The Peter, Paul and Mary song "Where Have All the Flowers Gone" asks a simple question of the Vietnam War: "When will we ever learn?"

The same question could be applied to subprime consumer lending.

Subprime consumer lending has reached its highest levels since the beginning of the financial crisis, according to a recent article in the Wall Street Journal. "Almost four of every 10 loans for autos, credit cards and personal borrowing went to subprime customers during the first 11 months of 2014," the paper reports.

In other words, the subprime consumer lending business, having failed to heed the lessons of the financial crisis, could be primed for another disaster.

Consumer lending is an activity with little margin for error. When I was a senior credit officer at a major financial institution, our division planned for a 1-2% after-tax return on assets for our consumer lending products. We set the loan pricing accordingly.

We had to forecast fee and interest income revenue as well as the costs of funding the loans and staff. This was difficult, but we had a good historic record to help us with our estimates. But we also had to estimate future credit losses — the most volatile component in loan profitability. An uptick of just a few percentage points in credit losses could move us into unprofitability.

Many lenders believe they can deal with the volatility of credit losses by using risk-adjusted pricing. If they know that the average subprime borrower has a higher risk for default, they charge the entire group more to compensate for this higher individual risk. That works as long as, on average, a group of subprime borrowers behaves according to expectations. But what happens when the whole group defaults at higher-than-expected rates?

The financial crisis has already given us our answer.

Part of the problem is that many lenders continue to rely too much on credit scores to predict credit losses. When I learned consumer lending in the 1980s, we focused on the "Three Cs" of credit: character, capacity, and collateral. Character, as summarized by a credit score, appears to dominate today's credit analysis. But the more pressing question, as I said in a recent HousingWire article, is whether customers can afford to take on new debt at a high interest rate.

Credit scores predict the statistical odds of that a customer will default. The median consumer FICO score of about 720 has a projected default risk of about 6%, while a "superprime" 780 score has a less than 1% default rate and a subprime 640 score has a 15% default rate, according to data from a 2012 study by the Consumer Financial Protection Bureau.

But what if, due to unforeseen events, the default rate projections are off by 30%? The subprime default rate increases to 20%. This can easily tip the scales to an unprofitable loan portfolio.

Income should be just as important a factor as credit scores in the loan approval process. And at a macro level, customer income remains a concern. The U.S. Bureau of Labor Statistics estimates that at the end of 2014, the labor "underutilization" rate was 12%, double the official unemployment rate.

Meanwhile, consumer debt ratios have remained constant at about 5% of disposable income since the start of the financial crisis, according to the Federal Reserve's "Household Debt Service and Financial Obligations Ratios report." While this may not seem like a burdensome ratio, let's put it in perspective. I went to the website and searched for subprime credit cards. I found a representative offer with an annual percentage rate of 29.99% (!). A borrower with a $5,000 credit card balance would pay about $250 per month in interest. For a borrower with $2,500 of monthly disposable income, that's a 10% debt ratio just for the incremental subprime debt.

The crisis taught us three hard lessons. First, the "Three C's" of credit matter. Reduced documentation, an overreliance on credit scores and defective appraisals produced loans that were designed to default.

Second, lenders need to have skin in the game. Lenders tolerated and even encouraged bad underwriting because they didn't have to hold risky loans on their books. Instead they sold them on the secondary market, and when the music stopped, the last man standing took the hit.

Lastly, everybody — from bankers and regulators to auditors and ratings agencies — needs to act with more common sense. All parties involved failed to ask basic questions about the consequences of extending credit to risky borrowers.

The mortgage industry seems to have learned these lessons. The subprime consumer lending industry, on the other hand, appears to be ignoring them.

I predict two bad outcomes as a result. A new wave of consumers will become overextended, default on their loans and further damage their credit. And the new nonbank lenders who have helped drive the increase in subprime lending are in for an unpleasant surprise.

As new entrants in a highly competitive market, they are likely subject to "adverse selection." They'll take on riskier customers in order to build their business.

The history of consumer lending is strewn with failed companies that thought they had the risk-based pricing model figured out. Providian Financial set the standard in subprime lending, only to be involved in extended litigation with the U.S. government and ultimately sold to Washington Mutual in 2005.

An old saying in lending is a "rising tide lifts all boats." The improving U.S. economy is indeed helping many borrowers. But the improvement is uneven. It's plausible that subprime borrowers will benefit the least, and risky lending will only make their financial futures more uncertain.

Neil Librock is a managing director at the Berkeley Research Group. He was previously a senior credit officer at Wells Fargo. The views and opinions expressed in this article are his own.