VoxEU Column International Finance

What is the value added of banks?

The financial system is like an organ in the body of the economy. But is it the heart or the appendix? This column, part of the Vox Debate on whether we need a financial sector, argues that we should measure the value banks create through their management of risk, not simply their bearing of risk. Under this measure, banks may well be less valuable to the economy.

Like an organ of the human body, the financial system calls most attention to itself when it malfunctions. But in normal times, is the financial system like the human heart, circulating essential capital throughout the economy? Or is it like the appendix, doing little when healthy but devastating when ill? Since the truth is probably somewhere in between, how can we calculate the contribution of finance to a modern economy? In particular, how much of the income received by financial institutions is compensation for actual services provided to their customers and how much is merely for taking on risk, such as funding risky loans with short-term borrowing?

Two kinds of income and “the bank that did nothing”

Financial companies process payments and other transactions for consumers and nonfinancial companies, screening and monitoring borrowers, as well as underwriting a variety of financial securities. The value of such services, net of purchases from other sectors, gives us the financial industry’s value added, which is the industry’s direct contribution to GDP.

It is difficult to measure even the dollar amount (let alone the inflation-adjusted real value) of financial companies’ value added.1 This is primarily because they often do not charge explicitly for services. Instead, they earn a spread between the interest rates received and the rates paid, as well as fees for writing derivatives contracts such as options and swaps. But earning interest is not in and of itself a productive activity that contributes to GDP. This is obviously sensible in the case of passive investors who buy market securities and then merely receive interest or dividends without producing new goods or services.

It seems only logical that the same principle should apply to financial institutions as well. They should not be counted as generating value added and contributing to GDP simply because they earn asset returns. The intuition behind this argument can be illustrated with a simple example. Call it “the bank that did nothing.” Imagine an entity – a “bank” – that did nothing other than recording on its balance sheet some risky loans as assets, funded mostly with short-term debt and a thin slice of equity.2 This bank would earn a handsome net income, at least during normal times, because of the higher (credit and liquidity) risk in its assets. However, by construction, we know this bank produced no services, and thus made no contribution to GDP.

In reality, financial institutions such as banks are not mere passive investors. They do hire workers and equipment to carry out various services. When the compensation for such services is not earned explicitly but bundled with asset returns that arise purely because of risk differentials, the question becomes how to tease out that portion of a bank’s overall income that is implicit revenue for services.

Value added of bank services, free of risk-based returns

The basic logic of the method for inferring the value added of bank services is intuitive. First, consider cases where a bank performs a transaction that earns it interest or fees, such as making a loan or underwriting a derivatives contract. We can ascertain what the bank would have to pay (in terms of either interest or fees) to purchase in the market a financial instrument of the same risk characteristics.3 Subtracting the value of, or interest on, the comparable market security from total income allows one to infer the implicit revenue the bank earns for services above and beyond investing passively in market securities. An analogous adjustment applies for transactions on which banks pay rather than receive interest or fees, such as those related to depositors.

The net income of a financial firm after these adjustments is a correct measure of its contribution to economic output. These adjustments strip out both risk-based returns and distortions due to explicit or implicit government subsidies. The adjusted financial sector output counts the value added of risk management but not gains from risk bearing.4 A bank will not be counted as producing more services merely because it holds more risky assets or underwrites more risky derivatives off the balance sheet, or obtains more funding through short-term debt. By comparison, if the bank arranges a suitable set of financial contracts that together deliver a customer a more desirable cash flow, then the risk-adjusted net income will capture the value added of the bank’s risk management services. If a bank under-prices risky contracts it underwrites to take advantage of government subsidy, it will be counted as producing a lower value of services.

Unfortunately, this approach is not used in the current system of GDP accounting. Currently, statisticians count banks as producing value simply by holding more risky assets, and they count all fee income as value added even if the fees simply compensate banks for bearing risks (for example, underwriting fees often contain the option value of price guarantees). This method would count “the bank that did nothing” as producing lots of services, and the more credit and liquidity risk it took on, the more productive it would become.

In my recent work with Susanto Basu and Robert Inklaar, we show that this problem is sizeable (see Basu et al. 2011). We modify the official measure in the US so that banks are no longer credited for passively receiving risk-based income.5 Making conservative assumptions, we show that the current official method overestimates the service output of the commercial banking industry by at least 21% (amounting to $116.8 billion in 2007:Q4 for example) and GDP by 0.3% ($52.9 billion in 2007:Q4 for example) between 1997 and 2007. The overstatement to GDP is mitigated because when the customer is another company, the official method simply re-assigns the firm’s value added to its bank, leaving GDP unchanged. As Figure 1 (reproduced from Figure 4 in our research) shows, the overstatement comes from counting term and credit risk premium earned by banks as part of their value added. (Banks’ assets, especially loans, are longer-dated and more risky than their liabilities.).

