This is how the AI bubble pops

 This is a long but fascinating video.  Cory Doctorow talks about what AI is, what it can and can't do, and why the AI bubble will pop and what it will look like when it does.  

He talks about "process knowledge" —  how firms depend on the unwritten knowledge that makes them work.  His example is how a machine breaks down, and how new workers are told that you contact the guy who used to fix that machine or its predecessor, and who has retired but for £50 will come in and fix it.  My example is Mrs McGinty in dispatch who knows whom to contact when things go wrong, who will be late, who can be relied on, and who needs to be chivvied.  When she retires or is fired (to save money!) dispatch stops working.  His point is that workers will be laid off because AI will be able to do their jobs, except it won't, so the workers will have to be rehired, only they will have moved on.  So critical process knowledge will be lost.

He and his interviewer compare how the railway boom popped, but the capital stock (the railways themselves, the rolling stock, the property infrastructure) remained.  All governments had to do was to pay off the creditors, and give the bankrupt companies enough money to get them restarted.  Once restarted, they could keep going by themselves.  But AI isn't like that.  There is no equivalent of railways tracks, and the rolling stock depreciates in a couple of years rather than over 30 years, because Nvidia continually improves the chips used by making them more and more specialised, but doesn't bother to retrofit its newer chips to the older ones, which means that every three years, data centres have to be levelled to the concrete and completely refitted.  There isn't enough money, private or public, which will be able to save these companies in the same way the railways were saved. 

He discusses how Nvidia sells chips, then lends the money to buy those chips to its customers ("circular finance"), but nevertheless still counts those "sales" as sales, and says it is making a profit on them, when its customers are unlikely ever to repay their debts.  The debt structures needed to fund data centres are vulnerable to a credit crunch, because most of the credit is provided by risk-averse investors who have written many escape clauses into the contracts.  These include connecting to the grid, diesel generators, sourcing water for cooling, completing construction on time, and if these targets aren't reached because too many data centres are being built or opposition is too strong, there will be a cascading collapse.

There are many more insights and examples, too many to try to summarise here.  But he is the first analyst I have seen who clearly really understands AI and can also communicate well with non-nerds like me.

My two key conclusions from watching this video are:

• AI is not going to take over the world.  It can't, because it's not actually intelligent.
• the AI bubble will pop, and that is likely soon, because credit is tightening, Central Banks are raising interest rates, and bond yields are soaring.

World bonds continue to sell off

The rout in bonds continues.  

In my experience, bond investors are realistic and canny and cynical.  Unlike share investors, they're not seduced by "blue sky" arguments.  They see improving growth, stubborn inflation, an endless Iran war, enormous supply of paper from the US, the risk of retaliation by the US's former allies.  All bad for bonds.

Rising government bond yields lead to rising cost of corporate debt, especially in the dodgy private credit market, which is funding the AI bubble.  And that will be complemented by rising discount rates.  Credit will tighten.  None of these things is good for growth.  Or the stock market.

Short-term chart — going to new cycle highs (US cycle high in late 23).

Long-term chart — near or past 20 year highs, decades-long bond bull market kaput, underpinnings of 20-year share-market bull market over.   

New tightening cycle beginning?

The chart shows different measures of Central Bank discount rates (= cash rates/bank rates/Fed Funds rate).

The green line is the GDP-weighted average.  It would be biased towards countries with a high percentage of world GDP, such as the US, Europe, Japan, the UK, etc.  It covers countries which make up 83% of world GDP.

The blue line shows the median discount rate for all the countries I monitor.  The median is the midpoint of a range of data points.  So, currently, the median D/R is 4.30%.  Half the countries of the world have a lower rate than 4.3, and half a higher one. 

The red line shows the unweighted average.  Because so many developing countries have high inflation and therefore a high bank rate, this average is skewed towards these countries.  But it does reflect the actual bank rate the different countries of the world face.   

All measures have started rising.  Long-dated bond yields are also rising (see lower chart).

