Sign up

Modeling mortgage refinancing decision.
Abstract:
We investigate whether a decision to refinance or not is influenced by socio-economic characteristics of different borrower groups. These characteristics may include immobility and lack of financial knowledge and information. We use the zip-codes of household borrowers as proxy for these socio-economic characteristics for loans originated in Los Angeles County. Our finding suggests that there is a relationship between socio-economic characteristics of borrowers and their refinancing decisions. Borrowers in certain geographic locations tend to miss significantly more refinancing opportunities, as defined by the interest rate difference between the original loan and market rates.

Keywords: refinancing decisions, transaction costs, socio-economic factors

Article Type:
Report
Subject:
Mortgages (Analysis)
Mortgages (Refinancing)
Authors:
Kim, Taewon
Loo, Jean
Wen, Min Ming
Pub Date:
12/01/2011
Publication:
Name: Journal of International Business and Economics Publisher: International Academy of Business and Economics Audience: Academic Format: Magazine/Journal Subject: Business, international; Computers Copyright: COPYRIGHT 2011 International Academy of Business and Economics ISSN: 1544-8037
Issue:
Date: Dec, 2011 Source Volume: 11 Source Issue: 4
Topic:
Computer Subject: Mortgages
Product:
Product Code: E565200 Mortgages
Geographic:
Geographic Scope: California Geographic Code: 1U9CA California
Accession Number:
272739732
Full Text:
1. INTRODUCTION

Refinancing is typically studied as a value-maximizing optimal decision based on financial variables such as the difference between original and current loan rates, the size of the loan and transaction costs. As such, refinancing is the result of an optimal financial decision that depends on interest rate fluctuations. A rational borrower would exercise the embedded call option to refinance when the market value of the mortgage exceeds the nominal balance plus any transaction costs.

We assume that borrowers are wealth-maximizing owner-occupants. They would choose one of the following alternatives at each period, to continue with the current mortgage, default or refinance. A decline in market and hence mortgage interest rates tends to prompt mortgage prepayment and refinancing. However, in the mortgage data that we use in this study, we observe that even when the decline in market interest rates is significant, refinancing behavior is quite varied within the areas served by same lending institutions.

Our research accords with the existing literature that the larger the loan amount and the higher the LTV, the higher the likelihood that the households will refinance. However, economic theory would also suggest that, the larger is the spread between the original loan rates and new market interest rates, the larger the incentive household borrowers would have to refinance. In our data in this study though, many loans were not refinanced when that spread was significantly large. We then conjecture that for these borrowers either the refinancing costs were so high as to outweigh the benefits of such refinancing, or that for some of them at least, the information about refinancing might not have been sufficiently available which would mean an ultimately high transaction cost, thereby leading them to forgo refinancing.

In this paper, we examine the explicit factors that might have contributed to these higher transaction costs, including points. The merit of this paper is that we also include an implicit factor, namely the socioeconomic status of borrowers. To this end, we use borrowers' house location as a proxy variable for their socio-economic status. That is, we use households' residential districts in the dataset as proxies for borrowers' transaction costs associated with refinancing. Our data consists of a total of 22,237 loans originated by the Bank of America in Los Angeles County between 1970 and 2002.

This way, we are able to explain a seemingly contradictory observation that high spreads between the original loan and market rates do not always trigger refinancing. In our dataset, the districts that on average had the highest original loan rates also had the lowest loan amounts along with the highest origination points. High points traditionally mean high transaction costs. In other words, the transaction costs of refinancing in certain districts seem so high as to outweigh the benefits of refinancing precisely when the interest rate reduction from refinancing would seem to be the greatest, all other things held equal. In this study, these districts are represented by the socio-economic factors that are associated with high transaction costs such as lack of financial information or lack of mobility.

The paper is organized as follows. Section 2 reviews previous research on mortgage prepayment and refinancing. Section 3 presents a refinancing model that recognizes implicit transaction costs associated with each district. Section 4 describes the data and methods used to test the model along with empirical results. Concluding remarks follow in Section 5.

2. LITERATURE REVIEW

Refinancing and default have largely been studied as an optimal financial decision based on interest rate fluctuations (Green and Shoven, 1986, Chen and Ling, 1989). A rational borrower would exercise the embedded call option to refinance when the market value of the mortgage exceeds the nominal balance plus any transaction costs (Kau and Kim, 1993).

