Lubos Pastor and Elisabeth Kempf on *Fifty Shades of QE* and the Implications of QE Research

Episode Summary

Executive Summary: The episode discusses Elizabeth Kempf and Lubos Pastor’s meta-analysis of 54 QE studies across the U.S., U.K., and euro area, finding that central bank-authored papers report larger and more significant effects of QE on output and inflation than academic papers. The conversation explores possible reasons—career concerns, institutional incentives, priors, and methodology—and considers what the findings imply for central bank research credibility and future QE evaluation.

Main Topics: Core findings of the QE meta-analysis (Priority: 5/5): The guests summarize their paper comparing 54 QE studies and show that central bank economists tend to find larger, more favorable effects of QE on output and inflation than academics. Incentives, career concerns, and institutional bias (Priority: 5/5): The discussion centers on whether central bank researchers face incentives—promotion prospects, institutional reputation, or policy alignment—that could shape results or language. Theoretical channels for QE effectiveness (Priority: 4/5): They review why QE may or may not work, including complete-markets critiques, portfolio balance, preferred habitat, market segmentation, and signaling/forward guidance channels. Methodological differences between central banks and academia (Priority: 4/5): Central bank studies more often use DSGE models, while academic studies more often use VARs; however, model choice does not fully explain the observed differences in findings. Career outcomes and evidence of promotion effects (Priority: 4/5): The paper finds associations between reporting larger QE effects and better career outcomes inside central banks, especially for more senior staff, suggesting possible career-concern mechanisms. Institutional parallels and external validation (Priority: 3/5): The guests connect their findings to broader issues in institutions like the World Bank and suggest that self-evaluation by public institutions may benefit from more external review. Policy relevance for tapering and balance-sheet reduction (Priority: 3/5): The hosts and guests discuss how the findings relate to current debates over tapering and QT, while emphasizing that the paper is about research bias rather than advising whether QE should end.

Key Arguments: Central bank-authored QE papers report larger estimated effects on both output and inflation than academic papers. All central bank papers in the sample found a statistically significant output effect, versus about half of academic papers. The differences remain even after accounting for methodology, suggesting that model choice does not fully explain the gap. Career concerns are a plausible mechanism: researchers inside central banks may worry that negative findings could reduce promotion prospects. Some evidence supports this mechanism, including stronger correlations for more senior central bankers and weaker QE estimates in the Bundesbank sample. A competing explanation is selection: people who choose central bank jobs may already have more favorable priors about policy effectiveness. QE appears more effective in crisis conditions, especially early rounds like QE1 and the 2020 pandemic response, than in normal times. The paper is not an argument for or against QE itself; it is a warning about how self-assessment research may be influenced by institutional incentives. External validation and independent review may be especially valuable when institutions evaluate their own policies. The observed differences have both statistical and economic significance, raising questions about how central bank research is produced and interpreted.

Data Points: Number of QE studies analyzed: 54 - Papers on QE in the U.S., U.K., and euro area studying output or inflation Central bank authors in sample: 60% - Share of authors primarily affiliated with a central bank Central bank papers with statistically significant output effect: 100% - All central bank papers in the sample reported significance on output Academic papers with statistically significant output effect: 50% - Only about half of academic papers reported significance on output Average peak output effect, all studies: 1.5% - Meta-analysis estimate of QE’s peak effect on output across all papers Average peak output effect, central bank studies: 1.75% - Peak output effect estimated in papers written by central bank economists Average peak output effect, academic studies: 1.0% - Peak output effect estimated in academic papers Median peak output effect, central bank studies: 1.5% - Median estimate among central bank studies Median peak output effect, academic studies: 1.0% - Median estimate among academic studies Peak price level effect, central bank studies: 1.8% - Estimated peak effect of QE on the price level Peak price level effect, academic studies: 0.5% - Estimated peak effect of QE on the price level Standardized peak price level effect, central bank studies: 0.24% - Price level increase from QE equal to 1% of GDP purchases Standardized peak price level effect, academic studies: 0.05% - Price level increase from QE equal to 1% of GDP purchases Standardized cumulative price level effect, central bank studies: 0.18% - End-of-period cumulative effect of standardized QE purchases Standardized cumulative price level effect, academic studies: -0.01% - End-of-period cumulative effect, effectively zero Career outcome association: 0.5 rank - One standard deviation larger output effect associated with about half a rank higher position in the central bank hierarchy Bundesbank sample size: 4 papers - Small number of Bundesbank papers limits statistical strength of that comparison

Pivotal Quotes: "central bank papers report larger effects of quantitative easing on both output and inflation" — Elizabeth Kempf: Summarizing the paper’s main result "career concerns have been shown in many settings to be quite important" — Elizabeth Kempf: Explaining a possible mechanism behind the findings "90-95% of central bank research is not research that evaluates the central bank's own policy" — Lubos Pastor: Clarifying that the study concerns a narrow but sensitive subset of self-evaluative research

Implications: The episode suggests that research on a central bank’s own policies may be systematically optimistic, so listeners should treat QE estimates cautiously. It also points to greater value for external review, transparency, and awareness of incentives when institutions study themselves.