The triple bubble (AI, economy and government deficit) in the USA

At the end of July, we pointed out that US equities were in some respects even more overvalued than during the telecoms and internet bubble in 2000 and the TINA (There Is No Alternative) bubble in 2021. The economic trends were positive in both of these years, while there are growing global recession risks in 2024. In addition, the geopolitical situation in the two earlier years was considerably more favorable than it is now. Unlike 2021, when interest rates in the USA were still at historic lows, there is now an alternative in the USA, namely US government bonds. In contrast to the year 2000, this time only the US equity market is heavily overpriced in a global comparison (chart 1). Was the price slide at the beginning of August a first warning sign?

Executive Summary:

The positive effects of artificial intelligence (AI) are overestimated both in terms of the large technology companies - most of which are based in the US - and their impact on long-term economic growth, while the opportunities for companies that can make good use of AI are underestimated. This means that the high valuation of US technology stocks compared to other stock markets can be seen as an overvaluation or even a bubble. This assessment is reinforced by the fact that valuation models with economic data such as US inflation and US and global economic growth, some of which can be calculated back over 100 years, also lead to negative earnings expectations for US equities up to 2034. The valuation comparison with US residential real estate and gold is also very unfavorable for US equities, as is our classic forecast model based on dividends, gross profits and book values, as well as the share price forecast based on the sentiment of US investors, who have never been as optimistic since 1947 as they are in the summer of 2024.

The AI fantasy of investors must not be impaired by bad news from this area and the US economy must not disappoint, otherwise there are considerable long-term price risks in the USA.

The valuation of the US equity market is strikingly high, both in terms of its own history and in comparison to the equity markets of the other major industrialized countries (chart 1) - US equities reach over 90% of the valuation level at the peak of the bubble in March 2000, while the other equity markets only reach between 40% and 51%. At 7.2% p.a., the annual growth rates of dividends, gross profits and book values of US companies since 1974 have been just as high as those of Japanese companies and barely higher than those of German companies (6.8% p.a., chart 2).

The rise in the valuation of US equities from 2022 onwards is linked to fantasies about the commercial opportunities of artificial intelligence (AI), which is being developed by major US technology companies in particular. Investors are likely to have pushed the share prices of these companies too high. The most important argument for this assertion lies in the history of the introduction of new high technologies (chart 3), which prompted the Nobel laureate in economics Robert Solow to make the famous remark back in 1987: "Computers are everywhere - except in productivity statistics." (Source: Wikipedia 2024, keyword productivity paradox).

Obviously, the productivity gains from the use of high technology cannot compensate for the decline in global economic growth that has been observed for decades (chart 3), especially not in the industrialized countries. It is also possible that the positive and negative effects of modern technologies on economic growth cancel each other out (see also Wikipedia 2024, keyword productivity paradox). In any case, history provides no justification whatsoever for assuming a sustained and immediate economic boom following the introduction of a new computer-based technology, meaning that the probability of such a boom is closer to 0% than 100%.

The next problem lies in the fact that large investments, such as those now being made by high-tech companies, have often produced weak returns or even high losses. An excellent current example is China, where the huge investments in residential real estate have led to enormous vacancy rates and therefore housing construction is now collapsing (see chart 4 as well as the capital market outlooks of September 2022(here) and November 2021(here), where we described China's coming problems early on). What is impressive is the fact that China is now making the same mistake for the second time, namely in the massive promotion of factories to build electric cars, solar panels, car batteries, ... . For someone who reveres Mao as the great wise leader of China, it may come as a surprise, but it is a simple realization of business economists that too much investment leads to overcapacity and this leads to losses. This is exactly what we are currently seeing in China's industry (Figure 5).

Examples of other investment bubbles from the recent past include the double bubble in the Japanese stock and real estate market until 1990, the bubble in real estate investments in East Germany until 1994, the bubble in telecom and internet stocks until 2000 and the subprime crisis, later called the financial crisis, until 2009, starting in the USA (for detailed information on these investment bubbles, see the Capital Market Outlook from October 2020, which you can find here ). The current investment bubble (in addition to factory construction in China) is currently being inflated in the AI sector, on three levels. It is generally known that many investors are concentrating on the shares of the large technology companies in the USA that are (also) active in the AI sector. Some investors are less familiar with the enormous sums that these companies have already invested or plan to invest in the AI sector. Examples include Google (planned investments in AI of over USD 100 billion, source: Demis Hassabis, CEO of Deepmind, Google's AI subsidiary, in April 2024) and Microsoft, which is also planning investments of over USD 100 billion (source: golem.de in March 2024). The third level is the enormous power consumption of AI. The respected Neue Zürcher Zeitung reported on February 16, 2024 that asking chat GPT could consume thirty times as much electricity as conventional "googling". It is therefore understandable that Google omits electricity consumption from its latest climate protection report, as reported by Swiss IT Magazine on July 3, 2024. The Macro Strategy Partners reported on 13.8.2024 that the price for the construction of power plant capacity (1 megawatt per day) has increased ninefold compared to the previous year.

