LIVE | LAGARDE'S HIDDEN MESSAGE: THE ECB'S PLAN FOR INFLATION AND RATES IN THE EUROZONE

Thank you very much, Madam Chair, and I reiterate the pleasure it is for me to be back in this room. I will switch to English, since I see surprised faces at seeing me speak in French. The subject of today's hearing focuses on the economic future of Europe. Artificial intelligence has the potential to transform the way we produce, work, and innovate. Companies are going to devote around 10% of total investment in 2026. And this is the result of a survey we regularly conduct with companies. We are talking about a quarter of lending to companies. AI could significantly improve Europe's productivity, competitiveness, and standard of living. But, and this is the question, it will also affect, to some extent. It is already affecting investment, work, markets, and inflation. That is why it is also important for monetary policy. Europe has a real opportunity to take advantage of this technology, but success is not automatic. We must seize the benefits while properly managing the risks. In my intervention today, I will first talk about the outlook for the euro area economy and explain our latest monetary policy decisions, and then I will talk about how AI may affect inflation and the economy overall.

First topic: the economic outlook. Despite the effects of the energy disruption, the euro area economy proved resilient, with solid real GDP growth in the second quarter of 2026. You have received a document with a series of charts, and then growth is discussed. Growth was broad-based across countries and sectors, and this is expected to continue in the second quarter of 2023. Manufacturing posted a solid result, backed by increased government spending on defense and infrastructure. Consumer confidence has recovered from its spring lows and has helped services to recover. And the increase in AI-related activity is visible in digital services, business investment, and exports. The labor market remains solid, with unemployment at 6.4% in July, although employment and labor force growth continue to slow, productivity has risen gradually. Over the medium term, consumption should be supported by a gradual decline in energy prices and by a solid labor market. Business investment should increasingly drive growth, and exports should benefit from stronger external demand. The baseline scenario of the ECB staff projections from September forecasts that the economy will grow by 0.9% in 2026, 1.4% in 2027, and 1.5% in 2028.

There are elements of what I am saying in the documents you have received. Headline inflation rose to 3.2% in August, from 2.9% in July. Energy inflation rose to 14.3%, from 10.3% in July. This increase reflects in particular the strong contribution of refining margins for liquid fuels especially, and a rise in energy commodity prices. At the same time, food price inflation fell to 1.1% in August, from 1.2% in July. Inflation excluding energy and food fell to 2.4%, due to a decline in services inflation, only partly offset by an increase in goods inflation. So far, wages have not shown a material response to the energy shock. Compensation per employee, which measures nominal wage growth, stood at 3.3% in the second quarter, compared with 3.6% in the first quarter. According to the baseline scenario of the September projections, headline inflation will average 3% in 2026, 2.5% in 2027, and 2.1% in 2028. For inflation excluding energy and food, the baseline projects 2.5% in 2026, 2.6% in 2027, and 2.3% in 2028. Inflation expectations at shorter horizons remain elevated, but most longer-term inflation expectation indicators are around 2%, which supports the stabilization of inflation around our medium-term target, and the outlook remains surrounded by great uncertainty with risks to the upside.

Monetary policy, the ECB's monetary policy orientation. In line with our commitment to ensure that inflation stabilizes at our 2% medium-term target, we decided to raise the ECB's three official interest rates by 25 basis points at the beginning of this month. I would like to explain the reasons for this decision. When we face energy shocks, the ECB has a very clear strategy. We do not react to energy prices; we only react if we see risks that the rise in energy prices is being embedded into inflation, and we assess these risks through a comprehensive evaluation based on three criteria. First, the inflation outlook. We examine our inflation forecasts and the risks surrounding them. Second, the dynamics of core inflation, which currently mainly refers to the extent to which the rise in energy prices is being passed through to other prices and wages. Third, the transmission of our monetary policy to borrowing costs and economic growth. Taking into account these three criteria I have just described, we see an increase in inflation ahead, but there is still no sign that it is really being embedded into the system. The inflation outlook will be higher in 2027 and 2028 than we expected a few months ago, mainly due to the rise in energy prices. However, at this point we still see no evidence that energy prices are contributing to wage increases. And although growth has been resilient since we last met, long-term interest rates have risen notably, which will slow growth and reduce transmission by more than foreseen in our September exercise. Taken all together, this means we remain on the middle path of monetary policy, that path I presented to you at the beginning of this year. That means that while it is true that the shock or disruption is too large to ride through, we consider that a measured response is the most appropriate way to keep inflation under control. Let us now turn to the question of artificial intelligence.

