关于「跨区套利带:价差能撑多宽,用什么守住」,《购买力平价与多币种尾数定价》给出的定量结论是什么?

结论:套利带是把一件商品从便宜的市场搬到贵的市场的全部成本:运费、汇差、转卖者的抽成、时间与风险、越过免税线的税。它有两条泄漏路线,客人自购受单件运费约束,在近邻市场最危险;代购受自身抽成约束,带宽很难低于四成左右。在本书的样本参数下,牌卡的平均带宽是 37.4%,礼盒只有 32.0%,单价越高,能容纳的价格水平落差越小。泄漏全部来自令吉市场:单靠提价,牌卡要涨到 139.90;限购、只寄本地与提价合用,只要涨到 109.90。数字商品没有运费,分档能否成立取决于对所有客人一致执行的地区检查。分市场定价在一般框架下属于三级价格歧视,守住它的是透明:币种写明,总价不藏费用,同市场同时同价,划线价真实,限购公开。泄漏比例用每周的警戒线监测,堵漏的效果用预先登记的比较检验。

数据口径:书中的汇率与价格水平都是写明的样本假设,不是实时数据;税费、标价与价格歧视的法规只写一般框架,不是法律或税务意见。 第八章让一个虚构牌卡品牌从 49 美元出发走完四步,最终价格是美国 49.00 美元、马来西亚 RM 149.00(被套利带从 RM 129.90 提上来)、新加坡 S$ 59.90、台湾 NT$ 1,080、日本 ¥5,580;典藏礼盒五地美元等值价的最大最小之比只有 1.1344,日常牌卡是 1.4536;模拟的汇率路径让日本的毛利率跌到 0.5433,最先触发复核;在虚构流量下,单个市场要分辨一成的毛利差距需要 97 到 257 周,所以测价只测大的决定,五地合并分析。

English version

对照数据表

下表数字由本章 Python 程序原样打印,可重新运行复现。 程序只用 Python 标准库与结论吻合的打印值: 32.0, 37.4

deck: band_ij in % of shop price P_j (row = buyer market i, column = shop market j)
USD SGD MYR TWD JPY P_i USD
USD -38.9%r 43.3%r 39.5%r 38.5%r 43.99
SGD37.9%r - 33.6%c 39.3%r 38.4%r 37.52
MYR37.9%r 23.1%c - 39.3%r 38.4%r 22.60
TWD37.9%r 36.4%c 42.9%r - 29.9%c 34.18
JPY37.9%r 38.7%r 43.0%r 33.4%c - 39.46
route: c = consumer, r = reseller at its best parcel size
LEAK USD buyers from MYR shop: spread 94.6% > band 43.3% (route r, n200)
LEAK SGD buyers from MYR shop: spread 66.0% > band 33.6% (route c, n1)
LEAK TWD buyers from MYR shop: spread 51.2% > band 42.9% (route r, n200)
LEAK JPY buyers from MYR shop: spread 74.6% > band 43.0% (route r, n200)
mean band over 20 ordered pairs37.4%
box: band_ij in % of shop price P_j (row = buyer market i, column = shop market j)
USD SGD MYR TWD JPY P_i USD
USD -34.4%r 34.9%r 34.6%r 34.4%r 280.00
SGD34.2%r - 25.9%c 31.4%c 32.4%c 255.64
MYR34.2%r 24.1%c - 31.4%c 32.4%c 203.62
TWD34.2%r 30.0%c 33.3%c - 27.7%c 231.01
JPY34.2%r 32.4%c 34.8%r 28.8%c - 255.03
route: c = consumer, r = reseller at its best parcel size
LEAK USD buyers from MYR shop: spread 37.5% > band 34.9% (route r, n200)
mean band over 20 ordered pairs32.0%
查看完整验算程序
# Chapter 5. p51: Maximum sustainable price spread for every ordered market pair, deck and box.
# SAMPLE parameters as in p50 (assumed, not live). For a buyer market i and a shop market j, with shop
# price P_j in USD at the p45 sample fx, the spread ceiling is the cheaper of two leak routes:
#   consumer   one unit, card fee + risk + waiting + parcel shipping + tax over de minimis
#   reseller   best parcel size n in 1..200: unit cost / (1 - fee - margin)
# band_ij = min(route cost) / P_j - 1. A price P_i above P_j * (1 + band_ij) invites a leak.
import math
from fractions import Fraction as F

