



































Microsoft Word - 2 proof


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

29 | P a g e  

THE COMPLEX INTERPLAY OF INCOME INEQUALITY AND TAX 
POLICY IN TAIWAN 

 
 

1Dr. Yen Chong and 2Professor Yu Chong 
1Associate Professor, Department of Economics, Chinese Culture University. 

2Professor, Department of Economics, Chinese Culture University. 
 

Abstract: This study delves into the complex relationship between democracy, taxation, and income 
redistribution, with a focus on blue-collar workers' political preferences. On the surface, one might 
expect that individuals with lower incomes would advocate for higher taxes and greater income 
redistribution, but this hypothesis doesn't align with the reality in many democracies where blue-
collar workers often support right-wing parties advocating limited redistribution. Existing literature 
provides diverse and inconclusive findings regarding the impact of democracy on tax policies and 
inequality. 
To address these discrepancies, this research employs a simplified case study: Taiwan's 
democratization. Taiwan's unique political landscape, defined by its relations with China and the 
sharp ideological divide between the KMT and DPP parties, makes it an ideal setting to explore the 
interplay between democratic elections and tax redistribution. Unlike many democracies, Taiwan's 
political discourse revolves primarily around the issue of "independence vs. reunification" with China, 
avoiding complicating factors like religious values and cultural differences. 
By analyzing Taiwan's case, this study aims to disentangle the complex web of factors that influence 
the relationship between democracy, taxation, and income redistribution. The clear political 
dichotomy in Taiwan provides a controlled environment to investigate the impact of national identity 
on this relationship, akin to conducting a controlled experiment in a well-defined laboratory. 
Keywords: democracy, taxation, income redistribution, blue-collar workers, Taiwan, national 
identity 
 
  
I Introduction   
Since the income of poor voters is less than the mean, then hypothetically they should prefer a tax rate 
of unity and fully redistribute all income to the mean. However, in many democracies, the real fact is 
that blue-collar workers tend to support right-wing parties despite these parties wanting to limit income 
redistribution. Although the literature attempting to explain this puzzle is expansive, the results are 
quite diverse for the sake of heterogeneous effects of democracy on tax and inequality as surveyed by 
Acemoglu et al. (2015). For instance, Olson (1993), McGuire and Olson (1996), and Niskanen (1997) 
show that, when democracy gives poor people the right to vote, their overall tax bills are lower than 
those in non-democracies. On the other hand, Meltzer and Richard (1981) argue that an expansion of 
democracy should lead to greater tax revenues and redistribution. Aidt et al. (2006) and Aidt and 

mailto:contact@americaserial.com
mailto:contact@americaserial.com


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

30 | P a g e  

Jensen (2009) also use historical panel data of democratization in Europe and find robust positive 
effects of suffrage on tax revenues as a percentage of GDP. 
There are good reasons for being skeptical about the earlier literature, since the effect of democracy on 
taxation identified in these models typically fails to “fully” capture the impact of omitted fixed effects, 
such as religious values, racial discrimination, and other sub-cultural traits. Due to the complicated 
interactions of these factors with democracy, the association between voting and tax redistribution 
becomes difficult to interpret and inconclusive. To avoid biases from these unobserved heterogeneities, 
our research uses a particularly simple case - Taiwan’s democratization -to filter out interactions that 
are likely to bias the estimates. The case of Taiwan is interesting based on its special political relations 
with China, the bitter ideological contradiction between the country’s two main political parties - the 
KMT (Kuomintang, forming the Pan-Blue Coalition) and the DPP (Democratic Progressive Party, 
forming the Pan-Green Coalition)-and the parties’ supporters. From the perspective of empirical 
studies, using this case to investigate the relationship between democratic election and tax 
redistribution is easy and correct as there are not any public debates on left-right dimension, racial 
discrimination, religious beliefs, and cultural difference in Taiwan. The “only” issue that generally 
matters in Taiwan’s politics is “independence vs. reunification” with China, which concerns contested 
national identities. This dichotomy gives rise to there being only two major political parties in Taiwan, 
with the pro-independence DPP and the anti-independence KMT. This clear and easy framework thus 
allows researchers to investigate the impact of national identity on the relation between election and 
taxation in an “other things being equal” environment, almost like a scientist conducting an experiment 
in a designated science laboratory.  
We specifically focus on the voter mobilization strategy of both parties and as how tight is the link 
between income inequality and voting behavior in Taiwan? Does the electoral competition between the 
two parties focus on the economic context in which voters’ preferences toward redistribution can be 
inferred from their incomes?    
Does the importance of non-economic issues (such as national identity) compete or even dominate that 
of economic redistribution? If so, then how do voters’ preferences on national identity correlate with 
their socioeconomic status such as income and education? More importantly, do poor voters place 
greater weight on noneconomic issue preferences than rich voters, as proposed by some Marxian 
thinkers, e.g., Roemer (1998)? Since these issues cannot be addressed without resorting to micro-level 
individual voter behaviors, we thus use Taiwan Social Image Survey(TSIS, hereafter) data and World 
Values Survey (WVS, hereafter) data, which cover income, education, and ideology patterns of more 
than 3,000 adult respondents across Taiwan society.  
Finally, our results show in Taiwan that poor voters tend to prefer the ideology of national identity 
rather than material benefits of redistribution, but rich voters want the opposite. This encourages both 
parties not only to disregard income inequality as a problem, but also to compete at cutting taxes and 
to offer other rich-friendly policies to cater to the rich. More importantly, our empirical study shows 
that this is the case in large part, because the median voters in the spectrum of 
unification/independence are wealthier than the polarized voters of Pan-Blue and Pan-Green (or the 
mean income of the population). This evidence provides an empirical support to the argument of 
Roemer (1998).    

mailto:contact@americaserial.com
mailto:contact@americaserial.com


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

31 | P a g e  

The remainder of this paper is organized in the following manner. Sections II and III discuss Taiwan’s 
politics and Taiwanese people’s national identity after democratization. Section IV models the linkages 
among income distribution, national identity, and taxation. Section V present data description and 
empirical findings. Section VI concludes the paper.   
II Taxation and Inequality after Taiwan’s Democratization 

The death of the KMT political strongman, Chiang Ching-kuo (蔣經國), was a pivot to divide Taiwanese 
politics into two regime types: autocracy and democracy. Figure 1 illustrates his death in 1988 as a 
watershed for the state capacity to tax in Taiwan. Before democratization, the Chiangs’ authoritarian 
regime empowered the government with enough capacity to enforce tax rules. During the period of 
autocracy, the ratio of tax to GDP exhibited a pronounced increasing trend due to a centralized 
strongman governance. However, as Taiwan began intensive democratization in the late 1980s, the 
politics of the island nation presented two critical trends.  
A. Weak state capacity to collect tax.  The development of democratization produces extraordinary 

pressures from various interest groups 淤 engaging in rent-seeking to ensure advantageous tax 
treatment, especially regarding capital gains. This in turn pushed government tax revenues to exhibit a 
downward trend from 1988 onwards. Figure 1 shows that the ratio of tax revenues to GDP has been on 
a significant long-term decline since the 1990s due to a series of tax cuts for rich capitalists and 

landlords, e.g., implementing an integrated income tax system in 1998 (兩稅合一), reducing the land 

value increment tax by 50% in 2002 and 2004 (土地增值稅減半徵收), permanently reducing the land 

value increment tax rate in 2005 (調降土地增值稅率), and reducing the rate of inheritance tax and gift 

tax from 50% to 10% in 2009 (調降遺贈稅率) and the tax rate of dividend income from 45% to 28% in  

