In the fall of 2016, Korean people were gripped by a series of corruption scandals which center on a shady relationship between their president and her unknown outside advisor in partnership with the government (in particular, the Blue House). Her presidency in the past four years was found to be fraught with several cases of willful infringement on government system and democratic order. The final unraveling of these political scandals came with a nosedive in the approval rating to the 5% level, pointing to ‘de facto’ zero support. Such low rating or trust in government is ravaging enough to paralyze the operation of national affairs, even setting aside criminal charges against the president and impeachment issues.

Trust in government is a resource of great importance for government to run national affairs. Why is it so important? To answer that, it is necessary to think about the definition and operation of democracy in which regardless of the type of government, sovereignty resides in the people and the people delegate a part of their power to representatives—who could be a political party led by an individual or an alliance of political parties—so that they could operate national affairs and make necessary policy decisions on behalf of the people. At every election, people vote for a group that they think will represent best the interest of its voters. Why to use such representative democracy system may vary but probably it is because general public have no expertise or information on national affairs.

In this context, the principal-agent theory, one of classic themes in economics, can be used to look into the issue of trust in government. As aforementioned, at election, people with sovereign rights—median voters—choose the one (president and the ruling party) they think will work successfully on national affairs on their behalf while failed candidates are left to serve a role as a check on power. The latter can be referred to as potential agents who are opposition parties and provide the public with noteworthy issues such as problems in national affairs run by the ruling party. In the meantime, the ruling party keeps sending out a message to the public emphasizing the appropriateness of their way of operating national affairs.

The proportion of people in agree with the information from the ruling party means the approval rating for then government and this approval rating determines the boundaries of the ruling party’s policy stance. On the other hand, considering the conflicting relationship between the ruling and opposition parties, information from opposition parties may be of little value to the public. What affects public’s judgement most in this process is as follows: whether competent press is present who can compile information from both sides and deliver it to the public in an objective manner and whether the press’ free speech is guaranteed. In short, trust in government is certainly of great importance for a democratic country to operate national affairs, but its ambiguity in concept and definition has defied several attempts for in-depth studies from more objective and scientific perspectives.

Thus, this study conducted big data analysis on Korea and the US (comparison target) to remove such ambiguity. Results were compared to analyze the determinants to trust in government in order to seek institutional basis that could strengthen it. The big data analysis is better effective in identifying diverse discussions with respect to trust in government than the conventional interview survey consisting mostly of closed questions.

Above all, this paper gathered necessary big data by typing in ‘government’ and ‘administration’ for search in online news articles and Twitter messages in the US and Korea. Samples of online news articles were released between 1995 and 2015 and those of Twitter messages between 2009 and 2014: a total 670,000 online news articles and 8,390,000 Twitter messages in Korea’s data sets and 530,000 news articles and 57,680,000 Twitter messages in the US data sets were gathered. After removing unnecessary elements, author extracted words in forms of noun, verb, adjective, adverb and root and these are categorized into three groups (ability, benevolence and integrity) according to the trust model proposed by Mayer et al. (1995).

First, as for Korea’s online news articles, of government-related words, school-related words form a separate community, pointing to Korea’s strong attention to education. Also, there is another large community consisting of ‘the US,’ ‘China,’ ‘Japan’ and North Korea-related words, suggesting that government issues have most to with export-driven economy and North Korea-related security and diplomacy. Korea’s Twitter messages exhibit the largest community of words such as ‘Moon Jae-in,’ ‘Park Geun-hye,’ ‘Lee Myung-bak’ and ‘rigged election,’ meaning that discussions are mostly regarding presidential election. Also, Sewol ferry-related words form a community (group) which contains emotional words such as ‘incompetent,’ ‘anxious’ and ‘furious,’ hinting at public sentiment towards the government’s response to the Sewol ferry sinking.

Then, US’ online news articles exhibit the largest community of ‘tax,’ ‘benefit’ and ‘aid,’ pointing to public’s high interest in government tax programs and benefits. Also, ‘health’ and ‘care’ were commonly mentioned, suggesting that the US government’s healthcare security program is of great significance to the public health insurance system. In US’ Twitter messages, the largest community is found to contain words about diverse themes and emotion.

