A passing reference from the encyclopedic Pseudoerasmus sent me to the work of Paul Caruana-Galizia, who has done an impressive series of papers quantifying historical income levels on a regional (rather than the usual national) basis for many countries.

His paper on China, written with Ye Ma, gives us provincial per-capita GDP for the late 19th and early 20th centuries — an essential resource for comparing patterns of development over a longer time period.

The paper inexplicably does not include a map, so I have made a couple myself:

(Note: In these maps, Hebei includes the present-day municipalities of Beijing and Tianjin, Jiangsu includes Shanghai, and Sichuan includes Chongqing.)

The decline of incomes across most provinces (with Jiangsu and Xinjiang notable exceptions) is visually very obvious. Regional inequality was also basically stable over this period, probably reflecting the absence of sustained growth. Here is some useful commentary from the paper for context:

How do these per capita income levels compare with Europe’s? In 1873, Heilongjiang was China’s richest province with a GDP per capita of $870. In 1870 Europe, there were eight regions out of a sample of 200 that had per capita incomes between $800 and $950. Of these eight, five were in Austria-Hungary, one in France (Haut-Garonne), one in Germany (East Prussia), and one in Spain (Extremadura). These comparisons add weight to the argument by Pomeranz that some European areas were at similar levels of development to Asian ones. The fact remains, however, that China’s richest province – some 43 per cent above the Chinese average in 1873 – was only as rich as Europe’s poorest regions. … By the close of the period, Jiangsu surpassed Heilongjiang as the richest Chinese province with an income of $853. The only comparable European region is Dalmatia ($874).

On a macro level, the results corroborate histories of national economic decline and stagnation. China’s mean provincial per capita income compound annual growth rate over the period was −0.20 per cent. The internal disorder and continuous shocks to the economy in the forms of rebellions, foreign wars, and natural disasters meant that most Chinese were better off at the end of the nineteenth century than they were at the start of the twentieth.

Ma and de Jong’s numbers show that Chinese real GDP per capita dropped by 10 per cent from 1873 to 1912. Looking to Europe for comparisons, we see that Italian regions enjoyed an annual growth in per capita income of some 1.2 per cent, similar to that in France. These figures fit with Gernet’s claim that China’s ‘tragic period’ coincided with Europe’s ‘acceleration.’

And here’s the original table for reference:

I’ve been interested in industrial clusters in China for a while, since I think they tell us a lot about underlying patterns of private-sector economic activity. Clusters are behind much of China’s decades-long success in exports, and more recently seem to be related to some fast-growing domestic service sectors as well.

A recent working paper by Xiwei Zhu et al., “Entrepreneurship and Industrial Clusters: Evidence from the China Industrial Census“,  makes some interesting points on this topic. The authors do some fancy math to identify clusters from data on the location of industrial firms, which results in the nice map below.

The results are broadly consistent with more anecdotal approaches to identifying: industrial clusters are most prevalent in the coastal provinces of Zhejiang, Jiangsu and Guangdong (but note that Sichuan far inland also does pretty well). Places with lots of clusters also tend to be places where the private sector is a larger part of the economy.

Why do clusters form in these places? Geography is part of the traditional explanation, and the authors do find that access to ports (i.e., access to world markets) contributes to the formation of clusters. But they also argue that what they call the historical “supply of entrepreneurs” is an important factor.

Since there were few recognized private companies in China before the 1990s, most founders of private-sector firms had to come from somewhere else–and state-owned or collective enterprises were a major source.

The authors use the number of firms in 1985 as an indicator of this historical potential for entrepreneurship. And they find that the number of all firms in 1985 is closely related with the number of private-sector firms in 2004:

The pattern suggests that the places where private-sector businesses flourished after liberalization were in fact those places where disguised private-sector businesses were already most prevalent. This fits in with historical evidence from the 1970s that pre-Communist commercial traditions and patterns often continued in the form of collective enterprises.

