2010-07-20

Serious Data Problems with Government Debt Comparison

The publication of statistics by one US Federal government agency has resulted in serious problems when comparing sovereign debt levels. The CIA World Factbook is one of the more important sources of statistical data on the internet and can usually be relied upon, and is the basis for the statistics quoted on the Wikipedia article listing countries by public debt. Nevertheless my own study of primary statistical sources, specifically official releases from Eurostat, have led me to the conclusion that the level of sovereign debt throughout the Eurozone and the European Union is actually lower, as a percentage of GDP, than the amount of sovereign debt in the US reported by the CIA.

If true, this has serious policy implications for both the United States and the European Union. It also has serious implications on market valuation of US and European sovereign debt. These implications depend upon how widely the CIA World Factbook error has been disseminated amongst policy makers, academics, econ bloggers and those who work in sovereign debt markets.

In this article I will first explain what these serious data problems are, then discuss the implications for the market and then finally offer a solution.

1. The Problem

US Federal Government Debt can be measured two ways: First is gross debt, where all debt is tallied up and measured; Second is Public debt, which measures the amount of debt owed to the public. This second measurement of debt removes what is called "Intragovernmental Holdings", which is debt owed by one government body to another. Treasury Direct runs a webpage which measures both public and intragovernmental debt. For sovereign bond markets, only public debt is traded, which means that any risk analysis of sovereign debt requires a comparison of the public debt to the nation's GDP. According to the CIA World Factbook, the size of US Federal Government public debt was estimated to be 52.9% of GDP in 2009. My own study of the data suggests that this level was around 54.1% in 2009 and, in March 2010, had reached 56.8% of GDP. By and large, therefore, the CIA figure for the United States is accurate.

To find sovereign debt data for European Union countries (including those in the Eurozone) Eurostat publishes a regular debt analysis. This latest release indicates gross government debt throughout the EU16 to be 78.7% of GDP in 2009, and 73.6% for the EU27 in 2009. Here is a comparison between selected sovereign debt levels as recorded by Eurostat and those recorded by the CIA:


At first glance these figures seem to match. There is obviously some differences in methodology being used and perhaps even timing (the CIA numbers are estimates whereas the Euostat figures are more up-to-date). These figures, however, are erroneous when used in comparison to The United States for the following reason:

European Debt figures measure Gross government debt, while the US figure measures Public Debt.

"Apples are therefore being compared to Oranges", to use the modern parlance. While the CIA points out in its country comparison that This entry records the cumulative total of all government borrowings less repayments that are denominated in a country's home currency, the figures for Europe are Gross debt levels while the figure for the United States is Public Debt. This means that, on the chart, the level of US sovereign debt appears much lower than those in Europe. If "Apples are to be compared to Apples", then either the Gross debt of European nations needs to be compared to the Gross debt of the United States, or the Public Debt of European Nations should be compared to the Public Debt of the United States. European governments do, after all, have their own levels of intragovernmental debt.

For the record, Gross debt of the United States in 2009 was 86.4% of GDP, higher than both the EU16 and EU27.

We can know that the European figures measure Gross debt because the Eurostat release explicitly says so. On page 2, footnote 1, it says: Government debt is the consolidated gross debt of the whole general government sector outstanding at the end of the year (at nominal value). The only reasonable conclusion to make is that the CIA has erred by not representing the correct data and has, instead, created statistics that seriously underestimates US sovereign debt and/or seriously overestimates European sovereign debt. Of course the CIA does point out that If data for Intra-government debt were added, "Gross Debt" would increase by about 30% of GDP, but does not adjust their chart or figures to reflect this.

2. The implications

At issue here is whether or not policy makers, academics, econ bloggers and those in the sovereign debt market have used CIA World Factbook figures in their calculations and have erroneously believed that US debt levels are considerably lower than those in Europe. Certainly a series on sovereign debt risks published at Calculated Risk has based much of its discussion upon this erroneous data, including a graph which clearly indicates low US debt at 53% in comparison to high levels of European nations. Since Calculated Risk is widely read and respected, not least by Nobel Prize winner Paul Krugman, the potential for misinformation is great.

At the time of writing (2010-07-20), markets have priced various European sovereign debt at higher risk of default. Currently Greek 10 year bonds are over 10% while safer German Euro bonds are at 2.65%. Forex Markets have also sold off the Euro in recent months as the European sovereign debt crisis continues. Yet it seems incongruous that the markets should rate the Eurozone as worse than the US Dollar since both sovereign debt and government deficits are higher in the US than they are in the Eurozone. But this assumes that the markets actually know these levels of debt. If the market is using the same erroneous data, then it stands to reason that European sovereign debt is seriously undervalued when compared to the US. This is not to say that Greek debt is risky, but it might be less risky than what the markets have priced it.

