Jump to ContentJump to Main Navigation
Varieties of Governance in ChinaMigration and Institutional Change in Chinese Villages$

Jie Lu

Print publication date: 2014

Print ISBN-13: 9780199378746

Published to Oxford Scholarship Online: October 2014

DOI: 10.1093/acprof:oso/9780199378746.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2017. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 27 February 2017

(p.210) Appendix 2 Multivariate Probit Regression (MPR)

(p.210) Appendix 2 Multivariate Probit Regression (MPR)

Source:
Varieties of Governance in China
Publisher:
Oxford University Press

As previously discussed, the provision of specific public goods is likely to be under the same budget and other socioeconomic constraints. Statistically, the error terms are not independent across the regression equations for specific public goods. To accommodate and model these correlated error terms, I follow the literature on public goods provision (L. Tsai 2007a, b) and adopt the more appropriate approach of seemingly unrelated regression (SUR) (Greene 2008; Zellner 1962, Chapter 14). The conventional SUR approach can deal only with continuous dependent variables, whereas all dependent variables here are dichotomous and, thus, cannot be analyzed with conventional SUR. Multivariate probit regression (MPR) is the appropriate and most efficient statistical modeling choice (Cappelari and Jenkins 2003).

As demonstrated by Greene (2008, pp. 817–826) and Pindyck and Rubinfeld (1998), MPR’s statistical structure is similar to that of bivariate probit regression (BPR). The only difference is that more than two dichotomous dependent variables are simultaneously estimated. This significantly increases the challenge for estimation and calls for more efficient algorithms and simulation methods. To effectively accommodate such statistical challenges, I follow the strategy of MPR and implement statistical analysis in the framework of structural equation modeling (SEM). More specifically, I specify five simultaneous equations for tap water, within-village paved roads, maintenance of irrigation projects, maintenance of other public facilities, and provision of other public welfare, respectively, and specify the error terms of all five dichotomous dependent variables to be correlated with each other as free parameters for simultaneous estimation. (p.211)

Table A2.1. Summary Statistics of Key Village-Level Variables

Variable Name

Mean

Mode

Std. Dev.

Max

Min

Total population

2215.46

1542.78

10,877

118

Total migrant workers

230.6

296.02

1946

1

Percentage of migrant workers

0.096

0.105

0.578

0.0001

Annual per capita income

3469.65

5048.35

86,000

200

Distance to township seat

6.12

6.27

65

0

Land per capita

1.38

1.2

7.58

0.057

Model village in self-governance

0

1

0

Public space in village

0

1

0

Rice as the key agricultural product

0

1

0

Source: 2008 National Village Survey (N = 356).