A Dire Forecast from Trump and Elon

The Three Things in the Warning Do Not Move Together

A warning that names three categories of wealth and predicts that all of them will be destroyed at once is making a claim about correlation, not about direction. The direction is the easy part. The correlation is the part that determines whether the warning describes anything that has ever happened.

The three categories in the standard version are equities, residential property and savings. They are habitually bundled because they are the three places most households hold wealth, and the bundling implies that a single event reaches all three.

The historical record does not support that implication as a general rule. It supports it in one specific configuration, which is worth identifying precisely, because that configuration is rare and has identifiable preconditions.

The clearest counterexample is the equity decline that began in 2000. The broad American index lost roughly half its value from peak to trough over about two and a half years. Residential property prices rose throughout that period, and rose substantially. Households whose wealth sat mainly in a house experienced those years very differently from households whose wealth sat mainly in a brokerage account.

The 2007 to 2009 episode ran the other way. Both fell, and fell together, which is why it is remembered as the example of everything going at once.

The difference between the two is not mysterious. In the first, the shock originated in equity valuations of a particular sector and the banking system was not impaired. In the second, the shock originated in the assets banks held against the housing stock, which transmitted it everywhere simultaneously.

That is the general condition: the three move together when the credit system is the channel, and otherwise they do not. A prediction that they will move together is therefore a prediction about the banking system rather than about market sentiment, and it should be assessed on banking evidence.

What follows is what happened the last time equities halved, why housing and equities are weakly linked, what the word savings is actually doing in the sentence, the configuration in which the correlation appears, and what broad stress looks like in published data.

What Happened the Last Time Equities Halved

The episode beginning in 2000 is the useful comparison because it was severe, protracted, and did not spread the way the warning assumes.

The broad index peaked in March 2000 and reached its low in October 2002. The decline from peak to trough was close to half, and the technology-heavy index fell considerably further.

Over the same period the national home price index rose. The rise was not marginal; it continued through the equity decline and accelerated afterward. Commentary at the time explicitly described households moving toward property as the safer alternative.

Employment weakened but the banking system did not. Bank failures over the period were few, deposit insurance was not tested at scale, and credit remained available to borrowers who qualified.

The recession that accompanied it was, by the standard dating, short. The equity decline outlasted the recession by a wide margin, which is itself informative: the market decline and the economic contraction were not the same event and did not share a timetable.

The lesson is narrow and it holds. A halving of equity values is compatible with rising house prices, functioning banks and available credit, because those are separate systems that a valuation correction in one sector does not automatically reach.

There is a further detail in that episode worth carrying. The equity decline was highly concentrated by sector. Broad indexes fell about half; the technology-weighted index fell roughly four-fifths; and several large sectors ended the period close to where they began. A statement that the market halved describes an index, and an index is a weighted average of components whose outcomes differed enormously.

Why Housing and Equities Are Weakly Linked

The structural reasons for the weak relationship are worth setting out, because they explain why the 2000 case was normal rather than anomalous.

A house is a consumption good as well as an asset. It is occupied, it cannot be sold in fractions, and the transaction costs of moving are large. That alone dampens the response of prices to sentiment.

Housing is local. There is no single national housing market, and price movements across metropolitan areas have historically diverged widely. A national index averages markets that can be moving in opposite directions.

Valuation runs through a different variable. House prices respond principally to the cost and availability of mortgage credit and to local income and supply, while equity prices respond to expected corporate earnings and the discount rate applied to them. The two share the interest rate and little else.

And the price data itself is slow. Repeat-sales indexes are built from completed transactions, published with a lag of months, and smoothed over multiple months of sales. A housing index cannot move sharply the way a quoted market can, because the instrument is not capable of it.

The composite effect is that the two asset classes respond to different drivers on different timescales, and the periods when they move together require a common cause strong enough to override all of that.

What the Word Savings Is Doing

The third item in the warning is the one that most repays attention, because it is not a single thing and its components have entirely different exposures.

Money in a deposit account at an insured institution is protected up to the standard maximum, currently two hundred and fifty thousand dollars per depositor, per insured bank, per ownership category. Its nominal value does not fall when markets fall. It is exposed to inflation, which is a real risk and a different one.

Money in a retirement account is not a category of asset at all. It is a tax wrapper containing whatever the holder chose — equity funds, bond funds, target-date funds, cash — and its behavior is determined entirely by those contents rather than by the wrapper.

Money in a brokerage account is exposed to the prices of what it holds. Separate protection covers the failure of the broker rather than declines in the value of the securities, which is a distinction that matters precisely when people become anxious about both at once.

A defined-benefit pension is a claim on an employer or a plan, backed in part by a federal guarantee corporation up to statutory limits, and its risk is the sponsor's solvency rather than the market's level.

Using one word for those four is what allows a warning to sound comprehensive. Separating them is what makes it assessable, and the separation takes a few minutes with a statement.

The Correlation That Appears Only in the Worst Case

Having established that the assets normally diverge, the conditions under which they converge deserve stating, because those conditions are the real content of any serious warning.

The mechanism is leverage in the financial system. When institutions hold assets financed with short-term borrowing, a fall in the value of those assets forces sales, and forced sales fall on whatever can be sold rather than on whatever has become overvalued.

That is why a shock originating in one asset class reaches others: the selling is determined by funding needs, not by views. The correlation between asset classes rises precisely when diversification is most wanted, which is a documented and much-studied regularity.

Housing enters this channel through the banking system specifically, because mortgages sit on bank balance sheets and in securities held by leveraged institutions. A decline in housing collateral impairs lenders, and impaired lenders withdraw credit from everything.

Deposits are the last thing to be affected and are affected differently: the exposure is institutional failure, which is what deposit insurance exists to address, and the record of insured depositors losing money is essentially empty.

There is an asymmetry in how quickly the channel opens and closes. Correlation rises abruptly under funding stress and subsides slowly afterward, which means the historical average correlation between two asset classes understates what happens in the episodes that matter and overstates what happens the rest of the time. A single number describing the relationship is misleading in both directions.

The practical implication is that a warning about all three at once is a warning about bank balance sheets. Anyone making it seriously would be pointing at bank balance sheets, and those are disclosed quarterly in considerable detail.

What Broad Stress Actually Looks Like in the Data

The indicators that would register a genuine systemic problem are published, frequent and free, and they move before headlines do.

Credit spreads are the first. The gap between yields on corporate debt and government debt of the same maturity widens when lenders demand more compensation for risk, and the series is available daily.

Bank lending standards are the second. A quarterly survey of senior loan officers asks banks directly whether they are tightening standards and whether demand is falling, and the responses lead changes in credit availability.

Funding market indicators are the third. Spreads between secured and unsecured short-term rates indicate whether institutions are willing to lend to one another unsecured, which is the measure that moved first in the episode everyone remembers.

Bank capital and asset quality are the fourth, reported quarterly in regulatory filings with delinquency rates, charge-offs and capital ratios by institution.

And the supervisory stress test results are the fifth, published annually with the scenarios stated, which means the regulator's own estimate of how the largest banks would perform under a severe downturn is a public document.

The general observation is that the three categories in the warning respond to different forces except in one configuration, that the configuration is a credit event rather than a sentiment event, and that every indicator which would signal it arriving is published on a known schedule by an agency with no product to sell.