On April 9, 2018, I wrote a Seeking Alpha article entitled "Why Memory Chips Won't Be Impacted By A Trade War In One Chart."
My key points from the bulleted summary were:
- Current bravado between the U.S. and China that could spark a trade war could also impact U.S. manufacturers of memory chips.
- Despite massive investments to make China self-sufficient in semiconductor production, China imports nearly all its memory chips from foreign companies.
- Any trade war that would impact the flow of memory chips into China would have a deleterious effect on its internal production of high-tech consumer products utilizing memory chips.
The memory market and the global economy has now had a year of data to absorb and analyze, since the U.S imposed a 10% tariff on $200 billion worth of China goods. Now that China-U.S. relations have further deteriorated and a new round of sanctions became effective May 10, 2019, I want to present data that suggests that the impact of Phase 1 was minimal and Phase 2 should be negligible.
Korea's Semiconductor Exports Improving
Korea's Ministry of Trade, Industry and Energy announced on May 1 that South Korea's exports fell less than expected in April even as they dropped for a fifth straight month. Exports from South Korea dropped by 2.05% year-on-year to $48.86 billion April 2019, following an 8.2% fall in March, while markets expected a 5.0% decrease.
The decline in April's exports was partially due to (1) a fall in cost of semiconductors and petrochemicals (2) slowing Chinese economy, (3) ongoing trade dispute between the US and China, and (4) stagnant world trade. Sales of semiconductors fell 13.5 percent, mostly due to falling prices of memory chips, prolonged inventory adjustments in data centers, and sluggish demand for smartphones.
Table 1 shows this change in semiconductor as well as other high-tech devices - wireless communication, flat panel displays, and computers. All exhibit the same trend except for wireless communications for April, which was due to a 43.7% YoY jump in mobile phone exports. Electronic products is the largest trading category, representing 28% of all trade.
Among trading partners, exports to China show the results of the slowing Chinese economy, which is impacting Japan's economy. China represents 25% of all Korea exports followed by the US at 12%. Besides these applications, Korea trade includes petrochemicals, general machinery, shipbuilding, and automobiles among others. US-bound exports rose for a seventh consecutive month, helped by a boost in demand for Korean cars and mobile phones.
Since most of the semiconductors manufactured in Korea are memory chips, and most are exported to China for insertion into smartphones, computers, TVs, and other electronic devices, I've plotted semiconductor exports and exports to China on the same graph (Chart 1).
Although exports to China include all products, the correlation with Korea's semiconductor shipments illustrates the close economic synergy between the two countries. This close relationship suggests:
- Once the memory overhang eases and prices start rising, export value in revenues will rise.
- Once the China economy improves and trade agreements are reached with the U.S., exports of semiconductors will improve. But Phase 2 of the trade war could delay improvements in China's economy. This could impact consumer items such as smartphones and TVs, but the large growth areas such as cloud servers and upcoming 5G networks won't be impacted.
Chart 2 shows IC imports into China from May 2018 through March 2019, a similar period to Chart 1. These two charts illustrate the synergy between Korean semiconductor exports (Chart 1) and China imports (Chart 2). The difference is due to the fact that China imports semiconductors from countries other than Korea.
Chart 3 shows the danger of making too many comparisons, because imports and exports are dependent on a variety of factors. Chart 3 is merely a duplicate of Chart 2, but going back to March 2016 rather than March 2018. One can interpret that the valley we see in February 2019 and attributed to the China trade problem also occurs in February 2018, February 2017, and possibly February 2016 are seasonal factors vs trade factors.
Korea's Semiconductor Exports Tied to ASPs
Chart 4 shows Korea's chip exports between 2010 and 2019, illustrating the rapid drop in chip exports of about 50% since late 2018 - the worst since 2009. The rise in exports, which is valued in revenue, is due to the run-up in average selling prices (ASPs) of memory chips since the end of 2016, as show in Chart 2. The blue line shows the server contract price and the correlation exports (red line). Note that this chart show data through March 2019 (latest data available) while chart 1 is for data through April 2019.
Memory Chip Production by Region
It is important to note where DRAMs and NAND are manufactured and how many go to China.
On a wafer start basis:
- For DRAMs, 70% of the 1.15 million wafer starts per month were made in Korea vs 10% in China (SK Hynix (OTC:HXSCL) in Wuxi) in 2018
- For NAND, 39% of the 1.56 million wafer starts per month were made in Korea vs 14% in China (Samsung (OTC:SSNLF) in Xi'an and Intel (INTC) in Dalian) in 2018
- Samsung and SK Hynix, except for Chinese fabs, make all their chips in Korea
On a revenue basis:
- For DRAMs, 43% were sold into China in 2018
- For NAND, 31% were sold into China in 2018
To be clear, the rise in Korean semiconductor exports are dependent of prices of semiconductors not unit shipments. Chart 5 shows global unit shipments of all DRAM and NAND chips from December 2017 through March 2019. Shipments for both types of memory chips were flat until Q3 of 2018. There was a rise in shipments for DRAMs as ASPs began their descent, but the sharp drop in Q4 (for NAND as well) as the impact of a slowdown in China and slower sales of smartphones and server spend, which were a catalyst for a buildup of inventory. The fact that there are several factors at play reduces the probability that the drop in shipments was primarily attributed to the China trade issue.
Stock Performance of Memory Chip Companies
Table 2 shows that memory companies Micron Technology (MU), Samsung Electronics, and SK Hynix compared the Philadelphia Stock Exchange (PHLX) by YTD, 1 Year Growth, and 2 Year Growth. The performance of both MU and HXSCL are comparable to the PHLX for YTD and 2 Year growth, but underperformed over the one year period between May 10, 2018 and May 10 2019. In contrast, SSNLF underperformed the SOX and competitors in all periods.
Chart 6 shows the SOX and MU stock price for the past two years. With the SOX at about 40 times MU's stock price, the impact of the recent memory cycle is difficult to discern on the chart. But if we look at Table 3, with data taken from Chart 6, we can see that for three critical dates - July 3, 2017 the start of the MU run-up, May 21, 2018 the MU peak, and December 17, 2018 the low point of MU. Note that the ratio of both values was comparable at 34.3 for the two end points and only 24.0 at the middle peak.
About two-thirds of the 50 mostly large-cap Kospi index firms that have announced first-quarter earnings have missed analysts' estimates through April 25, 2019, according to data compiled by Bloomberg. Tech stocks have the biggest weighting at almost 30 percent of the Kospi. Chart 7 shows that unlike the PHLX, which has no correlation with MU stock, the Kospi index (red line) shows a strong correlation with the ramp in stock of Samsung and SK Hynix.
Phase 1 of the China trade war may have impacted Korean exports but whether it impacted the memory market is not obvious. The downturn in the memory market, which resulted from overcapacity and excess inventory, sky high prices, and a pause in capex spend by cloud server companies merely coincided with the trade war, as there appears to be a minimal causal relationship.
Phase 2 of the China trade ware is just beginning. There could be a continued negative impact on consumer items utilizing memory chips, but the key drivers of memory in 2H 2019 are a resumption in capex spend by cloud server suppliers, increase in memory per smartphone, reduction in inventory overhang, and 5G networks.
As a result, memory chips should resume their growth in 2H 2019 and into 2020. Growth will be muted compared to the meteoric growth of 2017-2018 (Chart 4) because of several factors including expanded capacity increase through new fabs, greater bit growth, and the inauguration of a memory chip industry in China, which will temper ASP growth, particularly for low-end chips.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.
Editor's Note: This article discusses one or more securities that do not trade on a major U.S. exchange. Please be aware of the risks associated with these stocks.