Credit lending institutions often obtain credit data from third parties to make lending decisions. There are three major credit data providers (credit bureaus) in the United States. The credit reports that are obtained from these sources are not the same for the same individual for many reasons. The major sources of the data for these credit bureaus are the lending institutions themselves. Nevertheless, lending institutions do not report the performance of their customers to all three bureaus for cost considerations. As a result, there is discrepancy in the credit data pulled from credit bureaus. In this paper, we argue that variables based on merged credit bureau reports are more accurate than variables based on just single-bureau reports. Models estimated using data from individual credit bureaus have larger mean square errors relative to a model estimated using the merged data. Our results also show that the merge model possesses more predictive power than any of the individual credit bureau models. This is shown using Kolmogorov-Smirnov statistic and C-statistic.