Category: Data Source

In 2025, the U.S. raised average tariff duties from 2.4% to 9.6%, bringing protectionism to its highest level in eighty years. We explore the structure of these tariffs, estimate their short-run impacts, and summarize the growing literature on their effects. Across trade partners, the tariffs are correlated with trade deficits but not with geopolitical or strategic industrial goals, other than targeting China. In our baseline estimate, 90% of the tariffs are passed through to tariff-inclusive prices paid by U.S. importers. Incorporating the estimated price and trade responses into a static trade framework, we find an overall welfare impact ranging from a loss of 0.13% of GDP to a gain of 0.10%. These small net welfare impacts reflect sizable consumption losses roughly offset by income and revenue gains, with their sign hinging on whether U.S. terms-of-trade adjusted (on which the data are inconclusive). Among their stated rationales, the tariffs have been effective at raising federal revenue and diverting trade from China. However, it remains uncertain whether they will reduce the trade deficit, lower prices set by foreign exporters, promote manufacturing jobs, increase “friend-shoring” among aligned countries, or reshore key sectors; evidence from 2018-19 and 2025 indicators suggests a narrow path towards achieving these goals.

That is from Pablo D. Fajgelbaum & Amit Khandelwal.  I’ve said this before and I will repeat: if you love government revenue, the tariffs really are not so bad.  The biggest cost of the tariffs is that the government has found a new revenue source, and the Democrats will institutionalize this.  Classical liberals and libertarians have a coherent case against the tariffs, many other people do not, much as you might hear otherwise.

This paper introduces Entangled Time — a novel economic variable representing the simultaneous production-consumption state characterizing human engagement with algorithmic digital interfaces. We develop a formal equilibrium model in which rational agents allocate time to zero-price digital platforms, where their behavioral data constitutes unpriced cognitive labor driving AI capital formation. We demonstrate three principal results. First, under a non-stationary algorithmic resonance state formalized through a Preference Expansion Function, the marginal utility of interface time can be non-decreasing, violating Gossen’s First Law and generating a corner solution (Proposition~1). Second, the firm operating as an algorithmic monopsony facing perfectly inelastic labor supply optimally sets the fiat wage for digital labor equal to zero, substituting monetary compensation with endogenous digital utility (Proposition~2). Third, we define and calibrate Dark GDP — the aggregate value of uncompensated cognitive labor invisible to the System of National Accounts—and show it accounts for a measurable fraction of the secular decline in global labor share (Propositions~7–9). We establish equilibrium existence via Brouwer’s Fixed Point Theorem and propose an empirical identification strategy using privacy-mandate shocks as instruments for data extraction. Three institutional redesigns are proposed: an Algorithmic Monopsony Standard, a Pigouvian Algorithmic Severance Tax, and a Cognitive Depreciation Allowance.

That is all from Nav Vaidhyanathan, who estimates the value of these unpriced services may be in the range of $1.3 trillion.  Here is the easier to follow Substack version.  Speculative, but worth a ponder.

China’s share in US imports at 9% is back down to what it was right before China joined the WTO (2001).

From Gita Gopinath.

Addendum: If you are curious, here is GPT on how much is now shipped through third countries.