FSA 4 – Analysis of Financial Institutions

Most of the analytical machinery a candidate has met so far was built for manufacturers and retailers: firms that buy inputs, transform them, and sell the output. A bank does none of that. It buys money and sells money. Its inventory is a promise to repay, its cost of goods sold is interest, and its most important asset is a claim on somebody else’s future behaviour. Two features of that business model force a separate framework.

Systemic importance

The first feature is systemic importance. Banks sit between the providers of capital and the users of capital, and in doing so they weave a net of financial links across households, other banks, corporations and governments. Because the links run in every direction, the failure of one bank does not stop at that bank. It travels. The larger the institution and the denser its connections, the further the damage spreads, and the failure of a sufficiently large bank could in principle bring down the system that contains it.

Systemic risk is the formal name for that possibility: the risk that a disruption to financial services, caused by impairment of part or all of the financial system, produces serious negative consequences for the wider economy. Central to the idea is contagion, the spread of a shock from the institution, market or instrument where it started to other institutions, markets and places. Every kind of intermediary, market and piece of financial infrastructure can be systemically important to some degree. The 2008 global financial crisis pushed the problem onto the agenda of regulators in many countries at once, and a faltering economy can pass its difficulties on to healthier economies elsewhere.

Two consequences follow. Financial institutions are heavily regulated, with rules covering how much capital must be held, how much liquidity must be maintained and how risky the asset base may be. And deposits, the bulk of most banks’ liabilities, usually carry insurance up to a fixed ceiling provided by the government of whichever country the bank operates in. In December 2016 deposits made up more than 80% of what domestically chartered commercial banks in the United States owed in total. The reason deposit insurance exists is not charity: even the suspicion that a bank might fail to honour deposits can start a bank run, and a large sudden withdrawal can turn suspicion into an actual failure.

Financial assets rather than tangible assets

The second distinguishing feature is the nature of the assets. A manufacturer holds plant, inventory and equipment. A bank holds loans and securities. Financial assets carry direct exposure to credit risk, liquidity risk, market risk and interest rate risk, and unlike most tangible assets they are frequently carried at fair value for reporting purposes. In general, therefore, the reported values of a financial institution’s productive assets sit relatively close to market values, which is both an advantage to the analyst and a source of earnings volatility for the institution.

The population of financial institutions

The term covers a wide range of businesses, and services overlap heavily across them. Banks take deposits and make loans, but many also run investment management arms and sell derivatives, which are in effect insurance against adverse moves in interest rates, equities and currencies. Life insurers sell mortality protection, but they also sell savings vehicles. The categories below are illustrative rather than exhaustive, and the structure of the industry differs by country, with state ownership more common in some jurisdictions than others.

The list excludes supra-national organisations such as the World Bank, which brings together 189 member countries and lends via two arms, the International Bank for Reconstruction and Development and the International Development Association, because these are formed by member states for specific missions rather than offered as investments. Other examples are the Asian Development Bank and the Asian Infrastructure Investment Bank.

This reading concentrates on two of these groups: banks, defined broadly as deposit-taking and loan-making institutions, and insurers, split between property and casualty companies and life and health companies.

Banking and insurance are not systemically alike

The insurance market as a whole conducts a smaller share of its business across borders than banking does. The exception is reinsurance, which is largely international, and that exception matters: a reinsurer can be the link that ties institutions in different parts of the world together, which raises rather than lowers systemic vulnerability. A second structural difference is that the foreign branches of insurance companies are generally required to hold, inside a jurisdiction, assets sufficient to cover the policy liabilities written in that jurisdiction.

Beyond containing systemic risk, global and regional regulatory bodies exist to harmonise rules, standards and supervision. Consistency limits regulatory arbitrage, the practice by which a multinational exploits differences between jurisdictions to escape unfavourable regulation.

The Basel Committee and Basel III

Foremost among the global bodies working on financial stability stands the Basel Committee on Banking Supervision, set up in 1974 and hosted and supported by the Bank for International Settlements. Its members are central banks and the agencies responsible for supervising banks. The membership list as of July 2017 spans Argentina, Australia, Belgium, Brazil, Canada, the Chinese mainland, the European Union, France, Germany, Hong Kong SAR, India, Indonesia, Italy, Japan, Korea, Luxembourg, Mexico, the Netherlands, Russia, Saudi Arabia, Singapore, South Africa, Spain, Sweden, Switzerland, Turkey, the United Kingdom and the United States, with Chile, Malaysia and the United Arab Emirates as observers. Many countries are represented twice, once by the central bank and once by a separate prudential supervisor, which is itself a useful reminder of how fragmented supervision is in practice.

The Committee produced Basel III, the successor to Basel I and Basel II. Its stated purposes are to improve the banking sector’s capacity to absorb shocks from financial and economic stress whatever their origin, to improve risk management and governance, and to strengthen bank transparency and disclosure. Three highlights matter for analysis:

• Minimum capital. Basel III specifies the minimum share of risk-weighted assets that must be funded with equity capital. This prevents a bank from taking on so much leverage that it cannot survive loan losses.
• Minimum liquidity. Holdings of high-quality liquid assets must be sufficient for the liquidity a 30-day stress scenario would demand, so that a bank survives losing part of its funding or facing an outflow set off by off-balance-sheet commitments.
• Stable funding. A bank must hold a minimum amount of stable funding relative to its liquidity needs over a one-year horizon. Stability depends on the tenor of deposits, longer being more stable, and on the type of depositor, consumer deposits being more stable than interbank money.

By constraining leverage, Basel III pushed banks to focus on asset quality, to hold capital against operational and other risks, and to improve their risk assessment processes. It also changed the quality and composition of the capital base, strengthening its capacity to absorb losses so that losses are confined to the institution’s capital investors rather than passed to depositors, taxpayers or other institutions. That confinement is precisely what reduces contagion risk. Having written the framework, the Committee now monitors its adoption and implementation by member jurisdictions.

