A Better Way for Finance (and Others) to Handle Missing Data
As much as we’re awash in data, a huge problem for building predictive models is the information we don’t have.
A Better Way for Finance (and Others) to Handle Missing DataInvestors are inundated with news about business, politics, and the global economy—and use it to assess the health of the economy, set prices, and forecast returns. But in this flood of information, what in particular are investors paying attention to?
Carnegie Mellon’s Anisha Ghosh and Chicago Booth’s George M. Constantinides find empirical evidence that regardless of what information investors are considering, their focus manifests itself in price-level and labor-market variables.
The researchers start with the market price–dividend ratio, which is an important indicator of investors’ expectations about future dividend growth and discount rates. Many economists believe that the market price–dividend ratio is highly correlated with aggregate consumption growth, but Ghosh and Constantinides find otherwise—they see it as highly correlated with changes in price-level and labor-market variables.
Ghosh and Constantinides incorporate this observation in a model in which investors learn whether the economy is in recession, recovery, or expansion from changes in two prominent price-level and labor-market variables: the consumer-price index and average hourly earnings of production.
The empirical results highlight how much information these macroeconomic variables provide investors. The model explains the level and volatility of consumption and dividend growth, stock market returns, the price-dividend ratio, and the risk-free rate—and it performs well in predicting future returns, while a model with learning from consumption history does not. The findings suggest that changes in the variables the researchers identify have long-term economic implications.
Anisha Ghosh and George M. Constantinides, “What Information Drives Asset Prices?” Working paper, November 2016.
As much as we’re awash in data, a huge problem for building predictive models is the information we don’t have.
A Better Way for Finance (and Others) to Handle Missing DataNobel laureate Douglas W. Diamond and former Boston Fed president Eric Rosengren visit the Capitalisn’t podcast to discuss the collapse of Silicon Valley Bank.
Capitalisn’t: Silicon Valley Bank—The End of Banking As We Know It?Why did prices for US Treasurys drop along with stock prices in the spring of 2020?
How Financial Regulation Fed a Treasury Bond CrisisYour Privacy
We want to demonstrate our commitment to your privacy. Please review Chicago Booth's privacy notice, which provides information explaining how and why we collect particular information when you visit our website.