improving national statistical systems rebecca m. blank u.s. department of commerce
TRANSCRIPT
Improving National Statistical Systems
Rebecca M. Blank
U.S. Department of Commerce
Recent news has been bad
• International financial system came close to collapse
• Unemployment has risen sharply
• Output has fallen
All of this was unexpected
GDP Forecasts for 2008Forecasts as of Dec. 2007; year-over-year percent change
Canada France Germany Japan UK US-3
-2
-1
0
1
2
3
-3
-2
-1
0
1
2
3
Average of IMF, OECD, Consensus Actual
GDP Forecasts for 2009Forecasts as of Dec. 2008; year-over-year percent change
Canada* France Germany Japan UK US-6
-5
-4
-3
-2
-1
0
-6
-5
-4
-3
-2
-1
0
Average of IMF, OECD, Consensus Actual
* Based on first three quarters.
Key question
Can we (those of us responsible for economic data) do better?
Could we have provided better data that would have foreshadowed these economic problems or that would have allowed us to better understand the crisis as we were in the midst of it?
Important Note
Data problems did not cause this recession
• The question is not:
“Could we have reported better economic information?
• BUT…
“Could we have done a better job of reporting economic information?”
My Focus
Five U.S. examples:• Financial transactions• Credit information• Household wealth holdings• Industry information• Longitudinal information
Session title: “Are National Statistical Systems Effective?”
Answer to this is an unequivocal ‘yes’
They are good and getting better (and more coordinated across countries) all the time.
But all systems can be improved.
Example 1: Financial Transactions
This recession was generated by near-collapse in the global financial sector, in part due to bad debt, much of it originating in the U.S. housing market.
This financial crisis had many causes• Over-optimistic investors• Under-regulation by government• Lack of due diligence by private sector
But even once we knew there was a problem, it was difficult to use existing data to understand it.
Flow of funds data categories too aggregated. Within the U.S. we couldn’t tell which sectors held the bad debt or how much of it they held. And we couldn’t track it across national boundaries.
Possible Improvements
• More detail on the types of instruments that are held
• More detail on the institutions that are issuing those instruments
• Information on new issues and retirements of debt, as well as net flows
• Better ability to track this information across countries
Example 2: Credit Information
Financial collapse created a credit crisis:
Banks unwilling to make new loans until greater economic stability
Once recovery started, this led to substantial concern about credit constraints.
But it’s been very difficult to tell whether there are credit constraints and who they are affecting.
Need to Separate Demand and Supply for Credit
Little information on this:• In U.S., best data is a regular survey of senior
loan officers, asking for their judgment about credit supply versus credit demand.
• Most of our data just shows aggregate credit flows.
Net Borrowing by HouseholdsBillions of dollars, annual rate
2005 2007 2009-500
0
500
1,000
1,500
-500
0
500
1,000
1,500Mortgages
Consumer credit
Shading indicate recessions; end of recession is assumed to be July.
Potential Improvements
• Credit applications by type of applicant• Approvals by type of applicant
• Alternative sources of credit that are used if formal financial institutions are not lending.
Example 3: Household Wealth
Recession led to large declines in assets and wealth.
Potential effects:• Later retirement and higher labor force
participation among older workers• Less investment in children• Reduced income among those who rely on
pension investments
Timely wealth data is limited
Annual U.S. information comes from the Flow of Funds accounts, showing only aggregate information on wealth holdings among all households
Household wealth survey occurs only once every three years. Last done in 2007.
Household Net WorthTrillions of nominal dollars, year-end
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
0
20
40
60
80
0
20
40
60
80Homeowners'equity Deposits Tangible
assetsStocks, bonds, &other financial assets
2009 is third quarterFederal Reserve, Flow of funds
Possible Improvements
• Higher-frequency household-level data on wealth collections
• Distributional information in more aggregate data.
Example 4: Industry-level detail
•The recession has had a big negative effect on output, but some sectors have been hit harder than others.
•Useful data would let us see how different sectors are doing in a timely manner.
Manufacturing OutputIndex 2000 = 100
2000 2001 2002 2003 2004 2005 2006 2007 2008 200985
90
95
100
105
110
85
90
95
100
105
110
Shading represents recession. End of latest recession assumed to be July.
AnnualQuarterly
Potential Improvements
• Quarterly Industry GDP statistics• BEA has this in the current budget request to
Congress.
Example 5: Longitudinal Data
Longitudinal firm-level data can:• Look at firm births and deaths over the economic cycle• Look at which firms contract (or grow) more than others
in a time of economic change
Longitudinal household-level data can:• Show what happens to families when one member
experiences unemployment• Study unemployment spells• Investigate how changes in the housing market change
family behavior over time
Conclusions
Gaps in knowledge occur when data more aggregated. Ways to disaggregate:
• With higher frequency data• With more detailed categories of existing data• With more micro-level information• With more observations on the same unit over
time
Conclusions
Resources are limited, so this is not a plea for “let’s do it all.” Data gaps need to be prioritized.
But let’s use this crisis:• Seek gaps in the data we collect that would have
helped us to better understand the effects of rapid economic change
• Use this moment to argue for the additional resources needed to fill those gaps
Conclusions
• The purpose of the national data is to provide information. It is exactly in times of rapid change that the need for new and additional information is most visible.
• May this be a moment of opportunity to improve statistical systems so they will be even more helpful in the next economic cycle.