Christopher K. Ma

The Integration of Mortgage Markets and Capital Markets

In this paper, we develop a model to predict the impact of deregulation in the form of relaxing interest rate control on the integration between the mortgage credit market and the general credit market. The model is tested through the examination of the long-term Granger-like equilibrium relationship between mortgage interest rates and general interest rates in the pre-1980 regulated vs. the post-1980 deregulated periods. It is shown that the level of regulation, in the form of targeting general interest rate levels, contributes to the segmentation of the mortgage market from the capital market. To test this model, we compare the relationship between mortgage interest rates and general interest rates around 1980 where major control on interest rate levels in capital markets was lifted. Using Engle and Grangers procedure to overcome the estimation problem from nonstationarity in the interest rate series, we are able to find that the two interest rates were cointegrated after 1980 but not before. More importantly, it appears that the two markets were already integrated before the full development of the secondary mortgage markets between 1984 and 1987. Therefore, we conclude that the bulk of the integration between the mortgage and capital markets was completed as a result of the removal of interest rate controls around 1980, in contrast with previous studies that find integration occurred during the mid-1980s primarily as a result of the rapid development of the secondary mortgage markets. Copyright American Real Estate and Urban Economics Association.

Random Walks in World Money Rates

Interest rate changes in major industrialized countries are examined and found to exhibit significant deviations from random walks. When measured over short horizons, interest rate changes demonstrate significant negative serial correlation. As the time horizon is extended, the negative dependencies decline and interest rate changes approach random walks. In general, the evidence suggests that short-term interest rate changes in major industrialized countries follow a mean-reverting process.

Mean Reversions in GNMA Returns

The random-walk hypothesis is tested in the prices of mortgage-backed securities traded in the secondary market. Using the variance ratio test, the random-walk hypothesis is rejected for the daily GNMA bond return. We identify two components in the return series: a systematic component reflecting the market pricing on the expected information, and a noise term that represents the pricing on the unexpected information. After adjusting for the impact of bid-ask spread and thin trading on the price quotations, the evidence suggests that the short-horizon, weekly realized return, being dominated by the negative serial correlation of the random component, exhibits a mean-reverting process. However, it is also found that the noise term demonstrates significant positive serial correlation for holding periods of over two weeks. Thus, for longer-term returns, the realized return exhibits positive dependence. The implication is that the price of GNMA bonds did not react to unexpected information in a rational fashion in that the adjustment process is not instantaneous. Copyright American Real Estate and Urban Economics Association.