After all, AI has a significant disadvantage compared to today's successful internet business models, which were still in deficit 25 years ago, namely the presumed lack of a network effect (source: BCA, July 2024). Facebook, Amazon, Microsoft and Google benefit from the fact that people use their services because others also use them, which makes them interesting. This is unlikely to work with AI, as all AI models are ultimately trained on similar or the same data and may therefore hardly differ in their performance in the long term. You can therefore use one provider's model or that of a competitor if it is cheaper. Incidentally, the issue of paying for the data that AI companies need to train their models will also become more important, as the data used for this purpose has often been stolen up to now. The Canadian research service BCA, which has been in business since 1949, draws an interesting parallel to the AI issue, namely the airline business. Airlines are very important and useful for customers, but they are not a particularly profitable business model. The use of AI will enable many companies and industries, such as the healthcare sector, to make significant advances in productivity. The share prices of such companies are not currently inflated, so this idea does not yet seem to be a consensus.

Only if the optimistic expectations of holders of AI shares are fulfilled in the long term and if the economy continues to grow at least moderately in future (for the generally underestimated risk of recession, see the capital market outlook from July 2024(here)) could the valuation of US equities remain at the current level. However, if things do not turn out quite so well, the risks are considerable, even if the US does have some structural advantages over Europe. One example is the greater economic policy uncertainty in Europe since the start of the war in Ukraine, which is currently much lower in the US and Japan, which are far removed from the war zone (chart 6). Incidentally, this index has not existed for China since November 2023; the last value was a very high 743. This sharp rise in uncertainty in Europe since the 2008 financial crisis has shown a high correlation of +0.8 to the relative performance of US equities compared to European equities since 1987 (chart 7).

The correlation can fluctuate between -1 (both lines always run in opposite directions) and +1 (both lines always move in the same direction) - +0.8 is quite a high value. If the war ends or at least the Putin admirer Trump does not become US president, the blue and red lines in chart 7 are likely to move downwards.

Another structural advantage of the US to date has been its comparatively low government spending ratio (chart 8). The US government claims a significantly smaller share of what its citizens earn than the EU, for example, namely only 34% instead of 49%. The other side of the coin is the exploding national debt, which has risen from 56% of national income to 122% since 2000 (chart 9, blue line). By contrast, public debt in the eurozone has only grown moderately over the same period. It amounted to 69% of national income in 2000 and is currently at a level of 89% (source: Eurostat (Trading Economics), August 2024).

Chart 9 shows that the US national debt after the Second World War was just as high as it is today in relation to national income, but then fell very quickly. We should not get our hopes up that it will be the same this time. The vast majority of the debt at that time was incurred as a result of government spending on the war against Germany and Japan. After the end of the war, the high government deficits also disappeared immediately. Today's debts are the result of bloated social spending and constant economic stimulus. If you end this abruptly, there are immediately enormous political and economic problems. Another difference to today's situation was that in 1947, the post-war year, the private sector was only indebted to the tune of 47% of national income (private households: 18%, companies: 29%, chart 9). In 2024, this figure has more than tripled to 150%. Accordingly, total debt in the US currently stands at 272%; after the Second World War, it was only 153% (chart 10). Another negative difference to the post-war period is that the US Federal Reserve had pushed interest rates on 10-year government bonds to below 2% (green dots in chart 11) - this was probably remembered in the two coronavirus years (blue dots on the far right), but then forgotten again (red dots). Interest rates comparable to those in 2022 and 2023 were previously only seen in the US with government debt below 70%. Extremely high government debt at current interest rates (red dots) will lead to an enormous burden on the government budget in the medium term, which will weigh on growth through spending cuts.

Finally, the USA has another important advantage, namely a large capital market that is also productive for venture capital, which is an important reason why the USA produces so many innovative companies. In addition to the constant production of new regulations, the EU should perhaps also take a brief look at creating a uniform capital market. That is one of its natural tasks.

After taking a look into the distant past with regard to debt, we will now do the same with regard to the valuation of the stock market.

Until the First World War (1914 to 1918), there was hardly any inflation in the USA, as the US dollar, unlike European currencies, was still a gold currency even after the First World War (chart 12). In the 100 years since 1820, the average inflation rate reached only 0.7% p.a., while share prices rose by 1.3% annually. Thereafter, the inflation rate averaged 2.8% p.a., while share prices have risen by 6.4% annually since 1920 (chart 12). From the 1950s onwards, this was helped by the fact that companies no longer paid out the majority of their profits as dividends, as they had done until 1950 - hence the very high dividend yields during this period (chart 13) - but instead retained a growing proportion of their profits, allowing their capital to grow faster.

