Silicon Peninsula #018 — Korea's export surge needs a working-day check before an earnings model
Thursday, October 22, 2026 · For semiconductor analysts and supplier planning teams
The tape — strong activity, different denominators
Our conclusion: Korea's new preliminary trade data strengthen the case for investigating semiconductor demand, but they should enter a company model through a reconciliation sheet. First separate the reporting window, working-day adjustment and product category. Then ask which company disclosure can confirm the operating implications. A national customs total is useful evidence of cross-border activity; it is not a company's recognized revenue.
This edition checks the Korea Customs Service's October 21 release and SK hynix's newly published earnings-call invitation. Both dates matter: the trade observation window and the coming corporate reporting event serve different purposes. We omit an unverified overnight market close and make no English-exclusivity claim.
The decision sheet
| Evidence | What it supports | Action before changing a forecast |
|---|---|---|
| The customs release reports a large increase in export value. | Strong activity during the stated observation window. | Keep the raw and working-day-adjusted comparisons in separate fields. |
| Semiconductors account for a substantial export amount. | The sector is important to the trade result. | Request product mix, shipment and pricing evidence at company level. |
| A dated company earnings event is confirmed. | There is a specific next opportunity to test assumptions. | Prepare questions now; update only after the actual release. |
Signal 1 — compare like-for-like activity
For October 1–20, customs reports total exports of approximately $61.6 billion, up 105.4% year on year. The release also reports 65.9% growth in average daily export value, accounting for 13.0 working days versus 10.5 in the comparison period. These are national export measures, not semiconductor-only growth rates. Preliminary customs release.
Why it matters: The daily comparison still describes a strong increase, while giving a different answer to the question of activity per working day. Choosing the larger percentage without its denominator can distort a planning discussion. Calendar effects are part of the analysis, not a reason to dismiss the underlying improvement.
In a working model, preserve the observation dates, unit, preliminary status and working-day convention next to the result. Do not compare the partial October window with a complete calendar month and label the difference sequential growth. A consistent comparison series is more useful than an isolated record headline.
Signal 2 — chip export value is not a product-mix disclosure
The customs summary identifies semiconductor exports of approximately $27.5 billion during the same window. That establishes a sector-level trade amount. It does not, by itself, disclose a particular supplier's HBM mix, memory bit shipments or realized selling prices. Sector amount in the customs summary.
Why it matters: An analyst needs a bridge between national trade activity and the company variables that drive a model. Put volume, price, mix and recognition timing on separate lines. A higher export value can justify more focused diligence without answering every one of those questions.
For supplier planning, distinguish improved sector conditions from your own customer's delivery schedule. Keep committed orders, forecast requests and unconfirmed market opportunity separate. This extends the discipline in our FADU analysis: figures covering different periods and commercial stages should not be merged into a single revenue assumption.
Signal 3 — a scheduled test of the company-level story
SK hynix's October 21 invitation says its quarterly earnings press release will be issued before 09:00 KST on October 28, with the interpreted conference call at 09:00 KST that day. This is an official reporting schedule, not an earnings result or product forecast. Company invitation.
Why it matters: The invitation turns a broad monitoring task into a dated research assignment. Prepare questions on disclosed shipment trends, pricing commentary, product mix and capacity allocation. Ask which statements concern the reported quarter and which concern subsequent demand. Record unanswered questions rather than filling them with a national export growth rate.
Korea-only radar and today's number
English coverage of the trade release is already available, and the earnings invitation is an English primary source. Today's comparison is 105.4% versus 65.9%: raw total export growth versus working-day-adjusted daily export growth. Both belong beside the reporting window. Definitions and figures.
Seven-day calendar — what changes the assessment
On October 28, inspect the actual company release and call materials before updating the operating model. By October 29, revisit our reconciliation sheet and any trade-data revisions. The company event is confirmed; the follow-up review is our own checkpoint. More specific shipment, pricing or mix evidence can refine the company assessment. Without it, retain the sector signal and leave company attribution unresolved.
Silicon Peninsula reads Korean semiconductor filings and company sources, separates documented facts from interpretation, and records the next evidence to check. Reply with a company you want tracked. Our 30-day Korea Watchlist Pilot covers an agreed watchlist; scope is confirmed before payment.
Korea's chip supply chain, read from the source, on Monday, Wednesday and Friday, with additional material alerts.