At the same time, a rational house owner would default when the value of the house falls below the value of the mortgage balance (Kau and Kim, 1994). However, not every borrower is known to exercise this embedded default option even when it seems financially rational, possibly because default means more than just the loss of the house including a long-lasting damage to one's credit. Instead, default is more often motivated by "trigger events" such as divorce or unemployment (Deng, Quigley and Van Order, 2000).

Most option theoretic pricing models of prepayment and refinancing abstract from transaction costs (Cox, Ingersoll and Ross, 1985, Kau and Keenan, 1995). Accordingly, Virmani and Murphy (2010) argue that the spread of 1 % between the existing loan and refinancing rates works as well as any option pricing theoretic model in predicting refinancing behavior. Yet, it is also commonly observed that borrowers do not exercise their option to refinance, even though refinancing conditions are met (Giliberto and Thibodeau, 1989), which can only be explained by significant transaction costs.

Many studies in the literature use variables such as years lived in the current house or the age of the household as proxy variables to represent these transaction costs. The idea is that the longer the tenure in the same house or the older the age of the household, the less mobile the borrowers are, hence the higher the transaction costs (Clapp, Deng and An, 2004). Quigley (1987) includes household mobility as a factor influencing prepayment and refinancing by arguing that households' attachment to neighborhood increases over time.

Common to all these studies is that refinancing depends more than anything on the difference between the mortgage's nominal balance and its market value. We also employ financial variables such as the LTV and original loan amount in accordance with the existing literature. Our contribution to the literature is that we specifically focus on the geographical districts of household borrowers to represent different socio-economic factors as a distinguishing explanatory variable in refinancing behavior.

3. MODEL and HYPOTHESES

In our model, we set up a threshold at three levels, 150, 200 and 250 basis points respectively to represent the spread between the original loan rates and new market rates. Thus, our regression model is as follows in equation (1).

Count N = [beta]0+[beta]1D1t +[beta]2D2t +[beta]3D3t +[beta]4D4t +[beta]5D5t + [beta]6LTVit + [beta]7Pointit +[beta]8Rateit + [beta]9Amountit +[epsilon]it

(1)

Count N is defined as the number of refinancing opportunities missed by an individual loan i, when the spread of the original rate of an individual loan and the market rate is more than N basis points. In this model, three variables, Count 150, Count200 and Count250 measure the number of times that each mortgagor misses refinancing, given that the original loan rate exceeds the subsequently observed monthly market rates by 150, 200 and 250 basis points respectively. Each loan is ultimately either refinanced, remains active until it reaches the end of the observation period or defaults. The monthly 10-year Treasury yield is used as the market rate at each time. D1, D2, D3, D4 and D5 are the dummy variables to designate the five supervisorial districts of the Los Angeles County where the loan borrowers' houses are located.

Refinancing comes with the benefit of a reduction in net interest payments even as they are also accompanied by often significant transaction costs. In this study, we investigate whether these transaction costs, especially the implicit kind, are the determining factors in certain non-refinancing cases. In other words, we utilize the common economic theory that value maximizing homeowners with existing mortgage loans would choose to forgo a refinancing opportunity only and only if transaction costs outweigh the benefits from refinancing, assuming that they have access to such information.

Thus in our model, if a loan is not refinanced at the respective spread level of 150, 200 or 250 basis points, it is considered a refinancing opportunity missed. Our hypothesis is that an economic or social status may play an important role in refinancing decisions. Los Angeles County has diverse economic and ethnic populations. Borrowers in certain geographic locations tend to have a long history of residence in the same areas and hence less mobile (Quigley, 1987). It is also reasonable to believe that lower income groups have less access to financial information. Our data enables us to investigate a possible link between various population groups and their refinancing decisions.

Of all the transaction costs in refinancing, the points paid are among the major factors discouraging a refinancing application. As a result, all else equal, we expect a positive relationship between these point costs and the number of refinancing opportunities missed.

The economic theory suggests that larger original loans are more likely to be refinanced as a drop in mortgage rates would result in bigger savings for them. At the same time, these loans with a high balance would also incur high transaction costs in terms of point fees as they are charged proportionally to the amount to be refinanced. Consequently, whether larger loans are more or less likely to be refinanced would depend on whether the savings would outweigh the cost and vice versa. In other words, a priori, it is not clear whether the loans with larger balances would have a positive or negative relationship with the number of refinancing opportunities missed as defined here. A similar argument would apply to the loan to value ratio variable.