Figure 1. Imputed output of US commercial banks and risk compensation at current prices (1997 Q2 – 2007 Q4, $ billions)

Notes: "Deposit services" equals deposit service spread (cf. Figure 3 in Basu et al. 2011) times quarterly average deposit balance (see equation 7); "Loan services" equals loan service spead (cf. Figure 3) times quarterly average loan balance (equation 5), at all banks. "Term risk compensation" denotes the value of returns on loans and deposits due to maturities longer than 3 months. "Default risk compensation" denotes the value of pure default risk-based returns on loans.

One compelling indicator of the soundness of the risk-adjusted method is that the resulting estimates of bank value added imply more plausible estimates of the income share of capital and the return on fixed capital in the banking industry. As Table 1 shows, when banks’ value added is overstated, banking seems to generate great value for capital owners since its internal rate of return appears nearly twice as high as that of the overall private economy. Once returns due to term and credit risk are removed, banks look like the average for all US firms. (In fact, banks appear to generate slightly lower returns than the average firm, which is reasonable if holders of bank debt and equity are willing to accept lower rates of return because of explicit or implicit government guarantees, as suggested by Akerlof and Romer 1993.)

Table 1.The effect of risk adjustment: capital share and internal rate of return, average 1997-2007

Capital share in value added
Internal rate of return on fixed assets
Financial intermedation

Net of Risk-free Return

Net of Risk-adjusted Return
Private economy

Notes: Financial intermediation refers to NAICS industries 521 and 522 from the US Bureau of Economic Analysis (BEA) GDP by Industry data. The difference between BEA methodology and risk-adjusted is as in Table 1 of Basu et al. (2011). Private economy is as defined in the GDP by Industry data. The capital share in value added includes an estimate of the labor compensation of self-employed, assuming they earn the same average wage as employees. The internal rate of return on fixed assets is the return that equates capital compensation (defined as value added minus labour compensation) and the gross capital stock at current prices.

Concluding remarks

It appears likely that official statistics overstate the financial sector’s contribution to GDP, and we now have evidence that this is indeed the case. To the extent that market rates of return are available for risky assets held by financial firms, we have a method for removing returns due to risk to arrive at a ‘clean’ measure of the contribution that financial firms make to GDP. In order to estimate their productivity, however, we need to go further and estimate their inflation-adjusted real value added. Doing this requires the construction of proper deflators, an even more difficult task.6 Better measures of real financial output may overturn the current consensus that the financial industry accounted for a significant portion of the US productivity acceleration over the period 1995 to 2005.

Author's note: The opinions expressed here are those of the author, and do not necessarily represent those of the Federal Reserve Bank of Boston or the Federal Reserve System.


Akerlof, George A and Paul M Romer (1993), “Looting: The Economic Underworld of Bankruptcy for Profit”, Brookings Papers on Economic Activity, 2:1-73.

Alon, Titan, John Fernald, Robert Inklaar, and J Christina Wang (2011), “What is the value of bank output?”, Federal Reserve Bank of San Francisco Economic Letter, 16 May.

Basu, Susanto, Robert Inklaar, and J Christina Wang (2011), “The Value of Risk: Measuring the Services of U.S. Commercial Banks”, Economic Inquiry, 49(1):226-245.

Inklaar, Robert and J Christina Wang (2011), “Real Output of Bank Services: What Counts Is What Banks Do, Not What They Own”, Economica, forthcoming.

Wang, J Christina and Susanto Basu (2006), “Risk Bearing, Implicit Financial Services, and Specialization in the Financial Industry”, Federal Reserve Bank of Boston Public Policy Discussion Papers, No. 06-3.

Den Haan, Wouter (2011), “Why do we need a financial sector?”, VoxEU.org, 24 October.

Haldane, Andrew G and V Madouros (2011), “What is the contribution of the financial sector?”, VoxEU.org, 21 November.

1 See also Den Haan (2011) and Haldane and Madouros (2011).
2 This description may sound familiar; it essentially describes so-called Structured Investment Vehicles (SIVs) that held subprime mortgages, among other things. SIVs were heavily financed by short-term debt.
3 Difficulties arise when no similar market securities exist. In such cases, one reasonable solution is to use an average of the value of the most comparable market security and the value estimated from a widely accepted model.
4 See Wang and Basu (2006) for an in-depth argument for why risk bearing should best not be counted as a productive activity.
5 Alon et al. (2011) extend the estimates to the period since the financial crisis.
6 See Inklaar and Wang (2011) for empirical estimates of real output of some bank services using the U.S. data. They show that conceptual differences in how to measure real bank output lead to noticeably different empirical estimates.

9,659 Reads