 
Figure 1: Ratio of Tax Revenues to GDP (%)  

.   2019 

  

  

5  

7  

9  

11  

13  

15  

17  

19  

21  

mailto:contact@americaserial.com
mailto:contact@americaserial.com


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

32 | P a g e  

Notes: The black straight line indicates the year of Chiang’s death. The blue line indicates the ratio of 
total tax to GDP. The dotted line denotes the long-term trend. Source: Taiwan Statistical Data Book 
(2017) published by the National Development Council. 
Even worse, 70% of the increase in Taiwanese wealth has resulted from land appreciation and gains in 
securities, which are subject to low tax rates as well as various tax credits and exemptions. Therefore, 
the de facto income tax rate for capitalists and landlords is merely 8-10%, which is much lower than the 
maximum income tax rate (40%) or AMT (alternative minimum tax, 20%). Therefore, a series of tax 
cuts has reduced the ratio of tax revenue to GDP from 19% in 1990 to 13% in 2016, which is even lower 
than the 16% for Sub-Saharan Africa (2013 data), not to mention the 26% for the U.S. (2017), 31% for 
Japan (2015), and 34.3% (2017) for OECD countries.  
B. Worsening income distribution.  Before democratization, the Taiwan government used martial 
law to regulate strikes by workers and to repress trade unions. After democratization, the policies 
against poor laborers were mostly relaxed or lifted, and hence one would expect an improvement in 
income distribution when political power shifts to the poorer segments of society through the channel 
of election. However, after democratization in the 1980s, income inequality in Taiwan conversely began 
to significantly worsen. Figure 2 shows that the ratio of the richest 20%’s income to the poorest 20%’s 
income rose from 4.9 in 1989 to 6.1 in 2017. Although various factors influence the growth in the gap 
between rich and poor (e.g., globalization), there is no doubt that the government’s low tax rate with 
little redistribution has an important role. 
III Several Stylized Facts and National Identity   
This section presents several stylized facts about Taiwan politics as follows.  
A-1. Left/Right party cleavage is not apparent. Most Taiwanese people are under the impression that 
the KMT has an advantage in promoting economic growth, while the DPP’s advantage is in advocating 
social welfare. The literature of class politics (e.g., Hu and Lin, 2010) also shows that the bulk of DPP 
supporters are blue-collar workers, farmers, and self-employed citizens, while KMT supporters are 
mainly middle-class and capitalists. This can be seen by a question of the WVS survey,in which the 
respondents are asked to indicate their income positions. The answers are scaled from 1 (lowest income) 
to 10 (highest income). The evidence shows that the average income rank of PanBlue voters (5.3) is 
significantly higher than that of Pan-Green voters (4.5). The TSIS data also show that the median  

mailto:contact@americaserial.com
mailto:contact@americaserial.com


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

33 | P a g e  

 
Figure 2: Ratio of the Richest 20%’s Income to the Poorest 20%’s Income  
Notes: The black straight line indicates the year of Chiang’s death. The blue line indicates the ratio of 
the richest 20%’s income to the poorest 20%’s income. The dotted line denotes the long-term trend. 
Source: Database of the Directorate-General of Budget, Accounting and Statistics.  
In order to better serve its supporters, the DPP should theoretically seek to tax the rich and redistribute 
society’s wealth. However, as previously argued, the real fact is that both parties coincidentally impose 
a low tax (especially on capital gains) to benefit the rich and hence shift the tax burden to wage earners. 
Therefore, a welldesigned progressive tax system has never appeared in Taiwan. Both parties 
implement almost the same policies toward internal affairs, economic development, religion, and 
justice, except that the DPP opposes teaching Chinese history in favor of Taiwan history. This means 
the traditional left-right dichotomy cannot describe Taiwanese elections at all.   
A-2. Non-economic dimension increases in importance in Taiwan politics. Although Taiwan society 
lacks stable “class identification”, its politics are deeply influenced by “national identification”. Since 
1996, when Taiwan held its first presidential campaign, the issue of Taiwan independence has especially 
become the major focus in electoral politics and has dominated the issue of income redistribution. This 
makes the controversy of ideology as the major factor delineating political parties on either side of the 
‘‘Taiwan independence/China unification’’ axis. On the other hand, income redistribution becomes the 
accessory to national identity. This can be seen by the TSIS data that show respondents preferring to 
“maintain status quo indefinitely” decreased from 66.9% to 44.6% between 2009 and 2015. However, 
respondents who support moving toward independence or unification increased substantially (see 
Table 1). This evidence implies that the conflict of national identity has become more severe in Taiwan.   
Table 1: Taiwanese voters’ stances on the independence/unification issue  
 

 

monthly income of Pan-Blue supporters (NT$35,118) is higher than that of Pan-  
supporters  Green  

( NT$30,605).  
  
  
  
  
  

  
  
  
  
  
  
  
  
  
  

  

0 

1 

2 

3  

4 

5 

6 

7 

mailto:contact@americaserial.com
mailto:contact@americaserial.com


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

34 | P a g e  

Voters  2009   2010   2015   

Maintain status quo, but move toward 
independence  

24.70%  21.00%  38.50%  

Maintain status quo indefinitely  66.90%  69.60%  44.60%  

Maintain status quo, but move toward 
unification  

8.50%  9.40%  16.90%  

Sample size   1,043  1,084  1,088  

Notes: “Maintain status quo, but move toward independence” includes both “Independence as soon as 
possible” and “Maintain status quo & move toward independence.” “Maintain status quo, but move 
toward unification” includes both “Unification as soon as possible” and “Maintain status quo & move 
toward unification.” Data are collected from the TSIS dataset in 2009, 2010, and 2015.   
A-3. Both parties divergent in Taiwan independence, but convergent in low taxes. Although both the 
KMT and the DPP stand together in cutting taxes for the rich, their positions in the spectrum of Taiwan 
independence are quite different. The KMT is opposed to independence and leans toward reunification, 
while the DPP advocates or at least pays lip service to independence. This can be seen by the perception 
of the electorate toward both parties. Of PanGreen voters, 22 + 36 = 58% tend to support Taiwan 
independence as soon as possibleand slowly moving toward independence, while only 2.6 + 8.9 = 11.5% 
of Pan-Blue votersagree with these issues (see Table 2). One thing worth mentioning is that most Pan-
Blue voters and non-partisan voterstypically cling to the status quoindefinitely.The TSIS survey also 
shows that most Pan-Green voters call for a Taiwanese national identity, but only few Pan-Blue voters 
support unification with China. Hence, the real political debate should be “proindependence vs. anti-
independence,” rather than “independence vs. unification with China.” Based on this evidence, the rest 
of the paper assumes that both parties have heterogeneous preferences in “Taiwan intendance”, rather 
than “unification with China.”   
Table 2: Taiwanese voters’ stances on the independence/unification issue  

