That’ll do the trick (back-of-the-envelope calculations!). But it won’t necessarily be the Mexicans paying…
First consider the canonical analysis, where the US is considered a small country (in the context of US-Mexican trade):
The tariff t raises the price faced by US consumers, producers, by the entire 4% of the tax. The resulting deadweight (efficiency) loss is the pink area in panel (b). Let’s put some numbers to the graph.
US imports of goods and services from Mexico was $316.4 billion in 2015. If the US exits Nafta, the tariff rate reverts to MFN/NTR rates of about 4%, and the price elasticity of imports is 3.8% (Spearot, 2015, Table 2), then imports fall by $44.6 billion. Consumption side and producer side deadweight loss is about $0.9 billion, while tariff revenues equal $10.9 billion. In this depiction, the entire tax burden is borne by US consumers.
A more realistic interpretation is to assume the US and Mexico are roughly equal size (in trade, not GDP, terms). This is depicted in Figure 2.
For simplicity, assume supply and demand elasticities are the same, so half the incidence of the tax falls on the US, half on Mexico. Then import prices rise by 2%.
Imports fall from $316.4 billion to $294.1 billion. Tariff revenues are $11.8 billion; $5.9 billion of the tariff revenue is borne by US consumers. There is a terms of trade gain to the US, since we force Mexico to sell its goods to the US at a lower price.
“So-called President” Trump has suggested at various instances a 35% tariff. This would drive down imports to $163.2 billion. Total tariff revenues are $57.1 billion — more than enough to pay for building the wall (at $12-$40 billion). Of this amount, $28.6 billion is from American consumers.
All of this is mostly an illustrative exercise. It’s also important to recall that it’s a partial equilibrium exercise. GDP effects are not incorporated (both Keynesian and supply side). Nor are financial side implications, such as the impact on capital flows and exchange rates. And obviously, retaliation is not accounted for…
One important subtlety: the analysis presumes final goods; but possibly 40% of US imports from Mexico are intermediate goods.(p.6). Then it’s important to remember negative ERPs.