population growth since 1955

I recently had the misfortune to stumble across population data at the township level from 1955.  Since I have been stuck with a mild case of writer’s block for my real research, that discovery was more than enough of a distraction to take me happily away from trying to bang out yet another unsatisfying draft of that damn project.  Ah, procrastination!

We all know the broad outlines of population growth in Taiwan in the postwar period.  The population grew at a very high rate in the 1950s through 1970s.  It slowed down in the 1980s and 1990s and is barely growing at all these days.  This was accompanied by two broad internal migration patterns.  On the one hand, people moved from rural areas to urban areas.  On the other hand, the population shifted from the south to the north.  There is nothing new and surprising about these patterns, but, at least for me, it is jarring to see them in township level data.

Let me give an example.  In Dec 2013, Chiayi County had 529,229 people, and New Taipei City had 3,954,929, or nearly eight times as many people.  In 1955, Chiayi County had 618,098, while Taipei County only had 638,091.  It’s jarring to think that Chiayi was once just as populous as Taipei County.  Of course, we are cheating a little.  In 1955, Chiayi County included Chiayi City, while Taipei County covered Nangang 南港鎮, Neihu 內湖鄉, Mucha 木柵鄉, and Chingmei 景美鎮 Townships.  Lest you think that adding these areas of modern Taipei City puts Taipei County at an unfair advantage, keep in mind that Chiayi City had far more residents than those four townships put together (148,552 to 65,177).  Or think of this.  In 1955, Lioujiao Township 六腳鄉 had more people than Jhonghe Township 中和鄉 (39,672 to 37,875).  These days, Lioujiao has shrunk to only 24,947 residents.  In contrast, Jhonghe has exploded to 415,742.  Even that doesn’t tell the full story, since the present-day Yonghe District 永和區 (pop. 229,062) was part of 1955 Jhonghe.  So where the two areas were roughly equal in 1955, these days, Jhonghe/Yonghe is about 25 times bigger.

To give a broader picture, I reclassified the 1955 data according to today’s boundaries.  I also found similar data from 1981, which is roughly halfway between 1955 and 2013.  Since a table with over 350 townships would be large, I’ve aggregated them into (today’s) counties and cities.

	1955	1981	2013	81/55	13/81	13/55
Total	9098643	18193955	23373517	2.00	1.28	2.57
.
New Taipei City	572914	2354858	3954929	4.11	1.68	6.90
Taipei City	852670	2270983	2686516	2.66	1.18	3.15
Taichung City	772963	1651296	2701661	2.14	1.64	3.50
Tainan City	969881	1564360	1883208	1.61	1.20	1.94
Kaohsiung City	877264	2245517	2779877	2.56	1.24	3.17
Ilan County	299455	445472	458456	1.49	1.03	1.53
Taoyuan County	426522	1093621	2044023	2.56	1.87	4.79
Hsinchu County	252855	363408	530486	1.44	1.46	2.10
Miaoli County	386693	545608	565554	1.41	1.04	1.46
Changhua County	787137	1180612	1296013	1.50	1.10	1.65
Nantou County	349208	527538	517222	1.51	0.98	1.48
Yunlin County	593923	796968	707792	1.34	0.89	1.19
Chiayi County	469546	574451	529229	1.22	0.92	1.13
Pingtung County	552702	892107	852286	1.61	0.96	1.54
Taitung County	159631	281100	224821	1.76	0.80	1.41
Hualian County	210336	357530	333897	1.70	0.93	1.59
Penghu County	84502	105674	100400	1.25	0.95	1.19
Keelung City	187468	347828	374914	1.86	1.08	2.00
Hsinchu City	144421	284737	428483	1.97	1.50	2.97
Chiayi City	148552	251840	270872	1.70	1.08	1.82
Jinmen County		50248	120713		2.40
Lianjiang County		8199	12165		1.48

The overall population doubled in the first 26 years, so no cities or counties had negative growth (though some townships did).  However, Taipei County absorbed far more of the population growth than anywhere else, quadrupling in size.  In the recent 32 years overall population growth has been much slower, but the urbanization and northward migration patterns have continued.  This has meant negative population growth for most of the agricultural heartland.  The above table shows Nantou, Yunlin, and Chiayi to have negative growth.  It’s actually clearer than that.  Changhua and Tainan have positive growth, but it is all in the urban areas.  If you look at the old Tainan County, once you take out Yongkang City 永康市 (which is really an extension of the metro area), the rest of the county had negative growth.  Similarly, the southern two-thirds of Changhua County also lost population.  So basically all of south-central Taiwan lost population between 1981 and 2013.  Meanwhile, Taoyuan became the fastest growing area, followed by New Taipei and Taichung.  (I don’t know what to make of the Jinmen and Lianjiang numbers.  You can easily spin a story about the new tourism economies, but I’m not sure how much I trust the 1981 numbers.)

