Assessment of the fulfilment of the 2022 forecasts

(authors: Tatiana Keseliová, Tomáš Pokorný, Tomáš Šestořád)

The CNB works to ensure a high level of monetary policy credibility, accountability and transparency. Besides regularly assessing the achievement of the inflation target, usually in a box in the winter Monetary Policy Report (MPR), this involves reviewing the fulfilment of past forecasts. The reviews provide feedback on the use of the CNB’s forecasting and analytical system. The core of the system is the g3+ forecasting model, the basic unifying element used in preparing the CNB’s macroeconomic forecasts. The conclusions of the assessments of past forecasts are used to verify the model’s current settings and to consider potential adjustments to them. Such adjustments have also resulted an updated g3+ model, which will be the CNB’s core forecasting model from this MPR onwards.[1]

In this appendix, we first compare the CNB’s forecasts prepared in 2022 with the contemporaneous outlooks of other analytical institutions. We then assess the fulfilment of the CNB’s 2022 forecasts, starting with a comparison of their assumptions at the time (the exogenous inputs of the forecast) and the subsequently observed developments. We then compare the forecasted paths of the main domestic endogenous variables themselves with the data now known. The final passage is devoted to a hypothetical model simulation. It shows roughly what the forecast in the Summer 2022 MPR would have looked like had the subsequently observed, but then unknown, future paths of all the assumptions entering the forecast been known at the time it was prepared.

The 2022 forecasts compared to other institutions

This first section compares the CNB’s 2022 forecasts with other analytical institutions’ contemporaneous outlooks. Chart 1 shows the forecasted and subsequently observed main variables in whole-year terms for 2022 and 2023.

Chart 1 – Comparison of the CNB’s 2022 forecasts with those of other institutions (full-year data for 2022 and 2023)
The data sources are the CNB’s forecasts and the Ministry of Finance (MoF) surveys published in the 2022 Macroeconomic Forecast of the Czech Republic. The MoF surveys are based on the publicly available forecasts of 13 institutions, eight of them domestic (CNB, Czech Banking Association, MLSA and domestic commercial banks) and the others foreign (e.g. European Commission, OECD, IMF). For the purposes of this document, the CNB’s forecasts are excluded from the survey and the MoF’s forecasts are included.

Example: the blue CNB dot corresponding to “Spring 22” in the “GDP for 2022” chart shows the full-year GDP growth estimate for 2022 from the central bank forecast published in spring 2022 (i.e. the Spring 2022 MPR).

Chart 1 – Comparison of the CNB’s 2022 forecasts with those of other institutions (full-year data for 2022 and 2023)

The data gathered reveals that the forecasts of the institutions under review for consumer price inflation took the strength and robustness of the inflation pressures into account only gradually, while the CNB estimated them better overall than the other institutions. The forecasts were additionally surrounded by a high degree of uncertainty linked primarily with the length and impacts of the conflict in Ukraine. This uncertainty was most evident in the spring forecasts, when the institutions were forced to predict whether governments would succeed in securing a substitute for Russian gas. It was in the spring forecast that the CNB was in hindsight very pessimistic in its GDP projection, as – contrary to its expectations – countries managed to fill their gas storage facilities to solid levels ahead of the next winter. In its subsequent forecasts, though, the CNB’s predictions for economic activity in 2022 were only just below the outcomes. For 2023, by contrast, all the forecasts underestimated the effect of the monetary policy tightening at home and abroad, so the institutions under review predicted significantly higher growth than the CZSO estimate.[2] We should note, however, that the CNB assessed the 2023 downturn better than the other institutions, especially in the summer and autumn forecasts. In the case of wages, the CNB’s forecasts in 2022 were on average closer to what transpired than those of the other institutions. For 2023, the forecasts of both the CNB and the other institutions initially underestimated wage growth but subsequently predicted it quite well in comparison with the CZSO estimate now available.[3]

To sum up, the CNB’s forecasts for inflation and domestic economic activity were rather more accurate than those of the other analytical institutions in 2022 (except for the spring forecast for 2022). In the case of overall nominal wage growth, the CNB predicted 2022 more accurately than the other institutions, while the projections for last year can be assessed as comparably accurate.

