Search Results
Working Paper
Artificial Intelligence and Inflation Forecasts
We explore the ability of Large Language Models (LLMs) to produce in-sample conditional inflation forecasts during the 2019-2023 period. We use a leading LLM (Google AI's PaLM) to produce distributions of conditional forecasts at different horizons and compare these forecasts to those of a leading source, the Survey of Professional Forecasters (SPF). We find that LLM forecasts generate lower mean-squared errors overall in most years, and at almost all horizons. LLM forecasts exhibit slower reversion to the 2% inflation anchor.
Working Paper
Artificial Intelligence and Inflation Forecasts
We explore the ability of Large Language Models (LLMs) to produce conditional inflation forecasts during the 2019-2023 period. We use a leading LLM (Google AI's PaLM) to produce distributions of conditional forecasts at different horizons and compare these forecasts to those of a leading source, the Survey of Professional Forecasters (SPF). We find that LLM forecasts generate lower mean-squared errors overall in most years, and at almost all horizons. LLM forecasts exhibit slower reversion to the 2% inflation anchor. We argue that this method of generating forecasts is inexpensive and can be ...
Working Paper
Artificial Intelligence and Inflation Forecasts
We explore the ability of Large Language Models (LLMs) to produce in-sample conditional inflation forecasts during the 2019-2023 period. We use a leading LLM (Google AI's PaLM) to produce distributions of conditional forecasts at different horizons and compare these forecasts to those of a leading source, the Survey of Professional Forecasters (SPF). We find that LLM forecasts generate lower mean-squared errors overall in most years, and at almost all horizons. LLM forecasts exhibit slower reversion to the 2% inflation anchor.
Journal Article
On the Relative Performance of Inflation Forecasts
Inflation expectations constitute important components of macroeconomic models and monetary policy rules. We investigate the relative performance of consumer, professional, market-based, and model-based inflation forecasts. Consistent with the previous literature, professional forecasts most accurately predict one-year-ahead year-over-year inflation. Both consumers and professionals overestimate inflation over their respective sample periods. Market-based forecasts as measured by the swap market breakeven inflation rates significantly overestimate actual inflation; Treasury ...
Working Paper
A Financial New Keynesian Model
This paper solves a standard New Keynesian model in terms of risk-neutral expectations and estimates it using a cross-section of longer-dated financial assets at a single point in time. Inflation risk premia appear in the theory and cause inflation to deviate from its target on average. We re-estimate the model based on each day’s closing prices to capture high-frequency changes in the expected path of the economy. Our estimates show that financial markets reacted to the post-COVID surge in inflation with higher short-run inflation expectations, an increase in the inflation risk premium, ...
Journal Article
What’s the Best Measure of Economic Slack?
Different ways of measuring the economy’s unused capacity, or slack, can result in varying inflation forecasts. Estimates suggest that direct measures of labor market tightness, such as the ratio of job vacancies to unemployment or the rate of employee job switching, provide more accurate forecasts than commonly used measures, such as the unemployment rate or the output gap. Recent elevated values of these measures of labor market tightness suggest greater inflation pressure than is implied by the unemployment rate alone.
Working Paper
Artificial Intelligence and Inflation Forecasts
We explore the ability of Large Language Models (LLMs) to produce conditional inflation forecasts during the 2019-2023 period. We use a leading LLM (Google AI's PaLM) to produce distributions of conditional forecasts at different horizons and compare these forecasts to those of a leading source, the Survey of Professional Forecasters (SPF). We find that LLM forecasts generate lower mean-squared errors overall in most years, and at almost all horizons. LLM forecasts exhibit slower reversion to the 2% inflation anchor. We argue that this method of generating forecasts is inexpensive and can be ...
Working Paper
An Investigation into the Uncertainty Revision Process of Professional Forecasters
Following Manzan (2021), this paper examines how professional forecasters revise their fixed-event uncertainty (variance) forecasts and tests the Bayesian learning prediction that variance forecasts should decrease as the horizon shortens. We show that Manzan's (2021) use of first moment "efficiency" tests are not applicable to studying revisions of variance forecasts. Instead, we employ monotonicity tests developed by Patton and Timmermann (2012) in the first application of these tests to second moments of survey expectations. We find strong evidence that the variance forecasts are ...