Anxiety Rollercoaster

It seems that U.S. Agriculture has been overwhelmed with negative news in the last several months.  Concerns about trade tariffs, renegotiation of NAFTA, and implementation of the new Farm Bill have dominated  discussions in social media and news outlets.  This, in addition to the usual concerns over weather and commodity prices. I wanted to revisit, the sentiment analysis that I performed regarding U.S. agriculture that demonstrated the extreme negative reaction when trade tariffs were first announced back in March.    Scraping the recent news headlines (Yahoo News and Google News) for May 25 suggests that the same concerns remain (Figure 1).

FIgure 1. Word cloud from May 25, 2018 news headlines relating to U.S. Agriculture.

When examined for polarity, this day’s headlines scored 0.149 suggesting neutral to slightly positive sentiment (0.33 is the threshold for strong positive). The polarity scores for news headlines during the last several months ranged from a low of 0.005 on June 8 to a high of 0.39 on June 30 and exhibited extreme volatility during this period (Figure 2).  News headlines such as: “Agriculture caught in US-China trade dispute” , “Mexico Retaliates With Tariffs on US Agriculture” and “Canada to US: Explain that $30 billion farm spending war chest” are what drove the sentiment index to its lowest levels.  In contrast, the prominent stories that help push the index to its highest level had to do with renewed efforts to push a Farm Bill through the Senate.

Figure 2. Polarity Scores for U.S. Agriculture News Headlines.

Sentiment expressed in social media regarding U.S. agriculture exhibits the same volatility (Figure 3).  The longer-term view of twitter sentiment scores pertaining to U.S. agriculture show that for most days during the last several months viewpoints were neutral to moderately positive.  There were three brief periods of extreme negativity.  On or around April 2, 2018 having to do with trade concerns,  on May 17, 2018 in relation to the failure to pass the Farm Bill in the House of Representatives, and on June 20, 2018 as there was heighten concerns regarding the escalation of U.S.-China trade disputes.  Similarly, there were a few brief periods when sentiment moved towards strongly positive (above 0.50).  The monthly trend in the hybrid sentiment score was 0.37 for March, 0.27 for April,  0.34 for May, and 0.45 for June.  All four months were neutral-to-slightly positive and trending towards slightly positive.  Later we will look at alternative sentiment measures for U.S. agriculture to see if this positive trend is further substantiated.

Figure 3. Sentiment Scores for U.S. Agriculture Tweets, March-July, 2018

Most of the tweets had a well defined emotional content., with the most dominant emotion being anticipation (Figure 4).   Despite the high level of anxiety, positive emotions outnumbered negative emotions by nearly 2-to-1. Trust and joy were the most common positive emotions. The predominant negative emotions were fear, sadness, and disgust.

Figure 4. U.S. Agriculture Tweet Emotion, March – July, 2018

Alternative Agriculture Sentiment Measures

The Purdue/CME Group has, for several years, conducted monthly opinion surveys of agriculture producers to gauge economic sentiment.   The results are used to calculate an index called “The Ag Economy Barometer.”  It is designed to be a comprehensive measure of the health of the agricultural economy. In a recent report it was noted that “producers’ weakening perceptions of current conditions in the production agriculture sector, along with a decline in their expectations for future economic conditions” led to a sharp drop in the ag producer sentiment index in April 2018 and the second month in a row of declines.   The report also postulates that “the undercurrent of concern expressed by producers in March became more pronounced in April as the trade dispute with key export customer China continued.”  It was also noted that “the attitude shift identified in the survey extended beyond crops into animal agriculture.”

Looking at the recent history of monthly index values reveals that the largest monthly increase occurred in November 2016 (Figure 5).  Coincidently, a Presidential election was settled during the same time frame.  The May reading of the Purdue University/CME Group Ag Economy Barometer was 141, 16 points higher than in April 2018 and the highest barometer value since January 2017.

Figure 5. Purdue/CME Group Ag Economy Barometer Monthly Index Change, October 2015-June 2018.

To get a sense of what is driving the sentiment indicated in the Ag Economy Barometer, I examined the correlation with monthly indices of major commodity profit developed for Iowa  and the general prices received and paid indices from NASS (Figure 6).  The Ag Economy Barometer is most strongly correlated with the profitability of corn and soybeans and had a strongly negative correlation with cattle profitability.  No relationship was evident with the general prices revived index, while prices paid was negatively correlated.


Figure 6. Correlation Between Purdue/CME Ag Barometer and Commodity Indices

What about monthly changes? There are periods where the Ag Economy Barometer tracks closely with indices for several commodities (March 2016) and periods where they seem to be moving the opposite direction (Figure 7).

Figure 7. Comparison of the Purdue/CME With Iowa Commodity Profit Indices

The Ag Economy Barometer was found to be most closely associated with  corn and soybean profitability (Figure 8).  Corn profitability was nearly four times more influential in explaining the Ag Economy Barometer.   The results of the regression indicated that the model explained 66 percent of variance and that the model was a significant predictor of the Ag Economy Barometer, F(6,25) = 10.85.

Figure 8. Model Coefficients (PAB ~ CORN + SOYB + CATT + HOGS + PREC + PPAY)

Sentiment analysis is the measurement of positive and negative language. It is a way to assess written or spoken language to gauge if the expression is favorable, unfavorable, or neutral, and to what degree.   It provides one approach to discerning opinions of individuals or groups.  When done consistently, it provides a way to evaluate changes in topic views over time.  At least as it pertains to U.S. agriculture, I have demonstrated that sentiment does reflect changes in the underlying economic fundamentals (commodity prices, expenses, etc…) to some degree and that news headlines seem to strongly influence sentiment.  More importantly, sentiment about U.S. agriculture is extremely turbulent, which may not be surprising for an industry that is known for economic volatility.

Leave a Reply

Your email address will not be published. Required fields are marked *