Transcript What do we know and not know about potential output in the

Improving Consistency in Measuring Financial Services in
the National Accounts
John Fernald*
Federal Reserve Bank of San Francisco
* All views expressed are my own and do not necessarily reflect the views
of anyone else associated with the Federal Reserve System.
1
Financial crisis highlights key questions for which we need
reliable measurement
• How important is finance for economic growth?
• How does “financial innovation” and the rise of the shadow banking
system affect the economy?
• How risky are different parts of the financial system?
2
Finance matters to the economy primarily as an intermediate
input into other activities
Credit intermediation, securities, investments, and related
activities as share of GDP
Percent
Source: BEA Input-Output Tables
3
Financial crisis highlights key questions for which we need
reliable measurement
• How important is finance for economic growth?
• How does “financial innovation” and the rise of the shadow banking
system affect the economy?
• How risky are different parts of the financial system?
4
Shadow banking system has overtaken the traditional banking
system (but fallen off sharply recently)
Source: Pozsar, Adrian, Ashcraft, and Boesky (2010). Shadow banking liabilities include items
like total repo liabilities, net securities loaned, total liabilities of ABS issuers, and shares of money
market mutual funds.
5
Financial crisis highlights key questions for which we need
reliable measurement
• How important is finance for economic growth?
• How does “financial innovation” and the rise of the shadow banking
system affect the economy?
• How risky are different parts of the financial system?
6
One inconsistency: Treatment of risk across types of
financial service providers
• Financial services are complex and intangible, so measurement is
challenging
• Compensation for many services is implicit (through interest margins),
not through explicit fees
• Are risk premia part of the (nominal) compensation for services?
• For banks, the BEA says yes, for non-banks (and the shadow banking
system) the BEA says no
• A preferable, practical fix for the inconsistency: Apply a risk-based
“user-cost” approach to measuring nominal bank output
• Consistent with theory and across providers
• Conceptually similar to what BEA already does
• Improved consistency in measuring bank output automatically
improves consistency of non-financial statistics
7
Risk-based user cost on a bank loan
• Bank lends $1 million at 10 percent
• BEA’s imputation of nominal bank services:
(rL –rRef )Loan + fees = (10%- rRef )($1 million)
•
What reference rate? Want the opportunity cost of the funds.
• BEA uses something close to a risk-free rate, e.g., Treasury rates
• If Treasury rate is 2 percent, then imputed service value is 8 percent
($80,000)
• But the bank could have invested in a junk bond, or the stock market,
or gold. Those are all foregone opportunities as well!
8
BEA approach measures nominal output of borrowing firms
differently depending on financing choices
• Consider a commercial real estate firm with a $1 million building that
yields $100,000 in leasing revenue (with no labor)
• Suppose it borrows from the bank at 10 percent interest
• Implicitly buys $80,000 in intermediate financial services
• So the firm’s “value added” – the implicit rental cost on the $1 million
building—is $20,000
• Alternatively, it could borrow from the bond market
• Pay fees of, say, $20,000 to an investment bank and rating agency, pay 8
percent interest on the loan
• Firm is recorded as buying $20,000 in intermediate financial services, and
value added is $80,000
• The firm might be indifferent, and economics might be identical…but
measurement is not
9
Common-sense answer is to treat risk the same across
financing choices
• Wang-Basu-Fernald (2004) model a bank as if it were a bond market
and a rating agency
• The correct opportunity cost of the bank’s funds is what the fixed-income
market would charge on a similar loan
• (rL –rRef )Loan + fees = (10%- 8% )($1 million) =$20,000
• If the bank offers additional services beyond a rating agency, the riskadjusted user cost measure automatically captures that
10
Adjusting for the risk of bank loans is possible
(Basu-Inklaar-Wang, 2009)
11
Risk premium (for term and default risk) varies over time
(Basu-Inklaar-Wang, 2009)
12
Payoffs to better measurement of financial output
• Measuring value added of banking system properly allows more
accurate assessment of its economic contribution
• Especially important now, when high risk premia artificially inflate
BEA measure of bank value added (all else equal)
• Future risk premia will fall as asset markets stabilize, and current method
will make it appear that the banking system is declining in importance
• Distinguish bank’s “value creation” from riskiness of its existing
portfolio
• Allows market participants to better assess the risk of investing in banks
• Will help regulators to evaluate the health of the banking system more
accurately
13
Conclusion
• Key questions are hard to answer because of inconsistencies in the
national accounting for financial services
• Need accounting based on economic function, not type of institution
• A key principle of financial reform: Regulate based on economic
function in order to minimize regulatory arbitrage
• Improving nominal consistency is both preferable and practical
• Measuring real value of bank services is also difficult, and raises
conceptual issues
• Being explicit about the services that banks actually perform helps resolve
the difficulties and leads to better measurement of real output as well
14