Is China recovering?

The new gdp figure came in at 6.7%.  No matter what you believe “the real number” to be, this is probably more important:

Chen Xingdong, China chief economist at BNP Paribas, notes that first quarter growth was bolstered by industrial production, fixed-asset investment and an “astonishing” acceleration in construction starts while service sector growth moderated. That is exactly the opposite of what is supposed to be happening. “The pick-up in SOE investment and slowdown in private sector investment will cause problems for structural change,” Mr Chen said.

The first quarter this year also saw record credit expansion in China, even though most economists believe the country needs to be deleveraging. Here is David Keohane citing Wei Yao:

In Q1, increases in total credit exploded to CNY7.5tn, up 58% yoy and equivalent to 46.5% of nominal GDP – one of the highest ratios ever. Credit growth accelerated to 15.8% yoy to end-March, the quickest pace in 20 months.

Also from the FT (see the first link) is this:

Meanwhile, the “Mr Li got lucky” argument suggests that the most powerful player is not the country’s much feared president, Xi Jinping, but rather Janet Yellen, chair of the US Federal Reserve. According to this theory, Ms Yellen’s pause on US rate rises saved China from what looked like the beginning of run on the renminbi and Beijing’s foreign exchange reserves, which fell precipitously in January and February.

These falls moderated only after the Fed suggested it would not raise interest rates as aggressively as it had indicated late last year.

“The Fed’s reversal has taken a lot of pressure off the renminbi and without the currency looking like it’s going to collapse, people are feeling better,” said one Asian investment strategist, who asked not to be named.

The simplest China model for 2016 is this.  Due to the prevalence of SOEs and state influence in the economy, the country can in fact (for now) achieve almost any gdp target it wishes, at least within reason.  But it trades off the quantity of gdp for the quality of gdp, and this time — again — the Party opted for the relatively high growth figure.  That is bad news, not good news.

Here is an unpacking of some parts of the gdp number.  Here is Nerys Avery.


'But it trades off the quantity of gdp for the quality of gdp'

And yet, not so simply, quantity has a quality of its own.

Sadly, in this case, its a bad quality.

Can you unpack a little bit what you mean by "quality of GDP", or point to some introductory text on this? While I get intuitively what you mean, I'm having trouble making headway here in trying to model this precisely. Much of the literature that I can find goes back to Soviet times and the economic measurement issues there, but that has it's own gigantic set of confounding factors...

Marginal return on debt.

I think it means Tyler predicted a recession in China this year.

No Tyler was careful not to make any prediction that he could be held to.

Tyler from last August 25:

"You can think that China is in for a very bad recession, as I do."

Sumner says "Stop predicting recessions!":

Usefulness of what they get for the marginal renminbi.

If we printed an extra $10 trillion, we could launch a new WPA and pay all the unemployed to dig a bunch of ditches, or pay a bunch of contractors to do some BS projects, both of which would raise GDP. Neither, however, would contribute $10 trillion in value.

China is printing money and spending it on projects of dubious return. This gooses the growth figures but does not actually increase productive capacity, increasing the magnitude of the eventual snapback unless they can achieve enough actual growth to catch up.

I would guess that you have never been to China. Its difficult to say that their economy is not growing in real terms when you see all the new roads and buildings going up, plus all the people in the shiny new malls.

Also the Chinese are not in hyperinflation mode, so if they are "printing money" no-one seems to be being harmed.

This dumbed down Austrianism is all over the net, that "money printing" is leading to malinvestment. It always seemed a silly idea to me, when you see economies that are in hyperinflation, like Zimbabwe or 1920's Germany, excessive over investment was not their problem.

I have been to China, and all those roads and shiny new malls are great. The malls are private investment anyways.

The problem is the white elephant projects like conference centers in every tier 3 city that have to look like artwork.

I could point to the malinvestment in Las Vegas houses in 2006, but you'd claim "look at all those shiny granite countertops."