Though this particular paper is heavy on data and light on historical interpretation, I think it does contribute to a different narrative about China’s economic development. In such a narrative, China’s growth resurgence, at least in the 1980s and 1990s, is more about the flourishing of long-suppressed indigenous entrepreneurial traditions than the success of top-down development programs.

This week the last of the provinces reported their GDP data for 2015, so now the full set is available–which means more maps! I’m going to focus on provincial nominal GDP growth here, as the changes are more dramatic, and more interesting, than in real GDP growth. Here’s an example: in 2014, three provinces reported real GDP growth below 6% (Shanxi, Liaoning and Heilongjiang), and three provinces reported real GDP growth above 10.5% (Guizhou, Chongqing and Tibet). In 2015, the same three lagging provinces were the only ones to report real GDP growth below 6%, and the same three leaders were the only ones to report real GDP growth above 10.5%. So, not much change, right?

Not at all. Nominal GDP growth tells a different story, and a more intuitive one: there was a broader and deeper slowdown last year. In 2014, exactly one province had nominal GDP growth below 2%: the coal capital of Shanxi, with 1.3% growth. But in 2015, seven provinces reported nominal GDP growth below 2%: Shanxi again at 0.3%, but also Gansu, Liaoning, Heilongjiang, Xinjiang, Hebei and Inner Mongolia. In Gansu, nominal growth was actually negative, with its GDP declining 0.7%. Similarly, in 2014 there were 10 provinces that reported nominal GDP growth above 10.5%, but in 2015 that number shrank to three. The three remaining fast-growth provinces, again Guizhou, Chongqing and Tibet, have economies that are heavily dependent on government subsidies.

The regional pattern of the slowdown is pretty clear: it is deepest in the heavy-industry-dependent provinces of the north and far west. This fact, and the fact that the slowdown is so much deeper in nominal terms than in real terms, to me reinforces the point that China’s slowdown has its origins in the housing downturn and the knock-on effects in commodity prices.

How far back do the current economic problems of some Chinese provinces go? To a large extent the current downturn looks mainly resource-related, a result of the intense slowdown in housing construction. But it is also hard to avoid noticing that many of the places that are doing poorly have historical legacies that could be, shall we say, problematic. The three northeastern provinces, which were the cradle of state-owned industry in the 1950s and have been receiving heavy government support for decades, have been among the worst performing of all. Here are some interesting comments from Li Pumin of the National Development and Reform Commission, at a press conference on the Northeast in August 2015, which convey some of the flavor of the political discussion about the Northeast (my translation):

The General Secretary [Xi Jinping] in July and the Premier [Li Keqiang] in April each made inspection tours of the Northeast, and also held forums. This shows that Party Central and the State Council attach great importance to the revitalization of the Northeast. Why is it so important? …The Northeast was the first part of our country to be liberated, and it also has the best industrial base. During the First Five-Year Plan period when the Soviet Union helped build 156 projects, 58 of them were in the Northeast, or one-third of the total. This formed a large group of backbone enterprises in resources, energy, equipment manufacturing and national defense. As the “first born” of the People’s Republic, the Northeast made great contributions to building our country’s industrial system and the whole national economy. The comrades here today are perhaps rather young, but if you ask the older generation [they will tell you that] in the 1960s and 1970s, the Northeast produced a lot of equipment, technical information and personnel that supported the development of the mainland. …Also at that time the Northeast made other contributions, producing much oil, grain and timber to support China’s economic construction in the early days after Liberation. It should be said that as the “eldest son” of the People’s Republic, the Northeast has made great contributions, so the People’s Republic will not forget this during the current difficulties. This is the historical perspective.

Li’s mention of the Soviet projects in the 1950s piqued my interest–some possible data to work with! It was in fact not too hard to dig up a complete list of the projects, and from that make a few observations (there are a few lists on the internet, but I also confirmed these with printed sources). Indeed, the Northeast (aka Manchuria) received the lion’s share of the Soviet projects. These were the flagship projects of the early days of the planned economy, as China worked hand-in-hand with Soviet experts to replicate their model–which at the time, in the afterglow of the Allied victory in WWII, was seen as quite successful.

