Until recently, most European economies had witnessed relatively few bankruptcies, and their banks had enjoyed generous lending margins. This meant that the banks could succeed by avoiding truly bad loans: only minor benefit derived from being distinctively competent at risk rating, pricing, and monitoring. No longer is this the case. Throughout Europe, banks are now experiencing losses of unprecedented proportions. It is tempting to blame the current economic climate and hope that all will be better in a year or so, when the economy improves. But as a McKinsey study of credit risk management has found, both the underlying nature of the risk and its importance to a bank's profitability equation have fundamentally changed. So much so, in fact, that financial institutions need to develop a whole new technical and organizational approach to managing credit, which will radically alter the culture of traditional universal banks.

"Of course, financial ratios and local information are important. But one of my most reliable sources of judgment on credit risk is having my assistant observe the behavior of a client waiting for a credit negotiation meeting to begin." According to the senior credit officer of a major European bank, such "eye contact" methods have long characterized the art of lending. Together with limit hierarchies, multiple signature systems, and annual reviews, they helped banks to identify prospective bad loans. Most lenders, after all, were generalists skilled in relationship building, and they could, with training, provide adequate credit risk evaluation.

For the most part, this approach worked well enough in an environment of comfortable margins and low loss levels, allowing banks to focus improvement efforts on volume growth and geographic diversification. In the meantime, however, the underlying nature of credit risk has changed, driven by four external forces:

Although bankruptcies are cyclical, the current environment has produced, in both commercial and personal segments, two to three times as many defaults as in past recessions. This is related to the increase in debt built into balance sheets now as compared with the late 1970s. It is likely that bankruptcies will continue to be cyclical but at a structurally higher level.

• Disintermediation—the process by which companies access the capital markets directly—has continued to allow good credit to avoid the banking system entirely. This trend, combined with the rapid lending growth of the late 1980s, the mispricing of risks, and the development of new financial technologies such as securitization, has resulted in a deterioration of the balance sheet of the banking industry as a whole (Exhibit 1).
• Margins for all lending, especially large corporate loans, have fallen dramatically over the past decade. This means that both the level of losses and their volatility are now fundamental determinants of profitability. In the past, they were a relatively stable and predictable cost of doing business.
• The current recession has cut into the historic price stability of assets (particularly land and buildings) in a manner not seen in Europe for many years. This drop in wealth and income not only affects companies' ability to compete, but also means that the whole structure of bank collateral and guarantees is far less effective (Exhibit 2).

To be fair, banks have, to some extent, seen this coming, and have made improvements at the margin (for example, by installing rating systems and relationship profitability models) to help measure risks. Yet most have not addressed the real challenge of radically revising the organizational processes, structures, incentives, and culture of their credit functions.

The time has come to put such a challenge at the top of senior managers' agendas. The reason is simple: better credit decision making ensures lower losses, reduces their volatility, and minimizes the possibility of one proving catastrophic. More efficient systems and processes lead to lower costs and better service levels for customers. Our work in Europe, for example, indicates that a real change in organizational systems, incentives, and skills can reduce credit losses by 20 to 40 percent, cut credit operating costs by 10 to 20 percent, and halve turnaround time.

A new model

In practical terms, achieving this step change in credit capabilities involves three key challenges.

1. Making risks visible and measurable

Internal ratings often said more about the likes and dislikes of individual account officers and their superiors than about the underlying credit risks

From discussions with senior bank managers across Europe, we found a staggering lack of transparency in systems for measuring and monitoring credit risk exposures, whether of individual contracts or whole portfolios. Many banks had just finished—or were in the process of—installing rating systems to classify a borrower's creditworthiness. But the internal ratings often said more about the likes and dislikes of individual account officers and their superiors than about the underlying credit risks.

The systems currently in place seldom correlated well with default rates. Nor were they evenly applied across segments. Their criteria for up- and down-grading were not clearly established, and there was no consistent link between risk classification and margin. Moreover, at the level of integrated credit portfolios, little information existed on the current structure or likely evolution of the portfolio, apart from that mandated by regulatory reporting requirements. Successful efforts to calculate the concentration of risk by industry sector and region, or the concentration of asset size by borrower and product, were also rare.

If these systems are to function as they should, they must provide:

A consistent rating system across products and segments—that is, a system able to classify ratings according to an expected (and monitored) default rate; to ensure that the average default rates within a risk class are consistent over time; and to force rating classes to mean the same thing across different segments.

Credit scoring to complement (in commercial lines) or replace (in personal lines) the credit judgment of line officers. Many banks report benefits from credit scoring, at least on the personal side. The best scoring models are dynamic and have the ability to monitor and learn from results, as well as from the macroeconomic environment, so that internal values can be adjusted over time. Although scoring is not widely used on the commercial side, banks can gain substantial benefit by applying it in conjunction with lender judgment.

