Bloomberg Economics 以机器学习模型输入 2,200 万笔资料点、涵盖逾 50 项指标（从政治极化、所得不均到油价与人口结构），指出社群媒体渗透率高与人口中位数年龄低，会提高「不平等／失业／贪腐」升级为严重动乱的机率。尼泊尔在 9 月初抗议与镇压后，首相 KP Sharma Oli 于 9 月 8 日辞职，并排定 3 月 5 日选举；马达加斯加则因贫困与基础服务短缺而脆弱，世银资料显示约三分之二人口每日生活费低于 3 美元。
展望 2026 年，模型在 157 国中点名衣索比亚、中非共和国、安哥拉、瓜地马拉、刚果共和国与马来西亚等 6 国的动乱风险上升最快（仅次于尼泊尔与马达加斯加）。青年困境亦在恶化：以「青年失业＋5 年平均通膨」衡量的指标显示，辛巴威通膨 152%、阿根廷 103% 居全球前列；国际劳工组织称全球约每 4 名年轻人就有 1 人既未就业也未就学或受训。菲律宾被 IMF 指为 AI 替代风险约三分之一工作；印尼则有约每 7 名青年就有 1 人失业。
A Gen-Z-led protest wave is destabilizing governments worldwide, driven by soaring rents and living costs, underemployment, corruption, and widening inequality, and amplified through social media plus video-game/anime symbolism. Leaders have already fallen in Nepal, Madagascar, and Bulgaria, while youth-led unrest persists from Indonesia to Peru and Serbia. The Carnegie Protest Tracker counts 53 demonstrations of 10,000+ people across 33 economies this year, the highest since the tracker began in 2017.
Bloomberg Economics trained a machine-learning model on 22 million data points and 50+ indicators (from polarization and inequality to oil prices and age structure). Its results link higher social-media penetration and lower median age to greater risk that grievances over inequality, unemployment, and corruption escalate into intense civil unrest. Nepal’s early-September violence preceded Prime Minister KP Sharma Oli’s Sept. 8 resignation and a March 5 election, while Madagascar’s vulnerability is tied to poverty and weak basic services, with World Bank data showing about two-thirds living on under $3/day.
Looking to 2026, the model flags six countries—Ethiopia, Central African Republic, Angola, Guatemala, Republic of Congo, and Malaysia—as among the fastest-rising unrest risks within 157 tracked nations (behind only Nepal and Madagascar). A youth “misery” gauge (youth unemployment plus five-year average inflation) shows triple-digit inflation in Zimbabwe (152%) and Argentina (103%), and the ILO reports about 1 in 4 young people globally are neither working nor in education/training. IMF estimates roughly one-third of Philippine jobs are AI-exposed, while Indonesia’s youth unemployment is about 1 in 7.
