Essays on trends and cycles in China's and emerging economies

• Main Author: Bian, Tingbin
• Other Authors: Feng Qu
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2025
• Subjects: Social Sciences; Trends and cycles; China; Emerging economies
• Online Access: https://hdl.handle.net/10356/182736
• Institution: Nanyang Technological University

In this thesis, we aim to address the important issues related to long-run potential growth and short-run fluctuations in China and emerging economies (EEs). In specific, we focus on three questions. First, when does China’s potential GDP growth start to decline? Second, does the Phillips curve exist in China's economy? Third, is the real exchange rate (RER) procyclical in EEs? To achieve this, the first step involves decomposing the output into trends and cycles. However, traditional decomposition methods, typically designed for advanced economies (AEs), do not directly apply to EEs such as China. For example, Levy (2003) and Laurenceson and Rodgers (2010) demonstrate that EEs’ and China’s macroeconomic time series exhibit fluctuations predominantly concentrated in the low-frequency band, unlike hump-shaped spectrum density in the US. Therefore, our study develops a consumption-based decomposition method based on the Permanent Income Hypothesis. First, we give an overview of the thesis in Chapter 1. Then, we apply this method to three scenarios using China's aggregate time series data, provincial-level data, and cross-country panel data across Chapters 2-4. In Chapter 2, we study when China’s potential GDP starts to decline. Determining whether the recent slowdown is due to a decline in potential growth or a cyclical trough has significant policy implications. With the consumption-based method, we decompose China’s per capita GDP into trends and cycles from 1992Q1 to 2022Q4. Firstly, our findings indicate that although the trend growth starts declining in 2019, the actual growth begins to slow down earlier, in 2013. This slowdown in actual growth coincides with a decline in external demand. Secondly, four business cycle periods are identified using the estimated cycle with Bry and Boschan’s (1971) algorithm, highlighting two special features: a prolonged cycle from 2002 to 2019 driven by external demand, and consistent underperformance of China’s actual economy relative to its potential during the 1990s. Lastly, we validate our estimations by linking them to major historical events, comparing them with alternative decomposition methods, and evaluating cycles’ performance as output gaps. In Chapter 3, we investigate the existence of the Phillips curve in China's economy with provincial-level data. If it exists, the central government could consider issuing expansionary policies to stimulate the economy, especially given the low inflation rate in 2023. Meanwhile, provincial-level data help in capturing the same long-run inflation expectations across provinces, as they are all within the same monetary union. Given the primary challenge in estimating the Phillips curve is measuring two unobservable components: the inflation expectation and the output gap. To address this, we adopt Hazell et al.’s (2022) methods to control for inflation expectations using fixed effects and the consumption-based method to measure the output gap. A consistent positive relationship between the output gap and inflation is confirmed using provincial-level data in the baseline results. In contrast, a Phillips curve with backward-looking inflation expectations and national-level data does not reveal a significant relationship. Additionally, we confirm our baseline results using different price measures, truncation years, and discount rates. In Chapter 4, we re-examine whether the RER is procyclical in EEs as documented in Rothert (2020), who use HP filter for decomposition. Identifying distinct patterns between EEs and AEs helps to differentiate macroeconomic models and pinpoint the sources of business cycle fluctuations. Canova (1998) emphasizes, different decomposition methods can produce inconsistent business cycle facts. To address this, we employ a set of decomposition methods, including the HP filter, the bandpass filter, the unobserved components model, and the consumption-based method. Additionally, we extend Rothert’s (2020) analysis by including more countries and longer sample periods. Using quarterly data from 1980Q1 to 2023Q4 for 26 EEs, 14 AEs, and 7 G7 countries, our findings reveal that both the UCM and the consumption-based method suggest the RER is either mildly cyclical or acyclical, unlike the procyclical patterns identified by filter methods, potentially supporting Harvey and Jaeger’s (1993) argument that the HP filter can produce spurious correlations. In addition to the business cycle properties of RER, we also test two other distinct business cycle patterns between EEs and AEs: 1) the relative volatility of consumption and 2) trade balance cyclicity. Our results confirm the patterns only using the filter methods but reveal substantial differences using UCM and consumption-based methods at both the country group and the individual country level.