Strategic View: why Forecasters were Wrong on Russian Growth for 2022 and 2023, and why they probably will be for 2024

: With the start of Western sanctions against Russia at the end of February 2022, numerous catastrophic forecasts have been made about the Russian economy by both Western and some Russian economic institutions. However, neither for 2022 nor for 2023 were these predictions vindicated. Multiple errors can be traced to both technical difficulties associated with making serious forecasts in a rapidly changing situation, the nature of model used, and ideological or political bias. Even some Russian institutions demonstrated forecasting errors. The massive economic growth that started in Russia in the second quarter of 2023 obviously surpri-ses economists. One wonders if such a situation will repeat it 2024.


INTRODUCTION
Since the start of the Western sanctions against Russia at the end of February 2022, numerous catastrophic forecasts have been made about Russian economy.The effect of sanctions, or more accurately "measures of economic coercion" taken against Russia since the beginning of the conflict in Ukraine, has been the topic of much discussions 1,2,3,4 .Their effectiveness has been questioned 5 .As a matter of fact, the expe-rience of the first sanction episode against Russia (2014-2017) was not conclusive, and obviously the same applies to the second one 6,7,8,9,10 .
However, another problem then emerged.The "collective West" hoped that sanctions could lead to a massive failure of the Russian economy.The actual reality was quite different with Russia suffering a moderate recession in 2022 and enjoying a strong 2024.Том 4. № 1 https://doi.org/10.21603/2782-2435-2024-4-1-27-46   Сапир Ж. Стратегический взгляд... growth in 2023.This leads to a specific issue, which is quite disturbing for economists: what caused the growth forecasting errors made by both international economists and Russian institutions for 2022, 2023, and 2024?

RESULTS AND DISCUSSION Errors made on 2022
The sanctions taken by the EU or the United States have been significant 11,12 .They were imitated or taken up in part by a number of countries like Canada, Japan, Australia, Taiwan, South Korea, and Singapore.However, they were not taken up by all countries.Whether in Asia, Africa, or Latin America, many countries refuse to apply them, some of them denouncing the war waged by Russia on Ukraine.This list includes China, India, and Malaysia, as well as Mexico, Brazil, countries of the Persian Gulf, countries of the BRICS or candidates for BRICS membership, and Turkey, a country yet a member of NATO.This significantly weakens the scope of the current sanctions.The Western attempt to "isolate" Russia, either diplomatically or economically, has been at best a qualified success and, more probably, a failure.
Doomsday predictions have then been made by many, and especially by a team led by J. Sonnefeld at Yale University 13 .Their reasoning, little or badly sourced, seems to owe a lot to ideology and propaganda.Professor J. Galbraith debunked most of those arguments 14,15 .Similarly, the study carried out at the French Treasury Department by Mrs. Bénassy-Quéré provided some answers but did not avoid the bias of ideology either 16 .An overview of these forecasting errors was made in a paper published late in 2022: a team of colleagues from the Institute of Economic Forecasting of the Russian Academy of Sciences, led by professor Dmitry Kuvalin, studied the large dispersion of initial estimates 17 .
The World Bank then predicted a recession of -11.2% in April 2022, and again a recession of -8.9% in June (Table 1 18 ).So did some Russian institutions, e.g., the Ministry of Economic Development (-7.8% in May), the Institute of Economic Forecasting of the Russian Academy of Sciences, and the Central Bank of Russia.This study showed that the resilience of the Russian economy was largely underestimatedas showed by the actual results in late 2022 with a recession of only -2.1%. .They used partial statistics on household consumption, some of which were collected by private Russian companies.However, these data are extremely fragile.Consumption figures depend on preferences for consumption and savings, i.e., preferences that were naturally modified by the new situation.In addition, the evolution of GDP was also affected by other parameters, in particular, with regard to industry.Finally, they did not take into account the hypothesis of a shift from household consumption centered on individual consumption to consumption centered on collective goods (transport, infrastructure, etc.).This highlights the fragility of the study.Then, another paper supported the estimates computed by the two researchers.This paper relied on satellite data which showed that environmental pollution kept decreasing in some Russian regions in 2022, inferring that it proved a massive industrial recession 25 .
