimated total production capacity is expected to reach or exceed 200GW, and it will be about 170GW at the end of the third quarter. If the above-mentioned production capacity is converted to quarters, the demand in the fourth quarter will only exceed 40GW, and the demand in the first quarter of next year will only exceed 45GW, in order to ensure the basic full production of the silicon wafer link. Therefore, starting from a certain point at the end of the fourth quarter of this year (after the domestic “snap-up” shipment is completed), it is very likely that there will be a profit release in the silicon wafer link, which will promote the reduction of component costs.

We believe that starting from the end of the fourth quarter of this year, monocrystalline silicon wafers will “normalize” the phenomenon that the capacity utilization rate is lower than 100%, not due to maintenance or holidays. The reason for this situation is that monocrystalline silicon wafers are on the left side of the cost curve (lower cost and higher cost performance) in the process of replacing polycrystalline silicon wafers. In 2020, monocrystalline silicon wafers will be 100% replaced. One year of polycrystalline silicon wafers; therefore, when polycrystalline silicon wafers are completely replaced by monocrystalline silicon wafers, high-cost monocrystalline silicon wafers have become “marginal productivity.” “The technical route between monocrystalline silicon wafers and polycrystalline silicon wafers The resulting cost difference” will also be released once.

In view of the fact that the current higher cost silicon wafer companies have a gross profit margin of about 20% (lower cost silicon wafer companies have a gross profit margin of about 35%), most of the 10% period expense ratio is invariable costs (only a small part is Variable cost), depreciation (invariable cost) accounts for about 8% of the cost. When there is a serious excess of monocrystalline silicon wafers (for example, the capacity utilization rate of monocrystalline silicon wafers is less than 70%), it can be released The maximum profit margin is about 20% (corresponding to the tax-included price of 0.62 yuan / piece). For example, according to the calculation of 6.22W/piece of M6 silicon wafer, due to the oversupply of silicon wafer, the module price of 0.10 yuan/W can be released at most (Note: If it is a monocrystalline silicon wafer, the capacity utilization rate is significantly lower than 70%).

Fourth, the position of the cost curve in the fourth quarter

The position of the photovoltaic module cost curve in the fourth quarter was affected by the price of polysilicon, the profit of silicon wafer companies, and the price of bulk commodities. Since late July this year, polysilicon prices and bulk commodity prices have had the same influence, and the profits of silicon wafer companies have remained basically stable. However, from the fourth quarter of this year, polysilicon prices may be the first to see a significant decline due to the gradual and comprehensive recovery of Xinjiang silicon material plants, and the profits of silicon wafer companies are also expected to begin to be released due to the large-scale production of leading companies (Note: Fourth quarter The demand is good, so the profit of silicon wafers will not be fully released for the time being 0.10 yuan/W module price, unless there is a situation of concentrated sales of domestic and foreign inventory that we analyze below).

According to calculations by SOLARZOOM New Energy Think Tank, the current silicon consumption per W of photovoltaic modules is about 2.94 grams. Therefore, for every 10 yuan/kg decrease in the tax-included price of polysilicon, the corresponding component tax-included cost will decrease by about 0.03 yuan/W. In terms of glass, if double-glass modules are used, the price of glass will increase by 1 yuan/square meter, and the tax-included cost of the corresponding components will increase by 0.01 yuan/W.

In general, if silicon material prices and silicon wafer enterprise profits fluctuate downwards in the future, the downward and rightward shift of the cost curve will be significantly greater than the upward shift of the cost curve by the increase in the price of auxiliary materials. Unless there is a large-scale global inflation (which has not yet occurred), the probability that the fourth quarter cost curve of photovoltaic modules will move downward and to the right will be greater.

V. Inventory issues in the photovoltaic industry

Regarding whether there is inventory in the photovoltaic industry, there are different opinions in the photovoltaic industry. Regarding the definition of inventory, there are also different statistical calibers. SOLARZOOM new energy think tank believes that there are at least the following calibers:

(1) “Domestic manufacturers + traders + EPC + operators” inventory change = manufacturer’s production volume-domestic new installations-overseas new installations

(2) Domestic manufacturers’ inventory changes = manufacturer’s production volume-manufacturer’s shipments (including manufacturer’s own installation volume + sales volume)

(3) “Domestic manufacturers + traders + EPC + operators” domestic inventory change = manufacturer’s production volume-domestic new installed capacity-net export volume

(4) “Domestic manufacturers + traders + EPC + operators” overseas inventory changes = net exports-new overseas installations

In the above formula, due to data availability issues and delayed data publication issues, it is impossible to accurately calculate one by one. At present, more accurate data can be obtained: manufacturer’s production volume, domestic new installed capacity, and export volume (Note: Since there is almost no import of photovoltaics, so “export volume = net export volume” is approximately established). The above three data are as follows:

(1) According to SOLARZOOM new energy think tank module export data, module exports from January to May 2020 are 26.2GW; the estimated value of the same caliber for export in June is US$1.32 billion, and the estimated export scale data is 6.0GW, which can be calculated. In the first half of the year, module exports were 32.2GW, down 8% year-on-year. The error of the above data will not exceed ±1GW.

(2) According to Wang Bohua’s PPT “Development Situation and Prospects of the Photovoltaic Industry” on September 10 this year, China’s module output in the first half of 2020 was 53.3GW, an increase of 13.4% year-on-year, and the output of solar cells and silicon wafers reached 59GW and 75GW. . This data is the statistical data of industry associations with high accuracy.

