You are on page 1of 11

Jenny Lachira

03/24/11 Lab # 4

ABSTRACT:
The housing market is a very important indicator of the health of the economy. In the most recent crisis, this market was one of the first ones to collapse. Through the observation of data from years 1993 to 2005 it is seen that the changes on the rates of return on mortgage (mortgage yield) can determine the increase or decrease in the number of new single family houses sold. As this yield increases, the number of houses sold will decrease due to the unwillingness of buyers to pay higher interest rates on their loans. It is also helpful to observe other variables or indicators such as the Disposable Income per Capita (from the same time frame) and its positive correlation with the number of houses sold. Pre-Lab Questions What is a mortgage yield? It is the annual rate of return the holder of the bond (the financial institution which issues the mortgage or buys the mortgage from a third party in most cases smaller banks), gets on the amount of the investment throughout its life spam (the period of time it will take to repay the principal plus the accrued interest on the mortgage-backed security).

How should mortgage yields or rates affect housing? Positively?Negatively? It should have a negative relationship with the housing market in general. To explain this relationship better, let us say that if mortgage yields are high,financial institutions will be more willing to purchase these backed securities given the higher rate of return, and the issuer of the loan (mortgage) should be more confident he will get re-paid. However,to obtain a higher yield, higher interest rates are going to be applied to the loanerwhich will cause a decrease on the sales of new homes because, because generally speaking, buyers of houses don t want to pay high interest on their mortgage.

In general, how do you think the housing market has behaved over the past, say,decade?

From 2001 up to the 2008 recession, the housing market was booming. Nevertheless, this boom has been seen as the reason for the crash of the economy. Many institutions were approving loans without carefully reviewing the capacity of the buyer to repay the mortgage, and that caused the collapse of the housing market and the beginning of the recessionary period of this business cycle. Nowadays, the housing market is slowly recovering. According to the National Association of Realtors on a press release published on March 11th, the housing market showed a 5.3% increase compared to the previous year. It is important to also consider these indicators along with a recovering stock market (Dow Jones has been constantly increasing lately), and the growing confidence of the population on the recovery of the economy. ELASTICITY

Variable Yield

Estimate -234.048

Mean 6.97

Elasticity -1.82

Classification Elastic

PORTFOLIO

OLS estimates Dependent variable: Houses_Sold (1) 2529** (250.1) -234.0** (35.68) 13 0.7779 -76.34 13 0.0000 -86.68 (2) 896.9** (54.95)

const Yield n Adj. R2 lnL

Standard errors in parentheses * indicates significance at the 10 percent level ** indicates significance at the 5 percent level
Detailed Questions 1. What is the relationship between home sales and mortgage yields? Is this the relationship you expected? Plot the data as part of your analysis. Yes, the graph shows a negative relationship between Mortgage Yield and Houses Sold. Throughout years 1993 to 2005 yield has been constantly decreasing and as a consequence the number of houses sold went up.

Houses Sold versus Yield (with least squares fit) 1300 2005 Y = 2.53e+003 - 234.X 2004 1200

1100

2003

Houses Sold

1000

2002

900

2001 1998 1999

2000

1997 800 1996

700

1993

1994

1995

600 6 6.5 Mortgage Yield 7 7.5

2. - Formulate and test a hypothesis about the effect of mortgage rates on new home sales.What do you conclude? My hypothesis is that the quantity of houses sold has a negative relationship with yield. It goes up as yield goes down. My General empirical model would look like this;

Houses_Sold = H0: Ha: 0 1< 0


1=

0-

1Yield

+ Ei

Model 6: OLS, using observations 1993-2005 (T = 13) Dependent variable: Houses_Sold Coefficient 2528.78 -234.048 Std. Error 250.089 35.6762 t-ratio 10.1115 -6.5604 p-value <0.00001 0.00004

const Yield Mean dependent var Sum squared resid R-squared F(1, 11) Log-likelihood Schwarz criterion rho
Where,

*** ***

896.9231 95896.62 0.796441 43.03828 -76.33570 157.8013 0.422964

S.D. dependent var S.E. of regression Adjusted R-squared P-value(F) Akaike criterion Hannan-Quinn Durbin-Watson

198.1369 93.36956 0.777935 0.000041 156.6714 156.4392 0.791506


is the expected

is the number of houses sold when Yield is hypothetically zero; and

decrease on this number given an increase in the yield. Given the p-value is really small, we reject the null hypothesis and we can say that indeed, the yield has a negative effect on the number of houses sold.

3. Interpret the R2. R2 = 0.796 (a) What does it say about your model? It is telling me that 79.6% of the variation of the houses sold it is due to the change of mortgage yield. Given the date it could be a pretty accurate indicator to reach the conclusion of the negative relationship between houses sold and mortgage yield.

(b) How can you make the model better?What additional variables can you think of that should be included? Defend youranswer.

I would make the model better by adding the per capita disposable income per capita from the same period of time. Through this I pretend to see a positive relationship between the

number of houses sold and the change on disposable income because more disposable income means a greater purchase power.

4. Interpret the F-statistic.

P-value(F)

0.000041

Since the P-value is less than 0.05 we can reject H0 : Nave model is better. (a) What hypothesis does it test? H0 :Nave model is better Ha : Sophisticated model is better

(b) What do you conclude from it?


We conclude that the sophisticated model is better. 5. Are existing home sales elastic or inelastic with respect to the mortgage yield? They are elastic. (a) Does this make sense? Yes because if yields go up, buyers are not going to buy new homes at that moment. Instead, they will wait for prices and/or interest rate to decline.

