[Fact Check] What is the Truth Behind Statistics Korea Reporting Increased Semiconductor Production and Shipment Amid Ongoing Production Cuts?

Samsung Electronics and SK Hynix, leading the domestic semiconductor industry, are actively cutting production to balance memory semiconductor supply and demand. However, according to the industrial activity trends for August released by Statistics Korea on the 4th, both semiconductor production and shipments increased significantly. It is difficult to understand how production volume could increase while production cuts are underway. Why did such figures emerge?

Statistics Korea stated that semiconductor production rebounded after 13 months, raising expectations for the recovery of our manufacturing industry, which heavily depends on the semiconductor sector. The semiconductor production index for August, as announced by Statistics Korea, was 142.9 (base index, 2020=100), showing an 8.3% increase compared to a year ago and a 13.4% increase compared to July. This marks the first time in 13 months since July last year (14.9%) that semiconductor production has increased year-on-year. It appears that the semiconductor market is suddenly reviving.

However, contrary to the statistics, the actual semiconductor market remains poor. The semiconductor industry has maintained a production cut policy not only in August this year but also currently in October. Both Samsung Electronics and SK Hynix reported to Statistics Korea that semiconductor production decreased compared to last year. Although the worst phase of the vicious cycle of semiconductor inventory buildup and price drops has passed, inventory is still accumulating and price rebounds are slow. For these reasons, semiconductor companies are cutting production of both DRAM and NAND flash. This means that the environment is not one where the production and shipment volumes of the domestic semiconductor industry, which has a high proportion of memory semiconductors, could have increased compared to the same period last year.

Why do the numbers shown in the statistics fail to accurately reflect the industry's mood?

When compiling semiconductor production and shipment indices, Statistics Korea collects basic data not only from Samsung Electronics and SK Hynix but also from most domestic companies responsible for DRAM, NAND flash, system semiconductors, and back-end processes (such as packaging). They receive production value data by amount for about ten semiconductor product categories and apply weights by category to calculate the indices. In this process, Statistics Korea attempts to measure real production by removing semiconductor price fluctuations, dividing the amount-based data by the Producer Price Index (PPI) for each semiconductor category (denominator in the calculation).

The problem is that the PPI for memory semiconductors has sharply declined over the past year. According to the Bank of Korea's announced PPI, DRAM dropped from 100.78 in August last year to 55.25 in August this year, nearly halving. The flash memory PPI, which includes NAND, also plunged from 89.47 in August last year to 48.88 this August. For example, even if semiconductor companies report lower production value than last year, if the PPI has fallen sharply during that period, applying the lowered PPI in the conversion process to production statistics can cause a distortion where the final calculated index appears higher.

Additionally, the PPI used in index calculation did not fully capture the effect of semiconductor companies transitioning to new products. In the second half of the year, semiconductor companies focused on cutting production of legacy process products like DDR4 while increasing production of DDR5 or HBM product lines, which are priced more than twice as high. However, the numbers companies provide to Statistics Korea only include total DRAM production value without classification by high-end or low-end DRAM products. At the same time, the PPI applied in index calculation is uniform, so it cannot reflect the different price levels of high-end and low-end products. In other words, although prices differ by detailed product groups, applying a uniform PPI makes it appear as if the production index rises due to increased production of high-end new products, offsetting the impact of production cuts in low-end products.

Furthermore, even if companies cut memory semiconductor production, the production cycles vary by line and process, so the actual effect of production cuts is delayed, potentially causing a statistical illusion. It takes time from wafer input to finished semiconductor products. The industry expects the effect of production cuts to appear in as short as 2-3 months or as long as 6 months. For example, Samsung Electronics announced its semiconductor production cut policy in April, so it is difficult to say that the effect of production cuts is fully reflected in the current statistics. Because the production cut effect is not fully reflected, semiconductor production and shipments may temporarily appear to have increased compared to the previous month.

A semiconductor industry insider advised, "It is difficult to conclude that the semiconductor market has improved based solely on industrial activity statistics," adding, "It is necessary to comprehensively understand trends such as semiconductor price fluctuations, production volume, and export volume changes."

This content was produced with the assistance of AI translation services.