The Stronger the AI, the Lower the Wages: Your Education is Becoming the Most Expensive Devalued Asset

{#docs-internal-guid-685615de-7fff-77d7-3c8e-e5634f2f1982}

Author: Block Analytics Ltd X Merkle 3s Capital

A model projection shows that by 2030, the economy could produce one-third more output than a path without AI, while the total income for all workers would only increase by half a percentage point. The same projection also presents another outcome: AI changes almost nothing, and the unemployment rate remains unchanged from today. Both outcomes share the same set of already occurred data, with divergences only becoming apparent after 2027, and you must decide what to do before then. This article does not discuss its equations, but rather the results it calculates—four impacts on employment, education, investment, and society.

A Report That Does Not Intend to Comfort Anyone

The content of this article references The Anthropic Institute's "Economic Scenarios for Transformative AI" (September 2026). This is a working paper numbered 2026-02, with five authors including Charles I. Jones—a representative figure of semi-endogenous growth theory. The paper itself states that its views do not represent Anthropic's position and notes the use of Claude as a research and writing assistant.

What it does can be summarized in one sentence: it breaks down the impact of AI on cognitive work to the granularity of "tasks," requiring those replaced to spend time finding new positions, and then projects this to early 2030. The only mechanisms to remember are two: AI will not completely replace a profession; it takes away tasks piece by piece; those who lose tasks will not disappear into thin air; they will compete for others' positions. These two points determine the shape of all subsequent numbers.

The differences among three scenarios are simply how large a portion AI ultimately consumes.

Mild scenario: 4% of tasks in the entire economy are completed by AI, with GDP in 2030 being 1.6% higher than the path without AI, where AI is just an ordinary technology.

Significant scenario: 12% of tasks are taken over, roughly equivalent to one-fifth of cognitive tasks, with GDP up 8.3%, described in the report as "a pivotal technology more important than the internet."

Extreme scenario: 30% of tasks, about half of cognitive tasks, with GDP up 32.4%.

It must be clearly stated that these three scenarios are illustrative and not probabilistic; the report repeatedly declares in the abstract, introduction, and conclusion that they are not predictions. The only conclusion it offers is that today’s data cannot rule out any of them. Treating the extreme scenario as "Anthropic's prediction" is a complete misreading.

Its value lies not in predicting 2030, but in turning a vague debate into an arithmetic problem: If you believe X, then according to standard economics, you must accept Y!

Another easily overlooked detail is that what determines the outcome is not what the model can do, but to what extent it is actually used. The capability gap of AI in the three scenarios is less than three times, but when combined with "how many people are really using it and how deeply they are using it," the actual share of tasks executed by AI expands from 4% to 30%, becoming 7.5 times. Beyond the technology curve, there is a diffusion curve, which is the real amplifier.

The report includes a survey for comparison: 10,980 American adults, with the median implied scenario falling near the significant scenario—GDP +8.6%, cognitive employment -4.2%, unemployment rate 4.6%. However, reading this as "society has reached a consensus" is a statistical misreading. Regarding productivity gains, about 30% of respondents believe AI saves no time at all, while 49% believe it can save at least half; regarding Nobel-level discoveries, 40% believe they can never be achieved, while 39% believe they can be achieved before 2030. The median falls in the significant scenario not because most people believe in it, but because the number of believers in the mild and extreme scenarios is almost equal.

It should also be clarified what it does not consider. Robotics and physical automation are completely excluded, as are financial market turbulence, business cycles, and political economic factors. The report itself admits: "To a large extent, this is why we do not extend the analysis beyond 2030." This self-disclosure appears again later and is the most honest and unsettling statement in this report.

A report that provides no probabilities, no recommendations, and does not intend to comfort anyone is, in fact, harder to refute than any prediction!

Employment Impact: Who Will Lose Their Jobs and in What Form

First, let’s clarify who this impact hits. The report divides U.S. employment into two halves: the cognitive group includes 12 major categories—management, business finance, computer mathematics, construction engineering, science, community social services, law, education, arts and media, medical practice, sales, and office administration—accounting for 62.4% of employment in 2025; the remaining 37.6% includes medical assistance, security, catering, cleaning, personal care, agriculture, forestry, fishing, construction, installation and maintenance, production, and transportation.