I have a similar perception. But I'm not convinced that the artsy-looking aspect is particularly problematic. Perhaps it doesn't cost much more at all. It may not be any worse of a misallocation than vanity projects driven by executives in markets more dominated by the private sector ... and anyways, I do think the residents of the cities get value from looking at cool stuff.

Where I see more evidence of misallocations is make work projects, where they're digging up and replacing stuff that simply doesn't need to be replaced yet. Like, sidewalks in even cities like New York and London don't get repaired until damage is fairly noticeable, and the same it likely for a cracked tile in even a fairly high end hotel in those cities - but many many times I've seen such repairs go forward in unimportant locations where they wouldn't even be warranted in the central districts of world class cities.

I'm not saying that China is or will enter hyperinflation or is not growing. I'm saying that they are making investments at a suboptimal place on the growth curve.

The roads and buildings and malls are good, yes, but my argument is that the Chinese are either overspending on them or also investing in completely dubious projects, with the result that expected future economic activity per unit of investment is lower than it would be if they were investing more wisely and possibly less. If you define "quality of GDP" as "amount of future activity resulting from current activity" then they are indeed maximizing quantity over quality.

Whether that is a good or bad idea depends on various factors of which I am not fully conversant. It's not like the USA is doing very well on the quality-of-GDP score right now.

I think it is a valid question to wonder if they could invest more wisely. But I think most people are generally interested to suggest that the investments would be done more wisely if they were allocated by private markets instead of the government.

However, growth maximization is not the only objective. There is also an interest to address issues of regional inequality, with the hope (expectation?) that doing so will also contribute to growth through better integration with the broader economy.

There is also a lot of corrupt contracts handed to brothers-in-law, and a massive amount of opaque, untrustworthy accounting undergirding it all.

I think all of the following are true:
1) China is nominally growing significantly
2) Actual growth, in terms of an increased quantity of valuable and/or useful stuff, is substantially less
3) Actual growth is nevertheless positive
4) The ratio of actual growth to corrupt skimming/wastage is unsustainable
5) They are not yet at the breaking point where things will immediately change as a result of (4)
6) If they get to that point, things will become very bad very quickly
7) They are currently on pace to reach that point before reforming enough to miss it
8) Everything China does with its economy is unprecedented and difficult to accomplish. Reforming enough to miss the breaking point would be a historic achievement and is definitely possible

JB - I don't think corruption necessarily has to be such a huge macro-economy problem so long as contracts are not "excessively" padded and most of the capital stays in the country and gets directed into other production activities (or even if the money is allocated in a way that stimulates domestic demand).

I'm definitely not arguing that corruption is not bad for allocations. But it depends a lot on stuff like this. Nothing at all, for example, like the sort of corruption in Africa where political leaders may be siphoning off as much as 10-20% of GDP (guessing) and all of that capital leaves the country.

A lot of China's problems are unseen by casual observers.

When I was traveling to china a lot, I'd eat at a particular Japanese restaurant everyday, so I got to know the owners. One day I walk over, and their sign is covered up with a cheap banner.

I was puzzled and asked what the deal with that was. Turns out the city said their sign and every other sign on the street were now "illegal" and so everyone hung up cheap banners instead. (Their old signs were perfectly modest, actually.)

Now, maybe they were somehow illegal. Or more likely, someone was shaking them down, or as suggested a brother-in-laws sign business was hurting and they needed to drum up some business.

All that is malinvestment and might show up as GDP but isn't "quality" growth.

How many times would I walk around or drive past places oblivious of what was really going on? Its impossible for someone to get a handle on that kind of stuff. Nathan had another great example of unnecessary roadwork. You can't really tell that unless you use the road everyday.

Japan has spent gobs of money on mostly useless infrastructure projects for quite a while now without much success in reviving its economy. If China keeps doing the same (even assuming that relatively more projects will be worthwhile) it is not going to end well.

When I think of quality I think of more telemarketers, or the need to hire more police.