I continue in English, says Ms. Lagarde. Let me now turn to artificial intelligence, which is a transformative force. It can reshape production processes, business models, and the structures of the entire economy, but its macroeconomic effect is still uncertain globally. Its impact will be seen through several interconnected channels spanning both the demand and supply sides of the economy, and it will unfold over time. I will focus on two of the channels most relevant to the inflation outlook. First, productivity and investment, and then labor markets and income distribution. First, productivity and investment. AI has the potential to help companies produce more, and over time higher productivity could reduce costs. All else being equal, this should reduce inflationary pressures over the long term, but the magnitude and speed of that effect will depend on how it is all adopted in the economy. There are encouraging signs, but also work to be done. A recent ECB survey shows that by the end of 2020, 38% of euro area companies had already reported at least moderate AI use. However, only 7% reported significant use. We see that the data have evolved quite significantly over recent quarters. Unlocking AI's full potential will require significant investment. Europe needs to invest in innovation, but also in computing capacity, data centers, and energy. Companies must also adapt their processes and train their employees. AI-related investment is rising in Europe, and there is more still being prepared in companies, but we remain behind the United States. In parallel, there are European initiatives such as InvestEU and the European Growth Fund for companies that can help mobilize additional investment in AI infrastructure. EUR 30 billion, for example, for AI gigafactories, which is a good start. At the same time, AI investment activity will also depend on what happens in financial markets and on whether the risks associated with the AI financing boom materialize. Equity valuations worldwide are concentrated in a relatively small number of AI-related companies that are also rapidly increasing their funding through debt. A sharp reassessment of AI companies' prospects and the sustainability of their debt could lead to market corrections and spread to euro area investors and the economy overall. Second, AI will operate through labor markets and income distribution. AI will change the tasks people perform, the capabilities companies need, and how the benefits of AI are distributed will also matter. Currently, more than 50% of workers already use AI in their work, and so far, overall, companies continue to hire. In fact, survey evidence shows that companies that use AI more generally are more likely to hire new people than those companies that are in transition. But the important thing is, what do companies use AI for? They use it to support research, or rather, those that use AI to support research, innovation, and new products tend to hire more, while those that use it mainly to reduce labor costs hire less. Therefore, the central question is whether, over the long term and for the economy overall, AI will mainly complement workers or replace them, and this can affect incomes, demand, and ultimately inflation. But these effects remain difficult to predict. Historically, major technological advances have not reduced employment, but right now we still do not know whether the case of AI may be different. What should Europe then do to make the most of the AI growth opportunity while at the same time ensuring sovereignty? Well, first, Europe must invest in AI innovation. Innovative companies need broad, integrated markets, deep and accessible capital, significant computing capacity, and skilled workers, as well as abundant and affordable energy. I think you understand me. This means reducing the fragmentation of the single market, advancing the savings and investment union, and accelerating the energy transition. EU initiatives such as AI factories, testing facilities, and regulatory sandboxes can also help ideas move more quickly from research to commercialization, and there is still room to innovate in many areas, from greener chips to more efficient algorithms. Second, Europe must build greater independence. Europe must strengthen the critical, underdeveloped parts of the AI value chain. We must also ensure that development guarantees that we can rely on European technology for sensitive activities, in line with the objective of the technological sovereignty package. The goal is not self-sufficiency. We must ensure that essential functions can operate under any circumstances. At the same time, Europe must have a solid position in the global AI value chain for economic and geostrategic reasons. Third, Europe must deploy AI intensively. AI adoption must extend beyond a small group of technology leaders, but that alone is not enough. Companies must also use AI deeply and effectively. Among other things, this will require specific support for reskilling programs to address the critical shortage of AI-related skills and to ensure that the workforce can adapt to this transformative change. Implementing the AI strategy can support broader adoption, and we also look forward to the Commission's upcoming initiatives on industrial AI in priority sectors. In conclusion, for monetary policy, the task is clear. We must closely monitor and study how AI affects productivity, investment, labor markets, financial conditions, and inflation, so that we can continue to fulfill our mandate of maintaining price stability. For Europe, the direction is also clear. We must create conditions for AI to raise productivity and living standards, and we must ensure our sovereignty in this area. The precise effects of AI remain uncertain, but uncertainty is no reason for inaction. As Alan Turing, a pioneer in computing, already said or observed, we can only see a short distance ahead, but we can see many things that need to be done. The same applies today to artificial intelligence. Thank you.

