The unhappy degrowth: predictions and omens
The supporters of happy degrowth are emerging from their chrysalis (they are orthopterans) to accuse this Government of having implemented austerity, and therefore degrowth, which therefore, it seems, has in the meantime become unhappy.
After the mantra of fiscal drag (which I'll discuss in a later post, because I'd be fed up with absorbing a rate of bullshit and slight inaccuracies that are off the scale), the new mantra is this: "in 2026 we'll only grow by 0.8% because the Government has implemented austerity!"
We have already explored what austerity is in another post, where I have given you the trend in net public investments as a key to understanding it, which clearly indicates who, when, where and how implemented austerity:
Incidentally, I suggest you download these data from their original source (if you encounter any difficulties, let me know and I'll explain how), which is this one (the variable is UING and you can select it in the "subchapter" 3.4 "Net fixed capital formation"), because I'm quite convinced that since this non-existent blog brought it into the Debate that no one on the Commission has been going through a bad time, and surely a thorough "revision" of the data could be promoted to rewrite history, a bit like what happened with the OECD's labor market flexibility measures, which, as you may recall, were "revised" when someone showed you that our labor market was already sufficiently flexible and therefore the "reforms" were not necessary (you can find this history here and those unfamiliar with it would benefit from studying it).
Today, let's talk about the new mantra of low growth in 2026, starting with an obvious fact: to know what growth will be in 2026, we'll have to wait at least until 2028. That's how long economic statistics need to consolidate, given that GDP, as those of you older than you know, is a sample estimate, subject to progressive refinements. It's not a big surprise! Just visit the ISTAT website now to see that between the September 2024 and September 2025 editions, the real GDP growth rate in 2023 increased from 0.7% to 1%!
(Excuse me if I leave you to do the calculations as an exercise?)
Not bad, huh?
Incidentally, in the fall of 2022, the European Commission put it even more grimly, predicting that we would grow by just 0.3% in 2023 (you can find it here ). In short, the final but unconsolidated ISTAT figure for 2024 underestimated growth by 0.3 percentage points, while the European Commission's forecast at the end of 2022 underestimated growth in 2023 by 0.7 percentage points.
Having reiterated that I'm only interested in defending the right-wing government (I know them, they're good people, but we're dealing with other matters here), I'd like to say that if the Brussels idiots' forecasts are to be taken at face value, then first of all I don't understand why they're so upset over a respectable 0.8% achieved with Germany and France on the brink of a serious crisis. And above all, I don't understand why no one applauded Giorgetti when, in 2025, it was seen that in 2023 he had achieved 0.7 points more growth than expected!
It's no small feat to triple the growth you've been credited with!
The point, however, is obviously another.
Brussels' forecasting exercises have merely communicative and political relevance; they have little scientific value (they're not replicable, neither the reference database nor the model they were based on is provided), and above all, they're glaringly flawed and subject to considerable ideological bias . It would be appropriate to print them on Merkel's printer and use them accordingly, rather than using them for sterile class-biased polemics. But this is the opposition we can afford with the budget we have (next time I meet Giancarlo, I'll ask him to run a small deficit to buy us a better opposition!), and I must say that, seen from the right, it's no less stupid than when I saw it from the left…
So let's get into the nitty-gritty and analyze the track record of these forecasting exercises over the past few years. To avoid boring you (or myself), I've limited myself to examining those conducted since the blog opened, that is, since the fall of 2011. For those who forgot their calendars at home, we're now in the fall of 2025, and the forecasts just issued, which you can find here , refer to the three-year period 2025-2027. 2025 isn't over yet ( it's not over until it's over ), so we need to estimate it too: the Commission estimates it at 0.4%, like the IMF , while the OECD estimates it at 0.6% , and we'll know how it fared in 2027 (in September, when perhaps I'll have returned to a normal life). To assess the overall reliability of these forecasts, I downloaded all the reports starting from the one from autumn 2011 , which therefore forecast the years from 2011 to 2013 (you can find the archive here ).
I calculated the usual descriptive statistics for forecast quality (mean error, mean absolute error, root mean square error) for two forecasts: the cumulative one for the three-year period, and the one for the year immediately following the year in which the financial year is published (hence the one-step-ahead forecasts: the one for 2012 made in 2011, the one for 2013 made in 2012, etc.). Naturally, I considered the three-year period because I've often heard it mentioned by Democratic Party mourners ("we'll only grow 2% in the three-year period…"). The data are shown here:
where each set of forecasts is represented by a red dashed line joining three data points, and the consolidated data is instead represented by the black dashed line.
Let's start with the "one step forward" forecasts.
