As Brad Setser has painstakingly demonstrated we live in a world of de-reservification not de-dollarization. Net claims on the US are accumulated by way of state banks, pension funds, and other quasi-sovereign investors. Meanwhile in value terms, the growth in foreign holdings of US assets is increasingly dominated by equities rather than US Treasury bonds. The dollar today functions far less as a piggy-bank, rainy-day “reserve” currency and more as a vehicle for financial capital accumulation. Some call it a “portfolio currency”. I have dubbed our current era, the era of the "profit dollar".
This is an interesting assessment of the current dollar regime. We are not seeing investors hold dollars as a safe asset but as a place to invest for profit. Return opportunities are high, so market flows have gone into the dollar and then into equity markets.
The next part of this process is the great concern. What happens if equity returns decline? Will the foreign investors move from equities to Treasury bonds, or will the money move offshore? Some capital will likely move into Treasuries as a safe asset, but we’ll likely see money move into other safe havens, which will cause the dollar to decline. No dollar safety.
The paper "Interpretable machine learning for macro alpha: A news sentiment case study" uses the Global Database of Events, Language and Tone (GDELT) with FinBERT, a bidirectional encoder representation from Transformers (BERT), to construct daily sentiment indices that incorporate time dispersion and event impacts. These indices are compared using an XGBoost classifier against logistic regression to predict major currency and bond futures. The study finds that sentiment trends are also useful for predicting asset returns, as shown by Shapley Additive Explanations (SHAP). Domain-specific natural language processing with interpretable ML can add value to any market model and generate macro alpha.