This article probably it lines the view fairly succinctly:

Correct FT link (paywall)

Do the official numbers understate capital flight or have restrictions stemmed the flood (temporarily at least)? Anbang's decision not to invest a few more billion in the U.S. was voluntary only in the sense that the possible consequences were less appealing than the alternative. Unless and until the world comes to grips with the consequences of shifting trillions in production and income to China and similar places with much higher levels of inequality and savings rates (the latter due to both the culture and little or no safety net), the world economy will languish and the potential for financial, social, and political instability will be an ever-present danger. For us and them. In the meantime, there's always theory and math to keep economists busy.

If you consider east China and west China as two different economies, then inequality is not nearly as high. This is very relevant in a country with population of 1.3 billion, 3rd largest land mass.

Unpacking GDP number link doesn't work.

It's sometimes hard to tell increased collateralization (good, esp. in poorer economies) from zombie banking (bad), athough at a glance the latter appears to dominate here. The trade numbers and reserves have, at least, stopped falling. Remains to be seen whether this is an inflection point or a pause, but as long as the reserves are stable, they can probably get through this.

Remember, if China is issuing debt to keep zombie enterprises alive, that should affect the reserves, because China has to support the peg against the natural impetus of increasing money supply. Watch the reserves.

Overall I wonder if we're missing the forest for the number of trees, though -- it's probably a lot more important that Chinese fix its issues with political repression and air pollution, at this point those are the low-hanging fruit in terms of improving living conditions for the Chinese, probably bigger than any trends in GDP. over the next ten years

"It’s probably a lot more important that Chinese fix its issues with political repression and air pollution.." [SNIP]

Forget air pollution:

This is the 1st chapter in a thirty year unfolding nightmare.

"The new gdp figure came in at 6.7%"

Of course it did.

Analyzing Chinese figures is like consulting the Oracle at Delphi.

I just read Cowen's Upshot column for today. I think he phoned it in. I once drafted an opinion for my judge and, after a sleepless night, arrived early the next morning to retrieve the copies that had been distributed to the other judges before they could read the damn thing.

When has TC been right on China? Can we all admit TC is a smart guy with a lot of talent and foresight, but he's not good at analyzing things China?

Who's good at analyzing things China?


In 2000, I predicted China would become a proto-democracy between 2008 and 2015 as computers would sweep across the country and as standards of living would obviously continue to rise. The CCP would lose monopoly power during those years.

Was I prophet, or what?

I was off a little only because the damn communists are belligerent with respect to losing power. However, I talked to a few future China experts in history and political science back then and all of them said 1) the internet has nothing to do with a change to democracy and 2) it would happen in 2050, 2100 or "never."

When it happens in 2020, I win

I bet you didn't bet that an enormous number of people in China would have computers in their pockets connected to the internet with rich interfaces in 2000! The technology revolution has been far stronger than you expected and yet no democracy.

It is possible, and I would say likely, that no one gives a hoot about democracy when they are experiencing 7% growth. When that stops, and if it stops below western levels is when democracy may happen in China.

Funny you mention that...

I was telling a scientist friend in winter a bit about my predictions. He is 60 so appreciates the Cold War. I explained that I said to a few friends in 1990 that "email will be everywhere by 2000" - that is , all over the EU and the US. A few engineers laughed and said I had no idea what I was talking about and cute that a "theory guy" (physics student then, and I didn't know computer networking/engineering at all) would think that. They said maybe 2020. I missed the internet. But, hey! email is a major part of the internet, especially in 2000 before youtube, fast pdf files and facebook .

My scientist friend tossed his iphone on the table and said "But you didn't predict this!"

Nope. I just didn't think of how cell phones would evolve since not interesting to me in 2000. I did make a huge error by thinking texting in English wouldn't be popular since too cumbersome but maybe Japanese more since easier. So, wrong there, too

I just saw tons of Chinese chatting to each other in chat rooms and on sites (I just learned the word blog that year) but that it would sweep over China and they would demand elections as almost all evaded government detection.

