Felix Salmon

Many congratulations to the WSJ for putting this chart together; it says much more than the thousand-word accompanying article does. New York City restaurants are periodically graded by inspectors, and then given a grade based on how many violation points they have. And the chart demonstrates, more than any set of anecdotes ever could, that inspectors are trying very hard to give the restaurants the highest letter grade they can.

Now that’s not what the restaurants think, of course. And indeed the fiscal incentives run the other way: restaurants which get B or C grades get fined, and “the amount collected in fines has skyrocketed,” the WSJ reports, “to $42.4 million in the fiscal year that ended last month, from $32.7 million in the previous fiscal year.” (Does a 30% increase really count as “skyrocketed”? I’m not sure about that.)

One theory in the story doesn’t ring true to me at all: Columbia University statistician Roger Vaughan speculates “that restaurant owners learn, after the inspection, about the criteria they are graded on and what the thresholds of the grading system are, and do enough improvements so that at the next inspection they are able to move their scores below 14 violation points or below 28.” But restaurant inspections are much less predictable than that — it’s simply not possible to fine-tune the cleanliness of your kitchen to within a point or two.

Another Vaughan theory doesn’t compel me much either — that restaurant inspectors just don’t want the extra work involved in re-grading restaurants, something which happens to everybody getting a B or a C. The inspectors are salaried, and inspect a fixed number of restaurants: the grades they give out don’t affect that.

Which leaves the hypothesis that “pressure from restaurant groups” is driving the inspectors to give higher grades where it matters. That wouldn’t be the first time, of course, that a regulator was captured by the industry it was meant to be regulating. But of course neither restaurants nor inspectors would ever admit to such a thing, so it’s hard to nail down — until you generate a chart like the one above.

My feeling is that there could be a couple of other factors in play. Firstly, NYC restaurant inspections are harsh, and the inspectors know it. And I remember from my schooldays that the toughest and strictest teachers — the ones we were all most afraid of — were often the ones who ended up giving out the highest grades at the end of the year; something similar might be going on here.

And on top of that is the fact that a simple add-up-the-points heuristic is not always going to be the best way of judging the final grade a restaurant should get. The chart says to me that a restaurant inspection is a somewhat iterative process: you add up the points, you see where the letter grade is, you then compare that letter grade to the letter grade you feel that the restaurant as a whole deserves, and if there’s a difference, you shave off some points here or there. And the inspectors nearly always err on the side of generosity: they’ll shave off points to get you an A or a B, but they won’t be extra-tough to get you a B or a C.

This is no great scandal. Restaurants lose business if they don’t have an A rating, and getting a B or a C costs real money in fines, as well. Given what the public will think when they see a B or a C rating, it’s good that the restaurant inspector takes a minute to ask if that rating is really deserved.

Bad ratings are serious punishment, which shouldn’t just be handed out by an unfeeling computer. Judging, perforce, involves the exercise of judgment. What this chart really shows is that restaurant inspectors are human. Which is something I’m frankly quite grateful for.

Many thanks to Van Tsui and Scott Barber for putting this chart together for me. We’re all used to seeing yield curves — charts which show the yield, for any given credit, at various points along the maturity spectrum. This chart is different: it’s a price curve. It just shows the price at which Greek bonds are trading, plotted according to their maturity.

And it’s really odd.

To understand just how odd this chart is, it’s important to realize that in the Greek bond exchange, there’s only one menu of options for anybody holding a Greek bond. It doesn’t matter if your bond is maturing in six months or if it’s maturing in 26 years, the instruments you’re given the choice of swapping into are all exactly the same.

The way that the bond exchange has been structured, the new Greek bonds are all going to trade at roughly the same price, at least in the first instance. If they didn’t, then everybody would simply pile into the most valuable instrument. We won’t know exactly what price they’ll be trading at until they start changing hands, of course, but I’ve marked a range between 62 and 75 cents on the dollar in the chart. At 62 cents on the dollar the bonds would be trading at an exit yield of 13%; at 75 cents on the dollar they would be trading at an exit yield of 9%. Chances are, when the bonds start trading, they’ll be somewhere in that range.

For the bonds which are trading at or near that range, the exchange makes sense. You swap one bond for another bond worth pretty much the same amount, and Greece gets a bailout at the same time. Net-net, you’re better off.

But for bonds trading significantly above 75 cents on the dollar, there’s a lot of reason to stay out of the exchange. We’re talking the bonds maturing in the next year or two here — for them, you’d be much better off holding them to maturity, or even just selling them on the secondary market.

Odder still are the bonds trading at very low prices, below 40 cents on the dollar, or, in some cases, below even 10 cents on the dollar. What on earth are they doing down there?

