Thrilled my paper with @namalhotra is out at @apsrjournal! Have you always been captivated by the impacts of trade policy on political behavior, applications of causal machine learning, and shocks to soybean prices? Ok well regardless this is still the paper for you! A thread 🧵
Just published on APSR First View: “Policy Impact and Voter Mobilization: Evidence from Farmers’ Trade War Experiences”, by Jake Alton Jares and Neil Malhotra. cambridge.org/core/journals/…
7
37
157
36,187
Work on distributive politics has shown that incumbents in various settings steer disproportionate economic policy benefits toward “core voters” who are already supportive. But why send $ to people who are going to vote for you anyways, when you could target swing voters instead?
1
2
4
785
Scholars, politicians, & pundits have suggested that there might not be much of a tradeoff here if sending more $ to supporters increases their political engagement (i.e. greater turnout, contributions). But just how much can incumbents crank up core voter engagement this way?
1
4
363
We combine linked admin data and a salient policy shock to credibly estimate the effects of (better) individual policy outcomes on 2018 turnout and contributions among a core GOP constituency: farmers affected by Pres. Trump’s trade war and his corresponding compensation package.
1
5
340
In 2018, Trump *unilaterally* authorized sweeping tariffs on Chinese imports, triggering retaliatory tariffs that centered on US ag exports. Trump then rushed out the Market Facilitation Program (MFP), which distributed $8.6 billion in compensation aimed at making farmers whole.
1
2
298
To spoil the punchline, better (net) compensation outcomes had large effects on farmers’ propensity to view the MFP as helpful, but didn’t meaningfully boost political engagement. However, that’s not to say the shock didn’t move behavior at all! Read on for details.
1
1
7
492
Our main research design leverages mismatches between commodity-specific trade war harms and compensation under the MFP, as well as the timing of this joint policy shock: farmers were locked into their 2018 crop portfolios before they saw retaliatory tariffs on the horizon.
1
1
4
390
The MFP paid fixed per-unit rates on farmers’ production of each of nine ag commodities. But in the Trump admin’s rush to publicize the MFP before looming midterms, they devised payment rates that (idiosyncratically) undercompensated some crops while overcompensating others.
1
2
278
We quantify disparities using 10 ag econ studies of impact of retaliatory tariffs on prices US farmers received. Big contrast in 2 largest crops. Corn MFP rate amounted to tiny fraction of tariff-induced price decline; soybean MFP rate exceeded the tariff-induced price decline.
1
3
236
Farmers noticed the difference! American Soybean Assoc. responded with press release stating “Soy growers are very thankful.” National Corn Growers Assoc. expressed “disappointment that corn farmers… would receive virtually no relief” & called the MFP rate “woefully inadequate”
1
4
217
We estimate the effects of (better) compensation on individual, non-elite farmers’ attitudes using a February 2019 survey of 693 corn and soybean growers. Respondents were asked “How helpful do you think President Trump’s $12 billion trade relief plan will be to your farm?”
1
3
220
We estimate MFP payments and trade war losses from reported production, and thereby construct three measures of (net) policy outcomes: net MFP benefits in $, MFP benefits as a % of tariff-induced losses, and an indicator for whether farm was made whole.
1
2
218
But is salient (& attributable) variation in individual policy outcomes enough to move political behavior? To find out, we constructed these SAME three policy outcome measures using admin MFP records for 122,157 farms that we were able to link to L2’s national voter file records.
Aug 6, 2024 · 12:11 AM UTC
1
2
5
560
We assess whether better outcomes led farmers to reward Trump & the GOP w/ turnout and contributions in the 2018 midterms. Fortunately, the idiosyncratic nature of MFP’s treatment of particular crops means that policy outcomes are close to orthogonal to past political engagement.
1
3
190
However, to minimize bias we leverage “double machine learning” to estimate a PLR model that squeezes the max info out of individuals’ past histories of general/primary election turnout, contributions, and other rich pre-treatment covariates. (Also relieves us of researcher df)
1
1
4
381
Effects on turnout were very small! Moving a Republican farmer across the interquartile range of net MFP benefits ($391 to $6,110) only increased her turnout rate by 0.3 pp.
1
1
4
329
Altogether, we find that even vastly improved policy outcomes in our setting earned Republicans less mobilization among the targeted population than their campaigns might reap from some of the most economical and standard outreach tactics.
1
1
6
334
Better compensation outcomes also didn’t yield a campaign contribution advantage to the incumbent GOP. There were no meaningful increases in contribution rates to Republicans or Trump specifically; we also did not find any detectable decreases in contributions to Democrats.
1
1
6
374
So altogether, better policy outcomes did not seem to increase engagement relative to a counterfactual of worse outcomes. But maybe that’s not the only channel by which policy shocks affect behavior?
1
3
187
When we use a similar PLR model to compare turnout of farmers to non-farmers with similar turnout histories and demographics, we estimate a roughly 1 pp difference in turnout, which we attribute to the broader experience of the trade war and MFP.
1
3
190
Likewise, comparing farmers to non-farmers, it appears that the rollout of the MFP caused farmers who were prior GOP donors to become 1 pp more likely to donate to Republicans in the following 9 months.
1
3
300
I was surprised by the results of this study – particularly the headline null results. Obviously, this is only a case study of one policy shock, but nonetheless I personally was led to update my priors along two dimensions
1
4
145
First, the well-documented “core voter” targeting in US distributive politics may entail an electoral trade off. In some areas, we may benefit from moving (on the margin) from emphasizing the “electoral connection” towards Hacker and Pierson’s (2014) focus on “policy as prize.”
1
2
5
564
Second, in the modern polarized climate, even very salient economic policy shocks might affect political behavior more through sociotropic / post-materialist concerns rather than through individuals’ own dollars-and-cents outcomes.
1
1
5
323
I am grateful to many academics for their extensive feedback and advice on this paper. However, in this thread, I would like to particularly acknowledge the non-academic experts who spent their free time advising me on this project.
1
3
150
Thanks @nebraskaboydale for explaining the sources of plausibly random variation in year-to-year crop portfolios, as well as all kinds of practical details on the economics of row crop production. You gave me confidence I couldn’t get elsewhere that this project was worthwhile.
1
4
206
Thanks @RobinLinacre and the British Ministry of Justice for developing Splink as open-source software. Record linkage and deduplication via Splink was critical, and I’m very grateful for Robin spending his free time on a Friday night (!) helping me optimize some Splink settings.
5
209

