I will address the second part of the question, if you don't mind. Later, if you can repeat the first part to me bilaterally, that would be very useful. Regarding the impact of AI, thank you very much for that mention. I will certainly pass it on. The point is that there is a lot of uncertainty, and there are those who predict that the impact will be minimal, 0.1% perhaps, and others who are much more optimistic about the productivity gains from AI. So, one, too much uncertainty. Then it will depend on the level of diffusion, involvement in sectors, etc. But our estimates of productivity gains, as I say, are estimates between 0.3 and 0.4 percentage points per year over a 10-year period. That is a significant productivity gain that we see thanks to the massive use of AI. We are certainly somewhat less optimistic than our US counterparts, for example, regarding what productivity gains we can expect. But well, if this becomes reality, if these are not negligible figures, they could change many things. And the first question, if you don't mind repeating it, well, I'll try in English. National measures to limit energy prices, as inflation targets are influenced by them. Well, at the ECB we focus on price stability, and we see these developments. And with the Commission, we have foreseen that temporary, targeted support is what should be pursued. And we see, at the European level, that for example it is at 0.1% of GDP for the euro area, a relative increase compared with what we have seen in recent months. We were at 0.03, now we are at 0.1. Ah, but from what the Commission sees, from what we ourselves see, it is not necessarily temporary nor necessarily targeted. Well, there are other measures that can be taken for the most vulnerable and most exposed part of the population.

Thank you very much. I want to clarify whether you want to talk about artificial intelligence or inequality. I had two questions. First.

I think I have a problem with these headphones, but it may be the shape of my ear. Yes, we take into account climate change and the consequences of the necessary transition in our projections and the work we do, and we also want to anticipate the impact of artificial intelligence with respect to all the aspects I have already mentioned: productivity, the volume of investment that will be needed, the impact on employment and inflation as a result of all this. So yes, we take it into account in our projections, and we try to anticipate it, and we do so based on all the studies that have been published, based also on what the interested sector reports, the deployment of artificial intelligence, but also surveys, consumption surveys, everything within our reach to see what percentage of investment will go to artificial intelligence, to workforce training, the expected impact of artificial intelligence. There is something that reassures me, because this is changing before our eyes, both in innovation, its use, the risks it carries, cyber risk is something that worries us a lot, but what seems encouraging to me is that diffusion is also advancing. I remember that people said, when there is a data flow you need artificial intelligence, but the business world, especially SMEs, now take it into account when they invest in artificial intelligence. And in the future we will have to take this into account in our projection analyses. And if productivity increases as it is supposed to, then I believe this will certainly have a considerable impact on inflation and production.

Inequality will be a very important aspect of the results that artificial intelligence will produce, and the distribution of the benefits of artificial intelligence is not something that concerns the European Central Bank, but rather governments and parliaments, to assess where these benefits will end up. This is important for governments and for tax authorities, because the tax base needed to generate revenue, if we see that there will be less employment but more corporate profits, then the tax criteria will have to be redefined, and this will have to be decided by the finance ministries.