All of those bonds can be tendered into the exchange for new bonds which are likely to be worth at least 60 cents on the dollar: it’s free money. Just buy a long-dated low-coupon Greek bond, tender it into the exchange, sell your new bond, and double your money. Why hasn’t that price difference been arbitraged away? And, more generally, why aren’t the blue dots all arrayed in a nice straight line at roughly the level the market expects the new bonds to trade at?

I suspect that what’s going on here is that we’re seeing artifacts of an extremely illiquid market. Over the past year or so, as the Greek fiscal crisis got steadily worse, mark-to-market bond investors sold their bonds to banks, who were willing to pay a premium for them, partly because they were eligible to be used as collateral at the ECB. And the banks are holding on to their Greek bonds now, promising to tender them into the exchange. As a result, there’s no real liquid market in Greek bonds any more, and the prices seen in this chart aren’t available to any old bond investor looking for a quick flip.

Embedded in that mechanism is an implied quid pro quo. The banks will make a lot of money, at least on a mark-to-market basis, by tendering their long-dated Greek debt. In return for that profit, they will tender their short-dated Greek debt as well, even though doing so doesn’t make a lot of rational sense.

This helps explain why the official description of the bond exchange splits the par bond into two seemingly identical parts: a “Par Bond Exchange” where bondholders get a new 30-year bond with a step-up coupon rising from 4% to 5%; and a “Par Bond offered at par value as a Committed Financing Facility,” where bondholders roll maturing bonds into exactly the same thing.

Why make a distinction between the Par Bond Exchange and the Par Bond offered at par value, when both involve swapping your old bonds for the same new 30-year bond? Because the second one, the Committed Financing Facility, can be spun by the banks as them putting new money into the deal. And in a certain sense they are: they’re voluntarily giving up the extra money they could get by holding or selling their short-dated Greek debt.

All of this is in one sense much more complicated than it needs to be. Why construct a hidden and implicit quid pro quo, when you could do a normal bond exchange instead, and swap existing debt into new debt of similar or slightly higher value? Greek bonds, as the chart clearly shows, are worth anywhere from 10 cents on the dollar to 110 cents on the dollar: why should they all be swapped into the same thing?

On the other hand, this is at heart an old-fashioned London Club debt restructuring, as opposed to an Argentina- or Ecuador-style bond exchange. Bondholders look at their holdings on an instrument-by-instrument basis, while bankers are more prone to thinking about their exposure to any given credit as one big risk to be restructured.

So this exchange might well work, insofar as Greece’s debt is overwhelmingly held by banks. But don’t think of it as a template for future restructurings.

Thanks to Larry Summers for suggesting that I take a look at this chart. It wasn’t particularly easy to find, but it’s quite striking all the same. (And thanks very much to Roy Strom and Van Tsui, here at Thomson Reuters for putting it together.)

What we’re looking at here is a ratio of ratios: it’s the price/earnings ratio of the companies in the MSCI USA IT index, which covers technology companies, divided by the price/earnings ratio of the companies in the MSCI USA Industrials index. (For earnings, we’re using 12-month forward earnings — but we’re not really looking at the p/e ratios themselves here, just the ratio between the two ratios.)

Summers is absolutely right: this ratio is currently at an all-time low. The TR data goes back to 1994, so this chart encompasses 17 years, but I suspect you’d need to go back a lot further to find the last time this ratio was trading this low.

Importantly, the ratio is trading at less than 1: the market is saying that earnings at technology companies, which historically exhibit high growth, are worth less than earnings at established industrial companies.

Now this might be an artifact of the specific indices I’m using here: according to another chart sent to me by John Coogan, the ratio is still at an all-time low, but is above 1. Still, the fact is that the market clearly isn’t giving technology earnings the premium they’ve historically commanded.

Why might that be? Frankly I don’t really know. Maybe it’s a function of industrial earnings rebounding less quickly than earnings in the tech sector more generally. Maybe it’s largely an Apple thing. But there’s definitely an indication here that either industrial earnings are too expensive, or technology earnings are too cheap right now. Or both.

This is why I love FT Alphaville in general and Tracy Alloway in particular: she’ll dutifully read 14 pages into something entitled “The Basel Committee on Banking Supervision Joint Forum Report on Asset Securitisation Incentives” before coming across this chart and immediately realizing just how important it is.

I’ve put a bigger version here for people who want to pass it around in all its horrifying glory, but it’s also worth spelling things out, because it might not be immediately obvious.