Next looks into the list of words used as a synonym for the three words related to trust in government—government’s ability, benevolence and integrity aforementioned as three factors affecting trust in government. First, of ability-related words, ‘disappoint’ was mentioned with high similarity to ‘furious’ and ‘angry’ in Korea’s online news articles and ‘wistful’ and ‘expected’ in its Twitter messages while to 'frustrated', 'dissatisfied' and 'discouraged' in US’s online news articles and ‘appalling,’ ‘disgusting’ and ‘shame’ in its Twitter messages. Also, ‘incompetent’ was mentioned with high similarity to ‘self-righteousness’ and ‘immorality’ in Korea’s online news articles and ‘stupidity’ in its Twitter messages while to ‘arrogance’ and ‘corruption’ in both US’ online news articles and Twitter messages.

Above findings imply that the use frequency of words related to ‘disappoint’ and ‘incompetent’ is not very different between the US and Korea. Then, ‘reliable’ was mentioned with high similarity to ‘trust’ and ‘believe’ in Korea’s Twitter messages and ‘respect,’ ‘honorable’ and ‘guardian’ in its online news articles while to ‘accessibility,’ ‘stability’ and ‘safety’ in US’s Twitter messages and ‘sophisticate’ and ‘accurate’ in its online news articles. To put it another way, something ‘reliable’ is depicted in the image of a respected and honorable guardian in Korea’s news articles while the word infers something sophisticate and accurate in US’ news articles.

According to the analysis of words related to benevolence, ‘help’ was used with high similarity to ‘energy,’ ‘driving force’ and ‘influence’ in Korea’s online news articles and ‘support,’ ‘cooperation’ and ‘collaboration’ in its Twitter messages while to ‘reward’ and ‘incentive’ in US news articles and ‘entrepreneur’ and ‘investment’ in its Twitter messages. Thus, it can be said that ‘help’ and ‘support’ are similar in meaning to government-financed support which demands no prior effort from the private sector in Korea but induces efforts from the private sector in the US.

Also, ‘the vested’ was mentioned with high similarity to ‘power to name candidate’ and ‘political power’ in Korea’s news articles and ‘extreme conservative’ and ‘conservatism’ in its Twitter messages while to ‘farm owner’ and ‘ranch owner,’ along with a few names of Republican politicians, in US news articles and ‘corruption’ and ‘egoism’ and money-related words in its Twitter messages. Such stark difference between the US and Korea is probably due to the difference in the understanding of the vested—typified by the top-down candidate recommendation system in Korea whereas by traditional subsidy given to farms or ranches in the US.

The following summarizes the results of analysis on integrity-related words. ‘Transparent’ was used with high similarity to ‘thorough’ and ‘rigid’ in Korea’s news articles and ‘clean,’ ‘righteousness’ and ‘integrity’ in its Twitter messages while to ‘efficiency,’ ‘effectiveness’ and ‘competence’ in US news articles and Twitter messages. This suggests that ‘transparent’ is seen as an antonym for ‘corruptive’ in Korea whereas people in the US associate the word with competence or efficiency through transparent competition. Also, ‘manipulate’ was used with high similarity to ‘poll manipulation’ and ‘concealment’ in Korea’s news articles and to ‘distort’ in US news, too.

The sections above are a brief examination of the big data on trust in government reflected in SNS messages in the US and Korea. The result confirms a certain gap in understanding between two countries caused by their difference in institutional or historical experiences, but most findings in two countries have a lot in common. The analysis in this paper may not be sufficiently rigorous, but it can somewhat verify the universality of the concept of trust in government. However it should be noted that this paper does not preclude the possibility that the composition of concrete contents implied in the concept may vary from country to country.

Next is about determinants to trust in government. No systemic theories have been made so far as to how trust in government is developed. Thus, existing theories show huge differences depending on themes or available resources they apply. As aforementioned, trust in government is created by the public, meaning that it will be influenced by individual’s characteristics, experience and incentive system, society and political system to which the individual belongs, the government as an evaluation target and politicians’ characteristics. Taken all together, this paper analyzed determinants to trust in government and obtained results as follows.

First, an increase in per capital GDP causes a decrease in trust in government, but when the increase exceeds a certain level, the trust is found to rise with increased per capita GDP. A country’s economic growth rate moves in positive proportion to trust in government whereas people’s educational level is in negative proportion to trust in government. This is probably because trust in government tends to strengthen with increased educational level. Disasters or incidents that result in human casualties are found to heighten trust in government, as people grow aware of the necessity of the presence of government. According to the analysis, the more transparent government becomes, the more trust in government people have. The more freedom of speech, the less trust in government.