However, the information they published was much less convincing that it sounded.Chemical industry, one of the main culprits for air pollution, was down in Russia up to March 2023 as exports to EU countries went down.The decrease was -6.7% for the second quarter of 2022, -4.6% for the third quarter, and -5.9% for the fourth 26 .This would make for the decrease in pollution detected by satellites.Chemical production, e.g., resin, plastics, coking coal, refined products, Estimates for 2023 were actually following the same path as if no lessons had been learnt (Table 2 27 ,28,29,30,31,32,33,34,35 ).In a paper published in 47 .However, much of this growth took place in the last three quarters and obviously surprised Russian forecasters.Forecasters from Western international economic institutions were wrong about Russia's economic growth in 2023 -more or less the same way as they had been about 2022.
They systematically underestimated the dynamics of the Russian economy (Table 2).This persistence of errors, and frequently serious ones, raises several questions.Nevertheless, the causes of mistakes in growth forecasting seem to be different for Russians and Westerners.
What were the main causes of growth forecast errors?Some forecasting mistakes were certainly technical.
The models used seemed uncapable of integrating sudden changes in trajectory or new econometric relationships between production, investment, and internal demand.This is particularly the case for DSGE (Dynamic Stochastic General Equilibrium) models, frequently used by international institutions and Central Banks 48, 49,50 .These models are quite systematically used for forecasting growth 51 .Still, they are based on heavy microeconomic assumptions 52 .
Agents have behavior patterns they cannot change 53 .
However, all the microeconomic surveys carried out in Russia in 2022 demo nstrated that agents actually did change their behavior and that companies were maximizing production volumes in hope of gaining market share as a condition for future profit maximization.The use of DSGE models then proved to be an obstacle in taking into account the shift in trajectory, and particularly of the Russian economy, but also changes in economic policy 54,55 .
However, technical problems cannot explain everything.Ideology certainly had an impact.It could be traced to the following disbelief, whether temporary or lasting: Western economists almost openly hold the Russian government incapable of initiative or innovative thinking.
These errors can therefore be due to ideological causes 56 .Many ideological assumptions, e.g., "Russians cannot run their economy without us Westerners" or "The corrupt Russian government can neither adapt nor react to sanctions", are widespread in western countries and specifically in the EU administration 57,58,59  The continuity of this ideology can be found in several declarations made by the US and other Western officials in 2023 61,62 .
However, these "errors" could have a political cause, too, in relation to the conflict in Ukraine.They were intended to diffuse a representation of Russian weakness in order to convince Western public that the war would be easy to win, maybe, just by economic coercion, and that no hard sacrifice would be needed to support Ukraine 63 .Thus, when the OECD and the World Bank maintained their negative forecasts for 2023 as late as in June and July 2023 (Table 2), while the data on Russian economic growth had been accumulating since March-April 2023, we are right to ask ourselves which of these errors were technical and which were caused by an ideological or/and political bias.Conversely, the IMF was characterized by a certain neutrality, which allowed its forecasts, with three revisions, to approach the actual growth results.Of course, the IMF changed its forecasts quiteslowly.But it is to be remembered that the July revisions incorporated at best the figures published in June 2023 and even, most probably, only those from May.Likewise, the publication of data for the month of October was most likely representative of the data available for August.The Moscow Institute for Economic Forecasting announced a yearly growth of 2.5% in August 2023, which made the IMF's October forecast (2.25%) entirely realistic or at least not much worse than the Russian estimates.The relative slowness of the process of publishing growth forecasts at the IMF was due, on the one hand, to the scale of the data processed and, on the other hand, to the verification procedures, both technical and political, which necessarily take time.As in 2022, the IMF was characterized within Western economic institutions by the quality of its technical work, which was relatively little influenced by ideological or political issues.