The model only allows AI to impact the cognitive group; the efficiency of the other 37.6% of workers is completely unaffected by AI, and their wage changes come entirely from an indirect effect: the economy has become richer, while their type of labor has become relatively scarce. Therefore, everything seen later is a direct consequence of this setting, not an empirical finding. Rejecting the premise that "AI only impacts cognitive work" invalidates the entire projection—this point must be kept in mind before reading any numbers.

This dichotomy has three blind spots, which the report itself admits: "Any dichotomy is a simplification." The two major categories of medical practice and education contain a large amount of physical and interpersonal labor that cannot be performed remotely, leading to an overestimation of exposure; the salary and educational thresholds for sales and office administration are much lower than for management and professional positions, yet they are treated as the same type of person in the model, completely obscuring intra-group differentiation; there are also no distinctions of seniority, skills, or regions within the same group. These numbers describe two statistical aggregates, not two specific types of people.

{#docs-internal-guid-0639113a-7fff-1435-202c-3d38559904b0}

Next are the results. In the extreme scenario, the unemployment rate for cognitive positions is 17.9%, while the overall economic rate is 11.9%—these two numbers must be distinguished, as mixing them will overestimate the unemployment rate by about 6 percentage points. The report describes the cognitive position number as "stunning." For reference, the annual peak in the U.S. after the financial crisis was slightly below 10%, and around 8% during the COVID period; in the extreme scenario, even the overall unemployment rate exceeds any post-war recession.

The difference between mild and extreme is not a matter of degree, but of magnitude. The cognitive unemployment rate rises from 2.9% to 17.9%, a sixfold increase, while cognitive wages shift from +0.4% to -11.5%, a sign reversal, with no indicators in between that can partially offset the differences at both ends. Considering that in a path without AI, wages should rise by about 2% annually, in the extreme scenario, the absolute level of cognitive wages will be lower than mid-2026—after four years, it is not that wages are rising slowly, but that they are regressing.

Job losses are more concentrated than wage changes. In the extreme scenario, the number of cognitive positions is 21.5% less than mid-2026, with the total wages for this group down 31.0%. Meanwhile, wages for other groups rise by 33.6%, the largest single increase in the entire report. The report states that the total wage increase for other groups is "almost the same" as the decline for the cognitive group—total quantity conservation, just a change of personnel.

The form and intuition of the impact are also different. AI progresses task by task, so professions will not disappear entirely one day; they will first be hollowed out partially. The tasks that trigger replacement are neither the hardest nor the simplest, but those where the cost of AI execution drops the most while also being the most significant in terms of payroll. The remaining parts are left for humans; the signboard still hangs, the positions still exist, but fewer people are being hired.

There is another colder detail: whether AI replaces you or helps you is determined by the technology itself in this model, and companies can only passively accept it. In the three scenarios, the proportion of usage that "looks like replacement" is 50%, 75%, and 90%, respectively—corresponding to the "help you" space, which compresses from 50% down to 10%. The augmentation is a space that will be continuously squeezed, not a stable safe haven.

Your profession will not suddenly disappear one day; it will first be hollowed out by half, and then you will find yourself competing for a position with three others who have also been hollowed out!

The arrival of pain also has a sequence. When companies encounter the situation where "humans are more expensive than AI," they will not first lower salaries but will directly lay off employees—salary reductions happen very slowly, only covering half of the gap in a year, so layoffs precede salary cuts. The first group to suffer is those who are laid off; the second group is those who remain in their positions, with salaries being compressed to -11.5% annually, and since wages should have risen by 2% each year, the feeling is worse than the numbers suggest; the third group is those attempting to switch fields, with laid-off individuals flooding into other industries while those industries can absorb at most half of the gap each month, creating a mismatch that is itself a source of unemployment. Notably, by the simulation endpoint in 2030, the cognitive unemployment rate remains at 17.9%—adjustments are far from complete.

Finally, it must be said: these numbers should be seen as a lower bound, not an upper bound. The model assumes that those who switch industries immediately receive the full wages of the new industry, with no depreciation for seniority, no skill devaluation, no income interruption during training periods, no relocation costs, and no licensing or apprenticeship periods. The evidence cited in the report's calibration section indicates that real-world wage losses during reemployment typically range from 10% to 20%, concentrated among those switching careers. It mentions this literature by name and then states it was not included. Even if one is optimistic enough to believe that "becoming an electrician from a software engineer only requires time," the extreme scenario still yields a cognitive unemployment rate of 17.9%.