The bias toward the Northeast was perhaps even more dramatic than Li’s comments indicate. Many defense-related projects were built in two inland provinces, Shanxi and Shaanxi (Xi’an today is still one of the centers of China’s state-run aerospace industry). But civilian projects, mostly in heavy industry, were focused on the Northeast, which got 50 out of 106 projects (see the table below). The rest were mostly scattered around northern and inland China. What is remarkable about the map below is that not a single province on the southern or eastern coastal province was chosen as the site of a Soviet project in the 1950s. This is a clear sign how Mao and China’s early planners wanted to narrow the gap between the coast and the inland by channeling resources to the interior (keeping the defense industry far from potential attack was also part of the strategy.)

The northern bias of the early planned economy does seems to explain something about China today–anyone who has traveled around the country can attest to the obvious differences between north and south. Try doing a simple “billboard test”: the billboards beside airport highways in the north advertise giant state-owned enterprises; down in Guangdong and Zhejiang, they advertise private firms and trade fairs. The flavor of urban life is also very different, with more small businesses, more variety in domestic brands and more vibrant shopping streets in the south.

On the other hand, I haven’t been able to come up with a very strong quantitative relationship. There is a positive but pretty weak correlation between the number of Soviet projects a province had in the 1950s and the size of its state sector today, really nothing to write home about. As my index of the influence of state-owned enterprises shows, the provinces where the state sector dominates most are the giant cities of Beijing and Shanghai, and the more remote western provinces that depend mainly on government projects. While the regional policy of the 1950s focused on channeling resources to the north and the northeast, in more recent decades the west has been the favored recipient–and China in the 1990s had a lot more resources to work with than it did in the 1950s. The Soviet-style planning of the early 1950s laid down one regional pattern, but the Soviet model fell out of favor in the 1960s, and subsequent decades had different priorities. So today’s regional pattern is really an overlay of different influences at different times.

So going back to the question I asked at the beginning: do the roots of today’s problems go all the way back to the 1950s? On the data I have now, it would be rash to say that the Northeastern provinces were condemned to have a recession in the 2010s because the Soviets built a lot of factories there in the 1950s. On the other hand, I cannot say that these two facts are completely unrelated. I’m still trying to tease out exactly how they are connected.

 

Appendix. Here’s the complete regional breakdown of the Soviet assistance projects, most of which took place during 1952-60. Manchuria is in bold.

Province	No. of civilian projects	No. of defense projects	Total no. of projects
Liaoning	20	4	24
Heilongjiang	20	2	22
Jilin	10		10
Henan	9	1	10
Shaanxi	7	16	23
Shanxi	7	7	14
Gansu	7	1	8
Hebei	5		5
Yunnan	4		4
Inner Mongolia	3	2	5
Hubei	3	1	4
Jiangxi	3	1	4
Hunan	3		3
Sichuan	2	4	6
Beijing	1	4	5
Anhui	1		1
Xinjiang	1		1
Grand Total	106	43	149

(Note to nitpickers: yes, my table only shows 149 projects, rather than the 156 mentioned above. This is because in 1983 the official list of actually completed Soviet projects was revised to 150 from 156; one of those was a defense-related project that, for unexplained reasons, was counted as two projects for statistical purposes, and I didn’t bother to adjust for this.)

As you may have noticed I have a bit of a thing for maps. I often go out of my way to find a way to present statistics in map form, because it’s fun, and also because I think it’s important to understand regional variation in China (here’s one recent attempt). But that kind of map where different areas are colored in to convey information, known as choropleths to infographics aficionados, has some drawbacks when applied to Chinese provinces. Namely, the size of Chinese provinces varies hugely. A standard map is visually dominated by Xinjiang and Tibet, which happen to have few people and small economies, while the economic powerhouses of Beijing and Shanghai are barely visible.