(3) According to data from the Bureau of Energy, 11.5GW of newly installed photovoltaic capacity in the first half of 2020 is basically the same as 11.4GW in the first half of 2019. According to the calculation of the data of China Electricity Union from January to June and January to July, the newly installed photovoltaic capacity in July was about 2.9GW (so there is no particularly obvious phenomenon of “rush to install”). The above data is official data with high accuracy, but considering the issue of the capacity ratio of AC and DC calibers, it may correspond to about 10-15% of the additional installed capacity of components (corresponding to about 1.2-1.7GW).

Based on the above calculations: in the first half of 2020, the output of photovoltaic modules is 53.3GW, the domestic newly installed capacity is 11.5GW (the DC side may be +1.2~1.7GW more than the AC side), and the module export volume is 32.2GW, resulting in 7.9-8.4 GW (Error ±1GW) domestic “incremental” inventory. This is equivalent to one month’s demand level.

It should be noted that the above-mentioned stocks were formed during the first half of 2020 when the global epidemic was relatively severe. The component prices at that time were at a relatively low level. Judging from the current component prices, the above-mentioned inventory has already achieved certain benefits. But whether this part of the inventory will be sold on the market depends on the changes in component prices and the degree of support from the financial market.

We know that component inventory revenue = selling price-cost at the time of inventory formation-cost of capital = (sale price-price at the time of inventory formation) + (price at the time of inventory formation-cost at the time of inventory formation)-cost of capital = price change + Locked profit-cost of capital. Among them, the second item “locked profits” is fixed, the third item accumulates over time, and the only thing that changes is the first item “price change”. Therefore, since the formation of domestic inventory in the first half of 2020, due to the rise in component prices, this part of the industry’s inventory has been profitable (whether in the hands of manufacturers or channels). However, if the price of components drops slightly, the floating profit of these stocks will quickly narrow, which will lead to concentrated selling of their stocks.

Therefore, from the current point of view, judging the inflection point of the component price has an important guiding role in determining the subsequent inventory behavior and the speed of component price change. Once there is a clearer loosening of component prices, the following mechanisms will be triggered:

(1) Some domestic projects have increased the momentum of “rush to install”.

(2) Wait-and-see demand for “non-rush-installation” and part of “rush-installation” demand at home and abroad (expected that module prices will fall by more than 0.10 yuan/W in the first quarter of next year or 0.20 yuan/W in the first half of the year, so that the fourth quarter will not be Switch to rush to install in the first half of next year).

(3) Inventory sell-off.

Among them, the first mechanism is good for demand, the second mechanism is not good for demand, and the third mechanism increases supply. In view of the fact that the total amount of domestic incremental “snap-loading” and the scale of “non-snap-loading” demand at home and abroad may be basically the same, the main variable determining the price will be the domestic incremental 7.9-8.4 GW (±1GW) inventory When to sell.

In addition, the overseas epidemic situation is obvious in the first half of 2020, and the decline in overseas exports is only 8%. But how much overseas installed capacity in the first half of the year? At present, accurate global statistics are not yet available. Therefore, we cannot accurately estimate the level of overseas inventories. But if there are overseas stocks, the behavior of these stocks will inevitably change with changes in component prices.

6. What should the price of photovoltaic modules look like in the fourth quarter?

Based on the complete analysis above, we can draw the following conclusions:

(1) From a demand perspective, there was strong support in the fourth quarter (domestic + overseas high-end markets, overseas emerging markets cannot be counted on), and there is a strong rigid demand characteristic due to the existence of domestic “snap-up”.

(2) From the perspective of cost promotion, the price drop of polysilicon materials mainly depends on when the Xinjiang silicon materials plant will resume production (according to the research of SOLARZOOM think tank, it is expected to be at the end of September), and there is more room for the profit release of the silicon wafer segment , The price increase of bulk commodities may hedge some of the component price drops caused by the above mechanism.

(3) According to relatively accurate and clear data calculations, in the first half of 2020, approximately 7.9-8.4 GW (±1GW) of domestic inventory (partly in the hands of manufacturers and partly in the hands of channels) will be added in the country, and the level of overseas inventory increase Can’t calculate accurately yet. This part of the incremental inventory is currently profitable, but once the component prices start to fall, its inventory revenue may decline rapidly, which will cause its inventory behavior to change from “accumulation” to “selling.”

(4) Once the price of components drops in the future, there will be an increase in demand due to stimulating domestic “rush installation” and a decrease in demand caused by “non-rush installation” projects at home and abroad + “rush installation” projects wait and see, both of which are in scale It may be basically the same. However, compared with the domestic inventory scale of 7.9-8.4 GW (±1GW) + the overseas inventory that cannot be accurately calculated, the increase or decrease in the scale of real installed demand is not dominant. In the end, it is the behavior of inventory that determines how quickly component prices fall.

(5) We believe that if the price of components in the fourth quarter loosens or declines slightly due to the downward and rightward shift of the component supply curve, it is necessary to pay special attention to the changes in domestic and foreign inventory at that time. If the inventory at that time makes a positive feedback behavior of “chasing the rise and killing the fall” that is in line with its own profit maximization, it is necessary to pay attention to the possibility of “a small price loosening into a larger price adjustment”. Of course, the biggest factor that will ultimately determine the price of modules at the end of the fourth quarter of this year is the expected global new installed capacity in 2021; if the industry’s expectations for 2021 demand at that time were still optimistic or even extremely optimistic, inventory remained stable in the fourth quarter of this year Even the possibility of continuing to accumulate also exists.