The big-picture thing to remember when looking at this chart is something which I’ve said many times before — that it wasn’t an excess of greed and speculation which led to the financial crisis, but rather an excess of overcaution, with an attendant surge in demand for triple-A-rated bonds. On a micro level, triple-A securities are safer than any other securities. But on a macro level, they’re much more dangerous, precisely because they’re considered risk-free. They breed complacency and regulatory arbitrage, and they are a key ingredient in the cause of all big crises, which is leverage.

At the left-hand side of the chart we see that global issuance of triple-A bonds was more or less nonexistent back in the early 90s. All those Treasury bonds, all those agency securities from Fannie and Freddie, all that Japanese debt — add it all up, and it still comes to essentially zero by the standards of what seems normal today. Check out the left-hand y-axis: it goes up in $1 trillion increments. And we’re not talking stock, here, we’re talking flows: this chart is issuance per year.

(It’s pretty easy to see, looking at this chart, how a company like Pimco can find itself with over $1 trillion in assets under management: that’s now just a small fraction of the bonds issued each year.)

Now zoom back, and look at the chart as a whole: it’s going up and to the right, which says two things. Firstly, the amount of debt in the world is soaring. That’s a bad thing, because debt is much more systemically dangerous than equity. And secondly, the amount of triple-A debt in the world is soaring as well. Which is a worse thing, because triple-A debt is much more systemically dangerous than most other debt.

Then look at the green line. Triple-A debt wasn’t a huge part of the bond market back in the early 90s, but for the past decade it has invariably accounted for somewhere between 50% and 60% of total global fixed income issuance. That’s possibly the most horrifying bit of all: it simply defies credulity for anybody to be asked to believe that more than half the bonds issued in any given year are essentially free of any credit risk.

Finally, look at the way that the maroon bars — structured products, basically — have given way to a scarily large purple bar at the far right of the chart. That’s sovereign debt, and it tells you all you need to know about where the next crisis is likely to come from.

In a nutshell, triple-A debt is dangerous; there’s far too much of it; its growth seems out of control; and the triple-A problem has now become a sovereign-debt problem, in a world where sovereign-debt crises are the most damaging crises of all.

All that said, there are two things worth bearing in mind which make the chart slightly less horrific. The first is that for reasons I don’t understand, the chart ends in 2009, a crisis year when sovereigns pulled out all the stops in their attempt to prevent a global Depression. We’re more than halfway into 2011 at this point, there’s no good reason why the chart couldn’t include 2010 as well. And that might show 2009 as being a bit of an aberration. Does anybody have the numbers for total triple-A bond issuance in 2010, and how much of that was sovereign?

And secondly, any kind of debt-issuance chart is likely to go up and to the right to some extent, just because borrowing needs never go away, and old debt needs to get rolled over. The total stock of triple-A debt isn’t increasing by this many trillions of dollars per year, and it would be great to see a second chart of how much that is increasing, and how much of it is sovereign.

Still, flows matter. If sovereigns start being downgraded from triple-A status, debt is going to get a lot more expensive, and those rollovers — which cost very little in the current interest-rate environment — will really start to bite. And the invidious thing about debt is that it doesn’t go away. Deleveraging is painful, and is often accompanied by inflation or default. And the more debt you have to start with, the more painful deleveraging is going to be. Prepare yourselves.

Check out page 44 of the Joint Committee on Taxation report on the way that household debt is treated for tax purposes. I’ve put the table into chart form, to make it easier to see what’s going on. Apologies for the rather weird y-axis on the chart: it’s serving a double purpose, counting total returns for the left-hand column and dollars for the right hand column. I would have done a dual axis, but I was having difficulty making that work in Excel.

In any event, the big picture here is clear. Households earning more than $200,000 a year account for less than 10% of the returns, but get 30% of all the benefits. And households earning more than $100,000 a year get 69% of all the benefit. The mortgage-interest deduction might be a middle-class tax break, but realistically it’s an upper-middle-class tax break.

The JCT is also very clear on the two separate ways in which it’s fundamentally unfair, benefiting owners at the expense of renters:

The deduction for home mortgage interest reduces the after-tax cost of financing and maintaining a home. Because the Federal income tax allows taxpayers to deduct mortgage interest from their taxable income, but does not allow them to deduct rental payments, there is a financial incentive to buy rather than rent a home. Taxpayers are also allowed to exclude gains from the sale of their principal residences of up to $500,000 from gross income. There is no such exclusion for other types of investments, further reinforcing the financial incentive to buy rather than rent a home.