Since I do elections, let’s imagine how these numbers would translate into political power.  Imagine that today’s legislative electoral system of 73 district seats were imposed in 1955 and 1981.  (Yes, I realize that would change history slightly more than merely imposing 2013 administrative lines on 1955, but save the alternate history for another day.)  Here is how those 73 seats would break down. (Remember, everyone gets a minimum of one seat.)

	1955	1981	2013
New Taipei City	4	9	12
Taipei City	7	9	8
Taichung City	6	7	8
Tainan City	8	6	6
Kaohsiung City	7	9	9
Ilan County	2	2	1
Taoyuan County	3	4	6
Hsinchu County	2	1	2
Miaoli County	3	2	2
Changhua County	6	5	4
Nantou County	3	2	1
Yunlin County	5	3	2
Chiayi County	4	2	1
Pingtung County	4	4	3
Taitung County	1	1	1
Hualian County	2	1	1
Penghu County	1	1	1
Keelung City	1	1	1
Hsinchu City	1	1	1
Chiayi City	1	1	1
Jinmen County	1	1	1
Lianjiang County	1	1	1

In a nutshell, compare the heartland counties (Changhua, Nantou, Yunlin, Chiayi, and Tainan) to the emergent areas (New Taipei, Taoyuan).  In 1955, the former would have had 27 seats to the latter’s 7.  In 2013, the heartland areas were reduced to 15 while the emergent areas have 18.  Take a couple of minutes to digest that.  Keep in mind that in 2012 the DPP won 56% of the seats in the heartland areas but only 11% in the emergent areas.

A few sharp readers may have noted that the 2013 seats are a bit wrong.  Nantou, for example, actually has 2 seats.  This is true, but the current apportionment was done in 2006, and Nantou barely had enough population to secure its second seat.  Seven years later, Nantou’s population has declined.  That second seat is gone, and it isn’t coming back.  Fortunately for Nantou, because of the corrupt bargain that said that redistricting would only take place once every ten years, Nantou will keep its second seat until the 2020 election.  By Dec 2018, when the new seats are apportioned, Nantou, Chiayi County, and Pingtung will all have lost a seat.  Kaohsiung is also in danger of losing a seat.  As of today, the winners are Tainan and Hsinchu County.  However, since Taoyuan, New Taipei, and Taichung are growing so fast, one of them will win another seat and Hsinchu might lose its second seat to another of them.

[Esoteric note.  The above apportionment is done incorrectly by current rules.  Under current rules, aboriginal population should be subtracted so that aborigines are not double counted.  I did not do this because (a) I didn’t have aboriginal population data for 1955 and 1981 and I wanted to be consistent, and (b) this is a fairly new rule that I don’t think was in effect in 2006.  Under the current rules, Chiayi County would still have its second seat and Pingtung would have already lost a seat.  Regardless, by 2018 both Chiayi and Pingtung will be far below the threshold for the last seat.]

This gets at the northward migration, but I still don’t think I have adequately encapsulated the urbanization.  Let’s think about population density.  Of course, Taiwan is one of the most densely populated countries in the world.  Outside of city states such as Singapore, Taiwan trails only Bangladesh.  In 2013, Taiwan had 646 people per square kilometer.  In 1955 and 1981, it was 251 and 503.  Those numbers woefully underestimates how people actually live, since about three-fourths of the island is sparsely populated deep mountains.  Instead of looking at the aggregated national averages, consider township population densities.  If you sort the townships by population density, you can see what kind of environment the median person lived in.  So let’s look at the 25^th, 50^th, and 75^th percentiles for each year.