OK, due to CCP intransigence, I'm 4 months off so far. However, I think I'll be way closer than the Chinese experts, now historians and political scientists, etc, I talked too then who laughed at democracy in China before 2100 - one guy could see 2050 - maybe.


It is very hard to get these technology predictions right, if it were easy we would all be billionaires. E-mail seemed like a way bigger deal than the web back in the early 1990s since e mail was how we got our work done back then (and now).

As for China, again I'll say it is all about growth. No one cares about how the pie is being divided when we're all getting rich. It is when we are all fighting over scraps is when we care about democracy. At least that is my view. We'll see what happens when China's growth slows down. Although, sadly the timing may coincide with your prediction.

How is Japanese easier? You have to learn thousands of characters to be remotely literate, just like Chinese.

Alain -- it is actually sort of the opposite: few care about democracy or much else, when they're starving. Rich countries are almost by definition those in which the institutions so successfully promote trust and inclusiveness that democracy is inevitable.

I found the arguments that a rising middle class, or more especially a rising wealthy class, would raise pressures to distribute more political influence (regardless of whether this would fit with Western notions of democracy). That doesn't seem to have happened either.

The internet and smart phones have not led to democracy. But they definitely serve as a major check on the arbitrary powers of government, also help with accountability by enabling citizens to document and disseminate certain evidence of mistakes. The example that comes to mind is the Wenzhou train derailing which killed a couple hundred people - it is widely believe that no one would have learned about that if it were not for citizens taking photos of it and disseminating them online.

One thing is clear. If China becomes "democratic", it will be Chinese democracy, and will be very different from Western democracy.

I tend to consider that the CCP is not as anti-democratic as it is made out to be. Yes, it is authoritarian in some ways, and absolutely is politically repressive. But it is also sensitive to the demands of the population. In order to prevent revolutionary stirrings, it must deliver.

* I know about 2500 Japanese characters. With cellphones I could text American friends (they can read Japanese) and Japanese friends about twice as fast as in English. (you only need 1000) With Japanese, you don't have to spell out a word since most kanji words (two syllables) comes right up. I'd almost always use the two character word instead of some long Anglosized word.

* I disagree with Alain about growth. I assumed steady fast growth in 2000 with no slowdown, and it hasn't happened yet (7% is trivial). But I thought and still think the entire richer East coast will crack through the authoritarian system by 2015 because it holds them back so much.

* There isn't really "Chinese democracy" Are there checks and balances and can people vote in a national election? That's why I said a "proto-democracy" by 2015, but it hasn't happened.

I'm pretty doubtful that being able to fit more information on a page or spell with fewer letters has a significant impact at all (perhaps none) on how fast we can meaningfully absorb information. Perhaps typing is slightly faster. But if the amount of information that can be contained in a given space is relevant, wouldn't we just use smaller fonts?

Meanwhile, 26 letters are sufficient to start reading and writing. For practical purposes, even perfect mastery of a Sinic language still requires several years of very serious effort to be literate. Are the infinitessimally marginal gains (perhaps none) in typing or reading speed really worth years of effort just to be able to recognize the basis components of language? Even after learning 2500 characters, you still have to learn all those words one at a time (although it is slightly more likely that you can correctly guess the meaning of new words). It's kind of like having to learn both Greek and Latin in order to be able to guess the meaning of newly encountered words in English - and not even that's enough, you have to learn a fairly complex representation of the concept. For example, "pre" or "post". In English, you only have to learn the concept and the sound, in Chinese or Japanese, you also have to learn an entirely new way to write/read the thing.

Additional years of effort to gain occasional fractions of seconds. But phonetics don't work well in Chinese and Japanese because there are such a large number of words with the same sounds, so there is no reason for them to change much.

You need to visit Taiwan. Chinese democracy is not that different from anywhere else.

