This is the final part of this note (Part I, II, and III are here, here, and here). In this part, we simulate a large shock to core import prices (the “Trump2.0 tariffs” plan) and assess the possible reaction of the Fed. The bottom line is that a 10 percent shock to the level of core import prices is expected to be visible in CPI space (much less in PCE space), although the magnitude of the passthrough is very uncertain. As for the Fed, the shock implies a higher level of the FF rate at the end of the forecast horizon (takes out two cuts); the Fed is not expected to react (no hike) as the shock is temporary but keeps the FF rate higher.
A partial equilibrium analysis
Before showing the FRB-US simulations, we assess the shock in partial equilibrium using impulse response functions (IRF). We stress that what follows is an upper bound estimate for reasons discussed at the end of this section.
Figure 1 shows the IRF of core CPI to a shock of 1% to the level of core import prices (controlling for a set of variables). According to our estimate, 1% shock to core import prices results in about 8bps on core CPI. Therefore, for a shock of about 15-20% to the level of import prices, we would expect the level of core CPI to increase by about 100- 150bps after 3-4 quarters. Please, note that this is a shock to the level of the series, that is its growth rate would not be persistently higher, and we would go back to pre-tariff inflation after the shock dissipates.
Figure 1. IRF of core CPI (QoQ) to a 1% shock of core import prices
The above results are an upper bound estimate. Why? The reason is because the model assumes a full passthrough from core import prices to core consumers’ prices. In other words, it assumes that the distribution margins do not absorb any portion of the shock. But is this realistic? Not really. For instance, Cavallo et al. (2019) studied the Trump1.0 tariffs and concluded that while the passthrough of tariffs to core import prices (the price at the border) is 1, the passthrough to CPI was very, very low. From the paper (page 22): “a 10-percentage point tariff increase on a good is associated with a 0.44 percent increase in that good’s price relative to other goods in the same sector.” (Please note: 0.44% is nothing for a 10% tariff) Put it simply: it was hard to see much in the data, as the margins absorbed most of the shock. Of course, this time can be different but what is clear is that the estimate in Figure 1 is an upper bound, uncertainty is high, and in the end the passthrough can be much lower.
What about in PCE space? The estimated elasticity in core PCE space is about half of what shown in Figure 1. However, in PCE space things are more complicated than in CPI because on top of the retailers’ margins, one must consider the substitution that would be triggered by the tariffs. As a reminder: the PCE price index is an ideal Fisher price index, meaning it does account for substitution by changing the consumption weights every month (the CPI index is a Laspeyres index, so it does not account for substitution). There is high uncertainty right now about the extent of the substitution consumers could operate (i.e. how many goods have local productions in the US? How big are the retailers’ margins for each of the goods that will be subject to the tariffs? What is the consumers’ elasticity across varieties for each good? Etc..). But what is clear is that the final effect on inflation is likely to be lower (or way lower) than implied by a face-value passthrough.
What about in a general equilibrium model?
The IRF above are in partial equilibrium. To assess the general equilibrium, Figure 2 shows FRB-US simulations of a 10% shock to import prices with no retaliation. The red dashed line is the current baseline, while the gray solid line is the constructed scenario. As expected, a shock to import prices is inflationary (although the elasticity of core PCE to core import prices in FRB-US is lower than in partial equilibrium). The counterintuitive part is that GDP growth is higher than in the baseline, precisely because tariffs lower imports volume, and the unemployment rate is marginally lower. Consequently, the path of the FF rate is higher than in the baseline (about 50bps higher at the end of the medium-term).
Figure 2. FRB-US simulations to a 10% shock to import prices
How plausible are the results of the solid gray line? The answer is “it depends on the elasticity of imports (volume) to import prices. Because in the end the elasticity can be lower than expected (that is, it can be hard to switch away from imports subject to tariffs), we added a second scenario (the dashed green lines in Figure 2) in which we halved the FRB elasticity (for the nerds: here the reader can see the “EMO” equation of the model – the lower elasticity scenario halves the second coefficient in the EMO equation. (-.09 from -.18). The PMO equation governing import prices can be seen here). In this second scenario, while inflation continues to be higher, growth takes a hit, as private consumption and business investment (not shown) would be lower (technically, investment is lower because it is (also) affected by the lower GDP growth). In this scenario, the unemployment rate stays higher and the path of the FF rate is marginally more dovish than in the baseline even if inflation is higher (it is a quasi-stagflationary environment).
For the record: the path of the usual financial variables under these scenarios is here.
What about retaliation?
The necessary premise is that to us, it is not obvious how to model “retaliation” in FRB-US. In any case, we decided to model it as a 10% shock to the level of export prices (on top of the 10% shock to import prices). Results are here (and here for financial variables). Mechanically, the “retaliation” shock lowers growth, but the effect is quite small. For this reason, the “retaliation” shock does not alter much the conclusions explained above.
Conclusion
Uncertainty is high. As shown, modeling the Trump2.0 tariffs scenario is not an easy task. The shock can be large, and it would likely show up in core CPI space. But as mentioned, it can be way lower than 100 or 150bps on the YoY; a decent estimate could be 60-70bps, remember Cavallo et al. (2019), and even lower in PCE space. Under this assumption, the Fed would be in no hurry to hike, as the shock is expected to be transitory, but could take out a couple of cuts from the path of the FF rate. We will reassess the implications of the policy if and when the details will be available (including how many goods/sectors will be affected).