Educational Impact: Why Degrees Are Becoming Devalued Assets

The report does not mention education at all. However, the extreme scenario shows cognitive position wages down by -11.5% and other positions up by +33.6%. These two paths together are sufficient to structurally question the chain of "studying hard, obtaining degrees, and entering white-collar positions"—the ones harmed are precisely those with the longest education years and the highest wages, accounting for 62.4%. Over the past two centuries, the victims of technological progress have typically been low-skilled manual laborers, with education always regarded as a safe asset. This report depicts an impact that is completely opposite in direction.

The first layer is the evaporation of human capital rents. The reason why educational qualifications can command a premium is that they correspond to tasks that are scarce, valuable, and cannot be performed by others. When the unit cost of these tasks drops to less than half of what it was over four years, the years spent mastering them will correspondingly evaporate that rent. Moreover, the report indicates that this cost decline will continue deepening until 2030, with no signs of leveling off.

The second layer is the signaling value. Educational qualifications are valuable partly because they are expensive—those with strong abilities face lower acquisition costs, making them a useful filter. As AI levels the playing field for writing, analysis, and coding outputs, the distinguishing power of educational qualifications will inevitably decline. Employers will turn to harder-to-replicate evidence: work samples, verifiable delivery records, and face-to-face judgment. This layer is our inference; the report does not discuss the signaling mechanism.

The third layer is the most glaring. Historical experience suggests that for every two tasks automated, technology creates one new task, which is the source of the comforting phrase "technology always creates new jobs" that has been quoted for two hundred years. In three scenarios, this ratio is one new task for every two automated tasks, one for every four, and in the extreme scenario, it is zero—AI creates no new tasks for humans. That comforting phrase is explicitly shut down in this scenario.

It’s not that education is useless; it’s that the pricing of education is becoming cheaper!

Thus, the retraining window becomes a very unfriendly arithmetic problem. The report states that differentiation occurs after 2027, and "data from the next few years will tell us which scenario we are in." In reality, most licensed professions—nursing, electricians, plumbing, imaging technicians—require 2 to 5 years, while higher-level licenses like registered nurses and professional engineers require 5 to 10 years. If one starts retraining by the end of 2026, the earliest one could be licensed is between 2029 and 2031, which happens to fall in the period where differentiation has already solidified but transition has not yet been completed. If one waits for the 2028 data to "confirm" before acting, licensing will be delayed to 2031 to 2038, by which time the premiums of other groups will have already been diluted by early movers.

This creates an asymmetric gamble: if the final scenario is moderate, the loss of early transition is merely learning an additional skill; if the final scenario is extreme, the cost of not transitioning is finding a job in a context of a 17.9% cognitive unemployment rate, where the efficiency of cross-industry job searching is only about a quarter of what it is in normal times. It should also be noted that the report does not mention: the model assumes that transitioning does not require any retraining, so the income interruption during the training period is not reflected in those numbers.

What’s more troublesome is that the more "successful" the cognitive profession, the harder it is to transition out. The original data in the report shows that the probability of unemployed sales and office workers finding jobs in other sectors is 28%, while for management and professional personnel, it is only 10%. The deeper the accumulation of occupation-specific human capital, the higher the barrier to cross-sector transitions—this means that in this shock, the group most severely affected is precisely the one that is the hardest to rescue.

Whether to continue investing in English and communication skills should be viewed separately. The cost-effectiveness of pure language conversion skills is significantly declining—translation, editing, and cross-language writing are precisely the types of tasks where AI costs are dropping the fastest, placing them at the core of the impact. Investing in English as a "tool for information acquisition and text production" is investing in a skill that is rapidly depreciating.

However, the communication skills that serve as a vehicle for trust and negotiation are becoming more cost-effective. The reason is straightforward in the model: as the automated segments become cheaper, the non-automated segments become bottlenecks, and more money will flow into those areas of the entire process. In extreme scenarios, those professions that increase by 33.6% derive their value precisely from the necessity of human presence—negotiation, relationships, and team leadership value lies not in generating text but in having someone sign off on the results.

Therefore, our judgment is: if the goal is merely to read English materials and write English emails, it is not worth making a large time investment; if the goal is to negotiate in real-time in multinational scenarios, host meetings, and build trust, it is worth investing, but what is truly scarce is the communication skill itself; English is merely the medium. The most misleading assumption is treating "fluent speaking" as the goal itself.