Several other international bodies work alongside it. The Financial Stability Board draws on supervisors and regulators from the G–20 countries together with Hong Kong SAR, Spain, Singapore and Switzerland, and aims to identify systemic risk and coordinate the responses of national authorities. The International Association of Deposit Insurers exists to make deposit insurance arrangements work better. Insurance supervision worldwide is the concern of the International Association of Insurance Supervisors (IAIS), and the International Organization of Securities Commissions (IOSCO) works towards fair and efficient securities markets. The IAIS and IOSCO sit with the Basel Committee in a Joint Forum addressing issues common to banking, insurance and securities.

National regulators and the compliance burden

Global bodies work through national regulators, and it is those national bodies that hold authority over specific aspects of an institution’s operations. The network is genuinely complex, with overlapping and sometimes differing mandates, and membership of the various global bodies only partly overlaps. The International Association of Deposit Insurers has 83 member organisations, some of which are also Basel Committee members, such as the United States Federal Deposit Insurance Corporation (FDIC), and some of which are not, such as Singapore Deposit Insurance Corporation Ltd, and Germany’s Bundesverband deutscher Banken. In some countries a single body oversees both banking and insurance, as Japan’s Financial Services Agency does; in others insurance has its own regulator, as with the National Association of Insurance Commissioners (NAIC) in the United States, and the China Insurance Regulatory Commission.

The practical consequence for a global institution is a compliance load that grows with geographic reach. HSBC Holdings, one of the most international banks in the world, discloses supervision by roughly 400 separate central banks and regulators across the jurisdictions holding its offices, branches or subsidiaries, each with its own requirements and controls.

The remaining three sections on banking work the CAMELS framework through the financial statements of a real institution, Citigroup, at the end of 2016. Two caveats belong at the front. First, the evidence an analyst gathers varies by analyst: an equity investor may care far more about earnings and earnings quality than about capital adequacy, while a fixed-income investor may care far more about capital adequacy and liquidity than about earnings, and the type of investor determines what analysis is performed for each component. Second, although CAMELS has quantitative elements it is not a formulaic approach. Judgement and discretion shape both the testing done to gather evidence and the rating assigned once the evidence is in. What follows is one plausible set of analyses, not an exhaustive one. Each component below closes with a rating, where 1 is the highest and 5 the lowest.

Capital adequacy

Capital adequacy compares the proportion of assets funded by capital, with assets given varying risk weights. The assets are sorted into risk classes and the capital behind them into tiers, namely Common Equity Tier 1, Total Tier 1 and Tier 2.

Citigroup sits well inside the required limits in both years. The ratio slipped from 14.60% to 14.35%, and the cause is mostly the increase in deferred tax assets disallowed in the computation: the carry-forward deduction rose from $9,464 million to $12,802 million and the excess-over-limitations deduction from $2,652 million to $4,815 million. The three risk categories reconcile to the total in both years: 773,483 + 64,006 + 329,275 = 1,166,764.

Total Tier 1 Capital includes instruments meeting criteria based on subordination to deposits and other debt obligations, absence of a fixed maturity, and no requirement to pay dividends or interest without the bank’s full discretion. Preferred stock can be structured to meet those criteria, and that is what drove the improvement here: the Tier 1 ratio rose from 14.81% to 15.29%, mostly because of additional perpetual preferred stock qualifying for inclusion in 2016, from $16,571 million to $19,069 million, together with a fall in the disallowed deferred tax assets from $14,195 million to $8,535 million. Additional Tier 1 Capital more than quadrupled, from $2,558 million to $11,009 million.

Tier 2 Capital admits, within limits, part of the allowance for loan and lease losses along with other instruments ranking behind depositors and general creditors. The Total Capital Ratio improved from 16.69% to 17.33%, driven mostly by the increase in Total Tier 1 Capital and in qualifying subordinated debt. Checking the arithmetic: 178,387 + 23,759 = 202,146, and 202,146 ÷ 1,166,764 = 17.33%.

Assessment. Citigroup’s capital adequacy at the end of 2016 looks solidly positive. All three principal ratios exceed the minimums required to be considered well capitalised, and by a wide margin: 14.35% against 4.50%, 15.29% against 6.00%, and 17.33% against 8.00%. A rating of 1 could be justified.

Asset quality: composition

Asset quality matters greatly to a bank. As intermediaries, banks owe their existence to creating loans, and unsound credit policy can erode the capital base quickly in a downturn, straining liquidity and the ability to generate earnings, and making new lending difficult.

Bank assets rise in riskiness in three broad steps. One portion sits in highly liquid form: cash, balances at other banks, and holdings that turn into cash fast, repos and certain receivables among them. Little risk attaches to these. Next come investments made with cash deemed surplus to operating needs, classified under US GAAP and IFRS as available for sale, reported at fair value, or held to maturity, reported at amortised cost unless impaired. These are riskier than the liquid securities and reflect a management investment decision, but their value is fairly transparent and their reported value reflects realisability in cash, though establishing that takes more analytical effort for held-to-maturity securities. The riskiest and often largest class is the loan book, which embodies credit risk and management’s judgement in extending credit. The underwriting risks and those judgements are reflected in the allowance for loan losses.

This is where the external analyst is at a structural disadvantage. Some information is simply unavailable to an analyst or investor, whereas a supervisory examiner can see the bank from the inside, assess the soundness of loan and investment policies and procedures, review the construction and workings of internal controls, and examine how exceptions to credit policy are handled. The analyst is confined to circumstantial evidence that credit policies are sound and are being maintained, and that evidence sits in the financial statements without being obvious from the face of the balance sheet.