This means that credit is starting to dry up, which is typically not good for the world's share markets.  The initial impact of credit tightening tends to be felt in financial markets.  Later on, it spreads to the real economy.

So far, the tightening has been modest.  But if inflation remains stubbornly higher than CBs want, and the oil market remains tight, then the probability is that economic growth will, after the usual lag (~12 months) start to falter.

Meanwhile, the favourable winds driving the start market higher, are turning to headwinds.  And credit costs for the debt inflating the AI bubble are soaring.  And spending on AI has been driving economic growth in the US, and therefore also the world economy.  If (when!) credit for AI dries up, there will be a credit crunch and economic slump comparable to the 2008/2009 GFC.

There is a lot of risk out there.  Take care.

Will world interest rates fall further?

This chart shows the GDP-weighted world central bank discount rate (bank rate/Fed Funds rate/central bank lending rate) covering 83% of the world economy, as well as the median of world interest rates that I monitor . The median is the point in a series of data values at which half the observations are above and half below.  It is unweighted.  

The two different kinds of average interest rates usually move in more or less the same direction, though not always.  The GDP-weighted average will be dominated by the largest economies, the median is skewed more towards smaller economies, because they are more numerous.  Note how the median interest rate rose much more than the average before the 2009 GFC, likely making the recession worse; and how the jump in the median rate in 2011-2012 signalled that smaller economies were in trouble, worsening the downturn linked to the euro crisis.

Right now, the median is falling faster than the GDP-weighted average, which is consistent with my "small 15" average PMI index, which has been much weaker than the "big 8" index.  

Will interest rates fall further?  Well, yes, but not by much.  The ECB (European Central Bank) seems happy with its bank rate; the Fed is muttering about not cutting rates again; and the "small 15" PMI is rising fast, meaning smaller economies have become more reluctant to cut rates.  At the same time, world inflation has levelled off after falling from the post-Covid highs.  What is certain is that the low interest rates of the 2009 to 2021 years will not be reached again in this cycle.  Unless the AI-bubble pops .....

Clicking on the chart will make it easier to read.

World growth should be picking up, but ....

The chart below shows two different measures of average world Central Bank discount rates.  The first is a GDP-weighted average covering the countries which together make up 83% of world GDP, the other shows the median discount rate (unweighted by GDP).  Half (by number) of CB discount rates are below the median, and half above.

CBs raised their discount rates when inflation rose during the rebound from the covid crash, and have been cutting them as inflation fell and economic growth stopped.

Economies respond to falling (or rising) interest rates with a lag, which varies from cycle to cycle, and even from country to country, but which is something between 1 and 2 years, so we would expect world economic activity to be now accelerating.  Trump's trade war has complicated this normal process, and though its effect will be mostly felt in the US, it will have some effect on global growth.  The US is stagnating; in other countries on the whole, growth is picking up.  However, it may be too early to see just how bad swingeing US tariffs, combined with a US recession, will be for US trade partners.  If the effect is negative, you can expect CBs outside the US to cut rates faster, but the Fed may be unwilling to cut the Fed Funds rate as inflation accelerates.

As usual, click on the chart to see a clearer image. 

Two regional Fed indicators are strengthening

 Most indicators are pointing down, but a couple are rising.  The average of the Empire State and Philly Fed indicators is trending up after falling since the beginning of the year (Trump's tariff pagaille).  It's true, this only covers the NE United States, so isn't definitive.  However, it fits well with the average of the two national surveys by the ISM (Institute of Supply Management) and the PMI (Purchasing managers' index, from S&P Global):

Here's the thing:  if the US is re-accelerating, the Fed simply won't cut rates again.  Not while the rise in costs because of tariffs is still working its way through the system.  They do not want inflation to become embedded into the system, and if the economy is recovering, they have no need to.

We'll see.