In one case, we tested more than 20 different scoring models, which were largely based on balance sheet data such as ratios for cashflow to debt and interest expenses to operating margin. Prediction accuracy was quite good overall. The results properly identified 90 to 95 percent of the poor credit risks and misclassified only 2 to 3 percent of good clients as high risk. Such ratings should, of course, be only the starting point of underwriting decisions, followed by discussion about why the model rating should be overridden in this or that particular case.

A cross-functional "early warning system" to predict defaults at least a year in advance. The major leverage point in the credit monitoring process is being the first bank to recognize and react to a possible client bankruptcy. Late movers are rarely able to recover money without lengthy and expensive workout procedures.

Done well, an early warning system will produce a "watch list" that contains 70 to 80 percent of future defaults

Effective systems develop warning signals by looking at individual behavior—such as overdrafts—and combinations of behaviors—say, late payment and lack of activity. Properly read, these signals can provide an index that can be used to develop a watch list which is predictive a year in advance. Done well, an early warning system will produce a "watch list" that contains 70 to 80 percent of future defaults (Exhibit 3).

A relationship profitability model that allocates capital on a risk-adjusted basis. Many banks in Europe are now in the process of developing models that include interest spreads established from a matched opportunity rate—that is, funded at the same maturity—plus amortized fees, as well as data on associated products (such as cash management or trust) in which credit was the lead item. At a minimum, regulatory capital should be put against the relationship income thus generated. In the better systems, a risk-adjusted capital measure is also used. The crux is that the system must recognize both economic and risk capital (which is a function of expected loss, volatility, and co-variance with other assets) and should be applied across segments.

A transparent view of portfolio concentration. In well-managed banks, portfolio monitoring can be done at different organizational levels, as well as by geography (economic region), currency, type of exposure (secured versus unsecured), industry segment (using in-house criteria, since government classifications are generally not helpful), individual customer, and product.

2. Redesigning the credit process

The credit hierarchy of most banks reflects their organizational hierarchy, not the credit skills of individuals

Formerly, limit hierarchies determined the credit approval process at most banks, and discretion was largely a function of seniority and status. Making loans—the bigger the better—was evidence of organizational standing. Executives who had been promoted for their strong relationship and selling skills thus found that their discretionary limits rose with promotion, not because of a genuine advance in credit skills or an outstanding credit performance record. As a result, the credit hierarchy of most banks reflects their organizational hierarchy, not the credit skills of individuals.

A second traditional element of this process is the credit committee, which brings a broader range of experience and fresh pairs of eyes to a particular decision. But the committee approach is easily misused. In fact, it has spread so widely precisely because it diffuses individual responsibility. All too often, recommendations are simply rubber-stamped by committees, whose members are far removed from the facts of a case and have little time to dig deep into the company or industry.

Significant change in these traditional credit processes is required—change that will go to the root of many bank cultures. Designing an up-to-date credit capability involves:

Linking limit-setting authority closely to proven skills. In the United States, for example, many banks have concluded that it is no longer feasible to maintain the skills of hundreds of loan officers throughout their branch systems. They have elected to limit lending authority to a smaller number of regionally based credit experts. These underwriters are highly trained and carefully motivated specialists.

Centralizing credit decisions concentrates them in the hands of the most capable individuals and creates centers of information about credit markets

Centralizing credit decisions in this way means that only 70 to 100 people make virtually all the lending decisions—except, perhaps, for those related to large corporate clients. Not only does this concentrate credit decisions in the hands of the most capable individuals; it also creates centers of information about credit markets, making it possible to exploit areas that were previously too diffuse to handle. The banks that believe in this system argue that the loss of direct customer knowledge is more than compensated by better underwriting skills, more concentrated information, and greater objectivity in making decisions.

Another approach is to embark on a major skill-building program for all lenders, setting strict standards and regularly reviewing all limits. So far this approach has had mixed reviews. Some banks report success in boosting skills, but others have found the large-scale training and review procedures difficult to implement. In both cases—centralizing decisions and upgrading broad-based skill levels—limits must be earned. The goal is to decouple them completely from the organizational hierarchy.

Making credit skills a critical part of evaluation systems. Regardless of the degree of centralization, credit skills evaluation must be an explicit part of the staff assessment and promotion process. To be effective, this process must clearly spell out the skills that are expected. It must also measure, quantitatively where possible as well as qualitatively, a lending officer's performance in portfolio complexity, risk analysis, credit administration, monitoring, early problem recognition, and credit judgment. Exhibit 4 shows one example of how this could be accomplished.