 
Voters  Pan-Green (1)  No partisanship (2)  Pan-Blue (3)  
Independence as soon as possible  22.0%  9.5%  2.6%  
Maintain status quo & move toward 
independence  

36.0%  13.4%  8.9%  

Maintain status quo indefinitely  35.3%  69.8%  70.0%  
Maintain status quo & move toward 
unification  

5.0%  5.3%  13.9%  

Unification as soon as possible  1.7%  2.1%  4.6%  
Observation number   897   1,026  1,292  
Note: The sample covers3,215 respondents surveyed by the Taiwan Social Image Survey in the years 
2009, 2010, and 2015.   
A-4. Poor voters are concerned more on national identity, but the rich think otherwise. Some 
researchers find little to no significant effect of economic voting on the Taiwanese electorate (e.g., Hsieh 

mailto:contact@americaserial.com
mailto:contact@americaserial.com


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

35 | P a g e  

et al. 1996). To our mind, these studies generally fail to account for the heterogeneity of the electorate. 
Although conforming to their argument that poor and low-educated voters indeed are concerned more 
about national identity than economic redistribution, the rich and educated voters seem to be more 
concerned with taxation and reluctant to redistribute their income to the poor. This can be verified by 
the evidence provided by Table3, which uses TSIS data to show that the average years of schooling and 
average monthly income (12.65 years and NT$35,375) of the respondents who believe that national 
sovereignty is more important than economic interests are significantly lower than those with the 
opposite view (13.49 years and NT$40,218). Moreover, the education and income levels of respondents 
who support independence or unification as soon as possible (12.32 years and NT$34,151) are lower 
than those preferring to maintain the status quo currently or indefinitely (13.32 years and NT$39,126).  
A-5.  
Table 3: Education and income level of Taiwanese voters with different thinking on cross-strait 
negotiations  

 
Prefer economic interests to Prefer national sovereignty to  
  national sovereignty or are  
economic interests concerned with both  

Average  years 
 of  

 

schooling  12.65  13.49   

  
Average 
 monthly  

t value = 6.36   

income (NT$)  
  

35,375  40,218  
t value = 2.37   

  Support independence or  Maintain the status quo 
unification as soon as possible  currently or 
indefinitely  

Average  years 
 of  

 

schooling  12.32  13.32   

mailto:contact@americaserial.com
mailto:contact@americaserial.com


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

36 | P a g e  

  
Average 
 monthly  

t value = 4.93   

income (NT$)  
  

34,151  39,126  
t value = 2.79   

Notes: All data are collected from the TSIS datasets in 2009, 2010, and 2015.The pooled dataset 
includes 3,215 observations.   
Some rich people admittedly may be indistinguishable from the poor in their preferences on national 
identity dimension. However, it is reasonable to expect that a sufficient percentage of them cares more 
about economic redistribution in spite of their political ideology. In order to protect her economic 
interests, a rich PanGreen (or Pan-Blue) voter would vote for the KMT (or DPP) if the party proposes a 
relatively low tax rate with little redistribution. Hence, perhaps because of loss aversion, there is 

asymmetry whereby the rich base 淤 their votes primarily on their economic interests, whereas the 
dimension of religious values is more salient among the poor. 
At this point, even though the DPP (or KMT) has large numbers of poor pro-independence (or anti-
independence) electorate, as long as the sense of national identification is highly salient among these 
voters, both parties will have strong incentives to sacrifice the material benefits of their poor supporters 
in order to win the critical support of the rich voters. This unsurprisingly moves the tax policy outcome 
away from the poor constituency’s ideal point. As indicated by a DPP scholar (Chen, 2010), a political 
consultant of President Tsai, although the DPP self-motivates to represent the disadvantaged classes 
and attempts to change the electoral mobilization campaign from identity cleavage to class cleavage, 
this approach has not succeeded owing to a growing sense of national identity, which makes Taiwan 
independence become the main focus of the campaign. In order to win the election, even if the DPP 
wants to move to the left, it still cannot propose a tax policy with a high redistribution rate, since it 
needs support from the rich voters, especially the ones with a neutral perception of Taiwan 
independence. At this point, the issue of redistribution at best is only a secondary factor in the voting 
campaigns of Taiwan.   
A-5.Low-income voters with strong national identity tend to be DPP supporters.  One thing worth 
emphasizing is that the low-income voters with strong national identity are mostly DPP supporters, as 
previously described in Subsection III-A-1. In other words, Pan-Green with a strong sense of national 
consciousness are generally poorer than PanBlue or “secular” voters. At this point, if her ideological 
identity is stronger than her need for redistribution, then she will vote for the DPP no matter how low 
the tax rate is that the DPP proposes. This gives the DPP an incentive to cut the tax rate in order to 
please rich voters, even though it wants to turn to the left rather than right. 
B. Why do the poor not expropriate the rich?  One of the reasons that the Chiang regimes did not extend 
suffrage is because they believed that, were the poor to be given the vote, they would soon expropriate 
the rich industrialists. Nevertheless, after democratization, Taiwan’s suffrage has not engendered the 
expropriation of the rich through taxation, but rather has seen it go the other way round.  

mailto:contact@americaserial.com
mailto:contact@americaserial.com


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

37 | P a g e  

When addressing this issue, the literature mostly focuses on the following two explanations: (1) poor 
voters expect themselves or their children will someday become rich (e.g., Piketty, 1995; Benabou and 
Tirole, 2006); (2) the poor believe that there would be adverse effects to expropriating the rich, who 
have productive talents that would cease to be supplied under high tax rates, and all would consequently 
suffer (e.g., Shayo, 2009). However, the literature is hardly applicable to the case of Taiwan based on 
the following two reasons. First, most of these studies are performed under a one-dimensional spatial 
model, in which the policy space is about tax redistribution and voters’ preferences are driven primarily 
by their places in the income spectrum. Second, even some research studies, such as Poole and 
Rosenthal (1991), Roemer (1998), and De La O and Rodden (2008), treat a democratic election as a 
two-dimensional issue over redistribution and religion, in which (1) the proletarian voters want more 
redistribution; while the capitalists want the opposite; or (2) the religious voters want more police, 
illegalization of abortion, and the death penalty; while the secular voters want the respective opposites. 
However, in their models, the left and the right always have divergent tax policies in that the left 
proposes a higher tax rate than the right one. However, Taiwan’s election politics are different in that 
both the DPP and the KMT unanimously converge to the same low tax rate. Due to the fierce conflict 
over Taiwan independence, Taiwan’s politics neither involves issues regarding religion nor class 
consciousness, as stressed by those other authors. In other words, “pro-independence vs. anti-
independence” is the only issue that matters in Taiwan’s politics.  
In order to explain both parties (especially the DPP) favoring the rich over the poor and their failures 
to favor a high tax for redistribution, the researchers should focus on the differences in ideological 
patterns between rich and poor voters. 
We hence instead propose national identity as a possible explanation for the non-expropriation of the 
rich in Taiwan’s form of democracy. This approach leads to a two-dimensional platform in which 
political parties compete in a policy space with two issues: taxation (economic dimension) and national 
identity (non-economic dimension); and both have the potential to interfere with each other. Under 
Taiwan’s unique democratic regime, we believe that this specification may help to explain through 
which channel can national identity lead to reduced support for redistribution.   
C. National identity. Our viewpoint is quite consistent with the Marxist argument formalized by 
Roemer (1998), in which the explanation for the reduced support of the proletarian voters for 
redistribution is that they do not appreciate their economic self-interest in progressive taxation and 
redistribution. This unawareness provides politicians an opportunity to mobilize the voters more easily 
around the issue of group identities, causing the equilibrium amount of redistribution to decrease. In 
fact, a fairly robust literature has shown that non-economic factors (e.g., religion) are surprisingly 
powerful and stable predictors of voting behavior, whereas the importance of economic factors(e.g., 
class conflict) continues to decline. For instance, De La O and Rodden (2008) show that political 
religion may generate a “moral puritanism effect,” which induces an electorate to prefer lower 
redistribution. Scheve and Stasavage (2006) also argue that religious individuals may prefer lower 
levels of income redistribution, because they could derive psychic benefits from the religion that serves 
as a substitute for the welfare state in buffering individuals against adverse life events. Basically, this 
approach highlights a Marxian perspective that preferences on non-economic issues that have a 