The situation is this:
In the thirteen years under consideration, there was a significant prevalence of overestimations of growth (negative forecast errors). The average error is -0.6, but as is well known, this statistic says little about the accuracy of the model, since in the average error, positive and negative errors cancel each other out. For this reason, it is better to use the mean absolute error (where errors are all taken as absolute values, i.e., ignoring any negative signs) or the mean squared error (since squaring the error always yields a positive number), of which the square root is taken to ensure it is of an order of magnitude comparable to that of the variable being analyzed, or, better yet, the SMAPE, i.e., the symmetric mean absolute percentage error ( the explanation is here ), which is the indicator used in the famous M-Competitions (famous among those in the know, of course: but tell the truth, how much do you learn here?).
The mean absolute error is 1.7 and the root mean square error is 3.1. Considering that the average growth rate in the period under consideration was 0.5, we can say that the order of magnitude of the forecast error is more than three times that of the variable it is intended to forecast. Naturally, these have been difficult years, so a precaution must be taken: no one could have predicted the crash caused by COVID, so it may be useful to recalculate these statistics eliminating 2020 (not 2021, because unlike the crash, the rebound was predictable!). With this precaution, the mean error essentially approaches zero, the mean absolute error drops to 1.1, and the root mean square error to 1.7. These are still orders of magnitude considerable compared to the average of the phenomenon (which, if we exclude the crash in 2020, rises to 1.3, which means, for example, that the mean absolute percentage error is 85%)! Even using SMAPE, a more appropriate measure in the presence of extreme errors, the results are substantially similar: 85% or 75%, depending on whether or not the 2020 outlier is considered.
To give you an idea of what this means for the accuracy of European forecasts, this:
It is Appendix B of the third M-Competition ( The M3-Competition: results, conclusions and implications , published in 2000), and as you can see the order of magnitude of the SMAPE for one-step-forward forecasts is between 8% and 10%. That is, European forecasts are seven to ten times less accurate than forecasts made with univariate analysis methods.
In short, these forecasts are technically a dump, like everything that comes from up there.
In case you're interested, and since I promised them, I'll also provide you with the descriptive statistics of the cumulative growth forecasts for the three-year period:
And as they say (and it's predictable, considering the lengthening of the forecast horizon), I feel better! The SMAPE is reaching 99%, with glaring fixes abounding, and often ideologically connoted. Look, for example, at the bright future the Commission predicted for us in autumn 2011: in the three-year period 2011-2013, cumulative growth would have been 1.3 (I don't recall anyone tearing their clothes off whining about austerity: I remember a great praise of the Loden coat, of rigor, and of composure…). Instead, it was -4.2, with a forecast error of 5.5 (not bad, huh?). This means that the predicted 2.0, in addition to not even being on the podium of the worst cumulative three-year growth forecasts (the worst were, in order: 0.1% in 2013, 1.2% in 2019 – remember the criticisms of the yellow-red government? – and 1.3% ex aequo in 2011 and 2014), could just as easily be 0% or 4% for all we know (I'm betting on something close to 3%, but take note because we'll have to talk about it again in 2028, at #goofy17).
Do you understand how many shades of nonsense are contained in the dirge of the Democratic Party mourners about the lack of growth?
Well.
I'll return to a point. Having established that an imbecile is so regardless of one's vantage point (and therefore equally so from the right as from the left), I'd like to emphasize, my beloved followers, in my capacity as your ecclesiastical guru and guru, that I wouldn't care much about defending the "right-wing" government, as an intellectual exercise. I'm not Giorgia's press secretary or anyone's spin doctor, and besides, they know very well how to defend themselves when they're under attack (at least as well as how to get into trouble when they're not)! Here's a different problem: intellectual dishonesty and, above all, amateurism really bother me. Hearing predictions babbled about as if they were consolidated data, from someone who barely knows how to define the variables involved, someone who has never made a prediction in their life, nor has any idea of the relevant methods and tools, much less knowledge of the vast scientific literature on the subject, gets on my nerves after a while. It's true that we should be accustomed to this. But this is precisely what we shouldn't get used to, and precisely the opposite is what we should demand, given that in its name we have been harassed beyond human imagination: that only experts should pronounce.
If it was true for opinion leaders with 21,007 followers at the time of the pandemic, it could be true for university professors in the subject with 142,695 followers after the pandemic.
Or not?
This is a machine translation of a post (in Italian) written by Alberto Bagnai and published on Goofynomics at the URL https://goofynomics.blogspot.com/2025/11/la-decrescita-infelice-previsioni-e.html on Sat, 22 Nov 2025 15:46:00 +0000. Some rights reserved under CC BY-NC-ND 3.0 license.