SRP
TIME SERIES MODELS FOR FORECASTING NEW ONE-FAMILY HOUSES SOLD IN THE U.S.
INTRODUCTION
The housing market has been weak since its recent peak in 2005. Then, the sharp drop in the housing prices in 2007 contributed to the subprime loan crisis [1]. This dramatic change in the housing market not only affects the construction industry, but also may have a significant impact on the whole economy [3]. We are still in the midst of the housing problem with the increase in the delinquency rate and foreclosure rate.
In this paper, time-series models are specified to forecast new one-family houses sold in the U.S. Through this research, we try to predict the future development of this housing market.
THE DATA PATTERN OF NEW ONE-FAMILY HOUSES SOLD IN THE U.S.
Figure 1 shows the time-series data for the non-seasonally adjusted monthly new one-family houses sold (NHS) in the U.S. from January 1975 to January 2010, the most recent data available. The period is chosen to be long enough to reveal the cyclical fluctuations in this market. A time series is normally contain some or all of the following components, (1) trend, the long-term change in the level of the data, (2) seasonality, a regular variation in the level of the data that repeats itself at the same time every year, (3) cycles, wavelike fluctuations of the data around the trend, and (4) irregularity, the remaining and random variation in the data. From the figure, we can see that the NHS data have significant seasonality, cycles, and upward trend till 2005.
FIGURE 1
New One-Family Houses Sold (NHS), in thousands, in the U.S. January 1975 to January 2010
Data source: National Association of Realtors
In evaluating a time-series data, it is useful to analyze the autocorrelation which measures the correlation between successive observations over time. A k-period plot of autocorrelations is called an autocorrelation
INTRODUCTION
The housing market has been weak since its recent peak in 2005. Then, the sharp drop in the housing prices in 2007 contributed to the subprime loan crisis [1]. This dramatic change in the housing market not only affects the construction industry, but also may have a significant impact on the whole economy [3]. We are still in the midst of the housing problem with the increase in the delinquency rate and foreclosure rate.
In this paper, time-series models are specified to forecast new one-family houses sold in the U.S. Through this research, we try to predict the future development of this housing market.
THE DATA PATTERN OF NEW ONE-FAMILY HOUSES SOLD IN THE U.S.
Figure 1 shows the time-series data for the non-seasonally adjusted monthly new one-family houses sold (NHS) in the U.S. from January 1975 to January 2010, the most recent data available. The period is chosen to be long enough to reveal the cyclical fluctuations in this market. A time series is normally contain some or all of the following components, (1) trend, the long-term change in the level of the data, (2) seasonality, a regular variation in the level of the data that repeats itself at the same time every year, (3) cycles, wavelike fluctuations of the data around the trend, and (4) irregularity, the remaining and random variation in the data. From the figure, we can see that the NHS data have significant seasonality, cycles, and upward trend till 2005.
FIGURE 1
New One-Family Houses Sold (NHS), in thousands, in the U.S. January 1975 to January 2010
Data source: National Association of Realtors
In evaluating a time-series data, it is useful to analyze the autocorrelation which measures the correlation between successive observations over time. A k-period plot of autocorrelations is called an autocorrelation
References: [1] DiMartino, D. and Duca, J, (2007, Nov. 11). The Rise and Fall of Subprime Mortgages, Economic Letter- Insights from the Federal Reserve Bank of Dallas. [2] Forecast X 6.0. John Galt Solutions, Inc. [3] Wheelock, D. (2007) Housing Slump Could Lean Heavily on Economy. Federal Reserve Bank of St. Louis. [4]Wilson, H. and Keating, B. (2007) Business Forecasting. New York, NY: McGraw Hill/Irwin.