The CNB’s macroeconomic forecast serves as an important guide for the Bank Board when setting interest rates. The tool used to create the forecast is the g3+ core forecasting model. The baseline scenario of the forecast predicts the most probable future evolution of the domestic economy in the view of the CNB Monetary Department’s economists. The domestic interest rate path consistent with this ensures the achievement of the CNB’s inflation target at the monetary policy horizon.

Assessment of the fulfilment of the 2022 forecasts – assumptions

The key assumptions of the CNB’s macroeconomic forecast for the domestic economy are the outlooks for the foreign environment, fiscal policy and administered prices.[4] Of all the forecasts under assessment, the assumptions of the winter 2022 forecast turned out to be most off the mark, mainly because of Russia’s unexpected invasion of Ukraine, which greatly exacerbated the energy crisis in Europe (see Chart 2). The foreign assumptions of the spring and summer forecasts did include this event but only partially captured its upward effect on energy prices. Conversely, the assumptions of the autumn forecast turned out, with hindsight, to be overly pessimistic. The outlooks at the time were caught out by the mild winter coupled with the rapid sourcing of alternative (non-Russian) gas supplies to EU countries and by faster easing of the global supply chain disruption. This led to a larger-than-expected decrease in industrial producer price inflation in the effective euro area and, from spring 2023 onwards, even to a decline in producer prices, which the outlooks under assessment had not included.

Apart from the winter forecast, which had not anticipated the conflict in Ukraine and its impacts, economic growth in the effective euro area was rather higher than in the forecasts under assessment in 2022. In a broadly solid economic situation, the ECB reacted to the surprisingly rapid rise in prices with a marked tightening, which neither the financial markets, nor the assumptions of the central bank forecasts under assessment had expected. The foreign monetary policy tightening was subsequently reflected in stagnation of the Czech Republic’s main trading partner economies last year, whereas the assumptions of the CNB forecasts under assessment had expected a recovery.

The larger-than-expected rise in energy prices in 2022 led to an underestimated outlook for domestic administered prices on average (see Chart 2). The assumptions of the winter 2022 forecast deviate the most from the observed values. This forecast had not expected war to break out in Ukraine and hence could not have taken its upward impact on energy prices into account. In addition, annual administered price inflation was later affected by the unexpected temporary introduction of an energy savings tariff in 2022 Q4. The tariff was not announced until after the publication of the autumn 2022 forecast and so was not included in the assumptions of the forecasts under assessment.[5] The government support provided to households and firms to help with high energy bills, along with other expenditure induced by the war in Ukraine, led to markedly higher growth in nominal general government consumption relative to the assumptions of the forecasts under assessment.

Chart 2 – Selected forecast assumptions

Chart 2 – Selected forecast assumptions

Assessment of the fulfilment of the 2022 forecasts – key endogenous domestic variables

Except for the winter forecast, which had not included the effect of Russia’s unexpected invasion of Ukraine on GDP growth, domestic economic activity developed better than in the forecasts under assessment in 2022 and early 2023 (see Chart 3). The deviations of the observed values from the forecasts were due mainly to stronger-than-assumed foreign economic activity, reflected in higher exports. The forecasts under assessment had been predicting a recovery in mid-2023, but it failed to materialise due to an unexpected continuation of the stagnation abroad (especially in Germany). This was reflected in substantially lower export activity and weaker growth in household consumption due to worse sentiment, reflected, among other things, in a surprisingly higher saving rate.

Chart 3 – Forecasts of key domestic variables

Chart 3 – Forecasts of key domestic variables

Consumer inflation surged in the first half of 2022 and was thus above the level in the winter forecast at that time (see Chart 3). This was mainly due to cost effects, especially an unexpectedly sharp rise in energy prices linked with the outbreak of war in Ukraine. As time went on and the scale of the economic impacts of Russia’s aggression began to become clear, the forecasts started to converge to the subsequently published figures in the course of 2022. The view of the strength of the demand pressures also became more accurate. The forecasts initially underestimated customers’ willingness to accept higher prices. This even allowed sellers to raise prices faster than their costs were increasing. The summer and winter forecasts largely materialised. The exceptions were the fourth quarters of 2022 and 2023, when the aforementioned temporary introduction of an energy savings tariff played a role. The forecasts under assessment had not expected this tariff (nor, of course, the base effect it caused a year later).

