Interview of Aleš Michl, Governor
By Christopher Jeffery (Central Banking 19. 12. 2024)
CNB Governor Aleš Michl discussed policy mistakes, targeting M2, diversifying into equities and gold, and ending interest on minimum reserves in an interview with Central Banking journal.
The interview was conducted on November 15, 2024 and was published on the Central Banking website on December 19, 2024.
In your first public speech after taking office in July 2022, you promised to bring inflation down from 18% to 2% in two years. Given the external factors that were outside of your control that were feeding inflation, that was quite a commitment. Why did you do it? And were you worried that if you failed, the experience could result in a loss of reputation for the CNB?
It was essential for us to set a clear, ambitious goal for both the public and the markets, demonstrating the Czech National Bank’s strong commitment to restoring price stability. Yes, I was aware of the risk that we might not achieve this goal, but I considered that risk to be very small. My team and I conducted an analysis of demand in the economy. At that time, flow variables indicated a declining trend of demand and the savings rate of households was high.
As a result, instead of implementing another rate hike, we committed to maintaining rates higher for longer at 7%. We also introduced a new strategy to strengthen the koruna, aiming to reduce the cost of importing expensive raw materials. Additionally, we benefited from a gradual decline in raw material prices, which further supported our efforts to reach our inflation target.
At that time, the CNB’s dynamic stochastic general equilibrium model indicated that the central bank should raise interest rates from 7% to more than 10% to lower inflation expectations and inflation. But you ignored the DSGE model’s projections. What was your problem with the DSGE model and what reinforced your confidence in ignoring its projections?
Models are not yet perfect. We don’t have something like Skynet from Terminator, though I’m not sure I’d even want it. In the automotive industry, there’s a concept called ‘automation complacency’, where an overreliance on automation can lead to neglect in monitoring, potentially resulting in performance errors. Often, the best approach is to step back, apply judgment, and adapt to specific circumstances. Ultimately, we chose to ignore the DSGE model’s recommendation, a decision that has since proven correct. Here are three concrete reasons why:
First, DSGE models largely underestimate the role of money, which is why they failed to predict the recent inflation surge – and similarly failed during the Lehman Brothers crisis. I asked myself the questions: ‘Can monetary policy truly be conducted without considering the variable ‘M’?’ I have serious doubts, especially from a monetarist perspective. Interest rates had been near zero for a decade before Covid-19, and the goal was to stimulate inflation; expecting a sudden rate hike above 10% to work miracles struck me as unrealistic.
Second, when our repo rate was already at 7% and the European Central Bank rate was near zero, companies increasingly began borrowing in euros – a channel the model didn’t account for. This initially weakened our monetary policy transmission, but as the ECB began tightening, borrowing costs for these companies rose, further slowing demand.
Third, DSGE models are designed to function around a steady-state equilibrium, but we were far from this. The model assumes high inflation expectations would boost consumption in anticipation of rising prices. However, in reality, real consumption was falling sharply as inflation increased. Additionally, our team’s calculations showed inflation expectations in our country were adaptive rather than forward-looking, contrary to the DSGE model’s assumptions.
You also declared a ‘policy for a strong koruna’. This involved a commitment to keep rates high for longer (but not to raise rates as indicated by your DSGE model). How did you make an assessment about the effectiveness of this approach? Why did you not raise rates in a bid to make the adjustment quicker?
Quicker? Absolutely not. The model’s mechanism simply didn’t fit this specific situation, and I believe I’ve demonstrated that clearly. Here’s another way we assessed it: inflation in our country is a function of domestic prices, the exchange rate and import prices. To address the first variable, a higher-for-longer rate approach was sufficient, as domestic demand and money creation were already declining. For the second variable – the exchange rate – we aimed to keep it as strong as DSGE models largely underestimate the role of money, which is why they failed to predict the recent inflation surge possible, which is why we introduced the new ‘strong koruna’ strategy. This also affected the third variable by making imports cheaper.
This approach not only reduced the cost of importing then-expensive raw materials, but also applied necessary pressure on exporters, helping to contain inflationary pressures domestically. Importantly, as inflation began to decline, inflation expectations followed, which raised ex ante real rates and further tightened monetary policy.
Communicating your message to both the market and the public was critical. What tactics and channels did you use – including the ketchup bottle – and, with hindsight, what worked and what didn’t?
In my living room, I picked up a bottle of ketchup, recorded a video on my phone, and posted it on my X [formerly Twitter] account. This marked the beginning of a media campaign aimed at lowering inflation expectations and helping the public understand that part of the inflation caused by supply bottlenecks was temporary. My fantastic fellow board members were also actively involved in communicating with the public.