(b) Is it what you expect? Defend your answer. Yes, it is. Due to rising interest rates and if people predict that the economy is going to go through a crisis they wouldn t buy a new house given the fear of losing their jobs and not being able to pay their mortgages. This elasticity can be also explained because whoever does not own a house yet can rent and postpone the acquisition of his/her own for better financial times. (c) Suppose there is a 1% increase in mortgage rates. How much do home sales changein percentage terms? It would decrease by 1.82%.

6. What is the practical significance of your model? In other words, what can this be usedfor, it for anything at all? How would you answer the question: "So what?" Defendyour answer. This can be used to understand when it is the best or at least the right time to buy a new house. By looking and analyzing the yield on mortgage it could be possible to draw a conclusion on whether the interest rate on mortgage is going to be high or low and how this will affect the final amount of money to be paid at the end of the contract (generally 30 years). Basically, someone will want to wait until yields are somewhat low. 7. Based on your answer to Question 3b about other variables, collect one more variableof your choice and build another model with just the one new variable. In other words,your second model will have just one independent variable other than the mortgageyield.

Model 7: OLS, using observations 1993-2005 (T = 13) Dependent variable: Houses_Sold Coefficient -220.23 0.0453938 Std. Error 95.8382 0.00384287 t-ratio -2.2979 11.8125 p-value 0.04219 <0.00001

const Dispo_income Mean dependent var Sum squared resid R-squared F(1, 11) Log-likelihood Schwarz criterion rho

** ***

896.9231 34424.54 0.926927 139.5347 -69.67644 144.4828 0.657929

S.D. dependent var S.E. of regression Adjusted R-squared P-value(F) Akaike criterion Hannan-Quinn Durbin-Watson

198.1369 55.94197 0.920284 1.37e-07 143.3529 143.1206 0.717431

(a) State your null and alternative hypotheses for this new model. H0 : Disposable Income per capita has no effect on houses sold. Ha : Disposable income per capita has a positive effect on houses sold. Houses Sold: H0 : 1= 0 H1 : 1> 0
0+ 1Disposable

Incomei + Ei

(b) Interpret the regression output for this new model. How does it compare to the first one with the mortgage yield? The p-value for the independent variable is very small; hence the null hypothesis is rejected assuming the positive effect of Disposable Income per Capita on Houses Sold. The R2 value (0.927) indicates that the variation of the number of houses sold is 92.7% related to the disposable income per capita. The F-statistic suggests that using a sophisticated model is better than using a nave one with NO x . This is new variable holds a complete opposite relationship with the one observed in the Mortgage Yield (negative relationship). However both models agree on the use of the sophisticated model as the better one.

8. Build a model portfolio and select a "best" model. Why did you select this one?

OLS estimates Dependent variable: Houses_Sold (1) 2529** (250.1) -234.0** (35.68) (2) 896.9** (54.95) (3) -220.2** (95.84)

const Yield Dispo_in come n Adj. R2 lnL

0.04539** (0.003843) 13 0.7779 -76.34 13 0.0000 -86.68 13 0.9203 -69.68

Standard errors in parentheses * indicates significance at the 10 percent level ** indicates significance at the 5 percent level
From my perspective, I would choose Model 3 because it indicates a higher relation between the independent and dependent variable. The graph also show an exponential relationship meaning that the number of houses sold (y) will change relatively to the change in disposable income per capita (x)

rather than absolutely as it would happen in a linear relationship. This model can also suggests the willingness of people to spend on long term investments such as the purchase of a new home, in the long run this can help to predict the future of the economy.

Elasticity: 1.25 Variable Dispo_income Estimate 0.0454 Mean 24610 Elasticity 1.25 Classification Elastic

1300

1200

1100

Hous s Sol

1000

900

800

700

600 20000 22000 24000 26000 28000 30000 Dispos bl Income pero Capi a

Houses Sol vs Disposable Income ( i

leas s uares fi )

CORRELATION MATRIX

Correlation coefficients, using the observations 1993 - 2005 5% critical value (two-tailed) = 0.5529 for n = 13 Yield 1.0000 Houses_Sold Dispo_income -0.8924 -0.8548 Yield 1.0000 0.9628 Houses_Sold 1.0000 Dispo_income

This is a summary of the correlation between the independent variables and the dependent variable (Houses Sold). It is observed the high positive relationship between Disposable Income per Capita and the negative relation between Yield and Houses Sold.

DATA DICTIONARY

Variable
Mortgage Yield

Value
Percentage

Source
U.S Census Bureau, Table 1178 Yield

Nemonic

New Single Family Houses Sold Disposable Income per Capita

Thousands of units

U.S. Census Bureau

Houses_Sold

Dollars

Department of Labor and Workforce Development

Dispo_income

DATA

YEAR 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Yield Houses_Sold Dispo_income (percentage) (thousands) (dollars) 7.20 666 18862 7.49 670 19550 7.87 667 20286 7.80 757 21089 7.71 804 21941 7.07 886 23163 7.04 880 23974 7.52 877 25955 7.00 908 26817 6.43 973 27816 5.80 1086 28829 5.77 1203 30309 5.94 1283 31342

SUMMARY STATISTICS

Summary Statistics, using the observations 1993 2005 Variable Yield Houses_Sold Dispo_income Variable Yield Houses_Sold Dispo_income Mean 6.97231 896.923 24610.2 Std. Dev. 0.755504 198.137 4202.34 Median 7.07000 880.000 23974.0 C.V. 0.108358 0.220907 0.170756 Minimum 5.77000 666.000 18862.0 Skewness -0.523093 0.598932 0.169486 Maximum 7.87000 1283.00 31342.0 Ex. kurtosis -1.12130 -0.595244 -1.30256

You might also like