TESTING THE PREDICTABILITY OF CONSUMPTION GROWTH: EVIDENCE FROM CHINA

Abstract

(ProQuest: ... denotes formulae omitted.)1. INTRODUCTIONIn his early study, Chow (1985) reported strong evidence in favor of the Permanent Income Hypothesis (PIH) using annual observations in China from 1953 to 1982. Later, Chow (2010) re-evaluated his model for the post-economic reform regime, 1978 to 2006, then provided the same conclusion.1 Using further updated data, Chow (2011) reported weaker but similar supporting evidence for the PIH in China.We believe these findings are not convincing both theoretically and empirically. Under the PIH, optimizing consumers choose a stable path of consumption over their lifetime.2 This implies that those consumers must borrow whenever their realized current income falls below their permanent income. However, in the absence of perfect capital markets, poor consumers may not be able to access to credit markets (Shoji, Aoyagi, Kasahara, Sawada, and Ueyama, 2012), which implies that liquidity constrained consumers may fail to obey the PIH.One may claim that these consumers may still resort to informal credit markets as Guirkinger (2008) points out. We view these possibilities highly improbable in the case of China. For instance, using rural household data, Yuan and Xu (2015) show that the poor in China are severely excluded from not only formal but also informal credit markets. Therefore, it seems difficult to reconcile Chow's proposed indirect evidence of the PIH with such institutional facts about accessibility to credit markets in China.From an empirical point of view, we note that his empirical analysis focuses on the coefficient of the lagged consumption in an autoregressive (AR) model for consumption. Finding that the confidence band of that coefficient includes 1, he suggested that the PIH is consistent with Chinese annual data. However, his statistical inference may not be valid when consumption obeys an integrated process, because the conventional t-test he uses is invalid when consumption obeys an I(1) process.3 Further, his work does not implement any direct statistical tests for the predictability of consumption growth in China.This paper fills this gap by directly investigating the in-sample and the out-of-sample predictability of consumption changes in China as well as the postwar US data for comparison. Our findings suggest that consumption changes are highly predictable, which provides a clear contrast with findings shown in Chow (1985, 2010, 2011).The rest of the paper is organized as follows. Section 2 provides a data description and preliminary test results. In Section 3, we provide our major findings. Section 4 concludes.2 DATA AND PRELIMINARY ANALYSISWe obtained per capita disposable income ( ) and consumption expenditure ( ) data in China from China Statistical Yearbook (2012) following Chow (1985, 2010, 2011). Observations are annual and span from 1978 to 2012. It should be noted that observations prior to 1978 are excluded, because China began their major economic reforms since 1978, and the pre-economic reform regime data are often unreliable. We deflated all observations using the GDP deflator, obtained from the same source. All data are log transformed.The log real per capita disposable income and the log real consumption expenditures of nondurable goods and services in the US are obtained from the Federal Reserve Economic Data (FRED). Observations are quarterly and cover the period from 1952:Q1 to 2011:Q4.We first implement the augmented Dickey-Fuller (ADF) test for these variables. Results are reported in Table 1. The ADF test rejects the null of nonstationarity only for differenced series with an exception of y t in China when time trend is present. Overall, our test results imply that consumption and income are integrated series, which is also consistent with Chow's work.3. EMPIRICAL FINDINGS3.1. In-Sample AnalysisCampbell and Mankiw (1990) test the validity of the PIH in the US by empirically evaluating Hall's (1978) famous claim: Consumption follows a random walk process under the PIH. …