With respect to the original mortgage rate variable, one would expect that a higher original loan rate would go hand in hand with fewer refinancing opportunities missed, considering their potentially bigger interest payment savings. Thus, one would expect a negative sign. However, in general, a higher original rate, especially on a small loan amount, may just be a reflection of the poor credit quality of the borrower, in which case, a significant amount of refinancing cost is expected on that loan. This would result in a positive relationship between the original rate and the number of missing refinancing opportunities. For loans with higher original rates then, the relationship between the rate and the number of refinancing opportunities missed is not certain, and instead, once again, it would depend on the degree of trade-offs between the costs and benefits of refinancing.

4. DATA and EMPIRICAL RESULTS

The loans included in our sample are the mortgages originated by the Bank of America between 1970 and 2002 with the county FIPS (Federal Information Processing Standard) code of 6037, i.e. in Los Angeles County. There are a total of 22,237 loans. Our study is based on the observation period between 1986 and 2004 during which a large number of 15,218 loans were refinanced. This represents 68.44 percent of all loans. At the end of the observation period in this study, 6,524 (or 29.34%) remained still active while 495 (or 2.23%) ended in default.

Each loan in the data is identified by a dummy variable that represents one of the five supervisorial districts of the Los Angeles County. These districts are drawn in such a way that each has roughly two million people. The 1st and 2nd districts are in the heart of the Los Angeles County; the 1st district is known to be majority Latino, and the 2nd district has a large African American population. The 3rd district includes many coastal cities and the 4th district covers the southern most cities, with the 5th district comprising the northern most cities of the County. We investigate the impact of implicit costs on refinancing using the house district information of each mortgage borrower as a proxy for such implicit costs which naturally can vary from one supervisorial district to another, given the varied social status of the average resident in it. Thus we test whether borrowers in different geographic locations may arrive at significantly different refinancing decisions, even as they face the same market rate fluctuations. Implicit costs would include limited access to financial information and financing activities that are linked to socio- economic advantages or disadvantages as it is reasonable to believe that differences in implicit refinancing costs among five supervisorial districts may explain variation in refinancing decisions.

Table 1 above summarizes the descriptive statistics of the variables used in this study. The average mortgage borrower in the Los Angeles County from Bank of America during 1970 to 2002 paid 0.13% in points, had a loan-to-value ratio of 72.32%, the loan amount of $140,387 with the original mortgage rate of 8.05%. The average number of times mortgagors decided not to refinance when original mortgage rates exceeded subsequent monthly market rates by 150, 200 and 250 basis points is 49, 38 and 27, respectively. The refinanced group paid a lower average point, had a higher loan amount as well as a higher loan rate than average. The active group (borrowers who did not refinance) paid the highest point. The active group also had a lower than average loan-to-value ratio, lower than the average loan amount as well as the lowest loan rate. It is not surprising that the group that defaulted had the highest loan-to-value ratio and paid the highest loan rate. The refinanced group missed refinancing opportunities the least, and the active group the most, while the group that defaulted fell between those of the refinanced and the active.

All loans in the sample are divided into five supervisorial districts based on the zip code of each loan. Table 2 shows the means of the variables of all the loans in each district. It shows that there were fewer loans in the 1st and 2nd districts than in the 3rd, 4th and 5th districts. Loans in the 1st and 2nd districts, on average, paid higher points, had higher loan-to-value ratios, lower loan amounts, and higher

mortgage rates than the loans in the 3rd, 4th, and 5th districts. Most importantly for our purpose, loans in the 1st and 2nd districts missed more opportunities to refinance than the loans in the 3rd, 4th and 5th districts.

Tables 3 provides the descriptive information of loans grouped into the three groups of refinanced, active and defaulted respectively for each of the five supervisorial districts. The 1st and 2nd districts had approximately 60% of the loan refinanced while the 3rd, 4th and 5th districts had about 70%. Among the refinanced group, higher points were paid by the loans in the 1st and 2nd districts and lower points were paid by loans in the 3rd, 4th and 5th districts. The average loan amount was also lower in the 1st and 2nd districts than that in the 3rd, 4th and 5th districts. In addition, the average number of times refinancing was missed was higher in the 1st and 2nd districts than in the 3rd, 4th and 5th districts. Among the refinanced groups, the loan-to-value ratio and mortgage rate did not differ much across the five districts.

