I think it is important to understand that loans with high DTI were an enabler for speculation during the housing bubble, and this speculation pushed up house prices. So, although the authors argue high initial DTI loans are not a good predictor of defaults, the prevalence of high DTI loans was evidence of a bubble - and a good predictor of a housing bust.
DTI is the ratio of mortgage debt to income. For an individual borrower, it is not a reliable predictor, for two reasons. One is that people with the same income can have different spending and saving habits. Another is that income is often measured with error, sometimes with deliberate falsification but sometimes randomly. If I have enough salary income to qualify for a mortgage, I might not bother to include my capital income on my application.
Although DTI is a noisy indicator for the individual, a policy of accepting high DTI loans is likely to lead to higher default rates than a policy that is more restrictive. Thus, I agree that the high prevalence of such loans portended trouble.