Government behavior and trust: the case of China.

• Author: Yang, Peihong
• Position: Report

Social capital has become a critical term in the social sciences since Loury (1977) and Coleman's (1988) seminal studies. Coleman (1990) and Putnan, Leonardi, and Nanetti (1993) focus on the positive spillover effect of social capital. Fukuyama (1997) argues that only certain shared norms and values can be regarded as social capital. Putnan (2000), Ostrom (2000), and Bowles and Gintis (2002) highlight the network effect of social capital. All these studies demonstrate that trust is central to social capital.

Fafchamps (2004) argues that trust may be understood as an optimistic expectation or belief regarding other agents' behavior. The origin of trust, however, may wiry. Durkheim (2000) argues that trust comes from family ties. Platteau (1994a, 1994b) argues that it arises from general knowledge about the population of agents, the incentives they face, and the upbringing they have received. The former can be called personalized trust and the latter generalized trust. Glaeser et al. (2000) employ economic experiments to see how attitudes and background characteristics influence the choice of strategies.

Researchers are obsessed with the term social capital, even if it is very elusive, and contend that it is an important determinant of economic development. Arrow (1972) and Fukuyama (1995) have argued that the level of trust in a society strongly influences its economic performance. Knack and Keefer (1997) find that a one standard deviation increase in a measure of country-level trust increases economic growth by more than 0.5 standard deviation. Putnam, Leonardi, and Nanetti (1993) use the social capital level difference to explain the development gap between Northern and Southern Italy. LaPorta et al. (1997) find that across countries, a one standard deviation increase in the same measure of trust increases judicial efficiency by 0.7 of a standard deviation and reduces government corruption by 0.3 of a standard deviation. In most situations, trusting others enables economic agents to operate more efficiently--for example, by invoicing for goods they have delivered or by agreeing to stop hostilities. Whenever this is the case, generalized trust yields more efficient outcomes than personalized trust (Durlauf and Fafchamps 2004).

These studies, though interesting and somewhat persuasive, do not separate trust in government from trust in the general public. In modern society, government plays a prominent role in shaping everyday life. People's behavior will inevitably be influenced by government behavior. It is therefore inappropriate to define trust or social capital without looking at the interplay between government behavior and trust. This article views trust from an institutional perspective and examines the interplay between government behavior and trust across various regions in China. One of the key findings from the sample surveys is that positive and negative government behavior have a significant impact on the level of social trust: good government increases social trust while bad government diminishes trust.

Data Description

From 2006 to 2007, the Unirule Institute of Economics distributed 3,300 questionnaires to investigate the attitudes and feelings of households toward government in each provincial capital city of China. (1) To understand the trust levels in Chinese society and to explain its determinants, this article divides trust into two categories: trust in government and trust in people. Survey questions were based on the Likert scale, which is a commonly used scale to measure respondents' attitudes (see Kapes, Mastie, and Whitfield 1994). Table 1 describes the various indicators of trust used in the sample survey.

The reliability of the scale has an impact on the survey results and should be examined. There are a variety of ways in which reliability can be assessed (see De Vellis 1991, Carmines and Zeller 1979). The most commonly employed method is the use of Cronbach's alpha, which measures the proportion of scale variance that is communal, resulting from covariation among the items in the sample survey. Since the construct is presumed to cause each of the item scores, "good" items are positively correlated and alpha should be "high." If the items in the scale were completely orthogonal, scale variance would equal the stun of the individual item variances and alpha would take a value of zero (the lower bound). The upper bound for alpha approaches one, with values above 0.7 generally accepted as demonstrating that a scale is internally consistent or reliable (Nunnally 1978). The aim of scale purification is to obtain a high alpha, which implies a reliable scale. However, while elimination of an item with a low item-total correlation raises alpha, fewer items in a scale reduces alpha. Table 2 presents Cronbach's alpha for each of the questions/ items used in the sample survey. From an examination of Table 2, we find that the data gathered from the survey is reliable because it falls in the interval (0.8, 0.9), an interval deemed to be very reliable.

In order to examine the trust levels of Chinese society quantitatively, the questions are converted into scores from 100 (highest trust level) to 0 (lowest trust level). All components of each indicator are given equal weights because each component is important and valuable. Table 3 reports the summary statistics of trust in government and trust in people. As far as trust in government is concerned, Hangzhou, the capital city of Zhejiang province (one of the most market-oriented regions in China) has the highest score (69.21), while Shenyang (68.93) is second and Shanghai (67.76) third. Trust in...