The percentage of loans that remained active was significantly higher in the 1st and 2nd districts than in the 3rd, 4th and 5th districts. The active loans in the 1st and 2nd districts paid significantly higher points, had significantly higher loan-to-value ratios, and had somewhat higher loan rates and lower loan amounts than those of loans in the 3rd, 4th and 5th districts. The average number of times that refinancing was missed for these active loans were somewhat higher for loans in the 1st and 2nd districts than for loans in the 3rd, 4th and 5th districts.

The amount of defaulted loans accounted for about 2% of all loans in each district. Defaulted loans in the 1st and 2nd districts paid significantly lower points, and had higher loan-to-value ratios, lower loan amounts and somewhat lower loan rates than the loans in the 3rd, 4th and 5th districts. The number of refinancing opportunities missed was significantly lower for defaulted loans in the 1st and 2nd districts than the loans that defaulted in the 3rd, 4th and 5th districts, indicating that they defaulted at the earliest opportunity possible. It is possible that these borrowers defaulted prior to any financially viable refinancing opportunities became available.

As shown in Table 4, the regression results do indeed validate our hypothesis. The coefficients on loan origination points are positive and significant for all three levels of 150, 200 and 250 basis point spreads. It suggests that higher origination points represented higher transaction costs, thus resulting in more misses of refinancing opportunities. Naturally, the bigger the spread, the bigger is the explanatory power [R.sup.2].

The significantly negative coefficients for LTV and the original loan amounts are consistent with the hypothesis that loans with higher LTV and higher loan amounts would miss fewer refinancing opportunities due to the bigger savings potential. Here again, in accordance with economic theory, the fewer opportunities were missed, the bigger the spread was. With the original mortgage rate variable, the conclusion is not a clear cut. This reflects the fact that although the borrowers of higher original loan amounts would have had more incentives to refinance because of bigger potential savings in interest payments, those high rates probably were reflections of those borrowers' unfavorable credit rating and hence their lack of ability to refinance.

As for individual district variables, the positive and significant coefficients for the 1st and 2nd districts suggest that loans in these two districts missed more refinancing opportunities, as could be predicted. In contrast, the coefficients for the 3rd, 4th and the 5th districts were significantly negative. Loo et al. (2010) hypothesize that the geographic location might prove to be a statistically meaningful proxy for transaction costs that might prevent borrowers from prepaying and refinancing loans. Indeed, our results in this study vindicate that hypothesis. The loans in the 1st and 2nd districts missed refinancing opportunities significantly more times, when faced with the same market interest rate fluctuations, compared to their more affluent neighbors in the 3rd, 4th and the 5th districts.

5. CONCLUSION

In this paper, we compare refinancing decisions made by over 22,000 Bank of America mortgage borrowers in Los Angeles. We assign each loan a dummy variable to represent the loan's geographic location in one of the five supervisorial districts of the county. The 1st and 2nd districts are in the heart of the county and have long been populated by a majority Latino and African American population. The 3rd, 4th and 5th districts cover the coastal cities and the northern and southern most areas of the county with larger house values. We count the number of times each loan missed a refinancing opportunity when the spread between its original loan and the new market rates exceeds 150, 200 and 250 basis points respectively. Indeed, the average number of refinancing opportunities missed is higher in the first two districts than in the other districts both among the loans that were eventually refinanced and the loans that remained active.

Economic theory suggests that the more the borrowers have to gain from refinancing with lower interest rate loans, the more likely they would refinance. However, our results show that refinancing decisions are not always financially optimal and that high spreads between the original loan and new market rates do not always trigger refinancing. We explain this seemingly contradictory phenomenon by employing borrower groups' socio-economic characteristics which might trigger uneven implicit refinancing costs among them, such as accessibility to financing information and credit availability. By using the borrowers' zip-codes as a proxy variable to explain implicit refinancing costs, we find that loans in affluent districts indeed missed significantly fewer refinancing opportunities than in other districts even though high explicit refinancing costs alone would not justify such refinancing decisions.