After capturing the impacts of the war in Ukraine, the spring forecast – unlike the winter one – had identified a need for a further, greater tightening of monetary policy, which would have partially eased the elevated inflation pressures and allowed inflation to return faster to the target. However, the CNB decided not to respond fully to inflation pressures beyond the control of monetary policy (after considering, among other things, a simulation featuring a more distant monetary policy horizon than the standard one in the Spring 2022 MPR), so the observed rate path was lower (see Chart 3). For the summer and autumn forecasts, the Bank Board decided not to react with interest rates to part of the inflation pressures, owing to their cost-push and expected temporary nature.[6] Accordingly, the monetary policy horizon in the baseline scenario itself was moved forward by two quarters in the summer forecast and one quarter in the autumn forecast.[7] Owing to the receding inflation pressures at the longer horizon, the implied rate path in the summer 2022 forecast now predicted a decline in rates, although this did not happen. The autumn 2022 forecast had expected a similar easing of inflation pressures overall, but on condition of a further temporary increase in interest rates due to a sizeable upward revision of wage growth relative to the summer forecast. The Bank Board responded to this situation by leaving interest rates higher for longer rather than raising them further and then lowering them.

The koruna exchange rate followed a different course than the forecasts under assessment (see Chart 3). While the winter 2022 forecast had predicted continued appreciation of the koruna, in reality the exchange rate strengthened significantly less, due to disastrously worse sentiment in the region because of Russia’s aggression in Ukraine. The spring forecast had also predicted a temporary appreciation of the koruna on the back of rising rates, but a further worsening of financial market sentiment led conversely to a depreciation. However, it was dampened by actual and verbal foreign exchange interventions by the CNB (which intervened against the weakening koruna in the foreign exchange market in May and October 2022 by selling a total of EUR 25.56 billion of its foreign exchange reserves).[8] In late 2022 and early 2023, the koruna and other Central European currencies surprisingly firmed. This reflected the mild winter and the unexpectedly well-handled energy crisis in Europe. The koruna weakened following the formal ending of the intervention regime last August. This depreciation was also due in part to a narrowing of the interest rate differential vis-à-vis the euro area due to a tightening of ECB monetary policy and, at the end of the year, to the start of the domestic rate-cutting cycle as well. The CNB’s forecasts in the second half of 2022 had expected the decline in interest rates and the accompanying weakening of the koruna to start earlier.

Hypothetical Summer 2022 MPR forecast incorporating knowledge of the actual evolution of the assumptions

The factors-known simulation is a hypothetical version of the Summer 2022 MPR macroeconomic forecast. It tells us what the forecast would have looked like had the subsequently observed paths of the assumptions at the time been used instead of the assumptions themselves.[9] Like the forecast at the time, the hypothetical simulation was prepared with a monetary policy horizon 18–24 months[10] ahead. However, the simulation is not a fully fledged forecast, as it does not contain additional expert adjustments.

Comparing the hypothetical factors-known simulation with the Summer 2022 MPR forecast, headline inflation is initially below the original forecast in 2022 and then above it in 2023 (see Chart 4). The difference is due partly to the aforementioned introduction of the energy savings tariff, whose effects on administered prices were not included in the original forecast. The higher inflation in the hypothetical simulation in 2023 reflects faster growth in nominal general government consumption and higher administered price inflation. In the hypothetical simulation, the central bank reacts to this outlook by temporarily raising interest rates to return inflation to the 2% target at the end of the period assessed. To achieve this, tighter monetary conditions are needed over the entire horizon of the hypothetical forecast, as it assumes considerably higher foreign interest rates than the original forecast. A more positive foreign output gap, coupled with an unexpected foreign producer price correction in 2023, leads to a stronger koruna in the hypothetical simulation. Higher general government consumption and slightly stronger external demand foster higher domestic economic growth in 2022. Owing to a subsequent stagnation of external demand, the hypothetical factors-known forecast leads to lower domestic GDP growth than predicted by the authentic forecast under assessment.