Three observations follow from the composition. Liquid assets are the largest single group at 37.4% in 2016, slightly above the prior year, indicating greater liquidity. Investments at 19.7% of total assets are practically unchanged from 19.8%, and most of that money sits in available-for-sale paper measured at fair value. The consumer and corporate loan books carry the most risk, exceed a third of the balance sheet in each year, and rank second in size behind the liquid assets.

Asset quality: the investment portfolio

The analyst wants confidence that the investments, transparent in value though they are, reflect sound investment decisions, and that the loans came from similarly reasoned underwriting and are collectible at the stated amount.

Drawn from Note 13 to the 2016 Form 10-K. The final two columns restate the gross unrealised amounts against the cost invested.

Fair value of $299,424 million is below amortised cost of $300,269 million, a net difference of $845 million. State and local government paper contributes most of it, at $757 million. Losing 7% of value, that class was alone in exceeding a 2% loss. Two observations follow. Citigroup has not produced a net winning strategy with its available-for-sale investments, but the losses do not suggest reckless abandon either. Separately, forthcoming US GAAP rules remove the AFS label from marketable equities. Fair value measurement continues exactly as at the end of 2016, but from 2018 the gains and losses on remeasurement will run through the income statement rather than other comprehensive income. At 31 December 2016 Citigroup’s unrealised gains on AFS equity investments of $20 million exceeded its unrealised losses of $6 million, so on those market values the reclassification would benefit reported income.

Note 13 also ages the losses, which matters because a loss position that persists is likelier to represent impairment that is other than temporary. Losses rarely sit unchanged for years and then reverse without warning.

Just over half the losses, 54%, have existed for under 12 months, so they trouble an analyst less than the remainder. Of the $1,172 million of gross unrealised losses that are 12 months old or older, 60%, or $702 million, relate to state and municipal securities, which raises the worry that the biggest block of losses will end up being realised.

The same analysis can be run on the held-to-maturity portfolio, and although it is a much smaller share of total assets it still evidences the manager’s investment acumen. The result is consistent with the AFS review. Citigroup’s unrealised losses on HTM securities totalled $457 million, or 1.3% of the amount invested, and 82% of that, or $373 million, sat in positions whose losses had already run beyond 12 months, $180 million of it in state and local government paper. In dollar terms the HTM losses are far smaller than the AFS losses and in percentage terms they are minor, but their age is troubling. Problem assets do not usually improve with age, and a concentration of old losses may indicate management reluctance to report economic reality. Because HTM securities are carried at amortised cost, the classification obscures their fair value, and the age of the loss-making positions suggests impairment may be more severe than has been recognised.

Asset quality: the allowance for loan losses

Investments are neither as large nor as risky as loans, and loans cannot be assessed without working through the allowance for loan losses. Three accounts interact:

• The allowance for loan losses is a balance sheet account, a contra asset to loans, directly analogous to the allowance for bad debts that offsets accounts receivable at a non-financial company.
• The provision for loan losses is an income statement expense that increases the allowance.
• Charge-offs net of recoveries, the actual losses, reduce the allowance.

Total loans minus the allowance is the expected value of the loans, which is why the allowance matters so much for judging loan quality. It is also discretionary by its very nature. Underestimating it overstates both assets and net income, and almost every bank discloses the allowance among its most critical accounting estimates.

The way to test a discretionary number is to compare it with a less discretionary one. Net charge-offs are less discretionary indicators of loan quality than the allowance, but they have the disadvantage of being a confirming event: by the time a loan is charged off it has already gone bad. They carry a second disadvantage, in that charge-offs can be used in good times to store away earnings for later release through recoveries. Non-performing, or non-accrual, loans are loans not currently paying their contractual amounts due, which makes them a more objective measure of portfolio quality. Three ratios follow, each comparing a discretionary measure to a more objective one:

Assessment. Citigroup’s asset quality at the end of 2016 was mixed. Investment policy looks fairly conservative, but the age of some positions carrying unrealised losses suggests a possible denial of impairment. On loan quality, the ratio analysis leaves the consumer book looking well reserved and the corporate book not. A rating of 2.5, near the midpoint of the scale, could be assigned given the mixed signals.

Management capabilities

External investors observe only circumstantial evidence of management quality, and a good deal of that evidence sits in the proxy statement. Three observations from Citigroup’s 2016 proxy are worth recording.

• Citigroup aims for two-thirds independent board representation, where the New York Stock Exchange requires only a majority.
• The roles of CEO and chairman have been separate since 2009, a practice often viewed as good governance because it avoids conflicts of interest.
• Fourteen meetings of the Risk Management Committee took place in 2016, which shows attention being paid to one of the areas that matters most in a bank. The committee also created a subcommittee in 2016 to oversee data governance, data quality and data integrity, and that subcommittee met seven times in the year.

These are good practices, but they do not by themselves demonstrate strong management capability. What they demonstrate is that an environment exists in which strong management quality is permitted to flourish, which is a weaker claim and the correct one.

Related-party transactions are difficult to avoid at Citigroup’s scale. BlackRock and Vanguard each beneficially owned 5% or more of the outstanding common stock at 31 December 2016, and over 2016 Citigroup subsidiaries supplied both of them with lending, trading and other financial services in the ordinary course. The proxy states that the transactions were conducted on an arm’s-length basis with customary terms substantially the same as comparable transactions with unrelated third parties. Other related-party transactions are discussed in the Form 10-K and are routine for a company of this size.

On operational risk, the auditor issued an unqualified opinion on the effectiveness of the system of internal controls. That is evidence of a minimally satisfactory environment for management to operate in rather than a clear signal of competence; a qualified or negative opinion, by contrast, would be a serious concern for an investor.

Assessment. The board appears solidly constructed and appears to exert adequate control over managers, yet how the company actually performs also reports on the quality of its directors and managers, and asset quality was hardly a resounding result. A rating of 2 could be assigned.