Educational qualifications will not become obsolete; they are simply reverting from a ticket to enter the arena back to a decorative line on a resume!

Investment Impact: The Feast of Capital and Its Bill

This is the strongest signal in the entire report. In three scenarios, the total income of capital relative to the path without AI is +3.1%, +18.9%, and +81.4%; during the same period, the total income of labor is +0.6%, +1.4%, and +0.5%. This is not a share battle; it is two completely different curves.

How is the money flowing? The mechanism is simple to the point of being cruel: every task taken over by AI directly transfers all its corresponding wage expenditure to capital, regardless of how much cost is saved. The labor share drops from 60.0% to 59.4%, 56.1%, and 45.2%, while the capital share rises from 40.0% to 40.6%, 43.9%, and 54.8%. In the extreme scenario, the report states: "A full 15% of GDP has shifted from payments to labor to being captured as investment returns by capital." But what is more noteworthy is the significant scenario—labor share declines by 4 percentage points over four years, slightly less than the total decline in the U.S. over the entire forty years since 1980.

The return rate is also rising. The net capital return rate rises from a baseline of 6.5% to 6.6%, 7.0%, and 8.3%, with an increase of about 28% in the extreme scenario. This number is extremely sensitive to one assumption: how quickly capital can be accumulated. The report assumes that computing power can be financed globally and built within one to two years, thus giving a high elasticity of supply; if actual expansion is much slower, in the extreme scenario, net returns could reach 10.3%, while average wages could drop to −9.2%; if computing power is abundant, net returns would remain at 6.5%, while average wages would rise by 30.1%. The level of capital returns directly depends on how hard the physical bottleneck of computing power expansion is.

Productivity has indeed increased significantly, but the increase is in unexpected areas. Total factor productivity relative to the path without AI is higher by 0.7%, 3.1%, and 13.4%, with annual growth rates rising from 1.0% to 1.2%, 2.3%, and 7.3%; GDP annual growth rates rise from 2.0% to 2.4%, 5.4%, and 15.4%—the fastest year during the internet boom of the 1990s was 1999 at 4.7%, and the growth rate in the significant scenario has already surpassed that. Capital stock is higher by 2.3%, 13.8%, and 56.3%. However, the line "AI accelerates scientific research" contributes almost nothing: the stock of ideas relative to the path without AI is only higher by 0.07%, 0.20%, and 0.61%, with annual growth rates moving from 1.67% to only 1.69%, 1.76%, and 2.02%.

All the growth in this report comes almost entirely from making existing tasks cheaper, rather than discovering new things—"AI accelerates scientific research" as an investment argument before 2030 is not supported in this model!

If we were to rank the benefits in order, the basis for sorting should be physical bottlenecks rather than narrative heat: at the forefront are computing chips and accelerators, followed by electricity, transmission and distribution, advanced packaging and optical interconnects, then data centers and supporting construction, and finally cloud and inference service layers, with downstream applications coming last. The realization of downstream value will always lag behind upstream because it relies on the slow variable of "how many people are actually using it," and that curve only just took off in mid-2026.

The report also proactively acknowledges areas where it may have underestimated. It admits that the model lacks a key mechanism linking capital returns to growth rates, and that filling this gap "could significantly increase capital returns"—8.3% is likely still an underestimate. Other lower bounds include: innovation effects are underestimated because the model cannot generate self-accelerating feedback, and unemployment and income losses are underestimated because transitioning does not incur any wage discounts. A report that admits to being conservative in three directions is itself a piece of information.

Then comes the bill. The report has calculated: in the extreme scenario, to maintain cognitive workers' income at levels without AI, about 9% of GDP in transfer payments would be needed, "roughly equivalent to the sum of social security and federal healthcare"; after paying this amount, the rest of the economy would still be over 20% ahead of the path without AI. The total gain is about three times the loss of cognitive workers, and the report states, "in the textbook sense, the gains are sufficient to compensate for the losses."

The problem is that if this 9% is indeed raised through capital gains taxes, corporate taxes, or some form of computing power taxation, it directly erodes the benefit chain above. The report then states that such a scale of technical transfer is unprecedented, and based on past experiences of trade shocks on U.S. regions, it will not spontaneously occur. Worse still is the worst-case parameter combination in the report's footnotes: abilities and diffusion only reach the level of significant scenarios, but the replacement ratio is extremely high, cross-industry searches are very difficult, AI does not create new tasks, and capital expansion is limited; at this point, GDP is only 7.2% higher, while total labor income is actually 4.3% lower than the path without AI, requiring 84% of GDP gains for compensation. The logic of "first enlarge the cake, then divide it" basically fails in this corner.