mailto:contact@americaserial.com
mailto:contact@americaserial.com


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

38 | P a g e  

disproportionate pull on the poorest, least educated voters, whereas rich and educated voters are still 
more concerned with economic self-interest (Frank, 2004). 
Based on this line of argument, we believe that national identity can develop to support the formation 
of “political religion”, especially for poor voters. This implies that individuals with more religion in 
national identity would have significantly different preferences than secular individuals on both 
economic and non-economic dimensions, and these differences may become a powerful predictor of 
differences in their vote choices.  
D. Sociotropic and pocketbook voting. This subsection focuses on individual variations in the levels 
of education and income that influence voters to vary in their abilities to clearly attribute the 
responsibility for pecuniary changes to the government. In general, recognizing the linkage between 
politics and personal well-being requires higher cognitive ability than identifying the political relevance 
of national identity. Well-informed voters can easily link governmental actions and personal economic 
conditions, while such attribution is more difficult for the less informed (Duch, 2001; Gomez and 
Wilson, 2006). On the other hand, national identity is to some extent determined by ethnicity and 
language, which are manifest to anyone (Edwards, 2009). Therefore, concerning the ability to recognize 
the impact of politics on an individual’s material interests, pocketbook voting based on personal 
material interests is stronger among the informed electorate, while non-economic (sociotropic) voting 
based on national identity is more likely to occur among the less-informed (De La O and Rodden, 2008).  
In Taiwan politics, the above argument is supported by Choi (2010),who proposes that it is a well-
informed electorate rather than a less-informed one that can be involved in both sociotropic and 
pocketbook voting, because these voters can assign responsibility for both personal income and 
national identity to the government. On the other hand, low-educated individuals are less informed at 
being able to ascertain how governmental policy affects their personal incomes. They thus can only 
focus more on a sociotropic issue (Taiwanese identification), which more or less connects the subethnic 
cleavage between benshengren (Taiwanese people of local origin) and waishengren (Taiwanese people 
of recent China descent). For this reason, sociotropic voting should be more prevalent among the less-
informed electorate. Since this cohort constitutes a significant portion of diehard supporters of the DPP 
and KMT, both parties hence have strong incentives to mobilize voters by highlighting national identity 
to expand their social bases.  
IV. Two-Dimensional Voting model  
This section establishes a two-dimensional model grounded in the previous stylized facts to investigate 
the roles of national identity and redistribution in the competition between the KMT and the DPP. 
A. Tax and utility. Let the space of voters’ traits be Σ = 𝑌 × 𝐴, with generic element (𝑦, 𝑎). Here, is 
the set ofy, which is the pre-tax income of an individual, and A, taken to be real number line, is the set 
of national identity views a. We use the degree of a voter’s support for independence to measure 
national identity. The higher a is, the more she supports Taiwan independence. A voter’s material 
benefits can be proxied by after-tax income (1 − 𝑡)𝑦 + 𝑘, where t∈[0,1]is the tax rate, and:  
𝑘 = (𝑡 − 𝛽 ),               (1)  
in which 𝑦 is the mean of y, and t is a uniform tax rate on income. Equation (1) represents a standard 
model of redistribution financed by distortionary taxation (Bolton and Roland, 1997; Shayo, 2009), in 
which income taxation involves quadratic deadweight losses. These losses include a tax-levying cost, 

mailto:contact@americaserial.com
mailto:contact@americaserial.com


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

39 | P a g e  

government corruption, and resource misallocation driven by income tax (e.g., unproductive use of 
resources, capital and talent outflow, and rent-seeking by interest groups). 
The model includes two political parties: the DPP represents voters with traits(𝑦 , 𝑎 ); whereas the KMT 
represents voters with (𝑦 , 𝑎 ).They respectively propose a policy pair τ(𝑡, 𝑧), in which z is the party’s 
position toward support for Taiwan independence. Finally, the utility function of a voter with traits (𝑦, 
𝑎) over policies is (𝑡, 𝑧; 𝑦, 𝑎), and 𝑎 > 𝑎. 
B. Utility for different patterns of national identity.  Assume that both parties are heterogeneous in 
national identity such that, in a Hotelling line representation of Taiwan independence, the position of 
DPP (KMT) is denoted by 𝑧 (𝑧 ), and is located at the RHS (LHS) of the spectrum representing 𝑧 > 𝑧 or 
“pro-independence vs. antiindependence”. Furtherassume that both parties are not Downsian in that 
they wish to maximize the probability of winning the election in order to implement pro- or anti-
independence policies.This involves two conflicting incentives:  (1) both parties take different 
ideological preferences and have divergent incentives to adopt a platform that is truthful to their policy 
preferences; (2) however, since their preferences cannot be translated into real-world policy without 
winning the election, they thus have the incentive to move toward the middle so as to increase the 
chances of winning. These contrasting incentives (policy goals and electoral needs) create an outcome 
of “partial convergence” in that both parties propose similar policy packages except their most favored 
policies, i.e., pro- and anti-independence.  
C. Two-dimensional politico-economic model.  The DPP claims that it serves the interests of 
workers and farmers and always wins more votes from the poor than the KMT. However, in the real 
governance of taxation, the DPP has never adopted a tax policy that expropriates incomes from the rich 
and gives to the poor. The following twodimensional model modifies the setup of Roemer (1998) and 
uses a voter’s preference to explain why the DPP would not propose a highest possible tax rate 
(redistribution) at the cost of losing the votes of the better-off individuals. 
A voter’s optimization problem contains two issues:  taxation and national identity. Following Roemer 
(1998), we assume that a voter with an ideological view, a, has a utility function toward the 
government’s ideological position (𝑧; 𝑎) = − 𝛼⁄2(𝑧 − 𝑎) , in which a higher value of 𝑧 implies a more 
independence-leaning policy. A positive number 𝛼 ≥ 0 is used to measure the salience of the national 
identity. Finally, assume that the voter is risk neutral with the von Neumann–Morgenstern utility 
function, and her utility at tax rate t is the post-tax income and the government’s independence policy, 
z. Therefore, at policy (𝑡, 𝑧), the indirect utility function of a voter with the combination of economic 
income and political preference (𝑦, 𝑎) becomes:  
(𝑡, 𝑧; 𝑦, 𝑎) = (1 − 𝑡)𝑦 + (𝑡 − 𝛽 )𝑦 − (𝑧 − 𝑎) .    (2)  
A political party chooses tax rate (t) and ideology position (z) to maximize the probability of winning 
the election, meaning that it must do its best to satisfy the needs of the electorate. Differentiation gives 
the equilibrium conditions in a tie-in electoral vote:  
Δ𝑡 (𝑧  − 𝑎) 

