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