Investigation of the time-dependent dynamics between government revenue and expenditure in China: a wavelet approach

Abstract

Unlike previous studies on causal relationships between government revenue and expenditures in China, this study takes into consideration structural breaks in the data by performing wavelet decomposition prior to testing for Granger causality between the fiscal components. The use of wavelet decomposition is motivated by economic theories, which suggest allowing for different budgetary considerations at different time horizons, as well as by the existence of special properties in the data in the form of unit roots and structural breaks. The results from the Granger causality test when using the wavelet-decomposed quarterly data over the period 1980–2015 indicate that government revenue Granger-causes government expenditure (tax-and-spend hypothesis) in the wavelet scales of two to four quarters. The results also show that bidirectional causality (fiscal synchronisation) exists in the wavelet scale of eight to sixteen quarters. Understanding the causal relationships between revenue and expenditure at different time scales is important for formulating relevant policy measures in order to maintain fiscal sustainability in China.

Keywords:

• Government revenue
• government expenditure
• fiscal sustainability
• Granger causality
• wavelet analysis

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 For example, see Bloomberg (‘China faces biggest fiscal challenge since 1981’, 20 January 2015, at http://www.cnbc.com/2015/01/20/china-faces-biggest-fiscal-challenge-since-1981.html) and ‘China Backpedals on Fiscal Reform – Facing a weakened economy, the government lets borrowing resume’, 28 May 2015, at http://www.bloomberg.com/news/articles/2015-05-28/china-s-local-government-debt-fiscal-reform-reversed).

2 Time scales in a wavelet analysis change on a dyadic basis, which means that $2^j$ denotes a scale, j. This scale is then associated with $2^{j-1}$ to $2^j$ periods. The ‘periods’ in the wavelet scales are related to the data observation frequency. For example, regarding a daily time series, scale 1 means a time scale of 1–2 days. Scales 2 and 3 of the daily series would then mean time scales of 2–4 days and 4–8 days, respectively. Similarly, dealing with monthly data in wavelet analysis makes 1–2, 2–4 and 4–8 months at the time scales of 1, 2 and 3, respectively.

3 Unlike DWT, MODWT has a number of values for the averages at every scale level equal to the number of values in the original series, which is a useful property for our analysis. The wavelet transformation for MODWT is not an orthogonal one, however (Percival and Walden 2000, Ch. 5).

4 Data codes from the database are in the parentheses.

5 The difference between ADF and PP tests should be noted here. While the PP test does not require specific forms of the serial correlations in the data generation under the null hypothesis, the ADF test includes additional higher order lagged terms. This means that, if the autoregressive order is not correctly specified, the test will be either mis-sized or its power will suffer. This problem can be avoided in the PP test. Yet, if the autoregressive order is correctly specified, the PP test will be less powerful than the ADF test (Harris and Sollis 2003).

6 The lag length was decided using the AIC.

7 The R Waveslim package is found at: https://cran.r-project.org/web/packages/waveslim/index.html (2016-07-09).

Additional information

Notes on contributors

Hyunjoo Kim Karlsson

Hyunjoo Kim Karlsson is a senior lecturer in economics at Linnaeus University in Sweden. She received her doctoral degree in economics at Jönköping International Business School, Jönköping University, Sweden, in 2012. Her major area of research is in international economics and finance with a focus on emerging markets. Her recent publications are as below:

- Karlsson, H. K., K. Månsson, and P. Sjölander. 2018. “Investigation of the Nonlinear Behavior in Real Exchange Rates in Developing Regions.” Applied Economics Letters 25 (5): 335–339.

- Karlsson, H. K., Y. Li, and G. Shukur. 2018. “The Causal Nexus between Oil Prices, Interest Rates, and Unemployment in Norway Using Wavelet Methods.” Sustainability 10: 1–15.

- Karlsson, H. K., P. Karlsson, K. Månsson, and S. Pär. 2016. “Wavelet Quantile Analysis of Asymmetric Pricing on the Swedish Power Market.” Empirica 44: 1–12.

- Hacker, R. S., H. K. Karlsson, and K. Månsson. 2014. “The Relationship between Exchange Rates and Interest Rate Differential―A Wavelet Approach.” International Review of Economics and Finance.

- Karlsson, H. K., and S. R. Hacker. 2013. “Time-Varying Sensitivities of Sectoral Returns to Market Returns and Exchange Rate Movements.” Applied Financial Economics 23 (14): 1155–1168.

- Hacker, R. S., H. K. Karlsson, and K. Månsson. 2012. “The Relationship between Exchange Rate and Interest Rate Differential: A Wavelet Approach.” The World Economy 35 (9): 1162–1185.