Search Results
Working Paper
LLM on a Budget: Active Knowledge Distillation for Efficient Classification of Large Text Corpora
Large Language Models (LLMs) are highly accurate in classification tasks, however, substantial computational and financial costs hinder their large-scale deployment in dynamic environments. Knowledge Distillation (KD) where a LLM ""teacher"" trains a smaller and more efficient ""student"" model, offers a promising solution to this problem. However, the distillation process itself often remains costly for large datasets, since it requires the teacher to label a vast number of samples while incurring significant token consumption. To alleviate this challenge, in this work we explore the ...
Working Paper
News versus Sentiment : Predicting Stock Returns from News Stories
This paper uses a dataset of more than 900,000 news stories to test whether news can predict stock returns. We measure sentiment with a proprietary Thomson-Reuters neural network. We find that daily news predicts stock returns for only 1 to 2 days, confirming previous research. Weekly news, however, predicts stock returns for one quarter. Positive news stories increase stock returns quickly, but negative stories have a long delayed reaction. Much of the delayed response to news occurs around the subsequent earnings announcement.
Working Paper
What's the Story? A New Perspective on the Value of Economic Forecasts
We apply textual analysis tools to measure the degree of optimism versus pessimism of the text that describes Federal Reserve Board forecasts published in the Greenbook. The resulting measure of Greenbook text sentiment, ?Tonality,? is found to be strongly correlated, in the intuitive direction, with the Greenbook point forecast for key economic variables such as unemployment and inflation. We then examine whether Tonality has incremental power for predicting unemployment, GDP growth, and inflation up to four quarters ahead. We find it to have significant and substantive predictive power for ...
Working Paper
Evaluating the Conditionality of Judgmental Forecasts
We propose a framework to evaluate the conditionality of forecasts. The crux of our framework is the observation that a forecast is conditional if revisions to the conditioning factor are faithfully incorporated into the remainder of the forecast. We consider whether the Greenbook, Blue Chip, and the Survey of Professional Forecasters exhibit systematic biases in the manner in which they incorporate interest rate projections into the forecasts of other macroeconomic variables. We do not find strong evidence of systematic biases in the three economic forecasts that we consider, as the interest ...
Working Paper
The Power of Narratives in Economic Forecasts
We apply textual analysis tools to the narratives that accompany Federal Reserve Board economic forecasts to measure the degree of optimism versus pessimism expressed in those narratives. Text sentiment is strongly correlated with the accompanying economic point forecasts, positively for GDP forecasts and negatively for unemployment and inflation forecasts. Moreover, our sentiment measure predicts errors in FRB and private forecasts for GDP growth and unemployment up to four quarters out. Furthermore, stronger sentiment predicts tighter than expected monetary policy and higher future stock ...
Working Paper
Integrating Prediction and Attribution to Classify News
Recent modeling developments have created tradeoffs between attribution-based models, models that rely on causal relationships, and “pure prediction models†such as neural networks. While forecasters have historically favored one technology or the other based on comfort or loyalty to a particular paradigm, in domains with many observations and predictors such as textual analysis, the tradeoffs between attribution and prediction have become too large to ignore. We document these tradeoffs in the context of relabeling 27 million Thomson Reuters news articles published between 1996 ...
Working Paper
How sensitive is the economy to large interest rate increases? Evidence from the taper tantrum
The “taper tantrum” of 2013 represents one of the largest monetary policy shocks since the 1980s. During this episode, long-term interest rates spiked 100 basis points—a move unintentionally induced by policymakers. However, this had no observable negative effect on the overall U.S. economy. Output, employment, and other important variables, all performed either in line with or better than consensus forecasts, often improving considerably relative to their earlier trends. We conclude that, from low levels, a 100 basis point increase in long-term interest rates is probably too small to ...
Working Paper
Are Real Assets Owners Less Averse to Inflation? Evidence from Consumer Sentiments and Inflation Expectations
Using data from the University of Michigan Surveys of Consumers, we document a significant negative association between consumer sentiment and inflation expectations, controlling for prevailing inflation in the economy. We further show that consumer sentiments of homeowners and stockowners are more sensitive to expected inflation than those of other consumers, a disparity at odds with the notion that owning such assets provides hedges against inflation. Leveraging data from the Survey of Consumer Expectations, we find three factors that help account for this difference. First, assets owners' ...
Discussion Paper
Using Big Data in Finance: Example of Sentiment-Extraction from News Articles
There is much discussion and research in finance on using "big data" to understand market "sentiment."
Discussion Paper
Corporate Bond Issuers' Swap Exposure to Rising Interest Rates
United States corporate bond issuance has been elevated in recent years relative to historical standards, reflecting in part accommodative financing conditions at historically low rates.