=                          (3) Δ𝑧 (1 − 𝛽𝑡) − 𝑦 
Here, 𝑧  = (𝑧 +  )⁄2 represents the average of both parties’ attitudes toward independence, and hence𝑧 > 
𝑧  > 𝑧 . Equation (3) shows that the relationship between t and z depends on the distributions of a and 

mailto:contact@americaserial.com
mailto:contact@americaserial.com


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

40 | P a g e  

y. When the deadweight losses associated with the income tax is low, 𝛽 ≤ (𝑦 − 𝑦)⁄𝑡𝑦, a poor voter can 
benefit from the redistribution of higher tax rates:  

 𝑦   =  (1 − 𝑡) + (𝑡 − 𝛽 )𝑦 = (1 − 𝛽𝑡)𝑦  − 𝑦 > 0.(4)  
However, the following will show that if she prefers being a citizen of the Republic of Taiwan much 
more than being a beneficiary of redistribution, then her preference will encourage the DPP to choose 
a lower tax rate to appeal to rich voters rather than to enforce a high tax redistribution to represent its 
poor supporters’ economic interests. 
D.DPP’s electoral strategy toward poor Pan-Green supporters. Using equation (3), one may compute 
the precondition that a poor Pan-Green supporter (𝑎 > 𝑧 )will sacrifice her redistribution benefits and 
vote for the policy package of DPP, 𝜏 (𝑡 ,𝜏 ):  
Δ𝑡 [𝑦 − (1 − 𝛽𝑡)] 

 ≤ 𝑎 − 𝑧,                           (5) 𝛼Δ𝑧 
in which Δ𝑡 = 𝑡 − 𝑡 < 0 and Δ𝑧 = 𝑧 − 𝑧 > 0 . Since she is poor and “Green”, we have both  
𝑦 − (1 − 𝛽𝑡) ̅ < 0 and (𝑎 − 𝑧) ̅ > 0. However, under a Downsian model of political competition, regardless 
of her economic interests, the DPP may choose low tax redistribution to maximize the probability of 
winning the election, if the following two conditions hold.  
(1) The non-economic political dimension becomes more salient in the electoral arena(meaning that 
the value of α is large enough to ensure equation (5)), such that many poor Pan-Green voters will vote 
for the DPP even if the DPP sets a tax rate lower than the KMT.  
(2)The ideologically median voters concern themselves with economic benefits rather than political 
identity and hate to split their incomes with others. More importantly, as indicated by Roemer (1998), 
they must be wealthier than Pan-Green voters, andso a low tax rate not only caters to these median 
voters, but it also will not drive away poor polarized voters. 
This certainly encourages the DPP to increase z toward a more ideological pole and, in the meantime, 
to lower taxes to please rich voters. Hence, the more salient the national identification (α) issue is, the 
more likely that the DPP sets a lower rate than the KMT. 
As to the variable β, equation (5) shows that if low-income voters do not believe that the government is 
efficient and honest enough to implement redistribution, then an increase in β will discourage the DPP 
to propose alower tax rate.Regrettably, the WVS data show that the ratio of Taiwanese respondents who 
have “quite a lot confidence” in government has significantly dropped from 60.6% in 1995-1999 to 
39.4% in 2010 to 2014, implying that it is increasingly difficult for the political parties to cater to poor 
voters by tax redistribution.  
E. DPP’s electoral strategy toward ideologically neutral voters. Since ideologically neutral voters 
are in the median position of the independence spectrum, we have 𝑎 − 𝑧  ≅ 0. If the mean wealth of 
ideologically neutral voters is greater than that of the entire population, [𝑦 − (1 − 𝛽𝑡)𝑦] > 0, and if this 
cohort is concerned more about the redistribution issue than the others, then the DPP caneasily win 
their votes by setting Δ𝑡 = 𝑡 − 𝑡 ≤ 0to make the following equation hold true:  
Δ𝑡 [𝑦 − (1 − 𝛽𝑡)] 

 ≤ 𝑎 − 𝑧  ≅ 0                (6) 𝛼Δ𝑧 
Again, this unsurprisingly encourages the DPP to propose a lower taxation than that of the KMT. 

mailto:contact@americaserial.com
mailto:contact@americaserial.com


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

41 | P a g e  

F. KMT’s electoral strategy and Stackelberg equilibria. In the same way, we may expect that the 
KMT will choose the same low tax strategy as that of the DPP. The basic premise is that the “religious” 
issue is so salient that both parties can do nothing but to propose a low tax rate. This low tax rate 
equilibrium can be explained by a Stackelberg two-stage game. Assume that the DPP is the incumbent 
and the KMT is the challenger, where the challenger moves first. One may solve the game by backward 
induction and assume that the KMT sets a tax rate  . However, no matter how low 𝑡 is, in order to win 
the election, the DPP will set 𝑡 ≤ 𝑡 to cater to ideologically neutral voters. Likewise, the same is true 
forwhen the KMT is the incumbent. When operationalized as party alternation, hypothetically, if α → 
∞, then the convergence of this iteration may generate a unique electoral equilibrium 𝑡∗ = 𝑡∗ = 0. 
G. An intuitive explanation. Suppose that poor voters are concerned more with national identity 
than income redistribution; while the rich score the other way around. As long as the KMT (DPP) 
proposes a more antiindependence (pro-independence) policy than the DPP (KMT), there will be a 
significant number of poor voters who are so pro-independence (anti-independence) that they will not 
vote for the KMT (the DPP) even if it proposes a higher redistribution rate than that of the DPP (the 
KMT). Given that they prefer ideological orthodoxy to material interest and hold a belief, “Vote for the 