Homeowners also receive preferential treatment under U.S. tax law because the imputed rental income on owner-occupied housing (that is, the cost of rent which the taxpayer avoids by owning and occupying a home) is not taxed. Consider two taxpayers: one rents a home at a $1,000 monthly rate, and the other owns a home which carries a $1,000 monthly mortgage. All else equal, a renter pays taxes on a measure of income that includes the $1,000 used to pay rent and the homeowner pays taxes on a measure of income that does not include that same $1,000. If imputed rental income were included in income, it would be appropriate to allow a deduction for mortgage interest, property taxes, and depreciation as costs of earning that income. Because tax law allows taxpayers to deduct mortgage interest and property taxes to determine their taxable income but does not tax imputed rental income or allow them to deduct rental payment, it creates the incentive to buy rather than rent a home and to finance the acquisition with debt.

The mortgage-interest deduction should be abolished, of course — it’s a dreadful piece of public policy. Homeownership, especially during times of high unemployment, does more harm than good, and there’s not even any real evidence that the deduction actually increases homeownership, rather than just artificially making houses more expensive to buy.

But if we’re not going to abolish the mortgage-interest deduction, I like the idea that homeowners should be taxed on their imputed rental income. Think about it this way: I can give you a house, or I can give you the money to buy that house, or I can give you an income stream to pay the rent on that house. The tax consequences of the three are very different, and the last one is the worst: you have to pay income tax on the income stream, leaving you with less money for rent. But if you own a house, and get lots of valuable benefit from it every month, you don’t need to pay any tax on that benefit at all.

More realistically, however, we should just look at the $80 billion a year we’re spending on the mortgage-interest deduction and ask ourselves (a) whether we can afford it, and (b) whether it’s really the best possible way in which we could be spending $80 billion a year. The answer to both questions is clearly no. Especially since that money is going overwhelmingly to the richest households in America.

Last month, in a post headlined “how the mortgage industry lies with statistics,” I bemoaned the fact that I couldn’t find good real data to compare to the massaged data which went into this chart.

What this chart purports to show is that if you’re writing qualified mortgages, the default rate is low whatever the downpayment; it’s the non-qualified mortgages which see enormous default rates above 15%.

But now Glenn Costello of the Kroll Bond Rating Agency has taken the same data from CoreLogic and crunched it for me in exactly the way I requested of Anthony Guarino, the man who put the above chart together. And if you look at the data in a non-massaged way, it looks very different indeed:

The first thing to note is that all of the bars on this chart apply to qualified mortgages: the unqualified ones aren’t even included. And yet the y axis goes much higher than the official mortgage industry’s chart — one of the bars reaches a whopping 40%, about which more in a minute.

The greenish bars, on this chart, are all the mortgages with downpayments of 20% or more, broken down into three groups: downpayments of 20-25%; 25-30%; and more than 30%. (The chart actually shows LTV, or loan-to-value, which is the opposite of a downpayment: to get the downpayment, you subtract the LTV from 100.)

The worst performance for this group happened in 2006, for mortgages with a 20-25% downpayment: they ended up with a delinquency rate of 10.3%. That’s very high, and that single datapoint alone would suffice to show that the 20% downpayment level isn’t a guarantee of safety when it comes to mortgages.

But just look what happens when you compare the mortgages with a 20-25% downpayment to the mortgages with a 15-20% downpayment. Now we’re looking at the bluish bars — they range from 15-20% downpayments all the way to 3-5% downpayments. But for the time being, just look at the lightest blue bar and compare it to the darkest green bar. That’s where the 20% downpayment dividing line happens, and the difference is stark.

In 2002, 6.6% of mortgages with a 15-20% downpayment ended up in delinquency, compared to just 1.5% of mortgages with a 20-25% downpayment. That’s an increase of 340%. In 2003, the numbers are 5.7% and 1.8%; in 2004 they’re 8.0% and 3.5%; in 2005 they’re 13.3% and 8.3%; in 2006 they’re 21.2% and 10.3%; and in 2007 they’re 15.7% and 6.4%. In every case, the gap is huge; in 2006, it’s in double digits.

Remember, we’re talking about qualified mortgages here — the ones the mortgage industry claims are so safe that banks should be allowed to sell all of them off without keeping any skin in the game. All of these mortgages came with mortgage insurance, for instance. And now check out that tall bar from 2007: if you took out a qualified mortgage, that year, with a downpayment of between 3% and 5%, then you had a 40% chance of ending up in delinquency. There was clearly nothing safe about those mortgages at all; even the mortgages with no money down at all did better. (The final maroon bar shows mortgages with 0-3% down.)

The contrast with the official chart could hardly be starker. Let’s look at 2007. According to the mortgage industry, the worst performance seen in the qualified-mortgage universe was a delinquency rate of 6.3%, for loans with 5% or more in downpayment. The truth is that loans with 5-10% down in 2007 saw a delinquency rate of 25%. And you have to get up to a downpayment of 25-30% before you see a delinquency rate of less than 6.3%.