Comparing the hypothetical factors-known simulation with the historical outcome, we find that observed inflation was rather lower than in the simulation over the entire forecast horizon (see Chart 4). In reality, interest rates stayed higher for longer, suppressing the additional inflation pressures stemming from the depreciation pressure on the koruna. The stronger observed exchange rate of the koruna was additionally supported by actual and verbal foreign exchange interventions by the CNB, the impact of which the hypothetical forecast is unable to capture ex ante. Both components of the monetary conditions were thus tighter in reality, dampening economic activity. The stronger koruna led to a smaller quantity of goods and services being exported. The higher rates motivated people to save money. This was reflected in a larger decline in household consumption than in the hypothetical forecast. Overall, the factors-known simulation is closer to the historical outcomes than the summer 2022 forecast.

At the same time, the simulation shows that starting the rate-cutting cycle earlier would have led to a smaller contraction of the economy. This would have come at the cost of somewhat higher inflation, which, however, would also have fallen to the inflation target at the monetary policy horizon of the hypothetical forecast (i.e. in the first half of 2024).[11]

Chart 4 – Comparison of the forecast and the factors-known simulation (hypothetical forecast) in the Summer 2022 MPR with the observed data
light-grey area in charts shows Summer 2022 MPR forecast horizon

Chart 4 – Comparison of the forecast and the factors-known simulation (hypothetical forecast) in the Summer 2022 MPR with the observed data

[1] The updated g3+ core forecasting model was described in more detail in an appendix to the Winter 2024 MPR.

[2] The GDP growth figure will be revised by the CZSO simultaneously with the publication of the quarterly sectoral accounts in June 2024.

[3] The whole-year wage growth figure for 2023 will be revised by the CZSO in a news release on 4 June 2024.

[4] The specific indicators considered are industrial producer prices in the effective euro area (broken down into their core and energy components), foreign economic activity (the GDP trend and the output gap in the effective euro area), the USD/EUR cross rate, the Brent crude oil price and the 3M EURIBOR interest rate and its shadow component capturing the ECB’s unconventional monetary policy measures (asset purchases). The domestic assumptions include the outlook for administered prices and nominal government consumption along with its deflator and the fiscal impulse.

[5] This measure led to a decline in the price of energy and reduced administered price inflation in 2022 Q4. The lower base was conversely reflected in a jump in administered price inflation in the same quarter a year later, which temporarily interrupted the downward trend in annual consumer price inflation at the end of 2023.

[6] In the inflation targeting regime, the need for escape clauses (exemptions from hitting the inflation target) derives from the relatively frequent occurrence of shock changes in exogenous factors (particularly supply-side shocks) that are completely or largely outside the purview of the domestic central bank’s monetary policy. The escape clauses include major changes in world prices of raw materials, energy-producing materials and other commodities. Escape clauses are dealt with in more detail in an annex to Inflation Report III/2001.

[7] In both the summer and autumn 2022 forecasts, the monetary policy horizon that the Bank Board focused on when making its decisions was therefore located in the first half of 2024.

[8] At its monetary meeting on 3 August 2023, the Bank Board formally ended the intervention regime announced in May 2022 and resumed the programme of sales of part of the income on international reserves.

[9] The factors-known simulation thus contains information regarding different ex post observed paths of administered prices, general government consumption and foreign variables.

[10] The CNB normally considers a horizon 12–18 months ahead. In the summer 2022 forecast, however, the Bank Board decided not to react to part of the extraordinary cost factors from abroad and temporarily moved the horizon two quarters into the future.

[11] In late 2022 and early 2023, the Bank Board gradually returned to the standard monetary policy horizon. This led to a need for tighter interest rates in the forecasts at the time than in the summer 2022 forecast.