Some Russian errors, too…
Problems concerning the nature of forecasting models also explain why some Russian forecasters were also initially wrong (Tables 3 64 and 4), although to a much lesser extent than Western ones and with a much more rapid reaction during 2023 for the IEF.Two observations should immediately be made regarding the forecasts published by the Central Bank of the Russian Federation.First of all, the variation interval was quite considerable at the start of the period with 690 points in February 2023, followed by 580 points in March, before falling to 350 points in April and June and reaching around 150 points in October 2023.If at the very beginning of the year a large variation interval is not exceptional (usually coupled with a low probability range), it is still infrequent to be of such size.Remember that 680 points are far over an interval going from -2.1% (the actual results for 2022) to +2.5%.This indicates, at the very least, the instability in the model used by the Central Bank, which appears to have been inspired by the DSGE models, but also, most likely, an internal conflict within the forecasting team.
The data from the Institute of Economic Forecasting of the Russian Academy of Sciences (IEF-EAS) showed another problem.This institute used a different model, with no less than 2000 variables and 200 equations 74 .This model was based on a sequential iterative calculation of forecast indicators of economic dynamics in increments of one quarter 75 .The calculation of the dynamics of macroeconomic indicators is carried out in the logic of demand: population, business, and the state.
Demand is formed depending on the income level, as well as the structure and volume of savings of economic entities.According to the estimates of the team running the model, sanctions could be compared to a massive and exogeneous macroeconomic shock, something quite similar (in principle) to the COVID-19 crisis 76,77 .The crisis features induced by the sanctions were, for the period of negative GDP dynamics, associated both with restrictions on the supply of imported products and decisions of unfriendly countries to abandon Russian energy carriers and raw materials.In such conditions, a certain period of adaptation of the Russian economy to the changes in the structure of production, income, and prices was to be required.We understand, too, following the long-standing cooperation with colleagues of the IEF-RAS, that the researchers were wary of any optimism that would not have been really founded.Overoptimistic views are quite severely castigated among Russian economists, and for perfectly understandable reasons.This could introduce a downward bias in forecasts in Russia.If we turn to French institutions like the INSEE (National Institute for Statistics and Economic Studies) or the REXECODE (Research Center for the Expansion of the Economy and Business Development), we will find the same cautious approach, as well as some significant errors in their Available from: http://www.rexecode.fr/public/Analyses-et-previsions/A-noter/Croissance-en-France-de-la-prevision-economique-a-la-realisation#:~:text=Les%20instituts%20ont%20alors%20 effectué,2000%2D2022%20hors%202020 79 Cling J-P, Fayolle J. INSEE's economic forecasts since 1969.Economics and Statistics.1986;192:21-42.(In French)  By the way, there is another factor that has been pushing, for some time, "classical" models towards irrelevancy.It could be suspected, as early as in June 2022, that the Russian economy was much more constrained by supply than by demand 84,85 .Actually, we have seen such a situation develop in Western economies after the COVID-19 crisis.The fact that consumption has been restrained by lockdowns and that supply chains were disorganized by the pandemics created a situation where instant supply became more important than demand to control economies.
Combined effects of Western sanctions and military operations in Ukraine created such a situation for the Russian economy.But, if supply was more important than demand to understand economic dynamics, it had extremely important consequences for modelling growth.These factors have pushed the CEMI-CR451 to develop a much simpler forecasting model in an attempt to overcome difficulties encountered by much more complex ones.This relatively simple model was based on the significantly reduced numbers of relations between usable labor force and labor productivity, as well as on the impact of investments on it and production.This model has proved more robust, at least until the Russian economy settles again into a stable tra-jectory (where a more complex model is definitely superior) and again becomes an economy constrained by demand and not by supply.The low relevancy of classical demand factors could be deduced from the specific situation the Russian economy has faced since the end of spring 2022.Import-substitution mechanisms, demand for military hardware, and the fact that households refrained their demand for some time before increasing dramatically the way it was in France during the COVID-19 crisis -all these factors had 18 months to build up in a situation where supply was the most important factor constraining the economy.It could then be quite easily deduced, from some assumptions of the actual workforce and labor productivity, what the growth would be.