Ozzie recovery continues

The chart shows the PMIs for Australia from Judo Bank, smoothed with 5-month centred moving averages.  (Data through April).  Note the three downward spikes caused by Covid lockdowns in 2020, 2021 and the end of 2022.

Oz's main export markets are China and Japan, not the US.  So we won't be directly affected by Trump's tariffs.  But we could be indirectly affected by downturns in both countries, consequent on tariff increases; world growth is likely to slow, and may have already started to.

Given that inflation is within its target range (2 - 3%), our exports could be hit, and the A$ is strengthening against the US$, the RBA will likely cut rates a couple of times over the course of the next year.  You will be happy to hear that's also the consensus of my economist colleagues.

[Update:  Australia's PMI surveys are no longer sponsored by Judo bank, but by S&P Global themselves]

World interest cycle has clearly peaked

Despite some contra movements (for example, Brazil is raising its Central Bank rate), world interest rates have broadly peaked.   The ECB has just cut the discount rate for the Euro Area (the countries which have the Euro as their currency), which has been followed by other European countries.  The Fed Funds rate (the US's equivalent of the discount rate or bank rate) will prob'ly not be cut again in the near term because of the inflationary impact of Trump's tariff and immigration policies.  However, other countries will prob'ly cut their interest rates even faster to reduce the impact on growth of Trump's trade wars.

In the chart shows the GDP-weighted average discount rate for 83% of the world.  The other shows the unweighted median interest rate, including many additional countries in the analysis (e.g, Angola, Kenya, Ukraine, Argentina, and others.) 

Is the world economy really recovering?

 Or has it started a new recession?

I collect and maintain hundreds of time series from countries all round the world: North America, South America, Asia, Africa, Europe, Australasia.   I have a program which adds together all the series and produces a composite index of them all.   It's unweighted, but the number of component series is highest for the US, China and Europe.  Most countries are represented.  I monitor PMIs, industrial production, retail sales, unemployment rates, inflation rates, business and consumer confidence, money supply, car sales, and so on, for most of these countries, and for the big ones, many more.  For example, "jobs easy to find" in the USA, and "production of bullet trains" in China.  Where necessary, I seasonally adjust each time series.  My world composite index is made up of 430 series.

The result is shown below.  The year-on-year rate of change in my world composite index is compared to the rate of change in my weighted world index of industrial production.

The YoY % change in my world IP index is barely above zero, and for my composite index, it's off its lows, but is still below zero.  In other words, it's still falling, but is falling more slowly.

This is consistent with a sluggish recovery, but not a new recession.  This view is confirmed by my US leading indices, which point to the strong likelihood of a sustained recovery, though as I said in my previous post, my leading indices do not include fiscal stimuli.  Be that as it may, my US leading index leads the world economy by +-6 months, and it's only levelled off in the last couple of months.  This recent blip is not at all consistent with a deep recession.

So why have world stock markets fallen so sharply? The reason is probably because there was excessive optimism.  Investors and traders believed that there would be a strong recovery, and marked up share prices accordingly, misled perhaps by the rebound from "revenge spending" in services.  There was also a nice little bubble around AI, which has been hyped.  It reminds of the dot-com boom in the early 2000s.  The market is beginning to doubt that AI will make much money for the global IT behemoths.

I don't think there will be a renewed recession, what has been called a "double dip" recession.  But I do think there will be a very sluggish recovery for the next few months.  China is weakening; Europe is struggling, and the rate cuts by the ECB (European Central Bank) have been too little, too late; the US is slowing as fiscal stimulus drains out of the system, and the Fed should have started cutting rates a few months ago.  Emerging markets are in strife, because their currencies are falling because of  "the flight to quality", imposing new and unwelcome constraints on their economies.

My guess is that the fears engendered by the stock market plunge will bring forward and accelerate rate cuts, not because Central Banks care about investors per se, but because sharply falling share prices suggest that seasoned observers of the economy have seen what is happening more clearly than they have.  And of course, plunging share prices will affect business and consumer confidence.

Interesting times.  