Through interviews with Czech tabloids and posts on social media platforms, I used the analogy of a new ketchup bottle: when you turn it over, nothing comes out at first because the ketchup is stuck in the narrow neck. Then suddenly, a lot comes out at once, and eventually, it flows smoothly. This illustrated that, just as the ketchup eventually flows normally, we anticipated that supply chain issues would resolve themselves and prices would stabilise.
While I don’t have precise data to confirm which aspects of this campaign were most effective, the ketchup bottle even became a meme, and in today’s world, memes on X are powerful tools of communication. It really helped.
The koruna gained against the euro in 2023. How did this occur without any CNB interventions? The stronger currency made imports, including energy imports, relatively cheaper. But, if rates were even higher, couldn’t this effect have been even more pronounced?
We achieved the strongest koruna against the euro in history by spring2023 – without any interventions. This is significant because I’m not an advocate of exchange rate manipulation, which sets me apart from my predecessors. The market understood and trusted our strategy, which we communicated openly and transparently, and that was enough.
Sometimes, less is more in monetary policy. It’s more effective to maintain a steady, restrictive stance than to keep interest rates at zero for a decade and then expect inflation to be controlled with a large rate hike. That’s my logic. The yield curve also plays a key role; while we set the 14-day rates, it’s essential to maintain higher three- and five-year rates. At that point, further increases in short-term rates wouldn’t have significantly impacted medium-term rates, because the DSGE model recommended a quick increase followed shortly by a rapid decrease. My logic here is simple: we must avoid repeating the period of ultra-cheap mortgages and near-zero money rates.
Everyone talks about ‘data dependence’ these days. You decided to monitor the three-month moving average of monthly changes in the price index to ensure more sensitivity as to how inflation was changing versus the typical annual figures. Why did you do this and what data are you looking at today?
Data published today reflects the past, especially when we look at year-on-year inflation, so I prefer to monitor core inflation momentum and trends in various defined aggregates of M, particularly the three-month average of M2. These measures, combined with leading indicators, reveal patterns in domestic demand and foreign demand and, where relevant, core inflation.
You also developed an index of high-frequency indicators, such as electricity consumption and Google searches for unemployment benefits, etc, to determine demand in the economy. How well did these indicators work versus traditional indicators? Did the traditional indicators match up to the high-frequency indicators once the relevant data eventually became available?
We published a paper on this, and now the index is even available on Bloomberg, with its own ticker. It’s called Rushin, an index of Czech economic activity named in honour of Alois Rašín, a prominent economist and politician from Czechoslovakia’s founding era. His name is pronounced ‘Rasheen’ in Czech, and the modified term pays homage both to Rašín [a symbol of authority] and to the timeliness of the index. I even keep a picture of him in my office, and when someone from the bank requests excessive spending, I point to it and say: No. Austerity is necessary, just as Rašín advocated.
The index recently has performed very well. For example, in the second week of October, it showed the economy growing by 0.36% quarter-on-quarter, and two weeks later, the Czech Statistical Office’s flash estimate reported 0.3% – closely aligning with our projection. There are periods when it doesn’t provide as precise a forecast, so, as a model, it remains one of several useful tools for decision-making. Our analysts have incorporated the index into their nowcasting models and use it regularly as a reliable tool. It is one tool for decision making.
In the CNB’s fight against inflation, you frequently called for wage restraint. How important was it to reinforce this by freezing the salaries of board members and restricting staff wage rises to 4.5% – well below inflation at the time?
We needed to lead by example, demonstrating the same discipline of ourselves that we were asking of others. Wage restraint was essential to managing demand pressures and preventing second-round inflation effects. We communicated clearly that limiting wage growth was necessary for effective inflation control. This was when inflation was around 16% to 17%. By freezing board members’ salaries, reducing the number of ‘B-1’ [one level down from the board] managers from 17 to 14, and cutting overall staffing by 5.1% in 2023 – marking the first major job cuts at the CNB in a decade – we set a strong example. My goal is to stay at this level. Additionally, by capping staff wage increases at 4.5%, well below inflation at the time, we reinforced the credibility of our strategy and underscored the importance of restraint across the economy.
What gave you confidence to start cutting interest rates at the end of 2023? And how satisfying was it when annual inflation fell back to target earlier this year?
When we began cutting interest rates from 7% in December 2023, as our analysis of inflation momentum indicated that underlying inflation was stabilising near our 2% target and money aggregates, M and M2, weren’t growing significantly either. Our leading indicators, along with the Rushin Index, suggested the economy was close to a recession. This combination gave us confidence that gradual rate cuts models and use it regularly as a reliable tool would be appropriate. We also made it clear that our stance would remain as restrictive as necessary – and it still is. Now rates are at 4%,well above some neutral rate, if you believe in neutral rates, which are hard to calculate.
