Mirae Asset's chairman prescribes tax cuts for a KOSDAQ losing 1.8 firms a year

About 1.8 companies a year is the migration Park Hyeon-joo wants the tax code to stop ^[15][2], which is a trickle rather than a leak, and the figure that would tell you whether it matters is absent from the record: the Korea Exchange's tally of transfers arrives without the market value of the companies that made them ^[22], so a board losing its two largest names a year and a board losing its two most marginal look identical in the count.

The exit arithmetic is harder to wave away. Kakao Games at 384 billion won, HK inno.N at 596.9 billion and WCP at 432 billion raised 1.41 trillion won between them in the 2020 to 2022 window ^[8][20], and the four offerings that have cleared 100 billion won since Fadu's 192.3 billion in the third quarter of 2023 barely cleared it ^[7], so three deals from the old window raised something like three and a half times what four deals have raised since ^[20]. ChangXin Memory's STAR Market listing in July took in 66.6 billion yuan, which Seoul Economic Daily puts at about 13.2 trillion won, or roughly thirteen times the largest offering in the KOSDAQ's history ^[11][19].

The Chinese venture boards are also accelerating: 113.3 billion yuan of combined STAR and ChiNext proceeds in eight months annualises to about 170 billion, some 2.7 times the 2025 full year ^[18]. Park's own yardstick, Nasdaq above $42 trillion against roughly $60 trillion for the S&P 500 ^[6], does not survive inspection, because the report does not say whether either total excludes the other's members ^[23]; the domestic ratio is the sturdier version, with the KOSDAQ at 7.47% of Korean market capitalisation to the KOSPI's 92.53% ^[16], a growth board worth 8.1% of the main board where Nasdaq's total is 70% of the S&P 500's ^[17].

Whether a few points of corporate tax can outweigh the discount a KOSDAQ listing carries depends on the size of that discount, and nothing disclosed sizes it. The reading these numbers support, or rather the more interesting version of it, is that the constraint shows up in offering size rather than in listing address, and a policy aimed at address is priced against the wrong variable. What would break that view: if the four thin deals since 2023 reflect founders declining to list at all while a $3.4 trillion global venture pool ^[14] funds them privately, then venue tax rates are not the binding term and the incentive is a transfer to companies that already stayed. Park's timing claim is separate and testable, that public and private money moving into the semiconductor build-out now is the window ^[3]; the boards in Shanghai and Shenzhen are the ones currently converting it ^[12].