Causality between housing returns, inflation and economic growth with endogenous breaks

Abstract

This paper investigates the housing-macroeconomic nexus in Taiwan with endogenous structural breaks during 1991–2006. GDP and CPI are taken into consideration for examining the inflation hedging ability of Taiwan's housing returns and the contribution of the housing market to economic growth. The empirical results show that the growth of GDP actually affects inflation, but it does not cause the growth in housing returns. In particular, when taking the time trend into account, it is found that the effect of inflation on housing returns is negative and the effect of housing returns on inflation is positive. This evidence demonstrates the ineffectiveness of inflation hedging of Taiwan's housing during the period of study and the opportunistic characteristic of investors. In addition, the growth of the housing market is not beneficial for economic growth in the long-run, yet it leads to higher inflation in the short-run.

Keywords:

• housing returns
• inflation hedge
• economic growth
• structural breaks
• Taiwan

JEL Classifications::

• C22
• L85
• P44

Notes

Notes

1. The Sinyi housing index is established by Sinyi Realty Company, which is presently the biggest real estate brokerage in Taiwan. Using the transaction database of existing homes in the four biggest metropolises – Taipei city, Taipei County, Taichung City and Kaohsiung City – the housing index is built up by the Hedonic Price Theory. The Sinyi housing index came into effect in 1991, being the earliest index in Taiwan, relative to two other indices, the north area housing index and the Cathay housing index. Hence, the data that are provided by Sinyi housing index not only cover the longest but are also widely referenced as representative benchmarking. The data amply reflect the variational trend of the whole of Taiwan's housing prices. Additionally, the housing price index is adjusted quarterly, based on the actual transactions of the housing market.

2. The Taiwanese housing price represented by Sinyi Realty index is in nominal terms. The nominal housing price is used in this article for the following reasons: (1) the main purpose of this study is to explore the interaction among three variables – housing price index, CPI and real per capita GDP. If the housing index is formed in real terms instead of nominal terms, the error model or co-integration might arise. Moreover, the hedging effectiveness of housing price against inflation is impossible to fully grasp; (2) the housing index being used in real terms instead of nominal terms might ignore the existence of structural breaks of housing price index and CPI, causing the possible unreality of test results.

3. Generally, when the cointegration between variables exists, it is necessary to additionally consider the error correction term of the previous period, namely the long-run cointegration vector for analysis. The empirical results show that the cointegration does not exist between variables. We, therefore, use the VAR model to test for a causal effect.

4. The possible breaks might appear during 1994–1995, 1997–2000 and 2001–2003 for the following events: beginning in 1995, the relationship between Taiwan and mainland China deteriorated; accordingly, the economy was affected due to an increase in capital flight and an increase in emigration from Taiwan. The Asian financial crisis occurred in 1997. The government intervened in the loan market in 1988 (also found in Chen et al. 2006). In 2000, as the Democratic Progressive Party took over the government from the Nationalist Party, more capital was located outside Taiwan. Taiwan joined the WTO in 2002; consequently, in order to enable Taiwan's financial institutions to compete with foreign financial institutions and to handle Taiwan's financial competitive environment, the government passed the Financial Merger Institutions Law and Financial Holding Company Act in 2000 and in 2001, permitting cross-business operation.

5. In the formula for the ZA test, (x, y, z) are scalars (individual variables). Therefore, the structural break appears in the series of one individual variable.