	1955	1981	2013
25th	名間鄉	布袋鎮	田中鎮
25th	Mingjian, Nantou	Budai, Chiayi	Tianjhong, Changhua
25th	374	678	1245
.
50th	埤頭鄉	永靖鄉	豐原區
50th	Bitou, Changhua	Yongjing, Changhua	Fengyuan, Taichung
50th	632	1797	4032
.
75th	中和鄉	新莊市	桃園市
75th	Jhonghe, New Taipei	Sinjhuang, New Taipei	Taoyuan, Taoyuan
75th	1464	9680	11935

In 1955, the median person lived in a fairly rural place (by Taiwan standards).  Bitou was a classic farming community.  Even at the 75^th percentile, Jhonghe (as we have seen above) was hardly an urban place in 1955.  By 1981 Taiwan looked quite a bit different.  The 1981 25^th percentile township, Budai, was roughly as dense as the 1955 50^th percentile township.  The median person was still a farming community in southern Changhua, but now in a far denser community.  The biggest change was in the 75^th percentile.  At close to 10,000 people per square kilometer, Sihjhuang was clearly urban.  In 2013, the top quarter had not become much more densely packed.  In fact, many of the central city areas actually lost population during this period.  Taoyuan City in 2013 is not that different from Sihjhuang City in 1981.  The big change is in the next quartile.  By 2013, the median person lived in Fengyuan District, Taichung City.  Fengyuan is a quite urbanized area and is qualitatively different from the 1981 median township.  In general, between 1981 and 2013, the urbanization really occurred in this second quartile of the population.  Instead of packing ever more densely into the old city centers, people moved into new urban townships.  This led to dense cities surrounded by less dense but still urban peripheral areas.

Or we can cut the data in a slightly different way.  The following puts the population into five categories according to the population density of the township.  I think of the first categories as basically rural, the second as a small town with lots of rural characteristics, the middle category as a small city, the fourth as urban but not quite saturated, and the fifth category as utterly urban by any standards in the world.

	1955	1981	2013
	9,098,643	18,193,955	23,373,517
.
under 500	3,697,056	2,703,037	2,585,323
500-1000	2,385,844	3,776,873	2,673,947
1000-3000	1,588,445	4,493,555	4,989,507
3000-10000	651,092	2,938,115	6,783,986
over 10000	776,206	4,282,375	6,340,754
.
under 500	0.406	0.149	0.111
500-1000	0.262	0.208	0.114
1000-3000	0.175	0.247	0.213
3000-10000	0.072	0.161	0.290
over 10000	0.085	0.235	0.271

Think about these numbers in the context of democratization.  Democratization movements tend to be based in the cities, and this was also Taiwan’s experience.  When Lei Chen 雷震 attempted to build an opposition party in the late 1950s, he and his allies were not fighting on very fertile soil.  (It didn’t help that Hu Shih 胡適, the greatest voice for liberal democracy that the world has ever seen, paragon of marital virtue, and our Academia Sinica local deity, got cold feet and sold them out.)  By the time the next generation of activists tried in the late 1970s, Taiwan was a much more urban society.  Not coincidentally, they found much more support and repression was much more difficult.  Of course this sort of analysis is grossly oversimplifying things, but the degree of urbanization probably mattered a bit.

The last way I want to look at population trends is to look at the growth of individual metro areas.  Sometimes the political divisions are ridiculous.  For example, the official data say that Taipei has 2.6 million residents.  Balderdash.  The administrative lines do not reflect the actual urban area.  Taipei metro area is much, much larger.  Sometimes this goes the other way.  Drive down the west coast from Budai Township 布袋鎮 in (rural) Chiayi County to Beimen District 北門區 in (Direct Municipality) Tainan City.  I dare you to claim that you have suddenly made a transition from extremely rural to extremely urban.  Unless you see a sign, you probably won’t know you have entered a city at all.  So I like to reclassify metro areas subjectively by which districts I think are really part of the core city.  The most important thing I look at is population density.  However, I also consider geography (Does it have lots of mountains?) and whether you have to travel through a non-urban area to get to another urban area.  And because there are different levels of urbanization within a city, I subdivide the metro area into core and peripheral urban areas.  In the current Taipei metro area for example, I consider Beitou 北投 to be a part of the core city, while Danshuei 淡水 is peripheral.  Again, this is subjective, and there is a bit of guessing involved for 1955 and 1981.