Do you believe other central banks may be unwise not to be looking more closely at monetary aggregates?
I cannot comment about other central banks. But I did believe something was missing in our models. At the time of the Lehman Brothers collapse, it was the real estate market. Right now, it was monetary aggregates. Imagine you have 10 years of low interest rates, so you create money via new loans and mortgages. So, you just have to follow the money to understand the inflation.
When you took office in 2022, inflation was higher in the Czech Republic than in pretty much every other European Union nation. Why was inflation more elevated?
At the beginning of 2017, we at CNB made a mistake by creating an excess of money in the economy. The CNB massively sold koruna and bought euros to stimulate inflation. It was a paradox at the time. But this led to a significant increase in liquidity and this, combined with a public finance deficit, contributed to higher inflation – the highest core level in the EU, and beyond the problems that everyone faced with higher energy prices. We acknowledge this error, and while it’s in the past, it’s essential to learn from it and ensure it doesn’t happen again. Dwelling on it further won’t change the outcome; I can’t change the past, so I am focusing only on the future.
You have mentioned that massive selling of korunas to buy euros by the central bank in late 2016/early 2017 resulted in “free liquidity” in the banking sector almost doubling. You also noted that while this had little impact on the CPI, it drove up asset prices. What problems did this cause?
It’s very important to distinguish between time horizons. When liquidity in the banking sector is drastically increased, it may not immediately lead to higher CPI – especially in 2018/19 when banks, which want to earn some money, channel that liquidity into mortgages, which drives up property prices. However, in the medium and long term, these effects inevitably feed into the CPI, impacting overall inflation. The CNB has since taken steps to ensure we don’t repeat such monetary policies.
You have also mentioned that Milton Friedman’s claim that a restrictive monetary policy does not depend on the support of fiscal policy may not hold when there is excess bank liquidity. Can you please explain your thinking here, as well as the implications for the neutral interest rate?
Friedman’s claim was based on conditions with appropriate levels of bank liquidity. However, we are now in a high-inflation environment with excess liquidity. That is different to Friedman’s condition. Now, raising rates no longer constrains bank liquidity, as banks already hold ample reserves. In fact, higher rates add to their liquidity, as we must pay interest on these reserves. This increased bank liquidity makes it easier to finance government debt, while higher government spending on consumption further fuels inflation.
Austerity is essential to keep core inflation low, and in a small, open economy like the Czech Republic, it is the only way to keep headline inflation under control. In the future, austerity will be necessary to maintain core inflation slightly below 2%.
Do you believe central banks should also factor in asset prices when considering monetary policy – to have a broader mix of inputs? If they need to control asset prices, how should this be done?
Yes. Keeping real interest rates negative for too long can lead to asset price bubbles and fuel inflation. The policy mix should include, first, maintaining interest rates higher than pre-Covid levels for an extended period, potentially over a decade. Second, it is essential that governments balance their budgets, as failing to do so could lead to a second wave of inflation or asset price bubble. Another cost shock could trigger renewed inflationary pressures, causing asset prices to continue inflating and creating bubbles.
Should central banks also target asset prices in some way?
We created an index that included asset prices. We are monitoring asset prices. But we do not anticipate changing our target or our definition of our target in any way.
Does the CNB have any plans to shift its countercyclical capital buffer, change limits on the debt-to-income and debt-service-to-income ratios, or limit euro-denominated borrowings and derivatives exposures?
I don’t anticipate any changes in the foreseeable future.
The CNB has requested three independent assessments of its models: one on its core model and the interaction between the central bank’s monetary policy and government fiscal policy; another on the model and subsequent monetary policy recommendations; and a final one on the individual parts of the model and whether it is appropriate to have only one core model.
Why did you do this and what are the conclusions so far?
Our models did not fully capture the extent of the inflation surge, during which we experienced the highest core inflation in the EU. Previously, no-one within the CNB had called for such a review; we believed an external perspective was unnecessary, confident in our own expertise. However, an outside view can be highly valuable. Therefore, for the first time in CNB history, we commissioned three independent assessments to refine how we interpret model recommendations, manage uncertainties, and address the unique characteristics of the Czech economy within our framework. We published all these assessments transparently in mid-November.
We are still reviewing the results. But the assessments recommend that we reduce our reliance on a single model, explore alternatives to DSGE models, and expand economic research within the bank. These findings will guide our priorities as we implement these changes. We will start to develop some new models that can challenge the universal DSGE model and will establish a dedicated research team – one that thrives on challenges like this – to elevate our knowledge and capabilities at the CNB to the next level. That will be a priority.
Are you going to move to a semi-structural model, and potentially an agent-based model in the future?