Earnings: how much of the growth came from provisions

Earnings should be of high quality, and the main indicator of high quality is sustainability. Earnings are more sustainable when they do not depend on possibly opportunistic fine-tuning of discretionary estimates and are not reliant on non-recurring items or volatile revenue sources.

The allowance and the provision for loan losses are estimated amounts open to management discretion, and the provision can move a bank’s profitability sharply in a single year and steadily over many. The table sets Citigroup’s five-year change in pretax income against the change in total provisions for credit losses, which take in loan loss reserves alongside provisions for unfunded lending commitments and for policyholder benefits and claims.

Drawn from the five-year selected financial data in the 2016 and 2015 Forms 10-K.

Four observations follow, and they are uncomfortable ones.

• 2013 was the only year in which pretax income rose from the previous year while total provisions for credit losses fell. The $2,815 million decrease in the provisions drove 24% of the $11,637 million increase in pretax income.
• In 2016, 2014 and 2012 pretax income declined from the previous year. Each of those declines would have been more severe had it not been buffered by a decrease in total provisions.
• Across the five years, the change in total credit loss provisions contributed to improving pretax earnings in four of them. The single exception is 2015, when total provisions increased by $446 million, a negligible amount compared with the size of the decreases in the other four years.
• On a longer view, the five-year net change in provisions of $5,377 million accounts for 84% of the $6,381 million net change in pretax income, which indicates that not much profit growth happened anywhere else. The residual, the net difference of $1,004 million, is all that five years produced from the underlying business.

Earnings: the revenue mix

The second indicator of sustainability is the extent to which trading income features in the revenue stream. Trading income tends to be volatile and is not necessarily repeatable, whereas net interest income and fee-based income provide returning streams. An analyst should examine the composition of the revenue stream both to see whether it is growing and to identify what is driving the growth or the decline.

Total revenues in 2016 of $69,875 million are almost unchanged from the 2012 level of $69,530 million. At 10.9% of total revenues, trading income has been trending upward as a proportion over the five years. Rather than growing its sustainable, non-volatile revenues, Citigroup’s principal transactions line is moving the other way, rising both in absolute dollars and in relative importance. All other non-interest revenue in 2016 is at its lowest representative level since 2012, and although the net interest revenue proportion improved in 2016 it remains below where it stood in 2012.

Earnings: reading the net interest margin

Net interest revenue is the product of two management activities: managing the interest earned on loans and other interest-bearing assets, and managing the interest paid on deposits and other interest-bearing liabilities. Banks can create value through maturity transformation, borrowing on shorter terms than they lend. Lending long at a higher rate than the cost of short-term funding creates value, but the same structure destroys value if short-term funding markets seize up or the yield curve unexpectedly inverts, which is why a bank’s risk management and diversification practices are integral to the process rather than incidental to it.

Five observations come out of the asset side. The overall average rate earned slipped from 3.68% to 3.64%. One reason lies within loans, the single largest asset category: Citigroup sold its OneMain Financial subsidiary at the end of 2015, a United States consumer installment lending business with high yields, and that disposal pushed the rate earned in US offices down from 7.08% to 6.72%. Average US office loans nonetheless climbed from $354,434 million to $360,957 million after that sale, thanks to the Costco credit card book bought in mid-2016, which did not replace the OneMain interest income. Average loans outside the United States fell from $273,064 million to $262,715 million, partly on account of the sale, in the final quarter of 2015, of the Japanese retail banking and credit card operations. Trading account assets earned a better rate in 2016, 2.97% against 2.90%, but less capital was allocated to trading and less absolute interest income was earned from it. Finally, despite committing less capital to deposits with banks, $131,925 million against $133,853 million, the higher realised average rate lifted overall interest income, and the same pattern holds for the tax-exempt investments in US offices.

Citigroup’s cost of funding rose in every category of liability in 2016, and the increase was more pronounced outside the United States with the exception of deposit liabilities. The divergence between domestic and foreign performance carries through to the net interest margin at the bottom of the table. The margin on assets in US offices improved to 3.25% from 3.09%, while the margin outside the United States fell from 2.73% to 2.40%. The overseas margin has fallen without interruption since 2014, the year the dollar started to strengthen, and adverse translation effects have run alongside it.

An investor might reasonably exercise increased caution when watching management’s future foreign investment decisions. Nothing here confirms that overseas capital is being allocated well or that currency exposure is under control. Weaker returns may equally stem from macroeconomic conditions, such as lower and even negative yield curves overseas producing fewer profitable opportunities, and an investor should weigh those possibilities too.

One methodological caution applies to the whole average-balance analysis. It explains what happened inside a bank over a period, but it is not necessarily useful for projecting future earnings, because what was earned or paid on an average balance in a past period need have no bearing on next period’s figures. End-of-period balances and their associated rates usually make a better starting point for a forecast than average balance information.

Assessment. The quality of Citigroup’s earnings is not exceedingly high. Decreases in the provision for loan losses drove 84% of the pretax earnings increase over five years, which does nothing to relieve quality concerns, and the rising share of trading income does not instil confidence either. The margin analysis adds a third concern, with net interest margin sliding across three years and the non-US offices doing most of the damage. Earnings quality can reasonably be rated 3.

A brief overview of accounting for derivatives

Derivatives accounting is extensive, and the points below are a compressed summary of a complex topic, broadly applicable under both IFRS and US GAAP. They matter here because derivative positions sit behind much of what follows on liquidity and market sensitivity.