DPP, even if even if it makes us hungry!「肚子餓扁也要投阿扁」”, the DPP predictably will propose a 
low tax rate, thereby winning the votes of rich voters who are neutral toward independence, but are 
concerned about a tax increase or redistribution. From a welfare perspective, the DPP’s low taxation 
strategy not only can help to win the election, but also can maximize the expected welfare of both poor 
pro-independence and rich anti-redistribution constituents. However, this also makes the Pan-Green 
poor voters, who act against their material interests, become addicted to the opiate of ideological 
identity.  
V. Data Description and Empirical Setting 
A. Two preconditions. This section uses probit analysis to prove two propositions of previous 
studies: (1)People with higher income and better education tend to be ideologically neutral voters rather 
than polarized voters (no matter whether pro-independence or anti-independence). The reverse is also 
true for poor and loweducated voters.(2)People with higher income and better education tend to be 
concerned more about economic interest than political identity. The reverse is also true.  
B. Data description.  The data are drawn from TSIS and WVS that collected the information 
annually on respondents’ age, education, income, their attitudes toward economic opportunity, and 
political participation. Even more favorably, in the years of 2009, 2010, and 2015, the TSIS had special 
issues on “Taiwan independence vs. reunification with China,” which provide pertinent information 
concerning national identity. Across these three years, the TSIS dataset comprises1,043observations 
for 2009,1,084 for 2010, and 1,088 for 2015. Totally, we have a pooled dataset of 3,215 observations to 
perform the empirical study. On the other hand, the WVS dataset includes 2,789observations (707 for 
1995, 1,125 for 2006, and 957 for 2010). Table A-I in the Supplemental Materials provides the text of 
the questions for the key variables. Tables A-II, A-III, and A-IV list the definitions and descriptive 
statistics of each variable and how they are constructed based on the respondents’ answers to the 
survey. We now briefly describe variables used in this study.  
1.Median: This is a binary variable used to denote whether or not the respondent is a median voter in 
the spectrum of Taiwan independence. With reference to questions asked in TSIS, there are two related 

mailto:contact@americaserial.com
mailto:contact@americaserial.com


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

42 | P a g e  

to political ideology. The first one is regarding national identity in which the respondents are asked to 
answer the following “thermometer” question: “There are several views on the relationship between 
Taiwan and China. Which one do you prefer?” The response categories include (1) independence as 
soon as possible; (2) maintain status quo & move toward independence; (3) maintain status quo 
indefinitely; (4) maintain status quo & move toward unification; and (5) unification as soon as possible. 
The second one regards the preference for Taiwan’s relationship with China: “In cross-strait 
negotiations, which one do you think is more important, economic interests or national sovereignty?” 
(1) national sovereignty; (2) both; (3) economic interests. We use a respondent’s answers to these two 
questions to identify if she is a median voter. If her response for the first question is “maintain status 
quo”[(2)+(3)+(4)] and also her response for the second question is “both” or “economic interests” 
[(2)+(3)], then she is identified as a median voter, and the observation of her response will be recorded 
as 𝑀𝑒𝑑𝑖𝑎𝑛 = 1;otherwise,she is a polarized voter, and 𝑀𝑒𝑑𝑖𝑎𝑛 = 0. Hence, the median voters are those 
who prefer to maintain status quo and meanwhile are concerned about economic interests. 
2. Income and Education: We use the monthly income (20K, 40K, 60K, 80K, 100K)9of a respondent as 
the proxy for the income status of a respondent. Rows (1) and (2) in Table 4give the mean incomes of 
median voters and polarized voters. Row (5) also shows that the income of the median cohort is 
significantly higher than that of the polarized cohort.10The same conclusion also applies to the 
education level (Bachelor, Senior high, Junior high) (see the average years of schooling in Rows (6) to 
(8)). This preliminary evidence seems to indicate that median voters are wealthier and better educated 
than the polarized ones.   
Table 4: Salience of national identity in Taiwan  

 
(1) Mean Income of Median Voters (𝜇 )  NT$40,534  
(2) Mean Income of Polarized Voters (𝜇 )  NT$35,552  
(3) Size of Median Cohort  1,883  
(4) Size of Polarized Cohort  1,332  
(5) Pr (𝜇 ≥ 𝜇 )  99.99%  
(6) Average Level of Median Cohort’s Education (𝑒 )  13.57 years  
(7) Average Level of Polarized Cohort’s Education (𝑒 )  12.64 years  
(8) Pr (𝑒 ≥ 𝑒 )  99.99%  

 
Note: The figures are computed by using the data of TSIS. 
3. Exogenous Controls (X):  A set of exogenous variables is used to control for possible heterogeneity 
across observations. These individual-specific variables comprise gender (Female),marital status 
(Divorced, Married), employment status(Unemployment, Retired, Housekeeper, Student), year 
dummy (2015Y,2010Y), partisanship (Pan-Green, Pan-Blue), ethnicity (Aborigine, Waishengren, 
Benshengren), and resident districts (Eastern Taiwan, Southern Taiwan, Central Taiwan). 
C-1. Empirical results: neutral voters are wealthier and better educated. We first prove the proposition 
that higher income and better education tend to influence a voter to be neutral in political ideology. As 
our dependent variable (Median) is binary, we thus use a probit model to estimate the effects of income 
and education on Median.   

mailto:contact@americaserial.com
mailto:contact@americaserial.com


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

43 | P a g e  

            (𝑀𝑒𝑑𝑖𝑎𝑛 = 1〡𝐸𝑑𝑢𝑐𝑎𝑡𝑖𝑜𝑛, 𝐼𝑛𝑐𝑜𝑚𝑒, 𝑿) = 𝑃(𝑍 ≤ 𝛽 + 𝛽 𝐸𝑑𝑢𝑐𝑎𝑡𝑖𝑜𝑛 + 𝛽 𝐼𝑛𝑐𝑜𝑚𝑒 + 𝛀𝑿)           

= 𝐹(𝛽 + 𝛽 𝐸𝑑𝑢𝑐𝑎𝑡𝑖𝑜𝑛 + 𝛽 𝐼𝑛𝑐𝑜𝑚𝑒 + 𝛀𝑿),  
in which Median is a binary dependent variable, E is an individual’s education level, Y is her income, 

and X is a set of control variables.𝑃(𝑀𝑒𝑑𝑖𝑎𝑛 = 1〡𝐸, 𝑌, 𝑿)istheprobability that she is an ideologically 

neutral voter, given the values of 𝐸, 𝑌, 𝑿. 𝑍 is a standard normal variable (i.e., Z ∼ (0, 𝜎 )), and F is a 
standard normal CDF.   
Table 5 Empirical results of the Probit model for Median  

  
  

Model A  Model B  Model C  

 coeff  M.E.  coeff  M.E.  coeff  M.E.  

Intercept  0.307     0.231     0.466     

Age  -
0.008**   

-
0.0033   

-
0.008**   

-
0.0030   

-0.010**   -
0.0038   

Female  0.213**  0.0851   0.234**  0.0933   0.212**  0.0832   

Divorced  0.096   0.0382   0.147   0.0587   0.099   0.0390   

Married  0.021   0.0085   0.069   0.0274   0.029   0.0115   

  
9 These income variables are all binary in that, for instance, 20K denotes that the monthly income 
of a respondent is between NT$20,000 to NT$39,999; while40K denotes income between NT$40,000 
to NT$59,999. The reference group is no income and less than NT$20,000.  
10 Using the Central-limit-theorem test, we find that the probability of 𝜇 >𝜇 is at the significance 
level of 1%.  