So let’s all remember this chart, the next time anybody claims that you can have a safe mortgage with a low downpayment. Because the fact is that you can’t.

Yesterday something calling itself the Coalition for Sensible Housing policy put out a dense 13-page white paper entitled “Proposed Qualified Residential Mortgage Definition Harms Creditworthy Borrowers While Frustrating Housing Recovery”.

It’s all part of the lobbying campaign surrounding Dodd-Frank, and the eminently sensible idea that if a bank wants to securitize a bunch of mortgages, it has to keep at least 5% of those mortgages for itself. Somehow, in the course of putting Dodd-Frank together, an exception was carved out to that rule, called the Qualified Residential Mortgage, or QRM. For the small group of the most copper-bottomed mortgages, banks could sell off the whole lot, without having to retain 5%.

This gave the mortgage lobby an opening, and they’re attacking it aggressively. They want to open the QRM loophole as wide as possible, and are now kicking up a very loud fuss, complaining that consumers will be damaged if they can’t get access to a QRM loan. The main part of the QRM qualification that they’re upset about is the requirement for a significant downpayment, and so a central part of the lobby’s argument is that if you’re underwriting loans properly, increasing the downpayment doesn’t have much of an impact on delinquency rates. There’s other bits to the argument, too, such as the idea that non-QRM mortgages are going to be much more expensive, but for this post I’m just going to concentrate on the downpayment question.

The white paper explains — in bold type, on page 5 — that “boosting down payments in 5 percent increments has only a negligible impact on default rates”. It continues:

As shown in Table 3 (and in Attachment 2), moving from a 5 percent to a 10 percent down payment requirement on loans that already meet the defined QRM standard reduces the overall default experience by an average of only two- or three-tenths of one percent for each cohort year.

Of course, there are charts and tables. The table comes first:

This is so misleading and confusing that I’ve spent a large chunk of the past 24 hours trying to work out what on earth it’s actually saying, and where the data comes from. The raw data here is indeed being sourced from CoreLogic, and a company called Vertical Capital Solutions did analyze that data, in February 2010. The Vertical Capital report did not, however, have any of the information in this table. Indeed (and inconveniently, from the mortgage lobby’s point of view), it had a whole page which talks about how qualified loans have “substantially higher Delinquencies and Defaults on Qualified Loans with a LTV >80″. (Loan-to-value, or LTV, is the converse of the downpayment: the downpayment and the loan combined are 100% of the loan, since qualified mortgages by definition exclude piggyback loans were second mortgages are involved.)

What the Vertical Capital report does have is a chart of delinquency rates on qualified loans where the LTV is less than 80%, on page 7, and another chart where the LTV is more than 80%, on page 8. Put the two together, and you get something like this:

The difference in delinquency rates between the low-downpayment loans and the high-downpayment loans, here, ranges from 2.94 percentage points for the 2008 vintage, to 7.15 percentage points in 2006. Clearly much bigger differences than are implied in the white paper’s table. And if you look at the percentage increase in delinquency, it’s enormous: all of the delinquency rates more than double, with the lowest increase being 101% in 2006 and the highest being an amazing 502% in 2002.

The mortgage lobby’s own chart, of course, looks very different indeed. Here it is, from page 12 of the white paper:

What this chart purports to show is that non-qualified loans — the red bars — have very high delinquency rates, while qualified loans — the purple, green, and blue bars — have much lower and pretty similar delinquency rates, regardless of the downpayments they use.

But look more carefully. The non-qualified delinquency rates include all delinquencies for all non-qualified loans. But the qualified delinquency rates are not directly comparable, because all of them specifically exclude qualified mortgages with a downpayment of less than 5%.

I spent some time today talking to the man who put this chart together. (It’s sourced to Vertical Capital, but in fact these numbers came from Genworth’s own analysis of CoreLogic’s data, and Vertical Capital did none of this work.) His name is Anthony Guarino, and he’s the vice president of public policy at Genworth mortgage insurance — the company which initially commissioned the Vertical Capital report. I asked him, if he was showing the delinquency rates for all non-qualified loans, why wouldn’t he show the delinquency rates for all qualified loans? Well, he said, “the consortium didn’t want to even talk about zero downpayment mortgages. Why would we even show that? We’d lose credibility if we showed a qualified loan with no downpayment.”

