Strange Call Perhaps The Bank Of England Have Got It Wrong

A slightly esoteric topic to pick for this board, but as this involves economics and financial modelling, and I worked in the circles
of trading and risk management in banking, I thought I’d add a view. There are elements of “strange” here, but the Bank of England has a few internal issues to deal with. As does The Fed so I hear. Mainly centred around the economic modelling they use to make predictions and adjust policy.

Some of it Ben Bernanke could be useful at solving, and some, not. You have to bear in mind that Ben is not an econometrist (an expert in economic statistics). He is an economic historian, whose PhD was centred on the the failures of the Fed and the Gold Standard during The Great Depression of the 1920s/30s. In helping out with historical analysis he may be very useful.

Mandate of a Central Bank
The first thing he needs to understand, and will understand, is that the bank has a different mandate to the Fed, or even the ECB. He will already know that, as central bankers know the roles of other ones, and they regularly meet up at Davos / World Economic Forum, G-7, G-20 meetings. They also meet at the Bank of International Settlements (BIS) meetings in Switzerland where leading central bankers will hook up and discuss co-ordinated policies, on issues such as cryptocurrency, economic theory, economic policy and new econometric models.

Back to the bank’s mandate. The Bank of England’s mandate is to deliver price stability (low inflation, currently defined as 2% CPI). Providing that is achieved, the secondary mandates are to support the government’s economic objectives including growth and employment. But it is important to remember that unlike with The Fed, growth and employment objectives are secondary. If 2% CPI isn’t achieved, then growth and employment is not a factor in their decision making. The only thing that may override it is financial instability. If the system is in trouble then the Bank of England will do what it must to stabilise the system.

Faulty Modelling on Inflation
I don’t think it is odd they have called in an outsider. Inside the bank, they are still using outdated models such as The Phillips Curve to make predictions on inflation. The Phillips Curve had its use once upon a time but not anymore. The Phillips Curve makes predictions which have been faulty in this particular crisis. There are also in existence old Keynesian economics models, which are based on “closed loop” economies.

These has been defunct since the 1980s, especially with inflation predictions and links to the money supply, but many economists cling on to it because “It’s the best model we have got”. Since the 1980s, we have worked in “open loop” economies due to globalisation. But modelling global “open loop” economics is the holy grail, as to analyse global economies you need global data sets of economic variables. At the moment the central banks don’t have such mega data sets in full operation. So it makes sense that some peer review of each other’s work happens at least. That said, The Fed were equally guilty in misidentifying the spectre of high inflation, so the degree of how much Ben Bernanke can help is questionable.

Career Risk - Nobody Speaks Out
Why is the modelling faulty is something that fascinates me. A bit like scientists whose modelling is faulty, they cannot use anything else as an alternative if the alternative doesn’t exist, and so they cling behind the safety of institutionally backed existing modelling to protect their careers - even if they know the modelling is wrong. I don’t think this is sensible, because if a thermometer doesn’t work, a person of common sense would say “Stop using this, it doesn’t work and needs changing”, rather than “Keep using it, we have no better option”. Sometimes, academics lack common sense and get too hooked up on academic theories, and quantitative statisticians and economists can get too hooked up on precision of the decimal point, and forgetting the broader points.

Also, speaking out against the orthodoxy risks the attack of the wolfpack - a career risk many professionals won’t take, even though they know the existing models are insufficient. Doctors do it. Lawyers do it. Scientists do it. Teachers do it. Fund Managers do it. Economists do it. Hunker down. Back your tribe. Close rank. Yet we know these economists are incorrect because they failed to see the rise of the Long Term Capital Management crisis of 1999, Dotcom crash, Great Financial Crisis, Greek crisis, the effects of the CoVid crisis, and this inflation episode. Yet even I saw the effects of the GFC, CoVid crisis and this inflation crisis - and changed my portfolio to adjust for my view that inflation was on the horizon. That I can do that, and they cannot, should not happen - as they are higher qualified than me.