I think you misunderstood my post Nathen. First Japanese is very different compared with Chinese. Japanese is much harder to learn if a native English speaker because of the convoluted grammar, levels of politeness, etc. But everyday texting needs around 500 characters, 1500 max for serious Japanese.

By the way , it was 20 years ago I predicted that machine translation would end human translation. 2015 for French, Spanish, etc but 2020 for Japanese and Korean.

I read your post on Google translate and you forget that Indians are doing translations for much lower wages than Westerners.

Human translation is over in 2020

Well, I lived in Japan for 15 years.

I have no idea what you are talking about saying "you have to learn thousands of characters to be remotely literate, just like Chinese."

To be "remotely literate", you need about 1500 not "thousands" Highly literate Japanese know about 4,000+ characters Lit profs know maybe 6,000. But that includes alternate versions of flowers and fish. Unlike Chinese, Japanese characters almost always have two or three sounds.

Todd - Yes, I know that Indians are also doing a lot of translations.

However, I do not compete with them. They own the entire sum of new demand which is largely supported by machine translation, new demand from those who are only willing to pay half (or less) of the price of what old-school translators are getting paid. Indians have a rather particular form of English, one which is not desirable for any high-quality translation destined to a Western audience (the audience where the money/power is).

If this point is not obvious, consider that in many bids, the contractee is not even willing to consider a Canadian translator for a document destined to a British or American audience. How, then, could an Indian possibly compete to win jobs from such picky clients (and most clients worth working for are indeed very picky)?

As a matter of practice, I do not waste my time bidding on projects where the intermediary is an Indian agency. I think most good first world translators are the same (entering the market might be different though, since you may need to build a profile of happy customers, in which case you may need to work at Indian prices for a couple years).

For practical purposes, the "Indian translation market" serves as a backstop for things to bid on if you have a slow week and are bored, not a source of real competition.

Machine translation is OK to speed things up if you're translating something that has basically no new content, no new ideas, and just rehashes old ideas and rearranges them according to the needs of a current project. However, very few things which constitute high-value translations fall into this category. Consider that if there is basically nothing new in the original, then it is unlikely to be of any particular value. Five years ago I would have agreed with you, but seeing that the advances of machine translation over those five years is virtually zero in relation to most of the projects I work on (actually, the project I'm working on RIGHT NOW is somewhat of an exception, because there are a very large number of references to specific public programs and policy strategies that I'm not very familiar with, but for which there is a large existing body of translated documents), I'm very doubtful.

Besides the fact that the words in the original text have highly variable meaning depending on context, even imagining that machine translation gets very good at figuring out context, the output cannot possibly be better than the original. Often, in the process of translation, a translator can identify many sources of ambiguity in the original, and hence I market myself as a translator AND editor, and not just as a translator. The day that no one needs an editor will be the day that there is no market for high value translation. But how many writers do not need an editor? 1%? Less?

Machine translation is causing the market to grow, not decline. Among other things, many enter the market attracted by "Indian prices", get a few low grade translations, then bite the bullet and shell out the real money having thought through all the marketing and exposure possibilities that can only backfire with a low quality translation.

OK, 1500 characters in Chinese is often considered to be roughly sufficient to get the main meaning of most articles in the newspaper, and can be seen as a fairly low threshold for literacy. A fair few characters in Chinese also have several sounds depending on the contextual meaning. Most university graduates know about 3000 characters, and perhaps 100 or so are largely reserved for common family names.

Nathan, Japanese is not Chinese.

With 1200 Japanese characters, one can read 95% of a Japanese newspaper if of course one also knows the vocabulary and grammar. At 2000 characters, over 99% of a Japanese newspaper article can be understood. Magazine articles require closer to 2500+ kanji for 99%+ comprehension.


Interesting. In Japanese/English, machine translation advances since 2010 have been *yuge*!

My 1996 prediction when I finally saw the implications of the internet after using it for six months was that by 2006, computers will be a thousand times more powerful and the internet content will balloon. This would allow for very good statistical machine translation for French, Spanish, and German into English and jobs will start to disappear in those language pairs by 2009. For Japanese (my language) and Korean, it will be pushed back five years because the grammar is much different and subjects are often omitted.