Guarino was perfectly happy to tell me that by excluding all the loans with downpayments of less than 5%, “you’re throwing out the loans with the higher default rates. No one’s saying that downpayment doesn’t matter.” But compare that with the official tone of the white paper:

Based on data from CoreLogic Inc., nearly 25 million current homeowners would be denied access to a lower rate QRM to refinance their home because they do not currently have 25 percent equity in their homes… Even with a 5 percent minimum equity standard, almost 14 million existing homeowners – many undoubtedly with solid credit records – will be unable to obtain a QRM. In short, the proposed rule moves creditworthy, responsible homeowners into the higher cost non-QRM market.

This sounds very much as though even a 5% minimum downpayment is desperately unfair to millions of American homeowners; there’s no indication whatsoever, in the paper, that including a minimum downpayment of 5% in the definition of what constitutes a qualified mortgage might actually be a good idea. Yet when the consortium wants to publish a chart showing the delinquency rates of qualified mortgages, it’s very careful to first strip out any mortgages with a downpayment of less than 5%.

On top of that, the bars in the official chart all look very similar largely because they are very similar: the industry is essentially comparing a set of loans with itself, and declaring that there’s not a lot of difference. The purple bar is all the loans with a downpayment of more than 5%; the green bar is all the loans with a downpayment of more than 10%; and the blue bar is all the loans with a downpayment of more than 20%. The blue bar is a subset of the green bar, which in turn is a subset of the purple bar. The chart is designed, in other words, to look at similarities rather then differences.

I asked Guarino if he could send me the data sliced more naturally: how do loans with a downpayment of less than 5%, for instance, compare to loans with a downpayment of between 5% and 10%? To his credit, he did come back to me with some new data, even if it wasn’t exactly what I asked for: he refused to slice the loan tranches by year, as he did in this graph. Instead, he would only give me aggregate figures, for 2002-2008 and for 2002-2004. Here’s what they look like, charted:

When the mortgage industry starts complaining about the 14 million people who would be denied the chance to buy a qualified mortgage if they don’t have a 5% downpayment, it’s worth remembering that qualified mortgages for people who don’t have a 5% downpayment have a delinquency rate of 16% over the course of the whole housing cycle. (You can be sure the numbers were much higher still in 2006 and 2007, which is why Guarino didn’t give them to me.)

And you can see too why the 20% downpayment limit was put in place: it’s the point at which delinquencies fall to less than 5%. If you take one group of loans with a 20-25% downpayment, and a second group of loans with a 15-20% downpayment, then the second group, on these numbers will have a delinquency rate 56% higher than the first.

The big picture here is that QRM is a distraction, which really shouldn’t exist in the first place. But given that it does exist, the downpayment requirements embedded within it are perfectly sensible. The lower the downpayment, the more likely a loan is to become delinquent. By far. That’s a simple fact which the mortgage lobby will go to astonishing lengths to hide.

Update: Guarino responds in the comments.

Randy Kennedy has a bullish article on Artnet’s nascent art-auction business, which is doing better the second time round than it did the first time round, but which is still tiny. I’m skeptical: value in the art world is very much reliant upon the institutional authority of auction houses and galleries, and Artnet’s auction system is designed to strip out all of that information. It can work for fungible editioned works of relatively modest value, but I do think that most collectors are always going to want a bit more hand-holding before buying art, not to mention the opportunity to actually see the art object before buying it.

That said, the decline of the Sotheby’s and Christie’s duopoly is real, and this is a great opportunity for Artnet or anybody else to try to make a name in a new world with many more players. Here’s some data which Artnet pulled for me, showing total global auction sales, split between the duopoly and everybody else. While sales totals rise and fall, the one constant is the decline of the big guys’ market share:

Next year, it’s more likely than not that Sotheby’s and Christie’s between them will have less than half the global auction market for fine art (which is the data used for this chart). A large part of this is the rise of China, which has hundreds if not thousands of auction houses, and where the big two have very small market share. But even outside China I think that Sotheby’s and Christie’s are both vulnerable, in theory, to lighter-weight business models like Artnet’s auctions or art.sy or the VIP Art Fair. Many of them will fail — but some will succeed. The big two will be faced with a tough choice: do they compete directly with the new entrants, or do they protect their own high-margin core business?

Michael McDonough has this wonderful chart:

Clearly the “new normal” meme is on the decline, and equally clearly “kicking the can” is developing a life of its own, having won the battle of minds against the cutesy rhyming phrases (“extend and pretend,” “delay and pray,” “fake it till you make it”).

I prefer “kicking the can” to “new normal,” on the grounds that it actually contains the tiniest morsel of actual meaning. But still, the cliché has definitely reached the annoying stage at this point. Any suggestions for what is going to supplant it, once it begins its inevitable decline?

Update: Jim Ledbetter informs me that Kick the Can is a popular US children’s game, which has nothing to do with its economic meaning. If it’s not the game which is the source of the phrase, where and when and how did people start talking about “kicking the can down the road”? Is that something people do? And what’s in the can?