	1955	1981	2013
Total pop.	9098643	18135508	23240639
Metros	2022378	8313628	15574223
Metro %	0.222	0.458	0.670
.
Taipei	853013	3932376	6212153
Core	652002	3723211	5186256
Periphery	201011	209165	1025897
.
Kaohsiung	373394	1454764	2383943
Core	193279	1281226	1873940
Periphery	180115	173538	510003
.
Taichung	170851	890043	2640641
Core	170851	392727	1658427
Periphery	0	497316	982214
.
Tainan	229354	676593	1077341
Core	99191	373902	819485
Periphery	130163	302691	257856
.
Hsinchu	122137	243218	867536
Core	122137	243218	352619
Periphery	0	0	514917
.
Keelung	125077	265290	374914
Core	52464	135531	283013
Periphery	72613	129759	91901
.
Chiayi	148552	251840	270872
Core	148552	251840	270872
Periphery	0	0	0
.
Taoyuan	0	599504	1746823
Core	0	185257	1184980
Periphery	0	414247	561843

The top line number is that in 1955, less than a quarter of the population lived in a metro area while two thirds did in 2013.  Instead of 2.6 million, Taipei is redefined as 6.2 million.  It could be more if you were inclined to consider Keelung as part of Taipei.  I thought about that for a long time before deciding that Keelung is still a separate place.  It may not be for much longer.

According to my classification, Taichung, not Kaohsiung, is actually Taiwan’s second largest city.  Kaoshiung has a slightly larger core, but Taichung has a substantially larger periphery.  Even if you dispute the exact boundaries, the big point is that the common idea of Kaohsiung as clearly being Taiwan’s second largest city is wrong.  It used to be, but Taichung has caught up.  In 1955 and 1981, Tainan and Taichung jockeyed for the title of the third largest city.  Tainan lost that race and is no longer even number four.  Taoyuan has shot by and Hsinchu has almost caught up.

Let’s stop for a minute and consider Hsinchu.  It’s easy to miss Hsinchu’s growth because the Hsinchu metro area is split between Hsinchu City, Hsinchu County, and Miaoli County.  Also, note that Hsinchu has grown much faster after 1981.  The obvious explanation is the Hsinchu Science Park, which started operations in the early 1980s.  Think for a minute about the early section on legislative districts.  Did it strike you as odd that Hsinchu County was slated to gain a seat?  In fact, Hsinchu County has already passed Nantou and Chiayi Counties in population.  Historically, Nantou, Chiayi, and Miaoli were similarly sized, but unlike the other two, Miaoli has not experienced negative population growth.  Well, actually it has in most of the county.  Most of the county has seen the same sorts of negative growth that we see in the heartland areas.  However, the two townships adjacent to Hsinchu City, Jhunan 竹南鎮 and Toufen 頭份鎮, have added 60073 people since 1981, more than enough to give Miaoli overall positive growth.  When reapportionment arrives in 2018, Miaoli, unlike Nantou and Chiayi, will keep its second seat, and this will be due to the growth of the Hsinchu metro area.  The growth of Hsinchu is all the more striking when contrasted with the other two smaller cities, Keelung and Chiayi.  Both have grown, but Chiayi hasn’t even kept pace with overall population growth.  Keelung is perhaps even worse, given its proximity to Taipei.

Finally, there is Taoyuan.  In 1955, I really can’t identify a core city.  If I had been making this list in 1955, I would have chosen Ilan/Luodong 宜蘭市/羅東鎮, Changhua 彰化市, Pingtung 屏東市, and Hualian 花蓮市 before Taoyuan/Jhongli 桃園鎮/中壢鎮 as my next metro area.  The two weren’t all that densely populated and the Hakka/Minnan division would have made them much more distinct from each other than it does today.  Even today, there still isn’t a central urban area in Taoyuan in the same sense that there is in all the other metros.  Taoyuan is, however, a gigantic sprawl of fairly densely populated territory.  It adds up to a very large urbanized area, even if it doesn’t seem to have any focal point.  For 2013, I put Jhongli 中壢市, Pingjhen 平鎮市, and Bade 八德市 in the core.  It might be more accurate to put them in the periphery and leave the core with only the 415,414 people in Taoyuan City 桃園市.  If you’ve been to the Los Angeles area, Taoyuan metro is something like the Inland Empire.  It’s gritty, industrial, sprawling, and far removed from the “desirable” areas.  It’s also shockingly large; when considered on its own it is now one of the larger metro areas in the country.