• At inception, many derivative contracts create neither an asset nor a liability on the balance sheet, and neither a gain nor a loss on the income statement. An interest rate swap, for example, can involve exchanging future cash flows of equivalent present value. At inception the only accounting record required for every derivative contract is disclosure of its notional amount in the notes to the financial statements.
• Measuring the mark-to-market value of a contract creates an asset or a liability and thereafter increases or decreases the value of that asset or liability.
• Changes in the value are recorded either in profit and loss or in comprehensive income, depending on the classification.
• Two hedge classifications are available at this stage: a hedge of a cash flow, and a hedge of a net investment held in a foreign subsidiary. Either classification has to be supported by evidence of correlation with the risk hedged. Where a contract qualifies as one of these two hedge types, changes in its value go to other comprehensive income and are recognised in net income over the life of the hedged transaction.
• If the contract fails to qualify as a hedge and is instead free-standing, or if it is classified as the third hedge type, a fair value hedge, then changes in fair value are reported as income or expense in each reporting period. Immediate recognition in earnings rather than in other comprehensive income can produce unexpected earnings volatility and missed earnings targets. Depending on the transaction, a secondary effect of failing hedge qualification may be a requirement to post additional collateral or cash.

Liquidity position

A bank’s liquidity matters for its own survival in stress, and because banks are interdependent through interbank deposits and derivative counterparty relationships, it matters for other banks and possibly for an entire economy. Capital alone does not assure liquidity: enough of that capital has to be available in cash or near-cash to meet obligations when they fall due. Basel III introduced two liquidity standards precisely to provide that assurance.

The Liquidity Coverage Ratio fixes the smallest share of projected outflows a bank has to hold in highly liquid form. The outflow figure is the liquidity requirement projected over 30 stressed days, and assets counted towards it must be of high quality and convertible into cash on the spot. Net outflows are built by applying prescribed factors to each liability category, then offsetting inflows from assets that mature within the 30-day window, and banks must add an amount to allow for possible maturity mismatches between contractual outflows and inflows during the period to reach total net outflows. The minimum threshold is 100%; anything below indicates an inability to meet liquidity needs.

The ratio improved over the last two years, from 113% to 121%. The 2016 figure means Citigroup can withstand cash outflows 21% higher than its 30-day stressed liquidity needs or, expressed differently, can absorb stressed outflows for 36.3 days, which is 121% applied to 30 days. Either reading indicates adequate liquidity even before allowing for the remedial management steps that would very likely be taken in a genuine stress event.

Standard number two is the Net Stable Funding Ratio, which fixes how much of the required stable funding has to come out of what is available. Required stable funding depends on the composition and maturity of the asset base; available stable funding depends on the composition and maturity of the funding sources. The NSFR is in effect an inverted LCR: where the LCR evaluates short-term liquidity, the NSFR measures the available stable funding supporting longer-term, less liquid assets such as loans, and highly liquid assets do not enter the calculation at all. A ratio of 100% is again the minimum acceptable threshold.

The NSFR was not yet a required Basel III standard at the end of 2016, with final rules expected in 2017, so the table below is a rough calculation made without the various weightings for components of available and required stable funding that the final rules would introduce. It divides the estimated unweighted available stable funding by the estimated required stable funding, using Citigroup’s consolidated balance sheet amounts.

The estimated ratio has been relatively stable since the end of 2015, and available stable funding sits well above the minimum required funding. Confirming the arithmetic: 929.4 + 206.2 + 205.9 = 1,341.5, and 1,341.5 ÷ 1,122.0 = 120%.

Assessment. Citigroup’s liquidity position is very good on both ratios, and a rating of 1 is justifiable.

Sensitivity to market risk

Market risk works on a bank’s assets and liabilities without pause, and it reaches both earnings and liquidity, so an analyst has to work out how a hostile move in rates, currencies or other market variables would feed through. Required disclosures make several sensitivities assessable. The value at risk disclosure helps in assessing exposure to market factors, and VaR statistics can be effective indicators of trends in risk-taking within a company. Because calculation assumptions differ across companies, VaR is much less useful for comparing risk-taking between companies.

At a 99% confidence level Citigroup estimates how far a position or portfolio could fall in value over an assumed one-day holding period in normal market conditions, using a Monte Carlo simulation model to capture material risk sensitivities across asset classes and risk types. Its VaR includes positions measured at fair value but excludes investment securities classified as available for sale or held to maturity.

Because risks inside and between risk types are imperfectly correlated, the covariance adjustment pulls the daily total below the sum of its parts. Trading VaR covers mark-to-market and certain fair value option positions, sets aside certain hedges, and leaves out available-for-sale and accrual exposures. The specific risk component captures issuer-level equity and fixed income risk inside the measure, and the credit portfolio holds mark-to-market positions belonging to units outside trading.

Average trading VaR fell in 2016 to $70 million from $92 million, mainly because of changes in interest rate exposures from mark-to-market hedging activity, and average trading and credit portfolio VaR fell to $92 million from $117 million. Total trading and credit portfolio VaR did rise at year end, to $105 million from $98 million, yet that worst-case one-day number stays under 1% of continuing operations net income in each year, whether on an end-of-period basis at 0.7% or an average basis at 0.6%. Against equity the magnitude is smaller still, at 0.1% end-of-period and less than 0.1% on average. The important caveat is that VaR here is a single-day measure of a market shock, while market dislocations can persist for days, weeks or longer. VaR captures very short-term shocks well and says nothing about longer-lasting market damage.

A second disclosure estimates the sensitivity of Citigroup’s capital ratios to a $100 million change in each capital measure and to a $1 billion change in risk-weighted assets. These sensitivities consider only a single change at a time, and an event affecting more than one factor could have a far greater impact than the estimate suggests.

Advanced Approaches risk-weighted assets rest mainly on models and cover credit, market and operational risk together. Under the Standardized Approach operational risk-weighted assets drop out and supervisory weights are applied wholesale to broad credit exposure categories, which leaves Advanced Approaches credit risk-weighted assets the more risk sensitive of the two. The market risk component is computed on much the same basis either way.

On either calculation, moving capital by $100 million or risk-weighted assets by $1 billion barely registers beside the ratios reported at year end: 0.90 to 1.70 basis points against ratios measured in whole percentage points. These remain static measures that adjust for one impact at a time.