If you want to waste a bunch of time this morning, I can highly recommend playing around with the World Top Incomes Database. Click on “Graphics,” and knock yourself out — the database has a wealth of information, going all the way back to the 19th century, and creates lovely charts on the fly; it will even export them in .png format for you.

Here’s one I just put together, showing how in 1994 the top 0.5% of the US population started raking in a greater share of total income than the top 1% of the German population — and how the top 1% in the US are seeing their proportion of total income rise dramatically, even as their German counterparts are seeing their share of total income shrink.

In any case, do play around, it’s a fantastic resource with a massive database behind it. Thanks to Barry for the pointer!

As I secretly hoped that he might, Guan came to the rescue and provided me with exactly what I was looking for — and with Thomson Reuters data, no less! (It comes from SDC Platinum, I should probably befriend someone there.) I wanted a chart of the ratio of foreign IPOs to domestic ones, for U.S. companies, on a rolling five-year basis, to see whether the current level around 10% constitutes a big spike upwards. And the answer is that yes, it does:

Guan cautions that the data from before 1980 or so might not be particularly reliable, since it’s hard to know when a U.S. company lists abroad unless you’re a truly global company. But that doesn’t really matter: the proportion of IPOs of U.S. companies which took place abroad only cracked 2% for the first time in 1999. It stayed between 1% and 2.5% all the way from 1998 through 2004, and then it suddenly started spiking: 7.1% in 2005, 8.4% in 2006, 9.3% in 2007, and a whopping 15.7% in 2008, when 6 companies had IPOs abroad and only 38 managed the feat domestically.

On an absolute rather than percentage level, the record year was 2007, when there were 24 foreign IPOs; there’s a three-way tie for second place, with 17 foreign IPOs in each of 1999, 2005, and 2006.

In any case, the thick blue line is what I was looking for, and it’s going up and to the right about as fast as any five-year moving average is ever likely to.

My next project, which maybe I can find someone at SDC Platinum to help me with, is to have a look at all those U.S. companies which had an IPO abroad — there are 157 of them, altogether — and work out how many of them ended up getting a fully-fledged US listing. Could a listing on, say, London’s AIM end up being a reasonably common bunny slope for U.S. companies which want a cheaper and gentler introduction to the world of being public than a major listing on the New York Stock Exchange?

Remember the Pimco chart which showed emerging countries trading through the G7? Turns out it was a mistake: Pimco was using the SovX Western Europe index, and not the SovX G7 index that it referenced in the relevant footnote.

The Western Europe index includes all of the PIIGS, including Greece, so it’s hardly surprising that it’s trading pretty wide right now. But in fact, if you look at Markit’s official numbers, it’s not trading quite as wide as the emerging-markets index:

How come Markit has the WE index trading tight to EM, while Pimco has the indices the other way around? The answer is pretty technical: the official Markit indices only go back to 2009, while Pimco wanted data going back to 2003. So Pimco put together their own proxy for the indices, which is simply the arithmetic average of the various components in the index. The actual Markit methodology, by contrast, is a bit more complicated: the components are weighted by present value.

Where does this leave Pimco’s thesis? The original chart showed the spread between emerging and advanced economies, and it’s fair to consider all of western Europe to be advanced economies. So the big picture there is absolutely right. But if you just look at the spread between emerging economies and the safest G7 bonds, the movement there is far less dramatic:

There’s much less risk between emerging-market nations and the safest and richest sovereigns in the world than there used to be. But there’s still a significant spread there, of well over 100bp. (It was actually 137bp yesterday.) There’s probably no such thing as a risk-free asset any more. But if you’re looking for something close, you’re still better off in the G7 than you are in the emerging markets.

Thanks to Mohamed El-Erian for pointing this out in his latest Secular Outlook: the market risk spread on advanced economies now exceeds that on emerging economies.

Here’s Pimco’s explanation for what we’re seeing in this chart:

The difference in spreads shows the 5 year Markit CDX.EM.15 index minus the 5 year Markit iTraxx SovX G7 Index Spread. A positive number implies that Emerging Markets sovereign spreads are greater than Advanced Economy sovereign spreads. A negative number implies that Emerging Markets sovereign spreads are less than Advanced Economy sovereign spreads and therefore, the market implied credit risk for EM is lower when the spread is negative.

This is a very big deal, because the names in the EM.15 index are not exactly paragons of creditworthiness. Here’s the list: it starts with Argentina and Venezuela, and goes on from there, including countries like Panama, Russia, and Ukraine.