In 2002, I finally told friend who asked about that Japanese prediction to be concrete. I said "I can't see it happening before 2008 but will be surprised if machine translation doesn't wipe out most translation jobs by 2015."

Once again, I'm a bit off. What I really got wrong was the impact of indian translators who will work for $10/hour when most Japanese/English translators make $40 to $50/hour.

A German translator overheard my explanation to a friend in that small computer center and said I was a (naughty word)ing idiot. "Computers will never think." I said that wasn't the point and tried to explain what I thought statistical translatioion,n would look like.

But almost no one in the translation world has a science/stats background. They think translation is some kind of magical power.

Google Translate lets Indians (and others) who aren't great at Japanese get over hurdles that they would not have been able to get over before 2010 or so. It is pushing wages down for everyone. Japanese/English translators make about 30% less than in 2009 and the slide ain't stopping as Google Translate keeps improving. It is funny that many say it hasn't affected them except they forget about inflation.

Human translation will be mostly over by 2020. My 2015 prediction was a wee bit off, but like democracy in China, translators still think 2050 , 2100 or never.

No, the machine translation titanium wall goes up in 2020.

Actually, most of the translation I do is basically in economics and statistics. Considering that nearly every methodology is different, a computer is virtually guaranteed to screw up lots of stuff. It'll get stuff like "the confidence is interval is ..." done correctly, say, but not only is every context different, but every author has their own writing style. Since we all have our own idiosyncracies in writing, computers will still be making mistakes that translators wouldn't even when it is trusted for most purposes.

Maybe if you've got a routine monthly report. But you don't need AI for that, you can just do Word templates with macros to populate all the data into the respective language versions.

Even for stuff you'd think that computer should excel at already, like corporate financial sheets, well, every company has its own way of doing things, their own lingo, etc.

A lot of the big money (no one's getting rich, but size of the industry anyways) is in finance, banking, law, and engineering-related stuff. If you get it wrong, someone can get sued. I'd give it 50 years before lawyers or bankers would trust machine translation for anything but the most extremely routine of processes.

What American has ever been good at analyzing China?

I can think of no one.

Which Chinese have been good at analyzing China? I can think of no one.

Among other things, my understanding (a common one?) is that data sources (especially data collection methods) are poorly integrated/standardized, and lower level officials commonly fudge numbers to according to whatever current targets are.

A previous problem was targets partially driven by macro stability concerns, where local officials shouldn't shoot "too high", but in many areas they did indeed shoot "too high" for local development purposes, not caring about the effects on the macroeconomy, and systematically UNDER-reported growth to be closer to centrally determined objectives.

And so even Chinese authorities often refer to things like electricity consumption or freight movement as a proxy.

If their economy is chugging along at 7% growth, and our is limping along still, why is the trade deficit growing? grew by a billion dollars in Feb. How is that consistent with the paths of the two economies?

What do the two have to do with each other? Expand on your logic here.

So, China doing what China always does: invest in dubious "infrastructure" projects to prop up inefficient SOEs.

Their numbers, on the other hand, are about as trustworthy as Enron's.

So what's new here? At least the only interesting and true words in this post are that...China can get any GDP number it wants. Hence, the numbers mean little since it's all SOEs digging up and filling back in trenches.

"“The Fed’s reversal has taken a lot of pressure off the renminbi and without the currency looking like it’s going to collapse, people are feeling better"

Who would those people be? Maybe international currency traders dinking around with spreads? How many of those parasites are there anyway? No normal, ordinary person cares about putting pressure on the renminbi or relieving it. They just want a stable currency that maintains its value vis a vis eggs, cell phone contracts and air line fares. The idea that members of some pseudo-profession like currency traders can wreak havoc on the lives of the proles means that there should be a bounty on those guys.

Comments for this post are closed