Central Banks were too late raising interest rates after Covid, and are now too late cutting them.  Their staffs are paid quite a lot to get things so wrong.

World economy picks up

 I often forget that though I have lived through the events depicted in my charts, others haven't.  The chart below shows my calculation of the "Big 8" GDP-weighted PMI, which, coincidentally, covers about half my career in the financial markets.

So, the events: the 2001 recession, the GFC (2008/9), the euro crisis (2012/13), the 2015/16 pause, the 2016-17 Trump tax boom (nothing like deficit spending to give a growth sugar rush), the 18/19 slowdown as fiscal stimulus faded, the 2020 Covid crash, the 2020/21 post-Covid rebound with massive monetary and fiscal stimulus, the inevitable hangover in 2022/23 as fiscal stimulus faded, and interest rates were hiked, and war pushed up inflation, and now, the 2024 recovery.

My guess, after nearly 50 years in economics and the markets?  The recovery will continue.  But it won't be the steep slope of the post-covid recovery, but something shallower.  Which will stop CBs raising rates.  Will they cut rates?  Some will.  Most will --- because inflation is drifting lower and growth won't be fierce enough to push it back up again.  But inflation will be sticky downwards, for reasons I'll discuss in another post.  So the rate cuts won't be very large.

(Data through May 2024; the Big 8 are the USA, China, Japan, Euro area, India, Russia, Brazil, the UK, and they make up +- 70% of the world's economy.)

Asia booming

 This chart shows the average (GDP-weighted) PMI for Asia, i.e., Japan, China, India, Indonesia, Taiwan, Korea, Malaysia.  Still to add Thailand and Vietnam to the calculations.

Asia (+-30% of the world economy) is picking up nicely, rebounding from the impact of draconian Chinese covid lockdowns.

The implications for Central Bank policy are obvious.  

World economy continues to recover

 S&P Global has released the provisional ("flash") estimates of the PMI indices for May.  The PMI indices are among the earliest data points available for the state of the economy.  The survey asks whether sales, employment, orders, etc are up or down on last month, but not by how much.  S&P Global then produces country indices for manufacturing and non-manufacturing/services.

I take these time series, extreme adjust them, and add them together, each weighted by that country's weight in world GDP (using purchasing power parity, or PPP, exchange rates to value national currency real GDP). 

The Big 5 are: the USA, the UK, the Euro Zone (European countries which use the euro currency), Japan and India.   The big 8 adds China, Brazil, and Russia to this calculation.

The chart below shows the Big 5 and the Big 8 GDP-weighted PMI averages, with manufacturing and service PMIs averaged (= "whole economy").  Since we don't have "flash" PMI estimates for China, Brazil and Russia, Big 8 PMI is only available to April.

Clearly, the world economy is accelerating.  Not only is the Big 5 PMI above the 50% "recession line" indicating that the economy is advancing, but it is also rising, i.e., the economy is accelerating.

The markets' conclusion that interest rates are likely to fall more slowly is correct.  And it is also likely that inflation will fall more slowly, too.

US upturn gathers pace

 To disentangle the signal from the noise, I have first extreme-adjusted the individual time series (the ISM and the PMI survey data time series) and then I have averaged them.  Extreme-adjustment is an algorithm which removes or reduces the impact of extreme events.  In particular, over this period, it reduces the impact of the Covid Crash in early 2020.  In addition, averaging two statistically independent series reduces the random fluctuation (error term) in the resulting index.  Both these techniques reduce the "spikiness" of the time series, making the chart easier to read.

Right now, the PMI is giving a more robust picture of the US economy, and the ISM survey a slightly weaker view.  This pattern has been the other way around in the past, for example in 2017/18.  But the line to watch remains the thick green one.  It points to a renewed upturn, as the economy shakes off the impact of the Fed's policy tightening, as government spending and private investment on the back of IRA tax incentives push the economy forward.