This is in fact an extreme version of the problem faced when mapping the US, which also has some big empty states and small populous ones. A recent solution adopted by many infographics mavens is to make all the states the same size; the resulting “grid” maps, usually made up of squares, have been the talk of the infrographics community. Recently the folks over at NPR came up with an even cooler solution, which is to use hexagons rather than squares. This allows the resulting layout of tiles to more closely approximate geographic reality and the well-known shape of the US. I quite like it:

 

Naturally, this presented a challenge: can this same technique be applied to China? You betcha. After some playing around I came up with the following solution, which preserves both the chicken-shaped outline of China, and is reasonably accurate about the positions of individual provinces relative to each other. The biggest compromise with reality I made was to pull off the coastal municipalities (Beijing, Tianjin, Shanghai) from the main part of the map. This makes it easier to get the relative positions of the remaining provinces correct.

I’m pretty pleased with this (though suggestions for improvement are naturally welcome). Now for the test: does the hexagon tile map work better at presenting information visually? Here’s a map I produced earlier this year, which highlights the regional variation in economic growth across China.

And here’s the same statistical information presented in a hex tile grid map.

The differences are interesting. The traditional map presentation works quite nice visually, because the big swath of the slow-growing black and red provinces across the north half of the country conveys a daunting impression. The hex map, as intended, lessens the visual impact of the large but less-important western provinces, and makes it clearer what is happening in the municipalities. It also makes it easier to see how many provinces are growing slowly versus how many are growing more quickly (and has the added benefit of being able to fit in province names, which is hard to do legibly on a regular map). It’s a bit less scary as a result, since you can see that there are still more fast-growing ones than slow-growing ones. So it seems like the hex map is best for illustrating data where the important point is the number of provinces meeting various criteria. I’m not sure for this particular example I would prefer the hex map, but it was certainly fun to explore.

One of my recent side interests has been getting an understanding of how China’s population flows in recent decades relate to patterns further back in history. Mass migration has been a feature of Chinese history long before recent years’ headlines about migrant workers and urbanization. One example is the huge movement of people into Manchuria in the late 19th and early 20th century, which was in terms of the absolute number of people comparable to movement of people into the western United States from 1880-1950, and larger than the emigration from Ireland in the 19th century (according to statistics from Thomas Gottschang’s 1987 article; JSTOR link). After the founding of the People’s Republic in 1949, central planning and political campaigns also had big effects on the movement of people. Examples here are the heavy-industrialization drive of the 1950s, or the “sent-down youth” phenomenon of the 1960s, when millions of students were shipped from urban centers to more isolated provinces.

To try to give myself a way to better understand and visualize all this history, I went through a fairly simple exercise. I put together a spreadsheet of population by province going back to the 1950s from various Chinese statistical yearbooks. Then I looked at which provinces had a rising or falling share of population over time. If the natural rate of population growth does not vary too much from province to province, then a rising or falling share of national population should be a result of net in- or out-migration. The maps below are the result; they depict three big eras in Chinese population flows over the past six decades. The results are obviously sensitive to the periods chosen, which probably could be refined. I also ended up treating the entire reform era (post-1980) as one period as differences among the three decades were not great. (The arrows are just to help clarify the trend, and do not actually indicate the source and destination of migration flows–the data do not permit such precision.)

Hopefully the maps should mostly speak for themselves, but the shift in the direction of population flows over time is quite striking. The first years of Communist rule led to big flows of people into old and new industrial centers in north China and the Pearl and Yangtze River deltas. Some of those movements were then reversed in the chaos of the 1960s, which saw the coastal provinces lose people to a belt of inland provinces. With the reform era the influence of political campaigns declined, and the market forces helped draw people to growing urban centers where higher wages could be earned. The formation of big urban concentrations around Beijing, Shanghai and Guangzhou/Shenzhen is clearly visible in the map. Some regional patterns persist throughout the three periods, notably the net migration into Xinjiang and other western provinces, as well as the island of Hainan, which were places that were relatively lightly populated. I’m sure there is more to learn here, but I have to say these initial results are quite pleasing.