Assessment. Citigroup’s sensitivity to market impacts appears controlled and provides circumstantial evidence of effective risk management. A rating of 1 is defensible.

Building the composite

Once every component has been analysed and rated, the composite assessment can be assembled. The simplest approach adds the six ratings. A bank earning the best rating of 1 on every component scores 6; a bank earning the worst on every component scores 30. Dividing the score by 6 converts it into a composite rating, which is the arithmetic mean of the component ratings. Note that where every component receives the same rating, weighting makes no difference at all.

The arithmetic mean nonetheless fails to reflect the fact that some components matter more to some analysts than others. Depending on the focus of the analysis, an analyst-weighted composite score and rating can differ appreciably from the unweighted mean.

Insurance companies provide protection against adverse events, and they earn revenue from two distinct activities. The first is premiums, the amounts paid by the purchaser of the insurance product. The second is the investment income earned on the float, being premium already collected but not yet paid out in benefits. The industry divides into property and casualty writers (P&C) and life and health writers (L&H), and their products differ on two dimensions: contract duration and variability of claims. P&C policies are usually short term, with the final cost usually known within a year of the insured event, whereas L&H policies are usually longer term. P&C claims arrive lumpier and less evenly, since accidents and comparable unforeseeable events cause them, whereas L&H claims can be anticipated with much greater confidence, since across large populations they track fairly steady actuarial mortality rates.

For both types, the important areas for analysis are business profile, earnings characteristics, investment returns, liquidity and capitalisation. For P&C companies two further areas apply: the analysis of loss reserves and the combined ratio, which is an indicator of overall underwriting profitability.

One reporting caveat frames everything that follows. Some countries, the United States among them, require insurers to prepare financial reports under statutory accounting rules that differ from US GAAP and IFRS and that place greater emphasis on solvency. In the United States the National Association of Insurance Commissioners has built a system of ratios and guideline values for solvency monitoring, the Insurance Regulatory Information System (IRIS), whose ratios are based on statutory reports. The analysis in this lesson is based on US GAAP and IFRS financial reports.

Operations, products and distribution

P&C insurers provide risk management services to insured parties. For the price of a premium they protect those parties against losses many times greater than the premiums paid. They try to minimise payouts by exercising care in underwriting and charging an adequate price for the risk they bear, and they diversify by avoiding excessive concentration in one type of policy, market or customer. They may also transfer policies, whole or in part, to reinsurers, who deal only with risks already insured by other insurers and do not originate primary policies.

The duty to perform is comparatively short compared with life insurance. Policies are often written annually and the covered event is often known with certainty inside the policy period, as with fire or weather. Some insured events take far longer to surface: environmental harm occurring during the policy period may not become obvious until well after the policy has expired.

• Property insurance covers loss of or damage to property: buildings, motor vehicles, environmental damage and other tangible items of value. The causes of loss vary, from accident and fire to theft and catastrophe, and they determine which kind of policy applies.
• Casualty insurance, which also goes by the name liability insurance, covers legal liability arising from an insured event, meaning liability owed to a third party: a passenger, an employee or a bystander.
• Multiple peril policies arise because a single event can produce both kinds of loss. An automobile accident may destroy the car and injure the passengers at once.

Products may be sold as personal lines or commercial lines depending on the customer, and some products are sold in both. Types of P&C insurance run to motor property and liability cover, which is one product sold in both lines, plus homeowner cover, workers’ compensation, marine and reinsurance.

There are two distribution methods, and the difference matters for the cost structure. Direct writers employ their own sales and marketing people, and can also reach buyers online, by mail and other direct response routes, and through affinity groups bound by a common profession or interest. Agency writers sell through brokers and through agents who may be independent or exclusive. Because direct writers employ salaried sales and marketing staff, they carry higher fixed costs. Agency writers avoid that fixed cost and instead pay commissions to agents and brokers, which is a variable cost.

The underwriting cycle

Seen from a distance, the P&C business is cyclical. It is price sensitive, with many competitors willing to cut prices to win market share. According to A.M. Best, a United States insurance rating agency, there are approximately 1,200 P&C groups in the United States comprising approximately 2,650 P&C companies, and the largest 150 of those groups wrote roughly 92% of the consolidated industry net written premium in 2015.

The cycle runs as follows. Price cutting eventually destroys profitability, producing a soft pricing market, and insurers find themselves at an uncomfortably depleted level of capital. Competition then lessens and underwriting standards tighten, producing a hard pricing market. Premiums rise, profitability returns to more reasonable levels, the improved profitability attracts new entrants, and the cycle repeats.

The mechanism running the cycle is the combined ratio, the total insurance expenses divided by net premiums earned. When the industry ratio is low the market is hard, entrants arrive, prices are cut and the cycle turns down. The effect shows up beneath the line: lower premium prices reduce total net premiums earned, so the combined ratio rises, indicating a soft market. Competitors then leave, either because they choose to stop writing unprofitable business or because they fail.

The combined ratio and its components

For a single insurer, a combined ratio above 100% indicates an underwriting loss. Under Statutory Accounting Practices in the United States the combined ratio adds together two figures taken from statutory statements, one for underwriting losses and one for expenses. The first of these puts losses, defined as claims paid plus the change in loss reserves from beginning to end of period, divided by net premiums earned, and it indicates the quality of underwriting activity, meaning the decisions on whether to accept an application and what premium to charge. The expense ratio puts underwriting expenses, sales commissions and the related employee costs included, over net premiums written, and it reports operating efficiency in acquiring and managing underwriting business.

Companies sometimes report modified versions in their financial disclosures. Travelers, for one, puts net earned premiums below the line in its expense ratio, which follows US GAAP, and other companies present the calculation differently again. This is a real comparability trap, and it is worth checking the denominator before comparing two published combined ratios.

