The Last Strike: Age, Career Incentives and Taxation in China

Alt et al. (2011) shows that the effect of electoral accountability and that of political competence in the US gubernatorial election on change in taxation is approximately 4 percentage points and 3 percentage points, respectively.

Only Wang Weilu, the party secretary of Haikou, was promoted to be the secretary of the Hainan disciplinary and inspection commission when aged 58. Wang is a special case because he was a standing committee member of the Hainan province (vice minister-level) before becoming the party secretary of Haikou.

Landry et al. (2017) observe a significant effect of fiscal extraction on career advancement at the county level; however, they find mixed evidence on the effect of fiscal extraction at the prefectural level.

We show no significant change in promotion probability for prefectural leaders who are barely older than 50 (Table A.2 in supplementary material). We also show no significant changes in the fiscal extractions in pre-2003 data (Table A.3 in supplementary material). We realize that there are approximately 30% fewer observations in the pre-2003 period than there are in the 2003–2007 period. Such missing data issues can lead to nonsignificant results. To address this concern, we pool the pre-2003 and 2003–2007 data together and conduct an RD analysis. We show no significant jump in fiscal extraction (Table A.3 in supplementary material) and some significant jumps in promotional measures (Table A.5 in supplementary material). The results suggest that pre-2003 data drives the nonsignificance of the pooled regression results, which again suggests the effect of the institutionalization of political selection in 2003.

We extend our analysis to the sample of ethnic minority regions. However, the results on the discontinuous change in promotion probability do not hold in this sub-sample. Table A.6 in supplementary material shows the result, which suggests that the governing principle is different in ethnic minority regions.

We use a sharp RD design rather than the fuzzy design for the following reasons: The age threshold reflects the dramatic change in the expected opportunity of career advancement when officials enter their last promotion-eligible term: those who are younger than 50 have at least two chances to compete for promotion; however, those older than 50 have only one chance to compete for promotion. Officials in both age groups are eligible for promotion, but those older than 50 face more intense competition. We leverage this variation to estimate the effect of career incentives.

As our running variable is discrete, the estimation requires extrapolation of the trend in the outcome apart from the threshold. However, this approach will render our results sensitive to the choice of functional form (Crost et al. 2014; Gelman and Imbens 2018). Therefore, we use multiple functional forms to enhance the robustness.

Ideally, we should use monthly-level fiscal data. However, the available fiscal data are at the annual level.

If our theory is correct, we expect a sharp effort change in fiscal extraction in a short period of time before and after the officials aged over 50 are promoted. We choose various bandwidths between 2 and 5 years, which are narrow enough to compare observations closer to the cutoff but still contain enough observations to minimize noise.

Ideally, a direct quantitative test of this competence mechanism is to show the behavior changes of officials after passing the age threshold. Following Jiang (2018), we focus on the policy priority measured by topics in a government work report. However, policy priority topics are only available from 2005 to 2007, and certain topic measures (Administration, Culture, Development, and Welfare) are not relevant to tax extraction. Because of this data limitation, we do not observe significant difference in policy priorities across the age threshold (Table A.11).

Note that the results using the probit model are suggestive and should also be interpreted with caution due to the incidental parameter problem (Greene 2002).

Financial Times Chinese: http://​www.​ftchinese.​com/​story/​001058737?​full=​y&​archive.

Xinhua News. http://​news.​xinhuanet.​com/​politics/​2014-09/​01/​c_​1112313828.​html.

People’s Daily: http://​www.​zjzzgz.​gov.​cn/​art/​2018/​9/​11/​art_​1405238_​20996240.​html.