September 2015, A Repeat of September 2013?

  • Tuesday, September 22, 2015


September 2015, A Repeat of September 2013?

Evan A. Schnidman Ph.D.

Executive Summary

+ This commentary utilizes Prattle’s central bank sentiment scores to explore the similarities and differences between the decision to tighten policy by initiating tapering in 2013 and the potential decision to tighten policy by raising rates in 2015.

+ Using a proprietary algorithm trained on central banking language, Prattle’s methodology yields a “score” for every speech, interview, press release, and set of meeting minutes by Fed policymakers and the Committee as a whole. The scores represent the degree to which Fed communications are hawkish or dovish.  

+ Prattle’s data indicates clear similarities between the September 2013 decision to delay tapering asset purchases and current conditions that led into the September 2015 FOMC meeting where policymakers decided to delay a rate hike.

+ While Prattle can successfully forecast the Fed’s policy decisions, our data is even more useful for predicting market fluctuations in response to Fed communications. This is because our scoring methodology is rooted in the history of the market’s reaction to the Fed’s language.

+ Last week’s policy outcome played out exactly as Prattle projected and suggests a December rate hike is the most likely outcome, if history is to repeat itself. However, given the negative market reaction that ensued from December 2013 tapering, Prattle sees a January rate increase as almost as likely as a December hike.

Trends in the Prattle Fed Sentiment Index

The basis of Prattle’s forecasts are the scores our proprietary algorithm issues for every publicly available central bank communication. These scores indicate the level of hawkishness or dovishness of each communication. The time series of scores of all communications are combined and averaged to produce the Fed Sentiment Index: a quantified measure of the Fed’s attitude.

Prattle’s scores are normalized around 0.00 with negative numbers indicating dovish sentiment and positive numbers indicating hawkish sentiment. Most scores fall between -2 and +2 indicating two standard deviations from the norm. Figure 1 shows the raw scores of the Fed Index from January through September for both 2013 and 2015.

Figure 1: Prattle’s Fed Sentiment Index for 2013 and 2015

2013 & 2015

Although these lines follow different patterns, they share at least one common trait: the sentiment scores in September of both 2013 and 2015 are not at their respective year high-points.

A Closer Look at 2013

Prattle’s Fed Sentiment scores from 2013 suggest that sentiment was rising heading into the September FOMC meeting. As Figure 2 depicts, this rise in sentiment had been precipitous since a localized trough in July of 2013.

Figure 2: Prattle’s Fed Sentiment Index for 2013 (January-September)


The rise in sentiment in late summer of 2013 convinced many analysts that the Fed would announce reduced asset purchases at the September FOMC meeting. This talk of tapering was called into question, however, by steep declines in equity markets seen in the late summer. Despite market turmoil, many analysts saw rising hawkishness from the Fed as a sign that tapering was set to begin in September. But this view was challenged by the larger context: a simple look back at Fed sentiment from earlier that same year would have revealed that September’s hawkish mood had yet to match previous highs.

When tapering was first proposed in May of 2013, Fed sentiment had achieved a local peak, briefly averaging above 0.5. Due to a dip in sentiment over the mid-summer months, even though sentiment was rising going into the September meeting it had still not reached the same highs from earlier in the year. Simply, Fed personnel were not expressing the same level of optimism about the economy in September as they had been in May. The Fed Index data from that time period suggested that the Fed was not yet ready to initiate tapering, and that is exactly what happened: the Fed didn’t taper until December of that year.

2015 Looks Like a Repeat of 2013

Looking to a graph of Fed sentiment in 2015, we see a somewhat different story than 2013. Figure 3 depicts that at the start of 2015 average sentiment scores were very positive, above 0.6.

Figure 3: Prattle’s Fed Sentiment Index for 2015 (January-September)


This hawkish position to start 2015, on the heels of ending asset purchases, initiated the discussion of interest rate liftoff. However, just a few weeks later economic conditions deteriorated, and sentiment dropped precipitously. By the end of Q1 of 2015 many analysts were suggesting that cyclical seasonality was to blame for the weak winter data, and this led to a bounce in sentiment entering Q2 of 2015. Concerns over the Greece cut this rebound short, as the threat posed by Greece to the broader E.U. dominated the global economic storyline in the spring and early summer. Sentiment briefly rebounded once again in July as it appeared as though the Greek matter was resolved–at least temporarily. But anxiety over weak Chinese growth quickly squelched that resurgence as the Asian markets became a significant strain on the global economy in the late summer.

Although market turmoil accelerated in late August into September, Fed policymakers expressed increased optimism (hawkishness) around the time of the Jackson Hole conference. Some analysts took this rising sentiment in the face of significant market turmoil as a sign that Fed policymakers were readying themselves to raise rates at the September FOMC meeting. But, just like 2013, this rise in sentiment did not bring the overall trend back to the peaks from earlier in the year when discussions about liftoff began.

Despite the differences in trend between 2013 and 2015, this year’s story is essentially the same: although Fed sentiment was rising, it was not yet high enough to definitively suggest an imminent move to tighten policy, especially in the face of increased market turmoil and downside economic risks. Like 2013, the FOMC appears on track to tighten policy, but, also like 2013, September was out of the question and December is much more likely.


Back in July, when the majority of analysts were predicting a Fed rate hike and the market was pricing in a 70 percent chance of a rate increase in September, Prattle declared less than 50/50 odds of a rate hike in September, putting far higher likelihood on a rate hike in December. When pressed, Prattle estimated the chances of a September rate hike were 30 percent. By the time the FOMC issued their policy release on September 17, markets had come to agree with our assessment.

This judgement about the likely path of Fed policy was not made based on some abstract expertise or opaque model. This call was rooted in our algorithmic assessment of the sentiment of the central bank’s communications.

While this comprehensive, unbiased, quantitative Fed watching data is a valuable tool in projecting policy, it is a far more valuable tool in determining likely market response. This is due to the fact that the primary input in scaling all Prattle Central Bank Sentiment Data is historical market response. In other words, the history of the market’s reaction to the Fed’s language is what Prattle uses to create its lexicon of hawkish and dovish expressions.  New communications that reflect the language of historically hawkish communications are likely to be scored as hawkish as well, and, through machine learning, this lexicon is constantly updated. Therefore, although our sentiment data has proved to be a superb tool for predicting Fed policy, it is even better suited to the prediction of market movements in response to Fed policy.

Nevertheless, the curiosity of the day remains about when Fed policymakers will raise rates. To this question we see two possible answers:

+ The FOMC repeats the path of policy in 2013 and raises rates in December of 2015.

+ The FOMC takes caution from the negative market reaction they caused by initiating tapering in December of 2013, and they wait to hike rates until January of 2016.

Due to current trends in the Fed Index, we put slightly higher odds on a December rate increase than a January one, but in either case we still see our Fed Sentiment Index data as an even better indicator of asset prices than policy. Please feel free to contact Prattle ( for a free sample of historical data to test the investment implications of not only the Fed Index, but also our Sentiment Indexes from central banks around the world


The Prattle Team,
click here to subscribe to the signal blog