ECB should look out the window more often


On June 14th ECB Vice President Vitorio Constancio gave a speech at Utrecht University School of Economics on the ECB’s negative interest rate experiment. It was a most interesting lecture in which he argued the equilibrium real interest rate (the rate at which demand and supply for capital equalize) in Europe and much of the West may in fact be negative. Secular stagnation due to low productivity growth, labor replacing technological change and sluggish demand may have pushed demand below capacity and the real interest rate at which Western economies achieve full employment at stable inflation rates could then be negative. A most interesting proposition.

But to me that was not the most interesting thing this monetary policy maker said. In passing mr. Constancio said something that was far more intriguing and perhaps scary. There is a lot of criticism on the ECB policy of massive purchases of investment grade assets in the Eurozone. This so called Quantitative Easing is claimed to be unsuccessful because in a balance sheet recession there is no demand for additional credit and making it cheaper for banks to lend out money will simply not lead to more investment, more real expenditure, employment, wage pressure and inflation. That issue is fundamental and has been discussed elsewhere. SFL member prof. Harald Benink and prof. Wim Boonstra, chief economist for Dutch Rabobank, for example suggested channeling the new liquidity through the European Investment Bank (read more here) and others have suggested monetary finance of government expenditure or even outright giving newly created liquidity to consumers directly (helicopter money). All to ensure the extra liquidity actually translates into actual real expenditure. But according to mr. Constancio that was not necessary, illegal and impractical, respectively. Because the current policies work.

On the question how the ECB could be so sure its policies worked, mr. Constancio gave the audience an interesting view into policy evaluation at the ECB. First he dismissed the above criticism that the policy does not result in real expenditure as unfair and based on “just looking out the window”. His argument was that one can not criticize the ECB policies for not creating growth and investment without controlling for shocks that also affected the economy in the past 18 months. Fair enough. That is what we teach all students in econometrics 101. You have to control for all relevant variables if you want to identify the effect of a treatment.

But mr. Constancio’s alternative for “looking out the window”, as far as I understood it from his response, is not much better. According to him the ECB evaluates the success of its policies by simulating complicated models. In these models they create a counter factual to compute the economic outcomes with and without the policy. He stressed the fact that they use multiple models to be sure the measured impacts are robust across these models. It sounded very convincing.

But if you consider how the ECB came up with the idea to implement this policy in the first place, you should wake up and smell the coffee. I did not ask, but I would hope that the ECB ran simulations in these same models to get an idea of the effects of this policy before it decided to spend 80 billion a month. And I would hope they did so in a range of models and for a range of possible scenarios for key exogenous variables like oil price and world trade. But then it should come as no surprise that comparing observed data to a counter factual that is constructed with the same models leads you to conclude that your policy worked. Ok, you will have introduced the true data for exogenous variables (oil price, world trade etc.) in the ex post run, but if these fall in the scenarios you used ex ante, the result is a mere tautology. If you implement policies that work in your models under a wide range of key input variables, you can be sure that excluding those same policies in a counterfactual simulation gives you the same result. All such an evaluation then proves is that the policy works in theory and the models have not changed much in the last 18 months. But what if theory is misguided and the models are misspecified? Should that not be a concern, particularly when you are pushing the economy into unchartered territory with negative real interest rates.

As an economist I love models. And every disproven theory spells another decade of fun for academics. But as a policy maker perhaps occasionally looking out of the window is not such a bad idea after all.