If you look at the inflation-adjusted, i.e. real, share prices, you can see that they have followed a trend since 1920, which they exceeded significantly at times, e.g. in 2000, 2021 and 2024 (chart 14). However, there were also times when the price performance was significantly weaker than the trend. The future 10-year price gains in each of the last 104 years were -4% p.a. at a valuation level comparable to 2024 (76% above the trend) (chart 15).

An approach that compares the sharp rise in real share prices since 1950 with the development of national income arrives at similar results. In particular, the price trend roughly corresponds to economic growth if one takes into account the fact that US companies have increasingly turned to world markets after the Second World War and therefore not only the US economy but also half of the development of global economic output is taken into account (chart 16). Here too, future share price gains show a fairly close correlation with the deviation of share prices from economic growth; where share prices have risen by more than 30% - currently 65% - there have always been share price losses in the following decade (chart 17).

Unfortunately, the US equity market is not only overvalued compared to other equity markets (chart 1) or to the economic data of inflation (chart 14) and economic growth (chart 16), but also to the price of gold and US residential real estate. Since 1900, the price development of the three tangible assets equities, real estate and gold in the USA has been practically identical for 82 years, with considerable fluctuations (excluding dividends and rental income, chart 18). From 1982, however, share prices rose to seven times the value of the other two tangible assets by 2024 (chart 19).

This represents a massive overvaluation, because there are reasons for this strikingly good performance of US equities since 1982 that are not sustainable.

Chart 22 shows that US companies were able to massively increase the share of their after-tax profits in US national income from 5% to 11% during this period. This was not only the result of successful entrepreneurial activity, but also a consequence of globalization, which led to lower demands for wage increases due to the fear of companies relocating to cheaper foreign countries. Accordingly, the share of employee income in national income fell from 57% in 1982 to a low since 1947 of less than 52% (chart 21) - one of the causes of populism in the US. Another reason for the increase in profits was the fact that companies currently only have to pay 1% of national income in taxes, whereas the comparable figure from 1945, when the national debt was roughly the same as today (chart 9, blue line), was seven times as high (chart 22). However, the US government will not be able to permanently do without higher taxes on corporate profits - for financial as well as political reasons. In the meantime, presidential candidate Kamala Harris has called for an increase in corporate taxes from 21% to 28%.

The high valuation of US equities (chart 23) is likely to lead to annual price losses of 3% p.a. over the next 10 years, as was the case with similarly high valuations in 1999 and 2000 (chart 24). This is also supported by the unsustainable reasons for the high profit increases of the last 40 years (charts 20 to 22), the exaggerated euphoria surrounding the topic of artificial intelligence and the overvaluation compared to the normal course of real share prices (chart 14) and real economic output (chart 16). The massive outperformance compared to the other real assets, residential real estate and gold, is also very striking and probably not sustainable.

Investors' euphoria, as measured by the equity allocation in their liquid wealth (chart 25), has been a reliable indicator of an unfavorable future price trend since 1947 (chart 26).

The last anomaly we would like to highlight is the unprecedented sharp decline in the highly reliable Leading Economic Indicators of the US research firm The Conference Board since the turn of 2021/2022 in the three largest economic areas in the world, namely the US, China and the eurozone - including in July 2024 - while share prices have risen at the same time (charts 27 to 29, red oval). Economically, this opposing trend makes no sense, even if interest rates fall soon.

Conclusion:

Overall, the AI euphoria has led to an overvaluation of US equities compared to the history of the US equity market, other equity markets, other real assets and economic data such as consumer price trends and national income growth. The particularly favorable conditions for US corporate earnings growth since 1982 have also contributed to this overvaluation and are not sustainable. The same applies to the massive economic support provided by the US government's record-high new debt in peacetime, which will again exceed 7% of national income this year. Ambrose Evans-Pritchard, the international economics editor of the Daily Telegraph, a British newspaper founded in 1855, wrote in August 2024 that, based on the current record-high US government deficits for a normal economic phase, deficits in the order of 15% of national income (Maastricht criterion: 3%) are likely to be reached quickly in the event of an economic downturn.

By then at the latest, the hot air will escape from the triple bubble.

‍

As usual, we will repeat the key messages of our capital market outlook from three years ago so that you can get a feel for our long-term forecasts.

You can find the Capital Market Outlook from August 2021 here. In it, the energy transition was our topic. We came to the conclusion that limiting climate policy to the reduction of CO2 emissions is not the right way forward, as this would entail enormous costs that voters may not support in the long term. A mixture of reducing CO2 emissions, investing in research and development on green energy and mitigating the negative consequences of global warming and - as the most efficient measure - immediately ending subsidies for fossil fuels is much more efficient.

However, politicians will not want to take the most efficient route, namely a uniform CO2 tax with compensation for the resulting social hardship, because voters only understand and support the simple route (bans, subsidies for e-cars, etc.). This is comparable to housing policy, where, with the broad support of voters, it is preferable to maintain rent controls, threaten to expropriate apartment building owners and extend tenant protection further and further, making the construction of apartments increasingly unattractive.

‍

You can also download the capital market outlook here.