This would explain why the simple model used by the CEMI-CR451 proved apparently so effective.As early as in July 2023, the model forecasted a growth rate of 3.0% (with a variation interval of 100 points compared to 180 points for the BCR model), before settling on 3.3% (with an interval of 60 points in November), then on 3.6% (with an interval of 40 points) in December 2023.Likewise, the CEMI-CR451 forecast of December 2023 of growth for 2024 is 2.4% (with an interval of 140 points) while that of the Central Bank of Russia is 1.7% (with an interval of 150 points), while the forecast given by the IEF-ASR is 2.0%.Remember that the current forecasts from Western economic institutions are of the order of 1.0%.
The model used by the CEMI-CR451 was then effective but not fool-proof.Actually, the model severely underestimated the growth during the first quarter of 2023.It has to be explained what the reason were and why.

A real surprise
There is no dispute that colleagues from the IEF-ASR and the CEMI-451 were surprised by the events 2024.Vol 4. No 1 https://doi.org/10.21603/2782-2435-2024-4-1-27-46Sapir J. Strategic View: why Forecasters... that took place between February and April 2023.The same can be said for the VEB research institute.If the CEMI-CR541 and our colleagues in the IEF-ASR were expecting by the winter of 2022-2023 that Russian economy would return to growth in 2023, it was forecasted that such an event would happen in the third or even the fourth quarter of 2023.The discussions held at the November 2022 session of the French-Russian seminar proved that.We were then all surprised by the strength of this recovery and by the fact that it appeared so early, at least a quarter in advance (for the most optimistic of us), and, maybe, even two.One could think in hindsight that we all underestimated the dynamics of an economy no more constrained by demand, as well as the adjustment capabilities of Russian enterprises.It is also quite probable that the cumulative effects of investments 86 Compiled according to Rosstat data.
were underestimated.In the case of the CEMI-CR451, this surprise was related to two factors: -The growth in the employed population: it increased by 2 million in the first 10 months of 2023 when only 1 million was expected -at least, by the CEMI-CR451.Alexander Shirov said at the June 2022 session of the French-Russian seminar that sanctions would make it mandatory for Russia to increase the number of people employed.However, we clearly underestimated the total number; -The speed with which the labor productivity returned to its level of the end of 2021 and the very beginning of 2022 after a fall of -5% in the second and third quarters of 2022.This was a very positive surprise, but one that made our forecasts to be on the pessimistic side.
Figure 1 86 shows that the increase in employed population had two distinct phases.From March to  This, of course, had a direct impact on the growth of real wages (Fig. 2 87 ).The real wages decrea-  When the activity rate went over 96.5%, the real wages began to increase sharply.When the threat of unemployment disappeared, it led to a strong increase in household consumption that went adding its weight on the activity surge.It was this increase in household consumption that "surprised" economists.However, conditions for this increase were clearly in place by December 2022 -March 2023.Now, it is quite clear that we cannot expect a similar increase in 2024.Employed population seems to have reached its upper limit.Considering the growing immigration flow of employed population, it will be quite low for the coming year.Nevertheless, labor productivity has been growing steadily since March 2023 and is now returning to the level it showed in late 2021 -early 2022.
93 CEMI-CR451 computation and Rosstat. 94China's exports to Russia: Overall trend and key products [Internet].[cited 2023 Dec 22].Available from: https://silverado.org/news/silveradochina-s-exports-to-russiaThe combination of these two factors explains why the final results exceeded the forecasts made in the early spring or summer of 2023 and why the prospects for 2024 are reasonably good.The rapid growth in productivity demonstrated the progresses made by companies, be they private or public, in reorganizing themselves to increase production.This growth proves that the sanctions have faltered much more quickly than expected, having lost a very large part of their effects from the third quarter of 2023.We all expected this rather towards the end of the second quarter of 2024, or even by the end of 2024.Certainly, imports from China had played a certain role, but they could notexplain everything 94 .Still, the situation where the Russian economy was to be constrained by supply and not by demand is not to stay forever.That is why forecasts for 2024 are clearly less optimistic than those for 2023. of Russian economy.In general, economy is more akin to a living organism than to a mechanism, as the late professor Viktor V. Ivanter repeatedly stated 96,97 .This is what makes the reaction capacities of the productive system in times of crisis very difficult to predict indeed.However, for economists, it is always necessary to make forecasts at the risk of being wrong.In this matter, humility and modesty are always required.