 
Unemployment  -0.143  -0.0572  -0.176  -0.0701  -0.159  -

0.0625  
Retired  0.081  0.0323  0.106  0.0422  0.074  0.0289  

Housekeeper  0.000  0.0001  0.004  0.0017  -0.002  -
0.0009  

Student  -0.036  -0.0145  -0.070  -0.0279  -0.064  -
0.0253  

Bachelor  0.257**  0.1024  0.295**  0.1174  0.193*  0.0757  

Senior high  0.272**  0.1083  0.298**  0.1186  0.224**  0.0881  

Junior high  -0.043  -0.0171  -0.009  -0.0035  -0.066  -
0.0260  

100K  0.289**  0.1152  0.230*  0.0915  0.263**  0.1031  

80K  0.163  0.0652  0.141  0.0564  0.126  0.0493  

60K  0.133  0.0530  0.161  0.0640  0.131  0.0513  

mailto:contact@americaserial.com
mailto:contact@americaserial.com


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

44 | P a g e  

40K  0.044  0.0175  0.021  0.0082  0.022  0.0086  

20K  0.053  0.0212  0.023  0.0091  0.046  0.0181  

2015Y  -0.026  -0.0105  -0.027  -0.0827  -0.017  -
0.0066  

2010Y  -0.015  -0.0058  -0.083  -0.0332  -0.046  -
0.0181  

Pan-Green  -0.756**  -0.3015      -0.738**  -
0.2896  

Pan-Blue  0.354**  0.1411      0.326**  0.1279  

Aborigine      -0.246  -0.0981  -0.373  -
0.1463  

Waishengren      0.507**  0.2018  0.276**  0.1085  

Benshengren      -0.080  -0.0318  -0.048  -
0.0188  

Eastern      0.081  0.0322  -0.066  -
0.0261  

Southern      -0.092  -0.0365  -0.052  -
0.0206  

Central      -0.021  -0.0085  -0.057  -
0.0225  

 
Notes: If the respondent is a median voter, then the dep. variable=1; otherwise, the dep. variable=0. 
M.E. is the marginal probability effect. Symbols * and ** represent the p-value is smaller than 10% and 
5%, respectively.  
Table 5 presents the result of the effects of Education and Income on Median, while controlling for 
other exogenous variables. The result evidently shows that the highest variable of personal income 
(100K) has a significantly positive effect on Median. In other words, an individual with a monthly 
income greater than NT$100,000 tends to have no particular interest in national identity (no matter 
whether anti-independence or pro-independence). In terms of the magnitude of effect size, the 
likelihood that she will be an ideologically neutral voter is higher than others by 9.15~11.52%.  
Education also has a significantly positive effect to prevent an individual from taking extreme political 
positions. A college (and above) or senior high graduate voter is more likely to become a median voter 
than the others by 7.57~11.74% or 8.81~11.86%. This evidence provides concrete support for our 
proposition that ideologically neutral voters are wealthier and better educated than polarized voters.  
As to the estimation of other independent variables, the result shows that an individual’s partisanship 
is important in determining her political view. A respondent voting for the KMT (DPP) camp is more 
likely to be a median (polarized) voter. This evidence provides some support to our argument that the 
DPP has more polarized supporters than the KMT. Moreover, a voter whose father is a waishengren 
tends to take a more median political position than those of other groups. Finally, a young or female 
voter is also more likely to be a median voter.  

mailto:contact@americaserial.com
mailto:contact@americaserial.com


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

45 | P a g e  

 C-2. Neutral Voters Are Concerned More About Economic Interests. We now examine if a neutral voter 
is concerned more about economic interests. As TSIS does not explicitly ask participants to rank the 
importance of economic interests and national identity, we thus turn to WVS, in which two questions 
are somehow related to our research interests. The first one can be used to assess a respondent’s 
attitude on whether or not income distribution should be equal. The answers are scaled from 
1(completely agree with the statement “Income should be made more equal”) to 10 (completely agree 
with the statement “We need larger income differences as incentives for individual effort”).We take the 
Taiwanese respondents to construct a variable referred to as Income Difference, which is used to 
identify if a respondent would like to sacrifice her material interests to reduce income inequality. The 
second question is used to identify if a respondent expects the government to improve income 
distribution. The answers are also scaled from 1(completely agree with the statement “Government 
should take more responsibility to ensure that everyone is provided for”) to 10(completely agree with 
the statement “People should take more responsibility to provide for themselves”). We also use a 
respondent’s response to construct a variable referred to as Personal Responsibility, which is used to 
identify if she desires government policies to reduce income inequality. The larger the values of Income 
Difference and Personal Responsibility are, the less a respondent agrees with tax redistribution.   
Finally, Income Difference and Personal Responsibility, taken together, are used to identify if she is a 
voter who believes in individualism and does not want to redistribute income. If her responses to the 
two questions are both higher than 5(Income Difference>5 and Personal Responsibility>5), then she is 
identified as an individualistic voter and the observation is recorded as 𝐼𝑛𝑑𝑖𝑣𝑖𝑑𝑢𝑎𝑙𝑖𝑠𝑡𝑖𝑐 = 1; 
otherwise,𝑛𝑑𝑖𝑣𝑖𝑑𝑢𝑎𝑙𝑖𝑠𝑡𝑖𝑐 = 0. As this dependent variable is binary, we still use a probit model to 
estimate the effects of income and education on Individualistic.  
Table 6 presents the results of the effects of Education and Income on Individualistic, while controlling 
for other exogenous variables. The results evidently show that the family income has a significantly 
positive effect on Individualistic. An individual in the richest 10% of families is more likely to become 
an individualistic voter by 22~23% than one in the poorest 10% of families, implying that rich voters 
tend to support an anti-redistribution party. As to the impact of education on Individualistic, a college 
and above graduate voter (Bachelor) also has no interest in reducing income inequality. In terms of 
magnitude, the likelihood that she will be an individualistic voter is higher than others by 5.86~6.46%.   
Table 6 Empirical results of probit model for Individualistic  

 
  Model A  Model B  Model C  
  
coeff  M.E.  coeff  M.E.  coeff  M.E.  

Intercept  -0.878     -1.006     -1.002     

Age  0.001   0.0004   0.002   0.0005   0.002   0.0005   

Female  -0.009   -0.0033   -0.014   -0.0046   -0.013   -
0.0044   

Divorced  0.002   0.0007   0.002   0.0008   0.004   0.0012   

mailto:contact@americaserial.com
mailto:contact@americaserial.com


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

46 | P a g e  

Married  0.097   0.0349   0.096   0.0324   0.097   0.0329   

Unemployment  -0.086   -0.0310   -0.081   -0.0274   -0.080   -
0.0272   

Retired  0.079   0.0284   0.086   0.0291   0.086   0.0290   

Housekeeper  -0.031   -0.0112   -0.028   -0.0096   -0.029   -
0.0098   

Student  0.018   0.0064   0.024   0.0081   0.025   0.0085   

Bachelor  0.163*  0.0586   0.189*  0.0639   0.191*  0.0646   

Senior high  0.075   0.0269   0.097   0.0327   0.098   0.0333   

Junior high  0.089   0.0319   0.101   0.0340   0.101   0.0343   

Household income  0.063**  0.0226   0.064**  0.0216   0.064**  0.0218   

2010Y  -0.116   -0.0419   -0.136*  -0.0461   -0.136*  -
0.0462   

2006Y  0.025   0.0090   0.015   0.0050   0.014   0.0048   

Pan-Green  -0.033   -0.0119       -0.016   -
0.0054   

Pan-Blue  0.010   0.0036       -0.001   -
0.0004   

Aborigine      0.253   0.0853   0.256   0.0866   

Waishengren      -0.036   -0.0121   -0.034   -
0.0114   

Benshengren      0.114   0.0384   0.111   0.0377   

Eastern      -0.052   -0.0176   -0.051   -0.0171   

Southern      0.034   0.0114   0.032   0.0109   

Central      -0.057   -0.0191   -0.057   -
0.0194   

Notes: If the respondent is an individualistic voter, then the dep. variable=1; otherwise, the dep. 
variable=0. M.E. is the marginal probability effect. Symbols * and ** represent the p-value is smaller 
than 10% and 5%, respectively.  
The evidence above supports our proposition that wealthier and better educated voters are concerned 
more about their own economic interest. They tend to exhibit individualism, highlight personal 
functions, and believe that income inequality is one’s personal responsibility and not the government’s. 
Therefore, the government should not pursue income redistribution as a taxation policy. Since they 
generally prefer the median position in the ideological spectrum of Taiwan independence, both the DPP 
and KMT thus have no choice but to reduce taxes in order to cater to these voters. 