The moderate scenario is also worth mentioning separately: there, net returns only rise from 6.5% to 6.6%, and excess returns from the AI theme are virtually nonexistent. The report clearly states that this scenario cannot be ruled out today. A position that only holds in extreme scenarios is effectively placing a heavy bet under an example that doesn’t even warrant a probability.

The biggest risk on the investment side has never been whether AI will succeed, but rather who will benefit from the returns after success!

Social Impact: The Cake Has Grown, but Workers Have Not Taken More

The most counterintuitive statement in the entire report is: in the extreme scenario, the total income received by workers is even less than in the significant scenario. +0.5% compared to +1.4%, while the GDP growth jumped from 8.3% to 32.4%.

The report provides only one arithmetic explanation, which is also the only one retained in this article: the labor share drops from 60.0% to 45.2%, a decrease of about a quarter; GDP rises by a third. 0.45 multiplied by 1.32 is approximately equal to 0.60, with the two forces precisely offsetting each other. Workers receive almost the same thin slice of a much larger cake.

There is no monotonic relationship between GDP and labor income—when AI upgrades from "significant" to "extreme," the cake grows by a third, but workers take home nothing extra!

Average wage is the most dangerous indicator here. It shows +0.7%, +2.1%, and +9.7% across the three scenarios, seemingly improving all the way. However, within that +9.7% in the extreme scenario are the −11.5% from cognitive positions and +33.6% from other positions, and it does not even include the 17.9% of cognitive workers who are already unemployed—those are not in the denominator, so the average looks good.

There is another more insidious trap: wage stability does not equate to human security. The report conducted a series of tests, adjusting the switch for "can wages be cut" from the most flexible to the most rigid. At the most flexible end, cognitive wages plummeted by 42.2%, but the cognitive unemployment rate was only 2.6%, and the overall rate was only 3.1%, almost the same as in normal times, with GDP actually the highest at +36.6%. At the most rigid end, cognitive wages were actually +2.8%, higher than the path without AI, at the cost of cognitive unemployment soaring to 24.0% and overall to 15.2%. The same shock can manifest as "everyone's salary drops by 40%" or as "a quarter of people are unemployed, while the rest see salary increases."

Protecting wages does not equate to protecting people; preserved wages will only transfer the shock into unemployment, affecting those who are already off the list!

This also explains why "current macro data is still quite mild" cannot be used as evidence. The report acknowledges that "the labor market effects of AI in the macro data so far are relatively mild," but immediately points out that this statement can be read as "we are in a mild scenario" or "we are at the beginning of an extreme scenario." The three scenarios share the same set of readings in mid-2026, with previous GDP differences within a few tenths of a percentage point. The data from 2026 cannot distinguish between the three scenarios, but the data from 2028 can. In a reality where wages are slow to adjust downward, early signals will first appear in hiring volumes rather than wages—by the time the payroll looks bad, the adjustments have already been underway for a long time.

The form of wealth disparity will also change. It is no longer primarily "high salaries versus low salaries," but rather "those with assets versus those with only wages." Cognitive workers are precisely the typical group that is "highly paid but with low asset ratios"—their income is exposed to the damaged side, while their assets are not on the benefiting side; the risk is not dispersed but rather compounded. The internal labor dynamics are equally intense: the total wages for cognitive positions have decreased by 31.0%, while other positions have nearly recovered the same amount; the winners and losers are not the same group.

A common rebuttal is, "If everyone is out of money, who will buy things?" In this model, this question does not hold: the total output translates into someone's income, with only 15% of GDP shifting from payroll to capital gains, and the consumption and investment of capital owners fill the gap. However, the report admits in its notes that the model does not consider price rigidity and related demand effects, thus "cannot produce the kind of shock that depresses demand and amplifies the shock through negative feedback." The risk of consumption breakdown is not disproven but rather assumed away—neither can it be used to negate the report's conclusions, nor can it be assumed that the report proves it does not exist.