REFERENCES:

Chen, A. and D. Ling, "Optimal Mortgage Refinancing with Stochastic InterestRates", AREUEA Journal, 17, 278-299, 1989

Clapp, J. M., Y. Deng and X. An "Alternative Models for Competing Risks of Mortgage Termination", USC Lusk Center for Real Estate Working Paper No. 2004-1002, 2004

Cox, J.C., J.E. Ingersoll and S. A. Ross, "An Intertemporal General Equilibrium Model of Asset Prices," Econometrica53, 2, 363-384, 1985

Deng, Y., J. M. Quigley and R. Van Order, "Mortgage Terminations, Heterogeneity and the Exercise of Mortgage Options", Econometrica8, 2, 275-307, 2000

Gilberto, S.M. and T.G. Thibodeau, "Modeling Conventional Residential Mortgage Refinancing", Journal of Real Estate Finance and Economics, 2, 285-299, 1989

Green, J., and J.B. Shoven, "The Effect of Interest Rates on Mortgage Prepayment," the Journal of Money, Credit and Banking, 18, 41-59, 1986

Kau, J.B. and D.C. Keenan, "An Overview of the Option Theoretic Pricing of Mortgages," the Journal of Housing Research, 6, 2, 217-244, 1995

Kau, J.B. and T. Kim, "The Timing of Prepayment: A Theoretical Analysis," the Journal of Real Estate Finance and Economics, 7, 3, 221-28, 1993

Kau, J.B. and T. Kim, "Waiting to Default: The Value of Delay", the Journal of the American Real Estate and Urban Economics Association, 22, 3, 539-51, 1994

Loo, Jean, Taewon Kim and Min-Ming Wen, "Mortgage Termination in Los Angeles County", International Journal of Business Research 10, 3, 155-160, 2010

Quigley, J.M. "Interest Rate Variations, Mortgage Prepayments and Household Mobility", the Review of Economics and Statistics, 64, 4, 636-643, 1987

Virmani, S. and A. Murphy, "An Empirical Analysis of Residential Mortgage Refinancing Decision-Making", the Journal of Housing Research, 19, 2, 129-138, 2010

Taewon Kim, California State University-Los Angeles, Los Angeles, CA, USA

Jean Loo, California State University-Los Angeles, Los Angeles, CA, USA

Min Ming Wen, California State University-Los Angeles, Los Angeles, CA, USA

Dr. Taewon Kim earned her Ph.D. at the University of Georgia in 1986. Currently she is a professor of Finance and Real Estate at California State University Los Angeles.

Dr. Jean Loo earned her Ph.D. at the Ohio State University in 1984. Currently she is a professor of Finance at California State University Los Angeles.

Dr. Min-Ming Wen earned her Ph.D. at the University of Connecticut in 2004. Currently she is an assistant professor of Finance at California State University Los Angeles.
Table 1: Sample Mean of Points, LTV, Loan Amount, Original Rate,
Count150, Count200 and Count250

Los Angeles County     All Loans   Refinanced   Active      Defaulted

Number of loans        22,237      15,218       6,524       495
(Number of
  Observations in %)   100.00%     68.44%       29.34       2.23%
Points (%)             0.13        0.11         0.18        0.11
Loan-to-value
  (LTV) (%)            72.32       72.68        70.62       83.82
Loan Amount            $140,387    $146,912     $125,248    $139,297
Original Mortgage
  Rate (%)             8.05        8.22         7.55        9.17
Count 150              49          39           73          58
Count 200              38          29           58          49
Count 250              27          20           43          38

Table 2: Sample Mean of Points, LTV, Loan Amount, Original Rate,
Count 150, Count200 and Count250 for Loans in Five Supervisorial
Districts

                             1st District   2nd District   3rd District

Number of Loans              4756           4396           5416
Points (%)                   0.15           0.21           0.05
Loan-to-value (%)            73.30          73.72          71.39
Loan Amount                  123,008        135,202        152,310
Original Mortgage Rate (%)   8.07           8.06           8.04
Count 150                    52             53             47
Count 200                    41             42             36
Count 250                    30             30             25

                             4th District   5th District

Number of Loans              5379           7266
Points (%)                   0.17           0.08
Loan-to-value (%)            70.80          72.60
Loan Amount                  149,837        140,654
Original Mortgage Rate (%)   8.01           8.03
Count 150                    48             47
Count 200                    37             36
Count 250                    25             25