mailto:contact@americaserial.com
mailto:contact@americaserial.com


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

47 | P a g e  

C-3 Testing for Endogeneity. An individual’s partisanship is influenced by her preference on national 
identity (Median)or economic benefits (Individualistic), meaning that the causality might run from 
Median to partisanship, but not in the other direction. This raises an endogeneity concern, since a 
reverse causality exists between partisanship and Median(or Individualistic).To address this issue, we 
apply a two-stage probit model and use subethnic group ethnicity (Aborigine, Waishengren, 
Benshengren), resident districts (Eastern Taiwan, Southern Taiwan, Central Taiwan)as instrumental 
variables. The OIR test fails to reject the null hypothesis of no endogeneity. Since the two-stage 
estimation may lead to a loss of precision, we still stick to the estimation results presented in Tables5 
and 6 to preserve estimation efficiency.  
VI. Conclusion 
This research uses Taiwan’s democratization as the research subject to explore the relationship between 
democratic election and tax redistribution. We first explain that the policies of the KMT and DPP are 
convergent in terms of a tax cut for the rich, but divergent in the sense that both propose contrasting 
national identities. In order to explain this, our empirical study shows that, for some low-income voters, 
the salience of the national identity issue has surpassed that of the income redistribution issue such 
that national pride is associated with reduced support for redistribution. At this point, both parties have 
no choice but to reduce taxes in order to cater to rich voters. Therefore, under democracy, although the 
poor are the majority and they vote sincerely, the equilibrium tax rate is still not the tax most favorable 
to themselves .Our result thus provides some support to the Marxian argument that the poor—
especially the “religious” poor—do not pay attention to their economic interests when voting, whereas 
the wealthy tend to ignore the national identification dimension and vote based on their own interests.   
References 

Acemoglu D, S. Naidu, P. Restrepo, and J. Robinson. 2015. “Democracy, redistribution, and inequality.” 
Handbook of Income Distribution. 2: 1885–966.  

Aidt, T.S., Dutta, J., Loukoianova, E., 2006. “Democracy Comes to Europe: Franchise Expansion and 
Fiscal Outcomes 1830–1938European Economic Review 50 (2): 249–283.  

Aidt, T.S., Jensen, P.S., 2009. “Tax Structure, Size of Government, and the Extension of the Voting 
Franchise in Western Europe, 1860–1938.” International Tax and Public Finance 16 (3): 362–
394.  

Benabou, R. and J. Tirole. 2006. “Belief in a Just World and Redistributive Politics.” Quarterly Journal 
of Economics 12(2): 699–746.  

Bolton, P., and G. Roland. 1997. “The Breakup of Nations: A Political Economy Analysis.” Quarterly 
Journal of Economics 112(4): 1057-1090.  

Chen, M.T. 2010. ‘‘Five-Municipal Elections and the DPP Transition.’’ Annual Meeting and Symposium 
of the Taiwan Political Science Association, Taipei: Soochow University.  

mailto:contact@americaserial.com
mailto:contact@americaserial.com


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

48 | P a g e  

Choi, E. 2010. “Economic Voting in Taiwan: The Significance of Education and Lifetime Economic 
Experiences.” Asian Survey 50(5): 990-1010.  

De La O, A. and J. Rodden. 2008. “Does Religion Distract the Poor?: Income and Issue Voting Around 
the World.” 41(4-5): 437-476.   

Duch, R.M. 2001. “A Developmental Model of Heterogeneous Economic Voting in New Democracies.” 
American  

Political Science Review 95(4): 895-910.  

Edwards, J. 2009. Language and Identity: An Introduction. London: Cambridge University Press.  

Frank, T. 2004. What’s the Matter with Kansas? How Conservatives Won the Heart of America. New 
York: Metropolitan Books.  

Gomez, B. and J. Wilson. 2006. “Cognitive Heterogeneity and Economic Voting: A Comparative 
Analysis of Four Democratic Electorates.” American Journal of Political Science 50(1):127-145.   

Hsieh, J.F., D. Lacy, and E. Niou. 1996. “Retrospective and Prospective Voting in a One-Party-
Dominant Democracy: Taiwan’s 1996 Presidential Election,” Public Choice 97(3): 383–99.  

Hu, K.W. and T.H. Lin. 2010. ‘‘The Rejuvenation of Class Politics: Taiwan’s Democratic Consolidation 
and Class Votes, 1992–2004.’’ Friday Forum report, Academia Sinica, Institute of Sociology.  

Meltzer, A.M., and S. Richard.1981. “A Rational Theory of the Size of Government.” Journal of Political 
Economy.” 89(5): 914–927.  

Niskanen, W.A. 1997. “Autocratic, Democratic, and Optimal Government.” Economic Inquiry. 35(3): 
464–79.   

Olson, M. 1993. “Dictatorship, Democracy, and Development.” American Political Science Review. 
87:(3): 567–76.   

McGuire, M. and  M. Olson. 1996. “The Economics of Autocracy and Majority Rule: The Invisible Hand 
and the Use of Force.” Journal of Economic Literature. 34(1): 72-97.  

Piketty, T. 1995. “Social Mobility and Redistributive Politics.” Quarterly Journal of Economy 110: 
a9551–584).  

Poole, K.T., and H. Rosenthal. 1991. “Patterns of Congressional Voting.” American Journal of Political 
Science 35: 228 – 278.  

mailto:contact@americaserial.com
mailto:contact@americaserial.com


American Research Journal of Economics, Finance and Management 

Volume 10 Issue 4, October-December 2022 

ISSN: 2836-9416 

Impact Factor: 4.85 

Journal Homepage: https://americaserial.com/Journals/index.php/ARJEFM, 

Email: contact@americaserial.com 

Official Journal of America Serial Publication 

 

American Research Journal of Economics, Finance and Management 
https://americaserial.com/Journals/index.php/ARJEFM, Email: contact@americaserial.com 

49 | P a g e  

Roemer, R. 1998. “Why the Poor do not Expropriate the Rich: An Old Argument in New Garb.” Journal 
of Public Economics 70 (3):399–424.  

Scheve, K., and D. Stasavage, 2006. “Religion and Preferences for Social Insurance. Quarterly Journal 
of Political Science 1(3): 255-286.  

Shayo, M. 2009. “A Model of Social Identity with an Application to Political Economy: Nation, Class, 
and Redistribution.” American Political Science Review 103(2): 147-174.  

mailto:contact@americaserial.com
mailto:contact@americaserial.com