As far as the debate between deficit "hawks" and "doves" is concerned, I am certainly in disagreement with Paul Krugman and other economists over the need for more stimulus from the Federal Government. I am, therefore, a deficit "hawk". Yet all effective policy debate needs reliable data, so whether a person is a deficit "hawk" or "dove" is of no consequence in this case.

3. The Solution

One way that the Eurozone has standardised data is through the European System of Accounts (ESA95). This is what has been used in the Eurostat data report to ensure that "Apples are being compared to apples" when various EU nations report their data. Of course the US is not a part of the European Union and does not need to subscribe to ESA95, but the statistical reporters of both currency zones either need to agree on a common statistical method or at least produce data from both methods to allow for accurate comparison. Although I am not a statistician I rely heavily upon the data that statistics brings. Any errors, therefore, may be serious.

In doing my research for this issue, I discovered that German public debt is around 42.9% of GDP. Yet this number does not include the debt incurred by the Länder, the equivalent of US States - but neither does the US public debt figure found at Treasury Direct. ESA95 obviously forces member nations to tally up total government debt, including those from state and local governments. Some nations, including the UK and France, are unitary states and have no comparable devolution of fiscal policy (ie ability to tax) to lower governments. This means that debt figures for these nations will appear quite high in comparison. In order for Apples to be compared to apples, total government debt needs to be tallied up.

Another important piece of data is the level of net debt that nations have. Many nations have money saved up through various industry funds. According to the CIA World Factbook, my own country Australia has public debt of 17.6% of GDP. Yet this level is a gross figure because it does not take into account the Australian Government Future Fund, which resulted from various large budget surpluses during the Howard government's term in power: rather than using the fiscal surplus to buy back remaining sovereign debt, a market for low-risk securities was maintained while simultaneously using the budget surplus to invest back into the economy. A similar program exists in Norway, whose public debt is 60.6% of GDP according to the CIA, but which is more than exceeded by the amount saved.

Accurate economic information is essential for all involved in economics and finance. If international comparisons need to be made then the statistical agencies of OECD countries need to release data that can be compared easily and transparently. Sovereign debt data needs to be released and compared in three forms:
  • Gross total government debt (gross debt of all levels of government added together),
  • Public debt (gross total government debt minus intragovernmental holdings)
  • Net debt (public debt minus government savings)
Naturally there are differences of sovereign debt risk assessment depending upon the style of governance: A unitary state like France is responsible for all government debt while a Federal government like the USA (or maybe Germany) has advantages in not having to underwrite debt accrued by the States.

Conclusion

One inevitable question that arises from the erroneous data being disseminated is whether the CIA deliberately misrepresented the figures. While this is possible I am more of the view that the CIA simply made a mistake. Hanlon's Razor states Never attribute to malice that which is adequately explained by stupidity, and this certainly applies to this case. After all, if the CIA was actively trying to deceive world markets they did so in an awkward and amateur way.

Until the three forms of sovereign debt (listed above) become the norm, and until the CIA changes its website, comparisons between levels of sovereign debt will be, at best, opaque. Considering the trillions of dollars that make up the sovereign debt market, the current reporting of debt levels is intolerable and dangerous.

DIsclosure: The author does not directly own any US or Euro-denominated bonds. I do have fund managers looking after superannuation accounts though.

© 2010
Neil McKenzie Cameron

Creative Commons License
Serious Data Problems with Government Debt Comparison by Neil McKenzie Cameron is licensed under a Creative Commons Attribution 3.0 Unported License.

2 comments:

Jaap said...

I have been giving this a lot of thought as well (a few months ago, at the height of the Greek crisis). let's see if this can get mainstream...
at least we will need some debunking (or confirmation)

ginsbu said...

Just came across your post... You've neglected to understand the implication of *consolidated* gross debt measures as used by Eurostat, IMF, and OECD. When producing a consolidated gross debt figure intragovernmental debts are netted out, so they don't add to the total. These consolidated gross debt measures include lots of other liabilities, as well as state and local debts, which results in higher totals than US public debt figures. If you want intragovernmental debt included in the total, you should look to unconsolidated figures.

Here are some links getting into the issue:
http://www.imf.org/external/pubs/ft/gfs/manual/pdf/all.pdf
http://forums.imf.org/showthread.php?t=16860
http://epp.eurostat.ec.europa.eu/statistics_explained/index.php/Structure_of_government_debt
http://epp.eurostat.ec.europa.eu/cache/ITY_OFFPUB/KS-RA-09-017/EN/KS-RA-09-017-EN.PDF