The report explicitly excludes political economy but indirectly leaves three fate-determining variables: whether a country can build sufficient computing power and electricity, whether wage shocks manifest as salary cuts or unemployment, and whether there is the capacity for taxation and redistribution. From this, one can deduce the differentiation of outcomes—places with votes, fiscal space, and local computing and energy industries are most likely to achieve high growth with high transfers; places with votes but no computing industries will suffer capital outflows while domestic labor is harmed; regions with neither will simultaneously bear a decline in labor share and capital gains not materializing. This section is our deduction; the report itself only states, "What systems can bring about shared prosperity is beyond this framework."

Betting Before the Answer is Revealed

The logical starting point of the whole matter is just one sentence: the divergence will only show a winner in the data from 2027 to 2028, while most people need 2 to 5 years to complete a career transition—you must make decisions before knowing the answer. The following are our deductions based on the report's results; the report itself does not provide any personal advice.

For White-Collar Workers and Professionals, the sharpest points are these four.

First, break down your job into a task list and label what AI can currently do at each level—within this model, the average exposure level for cognitive groups is more than twenty times that of other groups; what you need to know is not "Is this profession at risk?" but rather "How much of my job falls on the side that is being taken away?"

Second, actively shift towards areas where "AI seems to help you rather than replace you": taking responsibility externally, coordinating across departments, making on-site judgments, and reducing purely delivery-based text and analytical outputs.

Third, establish verifiable delivery records—published projects, verifiable performance, externally verifiable results, as the distinction between degrees and titles is decreasing.

Fourth, conduct a stress test: assume the position disappears within 18 months; how many months can cash reserves cover? Even in significant scenarios, the cognitive unemployment rate rises from 2.9% to 4.5%, while in extreme scenarios, it faces 17.9%, with only a quarter of cross-industry search efficiency remaining.

For Parents, the same four points apply.

First, do not hastily change educational paths to "avoid AI" in the next year or two—mild scenarios cannot be ruled out today, and in that scenario, the cognitive unemployment rate is 2.9%, unchanged, with the expectation of overreaction leading to real losses.

Second, shift investment focus from knowledge stock to capabilities that require presence, responsibility, and trust-building; in extreme scenarios, the other groups' +33.6% increase is the largest single increase in the entire report.

Third, prioritize directions with licenses, apprenticeships, and on-site delivery attributes, but accept one limitation—the report excludes robots, and the author admits the shelf life of this conclusion by stating, "not extending the analysis beyond 2030"; shifting to blue-collar work is a transitional strategy, not a final strategy.

Fourth, candidly inform children about the magnitude of uncertainty; teaching children to make decisions amid uncertainty is more aligned with the conclusions of this report than teaching any specific skill.

For Those with Investable Assets, there are also four points.

First, write down which scenario your holdings correspond to—under mild scenarios, net returns only increase from 6.5% to 6.6%, and excess returns from AI themes are virtually nonexistent; this point alone can filter out many narratives.

Second, base judgments on observable readings rather than news; the data from 2027 to 2028 will reveal the winner, and any "consensus" before that is an illusion.

Third, understand that the order of beneficiaries is determined by physical bottlenecks: computing power, electricity, grid connection, packaging, and optical interconnects come first, followed by data centers and supporting construction, then cloud and inference services, and finally downstream applications.

Fourth, treat the political risks of redistribution as an independent risk factor—under extreme scenarios, compensation requires 9% of GDP, and under the worst parameter combination, it needs to consume 84% of GDP gains, both of which come from the report itself.

These three lists actually point to the same action. In extreme scenarios, labor is damaged while capital surges; holding capital itself is a hedge against one's own human capital; conversely, human capital concentrated in cognitive groups, with financial assets unrelated to the AI benefit chain, is the most vulnerable combination exposed to damage on both the income and asset sides.

The report also contains a rarely cited comparison: among respondents in cognitive professions, the median believes that only 27% of their work can be done as well as theirs by AI by 2030; yet the same group estimates that 71% of knowledge work will be covered when assessing eight common tasks. The public's judgment of macro trends is not poor; the problem lies in everyone exempting themselves.

Finally, there are two things not to do. Do not treat extreme scenarios as baseline assumptions to leverage— the report assigns no probabilities to any scenarios, and the extreme scenario is also the one with the highest political risk and the deepest wage declines; the parts that benefit capital and those that harm it appear together. Also, do not remain inactive just because "nothing has happened so far"; the report has made it very clear: this fact is compatible with both mild scenarios and the beginning of extreme scenarios.

Before knowing the answer, you must place your bet— and not placing a bet is itself a form of betting!

This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.