The distinction between the two premium measures is worth stating precisely, because several ratios turn on it. Net premiums written are direct premiums written after deducting whatever has been ceded away to other insurers. Billing normally happens in advance, twice a year say, and the premium is earned across the period the policy covers. Only the net premiums written that are earned over the relevant accounting period are net premiums earned.

Loss reserves

One critical expense for a P&C insurer comes from managing its loss reserves. Proper estimation of liabilities is essential to pricing policies, because underestimating loss reserves leads to undercharging for the risks assumed. Development of the reserves is based on historical information but also incorporates estimates about future losses, so it is a material account subject to management discretion whose improper estimation carries real consequences. If the reserves and the annual adjustments to them are too optimistic, policy pricing may be insufficient for the risk borne and insolvency may follow.

A second problem is duration. The longer the insurer’s obligation runs, the harder it becomes to estimate the loss reserve properly. Asbestos cover illustrates the point. Those policies were written well before the courts began handing down larger awards, so what insurers now experience bears little resemblance to what they priced for, and awards grew fast enough to make the matching reserves extremely hard to set.

Drawn from the insurance claims footnote in Travelers’ 2016 financial statements. The roll-forward is denominated in reserves net of reinsurance recoverables expected to reduce the ultimate liability; beginning and ending balances are shown gross and then reduced by recoverables to reach net reserves.

Four observations come out of the roll-forward.

• The liability is mostly short tail. Claims paid in 2016 for current-year events of $6,220 million are 39.7% of the $15,675 million of estimated current-year claims and claim adjustments, so most of the exposure runs off quickly. The pattern in the two earlier years is much the same.
• Reinsurance is used substantially. Reinsurance means one insurer handing part of its risk to another for a premium, with the cedant expecting the reinsurer to reimburse the losses. Travelers has passed on somewhere between 16.7% and 17.6% of gross loss reserves.
• Reserve increases dominate the income statement. Reserve additions, after netting the releases on earlier-year claims, weigh on the income statement more heavily than any other expense line. The $14,995 million added in 2016 equals 63.6% of the $23,572 million of total claims and expenses reported for the year.
• Downward revisions matter more than they look. The company reduced its prior-year claims estimates by $680 million in 2016, $817 million in 2015 and $885 million in 2014. Downward revisions indicate that initial recognised reserves are being estimated conservatively, but aggressive revisions can equally be a tool for managing earnings. Against total increases of roughly $15 billion the revisions look minor; against income before taxes they are not. They contributed 16.8% of income before income taxes in 2016, 17.2% in 2015 and 17.4% in 2014.

Investment returns and liquidity

P&C insurers face much uncertainty in the risks they insure and compete hard when pricing enters its hard stage. To counteract that uncertainty they invest the collected premiums conservatively, favouring steady-return, low-risk assets and avoiding low-liquidity investments.

Investments represent 70% of total assets in both 2016 and 2015. Fixed-maturity holdings run at roughly 86% of the portfolio in both years, with almost a further 7% in short-dated paper that stands in for cash. Equities account for just 1% each year, and real estate barely registers. Concentrations merit attention as they would for any company: for a P&C insurer, concentration by type, maturity, credit quality, industry, geographic location or single issuer should each be evaluated.

Dividing total investment income by invested assets, cash plus investments, gives an estimate of performance. The metric can be computed on two bases, with and without unrealised capital gains, and the difference between the two shows how much of the total investment result depends on unrealised gains.

Because P&C insurers must stand ready to meet policy payouts, and because the timing of those payouts is uncertain, liquidity is a priority in selecting assets. The typical low-risk, steady-return instruments they hold are liquid by nature, but an analysis should still consider the overall quality of the investments and the ease with which they can be converted to cash without affecting their value. The fair value hierarchy provides direct evidence: Level 1 values are based on readily available prices from liquid markets and indicate the most liquid securities; Level 2 values are based on less liquid conditions, with prices unavailable from a liquid market and inferred instead from similar securities trading in an active market; Level 3 values rest on models and assumptions because no active market exists, implying illiquidity.

Capitalisation

Unlike banking, where international risk-based capital standards have existed since 1988, no global standard existed for the insurance sector as of mid-2016, though the IAIS was then at work on a global risk-based capital standard for the sector. It is expected to carry a target capital adequacy floor, computed by setting qualifying capital over the risk-based capital that is required.

Capital standards nonetheless exist in individual jurisdictions. Europe adopted Solvency II in 2014. Among much else it sets capital floors, and an insurer that drops beneath one obliges the supervisor in that country to step in. The NAIC risk-based capital rules in the United States, which date from the 1990s, fix a capital floor that depends on how large an insurer is and what risks it runs. For a P&C writer the formula weighs asset risk, credit risk, underwriting risk and whatever else is relevant.

An L&H company earns its revenue by writing life and health cover, and at many firms by selling investment products and services alongside it. The second main revenue source is investment income. The five areas that framed the P&C discussion apply here too: business profile, earnings characteristics, investment returns, liquidity and capitalisation. What changes is the weight given to each, because the liabilities behave differently. L&H policies run longer and their claims are more predictable, and both facts flow through to how the assets are invested and how much capital the regulator demands.

Products and distribution

Life insurance products vary widely. At the simplest end, a premium buys coverage and the beneficiary receives payment when the insured dies. Under a term life policy a benefit falls due should the insured die inside the contract’s fixed term, and the policy lapses worthless if the insured outlives it. Other products combine a death benefit with a savings vehicle. Life insurers may also offer investment products such as annuities, with fixed payments or with variable payments linked to market returns.

Health-related products vary primarily by the type of coverage. Certain products meet named medical expenses and treatments, and others pay an income where the policyholder falls ill or is injured.

Distribution runs either directly to consumers through electronic media or through agents, and the agents may be employees of the company, exclusive agents or independent agents. Distribution through independent agents is more expensive for the insurer, but it minimises fixed costs and increases the flexibility to pursue growth opportunities, which is the same fixed-versus-variable trade-off that separates direct writers from agency writers in the P&C business.