One of the recommendations was to move to a semi-structural model. Another was to create one more model and then look at two results. I believe there are just three central banks in the world – and we are one of them – that uses just one universal DSGE model. It makes sense to look at other tools. We will reduce our headcount in some other departments and hire some new people, including from overseas, to do the research for us.
You have said you want the CNB to be profitable, to make up for past losses and, ultimately, pay a dividend to the state. On the liability side, the largest component is the liquidity held by commercial banks, including their minimum reserves. The CNB stopped paying interest on minimum reserves last year but does pay interest on excess reserves – what saving has this generated and did it cause any issues?
Our priority is low inflation and financial stability. However, I bring a background as an investment strategist with a PhD in economics, and I’ve built an investment fund before joining the CNB. I hate losses and like profitability – that’s in my DNA. We expect losses, but how could we move to address that? We aim for the expected return on our assets to eventually cover our liability costs. We’ve increased the expected return on the asset side, through investments in equities and gold. On the liability side, we’ve reduced costs by raising the minimum reserve requirements and setting a zero interest rate on these reserves. At the moment, overall, this reduces our costs of conducting monetary policy by about 10% yearly. There is not a lot more I can do, as 80% of the liabilities are liquidity and I cannot change the market rates, 20%are banknotes and then there are requirements, etc.
Are you likely to increase the zero-interest element?
We do not plan any other changes at the moment.
Are you concerned that unremunerated minimum reserves requirements may impede bank capital accumulation, restrict their credit provision, weigh on macro-financial stability and potentially reduce the predictability of monetary policy transmission?
The real risks lie elsewhere. Currently, our commercial banks earn about 40% of their interest income from the central bank. The risk is that banks are becoming complacent, lending less to the productive sector and focusing instead on government bonds and real estate lending. This shift could lead to asset price bubbles, and, in the long run, it may slow potential growth and contribute to higher inflation.
On the asset side, the CNB’s FX reserves now stand at around $150 billion. You have increased the CNB’s gold holdings from 8 tonnes to 46 tonnes – and 100 tonnes of gold is your goal. You also want to nearly double the CNB’s equity holdings to 30%. What are the expected diversification and income benefits from these changes?
We published a research paper on this topic. The findings suggest that if asset prices follow the patterns of the past 20 years, doubling our equity share would increase the expected return on the portfolio, with only a minor increase in volatility. Additionally, raising our gold holdings to 100 tonnes from almost zero could further enhance the expected return, while reducing our portfolio volatility, when measured in Czech koruna. And it is a great diversifier. Most of our gold is held at the Bank of England. It is very liquid as you can click to sell it and get the money immediately. We plan to implement the changes within the next three years.
Does the relatively high price of gold at more than $2,700 per ounce and record price levels for equities give you any pause for thought?
Of course, it would be better to buy gold at a lower price. But we are buying gold to diversify our foreign exchange reserves. We need it for the long term. We have a never-ending investment horizon as a central bank. That is why we are not timing the market. We are gradually increasing our stock holdings to enhance the expected income in our portfolio.
Can you explain why the CNB ended its holdings in renminbi assets?
We sold our holdings in renminbi-denominated government bonds because US bonds offered higher yields, and geopolitical risks surrounding the renminbi were increasing. This doesn’t mean we have exited renminbi investments permanently; if the right opportunity arises, we may reintroduce them into the portfolio. Our approach is not to make policy through our investments; our criteria are strictly based on portfolio strategy, focusing on return, risk and liquidity – not politics. While I may be a governor with an account on X, when it comes to managing our reserves, we follow a traditional, conservative and historically proven approach.
Were there any problems selling or repatriating funds when you exited your onshore Chinese government bonds?
We had a great experience. There were no problems exiting those Chinese bonds.
What impact is the election of Donald Trump likely to have on how the CNB manages its liquidity and investment traches of reserves?
None. We do not alter the structure of our FX reserves portfolio based on the outcome of any election.
Banknote counterfeiting increased in the Czech Republic in 2023. Is the CNB planning to take any steps to address this?
In 2023, the CNB identified 2,135 counterfeit banknotes, most of which lacked security features and were easily detectable through visual inspection. When compared internationally, this number is relatively low, especially considering the volume of cash in circulation. However, maintaining the integrity of our currency is paramount. To enhance security, we are considering the introduction of a new protective element for banknotes, which will be discuss at the bank board next year.
How would you assess the use of instant payments in the Czech Republic?
Our vision is to enable people and businesses to make online payments to each other seamlessly. We have achieved this with the Czech koruna, creating a new instant payment system with banks where transfers take just a few seconds. Instant payments are now widely accepted, my barber even uses them. Over the next two years, we’ll work with banks to ensure Czech clients can make instant payments in euros as well, both with each other and with clients in the Eurozone.