FX = {"USD": F(1), "SGD": F("1.33"), "MYR": F("4.42"), "TWD": F("31.6"), "JPY": F("149.0")}
LOCAL = {"deck": {"USD": F("43.99"), "SGD": F("49.90"), "MYR": F("99.90"), "TWD": F(1080), "JPY": F(5880)},
         "box": {"USD": F(280), "SGD": F(340), "MYR": F(900), "TWD": F(7300), "JPY": F(38000)}}
SHIP = {frozenset(("MYR", "SGD")): F(6), frozenset(("MYR", "TWD")): F(11), frozenset(("SGD", "TWD")): F(11),
        frozenset(("MYR", "JPY")): F(13), frozenset(("SGD", "JPY")): F(13), frozenset(("TWD", "JPY")): F(9)}
P = {"card": F("0.03"), "risk": F("0.02"), "wait": F("0.003"), "extra": F("1.20"), "handling": F("0.80"),
     "spread": F("0.01"), "fee": F("0.06"), "margin": F("0.10"), "tax": F("0.10"), "deminimis": F(100)}
WEIGHT = {"deck": F(1), "box": F(3)}  # extra-unit shipping multiple; the box is heavier (sample)


def ship(i, j):
    return F(24) if "USD" in (i, j) else SHIP[frozenset((i, j))]


def consumer(i, j, pj, prod):
    days = 12 if "USD" in (i, j) else 7
    tax = P["tax"] * pj if pj > P["deminimis"] else F(0)
    return pj * (1 + P["card"] + P["risk"] + days * P["wait"]) + ship(i, j) * WEIGHT[prod] + tax


def reseller(i, j, pj, prod):
    best = None
    for n in range(1, 201):
        tax = P["tax"] * n * pj if n * pj > P["deminimis"] else F(0)
        unit = (pj * (1 + P["spread"]) + (ship(i, j) * WEIGHT[prod] + P["extra"] * WEIGHT[prod] * (n - 1)) / n
                + P["handling"] + tax / n)
        s = unit / (1 - P["fee"] - P["margin"])
        if best is None or s < best[0]:
            best = (s, n)
    return best


if __name__ == "__main__":
    mk = list(FX)
    for prod in ("deck", "box"):
        usd = {c: LOCAL[prod][c] / FX[c] for c in mk}
        print("{}: band_ij in % of shop price P_j (row = buyer market i, column = shop market j)".format(
            prod))
        print("      " + "".join("{:>9}".format(c) for c in mk) + "   P_i USD")
        leaks = []
        for i in mk:
            cells = []
            for j in mk:
                if i == j:
                    cells.append("{:>9}".format("-"))
                    continue
                pj = usd[j]
                cons = consumer(i, j, pj, prod)
                res, n = reseller(i, j, pj, prod)
                route = "c" if cons <= res else "r"
                band = min(cons, res) / pj - 1
                cells.append("{:>8.1%}{}".format(float(band), route))
                gap = usd[i] / pj - 1
                if gap > band:
                    leaks.append((i, j, float(gap), float(band), route, n))
            print("{:<5} {} {:>8.2f}".format(i, "".join(cells), float(usd[i])))
        print("   route: c = consumer, r = reseller at its best parcel size")
        for i, j, gap, band, route, n in leaks:
            print("   LEAK {} buyers from {} shop: spread {:.1%} > band {:.1%} (route {}, n={})".format(
                i, j, gap, band, route, n if route == "r" else 1))
        if not leaks:
            print("   no leaks")
        print("   mean band over 20 ordered pairs: {:.1%}".format(math.fsum(
            float(min(consumer(i, j, usd[j], prod), reseller(i, j, usd[j], prod)[0]) / usd[j] - 1)
            for i in mk for j in mk if i != j) / 20))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《跨国购买力平价畸变、多币种心理尾数映射与东南亚台日定价矩阵母典》第 5 章「跨区套利带:价差能撑多宽,用什么守住」(版本 0.1.0). 5D Library. DOI 待发布

DOI 待发布 · 专著页正式发布后上线(本站暂无专著页) · 5DL-B-134 · v0.1.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026134,
  author    = {Fan, Yafu},
  title     = {Purchasing-Power-Parity Distortions, Multi-Currency Psychological Price Endings and the Southeast Asia–Taiwan–Japan Pricing Matrix},
  year      = {2026},
  publisher = {5D Library},
  edition   = {0.1.0},
  note      = {DOI pending}
}

本书其他章节结论