In the domain of trust in government, Korea is among the lower ranking countries in several surveys. To understand why this happens, look at Korea’s per capita GDP which is situated at the bottom curve of its ‘U-shaped’ relationship between trust in government. Then Korea’s educational level is much higher than other countries whose per capital GDP is similar to Korea’s. Not only that, anti-corruption level is very and Korea ranks lower than advanced countries in the domain of free speech—its ranking has been on a constant decline since 1995. The analysis found the negative correlation between free speech and trust in government, but a positive correlation was observed among countries with guaranteed freedom of speech. This is because trust in government weakens first after free speech exposes corruption of the public sector, but afterwards people will have more trust in government as corruption diminishes with strengthened monitoring by the press.

According to the analysis of online news articles using SNS big data sources in Korea and the US, when it comes to the issue of trust in government, people mention words related to government’s ability, benevolence and integrity—listed in the order of frequency of use. Both countries use emotionally neutral words the most often, followed by positive and then negative words. The proportion of positive words in Korea is relatively higher than in the US. This could mean a weak freedom of speech in Korea considering its lower trust in government than the US’. Taking into account the exposure frequency of news articles, the order of use frequency shows no change in the US but government’s integrity comes first before ability and benevolence in Korea. The result implies that the most significant qualities of government are morality, integrity, anti-corruption and fairness in the eyes of the Korean press.

The findings above are about the analysis of determinants to trust in government through cross-sectional international comparison. Next is about impacts brought to trust in government in the event of social and economic changes, political scandal and incidents or natural disaster that result in human causalities. It is often the case that the exiting administration experiences a fall in trust in government and the new entering administration enjoys an increase. When the global financial crisis occurred in 2008, the use frequency of words related to benevolence and integrity decreased while that of ability-related words increased. Similar tendency is observed during MERS outbreak and North Korea torpedo attacks on Cheonan warship and shell attack on Yeonpyeong Island.

When trust in government weakens, Korean news tend to use words related to integrity and benevolence more frequently whereas US news are found to use ability-related words more often. This implies that the fall in trust in government is understood as corruption-related in Korea but as incompetency-related in the US. If the press covers more corruption-related articles than before, it may be a sign of a fall in trust in government.

In concluding this summary, it will be important to suggest policy recommendations to improve trust in government using the findings obtained. Korea’s trust in government is very low compared to its per capita GDP level. This appears to be because public is highly educated, but the levels of the government’s anti-corruption and the press’ free speech are very low. Of these issues, focus needs to be on anti-corruption and free speech. The basic principles to enhance trust in government are heightening openness, transparency and democratic control of all public organizations. Institutional improvement measures to materialize these principles include the followings: adopting a bottom-up approach to the candidate recommendation system, enhancing openness in the public service recruitment system, transforming the judicial system into a more open and independent one, ensuring a democratic control of intelligence agencies, resolving the issue of government-controlled finance, preventing control over the press and adopting competition system to government-run enterprises and promoting their privatization.

발간사
요 약

제1장 빅데이터 분석을 통한 정부신뢰의 개념 정리 및 다국가 비교분석
　제1절 연구의 배경
　제2절 연구방법
　제3절 결 과

제2장 신뢰의 종류
　제1절 신뢰대상별 연구방법론
　제2절 신뢰대상에 따른 분류

제3장 정부신뢰의 개념과 구성요소
　제1절 정부신뢰 개념의 분해
　제2절 정부신뢰의 세부 개념
　제3절 정부신뢰 설문조사의 일관성

제4장 정부신뢰 결정요인
　제1절 신뢰대상별 연구방법론
　제2절 다국가 표본을 이용한 정부신뢰 결정요인 분석
　제3절 주요 국가의 정부신뢰 결정요인 수준

제5장 빅데이터 분석을 통한 한국과 미국의 정부신뢰 개념 비교연구
　제1절 빅데이터 분석을 통해 발견된 정부신뢰 연관어
　제2절 미디어 데이터를 통해 본 한⋅미 정부신뢰 개념의 중요도 비교

제6장 빅데이터를 이용한 한국과 미국의 정부신뢰에 관한 비교연구
　제1절 정부신뢰에 관한 미디어 변수의 추이: 사회경제적 환경 및 이벤트의 미디어 효과
　제2절 정부신뢰와 정부신뢰 연관어의 미디어 노출빈도와의 상관관계

제7장 결론 및 정책적 시사점

참고문헌
부 록