Because the life of a loan is often longer than the job tenure of the loan officer responsible for it, true accountability for loans and losses is rare

Creating transparent accountability in the underwriting and monitoring process. Because the life of a loan is often longer than the job tenure of the loan officer responsible for it, and because credit committees tend to dilute responsibility, true accountability for loans and losses is rare. Some banks have addressed this problem by tracking loans back to the original underwriters. Others have set up systems whereby new officers "buy" loan portfolios from their predecessors. Both these approaches, if used consistently with the evaluation objectives set out above, can provide the necessary accountability. The real test is whether senior managers can point to any loan in the portfolio and identify the one person—or the small team—with clear responsibility for it.

Devising a unique underwriting process for each customer segment. Most banks follow an undifferentiated underwriting process: all credits that come up through the branch system are subjected to the same mechanisms of review and the same hierarchy of credit committees. (In one case we found eight hierarchical filters between original application and final approval. This process was expensive, but added little value.)

Tailoring these processes to individual customer segments can take many different forms. Building an automated, centralized underwriting-by-scoring system for personal lines, small business lending, and mortgages is one example. Such an approach removes all credit responsibility from branches and creates a completely separate channel. Another example is to establish a highly professional organization geared to complex problem solving for medium to large corporate clients. The goal here is to create a meritocratic, team-based approach to each relationship, in which every team member has broad credit risk experience and several have deep specialities either by product or by industry.

Seventy to 80 percent of branch analysis is often duplicated by a central credit department

Standardizing back-office procedures while customizing processes and prices for each segment. When a set of unique credit processes is redesigned, it is possible to eliminate duplication of effort and improve efficiency through standardization. We have found, for example, that 70 to 80 percent of branch analysis is often duplicated by a central credit department, and that standardizing processes can cut up to 40 percent of the time underwriters spend on small business loans.

Large benefits can also be captured by customizing procedures and prices by risk class. Most banks find little connection between the risk rating of a relationship and the total return from it. This is largely because underwriting is done bi-modally: the credit is either underwritten or rejected. A better connection can be established by setting minimum prices for each risk class. This can be done analytically, by using bond equivalents to set market-required returns, or by senior management judgment informed by an internal rating system. In addition, collateral types, loan to value, and duration of product can all be specified by risk class.

In the same way, monitoring policy should not be uniform, but instead should provide more frequent reviews and financial updates for riskier clients. The practical effect is to shift review and monitoring time to high risk and/or very large loans, rather than dedicating equal amounts of time to all transactions. With care, it is possible to design the process such that fewer reviews are conducted overall and a much greater proportion of high risk and large loans gets considered in detail.

Establishing an independent workout function. The skills, procedures, and processes necessary for running a workout organization are fundamentally different from those relevant to underwriting and monitoring. Banks without separate workout capabilities often find that a sudden rise in bankruptcies swamps the whole network. When it does, origination quality and focus can easily slip. An independent workout function must do four things well:

• Establish clear rules for when to use a traditional workout process and when to apply a lower-cost, streamlined collection process via letters and phone calls. Successful banks may well have four or five separate processes, depending on cost and potential for collection.
• Efficiently carry out triage on the portfolio. Once a loan is in workout, prioritizing effort according to the urgency of the situation, the possible economic impact, and the probability of success is critical to achieving high recovery rates. Most workout units are far too reactive to day-to-day crises to set clear priorities based on bottom-line impact.
• Assemble the best skills. Each situation will require a specific combination of skills—real estate, legal, credit assessment, financial analysis, merchant banking, collateral evaluation, and basic project management among them. The exibility quickly to build an appropriate team (or to buy needed skills outside) is critical.
• Employ analytically driven decision rules. The basic choices facing a bank—restructuring, loan sales, foreclosures, or "do nothing"—should be evaluated according to the NPV of each option multiplied by its probability of success. Such analyses should take into account all costs, including operating and carrying costs, as well as return on assets (if any).

Monitoring risk limits and aggregations centrally. A strong central authority is needed to ensure that state of the art methodologies are used, to monitor overall risk response, and to counterbalance the credit authority invested in the front line. The objective should be to create an organizational tension that improves decision making without diffusing accountability. A well-designed central authority can provide maintenance and monitoring of limit systems and portfolio concentration; an early warning system authorized to place clients on the watch list; industry and microeconomic analysis to identify optimal portfolio composition; and secondary market evaluation and trading capability.

3. Managing credit portfolios proactively

Most bankers intuitively know that the law of large numbers and the basic tenets of portfolio theory apply to credit risk management. Yet it is surprising how few shape their credit portfolios to comply with them. In recent years, the difference in bottom-line performance between a bank that has 15 percent and one that has 8 percent of its portfolio in commercial real estate has been measured in hundreds of millions of dollars.