Meanwhile, the SovX G7 list is short and powerful: Germany, France, Japan, Italy, UK, and USA.

There are probably technical reasons why a group of AAA-rated sovereigns is trading wide of a group of much less creditworthy emerging markets in the CDS market. But the big message here is clear: the world is being turned upside-down. And most investors have yet to even start adjusting to these new realities.

Update: Turns out that the chart wasn’t measuring the G7 spread after all, but rather the Western Europe spread, which includes all the PIIGS.

How do we know that the world is getting happier? It’s drinking more beer! Here’s the chart, from a new paper by Liesbeth Colen and Johan Swinnen:

What we’re seeing here is largely the China effect — and, more generally, a world where poor people, once they reach a certain minimum income, start hitting the hops.

By all indications, we’re still in the early days of this trend, whereby countries slowly converge in terms of per-capita beer consumption. For while China and Russia are soaring, the main beer-drinking nations of the world are all in decline:

In middle and low income countries which experience growth, such as China, Russia, Poland and India, beer consumption grows. In rich countries, however, further growth has led to a reduction in beer consumption per capita.

This is an economics paper, so of course there has to be some kind of regression analysis — in this case OLS, or ordinary least squares:

Our first important result is that we do indeed find an inverted-U shaped relation between income and per capita beer consumption in all pooled OLS and fixed effects specifications. From the pooled OLS regressions (Table 3), we find that countries with higher levels of income initially consume more beer. Yet, the second order coefficient on income is negative, indicating that from a certain income level onwards, higher incomes lead to lower per capita beer consumption. The first and second order effects for income are strongly significant and the coefficients are quite robust across the different specifications.

The fixed effects regression results confirm this (Table 4), so the non-linear relationship for income holds not only between countries, but also within individual countries over time. As a country becomes richer, beer consumption rises, but when incomes continue to grow, beer consumption starts to decline at some income level. We calculated the turning point, i.e. the point where beer consumption starts declining with growing incomes, to be approximately 22,000 U.S. dollars per capita.

I would imagine that this relationship could also be found within the U.S. — that states increase their beer consumption as they grow to an income of about $22,000 per capita, and thereafter see their beer consumption drop as their wine consumption increases.

I can also imagine that we’re going to see a China-driven surge in global wine consumption when the middle class population there starts earning that kind of money. In the first instance, most Chinese wine consumption will probably be domestic, but over the long term it’s surely inevitable that wine imports into China will stop being concentrated at the high end of the market and will start lubricating China’s middle classes on an everyday basis. But that’s probably not going to happen for a decade or two yet.

(Via Florida)

Many thanks to Steven Guichard and Zubin Jelveh, who picked up my gauntlet and ran the numbers on whether payrolls numbers released at the beginning of the month are any less accurate than payrolls numbers which come out a bit later.

Here’s a chart from Steven, showing the average absolute change in payrolls numbers between the initial report and the final figure, according to the day of the month that the initial report was released.

And here’s a chart from Zubin, showing the same data in a slightly different form: in this case every revision is a dot. I’m not entirely sure what the outlier is on the right-hand side*, but the message of both charts is clear: there’s no indication at all that payrolls reports released later in the month are any more accurate than those released earlier in the month.

Steven also provided this intriguing graph, which shows that payroll data got significantly more accurate in the 80s and 90s, but seems to be getting less so since the crisis struck.

The average revision seems to be somewhere in the 50,000 range, which is consistent with the official error bars:

The confidence interval for the monthly change in total nonfarm employment from the establishment survey is on the order of plus or minus 100,000. Suppose the estimate of nonfarm employment increases by 50,000 from one month to the next. The 90-percent confidence interval on the monthly change would range from -50,000 to +150,000 (50,000 +/- 100,000). These figures do not mean that the sample results are off by these magnitudes, but rather that there is about a 90-percent chance that the “true” over-the-month change lies within this interval. Since this range includes values of less than zero, we could not say with confidence that nonfarm employment had, in fact, increased that month. If, however, the reported nonfarm employment rise was 250,000, then all of the values within the 90-percent confidence interval would be greater than zero. In this case, it is likely (at least a 90-percent chance) that nonfarm employment had, in fact, risen that month.

It’s important to remember here that there’s an enormous difference, as far as the markets are concerned, between a payrolls number of say 70,000 and a number of 170,000. But each one is within the confidence interval of the other. So the lesson, as I suspected, is that we should treat all payrolls reports with skepticism, but pay no attention to the day of the month they’re released.

*Update: In the comments, Zubin reveals that the outlier dates from December 1995, when the report was set to come out the 5th but was delayed because of a government shutdown. Anybody taking bets on when the payrolls report will be released next month?