Table 3: Sample Mean of Points, LTV, Loan Amount, Original Rate,
Count150, Count200 and Count250 for All, Refinanced, Defaulted and
Active Loans in All Five Supervisorial Districts

1st District           All Loans    Refinanced     Active    Defaulted

Number of Loans             4756          2951       1696          109
Percentage                  100%           62%        36%           2%
Points                      0.15          0.13       0.18        -0.01
Loan-to-value (%)           73.3          73.5      72.16        85.42
Loan Amount              123,008       127,128    115,199      132,992
Original Mortgage
  Rate (%)                  8.07           8.3        7.6         8.93
Count150                      52            40         73           51
Count200                      41            31         59           42
Count250                      30            21         44           31

2nd District           All Loans    Refinanced     Active    Defaulted

Number of Loans             4396          2678       1594          124
Percentage of Loans         100%           61%        36%           3%
Points                      0.21           0.2       0.24         0.06
Loan-to-value (%)          73.72          73.1      73.96        84.09
Loan Amount              135,203       142,728    122,605      134,587
Original Mortgage
  Rate (%)                  8.06          8.29        7.6         8.89
Count150                      53            40         73           53
Count200                      42            31         59           44
Count250                      31            21         46           34

3rd District           All Loans    Refinanced     Active    Defaulted

Number of Loans             5416          3964       1327          125
Percentage of Loans         100%           73%        25%           2%
Points (%)                  0.05          0.03       0.14         0.08
Loan-to-value (%)          71.39         72.24      67.84        81.84
Loan Amount              152,310       158,654    133,032      155,787
Original Mortgage
  Rate (%)                  8.04          8.18       7.49          9.4
Count150                      47            38         72           65
Count200                      36            28         57           57
Count250                      25            19         42           46

4th District           All Loans    Refinanced     Active    Defaulted

Number of Loans             5379          3876       1402          101
Percentage of Loans         100%           72%        26%           2%
Percentage of Loans         100%           72%        26%           2%
Points (%)                  0.17          0.15       0.23         0.26
Loan-to-value (%)           70.8         71.52      67.95        82.54
Loan Amount              149,837       155,282    135,400      141,289
Original Mortgage           8.02          8.17       7.51         9.16
  Rate (%)
Count150                      48            39         72           59
Count200                      37            29         57           50
Count250                      25            19         42           38

5th District           All Loans    Refinanced     Active    Defaulted

Number of Loans             7266          5167       1941          158
Percentage of Loans         100%           71%        27%           2%
Points                      0.08          0.06       0.13         0.09
Loan-to-value (%)           72.6         73.31      69.65        85.48
Loan Amount              140,654       146,958    124,571      132,090
Original Mortgage           8.03          8.19        7.5         9.27
  Rate (%)
Count150                      47            38         71           59
Count200                      36            28         56           50
Count250                      25            19         41           39

Table 4: Regression Results

                       Count_150                    Count_200

variables              Coefficient       t-value    Coefficient

Points                    2.34773 ***    9.38          1.24018 ***
LTV                      -0.16488 ***    -12.98       -0.13449 ***
Original Loan Amount   -0.0000672 ***    -20.94     -0.0000479 ***
Original Rate            18.27466 ***    90.59        19.15794 ***
D1                         1.0617 ***    1.57           1.5174 ***
D2                        2.01534 ***    2.88          2.72208 ***
D3                       -1.95333 ***    -2.87        -2.13281 ***
D4                       -1.59696 ***    -2.13        -2.10927 ***
D5                       -2.33785 ***    -3.64        -2.19731 ***
Adj_[R.sup.2]               0.3243                       0.3951

                                  Count_250

variables              t-value    Coefficient       t-value

Points                 5.66          0.42519 **     2.23
LTV                    -12.09       -0.09487 ***    -9.81
Original Loan Amount   -17.03     -0.0000376 ***    -15.39
Original Rate          108.3        18.59652 ***    120.96
D1                     2.54          1.63236 ***    3.15
D2                     4.41          2.71059 ***    5.05
D3                     -3.55        -1.71686 ***    -3.29
D4                     -3.2         -1.99039 ***    -3.48
D5                     -3.87        -1.83049 ***    -3.71
Adj_[R.sup.2]                         0.4279
Gale Copyright:
Copyright 2011 Gale, Cengage Learning. All rights reserved.