For large banks, careful portfolio diversification—so that the volatility as well as the expected losses are optimized—can have a major effect on profits, in addition to reducing the risk capital that must be held against portfolio volatility. For small banks, however, the law of large numbers leads logically not to diversification, but to a rigorous effort to syndicate and reinsure loans so as to avoid exposures that far exceed the banks' underlying capability to absorb losses.

In practice, good portfolio management rests on:

Even when concentration limits are known, there is rarely a well-defined process for managing exposure up or down

Setting explicit concentration limits. In many banks, portfolio composition is driven by the pattern of origination. Even when concentration limits are known, there is rarely a well-defined process for managing exposure up or down. Successful institutions tend to define asset composition targets according to expectations about loss rates, volatility, and co-variance with other asset pools, and to review and revise these targets at regular meetings of the top credit committee. Such reviews usually include a detailed analysis both of the projected economic health of industries and regions and of the sensitivity of the current book to potential economic shocks of various kinds.

These banks then translate their concentration limits into a policy on originations by using gradations of urgency. They set "gray" lists of asset types, where loans can be put on the book only if they are above a minimum (usually high) credit rating; "black" lists, where a credit can be added only if another loan of a lower rating is run down; and "red" lists, where all loans in the asset class are up for review and all loans below a certain credit rating are targeted for run-down.

Including management in budgeting and plans for growth. There is a well-known, consistent relationship between present loan growth and higher future loss rates. This generally comes about because banks grow faster than their core skills or in areas where they lack local knowledge, and so end up as the lender of last resort. Examples can be seen in the experience of UK banks in different regions, as shown in Exhibit 5. In organizational terms, this occurs because many banks set volume and growth targets through an internal budgeting process that does not include knowledgeable risk managers. Strategic planning departments, for example, often make unrealistic growth projections for new regions because they have overlooked the issue of risk.

Setting limits for loan size. Many banks carry loans on their books that are far too large for their equity base. A bank with 20 large, independent loans that have a default probability of 1 percent and that jointly account for a third of its $10 billion portfolio can cut its chance of catastrophic loss in half by reducing loan sizes so that 60 loans account for the top third of its portfolio. If, as is usually the case, there is some connection between these loans, the benefits of loan size limits are even greater.

Barriers to change

Historically, most banks have been under insufficient profit pressure to provide a mandate for action on risk management. Internal systems for profitability measurement have not helped. They have tended to subsidize the asset side of their balance sheets at the expense of liability in the absence of any (or at least any appropriately used) market opportunity rate-based transfer prices. The great prestige of senior credit officers, and the insular, turf-conscious "credit society" in which they grew up and over which they now preside, did not help. Nor did the frequent lack of relevant information, or the reluctance to provide it for fear of triggering restrictions on banks' freedom of action in regard to reserve policies.

Twenty to 30 percent of the loans made by many European banks lack proper documentation and fail to follow the procedures outlined in bank policy

Even where the apparatus of sound risk management existed, it was—and still is—often not implemented in a rigorous manner. Many banks have scoring models and profitability and rating classifications that they have designed at great expense, but that they neglect to use regularly and effectively as a factor in credit decisions. Reviews of closed files indicate that 20 to 30 percent of the loans made by many European banks lack proper documentation and fail to follow the procedures outlined in bank policy.

Thus, the barriers to change are powerful—but they are not insurmountable. We have found that the banks that have broken through them paid close attention to:

• Sending consistent organizational signals. Front-line officers often view calls for change in the management of credit risk as only a short-term reaction to the current environment. "One quarter, management wants growth; next quarter, it's higher quality," one relationship manager told us. To implement a solid and systematic credit process, a forceful, well-communicated policy must be consistently applied over a period of years in the strategic as well as day-to-day decisions of top management.
• Ensuring real leadership by senior managers. In every successful restructuring of the credit management process that we know, senior managers have publicly "bet their careers" on a new vision for their banks. They have also backed up their vision with tireless efforts to break down barriers to change, coordinating and driving through the necessary shifts in managerial responsibilities at all levels.
• Promoting collaborative, cross-functional efforts. These tasks cannot be effectively executed by simple delegation; they require the formation of a number of cross-functional teams focused on specific, measurable goals.
• Working from an integrated vision. Many credit programs are rejected by the line because senior managers focus on only one aspect of a highly interrelated process rather than on the entire system. Attempting to install a commercial credit scoring system without thinking through related changes in organizational structure, incentive systems, power balances, and service levels is a virtual invitation to the front line to delay or even sabotage the new system.

All this is a very tall order, calling for major change directly led by a bank's most senior executives. However, fundamentally upgrading credit risk capability produces substantial benefits in the short to medium term and, in the longer term, will be an essential requirement for success.