Understanding where a company’s revenue comes from, and how that has changed, is the starting point of the business profile. Diversification lowers risk, and an L&H company can spread itself across the sources of its revenue, the products it offers, the territories it covers, the channels it sells through and the assets it invests in.

Earnings characteristics

The major components of an L&H insurer’s expenses are benefit payments to policyholders under life insurance, other insurance policies, annuity contracts and other contracts. Some products that accumulate a cash value let the holder end the contract ahead of its stated maturity and take out the cash value built up. Such early cancellation is a contract surrender, and surrenders may create additional expense for the insurer.

As with P&C insurers, earnings reflect a number of accounting items requiring significant judgement and estimation.

• Future policyholder benefits and claims must be estimated on actuarial assumptions, for example about life expectancy. The amounts expensed in any period are affected both by the benefits actually paid and by interest on the estimated liability for future policyholder benefits.
• Acquisition costs for new and renewal business go onto the balance sheet and are then written off against the profits that business actually and prospectively earns, which ties reported earnings to a forecast.
• Securities valuation is a third area where accounting judgement can materially affect earnings, and it is discussed under investment returns below.

General profitability measures apply: return on assets, return on equity, growth and volatility of capital, and book value per share, along with pre-tax and post-tax operating margin, meaning operating profit as a percentage of total revenues, and pre-tax and post-tax operating return on assets and on equity. Most analysis goes further than these general measures, because L&H earnings are complex and because operational distortion and accounting estimates matter so much to reported results. Sector-specific metrics help. A.M. Best, for instance, works with benefits paid measured against the sum of net premiums written and deposits, and with commissions and expenses measured the same way.

Sector figures are drawn from the Federal Insurance Office industry report of September 2016.

In 2011 MetLife earned a higher return on average equity than the industry average, 12.20% against 4.70%, and a similar pretax operating return on average equity, 9.00% against 9.10%. After 2011, MetLife did not perform as well as the industry on either measure. Two questions follow from the table and neither is answered by it. The first is what caused the differences between MetLife and the industry. The second is why the pretax operating return on average equity and the return on average equity were similar for MetLife in 2014 and 2015, at 9.80% against 9.40% and 7.80% against 7.50%, when the two measures diverge widely in other years. Both require further investigation.

Earnings can also be distorted by accounting treatment. Mismatches between the valuation approach used for assets and that used for liabilities introduce distortion when interest rates move. In some cases significant distortions to reported earnings have arisen because assets are reported at current market values while liabilities are reported at fixed historical costs reflecting assumptions in place when the liabilities were booked.

Investment returns

For an L&H company the investment result is an important source of income, and three things dominate the evaluation: how far the portfolio is diversified, how it performs, and what interest rate exposure it carries. Portfolio liquidity matters too and is taken separately below.

Diversification starts with the allocation across asset classes and with an evaluation of how well that allocation corresponds to the insurer’s liabilities to policyholders. Because L&H claims are relatively predictable, these companies can more often seek the higher returns offered by riskier investments than P&C companies can. The cost is volatility: higher-yielding assets such as equity or real estate fluctuate in valuation more than debt investments. Returns have also been hard to come by through the decade of low interest rates, which narrowed the opportunity set and made an adequate risk-adjusted return on financial assets more difficult to earn. Concentration by type, maturity, low credit quality, industry, geographic location or single issuer can be a concern, particularly to rating agencies.

Performance is gauged as it would be for any portfolio, by setting investment income over invested assets, which means cash plus investments, and the measure can be computed using investment income plus realised gains and losses either with or without unrealised capital gains and losses. The usual gauge of interest rate risk sets asset duration alongside liability duration.

Liquidity

An L&H company’s liquidity requirements are driven by its liabilities to creditors and, primarily, by its liabilities to policyholders, which include both benefits and policy surrenders. Historically liquidity mattered less to life insurers because traditional life products were long term, but it has become more important as new products have been introduced. Liquidity arrives from operating cash flow and from how readily the investment assets convert, so any liquidity work has to take in the portfolio as a whole. Sub-investment-grade bonds and equity real estate holdings usually convert to cash less readily than investment-grade fixed income does.

Broadly, a liquidity measure sets the readier assets, cash and marketable securities among them, against the obligations falling due soonest. More elaborate measures, such as the liquidity model used by Standard & Poor’s, compare assets individually adjusted for assumptions about ready convertibility to cash against obligations individually adjusted for assumptions about the potential for withdrawals, with the adjusted amounts computed under both normal market conditions and stress. One familiar tool does not transfer. The ordinary current ratio cannot be applied straightforwardly here, since an L&H balance sheet frequently carries no split between current and non-current items in the first place.

Capitalisation

Just as for P&C writers, no worldwide risk-based capital rule governs L&H companies. Individual jurisdictions do specify the amount of capital an insurer must hold based on its risk profile, and if capital falls below the minimum a supervisory authority generally intervenes.

The differences between the two insurance businesses show up in the differences between their capital requirements, and both differences follow directly from the nature of the liabilities.

• Since claims at an L&H writer are regarded as more foreseeable than those at a P&C writer, the equity buffer required is smaller and the capital requirement correspondingly lighter.
• Life products frequently generate substantial interest rate exposure, so an L&H company’s risk-based capital calculation folds that exposure in. The P&C formula, organised around asset risk, credit risk and underwriting risk, does not have to weight it the same way.

That pair of adjustments is a good place to end, because it closes the circle the lesson opened with. A financial institution is analysed differently from a manufacturer because its liabilities determine what its assets must look like, how much capital must sit behind them and how quickly they must be convertible into cash. CAMELS for banks, the combined ratio and reserve analysis for P&C insurers, and duration matching and surrender behaviour for L&H insurers are all the same question asked in three settings: can this balance sheet meet its promises when the promises come due?