The Biases of Central Bank Research Papers

The 2008 Financial Crisis led central banks all over the world to make use of several monetary policy tools that were considered unorthodox before. Particularly, monetary authorities resorted to three policies. First, central banks started to provide long-term refinancing to the banking sector in order to boost lending. Second, forward guidance (i.e., the communication by monetary authorities of the future path of the policy rate) was introduced to reduce economic uncertainty via anchoring the expectations of economic agents. Finally, central banks expanded their balance sheets by purchasing long-term assets, a policy known as quantitative easing (QE).

These unconventional policies were aimed at overcoming the difficulties to conduct monetary policy resulting from the zero-lower bound, which can be defined as a situation in which the nominal policy rate (i.e., the rate that central banks use to fine-tune the economy) is at zero. The zero-lower bound prevents central banks from stimulating the economy by lowering the policy rate since nominal rates can’t go below zero.

Did these unconventional tools work? The available literature suggests that, overall, they were effective in raising inflation and output. However, a recent study casts doubt about the reliability of at least some of these results. According to a paper by four economists from the University of Chicago, the European Central Bank, and the National Bank of Slovakia, the available research on the impact of unconventional monetary policy is biased.

Particularly, the authors show that papers written by economists working in central banks tend to report more favorable results about the effectiveness of long-term refinancing operations, forward guidance, and QE than papers from academics. Furthermore, those researchers who find greater effects of unconventional monetary policy on GDP are more likely to experience career improvements within central banks.

Although the authors analyze several potential explanations, two factors may be driving these results. The most obvious one is the existence of conflicts of interest. When central banks assess the effectiveness of their own policies, they are acting as judge and jury. As a result, it is not surprising that results of central bank papers are generally in line with their policies. After all, a central bank economist presenting a paper whose main findings go against central bank policies may find it more difficult to get her article published or even to be promoted to a more senior position.

Another plausible explanation is closely linked to the idea of self-selection bias, a statistical bias that emerges when there are systematic differences between participants and non-participants in a research project. In the case in question, there would be self-selection bias if economists who apply for and accept jobs at central banks mostly agree with central bank policies. Were this the case, we would expect research output by central bank economists to endorse central bank policies.

It’s difficult to elucidate which of these two mechanisms prevails over the other. The fact that career improvement depends at least in part on providing favorable evidence on central bank policy seems to support the conflict-of-interest hypothesis. However, the possibility that self-selection bias may be playing an important role in shaping these results cannot be ruled out.

The Bundesbank is a good example that at least one of these explanations holds water. The Bundesbank is known for having consistently opposed ECB’s easy policies over the last decade. According to the above hypotheses, this should result in Bundesbank economists finding that unconventional monetary policies are not as effective as reported by researchers working in other central banks. And this is exactly what the authors show. Bundesbank papers find substantially smaller effects of unconventional policies on output compared to papers from other Eurozone and non-Eurozone central banks.

As pointed out by the authors themselves, these results shouldn’t lead us to dismiss the evidence produced by central banks regarding their own policies. Yet these results should warn us of two things. First, economics suffers from the same biases and conflicts of interests than other fields of knowledge. Second, these biases and conflicts of interest may potentially undermine the credibility of some results.