Modelling Errors: Qualitative vs Quantitative models
One of the faults I know happens in investment banks and central banks is that PhD mathematicians and economists always insist on good data sets and precision before taking decisions. These are quantitative models, with long time-series data sets, such as Monte Carlo simulations. Unless economists have such large quantitative data sets like this on a subject any decisions are often dismissed as “unreliable” for taking decisions. This style is consistent with the “data is everything” or “not enough evidence” mindset that makes some people think they are clever. There is a huge difference between gathering a lot of data, and the interpretation of that data, let alone the decision-making that arises from it.

Quantitative data and “sufficient evidence” is great in isolation if you’re a lawyer or a data scientist, but not if you work in intelligence services, the military, or assessing political risk, where you need to be aware a lot earlier of a change in environment, or if you are assessing forward-looking risks such as a change in inflation environment.

Using quantitative data for a forward-looking risk is like saying “I need a car to hit me 100 times to conclude that a car crash hurts”. A qualitative model such as a Bayesian model would be better for assessing that risk: “Ah the road is always clear here and I always run across it, but I have a single high risk data observation that a car is coming towards me at high speed. This tells me I need to do something different”, and so you do. When crossing a road, Artificial Intelligencs would choose the Bayesian qualitative model for crossing a road, not a quantitative model.

Bayesian Models with Inflation
So a qualitative model such as a Bayesian model, used alongside quantitative models and historical model would be a better combination for assessing inflation. When building a risk management model with a few advanced mathematicians for a bank in 2013, I was taught by a superb risk manager (who was a former trader) at that bank, to combine “quantitative models” with “qualitative models” (Bayesian models) and “historical analysis” for good decision-making. It enabled the bank to adjust its portfolio to changes in interest rates, inflation, stock market crashes, bond market crashes, war shocks and pandemic outbreaks, using qualitative and historical analysis to compliment its quantitative analysis. A 360 degree view.

A Bayesian model only requires a single large unusual observations, or a few unusual observations of abnormal activity to trigger alertness to a change of decision.

Historical Models with Inflation
Historical modelling is also useful. So how did I know that inflation was not temporary, and was going a lot higher, when the Bank waited until the inflation data went up and up and up before they took action.

Well as someone who studied economic history, as well as economic theory, I knew from history that after the second world war, that global supply chains were damaged and inflation let rip for many years. It was a supply shock. So I knew that prices would shoot up after the Ukraine war due to grains, gas and oil markets being affected, and this would have second round effects in clothing, processed food and white goods produced using plastic. I had no data, but I knew enough from history, and knew my supply and demand economics theory well enough. In the Bank, and a few investment banks, historical analysis would not suit economists, and a lack of data from the second world war would also obstruct such a conclusion. Yet it was the correct decision.

I also knew that globalisation due to global supply chains had driven inflation down since the 1990s which is why interest rates have fallen. For reasons of US-China conflict, US re-shoring / friend-shoring, US-Taiwan semi-conductor shortage, China lockdown, Iran-Suez crisis, global Co-Vid lockdowns, I had a second reason for believing that supply would be hit. But I had no plentiful data for that - just enough observations to tell me that de-globalisation of supply chains were working in reverse. Simple cause-effect reversal. Without big quantitative data sets to prove it, this conclusion would not pass the econometrists in a major bank or a central bank. But using a historical model helped make me correct.

Decision-making systems: Can it ever be perfect?
So decision making and the systems behind it is what will be in play when Ben Bernanke
looks into this. In my view they need to move to a combination of historical and qualitative analysis to complement quantitative analysis to improve their inflation forecasting. But there can never be a guarantee you will always get i perfect. Using a combination of the above will make it better, but not perfect.

But what you need them to do to do a perfect job is to have the ability to model a global economy in real time, and with precision, to take near-perfect decisions. I don’t think you can do that.