Good news for Biden's re-election.    But it also means that the Fed will be reluctant to cut rates, until it gets evidence that inflation is continuing to decline.

US core PCE inflation remains elevated

 The chart below shows the year-on-year increase in the core PCE (personal consumption expenditure) deflator.  This is one of the Fed's key inflation indicators.  

Though down from Q4/22 levels, it's no longer falling.  This points towards further rate rises by the Fed.

click on chart to see clearer image

Just how deep will the US recession be?

Readers of this blog will know that I have worried about this issue for many months now.  

Here are two more charts which strongly suggest that a deep recession is possible.   They show data back to 1970.

The first chart shows the 12-month change (absolute, not percentage) in the Fed Funds rate.   The Fed Funds rate is the rate the Federal Reserve Bank charges for overnight borrowings by banks from the Federal Reserve system.  It is the bellwether which sets the level of interest rates in the market.   In the chart, I have plotted it upside down, because when interest rates rise, the economy slows, and when they fall, the economy accelerates.  I have also moved it forward, that is, I have plotted it with an 18-month lag, because it takes a long time for the change in rates to take effect.  

The QCI is a monthly coinciding index (i.e., it coincides with the business cycle), which closely tracks real GDP.

Notice how close the relationship between the change in rates and economic activity is, except for the GFC (2008/2009), where the downturn was materially worsened by the half-witted orgy of ill-advised mortgage lending by the banks in the years prior to the Fed raising rates.  The inevitable hangover took down the banking system and everything else with it.

The other apparent break in the relationship was caused by the Covid Crash of early 2020.  This also caused the spike in the QCI in early 2021 via the year-on-year calculation.

The rise in rates (shown in the chart as a fall, because, remember, it's inverted) is consistent with the deep recessions of 1973-1975 and 1980-1982.  The only good news is that this indicator (the inverted change in Fed Funds) is bottoming---provided the Fed doesn't raise the Fed Funds rate from now on.

Click on chart to see clearer image

The second chart shows the rate of change in real (i.e., after allowing for inflation) money supply.  The data for both M1 and M2 appear to have been distorted in early 2020 by a definitional/regulatory change by the Fed affecting the classification of chequing and savings accounts.   It's partly this change which may have caused the spike in the year-on-year rate in 2020/21.  But the spike was prolly also caused by the Fed flooding the system with liquidity.  However, this distortion is now passed.

Real money supply is falling faster than it has done at any time in the last 63 years (1960-1970 not shown in the chart).   It's falling faster than it did before the 1973–1975 and 1980–1982 recessions, and they were, before the GFC, the deepest recessions the USA has experienced since the Great Depression.  The implication is that the US is likely to experience a recession as deep as those two.  In other words, not a "soft landing".    See this interesting analysis from Reuters.

Click on chart to see clearer image.
Note: money supply charts not plotted with a lag.

So I remain sure that the US will experience a recession this year, with a peak-to-trough fall in real GDP of 3-5%, and a rise in the unemployment rate of 5-6%.

What could prevent these outcomes?   Well, Covid has distorted many indicators, and many economic relationships.  So time-honoured linkages and drivers may no longer work.  I think that, though possible, this is very doubtful.  

Of the Big 8, the US, Europe, the UK, Brazil and Russia are themselves likely to go into recession or are already in one this year.  China is rebounding, and India is still growing strongly.  On balance, the rest of the world won't save the US.  In the past, it usually went the other way---US recessions transmitted themselves to the rest of the world.   (The China rebound will likely reduce the depth of Australia's recession, though our Reserve Bank has also raised rates by too much.) 

During the 1973-1975 recession, the S&P500 fell by 44% peak to trough.  During the 1980-1982 recession, it rose at first before falling by 20%.   During the GFC, it fell by 57%.  However, the banks are much better capitalised now than they were when the GFC began, and the kind of financial crisis we experienced then is less likely (but by no means impossible) today.  So far this cycle, the S&P500 is down just 14%.  The risks, surely, are on the downside.