We have to start from the view that our mandate establishes that the main priority, the main objective, is price stability, and logically this then affects price stability in investment, hiring, capital allocation. This is our main contribution as a central bank, but we take it into account in our projections because we carry out research, analytical work, I think we can contribute to identifying the risks and the measures that others must take. Thank you very much.

Thank you for your answer. Well yes, it was a real pleasure to welcome you to Frankfurt last week, all of you, the ECON committee. The savings and investment union is not a primary responsibility of the European Central Bank, but rather yours, the Commission's. I suppose there will be many meetings between the institutions, trilogues. You are concerned about how this project is progressing, securitization, which must be a first step towards this savings and investment union. How do we see it? Everything goes hand in hand. If we want a robust and sovereign European Union, a euro area, everyone has to play their part. We want to do our part from the European Central Bank. We want to ensure the independence of the European Central Bank. We are developing the digital euro in its wholesale form, retail, how to organize the clearing and settlement of these tokenized payments. We are also going to extend our repo lines beyond the euro area, and we want a robust euro. That said, having a broad savings and investment union that reduces the fragmentation we suffer from. For example, look at supervision: 52 individual supervisory authorities and so many interpretations of the directives, and this is a headache for those who want to invest in Europe, and they are concrete obstacles to these investments because this goes against the single supervision that was the purpose of the Single Supervisory Board, the Single Supervisory Mechanism. And I think this will widen the capital base for the European Union, because the alternative is for capital to be deducted, to leave here for other places, and for that we need, yes indeed, to calibrate the savings and investment union well so that capital reaches all parts.

I understand you too, but the sum will be greater, the result will be greater than the sum of the individual capital markets we have now. That is what investors all over the world confirm to us.

This takes me back to 2008, everyone looking for capital. Right now, the host or home country problem can be dealt with by other entities. We deal with price stability, and then we have the Single Supervisory Board, but we can combine several elements such as licenses, derogations, and guarantees. These exemptions already exist, but we also need ad hoc guarantees, and I think they could go hand in hand to address the concern of host country authorities who fear that capital reserves in subsidiaries will be captured by the head office in the home country. But bringing together the resolution authority, as well as sufficient liquidity in the resolution fund, and an authority that deals with this, I think addresses this problem and is necessary for the banking union to advance, but we also need to reduce fragmentation. Thank you very much.

Thank you very much for paying such close attention to our decisions and how we have tried to justify and explain them. I may be repetitive, but we have tried to capture, as far as possible, the indicators we use meeting by meeting, based on data, to justify our decisions. And I refer you back to something I already told you: we use three criteria, the inflation outlook, that is, our inflation projection, and we had a fairly substantial projection for the euro. Then, second element, risk, also the dynamism of core inflation, which is the energy crisis. And we see whether this translates into effects...

Derivatives, such as the evolution of wages, but also the cost of credit. And we see, based on the September criteria, that we have more inflation, higher inflation than we forecast in June, but we also don't see it already perpetuating—for example, the collective bargaining negotiations in some of the states, in Germany, in France, in Italy, or other countries.

Is that what could have an impact on a knock-on effect? And we don't think it could lead to inflation, an inflation impulse. In any case, it would be too late if that were the case, but we see no indication that it translates into a second round of inflation.

For 2027 and 2028, the inflation projection is higher. We have updated our June projections, but we also see no signs today that energy prices are generating an increase in wages.

We have updated our projections for 2026 and 2027 since the last meeting, and we have seen new developments, for example, in the long term.

That will have an impact on growth, which will be slower than we previously thought. Based on all of that, we make our decisions, and we also analyze what is happening in the market. And we have cross-checked this with the three projections, with the three scenarios, and the decision is justified by the most conservative projection and the other two, which are more severe.